diff options
Diffstat (limited to 'third_party/rust/serde_json/src')
39 files changed, 17119 insertions, 0 deletions
diff --git a/third_party/rust/serde_json/src/de.rs b/third_party/rust/serde_json/src/de.rs new file mode 100644 index 0000000000..88d0f2624d --- /dev/null +++ b/third_party/rust/serde_json/src/de.rs @@ -0,0 +1,2606 @@ +//! Deserialize JSON data to a Rust data structure. + +use crate::error::{Error, ErrorCode, Result}; +#[cfg(feature = "float_roundtrip")] +use crate::lexical; +use crate::number::Number; +use crate::read::{self, Fused, Reference}; +use alloc::string::String; +use alloc::vec::Vec; +#[cfg(feature = "float_roundtrip")] +use core::iter; +use core::iter::FusedIterator; +use core::marker::PhantomData; +use core::result; +use core::str::FromStr; +use serde::de::{self, Expected, Unexpected}; +use serde::forward_to_deserialize_any; + +#[cfg(feature = "arbitrary_precision")] +use crate::number::NumberDeserializer; + +pub use crate::read::{Read, SliceRead, StrRead}; + +#[cfg(feature = "std")] +pub use crate::read::IoRead; + +////////////////////////////////////////////////////////////////////////////// + +/// A structure that deserializes JSON into Rust values. +pub struct Deserializer<R> { + read: R, + scratch: Vec<u8>, + remaining_depth: u8, + #[cfg(feature = "float_roundtrip")] + single_precision: bool, + #[cfg(feature = "unbounded_depth")] + disable_recursion_limit: bool, +} + +impl<'de, R> Deserializer<R> +where + R: read::Read<'de>, +{ + /// Create a JSON deserializer from one of the possible serde_json input + /// sources. + /// + /// Typically it is more convenient to use one of these methods instead: + /// + /// - Deserializer::from_str + /// - Deserializer::from_slice + /// - Deserializer::from_reader + pub fn new(read: R) -> Self { + Deserializer { + read, + scratch: Vec::new(), + remaining_depth: 128, + #[cfg(feature = "float_roundtrip")] + single_precision: false, + #[cfg(feature = "unbounded_depth")] + disable_recursion_limit: false, + } + } +} + +#[cfg(feature = "std")] +impl<R> Deserializer<read::IoRead<R>> +where + R: crate::io::Read, +{ + /// Creates a JSON deserializer from an `io::Read`. + /// + /// Reader-based deserializers do not support deserializing borrowed types + /// like `&str`, since the `std::io::Read` trait has no non-copying methods + /// -- everything it does involves copying bytes out of the data source. + pub fn from_reader(reader: R) -> Self { + Deserializer::new(read::IoRead::new(reader)) + } +} + +impl<'a> Deserializer<read::SliceRead<'a>> { + /// Creates a JSON deserializer from a `&[u8]`. + pub fn from_slice(bytes: &'a [u8]) -> Self { + Deserializer::new(read::SliceRead::new(bytes)) + } +} + +impl<'a> Deserializer<read::StrRead<'a>> { + /// Creates a JSON deserializer from a `&str`. + pub fn from_str(s: &'a str) -> Self { + Deserializer::new(read::StrRead::new(s)) + } +} + +macro_rules! overflow { + ($a:ident * 10 + $b:ident, $c:expr) => { + match $c { + c => $a >= c / 10 && ($a > c / 10 || $b > c % 10), + } + }; +} + +pub(crate) enum ParserNumber { + F64(f64), + U64(u64), + I64(i64), + #[cfg(feature = "arbitrary_precision")] + String(String), +} + +impl ParserNumber { + fn visit<'de, V>(self, visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + match self { + ParserNumber::F64(x) => visitor.visit_f64(x), + ParserNumber::U64(x) => visitor.visit_u64(x), + ParserNumber::I64(x) => visitor.visit_i64(x), + #[cfg(feature = "arbitrary_precision")] + ParserNumber::String(x) => visitor.visit_map(NumberDeserializer { number: x.into() }), + } + } + + fn invalid_type(self, exp: &dyn Expected) -> Error { + match self { + ParserNumber::F64(x) => de::Error::invalid_type(Unexpected::Float(x), exp), + ParserNumber::U64(x) => de::Error::invalid_type(Unexpected::Unsigned(x), exp), + ParserNumber::I64(x) => de::Error::invalid_type(Unexpected::Signed(x), exp), + #[cfg(feature = "arbitrary_precision")] + ParserNumber::String(_) => de::Error::invalid_type(Unexpected::Other("number"), exp), + } + } +} + +impl<'de, R: Read<'de>> Deserializer<R> { + /// The `Deserializer::end` method should be called after a value has been fully deserialized. + /// This allows the `Deserializer` to validate that the input stream is at the end or that it + /// only has trailing whitespace. + pub fn end(&mut self) -> Result<()> { + match tri!(self.parse_whitespace()) { + Some(_) => Err(self.peek_error(ErrorCode::TrailingCharacters)), + None => Ok(()), + } + } + + /// Turn a JSON deserializer into an iterator over values of type T. + pub fn into_iter<T>(self) -> StreamDeserializer<'de, R, T> + where + T: de::Deserialize<'de>, + { + // This cannot be an implementation of std::iter::IntoIterator because + // we need the caller to choose what T is. + let offset = self.read.byte_offset(); + StreamDeserializer { + de: self, + offset, + failed: false, + output: PhantomData, + lifetime: PhantomData, + } + } + + /// Parse arbitrarily deep JSON structures without any consideration for + /// overflowing the stack. + /// + /// You will want to provide some other way to protect against stack + /// overflows, such as by wrapping your Deserializer in the dynamically + /// growing stack adapter provided by the serde_stacker crate. Additionally + /// you will need to be careful around other recursive operations on the + /// parsed result which may overflow the stack after deserialization has + /// completed, including, but not limited to, Display and Debug and Drop + /// impls. + /// + /// *This method is only available if serde_json is built with the + /// `"unbounded_depth"` feature.* + /// + /// # Examples + /// + /// ``` + /// use serde::Deserialize; + /// use serde_json::Value; + /// + /// fn main() { + /// let mut json = String::new(); + /// for _ in 0..10000 { + /// json = format!("[{}]", json); + /// } + /// + /// let mut deserializer = serde_json::Deserializer::from_str(&json); + /// deserializer.disable_recursion_limit(); + /// let deserializer = serde_stacker::Deserializer::new(&mut deserializer); + /// let value = Value::deserialize(deserializer).unwrap(); + /// + /// carefully_drop_nested_arrays(value); + /// } + /// + /// fn carefully_drop_nested_arrays(value: Value) { + /// let mut stack = vec![value]; + /// while let Some(value) = stack.pop() { + /// if let Value::Array(array) = value { + /// stack.extend(array); + /// } + /// } + /// } + /// ``` + #[cfg(feature = "unbounded_depth")] + #[cfg_attr(docsrs, doc(cfg(feature = "unbounded_depth")))] + pub fn disable_recursion_limit(&mut self) { + self.disable_recursion_limit = true; + } + + fn peek(&mut self) -> Result<Option<u8>> { + self.read.peek() + } + + fn peek_or_null(&mut self) -> Result<u8> { + Ok(tri!(self.peek()).unwrap_or(b'\x00')) + } + + fn eat_char(&mut self) { + self.read.discard(); + } + + fn next_char(&mut self) -> Result<Option<u8>> { + self.read.next() + } + + fn next_char_or_null(&mut self) -> Result<u8> { + Ok(tri!(self.next_char()).unwrap_or(b'\x00')) + } + + /// Error caused by a byte from next_char(). + #[cold] + fn error(&self, reason: ErrorCode) -> Error { + let position = self.read.position(); + Error::syntax(reason, position.line, position.column) + } + + /// Error caused by a byte from peek(). + #[cold] + fn peek_error(&self, reason: ErrorCode) -> Error { + let position = self.read.peek_position(); + Error::syntax(reason, position.line, position.column) + } + + /// Returns the first non-whitespace byte without consuming it, or `None` if + /// EOF is encountered. + fn parse_whitespace(&mut self) -> Result<Option<u8>> { + loop { + match tri!(self.peek()) { + Some(b' ') | Some(b'\n') | Some(b'\t') | Some(b'\r') => { + self.eat_char(); + } + other => { + return Ok(other); + } + } + } + } + + #[cold] + fn peek_invalid_type(&mut self, exp: &dyn Expected) -> Error { + let err = match self.peek_or_null().unwrap_or(b'\x00') { + b'n' => { + self.eat_char(); + if let Err(err) = self.parse_ident(b"ull") { + return err; + } + de::Error::invalid_type(Unexpected::Unit, exp) + } + b't' => { + self.eat_char(); + if let Err(err) = self.parse_ident(b"rue") { + return err; + } + de::Error::invalid_type(Unexpected::Bool(true), exp) + } + b'f' => { + self.eat_char(); + if let Err(err) = self.parse_ident(b"alse") { + return err; + } + de::Error::invalid_type(Unexpected::Bool(false), exp) + } + b'-' => { + self.eat_char(); + match self.parse_any_number(false) { + Ok(n) => n.invalid_type(exp), + Err(err) => return err, + } + } + b'0'..=b'9' => match self.parse_any_number(true) { + Ok(n) => n.invalid_type(exp), + Err(err) => return err, + }, + b'"' => { + self.eat_char(); + self.scratch.clear(); + match self.read.parse_str(&mut self.scratch) { + Ok(s) => de::Error::invalid_type(Unexpected::Str(&s), exp), + Err(err) => return err, + } + } + b'[' => de::Error::invalid_type(Unexpected::Seq, exp), + b'{' => de::Error::invalid_type(Unexpected::Map, exp), + _ => self.peek_error(ErrorCode::ExpectedSomeValue), + }; + + self.fix_position(err) + } + + fn deserialize_number<V>(&mut self, visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + let peek = match tri!(self.parse_whitespace()) { + Some(b) => b, + None => { + return Err(self.peek_error(ErrorCode::EofWhileParsingValue)); + } + }; + + let value = match peek { + b'-' => { + self.eat_char(); + tri!(self.parse_integer(false)).visit(visitor) + } + b'0'..=b'9' => tri!(self.parse_integer(true)).visit(visitor), + _ => Err(self.peek_invalid_type(&visitor)), + }; + + match value { + Ok(value) => Ok(value), + Err(err) => Err(self.fix_position(err)), + } + } + + fn scan_integer128(&mut self, buf: &mut String) -> Result<()> { + match tri!(self.next_char_or_null()) { + b'0' => { + buf.push('0'); + // There can be only one leading '0'. + match tri!(self.peek_or_null()) { + b'0'..=b'9' => Err(self.peek_error(ErrorCode::InvalidNumber)), + _ => Ok(()), + } + } + c @ b'1'..=b'9' => { + buf.push(c as char); + while let c @ b'0'..=b'9' = tri!(self.peek_or_null()) { + self.eat_char(); + buf.push(c as char); + } + Ok(()) + } + _ => Err(self.error(ErrorCode::InvalidNumber)), + } + } + + #[cold] + fn fix_position(&self, err: Error) -> Error { + err.fix_position(move |code| self.error(code)) + } + + fn parse_ident(&mut self, ident: &[u8]) -> Result<()> { + for expected in ident { + match tri!(self.next_char()) { + None => { + return Err(self.error(ErrorCode::EofWhileParsingValue)); + } + Some(next) => { + if next != *expected { + return Err(self.error(ErrorCode::ExpectedSomeIdent)); + } + } + } + } + + Ok(()) + } + + fn parse_integer(&mut self, positive: bool) -> Result<ParserNumber> { + let next = match tri!(self.next_char()) { + Some(b) => b, + None => { + return Err(self.error(ErrorCode::EofWhileParsingValue)); + } + }; + + match next { + b'0' => { + // There can be only one leading '0'. + match tri!(self.peek_or_null()) { + b'0'..=b'9' => Err(self.peek_error(ErrorCode::InvalidNumber)), + _ => self.parse_number(positive, 0), + } + } + c @ b'1'..=b'9' => { + let mut significand = (c - b'0') as u64; + + loop { + match tri!(self.peek_or_null()) { + c @ b'0'..=b'9' => { + let digit = (c - b'0') as u64; + + // We need to be careful with overflow. If we can, + // try to keep the number as a `u64` until we grow + // too large. At that point, switch to parsing the + // value as a `f64`. + if overflow!(significand * 10 + digit, u64::max_value()) { + return Ok(ParserNumber::F64(tri!( + self.parse_long_integer(positive, significand), + ))); + } + + self.eat_char(); + significand = significand * 10 + digit; + } + _ => { + return self.parse_number(positive, significand); + } + } + } + } + _ => Err(self.error(ErrorCode::InvalidNumber)), + } + } + + fn parse_number(&mut self, positive: bool, significand: u64) -> Result<ParserNumber> { + Ok(match tri!(self.peek_or_null()) { + b'.' => ParserNumber::F64(tri!(self.parse_decimal(positive, significand, 0))), + b'e' | b'E' => ParserNumber::F64(tri!(self.parse_exponent(positive, significand, 0))), + _ => { + if positive { + ParserNumber::U64(significand) + } else { + let neg = (significand as i64).wrapping_neg(); + + // Convert into a float if we underflow, or on `-0`. + if neg >= 0 { + ParserNumber::F64(-(significand as f64)) + } else { + ParserNumber::I64(neg) + } + } + } + }) + } + + fn parse_decimal( + &mut self, + positive: bool, + mut significand: u64, + exponent_before_decimal_point: i32, + ) -> Result<f64> { + self.eat_char(); + + let mut exponent_after_decimal_point = 0; + while let c @ b'0'..=b'9' = tri!(self.peek_or_null()) { + let digit = (c - b'0') as u64; + + if overflow!(significand * 10 + digit, u64::max_value()) { + let exponent = exponent_before_decimal_point + exponent_after_decimal_point; + return self.parse_decimal_overflow(positive, significand, exponent); + } + + self.eat_char(); + significand = significand * 10 + digit; + exponent_after_decimal_point -= 1; + } + + // Error if there is not at least one digit after the decimal point. + if exponent_after_decimal_point == 0 { + match tri!(self.peek()) { + Some(_) => return Err(self.peek_error(ErrorCode::InvalidNumber)), + None => return Err(self.peek_error(ErrorCode::EofWhileParsingValue)), + } + } + + let exponent = exponent_before_decimal_point + exponent_after_decimal_point; + match tri!(self.peek_or_null()) { + b'e' | b'E' => self.parse_exponent(positive, significand, exponent), + _ => self.f64_from_parts(positive, significand, exponent), + } + } + + fn parse_exponent( + &mut self, + positive: bool, + significand: u64, + starting_exp: i32, + ) -> Result<f64> { + self.eat_char(); + + let positive_exp = match tri!(self.peek_or_null()) { + b'+' => { + self.eat_char(); + true + } + b'-' => { + self.eat_char(); + false + } + _ => true, + }; + + let next = match tri!(self.next_char()) { + Some(b) => b, + None => { + return Err(self.error(ErrorCode::EofWhileParsingValue)); + } + }; + + // Make sure a digit follows the exponent place. + let mut exp = match next { + c @ b'0'..=b'9' => (c - b'0') as i32, + _ => { + return Err(self.error(ErrorCode::InvalidNumber)); + } + }; + + while let c @ b'0'..=b'9' = tri!(self.peek_or_null()) { + self.eat_char(); + let digit = (c - b'0') as i32; + + if overflow!(exp * 10 + digit, i32::max_value()) { + let zero_significand = significand == 0; + return self.parse_exponent_overflow(positive, zero_significand, positive_exp); + } + + exp = exp * 10 + digit; + } + + let final_exp = if positive_exp { + starting_exp.saturating_add(exp) + } else { + starting_exp.saturating_sub(exp) + }; + + self.f64_from_parts(positive, significand, final_exp) + } + + #[cfg(feature = "float_roundtrip")] + fn f64_from_parts(&mut self, positive: bool, significand: u64, exponent: i32) -> Result<f64> { + let f = if self.single_precision { + lexical::parse_concise_float::<f32>(significand, exponent) as f64 + } else { + lexical::parse_concise_float::<f64>(significand, exponent) + }; + + if f.is_infinite() { + Err(self.error(ErrorCode::NumberOutOfRange)) + } else { + Ok(if positive { f } else { -f }) + } + } + + #[cfg(not(feature = "float_roundtrip"))] + fn f64_from_parts( + &mut self, + positive: bool, + significand: u64, + mut exponent: i32, + ) -> Result<f64> { + let mut f = significand as f64; + loop { + match POW10.get(exponent.wrapping_abs() as usize) { + Some(&pow) => { + if exponent >= 0 { + f *= pow; + if f.is_infinite() { + return Err(self.error(ErrorCode::NumberOutOfRange)); + } + } else { + f /= pow; + } + break; + } + None => { + if f == 0.0 { + break; + } + if exponent >= 0 { + return Err(self.error(ErrorCode::NumberOutOfRange)); + } + f /= 1e308; + exponent += 308; + } + } + } + Ok(if positive { f } else { -f }) + } + + #[cfg(feature = "float_roundtrip")] + #[cold] + #[inline(never)] + fn parse_long_integer(&mut self, positive: bool, partial_significand: u64) -> Result<f64> { + // To deserialize floats we'll first push the integer and fraction + // parts, both as byte strings, into the scratch buffer and then feed + // both slices to lexical's parser. For example if the input is + // `12.34e5` we'll push b"1234" into scratch and then pass b"12" and + // b"34" to lexical. `integer_end` will be used to track where to split + // the scratch buffer. + // + // Note that lexical expects the integer part to contain *no* leading + // zeroes and the fraction part to contain *no* trailing zeroes. The + // first requirement is already handled by the integer parsing logic. + // The second requirement will be enforced just before passing the + // slices to lexical in f64_long_from_parts. + self.scratch.clear(); + self.scratch + .extend_from_slice(itoa::Buffer::new().format(partial_significand).as_bytes()); + + loop { + match tri!(self.peek_or_null()) { + c @ b'0'..=b'9' => { + self.scratch.push(c); + self.eat_char(); + } + b'.' => { + self.eat_char(); + return self.parse_long_decimal(positive, self.scratch.len()); + } + b'e' | b'E' => { + return self.parse_long_exponent(positive, self.scratch.len()); + } + _ => { + return self.f64_long_from_parts(positive, self.scratch.len(), 0); + } + } + } + } + + #[cfg(not(feature = "float_roundtrip"))] + #[cold] + #[inline(never)] + fn parse_long_integer(&mut self, positive: bool, significand: u64) -> Result<f64> { + let mut exponent = 0; + loop { + match tri!(self.peek_or_null()) { + b'0'..=b'9' => { + self.eat_char(); + // This could overflow... if your integer is gigabytes long. + // Ignore that possibility. + exponent += 1; + } + b'.' => { + return self.parse_decimal(positive, significand, exponent); + } + b'e' | b'E' => { + return self.parse_exponent(positive, significand, exponent); + } + _ => { + return self.f64_from_parts(positive, significand, exponent); + } + } + } + } + + #[cfg(feature = "float_roundtrip")] + #[cold] + fn parse_long_decimal(&mut self, positive: bool, integer_end: usize) -> Result<f64> { + let mut at_least_one_digit = integer_end < self.scratch.len(); + while let c @ b'0'..=b'9' = tri!(self.peek_or_null()) { + self.scratch.push(c); + self.eat_char(); + at_least_one_digit = true; + } + + if !at_least_one_digit { + match tri!(self.peek()) { + Some(_) => return Err(self.peek_error(ErrorCode::InvalidNumber)), + None => return Err(self.peek_error(ErrorCode::EofWhileParsingValue)), + } + } + + match tri!(self.peek_or_null()) { + b'e' | b'E' => self.parse_long_exponent(positive, integer_end), + _ => self.f64_long_from_parts(positive, integer_end, 0), + } + } + + #[cfg(feature = "float_roundtrip")] + fn parse_long_exponent(&mut self, positive: bool, integer_end: usize) -> Result<f64> { + self.eat_char(); + + let positive_exp = match tri!(self.peek_or_null()) { + b'+' => { + self.eat_char(); + true + } + b'-' => { + self.eat_char(); + false + } + _ => true, + }; + + let next = match tri!(self.next_char()) { + Some(b) => b, + None => { + return Err(self.error(ErrorCode::EofWhileParsingValue)); + } + }; + + // Make sure a digit follows the exponent place. + let mut exp = match next { + c @ b'0'..=b'9' => (c - b'0') as i32, + _ => { + return Err(self.error(ErrorCode::InvalidNumber)); + } + }; + + while let c @ b'0'..=b'9' = tri!(self.peek_or_null()) { + self.eat_char(); + let digit = (c - b'0') as i32; + + if overflow!(exp * 10 + digit, i32::max_value()) { + let zero_significand = self.scratch.iter().all(|&digit| digit == b'0'); + return self.parse_exponent_overflow(positive, zero_significand, positive_exp); + } + + exp = exp * 10 + digit; + } + + let final_exp = if positive_exp { exp } else { -exp }; + + self.f64_long_from_parts(positive, integer_end, final_exp) + } + + // This cold code should not be inlined into the middle of the hot + // decimal-parsing loop above. + #[cfg(feature = "float_roundtrip")] + #[cold] + #[inline(never)] + fn parse_decimal_overflow( + &mut self, + positive: bool, + significand: u64, + exponent: i32, + ) -> Result<f64> { + let mut buffer = itoa::Buffer::new(); + let significand = buffer.format(significand); + let fraction_digits = -exponent as usize; + self.scratch.clear(); + if let Some(zeros) = fraction_digits.checked_sub(significand.len() + 1) { + self.scratch.extend(iter::repeat(b'0').take(zeros + 1)); + } + self.scratch.extend_from_slice(significand.as_bytes()); + let integer_end = self.scratch.len() - fraction_digits; + self.parse_long_decimal(positive, integer_end) + } + + #[cfg(not(feature = "float_roundtrip"))] + #[cold] + #[inline(never)] + fn parse_decimal_overflow( + &mut self, + positive: bool, + significand: u64, + exponent: i32, + ) -> Result<f64> { + // The next multiply/add would overflow, so just ignore all further + // digits. + while let b'0'..=b'9' = tri!(self.peek_or_null()) { + self.eat_char(); + } + + match tri!(self.peek_or_null()) { + b'e' | b'E' => self.parse_exponent(positive, significand, exponent), + _ => self.f64_from_parts(positive, significand, exponent), + } + } + + // This cold code should not be inlined into the middle of the hot + // exponent-parsing loop above. + #[cold] + #[inline(never)] + fn parse_exponent_overflow( + &mut self, + positive: bool, + zero_significand: bool, + positive_exp: bool, + ) -> Result<f64> { + // Error instead of +/- infinity. + if !zero_significand && positive_exp { + return Err(self.error(ErrorCode::NumberOutOfRange)); + } + + while let b'0'..=b'9' = tri!(self.peek_or_null()) { + self.eat_char(); + } + Ok(if positive { 0.0 } else { -0.0 }) + } + + #[cfg(feature = "float_roundtrip")] + fn f64_long_from_parts( + &mut self, + positive: bool, + integer_end: usize, + exponent: i32, + ) -> Result<f64> { + let integer = &self.scratch[..integer_end]; + let fraction = &self.scratch[integer_end..]; + + let f = if self.single_precision { + lexical::parse_truncated_float::<f32>(integer, fraction, exponent) as f64 + } else { + lexical::parse_truncated_float::<f64>(integer, fraction, exponent) + }; + + if f.is_infinite() { + Err(self.error(ErrorCode::NumberOutOfRange)) + } else { + Ok(if positive { f } else { -f }) + } + } + + fn parse_any_signed_number(&mut self) -> Result<ParserNumber> { + let peek = match tri!(self.peek()) { + Some(b) => b, + None => { + return Err(self.peek_error(ErrorCode::EofWhileParsingValue)); + } + }; + + let value = match peek { + b'-' => { + self.eat_char(); + self.parse_any_number(false) + } + b'0'..=b'9' => self.parse_any_number(true), + _ => Err(self.peek_error(ErrorCode::InvalidNumber)), + }; + + let value = match tri!(self.peek()) { + Some(_) => Err(self.peek_error(ErrorCode::InvalidNumber)), + None => value, + }; + + match value { + Ok(value) => Ok(value), + // The de::Error impl creates errors with unknown line and column. + // Fill in the position here by looking at the current index in the + // input. There is no way to tell whether this should call `error` + // or `peek_error` so pick the one that seems correct more often. + // Worst case, the position is off by one character. + Err(err) => Err(self.fix_position(err)), + } + } + + #[cfg(not(feature = "arbitrary_precision"))] + fn parse_any_number(&mut self, positive: bool) -> Result<ParserNumber> { + self.parse_integer(positive) + } + + #[cfg(feature = "arbitrary_precision")] + fn parse_any_number(&mut self, positive: bool) -> Result<ParserNumber> { + let mut buf = String::with_capacity(16); + if !positive { + buf.push('-'); + } + self.scan_integer(&mut buf)?; + if positive { + if let Ok(unsigned) = buf.parse() { + return Ok(ParserNumber::U64(unsigned)); + } + } else { + if let Ok(signed) = buf.parse() { + return Ok(ParserNumber::I64(signed)); + } + } + Ok(ParserNumber::String(buf)) + } + + #[cfg(feature = "arbitrary_precision")] + fn scan_or_eof(&mut self, buf: &mut String) -> Result<u8> { + match tri!(self.next_char()) { + Some(b) => { + buf.push(b as char); + Ok(b) + } + None => Err(self.error(ErrorCode::EofWhileParsingValue)), + } + } + + #[cfg(feature = "arbitrary_precision")] + fn scan_integer(&mut self, buf: &mut String) -> Result<()> { + match tri!(self.scan_or_eof(buf)) { + b'0' => { + // There can be only one leading '0'. + match tri!(self.peek_or_null()) { + b'0'..=b'9' => Err(self.peek_error(ErrorCode::InvalidNumber)), + _ => self.scan_number(buf), + } + } + b'1'..=b'9' => loop { + match tri!(self.peek_or_null()) { + c @ b'0'..=b'9' => { + self.eat_char(); + buf.push(c as char); + } + _ => { + return self.scan_number(buf); + } + } + }, + _ => Err(self.error(ErrorCode::InvalidNumber)), + } + } + + #[cfg(feature = "arbitrary_precision")] + fn scan_number(&mut self, buf: &mut String) -> Result<()> { + match tri!(self.peek_or_null()) { + b'.' => self.scan_decimal(buf), + e @ b'e' | e @ b'E' => self.scan_exponent(e as char, buf), + _ => Ok(()), + } + } + + #[cfg(feature = "arbitrary_precision")] + fn scan_decimal(&mut self, buf: &mut String) -> Result<()> { + self.eat_char(); + buf.push('.'); + + let mut at_least_one_digit = false; + while let c @ b'0'..=b'9' = tri!(self.peek_or_null()) { + self.eat_char(); + buf.push(c as char); + at_least_one_digit = true; + } + + if !at_least_one_digit { + match tri!(self.peek()) { + Some(_) => return Err(self.peek_error(ErrorCode::InvalidNumber)), + None => return Err(self.peek_error(ErrorCode::EofWhileParsingValue)), + } + } + + match tri!(self.peek_or_null()) { + e @ b'e' | e @ b'E' => self.scan_exponent(e as char, buf), + _ => Ok(()), + } + } + + #[cfg(feature = "arbitrary_precision")] + fn scan_exponent(&mut self, e: char, buf: &mut String) -> Result<()> { + self.eat_char(); + buf.push(e); + + match tri!(self.peek_or_null()) { + b'+' => { + self.eat_char(); + buf.push('+'); + } + b'-' => { + self.eat_char(); + buf.push('-'); + } + _ => {} + } + + // Make sure a digit follows the exponent place. + match tri!(self.scan_or_eof(buf)) { + b'0'..=b'9' => {} + _ => { + return Err(self.error(ErrorCode::InvalidNumber)); + } + } + + while let c @ b'0'..=b'9' = tri!(self.peek_or_null()) { + self.eat_char(); + buf.push(c as char); + } + + Ok(()) + } + + fn parse_object_colon(&mut self) -> Result<()> { + match tri!(self.parse_whitespace()) { + Some(b':') => { + self.eat_char(); + Ok(()) + } + Some(_) => Err(self.peek_error(ErrorCode::ExpectedColon)), + None => Err(self.peek_error(ErrorCode::EofWhileParsingObject)), + } + } + + fn end_seq(&mut self) -> Result<()> { + match tri!(self.parse_whitespace()) { + Some(b']') => { + self.eat_char(); + Ok(()) + } + Some(b',') => { + self.eat_char(); + match self.parse_whitespace() { + Ok(Some(b']')) => Err(self.peek_error(ErrorCode::TrailingComma)), + _ => Err(self.peek_error(ErrorCode::TrailingCharacters)), + } + } + Some(_) => Err(self.peek_error(ErrorCode::TrailingCharacters)), + None => Err(self.peek_error(ErrorCode::EofWhileParsingList)), + } + } + + fn end_map(&mut self) -> Result<()> { + match tri!(self.parse_whitespace()) { + Some(b'}') => { + self.eat_char(); + Ok(()) + } + Some(b',') => Err(self.peek_error(ErrorCode::TrailingComma)), + Some(_) => Err(self.peek_error(ErrorCode::TrailingCharacters)), + None => Err(self.peek_error(ErrorCode::EofWhileParsingObject)), + } + } + + fn ignore_value(&mut self) -> Result<()> { + self.scratch.clear(); + let mut enclosing = None; + + loop { + let peek = match tri!(self.parse_whitespace()) { + Some(b) => b, + None => { + return Err(self.peek_error(ErrorCode::EofWhileParsingValue)); + } + }; + + let frame = match peek { + b'n' => { + self.eat_char(); + tri!(self.parse_ident(b"ull")); + None + } + b't' => { + self.eat_char(); + tri!(self.parse_ident(b"rue")); + None + } + b'f' => { + self.eat_char(); + tri!(self.parse_ident(b"alse")); + None + } + b'-' => { + self.eat_char(); + tri!(self.ignore_integer()); + None + } + b'0'..=b'9' => { + tri!(self.ignore_integer()); + None + } + b'"' => { + self.eat_char(); + tri!(self.read.ignore_str()); + None + } + frame @ b'[' | frame @ b'{' => { + self.scratch.extend(enclosing.take()); + self.eat_char(); + Some(frame) + } + _ => return Err(self.peek_error(ErrorCode::ExpectedSomeValue)), + }; + + let (mut accept_comma, mut frame) = match frame { + Some(frame) => (false, frame), + None => match enclosing.take() { + Some(frame) => (true, frame), + None => match self.scratch.pop() { + Some(frame) => (true, frame), + None => return Ok(()), + }, + }, + }; + + loop { + match tri!(self.parse_whitespace()) { + Some(b',') if accept_comma => { + self.eat_char(); + break; + } + Some(b']') if frame == b'[' => {} + Some(b'}') if frame == b'{' => {} + Some(_) => { + if accept_comma { + return Err(self.peek_error(match frame { + b'[' => ErrorCode::ExpectedListCommaOrEnd, + b'{' => ErrorCode::ExpectedObjectCommaOrEnd, + _ => unreachable!(), + })); + } else { + break; + } + } + None => { + return Err(self.peek_error(match frame { + b'[' => ErrorCode::EofWhileParsingList, + b'{' => ErrorCode::EofWhileParsingObject, + _ => unreachable!(), + })); + } + } + + self.eat_char(); + frame = match self.scratch.pop() { + Some(frame) => frame, + None => return Ok(()), + }; + accept_comma = true; + } + + if frame == b'{' { + match tri!(self.parse_whitespace()) { + Some(b'"') => self.eat_char(), + Some(_) => return Err(self.peek_error(ErrorCode::KeyMustBeAString)), + None => return Err(self.peek_error(ErrorCode::EofWhileParsingObject)), + } + tri!(self.read.ignore_str()); + match tri!(self.parse_whitespace()) { + Some(b':') => self.eat_char(), + Some(_) => return Err(self.peek_error(ErrorCode::ExpectedColon)), + None => return Err(self.peek_error(ErrorCode::EofWhileParsingObject)), + } + } + + enclosing = Some(frame); + } + } + + fn ignore_integer(&mut self) -> Result<()> { + match tri!(self.next_char_or_null()) { + b'0' => { + // There can be only one leading '0'. + if let b'0'..=b'9' = tri!(self.peek_or_null()) { + return Err(self.peek_error(ErrorCode::InvalidNumber)); + } + } + b'1'..=b'9' => { + while let b'0'..=b'9' = tri!(self.peek_or_null()) { + self.eat_char(); + } + } + _ => { + return Err(self.error(ErrorCode::InvalidNumber)); + } + } + + match tri!(self.peek_or_null()) { + b'.' => self.ignore_decimal(), + b'e' | b'E' => self.ignore_exponent(), + _ => Ok(()), + } + } + + fn ignore_decimal(&mut self) -> Result<()> { + self.eat_char(); + + let mut at_least_one_digit = false; + while let b'0'..=b'9' = tri!(self.peek_or_null()) { + self.eat_char(); + at_least_one_digit = true; + } + + if !at_least_one_digit { + return Err(self.peek_error(ErrorCode::InvalidNumber)); + } + + match tri!(self.peek_or_null()) { + b'e' | b'E' => self.ignore_exponent(), + _ => Ok(()), + } + } + + fn ignore_exponent(&mut self) -> Result<()> { + self.eat_char(); + + match tri!(self.peek_or_null()) { + b'+' | b'-' => self.eat_char(), + _ => {} + } + + // Make sure a digit follows the exponent place. + match tri!(self.next_char_or_null()) { + b'0'..=b'9' => {} + _ => { + return Err(self.error(ErrorCode::InvalidNumber)); + } + } + + while let b'0'..=b'9' = tri!(self.peek_or_null()) { + self.eat_char(); + } + + Ok(()) + } + + #[cfg(feature = "raw_value")] + fn deserialize_raw_value<V>(&mut self, visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + self.parse_whitespace()?; + self.read.begin_raw_buffering(); + self.ignore_value()?; + self.read.end_raw_buffering(visitor) + } +} + +impl FromStr for Number { + type Err = Error; + + fn from_str(s: &str) -> result::Result<Self, Self::Err> { + Deserializer::from_str(s) + .parse_any_signed_number() + .map(Into::into) + } +} + +#[cfg(not(feature = "float_roundtrip"))] +static POW10: [f64; 309] = [ + 1e000, 1e001, 1e002, 1e003, 1e004, 1e005, 1e006, 1e007, 1e008, 1e009, // + 1e010, 1e011, 1e012, 1e013, 1e014, 1e015, 1e016, 1e017, 1e018, 1e019, // + 1e020, 1e021, 1e022, 1e023, 1e024, 1e025, 1e026, 1e027, 1e028, 1e029, // + 1e030, 1e031, 1e032, 1e033, 1e034, 1e035, 1e036, 1e037, 1e038, 1e039, // + 1e040, 1e041, 1e042, 1e043, 1e044, 1e045, 1e046, 1e047, 1e048, 1e049, // + 1e050, 1e051, 1e052, 1e053, 1e054, 1e055, 1e056, 1e057, 1e058, 1e059, // + 1e060, 1e061, 1e062, 1e063, 1e064, 1e065, 1e066, 1e067, 1e068, 1e069, // + 1e070, 1e071, 1e072, 1e073, 1e074, 1e075, 1e076, 1e077, 1e078, 1e079, // + 1e080, 1e081, 1e082, 1e083, 1e084, 1e085, 1e086, 1e087, 1e088, 1e089, // + 1e090, 1e091, 1e092, 1e093, 1e094, 1e095, 1e096, 1e097, 1e098, 1e099, // + 1e100, 1e101, 1e102, 1e103, 1e104, 1e105, 1e106, 1e107, 1e108, 1e109, // + 1e110, 1e111, 1e112, 1e113, 1e114, 1e115, 1e116, 1e117, 1e118, 1e119, // + 1e120, 1e121, 1e122, 1e123, 1e124, 1e125, 1e126, 1e127, 1e128, 1e129, // + 1e130, 1e131, 1e132, 1e133, 1e134, 1e135, 1e136, 1e137, 1e138, 1e139, // + 1e140, 1e141, 1e142, 1e143, 1e144, 1e145, 1e146, 1e147, 1e148, 1e149, // + 1e150, 1e151, 1e152, 1e153, 1e154, 1e155, 1e156, 1e157, 1e158, 1e159, // + 1e160, 1e161, 1e162, 1e163, 1e164, 1e165, 1e166, 1e167, 1e168, 1e169, // + 1e170, 1e171, 1e172, 1e173, 1e174, 1e175, 1e176, 1e177, 1e178, 1e179, // + 1e180, 1e181, 1e182, 1e183, 1e184, 1e185, 1e186, 1e187, 1e188, 1e189, // + 1e190, 1e191, 1e192, 1e193, 1e194, 1e195, 1e196, 1e197, 1e198, 1e199, // + 1e200, 1e201, 1e202, 1e203, 1e204, 1e205, 1e206, 1e207, 1e208, 1e209, // + 1e210, 1e211, 1e212, 1e213, 1e214, 1e215, 1e216, 1e217, 1e218, 1e219, // + 1e220, 1e221, 1e222, 1e223, 1e224, 1e225, 1e226, 1e227, 1e228, 1e229, // + 1e230, 1e231, 1e232, 1e233, 1e234, 1e235, 1e236, 1e237, 1e238, 1e239, // + 1e240, 1e241, 1e242, 1e243, 1e244, 1e245, 1e246, 1e247, 1e248, 1e249, // + 1e250, 1e251, 1e252, 1e253, 1e254, 1e255, 1e256, 1e257, 1e258, 1e259, // + 1e260, 1e261, 1e262, 1e263, 1e264, 1e265, 1e266, 1e267, 1e268, 1e269, // + 1e270, 1e271, 1e272, 1e273, 1e274, 1e275, 1e276, 1e277, 1e278, 1e279, // + 1e280, 1e281, 1e282, 1e283, 1e284, 1e285, 1e286, 1e287, 1e288, 1e289, // + 1e290, 1e291, 1e292, 1e293, 1e294, 1e295, 1e296, 1e297, 1e298, 1e299, // + 1e300, 1e301, 1e302, 1e303, 1e304, 1e305, 1e306, 1e307, 1e308, +]; + +macro_rules! deserialize_number { + ($method:ident) => { + fn $method<V>(self, visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + self.deserialize_number(visitor) + } + }; +} + +#[cfg(not(feature = "unbounded_depth"))] +macro_rules! if_checking_recursion_limit { + ($($body:tt)*) => { + $($body)* + }; +} + +#[cfg(feature = "unbounded_depth")] +macro_rules! if_checking_recursion_limit { + ($this:ident $($body:tt)*) => { + if !$this.disable_recursion_limit { + $this $($body)* + } + }; +} + +macro_rules! check_recursion { + ($this:ident $($body:tt)*) => { + if_checking_recursion_limit! { + $this.remaining_depth -= 1; + if $this.remaining_depth == 0 { + return Err($this.peek_error(ErrorCode::RecursionLimitExceeded)); + } + } + + $this $($body)* + + if_checking_recursion_limit! { + $this.remaining_depth += 1; + } + }; +} + +impl<'de, 'a, R: Read<'de>> de::Deserializer<'de> for &'a mut Deserializer<R> { + type Error = Error; + + #[inline] + fn deserialize_any<V>(self, visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + let peek = match tri!(self.parse_whitespace()) { + Some(b) => b, + None => { + return Err(self.peek_error(ErrorCode::EofWhileParsingValue)); + } + }; + + let value = match peek { + b'n' => { + self.eat_char(); + tri!(self.parse_ident(b"ull")); + visitor.visit_unit() + } + b't' => { + self.eat_char(); + tri!(self.parse_ident(b"rue")); + visitor.visit_bool(true) + } + b'f' => { + self.eat_char(); + tri!(self.parse_ident(b"alse")); + visitor.visit_bool(false) + } + b'-' => { + self.eat_char(); + tri!(self.parse_any_number(false)).visit(visitor) + } + b'0'..=b'9' => tri!(self.parse_any_number(true)).visit(visitor), + b'"' => { + self.eat_char(); + self.scratch.clear(); + match tri!(self.read.parse_str(&mut self.scratch)) { + Reference::Borrowed(s) => visitor.visit_borrowed_str(s), + Reference::Copied(s) => visitor.visit_str(s), + } + } + b'[' => { + check_recursion! { + self.eat_char(); + let ret = visitor.visit_seq(SeqAccess::new(self)); + } + + match (ret, self.end_seq()) { + (Ok(ret), Ok(())) => Ok(ret), + (Err(err), _) | (_, Err(err)) => Err(err), + } + } + b'{' => { + check_recursion! { + self.eat_char(); + let ret = visitor.visit_map(MapAccess::new(self)); + } + + match (ret, self.end_map()) { + (Ok(ret), Ok(())) => Ok(ret), + (Err(err), _) | (_, Err(err)) => Err(err), + } + } + _ => Err(self.peek_error(ErrorCode::ExpectedSomeValue)), + }; + + match value { + Ok(value) => Ok(value), + // The de::Error impl creates errors with unknown line and column. + // Fill in the position here by looking at the current index in the + // input. There is no way to tell whether this should call `error` + // or `peek_error` so pick the one that seems correct more often. + // Worst case, the position is off by one character. + Err(err) => Err(self.fix_position(err)), + } + } + + fn deserialize_bool<V>(self, visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + let peek = match tri!(self.parse_whitespace()) { + Some(b) => b, + None => { + return Err(self.peek_error(ErrorCode::EofWhileParsingValue)); + } + }; + + let value = match peek { + b't' => { + self.eat_char(); + tri!(self.parse_ident(b"rue")); + visitor.visit_bool(true) + } + b'f' => { + self.eat_char(); + tri!(self.parse_ident(b"alse")); + visitor.visit_bool(false) + } + _ => Err(self.peek_invalid_type(&visitor)), + }; + + match value { + Ok(value) => Ok(value), + Err(err) => Err(self.fix_position(err)), + } + } + + deserialize_number!(deserialize_i8); + deserialize_number!(deserialize_i16); + deserialize_number!(deserialize_i32); + deserialize_number!(deserialize_i64); + deserialize_number!(deserialize_u8); + deserialize_number!(deserialize_u16); + deserialize_number!(deserialize_u32); + deserialize_number!(deserialize_u64); + #[cfg(not(feature = "float_roundtrip"))] + deserialize_number!(deserialize_f32); + deserialize_number!(deserialize_f64); + + #[cfg(feature = "float_roundtrip")] + fn deserialize_f32<V>(self, visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + self.single_precision = true; + let val = self.deserialize_number(visitor); + self.single_precision = false; + val + } + + fn deserialize_i128<V>(self, visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + let mut buf = String::new(); + + match tri!(self.parse_whitespace()) { + Some(b'-') => { + self.eat_char(); + buf.push('-'); + } + Some(_) => {} + None => { + return Err(self.peek_error(ErrorCode::EofWhileParsingValue)); + } + }; + + tri!(self.scan_integer128(&mut buf)); + + let value = match buf.parse() { + Ok(int) => visitor.visit_i128(int), + Err(_) => { + return Err(self.error(ErrorCode::NumberOutOfRange)); + } + }; + + match value { + Ok(value) => Ok(value), + Err(err) => Err(self.fix_position(err)), + } + } + + fn deserialize_u128<V>(self, visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + match tri!(self.parse_whitespace()) { + Some(b'-') => { + return Err(self.peek_error(ErrorCode::NumberOutOfRange)); + } + Some(_) => {} + None => { + return Err(self.peek_error(ErrorCode::EofWhileParsingValue)); + } + } + + let mut buf = String::new(); + tri!(self.scan_integer128(&mut buf)); + + let value = match buf.parse() { + Ok(int) => visitor.visit_u128(int), + Err(_) => { + return Err(self.error(ErrorCode::NumberOutOfRange)); + } + }; + + match value { + Ok(value) => Ok(value), + Err(err) => Err(self.fix_position(err)), + } + } + + fn deserialize_char<V>(self, visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + self.deserialize_str(visitor) + } + + fn deserialize_str<V>(self, visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + let peek = match tri!(self.parse_whitespace()) { + Some(b) => b, + None => { + return Err(self.peek_error(ErrorCode::EofWhileParsingValue)); + } + }; + + let value = match peek { + b'"' => { + self.eat_char(); + self.scratch.clear(); + match tri!(self.read.parse_str(&mut self.scratch)) { + Reference::Borrowed(s) => visitor.visit_borrowed_str(s), + Reference::Copied(s) => visitor.visit_str(s), + } + } + _ => Err(self.peek_invalid_type(&visitor)), + }; + + match value { + Ok(value) => Ok(value), + Err(err) => Err(self.fix_position(err)), + } + } + + fn deserialize_string<V>(self, visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + self.deserialize_str(visitor) + } + + /// Parses a JSON string as bytes. Note that this function does not check + /// whether the bytes represent a valid UTF-8 string. + /// + /// The relevant part of the JSON specification is Section 8.2 of [RFC + /// 7159]: + /// + /// > When all the strings represented in a JSON text are composed entirely + /// > of Unicode characters (however escaped), then that JSON text is + /// > interoperable in the sense that all software implementations that + /// > parse it will agree on the contents of names and of string values in + /// > objects and arrays. + /// > + /// > However, the ABNF in this specification allows member names and string + /// > values to contain bit sequences that cannot encode Unicode characters; + /// > for example, "\uDEAD" (a single unpaired UTF-16 surrogate). Instances + /// > of this have been observed, for example, when a library truncates a + /// > UTF-16 string without checking whether the truncation split a + /// > surrogate pair. The behavior of software that receives JSON texts + /// > containing such values is unpredictable; for example, implementations + /// > might return different values for the length of a string value or even + /// > suffer fatal runtime exceptions. + /// + /// [RFC 7159]: https://tools.ietf.org/html/rfc7159 + /// + /// The behavior of serde_json is specified to fail on non-UTF-8 strings + /// when deserializing into Rust UTF-8 string types such as String, and + /// succeed with non-UTF-8 bytes when deserializing using this method. + /// + /// Escape sequences are processed as usual, and for `\uXXXX` escapes it is + /// still checked if the hex number represents a valid Unicode code point. + /// + /// # Examples + /// + /// You can use this to parse JSON strings containing invalid UTF-8 bytes, + /// or unpaired surrogates. + /// + /// ``` + /// use serde_bytes::ByteBuf; + /// + /// fn look_at_bytes() -> Result<(), serde_json::Error> { + /// let json_data = b"\"some bytes: \xe5\x00\xe5\""; + /// let bytes: ByteBuf = serde_json::from_slice(json_data)?; + /// + /// assert_eq!(b'\xe5', bytes[12]); + /// assert_eq!(b'\0', bytes[13]); + /// assert_eq!(b'\xe5', bytes[14]); + /// + /// Ok(()) + /// } + /// # + /// # look_at_bytes().unwrap(); + /// ``` + /// + /// Backslash escape sequences like `\n` are still interpreted and required + /// to be valid. `\u` escape sequences are required to represent a valid + /// Unicode code point or lone surrogate. + /// + /// ``` + /// use serde_bytes::ByteBuf; + /// + /// fn look_at_bytes() -> Result<(), serde_json::Error> { + /// let json_data = b"\"lone surrogate: \\uD801\""; + /// let bytes: ByteBuf = serde_json::from_slice(json_data)?; + /// let expected = b"lone surrogate: \xED\xA0\x81"; + /// assert_eq!(expected, bytes.as_slice()); + /// Ok(()) + /// } + /// # + /// # look_at_bytes(); + /// ``` + fn deserialize_bytes<V>(self, visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + let peek = match tri!(self.parse_whitespace()) { + Some(b) => b, + None => { + return Err(self.peek_error(ErrorCode::EofWhileParsingValue)); + } + }; + + let value = match peek { + b'"' => { + self.eat_char(); + self.scratch.clear(); + match tri!(self.read.parse_str_raw(&mut self.scratch)) { + Reference::Borrowed(b) => visitor.visit_borrowed_bytes(b), + Reference::Copied(b) => visitor.visit_bytes(b), + } + } + b'[' => self.deserialize_seq(visitor), + _ => Err(self.peek_invalid_type(&visitor)), + }; + + match value { + Ok(value) => Ok(value), + Err(err) => Err(self.fix_position(err)), + } + } + + #[inline] + fn deserialize_byte_buf<V>(self, visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + self.deserialize_bytes(visitor) + } + + /// Parses a `null` as a None, and any other values as a `Some(...)`. + #[inline] + fn deserialize_option<V>(self, visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + match tri!(self.parse_whitespace()) { + Some(b'n') => { + self.eat_char(); + tri!(self.parse_ident(b"ull")); + visitor.visit_none() + } + _ => visitor.visit_some(self), + } + } + + fn deserialize_unit<V>(self, visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + let peek = match tri!(self.parse_whitespace()) { + Some(b) => b, + None => { + return Err(self.peek_error(ErrorCode::EofWhileParsingValue)); + } + }; + + let value = match peek { + b'n' => { + self.eat_char(); + tri!(self.parse_ident(b"ull")); + visitor.visit_unit() + } + _ => Err(self.peek_invalid_type(&visitor)), + }; + + match value { + Ok(value) => Ok(value), + Err(err) => Err(self.fix_position(err)), + } + } + + fn deserialize_unit_struct<V>(self, _name: &'static str, visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + self.deserialize_unit(visitor) + } + + /// Parses a newtype struct as the underlying value. + #[inline] + fn deserialize_newtype_struct<V>(self, name: &str, visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + #[cfg(feature = "raw_value")] + { + if name == crate::raw::TOKEN { + return self.deserialize_raw_value(visitor); + } + } + + let _ = name; + visitor.visit_newtype_struct(self) + } + + fn deserialize_seq<V>(self, visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + let peek = match tri!(self.parse_whitespace()) { + Some(b) => b, + None => { + return Err(self.peek_error(ErrorCode::EofWhileParsingValue)); + } + }; + + let value = match peek { + b'[' => { + check_recursion! { + self.eat_char(); + let ret = visitor.visit_seq(SeqAccess::new(self)); + } + + match (ret, self.end_seq()) { + (Ok(ret), Ok(())) => Ok(ret), + (Err(err), _) | (_, Err(err)) => Err(err), + } + } + _ => Err(self.peek_invalid_type(&visitor)), + }; + + match value { + Ok(value) => Ok(value), + Err(err) => Err(self.fix_position(err)), + } + } + + fn deserialize_tuple<V>(self, _len: usize, visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + self.deserialize_seq(visitor) + } + + fn deserialize_tuple_struct<V>( + self, + _name: &'static str, + _len: usize, + visitor: V, + ) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + self.deserialize_seq(visitor) + } + + fn deserialize_map<V>(self, visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + let peek = match tri!(self.parse_whitespace()) { + Some(b) => b, + None => { + return Err(self.peek_error(ErrorCode::EofWhileParsingValue)); + } + }; + + let value = match peek { + b'{' => { + check_recursion! { + self.eat_char(); + let ret = visitor.visit_map(MapAccess::new(self)); + } + + match (ret, self.end_map()) { + (Ok(ret), Ok(())) => Ok(ret), + (Err(err), _) | (_, Err(err)) => Err(err), + } + } + _ => Err(self.peek_invalid_type(&visitor)), + }; + + match value { + Ok(value) => Ok(value), + Err(err) => Err(self.fix_position(err)), + } + } + + fn deserialize_struct<V>( + self, + _name: &'static str, + _fields: &'static [&'static str], + visitor: V, + ) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + let peek = match tri!(self.parse_whitespace()) { + Some(b) => b, + None => { + return Err(self.peek_error(ErrorCode::EofWhileParsingValue)); + } + }; + + let value = match peek { + b'[' => { + check_recursion! { + self.eat_char(); + let ret = visitor.visit_seq(SeqAccess::new(self)); + } + + match (ret, self.end_seq()) { + (Ok(ret), Ok(())) => Ok(ret), + (Err(err), _) | (_, Err(err)) => Err(err), + } + } + b'{' => { + check_recursion! { + self.eat_char(); + let ret = visitor.visit_map(MapAccess::new(self)); + } + + match (ret, self.end_map()) { + (Ok(ret), Ok(())) => Ok(ret), + (Err(err), _) | (_, Err(err)) => Err(err), + } + } + _ => Err(self.peek_invalid_type(&visitor)), + }; + + match value { + Ok(value) => Ok(value), + Err(err) => Err(self.fix_position(err)), + } + } + + /// Parses an enum as an object like `{"$KEY":$VALUE}`, where $VALUE is either a straight + /// value, a `[..]`, or a `{..}`. + #[inline] + fn deserialize_enum<V>( + self, + _name: &str, + _variants: &'static [&'static str], + visitor: V, + ) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + match tri!(self.parse_whitespace()) { + Some(b'{') => { + check_recursion! { + self.eat_char(); + let value = tri!(visitor.visit_enum(VariantAccess::new(self))); + } + + match tri!(self.parse_whitespace()) { + Some(b'}') => { + self.eat_char(); + Ok(value) + } + Some(_) => Err(self.error(ErrorCode::ExpectedSomeValue)), + None => Err(self.error(ErrorCode::EofWhileParsingObject)), + } + } + Some(b'"') => visitor.visit_enum(UnitVariantAccess::new(self)), + Some(_) => Err(self.peek_error(ErrorCode::ExpectedSomeValue)), + None => Err(self.peek_error(ErrorCode::EofWhileParsingValue)), + } + } + + fn deserialize_identifier<V>(self, visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + self.deserialize_str(visitor) + } + + fn deserialize_ignored_any<V>(self, visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + tri!(self.ignore_value()); + visitor.visit_unit() + } +} + +struct SeqAccess<'a, R: 'a> { + de: &'a mut Deserializer<R>, + first: bool, +} + +impl<'a, R: 'a> SeqAccess<'a, R> { + fn new(de: &'a mut Deserializer<R>) -> Self { + SeqAccess { de, first: true } + } +} + +impl<'de, 'a, R: Read<'de> + 'a> de::SeqAccess<'de> for SeqAccess<'a, R> { + type Error = Error; + + fn next_element_seed<T>(&mut self, seed: T) -> Result<Option<T::Value>> + where + T: de::DeserializeSeed<'de>, + { + let peek = match tri!(self.de.parse_whitespace()) { + Some(b']') => { + return Ok(None); + } + Some(b',') if !self.first => { + self.de.eat_char(); + tri!(self.de.parse_whitespace()) + } + Some(b) => { + if self.first { + self.first = false; + Some(b) + } else { + return Err(self.de.peek_error(ErrorCode::ExpectedListCommaOrEnd)); + } + } + None => { + return Err(self.de.peek_error(ErrorCode::EofWhileParsingList)); + } + }; + + match peek { + Some(b']') => Err(self.de.peek_error(ErrorCode::TrailingComma)), + Some(_) => Ok(Some(tri!(seed.deserialize(&mut *self.de)))), + None => Err(self.de.peek_error(ErrorCode::EofWhileParsingValue)), + } + } +} + +struct MapAccess<'a, R: 'a> { + de: &'a mut Deserializer<R>, + first: bool, +} + +impl<'a, R: 'a> MapAccess<'a, R> { + fn new(de: &'a mut Deserializer<R>) -> Self { + MapAccess { de, first: true } + } +} + +impl<'de, 'a, R: Read<'de> + 'a> de::MapAccess<'de> for MapAccess<'a, R> { + type Error = Error; + + fn next_key_seed<K>(&mut self, seed: K) -> Result<Option<K::Value>> + where + K: de::DeserializeSeed<'de>, + { + let peek = match tri!(self.de.parse_whitespace()) { + Some(b'}') => { + return Ok(None); + } + Some(b',') if !self.first => { + self.de.eat_char(); + tri!(self.de.parse_whitespace()) + } + Some(b) => { + if self.first { + self.first = false; + Some(b) + } else { + return Err(self.de.peek_error(ErrorCode::ExpectedObjectCommaOrEnd)); + } + } + None => { + return Err(self.de.peek_error(ErrorCode::EofWhileParsingObject)); + } + }; + + match peek { + Some(b'"') => seed.deserialize(MapKey { de: &mut *self.de }).map(Some), + Some(b'}') => Err(self.de.peek_error(ErrorCode::TrailingComma)), + Some(_) => Err(self.de.peek_error(ErrorCode::KeyMustBeAString)), + None => Err(self.de.peek_error(ErrorCode::EofWhileParsingValue)), + } + } + + fn next_value_seed<V>(&mut self, seed: V) -> Result<V::Value> + where + V: de::DeserializeSeed<'de>, + { + tri!(self.de.parse_object_colon()); + + seed.deserialize(&mut *self.de) + } +} + +struct VariantAccess<'a, R: 'a> { + de: &'a mut Deserializer<R>, +} + +impl<'a, R: 'a> VariantAccess<'a, R> { + fn new(de: &'a mut Deserializer<R>) -> Self { + VariantAccess { de } + } +} + +impl<'de, 'a, R: Read<'de> + 'a> de::EnumAccess<'de> for VariantAccess<'a, R> { + type Error = Error; + type Variant = Self; + + fn variant_seed<V>(self, seed: V) -> Result<(V::Value, Self)> + where + V: de::DeserializeSeed<'de>, + { + let val = tri!(seed.deserialize(&mut *self.de)); + tri!(self.de.parse_object_colon()); + Ok((val, self)) + } +} + +impl<'de, 'a, R: Read<'de> + 'a> de::VariantAccess<'de> for VariantAccess<'a, R> { + type Error = Error; + + fn unit_variant(self) -> Result<()> { + de::Deserialize::deserialize(self.de) + } + + fn newtype_variant_seed<T>(self, seed: T) -> Result<T::Value> + where + T: de::DeserializeSeed<'de>, + { + seed.deserialize(self.de) + } + + fn tuple_variant<V>(self, _len: usize, visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + de::Deserializer::deserialize_seq(self.de, visitor) + } + + fn struct_variant<V>(self, fields: &'static [&'static str], visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + de::Deserializer::deserialize_struct(self.de, "", fields, visitor) + } +} + +struct UnitVariantAccess<'a, R: 'a> { + de: &'a mut Deserializer<R>, +} + +impl<'a, R: 'a> UnitVariantAccess<'a, R> { + fn new(de: &'a mut Deserializer<R>) -> Self { + UnitVariantAccess { de } + } +} + +impl<'de, 'a, R: Read<'de> + 'a> de::EnumAccess<'de> for UnitVariantAccess<'a, R> { + type Error = Error; + type Variant = Self; + + fn variant_seed<V>(self, seed: V) -> Result<(V::Value, Self)> + where + V: de::DeserializeSeed<'de>, + { + let variant = tri!(seed.deserialize(&mut *self.de)); + Ok((variant, self)) + } +} + +impl<'de, 'a, R: Read<'de> + 'a> de::VariantAccess<'de> for UnitVariantAccess<'a, R> { + type Error = Error; + + fn unit_variant(self) -> Result<()> { + Ok(()) + } + + fn newtype_variant_seed<T>(self, _seed: T) -> Result<T::Value> + where + T: de::DeserializeSeed<'de>, + { + Err(de::Error::invalid_type( + Unexpected::UnitVariant, + &"newtype variant", + )) + } + + fn tuple_variant<V>(self, _len: usize, _visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + Err(de::Error::invalid_type( + Unexpected::UnitVariant, + &"tuple variant", + )) + } + + fn struct_variant<V>(self, _fields: &'static [&'static str], _visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + Err(de::Error::invalid_type( + Unexpected::UnitVariant, + &"struct variant", + )) + } +} + +/// Only deserialize from this after peeking a '"' byte! Otherwise it may +/// deserialize invalid JSON successfully. +struct MapKey<'a, R: 'a> { + de: &'a mut Deserializer<R>, +} + +macro_rules! deserialize_integer_key { + ($method:ident => $visit:ident) => { + fn $method<V>(self, visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + self.de.eat_char(); + self.de.scratch.clear(); + let string = tri!(self.de.read.parse_str(&mut self.de.scratch)); + match (string.parse(), string) { + (Ok(integer), _) => visitor.$visit(integer), + (Err(_), Reference::Borrowed(s)) => visitor.visit_borrowed_str(s), + (Err(_), Reference::Copied(s)) => visitor.visit_str(s), + } + } + }; +} + +impl<'de, 'a, R> de::Deserializer<'de> for MapKey<'a, R> +where + R: Read<'de>, +{ + type Error = Error; + + #[inline] + fn deserialize_any<V>(self, visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + self.de.eat_char(); + self.de.scratch.clear(); + match tri!(self.de.read.parse_str(&mut self.de.scratch)) { + Reference::Borrowed(s) => visitor.visit_borrowed_str(s), + Reference::Copied(s) => visitor.visit_str(s), + } + } + + deserialize_integer_key!(deserialize_i8 => visit_i8); + deserialize_integer_key!(deserialize_i16 => visit_i16); + deserialize_integer_key!(deserialize_i32 => visit_i32); + deserialize_integer_key!(deserialize_i64 => visit_i64); + deserialize_integer_key!(deserialize_i128 => visit_i128); + deserialize_integer_key!(deserialize_u8 => visit_u8); + deserialize_integer_key!(deserialize_u16 => visit_u16); + deserialize_integer_key!(deserialize_u32 => visit_u32); + deserialize_integer_key!(deserialize_u64 => visit_u64); + deserialize_integer_key!(deserialize_u128 => visit_u128); + + #[inline] + fn deserialize_option<V>(self, visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + // Map keys cannot be null. + visitor.visit_some(self) + } + + #[inline] + fn deserialize_newtype_struct<V>(self, name: &'static str, visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + #[cfg(feature = "raw_value")] + { + if name == crate::raw::TOKEN { + return self.de.deserialize_raw_value(visitor); + } + } + + let _ = name; + visitor.visit_newtype_struct(self) + } + + #[inline] + fn deserialize_enum<V>( + self, + name: &'static str, + variants: &'static [&'static str], + visitor: V, + ) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + self.de.deserialize_enum(name, variants, visitor) + } + + #[inline] + fn deserialize_bytes<V>(self, visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + self.de.deserialize_bytes(visitor) + } + + #[inline] + fn deserialize_byte_buf<V>(self, visitor: V) -> Result<V::Value> + where + V: de::Visitor<'de>, + { + self.de.deserialize_bytes(visitor) + } + + forward_to_deserialize_any! { + bool f32 f64 char str string unit unit_struct seq tuple tuple_struct map + struct identifier ignored_any + } +} + +////////////////////////////////////////////////////////////////////////////// + +/// Iterator that deserializes a stream into multiple JSON values. +/// +/// A stream deserializer can be created from any JSON deserializer using the +/// `Deserializer::into_iter` method. +/// +/// The data can consist of any JSON value. Values need to be a self-delineating value e.g. +/// arrays, objects, or strings, or be followed by whitespace or a self-delineating value. +/// +/// ``` +/// use serde_json::{Deserializer, Value}; +/// +/// fn main() { +/// let data = "{\"k\": 3}1\"cool\"\"stuff\" 3{} [0, 1, 2]"; +/// +/// let stream = Deserializer::from_str(data).into_iter::<Value>(); +/// +/// for value in stream { +/// println!("{}", value.unwrap()); +/// } +/// } +/// ``` +pub struct StreamDeserializer<'de, R, T> { + de: Deserializer<R>, + offset: usize, + failed: bool, + output: PhantomData<T>, + lifetime: PhantomData<&'de ()>, +} + +impl<'de, R, T> StreamDeserializer<'de, R, T> +where + R: read::Read<'de>, + T: de::Deserialize<'de>, +{ + /// Create a JSON stream deserializer from one of the possible serde_json + /// input sources. + /// + /// Typically it is more convenient to use one of these methods instead: + /// + /// - Deserializer::from_str(...).into_iter() + /// - Deserializer::from_slice(...).into_iter() + /// - Deserializer::from_reader(...).into_iter() + pub fn new(read: R) -> Self { + let offset = read.byte_offset(); + StreamDeserializer { + de: Deserializer::new(read), + offset, + failed: false, + output: PhantomData, + lifetime: PhantomData, + } + } + + /// Returns the number of bytes so far deserialized into a successful `T`. + /// + /// If a stream deserializer returns an EOF error, new data can be joined to + /// `old_data[stream.byte_offset()..]` to try again. + /// + /// ``` + /// let data = b"[0] [1] ["; + /// + /// let de = serde_json::Deserializer::from_slice(data); + /// let mut stream = de.into_iter::<Vec<i32>>(); + /// assert_eq!(0, stream.byte_offset()); + /// + /// println!("{:?}", stream.next()); // [0] + /// assert_eq!(3, stream.byte_offset()); + /// + /// println!("{:?}", stream.next()); // [1] + /// assert_eq!(7, stream.byte_offset()); + /// + /// println!("{:?}", stream.next()); // error + /// assert_eq!(8, stream.byte_offset()); + /// + /// // If err.is_eof(), can join the remaining data to new data and continue. + /// let remaining = &data[stream.byte_offset()..]; + /// ``` + /// + /// *Note:* In the future this method may be changed to return the number of + /// bytes so far deserialized into a successful T *or* syntactically valid + /// JSON skipped over due to a type error. See [serde-rs/json#70] for an + /// example illustrating this. + /// + /// [serde-rs/json#70]: https://github.com/serde-rs/json/issues/70 + pub fn byte_offset(&self) -> usize { + self.offset + } + + fn peek_end_of_value(&mut self) -> Result<()> { + match tri!(self.de.peek()) { + Some(b' ') | Some(b'\n') | Some(b'\t') | Some(b'\r') | Some(b'"') | Some(b'[') + | Some(b']') | Some(b'{') | Some(b'}') | Some(b',') | Some(b':') | None => Ok(()), + Some(_) => { + let position = self.de.read.peek_position(); + Err(Error::syntax( + ErrorCode::TrailingCharacters, + position.line, + position.column, + )) + } + } + } +} + +impl<'de, R, T> Iterator for StreamDeserializer<'de, R, T> +where + R: Read<'de>, + T: de::Deserialize<'de>, +{ + type Item = Result<T>; + + fn next(&mut self) -> Option<Result<T>> { + if R::should_early_return_if_failed && self.failed { + return None; + } + + // skip whitespaces, if any + // this helps with trailing whitespaces, since whitespaces between + // values are handled for us. + match self.de.parse_whitespace() { + Ok(None) => { + self.offset = self.de.read.byte_offset(); + None + } + Ok(Some(b)) => { + // If the value does not have a clear way to show the end of the value + // (like numbers, null, true etc.) we have to look for whitespace or + // the beginning of a self-delineated value. + let self_delineated_value = match b { + b'[' | b'"' | b'{' => true, + _ => false, + }; + self.offset = self.de.read.byte_offset(); + let result = de::Deserialize::deserialize(&mut self.de); + + Some(match result { + Ok(value) => { + self.offset = self.de.read.byte_offset(); + if self_delineated_value { + Ok(value) + } else { + self.peek_end_of_value().map(|_| value) + } + } + Err(e) => { + self.de.read.set_failed(&mut self.failed); + Err(e) + } + }) + } + Err(e) => { + self.de.read.set_failed(&mut self.failed); + Some(Err(e)) + } + } + } +} + +impl<'de, R, T> FusedIterator for StreamDeserializer<'de, R, T> +where + R: Read<'de> + Fused, + T: de::Deserialize<'de>, +{ +} + +////////////////////////////////////////////////////////////////////////////// + +fn from_trait<'de, R, T>(read: R) -> Result<T> +where + R: Read<'de>, + T: de::Deserialize<'de>, +{ + let mut de = Deserializer::new(read); + let value = tri!(de::Deserialize::deserialize(&mut de)); + + // Make sure the whole stream has been consumed. + tri!(de.end()); + Ok(value) +} + +/// Deserialize an instance of type `T` from an IO stream of JSON. +/// +/// The content of the IO stream is deserialized directly from the stream +/// without being buffered in memory by serde_json. +/// +/// When reading from a source against which short reads are not efficient, such +/// as a [`File`], you will want to apply your own buffering because serde_json +/// will not buffer the input. See [`std::io::BufReader`]. +/// +/// It is expected that the input stream ends after the deserialized object. +/// If the stream does not end, such as in the case of a persistent socket connection, +/// this function will not return. It is possible instead to deserialize from a prefix of an input +/// stream without looking for EOF by managing your own [`Deserializer`]. +/// +/// Note that counter to intuition, this function is usually slower than +/// reading a file completely into memory and then applying [`from_str`] +/// or [`from_slice`] on it. See [issue #160]. +/// +/// [`File`]: https://doc.rust-lang.org/std/fs/struct.File.html +/// [`std::io::BufReader`]: https://doc.rust-lang.org/std/io/struct.BufReader.html +/// [`from_str`]: ./fn.from_str.html +/// [`from_slice`]: ./fn.from_slice.html +/// [issue #160]: https://github.com/serde-rs/json/issues/160 +/// +/// # Example +/// +/// Reading the contents of a file. +/// +/// ``` +/// use serde::Deserialize; +/// +/// use std::error::Error; +/// use std::fs::File; +/// use std::io::BufReader; +/// use std::path::Path; +/// +/// #[derive(Deserialize, Debug)] +/// struct User { +/// fingerprint: String, +/// location: String, +/// } +/// +/// fn read_user_from_file<P: AsRef<Path>>(path: P) -> Result<User, Box<dyn Error>> { +/// // Open the file in read-only mode with buffer. +/// let file = File::open(path)?; +/// let reader = BufReader::new(file); +/// +/// // Read the JSON contents of the file as an instance of `User`. +/// let u = serde_json::from_reader(reader)?; +/// +/// // Return the `User`. +/// Ok(u) +/// } +/// +/// fn main() { +/// # } +/// # fn fake_main() { +/// let u = read_user_from_file("test.json").unwrap(); +/// println!("{:#?}", u); +/// } +/// ``` +/// +/// Reading from a persistent socket connection. +/// +/// ``` +/// use serde::Deserialize; +/// +/// use std::error::Error; +/// use std::net::{TcpListener, TcpStream}; +/// +/// #[derive(Deserialize, Debug)] +/// struct User { +/// fingerprint: String, +/// location: String, +/// } +/// +/// fn read_user_from_stream(tcp_stream: TcpStream) -> Result<User, Box<dyn Error>> { +/// let mut de = serde_json::Deserializer::from_reader(tcp_stream); +/// let u = User::deserialize(&mut de)?; +/// +/// Ok(u) +/// } +/// +/// fn main() { +/// # } +/// # fn fake_main() { +/// let listener = TcpListener::bind("127.0.0.1:4000").unwrap(); +/// +/// for stream in listener.incoming() { +/// println!("{:#?}", read_user_from_stream(stream.unwrap())); +/// } +/// } +/// ``` +/// +/// # Errors +/// +/// This conversion can fail if the structure of the input does not match the +/// structure expected by `T`, for example if `T` is a struct type but the input +/// contains something other than a JSON map. It can also fail if the structure +/// is correct but `T`'s implementation of `Deserialize` decides that something +/// is wrong with the data, for example required struct fields are missing from +/// the JSON map or some number is too big to fit in the expected primitive +/// type. +#[cfg(feature = "std")] +#[cfg_attr(docsrs, doc(cfg(feature = "std")))] +pub fn from_reader<R, T>(rdr: R) -> Result<T> +where + R: crate::io::Read, + T: de::DeserializeOwned, +{ + from_trait(read::IoRead::new(rdr)) +} + +/// Deserialize an instance of type `T` from bytes of JSON text. +/// +/// # Example +/// +/// ``` +/// use serde::Deserialize; +/// +/// #[derive(Deserialize, Debug)] +/// struct User { +/// fingerprint: String, +/// location: String, +/// } +/// +/// fn main() { +/// // The type of `j` is `&[u8]` +/// let j = b" +/// { +/// \"fingerprint\": \"0xF9BA143B95FF6D82\", +/// \"location\": \"Menlo Park, CA\" +/// }"; +/// +/// let u: User = serde_json::from_slice(j).unwrap(); +/// println!("{:#?}", u); +/// } +/// ``` +/// +/// # Errors +/// +/// This conversion can fail if the structure of the input does not match the +/// structure expected by `T`, for example if `T` is a struct type but the input +/// contains something other than a JSON map. It can also fail if the structure +/// is correct but `T`'s implementation of `Deserialize` decides that something +/// is wrong with the data, for example required struct fields are missing from +/// the JSON map or some number is too big to fit in the expected primitive +/// type. +pub fn from_slice<'a, T>(v: &'a [u8]) -> Result<T> +where + T: de::Deserialize<'a>, +{ + from_trait(read::SliceRead::new(v)) +} + +/// Deserialize an instance of type `T` from a string of JSON text. +/// +/// # Example +/// +/// ``` +/// use serde::Deserialize; +/// +/// #[derive(Deserialize, Debug)] +/// struct User { +/// fingerprint: String, +/// location: String, +/// } +/// +/// fn main() { +/// // The type of `j` is `&str` +/// let j = " +/// { +/// \"fingerprint\": \"0xF9BA143B95FF6D82\", +/// \"location\": \"Menlo Park, CA\" +/// }"; +/// +/// let u: User = serde_json::from_str(j).unwrap(); +/// println!("{:#?}", u); +/// } +/// ``` +/// +/// # Errors +/// +/// This conversion can fail if the structure of the input does not match the +/// structure expected by `T`, for example if `T` is a struct type but the input +/// contains something other than a JSON map. It can also fail if the structure +/// is correct but `T`'s implementation of `Deserialize` decides that something +/// is wrong with the data, for example required struct fields are missing from +/// the JSON map or some number is too big to fit in the expected primitive +/// type. +pub fn from_str<'a, T>(s: &'a str) -> Result<T> +where + T: de::Deserialize<'a>, +{ + from_trait(read::StrRead::new(s)) +} diff --git a/third_party/rust/serde_json/src/error.rs b/third_party/rust/serde_json/src/error.rs new file mode 100644 index 0000000000..1875ef08b0 --- /dev/null +++ b/third_party/rust/serde_json/src/error.rs @@ -0,0 +1,445 @@ +//! When serializing or deserializing JSON goes wrong. + +use crate::io; +use alloc::boxed::Box; +use alloc::string::{String, ToString}; +use core::fmt::{self, Debug, Display}; +use core::result; +use core::str::FromStr; +use serde::{de, ser}; +#[cfg(feature = "std")] +use std::error; + +/// This type represents all possible errors that can occur when serializing or +/// deserializing JSON data. +pub struct Error { + /// This `Box` allows us to keep the size of `Error` as small as possible. A + /// larger `Error` type was substantially slower due to all the functions + /// that pass around `Result<T, Error>`. + err: Box<ErrorImpl>, +} + +/// Alias for a `Result` with the error type `serde_json::Error`. +pub type Result<T> = result::Result<T, Error>; + +impl Error { + /// One-based line number at which the error was detected. + /// + /// Characters in the first line of the input (before the first newline + /// character) are in line 1. + pub fn line(&self) -> usize { + self.err.line + } + + /// One-based column number at which the error was detected. + /// + /// The first character in the input and any characters immediately + /// following a newline character are in column 1. + /// + /// Note that errors may occur in column 0, for example if a read from an IO + /// stream fails immediately following a previously read newline character. + pub fn column(&self) -> usize { + self.err.column + } + + /// Categorizes the cause of this error. + /// + /// - `Category::Io` - failure to read or write bytes on an IO stream + /// - `Category::Syntax` - input that is not syntactically valid JSON + /// - `Category::Data` - input data that is semantically incorrect + /// - `Category::Eof` - unexpected end of the input data + pub fn classify(&self) -> Category { + match self.err.code { + ErrorCode::Message(_) => Category::Data, + ErrorCode::Io(_) => Category::Io, + ErrorCode::EofWhileParsingList + | ErrorCode::EofWhileParsingObject + | ErrorCode::EofWhileParsingString + | ErrorCode::EofWhileParsingValue => Category::Eof, + ErrorCode::ExpectedColon + | ErrorCode::ExpectedListCommaOrEnd + | ErrorCode::ExpectedObjectCommaOrEnd + | ErrorCode::ExpectedSomeIdent + | ErrorCode::ExpectedSomeValue + | ErrorCode::InvalidEscape + | ErrorCode::InvalidNumber + | ErrorCode::NumberOutOfRange + | ErrorCode::InvalidUnicodeCodePoint + | ErrorCode::ControlCharacterWhileParsingString + | ErrorCode::KeyMustBeAString + | ErrorCode::LoneLeadingSurrogateInHexEscape + | ErrorCode::TrailingComma + | ErrorCode::TrailingCharacters + | ErrorCode::UnexpectedEndOfHexEscape + | ErrorCode::RecursionLimitExceeded => Category::Syntax, + } + } + + /// Returns true if this error was caused by a failure to read or write + /// bytes on an IO stream. + pub fn is_io(&self) -> bool { + self.classify() == Category::Io + } + + /// Returns true if this error was caused by input that was not + /// syntactically valid JSON. + pub fn is_syntax(&self) -> bool { + self.classify() == Category::Syntax + } + + /// Returns true if this error was caused by input data that was + /// semantically incorrect. + /// + /// For example, JSON containing a number is semantically incorrect when the + /// type being deserialized into holds a String. + pub fn is_data(&self) -> bool { + self.classify() == Category::Data + } + + /// Returns true if this error was caused by prematurely reaching the end of + /// the input data. + /// + /// Callers that process streaming input may be interested in retrying the + /// deserialization once more data is available. + pub fn is_eof(&self) -> bool { + self.classify() == Category::Eof + } +} + +/// Categorizes the cause of a `serde_json::Error`. +#[derive(Copy, Clone, PartialEq, Eq, Debug)] +pub enum Category { + /// The error was caused by a failure to read or write bytes on an IO + /// stream. + Io, + + /// The error was caused by input that was not syntactically valid JSON. + Syntax, + + /// The error was caused by input data that was semantically incorrect. + /// + /// For example, JSON containing a number is semantically incorrect when the + /// type being deserialized into holds a String. + Data, + + /// The error was caused by prematurely reaching the end of the input data. + /// + /// Callers that process streaming input may be interested in retrying the + /// deserialization once more data is available. + Eof, +} + +#[cfg(feature = "std")] +#[allow(clippy::fallible_impl_from)] +impl From<Error> for io::Error { + /// Convert a `serde_json::Error` into an `io::Error`. + /// + /// JSON syntax and data errors are turned into `InvalidData` IO errors. + /// EOF errors are turned into `UnexpectedEof` IO errors. + /// + /// ``` + /// use std::io; + /// + /// enum MyError { + /// Io(io::Error), + /// Json(serde_json::Error), + /// } + /// + /// impl From<serde_json::Error> for MyError { + /// fn from(err: serde_json::Error) -> MyError { + /// use serde_json::error::Category; + /// match err.classify() { + /// Category::Io => { + /// MyError::Io(err.into()) + /// } + /// Category::Syntax | Category::Data | Category::Eof => { + /// MyError::Json(err) + /// } + /// } + /// } + /// } + /// ``` + fn from(j: Error) -> Self { + if let ErrorCode::Io(err) = j.err.code { + err + } else { + match j.classify() { + Category::Io => unreachable!(), + Category::Syntax | Category::Data => io::Error::new(io::ErrorKind::InvalidData, j), + Category::Eof => io::Error::new(io::ErrorKind::UnexpectedEof, j), + } + } + } +} + +struct ErrorImpl { + code: ErrorCode, + line: usize, + column: usize, +} + +pub(crate) enum ErrorCode { + /// Catchall for syntax error messages + Message(Box<str>), + + /// Some IO error occurred while serializing or deserializing. + Io(io::Error), + + /// EOF while parsing a list. + EofWhileParsingList, + + /// EOF while parsing an object. + EofWhileParsingObject, + + /// EOF while parsing a string. + EofWhileParsingString, + + /// EOF while parsing a JSON value. + EofWhileParsingValue, + + /// Expected this character to be a `':'`. + ExpectedColon, + + /// Expected this character to be either a `','` or a `']'`. + ExpectedListCommaOrEnd, + + /// Expected this character to be either a `','` or a `'}'`. + ExpectedObjectCommaOrEnd, + + /// Expected to parse either a `true`, `false`, or a `null`. + ExpectedSomeIdent, + + /// Expected this character to start a JSON value. + ExpectedSomeValue, + + /// Invalid hex escape code. + InvalidEscape, + + /// Invalid number. + InvalidNumber, + + /// Number is bigger than the maximum value of its type. + NumberOutOfRange, + + /// Invalid unicode code point. + InvalidUnicodeCodePoint, + + /// Control character found while parsing a string. + ControlCharacterWhileParsingString, + + /// Object key is not a string. + KeyMustBeAString, + + /// Lone leading surrogate in hex escape. + LoneLeadingSurrogateInHexEscape, + + /// JSON has a comma after the last value in an array or map. + TrailingComma, + + /// JSON has non-whitespace trailing characters after the value. + TrailingCharacters, + + /// Unexpected end of hex escape. + UnexpectedEndOfHexEscape, + + /// Encountered nesting of JSON maps and arrays more than 128 layers deep. + RecursionLimitExceeded, +} + +impl Error { + #[cold] + pub(crate) fn syntax(code: ErrorCode, line: usize, column: usize) -> Self { + Error { + err: Box::new(ErrorImpl { code, line, column }), + } + } + + // Not public API. Should be pub(crate). + // + // Update `eager_json` crate when this function changes. + #[doc(hidden)] + #[cold] + pub fn io(error: io::Error) -> Self { + Error { + err: Box::new(ErrorImpl { + code: ErrorCode::Io(error), + line: 0, + column: 0, + }), + } + } + + #[cold] + pub(crate) fn fix_position<F>(self, f: F) -> Self + where + F: FnOnce(ErrorCode) -> Error, + { + if self.err.line == 0 { + f(self.err.code) + } else { + self + } + } +} + +impl Display for ErrorCode { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + match self { + ErrorCode::Message(msg) => f.write_str(msg), + ErrorCode::Io(err) => Display::fmt(err, f), + ErrorCode::EofWhileParsingList => f.write_str("EOF while parsing a list"), + ErrorCode::EofWhileParsingObject => f.write_str("EOF while parsing an object"), + ErrorCode::EofWhileParsingString => f.write_str("EOF while parsing a string"), + ErrorCode::EofWhileParsingValue => f.write_str("EOF while parsing a value"), + ErrorCode::ExpectedColon => f.write_str("expected `:`"), + ErrorCode::ExpectedListCommaOrEnd => f.write_str("expected `,` or `]`"), + ErrorCode::ExpectedObjectCommaOrEnd => f.write_str("expected `,` or `}`"), + ErrorCode::ExpectedSomeIdent => f.write_str("expected ident"), + ErrorCode::ExpectedSomeValue => f.write_str("expected value"), + ErrorCode::InvalidEscape => f.write_str("invalid escape"), + ErrorCode::InvalidNumber => f.write_str("invalid number"), + ErrorCode::NumberOutOfRange => f.write_str("number out of range"), + ErrorCode::InvalidUnicodeCodePoint => f.write_str("invalid unicode code point"), + ErrorCode::ControlCharacterWhileParsingString => { + f.write_str("control character (\\u0000-\\u001F) found while parsing a string") + } + ErrorCode::KeyMustBeAString => f.write_str("key must be a string"), + ErrorCode::LoneLeadingSurrogateInHexEscape => { + f.write_str("lone leading surrogate in hex escape") + } + ErrorCode::TrailingComma => f.write_str("trailing comma"), + ErrorCode::TrailingCharacters => f.write_str("trailing characters"), + ErrorCode::UnexpectedEndOfHexEscape => f.write_str("unexpected end of hex escape"), + ErrorCode::RecursionLimitExceeded => f.write_str("recursion limit exceeded"), + } + } +} + +impl serde::de::StdError for Error { + #[cfg(feature = "std")] + fn source(&self) -> Option<&(dyn error::Error + 'static)> { + match &self.err.code { + ErrorCode::Io(err) => Some(err), + _ => None, + } + } +} + +impl Display for Error { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + Display::fmt(&*self.err, f) + } +} + +impl Display for ErrorImpl { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + if self.line == 0 { + Display::fmt(&self.code, f) + } else { + write!( + f, + "{} at line {} column {}", + self.code, self.line, self.column + ) + } + } +} + +// Remove two layers of verbosity from the debug representation. Humans often +// end up seeing this representation because it is what unwrap() shows. +impl Debug for Error { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + write!( + f, + "Error({:?}, line: {}, column: {})", + self.err.code.to_string(), + self.err.line, + self.err.column + ) + } +} + +impl de::Error for Error { + #[cold] + fn custom<T: Display>(msg: T) -> Error { + make_error(msg.to_string()) + } + + #[cold] + fn invalid_type(unexp: de::Unexpected, exp: &dyn de::Expected) -> Self { + if let de::Unexpected::Unit = unexp { + Error::custom(format_args!("invalid type: null, expected {}", exp)) + } else { + Error::custom(format_args!("invalid type: {}, expected {}", unexp, exp)) + } + } +} + +impl ser::Error for Error { + #[cold] + fn custom<T: Display>(msg: T) -> Error { + make_error(msg.to_string()) + } +} + +// Parse our own error message that looks like "{} at line {} column {}" to work +// around erased-serde round-tripping the error through de::Error::custom. +fn make_error(mut msg: String) -> Error { + let (line, column) = parse_line_col(&mut msg).unwrap_or((0, 0)); + Error { + err: Box::new(ErrorImpl { + code: ErrorCode::Message(msg.into_boxed_str()), + line, + column, + }), + } +} + +fn parse_line_col(msg: &mut String) -> Option<(usize, usize)> { + let start_of_suffix = match msg.rfind(" at line ") { + Some(index) => index, + None => return None, + }; + + // Find start and end of line number. + let start_of_line = start_of_suffix + " at line ".len(); + let mut end_of_line = start_of_line; + while starts_with_digit(&msg[end_of_line..]) { + end_of_line += 1; + } + + if !msg[end_of_line..].starts_with(" column ") { + return None; + } + + // Find start and end of column number. + let start_of_column = end_of_line + " column ".len(); + let mut end_of_column = start_of_column; + while starts_with_digit(&msg[end_of_column..]) { + end_of_column += 1; + } + + if end_of_column < msg.len() { + return None; + } + + // Parse numbers. + let line = match usize::from_str(&msg[start_of_line..end_of_line]) { + Ok(line) => line, + Err(_) => return None, + }; + let column = match usize::from_str(&msg[start_of_column..end_of_column]) { + Ok(column) => column, + Err(_) => return None, + }; + + msg.truncate(start_of_suffix); + Some((line, column)) +} + +fn starts_with_digit(slice: &str) -> bool { + match slice.as_bytes().first() { + None => false, + Some(&byte) => byte >= b'0' && byte <= b'9', + } +} diff --git a/third_party/rust/serde_json/src/features_check/error.rs b/third_party/rust/serde_json/src/features_check/error.rs new file mode 100644 index 0000000000..22e58235c7 --- /dev/null +++ b/third_party/rust/serde_json/src/features_check/error.rs @@ -0,0 +1 @@ +"serde_json requires that either `std` (default) or `alloc` feature is enabled" diff --git a/third_party/rust/serde_json/src/features_check/mod.rs b/third_party/rust/serde_json/src/features_check/mod.rs new file mode 100644 index 0000000000..d12032cef8 --- /dev/null +++ b/third_party/rust/serde_json/src/features_check/mod.rs @@ -0,0 +1,13 @@ +//! Shows a user-friendly compiler error on incompatible selected features. + +#[allow(unused_macros)] +macro_rules! hide_from_rustfmt { + ($mod:item) => { + $mod + }; +} + +#[cfg(not(any(feature = "std", feature = "alloc")))] +hide_from_rustfmt! { + mod error; +} diff --git a/third_party/rust/serde_json/src/io/core.rs b/third_party/rust/serde_json/src/io/core.rs new file mode 100644 index 0000000000..465ab8b249 --- /dev/null +++ b/third_party/rust/serde_json/src/io/core.rs @@ -0,0 +1,79 @@ +//! Reimplements core logic and types from `std::io` in an `alloc`-friendly +//! fashion. + +use alloc::vec::Vec; +use core::fmt::{self, Display}; +use core::result; + +pub enum ErrorKind { + Other, +} + +// IO errors can never occur in no-std mode. All our no-std IO implementations +// are infallible. +pub struct Error; + +impl Display for Error { + fn fmt(&self, _formatter: &mut fmt::Formatter<'_>) -> fmt::Result { + unreachable!() + } +} + +impl Error { + pub(crate) fn new(_kind: ErrorKind, _error: &'static str) -> Error { + Error + } +} + +pub type Result<T> = result::Result<T, Error>; + +pub trait Write { + fn write(&mut self, buf: &[u8]) -> Result<usize>; + + fn write_all(&mut self, buf: &[u8]) -> Result<()> { + // All our Write impls in no_std mode always write the whole buffer in + // one call infallibly. + let result = self.write(buf); + debug_assert!(result.is_ok()); + debug_assert_eq!(result.unwrap_or(0), buf.len()); + Ok(()) + } + + fn flush(&mut self) -> Result<()>; +} + +impl<W: Write> Write for &mut W { + #[inline] + fn write(&mut self, buf: &[u8]) -> Result<usize> { + (*self).write(buf) + } + + #[inline] + fn write_all(&mut self, buf: &[u8]) -> Result<()> { + (*self).write_all(buf) + } + + #[inline] + fn flush(&mut self) -> Result<()> { + (*self).flush() + } +} + +impl Write for Vec<u8> { + #[inline] + fn write(&mut self, buf: &[u8]) -> Result<usize> { + self.extend_from_slice(buf); + Ok(buf.len()) + } + + #[inline] + fn write_all(&mut self, buf: &[u8]) -> Result<()> { + self.extend_from_slice(buf); + Ok(()) + } + + #[inline] + fn flush(&mut self) -> Result<()> { + Ok(()) + } +} diff --git a/third_party/rust/serde_json/src/io/mod.rs b/third_party/rust/serde_json/src/io/mod.rs new file mode 100644 index 0000000000..9dee4a0685 --- /dev/null +++ b/third_party/rust/serde_json/src/io/mod.rs @@ -0,0 +1,20 @@ +//! A tiny, `no_std`-friendly facade around `std::io`. +//! Reexports types from `std` when available; otherwise reimplements and +//! provides some of the core logic. +//! +//! The main reason that `std::io` hasn't found itself reexported as part of +//! the `core` crate is the `std::io::{Read, Write}` traits' reliance on +//! `std::io::Error`, which may contain internally a heap-allocated `Box<Error>` +//! and/or now relying on OS-specific `std::backtrace::Backtrace`. + +pub use self::imp::{Error, ErrorKind, Result, Write}; + +#[cfg(not(feature = "std"))] +#[path = "core.rs"] +mod imp; + +#[cfg(feature = "std")] +use std::io as imp; + +#[cfg(feature = "std")] +pub use std::io::{Bytes, Read}; diff --git a/third_party/rust/serde_json/src/iter.rs b/third_party/rust/serde_json/src/iter.rs new file mode 100644 index 0000000000..9792916dcc --- /dev/null +++ b/third_party/rust/serde_json/src/iter.rs @@ -0,0 +1,70 @@ +use crate::io; + +pub struct LineColIterator<I> { + iter: I, + + /// Index of the current line. Characters in the first line of the input + /// (before the first newline character) are in line 1. + line: usize, + + /// Index of the current column. The first character in the input and any + /// characters immediately following a newline character are in column 1. + /// The column is 0 immediately after a newline character has been read. + col: usize, + + /// Byte offset of the start of the current line. This is the sum of lengths + /// of all previous lines. Keeping track of things this way allows efficient + /// computation of the current line, column, and byte offset while only + /// updating one of the counters in `next()` in the common case. + start_of_line: usize, +} + +impl<I> LineColIterator<I> +where + I: Iterator<Item = io::Result<u8>>, +{ + pub fn new(iter: I) -> LineColIterator<I> { + LineColIterator { + iter, + line: 1, + col: 0, + start_of_line: 0, + } + } + + pub fn line(&self) -> usize { + self.line + } + + pub fn col(&self) -> usize { + self.col + } + + pub fn byte_offset(&self) -> usize { + self.start_of_line + self.col + } +} + +impl<I> Iterator for LineColIterator<I> +where + I: Iterator<Item = io::Result<u8>>, +{ + type Item = io::Result<u8>; + + fn next(&mut self) -> Option<io::Result<u8>> { + match self.iter.next() { + None => None, + Some(Ok(b'\n')) => { + self.start_of_line += self.col + 1; + self.line += 1; + self.col = 0; + Some(Ok(b'\n')) + } + Some(Ok(c)) => { + self.col += 1; + Some(Ok(c)) + } + Some(Err(e)) => Some(Err(e)), + } + } +} diff --git a/third_party/rust/serde_json/src/lexical/algorithm.rs b/third_party/rust/serde_json/src/lexical/algorithm.rs new file mode 100644 index 0000000000..a2cbf18aff --- /dev/null +++ b/third_party/rust/serde_json/src/lexical/algorithm.rs @@ -0,0 +1,193 @@ +// Adapted from https://github.com/Alexhuszagh/rust-lexical. + +//! Algorithms to efficiently convert strings to floats. + +use super::bhcomp::*; +use super::cached::*; +use super::errors::*; +use super::float::ExtendedFloat; +use super::num::*; +use super::small_powers::*; + +// FAST +// ---- + +/// Convert mantissa to exact value for a non-base2 power. +/// +/// Returns the resulting float and if the value can be represented exactly. +pub(crate) fn fast_path<F>(mantissa: u64, exponent: i32) -> Option<F> +where + F: Float, +{ + // `mantissa >> (F::MANTISSA_SIZE+1) != 0` effectively checks if the + // value has a no bits above the hidden bit, which is what we want. + let (min_exp, max_exp) = F::exponent_limit(); + let shift_exp = F::mantissa_limit(); + let mantissa_size = F::MANTISSA_SIZE + 1; + if mantissa == 0 { + Some(F::ZERO) + } else if mantissa >> mantissa_size != 0 { + // Would require truncation of the mantissa. + None + } else if exponent == 0 { + // 0 exponent, same as value, exact representation. + let float = F::as_cast(mantissa); + Some(float) + } else if exponent >= min_exp && exponent <= max_exp { + // Value can be exactly represented, return the value. + // Do not use powi, since powi can incrementally introduce + // error. + let float = F::as_cast(mantissa); + Some(float.pow10(exponent)) + } else if exponent >= 0 && exponent <= max_exp + shift_exp { + // Check to see if we have a disguised fast-path, where the + // number of digits in the mantissa is very small, but and + // so digits can be shifted from the exponent to the mantissa. + // https://www.exploringbinary.com/fast-path-decimal-to-floating-point-conversion/ + let small_powers = POW10_64; + let shift = exponent - max_exp; + let power = small_powers[shift as usize]; + + // Compute the product of the power, if it overflows, + // prematurely return early, otherwise, if we didn't overshoot, + // we can get an exact value. + let value = mantissa.checked_mul(power)?; + if value >> mantissa_size != 0 { + None + } else { + // Use powi, since it's correct, and faster on + // the fast-path. + let float = F::as_cast(value); + Some(float.pow10(max_exp)) + } + } else { + // Cannot be exactly represented, exponent too small or too big, + // would require truncation. + None + } +} + +// MODERATE +// -------- + +/// Multiply the floating-point by the exponent. +/// +/// Multiply by pre-calculated powers of the base, modify the extended- +/// float, and return if new value and if the value can be represented +/// accurately. +fn multiply_exponent_extended<F>(fp: &mut ExtendedFloat, exponent: i32, truncated: bool) -> bool +where + F: Float, +{ + let powers = ExtendedFloat::get_powers(); + let exponent = exponent.saturating_add(powers.bias); + let small_index = exponent % powers.step; + let large_index = exponent / powers.step; + if exponent < 0 { + // Guaranteed underflow (assign 0). + fp.mant = 0; + true + } else if large_index as usize >= powers.large.len() { + // Overflow (assign infinity) + fp.mant = 1 << 63; + fp.exp = 0x7FF; + true + } else { + // Within the valid exponent range, multiply by the large and small + // exponents and return the resulting value. + + // Track errors to as a factor of unit in last-precision. + let mut errors: u32 = 0; + if truncated { + errors += u64::error_halfscale(); + } + + // Multiply by the small power. + // Check if we can directly multiply by an integer, if not, + // use extended-precision multiplication. + match fp + .mant + .overflowing_mul(powers.get_small_int(small_index as usize)) + { + // Overflow, multiplication unsuccessful, go slow path. + (_, true) => { + fp.normalize(); + fp.imul(&powers.get_small(small_index as usize)); + errors += u64::error_halfscale(); + } + // No overflow, multiplication successful. + (mant, false) => { + fp.mant = mant; + fp.normalize(); + } + } + + // Multiply by the large power + fp.imul(&powers.get_large(large_index as usize)); + if errors > 0 { + errors += 1; + } + errors += u64::error_halfscale(); + + // Normalize the floating point (and the errors). + let shift = fp.normalize(); + errors <<= shift; + + u64::error_is_accurate::<F>(errors, fp) + } +} + +/// Create a precise native float using an intermediate extended-precision float. +/// +/// Return the float approximation and if the value can be accurately +/// represented with mantissa bits of precision. +#[inline] +pub(crate) fn moderate_path<F>( + mantissa: u64, + exponent: i32, + truncated: bool, +) -> (ExtendedFloat, bool) +where + F: Float, +{ + let mut fp = ExtendedFloat { + mant: mantissa, + exp: 0, + }; + let valid = multiply_exponent_extended::<F>(&mut fp, exponent, truncated); + (fp, valid) +} + +// FALLBACK +// -------- + +/// Fallback path when the fast path does not work. +/// +/// Uses the moderate path, if applicable, otherwise, uses the slow path +/// as required. +pub(crate) fn fallback_path<F>( + integer: &[u8], + fraction: &[u8], + mantissa: u64, + exponent: i32, + mantissa_exponent: i32, + truncated: bool, +) -> F +where + F: Float, +{ + // Moderate path (use an extended 80-bit representation). + let (fp, valid) = moderate_path::<F>(mantissa, mantissa_exponent, truncated); + if valid { + return fp.into_float::<F>(); + } + + // Slow path, fast path didn't work. + let b = fp.into_downward_float::<F>(); + if b.is_special() { + // We have a non-finite number, we get to leave early. + b + } else { + bhcomp(b, integer, fraction, exponent) + } +} diff --git a/third_party/rust/serde_json/src/lexical/bhcomp.rs b/third_party/rust/serde_json/src/lexical/bhcomp.rs new file mode 100644 index 0000000000..1f2a7bbdee --- /dev/null +++ b/third_party/rust/serde_json/src/lexical/bhcomp.rs @@ -0,0 +1,218 @@ +// Adapted from https://github.com/Alexhuszagh/rust-lexical. + +//! Compare the mantissa to the halfway representation of the float. +//! +//! Compares the actual significant digits of the mantissa to the +//! theoretical digits from `b+h`, scaled into the proper range. + +use super::bignum::*; +use super::digit::*; +use super::exponent::*; +use super::float::*; +use super::math::*; +use super::num::*; +use super::rounding::*; +use core::{cmp, mem}; + +// MANTISSA + +/// Parse the full mantissa into a big integer. +/// +/// Max digits is the maximum number of digits plus one. +fn parse_mantissa<F>(integer: &[u8], fraction: &[u8]) -> Bigint +where + F: Float, +{ + // Main loop + let small_powers = POW10_LIMB; + let step = small_powers.len() - 2; + let max_digits = F::MAX_DIGITS - 1; + let mut counter = 0; + let mut value: Limb = 0; + let mut i: usize = 0; + let mut result = Bigint::default(); + + // Iteratively process all the data in the mantissa. + for &digit in integer.iter().chain(fraction) { + // We've parsed the max digits using small values, add to bignum + if counter == step { + result.imul_small(small_powers[counter]); + result.iadd_small(value); + counter = 0; + value = 0; + } + + value *= 10; + value += as_limb(to_digit(digit).unwrap()); + + i += 1; + counter += 1; + if i == max_digits { + break; + } + } + + // We will always have a remainder, as long as we entered the loop + // once, or counter % step is 0. + if counter != 0 { + result.imul_small(small_powers[counter]); + result.iadd_small(value); + } + + // If we have any remaining digits after the last value, we need + // to add a 1 after the rest of the array, it doesn't matter where, + // just move it up. This is good for the worst-possible float + // representation. We also need to return an index. + // Since we already trimmed trailing zeros, we know there has + // to be a non-zero digit if there are any left. + if i < integer.len() + fraction.len() { + result.imul_small(10); + result.iadd_small(1); + } + + result +} + +// FLOAT OPS + +/// Calculate `b` from a a representation of `b` as a float. +#[inline] +pub(super) fn b_extended<F: Float>(f: F) -> ExtendedFloat { + ExtendedFloat::from_float(f) +} + +/// Calculate `b+h` from a a representation of `b` as a float. +#[inline] +pub(super) fn bh_extended<F: Float>(f: F) -> ExtendedFloat { + // None of these can overflow. + let b = b_extended(f); + ExtendedFloat { + mant: (b.mant << 1) + 1, + exp: b.exp - 1, + } +} + +// ROUNDING + +/// Custom round-nearest, tie-event algorithm for bhcomp. +#[inline] +fn round_nearest_tie_even(fp: &mut ExtendedFloat, shift: i32, is_truncated: bool) { + let (mut is_above, mut is_halfway) = round_nearest(fp, shift); + if is_halfway && is_truncated { + is_above = true; + is_halfway = false; + } + tie_even(fp, is_above, is_halfway); +} + +// BHCOMP + +/// Calculate the mantissa for a big integer with a positive exponent. +fn large_atof<F>(mantissa: Bigint, exponent: i32) -> F +where + F: Float, +{ + let bits = mem::size_of::<u64>() * 8; + + // Simple, we just need to multiply by the power of the radix. + // Now, we can calculate the mantissa and the exponent from this. + // The binary exponent is the binary exponent for the mantissa + // shifted to the hidden bit. + let mut bigmant = mantissa; + bigmant.imul_pow10(exponent as u32); + + // Get the exact representation of the float from the big integer. + let (mant, is_truncated) = bigmant.hi64(); + let exp = bigmant.bit_length() as i32 - bits as i32; + let mut fp = ExtendedFloat { mant, exp }; + fp.round_to_native::<F, _>(|fp, shift| round_nearest_tie_even(fp, shift, is_truncated)); + into_float(fp) +} + +/// Calculate the mantissa for a big integer with a negative exponent. +/// +/// This invokes the comparison with `b+h`. +fn small_atof<F>(mantissa: Bigint, exponent: i32, f: F) -> F +where + F: Float, +{ + // Get the significant digits and radix exponent for the real digits. + let mut real_digits = mantissa; + let real_exp = exponent; + debug_assert!(real_exp < 0); + + // Get the significant digits and the binary exponent for `b+h`. + let theor = bh_extended(f); + let mut theor_digits = Bigint::from_u64(theor.mant); + let theor_exp = theor.exp; + + // We need to scale the real digits and `b+h` digits to be the same + // order. We currently have `real_exp`, in `radix`, that needs to be + // shifted to `theor_digits` (since it is negative), and `theor_exp` + // to either `theor_digits` or `real_digits` as a power of 2 (since it + // may be positive or negative). Try to remove as many powers of 2 + // as possible. All values are relative to `theor_digits`, that is, + // reflect the power you need to multiply `theor_digits` by. + + // Can remove a power-of-two, since the radix is 10. + // Both are on opposite-sides of equation, can factor out a + // power of two. + // + // Example: 10^-10, 2^-10 -> ( 0, 10, 0) + // Example: 10^-10, 2^-15 -> (-5, 10, 0) + // Example: 10^-10, 2^-5 -> ( 5, 10, 0) + // Example: 10^-10, 2^5 -> (15, 10, 0) + let binary_exp = theor_exp - real_exp; + let halfradix_exp = -real_exp; + let radix_exp = 0; + + // Carry out our multiplication. + if halfradix_exp != 0 { + theor_digits.imul_pow5(halfradix_exp as u32); + } + if radix_exp != 0 { + theor_digits.imul_pow10(radix_exp as u32); + } + if binary_exp > 0 { + theor_digits.imul_pow2(binary_exp as u32); + } else if binary_exp < 0 { + real_digits.imul_pow2(-binary_exp as u32); + } + + // Compare real digits to theoretical digits and round the float. + match real_digits.compare(&theor_digits) { + cmp::Ordering::Greater => f.next_positive(), + cmp::Ordering::Less => f, + cmp::Ordering::Equal => f.round_positive_even(), + } +} + +/// Calculate the exact value of the float. +/// +/// Note: fraction must not have trailing zeros. +pub(crate) fn bhcomp<F>(b: F, integer: &[u8], mut fraction: &[u8], exponent: i32) -> F +where + F: Float, +{ + // Calculate the number of integer digits and use that to determine + // where the significant digits start in the fraction. + let integer_digits = integer.len(); + let fraction_digits = fraction.len(); + let digits_start = if integer_digits == 0 { + let start = fraction.iter().take_while(|&x| *x == b'0').count(); + fraction = &fraction[start..]; + start + } else { + 0 + }; + let sci_exp = scientific_exponent(exponent, integer_digits, digits_start); + let count = F::MAX_DIGITS.min(integer_digits + fraction_digits - digits_start); + let scaled_exponent = sci_exp + 1 - count as i32; + + let mantissa = parse_mantissa::<F>(integer, fraction); + if scaled_exponent >= 0 { + large_atof(mantissa, scaled_exponent) + } else { + small_atof(mantissa, scaled_exponent, b) + } +} diff --git a/third_party/rust/serde_json/src/lexical/bignum.rs b/third_party/rust/serde_json/src/lexical/bignum.rs new file mode 100644 index 0000000000..f9551f534f --- /dev/null +++ b/third_party/rust/serde_json/src/lexical/bignum.rs @@ -0,0 +1,33 @@ +// Adapted from https://github.com/Alexhuszagh/rust-lexical. + +//! Big integer type definition. + +use super::math::*; +use alloc::vec::Vec; + +/// Storage for a big integer type. +#[derive(Clone, PartialEq, Eq)] +pub(crate) struct Bigint { + /// Internal storage for the Bigint, in little-endian order. + pub(crate) data: Vec<Limb>, +} + +impl Default for Bigint { + fn default() -> Self { + Bigint { + data: Vec::with_capacity(20), + } + } +} + +impl Math for Bigint { + #[inline] + fn data(&self) -> &Vec<Limb> { + &self.data + } + + #[inline] + fn data_mut(&mut self) -> &mut Vec<Limb> { + &mut self.data + } +} diff --git a/third_party/rust/serde_json/src/lexical/cached.rs b/third_party/rust/serde_json/src/lexical/cached.rs new file mode 100644 index 0000000000..ef5a9fe54d --- /dev/null +++ b/third_party/rust/serde_json/src/lexical/cached.rs @@ -0,0 +1,82 @@ +// Adapted from https://github.com/Alexhuszagh/rust-lexical. + +//! Cached powers trait for extended-precision floats. + +use super::cached_float80; +use super::float::ExtendedFloat; + +// POWERS + +/// Precalculated powers that uses two-separate arrays for memory-efficiency. +#[doc(hidden)] +pub(crate) struct ExtendedFloatArray { + // Pre-calculated mantissa for the powers. + pub mant: &'static [u64], + // Pre-calculated binary exponents for the powers. + pub exp: &'static [i32], +} + +/// Allow indexing of values without bounds checking +impl ExtendedFloatArray { + #[inline] + pub fn get_extended_float(&self, index: usize) -> ExtendedFloat { + let mant = self.mant[index]; + let exp = self.exp[index]; + ExtendedFloat { mant, exp } + } + + #[inline] + pub fn len(&self) -> usize { + self.mant.len() + } +} + +// MODERATE PATH POWERS + +/// Precalculated powers of base N for the moderate path. +#[doc(hidden)] +pub(crate) struct ModeratePathPowers { + // Pre-calculated small powers. + pub small: ExtendedFloatArray, + // Pre-calculated large powers. + pub large: ExtendedFloatArray, + /// Pre-calculated small powers as 64-bit integers + pub small_int: &'static [u64], + // Step between large powers and number of small powers. + pub step: i32, + // Exponent bias for the large powers. + pub bias: i32, +} + +/// Allow indexing of values without bounds checking +impl ModeratePathPowers { + #[inline] + pub fn get_small(&self, index: usize) -> ExtendedFloat { + self.small.get_extended_float(index) + } + + #[inline] + pub fn get_large(&self, index: usize) -> ExtendedFloat { + self.large.get_extended_float(index) + } + + #[inline] + pub fn get_small_int(&self, index: usize) -> u64 { + self.small_int[index] + } +} + +// CACHED EXTENDED POWERS + +/// Cached powers as a trait for a floating-point type. +pub(crate) trait ModeratePathCache { + /// Get cached powers. + fn get_powers() -> &'static ModeratePathPowers; +} + +impl ModeratePathCache for ExtendedFloat { + #[inline] + fn get_powers() -> &'static ModeratePathPowers { + cached_float80::get_powers() + } +} diff --git a/third_party/rust/serde_json/src/lexical/cached_float80.rs b/third_party/rust/serde_json/src/lexical/cached_float80.rs new file mode 100644 index 0000000000..9beda3ddb1 --- /dev/null +++ b/third_party/rust/serde_json/src/lexical/cached_float80.rs @@ -0,0 +1,206 @@ +// Adapted from https://github.com/Alexhuszagh/rust-lexical. + +//! Cached exponents for basen values with 80-bit extended floats. +//! +//! Exact versions of base**n as an extended-precision float, with both +//! large and small powers. Use the large powers to minimize the amount +//! of compounded error. +//! +//! These values were calculated using Python, using the arbitrary-precision +//! integer to calculate exact extended-representation of each value. +//! These values are all normalized. + +use super::cached::{ExtendedFloatArray, ModeratePathPowers}; + +// LOW-LEVEL +// --------- + +// BASE10 + +const BASE10_SMALL_MANTISSA: [u64; 10] = [ + 9223372036854775808, // 10^0 + 11529215046068469760, // 10^1 + 14411518807585587200, // 10^2 + 18014398509481984000, // 10^3 + 11258999068426240000, // 10^4 + 14073748835532800000, // 10^5 + 17592186044416000000, // 10^6 + 10995116277760000000, // 10^7 + 13743895347200000000, // 10^8 + 17179869184000000000, // 10^9 +]; +const BASE10_SMALL_EXPONENT: [i32; 10] = [ + -63, // 10^0 + -60, // 10^1 + -57, // 10^2 + -54, // 10^3 + -50, // 10^4 + -47, // 10^5 + -44, // 10^6 + -40, // 10^7 + -37, // 10^8 + -34, // 10^9 +]; +const BASE10_LARGE_MANTISSA: [u64; 66] = [ + 11555125961253852697, // 10^-350 + 13451937075301367670, // 10^-340 + 15660115838168849784, // 10^-330 + 18230774251475056848, // 10^-320 + 10611707258198326947, // 10^-310 + 12353653155963782858, // 10^-300 + 14381545078898527261, // 10^-290 + 16742321987285426889, // 10^-280 + 9745314011399999080, // 10^-270 + 11345038669416679861, // 10^-260 + 13207363278391631158, // 10^-250 + 15375394465392026070, // 10^-240 + 17899314949046850752, // 10^-230 + 10418772551374772303, // 10^-220 + 12129047596099288555, // 10^-210 + 14120069793541087484, // 10^-200 + 16437924692338667210, // 10^-190 + 9568131466127621947, // 10^-180 + 11138771039116687545, // 10^-170 + 12967236152753102995, // 10^-160 + 15095849699286165408, // 10^-150 + 17573882009934360870, // 10^-140 + 10229345649675443343, // 10^-130 + 11908525658859223294, // 10^-120 + 13863348470604074297, // 10^-110 + 16139061738043178685, // 10^-100 + 9394170331095332911, // 10^-90 + 10936253623915059621, // 10^-80 + 12731474852090538039, // 10^-70 + 14821387422376473014, // 10^-60 + 17254365866976409468, // 10^-50 + 10043362776618689222, // 10^-40 + 11692013098647223345, // 10^-30 + 13611294676837538538, // 10^-20 + 15845632502852867518, // 10^-10 + 9223372036854775808, // 10^0 + 10737418240000000000, // 10^10 + 12500000000000000000, // 10^20 + 14551915228366851806, // 10^30 + 16940658945086006781, // 10^40 + 9860761315262647567, // 10^50 + 11479437019748901445, // 10^60 + 13363823550460978230, // 10^70 + 15557538194652854267, // 10^80 + 18111358157653424735, // 10^90 + 10542197943230523224, // 10^100 + 12272733663244316382, // 10^110 + 14287342391028437277, // 10^120 + 16632655625031838749, // 10^130 + 9681479787123295682, // 10^140 + 11270725851789228247, // 10^150 + 13120851772591970218, // 10^160 + 15274681817498023410, // 10^170 + 17782069995880619867, // 10^180 + 10350527006597618960, // 10^190 + 12049599325514420588, // 10^200 + 14027579833653779454, // 10^210 + 16330252207878254650, // 10^220 + 9505457831475799117, // 10^230 + 11065809325636130661, // 10^240 + 12882297539194266616, // 10^250 + 14996968138956309548, // 10^260 + 17458768723248864463, // 10^270 + 10162340898095201970, // 10^280 + 11830521861667747109, // 10^290 + 13772540099066387756, // 10^300 +]; +const BASE10_LARGE_EXPONENT: [i32; 66] = [ + -1226, // 10^-350 + -1193, // 10^-340 + -1160, // 10^-330 + -1127, // 10^-320 + -1093, // 10^-310 + -1060, // 10^-300 + -1027, // 10^-290 + -994, // 10^-280 + -960, // 10^-270 + -927, // 10^-260 + -894, // 10^-250 + -861, // 10^-240 + -828, // 10^-230 + -794, // 10^-220 + -761, // 10^-210 + -728, // 10^-200 + -695, // 10^-190 + -661, // 10^-180 + -628, // 10^-170 + -595, // 10^-160 + -562, // 10^-150 + -529, // 10^-140 + -495, // 10^-130 + -462, // 10^-120 + -429, // 10^-110 + -396, // 10^-100 + -362, // 10^-90 + -329, // 10^-80 + -296, // 10^-70 + -263, // 10^-60 + -230, // 10^-50 + -196, // 10^-40 + -163, // 10^-30 + -130, // 10^-20 + -97, // 10^-10 + -63, // 10^0 + -30, // 10^10 + 3, // 10^20 + 36, // 10^30 + 69, // 10^40 + 103, // 10^50 + 136, // 10^60 + 169, // 10^70 + 202, // 10^80 + 235, // 10^90 + 269, // 10^100 + 302, // 10^110 + 335, // 10^120 + 368, // 10^130 + 402, // 10^140 + 435, // 10^150 + 468, // 10^160 + 501, // 10^170 + 534, // 10^180 + 568, // 10^190 + 601, // 10^200 + 634, // 10^210 + 667, // 10^220 + 701, // 10^230 + 734, // 10^240 + 767, // 10^250 + 800, // 10^260 + 833, // 10^270 + 867, // 10^280 + 900, // 10^290 + 933, // 10^300 +]; +const BASE10_SMALL_INT_POWERS: [u64; 10] = [ + 1, 10, 100, 1000, 10000, 100000, 1000000, 10000000, 100000000, 1000000000, +]; +const BASE10_STEP: i32 = 10; +const BASE10_BIAS: i32 = 350; + +// HIGH LEVEL +// ---------- + +const BASE10_POWERS: ModeratePathPowers = ModeratePathPowers { + small: ExtendedFloatArray { + mant: &BASE10_SMALL_MANTISSA, + exp: &BASE10_SMALL_EXPONENT, + }, + large: ExtendedFloatArray { + mant: &BASE10_LARGE_MANTISSA, + exp: &BASE10_LARGE_EXPONENT, + }, + small_int: &BASE10_SMALL_INT_POWERS, + step: BASE10_STEP, + bias: BASE10_BIAS, +}; + +/// Get powers from base. +pub(crate) fn get_powers() -> &'static ModeratePathPowers { + &BASE10_POWERS +} diff --git a/third_party/rust/serde_json/src/lexical/digit.rs b/third_party/rust/serde_json/src/lexical/digit.rs new file mode 100644 index 0000000000..882aa9eef2 --- /dev/null +++ b/third_party/rust/serde_json/src/lexical/digit.rs @@ -0,0 +1,15 @@ +// Adapted from https://github.com/Alexhuszagh/rust-lexical. + +//! Helpers to convert and add digits from characters. + +// Convert u8 to digit. +#[inline] +pub(crate) fn to_digit(c: u8) -> Option<u32> { + (c as char).to_digit(10) +} + +// Add digit to mantissa. +#[inline] +pub(crate) fn add_digit(value: u64, digit: u32) -> Option<u64> { + value.checked_mul(10)?.checked_add(digit as u64) +} diff --git a/third_party/rust/serde_json/src/lexical/errors.rs b/third_party/rust/serde_json/src/lexical/errors.rs new file mode 100644 index 0000000000..cad4bd3d58 --- /dev/null +++ b/third_party/rust/serde_json/src/lexical/errors.rs @@ -0,0 +1,133 @@ +// Adapted from https://github.com/Alexhuszagh/rust-lexical. + +//! Estimate the error in an 80-bit approximation of a float. +//! +//! This estimates the error in a floating-point representation. +//! +//! This implementation is loosely based off the Golang implementation, +//! found here: +//! https://golang.org/src/strconv/atof.go + +use super::float::*; +use super::num::*; +use super::rounding::*; + +pub(crate) trait FloatErrors { + /// Get the full error scale. + fn error_scale() -> u32; + /// Get the half error scale. + fn error_halfscale() -> u32; + /// Determine if the number of errors is tolerable for float precision. + fn error_is_accurate<F: Float>(count: u32, fp: &ExtendedFloat) -> bool; +} + +/// Check if the error is accurate with a round-nearest rounding scheme. +#[inline] +fn nearest_error_is_accurate(errors: u64, fp: &ExtendedFloat, extrabits: u64) -> bool { + // Round-to-nearest, need to use the halfway point. + if extrabits == 65 { + // Underflow, we have a shift larger than the mantissa. + // Representation is valid **only** if the value is close enough + // overflow to the next bit within errors. If it overflows, + // the representation is **not** valid. + !fp.mant.overflowing_add(errors).1 + } else { + let mask: u64 = lower_n_mask(extrabits); + let extra: u64 = fp.mant & mask; + + // Round-to-nearest, need to check if we're close to halfway. + // IE, b10100 | 100000, where `|` signifies the truncation point. + let halfway: u64 = lower_n_halfway(extrabits); + let cmp1 = halfway.wrapping_sub(errors) < extra; + let cmp2 = extra < halfway.wrapping_add(errors); + + // If both comparisons are true, we have significant rounding error, + // and the value cannot be exactly represented. Otherwise, the + // representation is valid. + !(cmp1 && cmp2) + } +} + +impl FloatErrors for u64 { + #[inline] + fn error_scale() -> u32 { + 8 + } + + #[inline] + fn error_halfscale() -> u32 { + u64::error_scale() / 2 + } + + #[inline] + fn error_is_accurate<F: Float>(count: u32, fp: &ExtendedFloat) -> bool { + // Determine if extended-precision float is a good approximation. + // If the error has affected too many units, the float will be + // inaccurate, or if the representation is too close to halfway + // that any operations could affect this halfway representation. + // See the documentation for dtoa for more information. + let bias = -(F::EXPONENT_BIAS - F::MANTISSA_SIZE); + let denormal_exp = bias - 63; + // This is always a valid u32, since (denormal_exp - fp.exp) + // will always be positive and the significand size is {23, 52}. + let extrabits = if fp.exp <= denormal_exp { + 64 - F::MANTISSA_SIZE + denormal_exp - fp.exp + } else { + 63 - F::MANTISSA_SIZE + }; + + // Our logic is as follows: we want to determine if the actual + // mantissa and the errors during calculation differ significantly + // from the rounding point. The rounding point for round-nearest + // is the halfway point, IE, this when the truncated bits start + // with b1000..., while the rounding point for the round-toward + // is when the truncated bits are equal to 0. + // To do so, we can check whether the rounding point +/- the error + // are >/< the actual lower n bits. + // + // For whether we need to use signed or unsigned types for this + // analysis, see this example, using u8 rather than u64 to simplify + // things. + // + // # Comparisons + // cmp1 = (halfway - errors) < extra + // cmp1 = extra < (halfway + errors) + // + // # Large Extrabits, Low Errors + // + // extrabits = 8 + // halfway = 0b10000000 + // extra = 0b10000010 + // errors = 0b00000100 + // halfway - errors = 0b01111100 + // halfway + errors = 0b10000100 + // + // Unsigned: + // halfway - errors = 124 + // halfway + errors = 132 + // extra = 130 + // cmp1 = true + // cmp2 = true + // Signed: + // halfway - errors = 124 + // halfway + errors = -124 + // extra = -126 + // cmp1 = false + // cmp2 = true + // + // # Conclusion + // + // Since errors will always be small, and since we want to detect + // if the representation is accurate, we need to use an **unsigned** + // type for comparisons. + + let extrabits = extrabits as u64; + let errors = count as u64; + if extrabits > 65 { + // Underflow, we have a literal 0. + return true; + } + + nearest_error_is_accurate(errors, fp, extrabits) + } +} diff --git a/third_party/rust/serde_json/src/lexical/exponent.rs b/third_party/rust/serde_json/src/lexical/exponent.rs new file mode 100644 index 0000000000..6fc51977ed --- /dev/null +++ b/third_party/rust/serde_json/src/lexical/exponent.rs @@ -0,0 +1,50 @@ +// Adapted from https://github.com/Alexhuszagh/rust-lexical. + +//! Utilities to calculate exponents. + +/// Convert usize into i32 without overflow. +/// +/// This is needed to ensure when adjusting the exponent relative to +/// the mantissa we do not overflow for comically-long exponents. +#[inline] +fn into_i32(value: usize) -> i32 { + if value > i32::max_value() as usize { + i32::max_value() + } else { + value as i32 + } +} + +// EXPONENT CALCULATION + +// Calculate the scientific notation exponent without overflow. +// +// For example, 0.1 would be -1, and 10 would be 1 in base 10. +#[inline] +pub(crate) fn scientific_exponent( + exponent: i32, + integer_digits: usize, + fraction_start: usize, +) -> i32 { + if integer_digits == 0 { + let fraction_start = into_i32(fraction_start); + exponent.saturating_sub(fraction_start).saturating_sub(1) + } else { + let integer_shift = into_i32(integer_digits - 1); + exponent.saturating_add(integer_shift) + } +} + +// Calculate the mantissa exponent without overflow. +// +// Remove the number of digits that contributed to the mantissa past +// the dot, and add the number of truncated digits from the mantissa, +// to calculate the scaling factor for the mantissa from a raw exponent. +#[inline] +pub(crate) fn mantissa_exponent(exponent: i32, fraction_digits: usize, truncated: usize) -> i32 { + if fraction_digits > truncated { + exponent.saturating_sub(into_i32(fraction_digits - truncated)) + } else { + exponent.saturating_add(into_i32(truncated - fraction_digits)) + } +} diff --git a/third_party/rust/serde_json/src/lexical/float.rs b/third_party/rust/serde_json/src/lexical/float.rs new file mode 100644 index 0000000000..2d434a29cb --- /dev/null +++ b/third_party/rust/serde_json/src/lexical/float.rs @@ -0,0 +1,183 @@ +// Adapted from https://github.com/Alexhuszagh/rust-lexical. + +// FLOAT TYPE + +use super::num::*; +use super::rounding::*; +use super::shift::*; + +/// Extended precision floating-point type. +/// +/// Private implementation, exposed only for testing purposes. +#[doc(hidden)] +#[derive(Clone, Copy, Debug, PartialEq, Eq)] +pub(crate) struct ExtendedFloat { + /// Mantissa for the extended-precision float. + pub mant: u64, + /// Binary exponent for the extended-precision float. + pub exp: i32, +} + +impl ExtendedFloat { + // PROPERTIES + + // OPERATIONS + + /// Multiply two normalized extended-precision floats, as if by `a*b`. + /// + /// The precision is maximal when the numbers are normalized, however, + /// decent precision will occur as long as both values have high bits + /// set. The result is not normalized. + /// + /// Algorithm: + /// 1. Non-signed multiplication of mantissas (requires 2x as many bits as input). + /// 2. Normalization of the result (not done here). + /// 3. Addition of exponents. + pub(crate) fn mul(&self, b: &ExtendedFloat) -> ExtendedFloat { + // Logic check, values must be decently normalized prior to multiplication. + debug_assert!((self.mant & u64::HIMASK != 0) && (b.mant & u64::HIMASK != 0)); + + // Extract high-and-low masks. + let ah = self.mant >> u64::HALF; + let al = self.mant & u64::LOMASK; + let bh = b.mant >> u64::HALF; + let bl = b.mant & u64::LOMASK; + + // Get our products + let ah_bl = ah * bl; + let al_bh = al * bh; + let al_bl = al * bl; + let ah_bh = ah * bh; + + let mut tmp = (ah_bl & u64::LOMASK) + (al_bh & u64::LOMASK) + (al_bl >> u64::HALF); + // round up + tmp += 1 << (u64::HALF - 1); + + ExtendedFloat { + mant: ah_bh + (ah_bl >> u64::HALF) + (al_bh >> u64::HALF) + (tmp >> u64::HALF), + exp: self.exp + b.exp + u64::FULL, + } + } + + /// Multiply in-place, as if by `a*b`. + /// + /// The result is not normalized. + #[inline] + pub(crate) fn imul(&mut self, b: &ExtendedFloat) { + *self = self.mul(b); + } + + // NORMALIZE + + /// Normalize float-point number. + /// + /// Shift the mantissa so the number of leading zeros is 0, or the value + /// itself is 0. + /// + /// Get the number of bytes shifted. + #[inline] + pub(crate) fn normalize(&mut self) -> u32 { + // Note: + // Using the cltz intrinsic via leading_zeros is way faster (~10x) + // than shifting 1-bit at a time, via while loop, and also way + // faster (~2x) than an unrolled loop that checks at 32, 16, 4, + // 2, and 1 bit. + // + // Using a modulus of pow2 (which will get optimized to a bitwise + // and with 0x3F or faster) is slightly slower than an if/then, + // however, removing the if/then will likely optimize more branched + // code as it removes conditional logic. + + // Calculate the number of leading zeros, and then zero-out + // any overflowing bits, to avoid shl overflow when self.mant == 0. + let shift = if self.mant == 0 { + 0 + } else { + self.mant.leading_zeros() + }; + shl(self, shift as i32); + shift + } + + // ROUND + + /// Lossy round float-point number to native mantissa boundaries. + #[inline] + pub(crate) fn round_to_native<F, Algorithm>(&mut self, algorithm: Algorithm) + where + F: Float, + Algorithm: FnOnce(&mut ExtendedFloat, i32), + { + round_to_native::<F, _>(self, algorithm); + } + + // FROM + + /// Create extended float from native float. + #[inline] + pub fn from_float<F: Float>(f: F) -> ExtendedFloat { + from_float(f) + } + + // INTO + + /// Convert into default-rounded, lower-precision native float. + #[inline] + pub(crate) fn into_float<F: Float>(mut self) -> F { + self.round_to_native::<F, _>(round_nearest_tie_even); + into_float(self) + } + + /// Convert into downward-rounded, lower-precision native float. + #[inline] + pub(crate) fn into_downward_float<F: Float>(mut self) -> F { + self.round_to_native::<F, _>(round_downward); + into_float(self) + } +} + +// FROM FLOAT + +// Import ExtendedFloat from native float. +#[inline] +pub(crate) fn from_float<F>(f: F) -> ExtendedFloat +where + F: Float, +{ + ExtendedFloat { + mant: u64::as_cast(f.mantissa()), + exp: f.exponent(), + } +} + +// INTO FLOAT + +// Export extended-precision float to native float. +// +// The extended-precision float must be in native float representation, +// with overflow/underflow appropriately handled. +#[inline] +pub(crate) fn into_float<F>(fp: ExtendedFloat) -> F +where + F: Float, +{ + // Export floating-point number. + if fp.mant == 0 || fp.exp < F::DENORMAL_EXPONENT { + // sub-denormal, underflow + F::ZERO + } else if fp.exp >= F::MAX_EXPONENT { + // overflow + F::from_bits(F::INFINITY_BITS) + } else { + // calculate the exp and fraction bits, and return a float from bits. + let exp: u64; + if (fp.exp == F::DENORMAL_EXPONENT) && (fp.mant & F::HIDDEN_BIT_MASK.as_u64()) == 0 { + exp = 0; + } else { + exp = (fp.exp + F::EXPONENT_BIAS) as u64; + } + let exp = exp << F::MANTISSA_SIZE; + let mant = fp.mant & F::MANTISSA_MASK.as_u64(); + F::from_bits(F::Unsigned::as_cast(mant | exp)) + } +} diff --git a/third_party/rust/serde_json/src/lexical/large_powers.rs b/third_party/rust/serde_json/src/lexical/large_powers.rs new file mode 100644 index 0000000000..c63ce1cf21 --- /dev/null +++ b/third_party/rust/serde_json/src/lexical/large_powers.rs @@ -0,0 +1,9 @@ +// Adapted from https://github.com/Alexhuszagh/rust-lexical. + +//! Precalculated large powers for limbs. + +#[cfg(limb_width_32)] +pub(crate) use super::large_powers32::*; + +#[cfg(limb_width_64)] +pub(crate) use super::large_powers64::*; diff --git a/third_party/rust/serde_json/src/lexical/large_powers32.rs b/third_party/rust/serde_json/src/lexical/large_powers32.rs new file mode 100644 index 0000000000..7991197262 --- /dev/null +++ b/third_party/rust/serde_json/src/lexical/large_powers32.rs @@ -0,0 +1,183 @@ +// Adapted from https://github.com/Alexhuszagh/rust-lexical. + +//! Precalculated large powers for 32-bit limbs. + +/// Large powers (&[u32]) for base5 operations. +const POW5_1: [u32; 1] = [5]; +const POW5_2: [u32; 1] = [25]; +const POW5_3: [u32; 1] = [625]; +const POW5_4: [u32; 1] = [390625]; +const POW5_5: [u32; 2] = [2264035265, 35]; +const POW5_6: [u32; 3] = [2242703233, 762134875, 1262]; +const POW5_7: [u32; 5] = [3211403009, 1849224548, 3668416493, 3913284084, 1593091]; +const POW5_8: [u32; 10] = [ + 781532673, 64985353, 253049085, 594863151, 3553621484, 3288652808, 3167596762, 2788392729, + 3911132675, 590, +]; +const POW5_9: [u32; 19] = [ + 2553183233, 3201533787, 3638140786, 303378311, 1809731782, 3477761648, 3583367183, 649228654, + 2915460784, 487929380, 1011012442, 1677677582, 3428152256, 1710878487, 1438394610, 2161952759, + 4100910556, 1608314830, 349175, +]; +const POW5_10: [u32; 38] = [ + 4234999809, 2012377703, 2408924892, 1570150255, 3090844311, 3273530073, 1187251475, 2498123591, + 3364452033, 1148564857, 687371067, 2854068671, 1883165473, 505794538, 2988060450, 3159489326, + 2531348317, 3215191468, 849106862, 3892080979, 3288073877, 2242451748, 4183778142, 2995818208, + 2477501924, 325481258, 2487842652, 1774082830, 1933815724, 2962865281, 1168579910, 2724829000, + 2360374019, 2315984659, 2360052375, 3251779801, 1664357844, 28, +]; +const POW5_11: [u32; 75] = [ + 689565697, 4116392818, 1853628763, 516071302, 2568769159, 365238920, 336250165, 1283268122, + 3425490969, 248595470, 2305176814, 2111925499, 507770399, 2681111421, 589114268, 591287751, + 1708941527, 4098957707, 475844916, 3378731398, 2452339615, 2817037361, 2678008327, 1656645978, + 2383430340, 73103988, 448667107, 2329420453, 3124020241, 3625235717, 3208634035, 2412059158, + 2981664444, 4117622508, 838560765, 3069470027, 270153238, 1802868219, 3692709886, 2161737865, + 2159912357, 2585798786, 837488486, 4237238160, 2540319504, 3798629246, 3748148874, 1021550776, + 2386715342, 1973637538, 1823520457, 1146713475, 833971519, 3277251466, 905620390, 26278816, + 2680483154, 2294040859, 373297482, 5996609, 4109575006, 512575049, 917036550, 1942311753, + 2816916778, 3248920332, 1192784020, 3537586671, 2456567643, 2925660628, 759380297, 888447942, + 3559939476, 3654687237, 805, +]; +const POW5_12: [u32; 149] = [ + 322166785, 3809044581, 2994556223, 1239584207, 3962455841, 4001882964, 3053876612, 915114683, + 2783289745, 785739093, 4253185907, 3931164994, 1370983858, 2553556126, 3360742076, 2255410929, + 422849554, 2457422215, 3539495362, 1720790602, 1908931983, 1470596141, 592794347, 4219465164, + 4085652704, 941661409, 2534650953, 885063988, 2355909854, 2812815516, 767256131, 3821757683, + 2155151105, 3817418473, 281116564, 2834395026, 2821201622, 2524625843, 1511330880, 2572352493, + 330571332, 2951088579, 2730271766, 4044456479, 4212286644, 2444937588, 3603420843, 2387148597, + 1142537539, 3299235429, 1751012624, 861228086, 2873722519, 230498814, 1023297821, 2553128038, + 3421129895, 2651917435, 2042981258, 1606787143, 2228751918, 447345732, 1930371132, 1784132011, + 3612538790, 2275925090, 2487567871, 1080427616, 2009179183, 3383506781, 3899054063, 1950782960, + 2168622213, 2717674390, 3616636027, 2079341593, 1530129217, 1461057425, 2406264415, 3674671357, + 2972036238, 2019354295, 1455849819, 1866918619, 1324269294, 424891864, 2722422332, 2641594816, + 1400249021, 3482963993, 3734946379, 225889849, 1891545473, 777383150, 3589824633, 4117601611, + 4220028667, 334453379, 1083130821, 1060342180, 4208163139, 1489826908, 4163762246, 1096580926, + 689301528, 2336054516, 1782865703, 4175148410, 3398369392, 2329412588, 3001580596, 59740741, + 3202189932, 3351895776, 246185302, 718535188, 3772647488, 4151666556, 4055698133, 2461934110, + 2281316281, 3466396836, 3536023465, 1064267812, 2955456354, 2423805422, 3627960790, 1325057500, + 3876919979, 2009959531, 175455101, 184092852, 2358785571, 3842977831, 2485266289, 487121622, + 4159252710, 4075707558, 459389244, 300652075, 2521346588, 3458976673, 888631636, 2076098096, + 3844514585, 2363697580, 3729421522, 3051115477, 649395, +]; +const POW5_13: [u32; 298] = [ + 711442433, 3564261005, 2399042279, 4170849936, 4010295575, 1423987028, 330414929, 1349249065, + 4213813618, 3852031822, 4040843590, 2154565331, 3094013374, 1159028371, 3227065538, 2115927092, + 2085102554, 488590542, 2609619432, 3602898805, 3812736528, 3269439096, 23816114, 253984538, + 1035905997, 2942969204, 3400787671, 338562688, 1637191975, 740509713, 2264962817, 3410753922, + 4162231428, 2282041228, 1759373012, 3155367777, 4278913285, 1420532801, 1981002276, 438054990, + 1006507643, 1142697287, 1332538012, 2029019521, 3949305784, 818392641, 2491288846, 2716584663, + 3648886102, 556814413, 444795339, 4071412999, 1066321706, 4253169466, 2510832316, 672091442, + 4083256000, 2165985028, 1841538484, 3549854235, 364431512, 3707648143, 1162785440, 2268641545, + 281340310, 735693841, 848809228, 1700785200, 2919703985, 4094234344, 58530286, 965505005, + 1000010347, 3381961808, 3040089923, 1973852082, 2890971585, 1019960210, 4292895237, 2821887841, + 3756675650, 3951282907, 3885870583, 1008791145, 503998487, 1881258362, 1949332730, 392996726, + 2012973814, 3970014187, 2461725150, 2942547730, 3728066699, 2766901132, 3778532841, 1085564064, + 2278673896, 1116879805, 3448726271, 774279411, 157211670, 1506320155, 531168605, 1362654525, + 956967721, 2148871960, 769186085, 4186232894, 2055679604, 3248365487, 3981268013, 3975787984, + 2489510517, 3309046495, 212771124, 933418041, 3371839114, 562115198, 1853601831, 757336096, + 1354633440, 1486083256, 2872126393, 522920738, 1141587749, 3210903262, 1926940553, 3054024853, + 2021162538, 2262742000, 1877899947, 3147002868, 669840763, 4158174590, 4238502559, 1023731922, + 3386840011, 829588074, 3449720188, 2835142880, 2999162007, 813056473, 482949569, 638108879, + 3067201471, 1026714238, 4004452838, 2383667807, 3999477803, 771648919, 630660440, 3827121348, + 176185980, 2878191002, 2666149832, 3909811063, 2429163983, 2665690412, 907266128, 4269332098, + 2022665808, 1527122180, 3072053668, 1072477492, 3006022924, 549664855, 2800340954, 37352654, + 1212772743, 2711280533, 3029527946, 2511120040, 1305308377, 3474662224, 4226330922, 442988428, + 954940108, 3274548099, 4212288177, 2688499880, 3982226758, 3922609956, 1279948029, 1939943640, + 3650489901, 2733364929, 2494263275, 1864579964, 1225941120, 2390465139, 1267503249, 3533240729, + 904410805, 2842550015, 2517736241, 1796069820, 3335274381, 673539835, 1924694759, 3598098235, + 2792633405, 16535707, 3703535497, 3592841791, 2929082877, 1317622811, 294990855, 1396706563, + 2383271770, 3853857605, 277813677, 277580220, 1101318484, 3761974115, 1132150143, 2544692622, + 3419825776, 743770306, 1695464553, 1548693232, 2421159615, 2575672031, 2678971806, 1591267897, + 626546738, 3823443129, 267710932, 1455435162, 2353985540, 3248523795, 335348168, 3872552561, + 2814522612, 2634118860, 3503767026, 1301019273, 1414467789, 722985138, 3070909565, 4253482569, + 3744939841, 558142907, 2229819389, 13833173, 77003966, 2763671364, 3905603970, 2931990126, + 2280419384, 1879090457, 2934846267, 4284933164, 2331863845, 62191163, 3178861020, 1522063815, + 785672270, 1215568492, 2936443917, 802972489, 2956820173, 3916732783, 2893572089, 1391232801, + 3168640330, 2396859648, 894950918, 1103583736, 961991865, 2807302642, 305977505, 3054505899, + 1048256994, 781017659, 2459278754, 3164823415, 537658277, 905753687, 464963300, 4149131560, + 1029507924, 2278300961, 1231291503, 414073408, 3630740085, 2345841814, 475358196, 3258243317, + 4167625072, 4178911231, 2927355042, 655438830, 3138378018, 623200562, 2785714112, 273403236, + 807993669, 98, +]; +const POW5_14: [u32; 595] = [ + 1691320321, 2671006246, 1682531301, 2072858707, 1240508969, 3108358191, 1125119096, 2470144952, + 1610099978, 1690632660, 1941696884, 2663506355, 1006364675, 3909158537, 4147711374, 1072663936, + 4078768933, 745751659, 4123687570, 471458681, 655028926, 4113407388, 3945524552, 985625313, + 1254424514, 2127508744, 570530434, 945388122, 3194649404, 2589065070, 2731705399, 202030749, + 2090780394, 3348662271, 1481754777, 1130635472, 4025144705, 1924486271, 2578567861, 125491448, + 1558036315, 994248173, 3817216711, 763950077, 1030439870, 959586474, 3845661701, 483795093, + 1637944470, 2275463649, 3398804829, 1758016486, 2665513698, 2004912571, 1094885097, 4223064276, + 3307819021, 651121777, 1757003305, 3603542336, 129917786, 2215974994, 3042386306, 2205352757, + 3944939700, 3710987569, 97967515, 1217242524, 930630949, 3660328512, 1787663098, 1784141600, + 2500542892, 4034561586, 3444961378, 785043562, 3869499367, 885623728, 2625011087, 3053789617, + 1965731793, 3900511934, 2648823592, 3851062028, 3321968688, 799195417, 1011847510, 1369129160, + 1348009103, 2876796955, 2915408967, 3305284948, 263399535, 1715990604, 2645821294, 1587844552, + 2624912049, 3035631499, 2306636348, 3499275462, 675152704, 854794152, 4004972748, 1739996642, + 1333476491, 4012621867, 3658792931, 3297985728, 2864481726, 3066357406, 785287846, 1671499798, + 433044045, 1919608025, 264833858, 3999983367, 1116778570, 1301982149, 4213901070, 4081649357, + 536169226, 1389008649, 188923873, 373495152, 2551132278, 1800758715, 3951840330, 2632334454, + 3118778225, 1034046547, 1862428410, 3037609062, 1994608505, 29051798, 2571685694, 264151332, + 2260643090, 2717535964, 3508441116, 3283713017, 1903365635, 923575694, 1219598101, 2288281570, + 3676533911, 1014136356, 555142354, 2389170030, 4185108175, 884862419, 836141292, 2957159173, + 1997444768, 4233903127, 2876184692, 3089125070, 1480848293, 1097600237, 299700527, 2507669891, + 2982628312, 2114881043, 2529576251, 2812279824, 2987750993, 4241938954, 2204775591, 1037094060, + 829315638, 1231047149, 52608178, 3735136637, 3455232602, 962039123, 488286513, 50685385, + 3516451821, 843975207, 1572355722, 675489076, 2428445672, 1555117248, 3708476086, 10375249, + 4172112346, 2117510871, 2227658327, 3187664554, 3050656558, 328034318, 3179601324, 1247769761, + 3439263953, 1431538938, 2962525068, 1213366289, 3813013550, 2651093719, 1860661503, 3933716208, + 264320617, 789980519, 2257856172, 102000748, 977269860, 1113845122, 3008928583, 1461738106, + 557786285, 2926560363, 1038106190, 3643478847, 828004507, 457818698, 1933056971, 373408056, + 2076808229, 3160935130, 2781854874, 2519636100, 177606000, 4237103862, 3977834316, 1621936232, + 2599050516, 319893558, 3343370366, 765044144, 976657331, 7026264, 294277429, 3829376742, + 3029627280, 2705178718, 3614653880, 230519152, 3288033233, 293525479, 3805751881, 3227511198, + 2520308544, 3648103003, 1111086184, 437622105, 2232033852, 3239146386, 584244184, 1450926016, + 2462430443, 3226534010, 298582169, 4214576928, 1762099469, 964985185, 1585788148, 1641127666, + 787006566, 2315956284, 3258232694, 2275058964, 2541003317, 1508235863, 2613339827, 4080647514, + 1152057965, 3149266279, 731345410, 914737650, 65395712, 1884566942, 1379520432, 2611027720, + 4163073378, 2619704967, 2746552541, 1388822415, 3005141199, 843440249, 4288674003, 3136174279, + 4051522914, 4144149433, 3427566947, 3419023197, 3758479825, 3893877676, 96899594, 1657725776, + 253618880, 434129337, 1499045748, 2996992534, 4036042074, 2110713869, 906222950, 928326225, + 2541827893, 1604330202, 226792470, 4022228930, 815850898, 1466012310, 3377712199, 292769859, + 2822055597, 3225701344, 3052947004, 385831222, 705324593, 4030158636, 3540280538, 2982120874, + 2136414455, 255762046, 3852783591, 3262064164, 2358991588, 3756586117, 4143612643, 3326743817, + 2897365738, 807711264, 3719310016, 3721264861, 3627337076, 944539331, 3640975513, 3712525681, + 1162911839, 2008243316, 2179489649, 2867584109, 261861553, 3570253908, 2062868357, 2220328623, + 3857004679, 3744109002, 4138041873, 1451860932, 2364975637, 2802161722, 2680106834, 753401584, + 1223182946, 1245401957, 4163377735, 3565815922, 2216942838, 4036140094, 71979081, 3924559643, + 400477238, 551750683, 1174153235, 859969898, 1185921017, 1711399735, 812991545, 4051735761, + 3549118738, 1631653329, 3631835958, 3648867800, 1206500363, 2155893137, 361030362, 3454286017, + 2505909489, 1083595169, 453595313, 1510564703, 1706163902, 1632924345, 1381875722, 1661526119, + 1082778324, 3571910052, 1140625929, 851544870, 1145546234, 2938573139, 907528924, 1304752338, + 1764668294, 1788942063, 1700368828, 104979467, 1413911959, 3327497828, 1956384744, 1272712474, + 2815637534, 3307809377, 1320574940, 1111968962, 4073107827, 434096622, 169451929, 3201183459, + 3331028877, 2852366972, 3369830128, 2924794558, 3106537952, 3739481231, 1612955817, 4138608722, + 2721281595, 2755775390, 843505117, 982234295, 1157276611, 814674632, 4246504726, 3532006708, + 992340967, 1647538031, 204696133, 193866982, 3899126129, 300851698, 1379496684, 1759463683, + 1354782756, 1374637239, 3410883240, 1073406229, 3038431791, 1053909855, 3607043270, 173719711, + 3733903830, 171820911, 1573050589, 932781534, 4183534770, 2158849555, 372245998, 3573073830, + 841339264, 2759200520, 1610547277, 2603293319, 3890906486, 1557138278, 3964109906, 677238797, + 537994297, 1124184993, 4287078344, 4207654540, 2943022776, 2977947524, 3255359985, 4098397558, + 2274666217, 2915862060, 243524940, 2467726756, 2869020032, 507521339, 3403121914, 522051455, + 1803903108, 3471254194, 473535371, 1948602036, 3352095732, 3116527002, 1795743673, 775867940, + 2551469548, 3757442064, 3162525227, 3765412747, 3040105484, 1927625810, 48214767, 2997207130, + 1342349989, 2536583992, 1501320191, 3592287317, 887432730, 967585477, 3334212779, 948663609, + 1064513472, 15386372, 2465931737, 3230242590, 3036652803, 2063155087, 1927500726, 2821790499, + 2187774383, 501520074, 3688568496, 3606711121, 2576459247, 3176542345, 378322447, 156541411, + 1400607301, 1406179107, 677848877, 2253753529, 193196070, 4207435024, 4166396241, 509467541, + 2906024136, 1221753746, 3375413222, 431327897, 2749265123, 2848827671, 3412997614, 2051920238, + 1283516885, 1300498239, 1957256104, 2634010560, 3531900395, 360276850, 1461184973, 2012063967, + 2873572430, 2914608609, 4289554777, 1539331673, 1859532928, 4213441063, 538215691, 3512720863, + 4258743698, 3040408445, 982396546, 343095663, 4138069496, 1021581857, 214185242, 1968079460, + 2864275059, 3347192726, 4096783459, 3259169450, 3707808869, 142485006, 399610869, 230556456, + 2219467721, 4191227798, 2242548189, 3136366572, 179755707, 3464881829, 452317775, 3887426070, + 3446430233, 1473370015, 1576807208, 3964523248, 419325089, 2373067114, 1596072055, 1928415752, + 3635452689, 1005598891, 3335462724, 3290848636, 3669078247, 1178176812, 2110774376, 3068593619, + 1253036518, 908857731, 3631223047, 4138506423, 2903592318, 3596915748, 3289036113, 3721512676, + 2704409359, 3386016968, 3676268074, 2185259502, 1096257611, 3360076717, 3548676554, 170167319, + 3360064287, 3899940843, 9640, +]; + +pub(crate) const POW5: [&'static [u32]; 14] = [ + &POW5_1, &POW5_2, &POW5_3, &POW5_4, &POW5_5, &POW5_6, &POW5_7, &POW5_8, &POW5_9, &POW5_10, + &POW5_11, &POW5_12, &POW5_13, &POW5_14, +]; diff --git a/third_party/rust/serde_json/src/lexical/large_powers64.rs b/third_party/rust/serde_json/src/lexical/large_powers64.rs new file mode 100644 index 0000000000..ee36561088 --- /dev/null +++ b/third_party/rust/serde_json/src/lexical/large_powers64.rs @@ -0,0 +1,625 @@ +// Adapted from https://github.com/Alexhuszagh/rust-lexical. + +//! Precalculated large powers for 64-bit limbs. + +/// Large powers (&[u64]) for base5 operations. +const POW5_1: [u64; 1] = [5]; +const POW5_2: [u64; 1] = [25]; +const POW5_3: [u64; 1] = [625]; +const POW5_4: [u64; 1] = [390625]; +const POW5_5: [u64; 1] = [152587890625]; +const POW5_6: [u64; 2] = [3273344365508751233, 1262]; +const POW5_7: [u64; 3] = [7942358959831785217, 16807427164405733357, 1593091]; +const POW5_8: [u64; 5] = [ + 279109966635548161, + 2554917779393558781, + 14124656261812188652, + 11976055582626787546, + 2537941837315, +]; +const POW5_9: [u64; 10] = [ + 13750482914757213185, + 1302999927698857842, + 14936872543252795590, + 2788415840139466767, + 2095640732773017264, + 7205570348933370714, + 7348167152523113408, + 9285516396840364274, + 6907659600622710236, + 349175, +]; +const POW5_10: [u64; 19] = [ + 8643096425819600897, + 6743743997439985372, + 14059704609098336919, + 10729359125898331411, + 4933048501514368705, + 12258131603170554683, + 2172371001088594721, + 13569903330219142946, + 13809142207969578845, + 16716360519037769646, + 9631256923806107285, + 12866941232305103710, + 1397931361048440292, + 7619627737732970332, + 12725409486282665900, + 11703051443360963910, + 9947078370803086083, + 13966287901448440471, + 121923442132, +]; +const POW5_11: [u64; 38] = [ + 17679772531488845825, + 2216509366347768155, + 1568689219195129479, + 5511594616325588277, + 1067709417009240089, + 9070650952098657518, + 11515285870634858015, + 2539561553659505564, + 17604889300961091799, + 14511540856854204724, + 12099083339557485471, + 7115240299237943815, + 313979240050606788, + 10004784664717172195, + 15570268847930131473, + 10359715202835930803, + 17685054012115162812, + 13183273382855797757, + 7743260039872919062, + 9284593436392572926, + 11105921222066415013, + 18198799323400703846, + 16314988383739458320, + 4387527177871570570, + 8476708682254672590, + 4925096874831034057, + 14075687868072027455, + 112866656203221926, + 9852830467773230418, + 25755239915196746, + 2201493076310172510, + 8342165458688466438, + 13954006576066379050, + 15193819059903295636, + 12565616718911389531, + 3815854855847885129, + 15696762163583540628, + 805, +]; +const POW5_12: [u64; 75] = [ + 16359721904723189761, + 5323973632697650495, + 17187956456762001185, + 3930387638628283780, + 3374723710406992273, + 16884225088663222131, + 10967440051041439154, + 9686916182456720060, + 10554548046311730194, + 7390739362393647554, + 6316162333127736719, + 18122464886584070891, + 4044404959645932768, + 3801320885861987401, + 12080950653257274590, + 16414324262488991299, + 16395687498836410113, + 12173633940896186260, + 10843185433142632150, + 11048169832730399808, + 12674828934734683716, + 17370808310130582550, + 10500926985433408692, + 10252725158410704555, + 14170108270502067523, + 3698946465517688080, + 989984870770509463, + 10965601426733943069, + 11389898658438335655, + 6901098232861256586, + 1921335291173932590, + 7662788640922083388, + 9775023833308395430, + 4640401278902814207, + 14532050972198413359, + 8378549018693130223, + 11672322628395371653, + 8930704142764178555, + 6275193859483102017, + 15782593304269205087, + 8673060659034172558, + 8018354414354334043, + 1824896661540749038, + 11345563346725559868, + 14959216444480821949, + 970189517688324683, + 3338835207603007873, + 17684964260791738489, + 1436466329061721851, + 4554134986752476101, + 6398757850768963907, + 4709779218751158342, + 10033277748582410264, + 17932125878679265063, + 10004750887749091440, + 256584531835386932, + 14396282740722731628, + 3086085133731396950, + 17831272085689600064, + 10573926491412564693, + 14888061047859191737, + 4570995450261499817, + 10410165022312935266, + 5691078631447480790, + 8632710455805418155, + 790672778942823293, + 16505464105756800547, + 2092171438149740401, + 17505030673829275878, + 1291290830058928444, + 14856191690683232796, + 8916773426496500052, + 10152003807578858265, + 13104441193763861714, + 649395, +]; +const POW5_13: [u64; 149] = [ + 15308384451594534913, + 17913664074042735335, + 6115977719198531863, + 5794980608663993169, + 16544350702855106930, + 9253787637781258566, + 4977988951675168190, + 9087837664087448770, + 2098480401110016986, + 15474332540882100712, + 14042133997396540944, + 1090855284423485362, + 12639956485351058381, + 1454115676006639319, + 3180465001342538023, + 14649076551958697729, + 9801292446545910916, + 13552201410826594004, + 6101141927469189381, + 1881431857880609316, + 4907847477899433595, + 8714572486973123228, + 3514969632331374520, + 11667642286891470094, + 2391499697425323350, + 17486585679659076043, + 18267223761882105642, + 2886610765822313148, + 9302834862968900288, + 15246507846733637044, + 15924227519624562840, + 9743741243284697760, + 3159780987244964246, + 7304816812369628428, + 17584602612559717809, + 4146812420657846766, + 14525415362681041515, + 8477630142371600195, + 4380695748062263745, + 12119915994367943173, + 16970630866565485122, + 4332724980155264503, + 8079943140620527639, + 1687908087554405626, + 17051081099834002166, + 12638146269730763230, + 11883749876933445771, + 4662462156371383785, + 4796962238316531176, + 3325504751659868927, + 6469595803187862550, + 5852556621152583005, + 9229334792448387881, + 17979733373938620709, + 13951623534175792756, + 17075879371091039277, + 14212246479457938037, + 4008999959804158260, + 2414266395366403722, + 3252733766253918247, + 6382678985007829216, + 2245927470982310841, + 13790724502051307301, + 13116936866733148041, + 9718402891306794538, + 13516274400356104875, + 17859223875778049403, + 4396895129099725471, + 3563053650368467915, + 12176845952536972668, + 3492050964335269015, + 2740656767075170753, + 4409704077614761919, + 10237775279597492710, + 3314206875098230827, + 16437361028114095448, + 12361736225407656572, + 16792510651790145480, + 11449053143229929935, + 18336641737580333136, + 6558939822118891088, + 4606255756908155300, + 2360792578991605004, + 160428430149144538, + 11644861220729221511, + 10785178451159739786, + 14923560618031934681, + 1902620814992781610, + 14064076995338910412, + 11547019064112212657, + 16847481479966225734, + 8331994491163145469, + 11739712981738851885, + 8008309968651120619, + 10266969595459035264, + 15175153381217702033, + 12208659352573720245, + 7714061140750342961, + 2892831567213510541, + 15453714249045017319, + 71020323573871677, + 15431137995750602633, + 5659146884637671933, + 5998809010488554503, + 16552192379299157850, + 1192197967194298797, + 16157555793424861524, + 10929371590994640255, + 3194469143425738352, + 6651586784672005225, + 11062427140788057791, + 6834443579468668318, + 16421563197797455922, + 6251046422506172884, + 13952303462156793860, + 16632486601871393224, + 11313454360291325172, + 5587835232504462834, + 3105197524618514637, + 18268568531031972989, + 2397205535804309313, + 59413027864729597, + 11869878125348715710, + 12592801707270523266, + 8070632061321113656, + 18403647807860650811, + 267109013517069093, + 6537214311028855260, + 5220826919973709902, + 3448740582779163661, + 16822239213112884941, + 5975299384311048185, + 10294433804430712138, + 4739856055412448774, + 12057273038326387897, + 13119002941950056609, + 3354445304051737058, + 13592813067499314594, + 3890182464434078629, + 17820384357466425060, + 9785228118969879380, + 1778431746734556271, + 10075313876350055029, + 13994048489400919028, + 17948287074199726448, + 2815088342305858722, + 2676626035777198370, + 1174257960026283968, + 421714788677, +]; +const POW5_14: [u64; 298] = [ + 11471884475673051137, + 8902860357476377573, + 13350296775839230505, + 10609191786344608888, + 7261211985859587338, + 11439672689354862964, + 16789708072300570627, + 4607056528866348430, + 3202978990421512997, + 2024899620433984146, + 17666950207239811774, + 4233228489390288200, + 9137580478688460738, + 4060411066587388546, + 11119949806060600124, + 867715462473090103, + 14382394941384869610, + 4856042377419278489, + 8265605599571137921, + 538981667666252469, + 4270263388700786523, + 3281140600308898503, + 4121392524544394174, + 2077884106245940229, + 9773041957329767574, + 7550623316597646685, + 8611033926449791714, + 18137922955420802793, + 2796546741236224013, + 15477096484628446761, + 9517540128113714010, + 9471917970500821378, + 15938570248662483124, + 5228016831978462619, + 15720991252586974501, + 7662829825220776698, + 17328310068068434348, + 3371736428170309730, + 3803724952191098855, + 13115926536504376719, + 16752571196153442257, + 16540185467776259880, + 3432518182450051120, + 5880364967211798870, + 12355748840305392783, + 14196090758536469575, + 7370123524686686319, + 6819740424617592686, + 13037938013537368753, + 15029273671291927100, + 3671312928327205696, + 7473228676544792780, + 17234079691312938123, + 14164740848093544419, + 13169904779481875902, + 7179036968465894054, + 8244653688947194445, + 17179797746073799490, + 5591970751047577674, + 17530550506268329742, + 5965746721852312330, + 1604149463243472865, + 7734199791463116918, + 11305790396015856714, + 4441196105025505137, + 13046431581185664762, + 124776524294606713, + 1134521334706523966, + 11671728093344476434, + 14103440020972933148, + 3966727403013869059, + 9828094508409132821, + 4355682486381147287, + 10261407143988481234, + 3800455155249557199, + 12700901937937547500, + 18184475466894579360, + 13267691151779895412, + 4714157123477697445, + 10770360171308585263, + 9083344917597998040, + 12078649873810212155, + 18218989082046199377, + 4454285072780637351, + 5287307245618354742, + 16042289702059031730, + 4131926574212754010, + 217692071448455473, + 3624845916216282093, + 2901203491797614218, + 6679177724033967080, + 44561358851332790, + 9094639944041587162, + 13690915012276084311, + 1408896670826320686, + 5359130319612337580, + 6148412925099835601, + 5211368532286409612, + 11386360825549027374, + 16895182466965795071, + 3392940493846427241, + 438089879085393580, + 4783928372776399972, + 6278117363595909959, + 12569481049412674733, + 15648622492570893902, + 1966316336235305115, + 1603775390515993547, + 13576113010204316709, + 10821754650102840474, + 18198222517222903152, + 6966163076615302988, + 1373932372410129684, + 3285839581819684990, + 30177575069719475, + 16447047871247307061, + 11618654126674833808, + 990072222556306872, + 1260682336135768017, + 13862055046689532489, + 15668483092844698432, + 1879572630092764264, + 13912027797058626108, + 6231679788219816920, + 13857858054844167403, + 18101470072534728857, + 4144579812461609229, + 7048589655616599284, + 9946956499532694630, + 9771303850109874038, + 6477823708780339765, + 17526247621747041971, + 13525995675852669549, + 3928768291901239810, + 8094153383078124544, + 11214278667728965552, + 11251547162596832610, + 5964946855123292381, + 3622548288590237903, + 13469765967150053587, + 17798986288523466082, + 14684592818807932259, + 16724077276802963921, + 7119877993753121290, + 1864571304902781632, + 12871984921385213812, + 9065447042604670298, + 3987130777300360550, + 6890545752116901685, + 17275341711601865750, + 6296474927799264658, + 1257436973037243463, + 13854281781965301421, + 1657132483318662716, + 17309399540017292849, + 12808111630089217242, + 1098489625264462071, + 14010458905686364135, + 16134414519481621220, + 14288255900328821475, + 3469093466388187882, + 15982710881468295872, + 4056765540058056052, + 15945176389096104089, + 8625339365793505375, + 12316179968863788913, + 15334123773538054321, + 9536238824220581765, + 16080825720106203271, + 6235695225418121745, + 12035192956458019349, + 3235835166714703698, + 5348960676912581218, + 15315062772709464647, + 17335089708021308662, + 16855855317958414409, + 2369751139431140406, + 3693542588628609043, + 7350405893393987577, + 17402072586341663801, + 7007897690013647122, + 15671767872059304758, + 9259490518292347915, + 14836045474406130394, + 4654005815464502513, + 6487825998330548401, + 7013356660323385022, + 7136200343936679946, + 15341236858676437716, + 3657357368867197449, + 12621075530054608378, + 5603868621997066972, + 7683447656788439942, + 450883379216880060, + 14291494350184945047, + 5466258454997635048, + 14206933098432772126, + 4775870327277641692, + 1864430798867181939, + 13748978265070608793, + 12250822864261576589, + 12561896977498605296, + 16060949594257359328, + 17775189113543311529, + 11835965177892927035, + 4218664174878121437, + 3499000902478111683, + 15169853304359126294, + 7076121963053575143, + 832652347668916805, + 1292148207755194737, + 7556838978364207852, + 5904021986723518500, + 4610244652288570024, + 4526508363195533871, + 746120481022614726, + 737965197247830486, + 4006266184415762653, + 9272188239892688050, + 15346235246415709678, + 11850675997347533184, + 11181059668610842701, + 6687857983250662774, + 2908718488661492818, + 4828337780126983225, + 18071738646453002184, + 12790187227727197880, + 17602483480871623153, + 12523532189621855977, + 10598805712727696716, + 2179787555896149376, + 2242193929457337594, + 14908923241136742532, + 8369182018012550027, + 13385381554043022324, + 3332327430110633913, + 16138090784046208492, + 16172324607469047339, + 8279089815915615244, + 12872906602736235247, + 10894545290539475621, + 15428756545851905023, + 4155747980686992922, + 4074479178894544043, + 66083965608603584, + 13873786284662268377, + 8861183628277687555, + 12119497911296021430, + 2154012318305274287, + 15490706314503067312, + 13643145488710608367, + 672340241093017103, + 6039493278284091973, + 9679797700977436461, + 18070795828318171174, + 2188146431134935377, + 5247392385741514952, + 1852539214842869734, + 12235621681634112739, + 8812930319623534062, + 5585597406294108629, + 11312989214475901864, + 1547377291787797995, + 8641748937186208205, + 12518148659168623694, + 6611379197521520985, + 18096591571068008576, + 15087021227100112139, + 13058454842015958418, + 1473584652966833794, + 4387660670140018168, + 8452836916843525402, + 14376083294443363955, + 13998026203969090659, + 611968444648172645, + 990232438801273845, + 18001186324715561929, + 13470591857250177501, + 14881554140239420091, + 16696367836720124495, + 6328076032778459673, + 17027497695968504616, + 10192245646262428833, + 8282482589527318647, + 4319014353374321425, + 14134087271041670980, + 5060230880114618599, + 13179509240430058600, + 3903514232614801894, + 17774749744702165255, + 15448635507030969726, + 15983775238358480209, + 14542832143965487887, + 9385618098039514666, + 14431419612662304843, + 730863073501675978, + 16750118380379734815, + 9640, +]; + +pub(crate) const POW5: [&[u64]; 14] = [ + &POW5_1, &POW5_2, &POW5_3, &POW5_4, &POW5_5, &POW5_6, &POW5_7, &POW5_8, &POW5_9, &POW5_10, + &POW5_11, &POW5_12, &POW5_13, &POW5_14, +]; diff --git a/third_party/rust/serde_json/src/lexical/math.rs b/third_party/rust/serde_json/src/lexical/math.rs new file mode 100644 index 0000000000..37cc1d24ad --- /dev/null +++ b/third_party/rust/serde_json/src/lexical/math.rs @@ -0,0 +1,886 @@ +// Adapted from https://github.com/Alexhuszagh/rust-lexical. + +//! Building-blocks for arbitrary-precision math. +//! +//! These algorithms assume little-endian order for the large integer +//! buffers, so for a `vec![0, 1, 2, 3]`, `3` is the most significant limb, +//! and `0` is the least significant limb. + +use super::large_powers; +use super::num::*; +use super::small_powers::*; +use alloc::vec::Vec; +use core::{cmp, iter, mem}; + +// ALIASES +// ------- + +// Type for a single limb of the big integer. +// +// A limb is analogous to a digit in base10, except, it stores 32-bit +// or 64-bit numbers instead. +// +// This should be all-known 64-bit platforms supported by Rust. +// https://forge.rust-lang.org/platform-support.html +// +// Platforms where native 128-bit multiplication is explicitly supported: +// - x86_64 (Supported via `MUL`). +// - mips64 (Supported via `DMULTU`, which `HI` and `LO` can be read-from). +// +// Platforms where native 64-bit multiplication is supported and +// you can extract hi-lo for 64-bit multiplications. +// aarch64 (Requires `UMULH` and `MUL` to capture high and low bits). +// powerpc64 (Requires `MULHDU` and `MULLD` to capture high and low bits). +// +// Platforms where native 128-bit multiplication is not supported, +// requiring software emulation. +// sparc64 (`UMUL` only supported double-word arguments). + +// 32-BIT LIMB +#[cfg(limb_width_32)] +pub type Limb = u32; + +#[cfg(limb_width_32)] +pub const POW5_LIMB: &[Limb] = &POW5_32; + +#[cfg(limb_width_32)] +pub const POW10_LIMB: &[Limb] = &POW10_32; + +#[cfg(limb_width_32)] +type Wide = u64; + +// 64-BIT LIMB +#[cfg(limb_width_64)] +pub type Limb = u64; + +#[cfg(limb_width_64)] +pub const POW5_LIMB: &[Limb] = &POW5_64; + +#[cfg(limb_width_64)] +pub const POW10_LIMB: &[Limb] = &POW10_64; + +#[cfg(limb_width_64)] +type Wide = u128; + +/// Cast to limb type. +#[inline] +pub(crate) fn as_limb<T: Integer>(t: T) -> Limb { + Limb::as_cast(t) +} + +/// Cast to wide type. +#[inline] +fn as_wide<T: Integer>(t: T) -> Wide { + Wide::as_cast(t) +} + +// SPLIT +// ----- + +/// Split u64 into limbs, in little-endian order. +#[inline] +#[cfg(limb_width_32)] +fn split_u64(x: u64) -> [Limb; 2] { + [as_limb(x), as_limb(x >> 32)] +} + +/// Split u64 into limbs, in little-endian order. +#[inline] +#[cfg(limb_width_64)] +fn split_u64(x: u64) -> [Limb; 1] { + [as_limb(x)] +} + +// HI64 +// ---- + +// NONZERO + +/// Check if any of the remaining bits are non-zero. +#[inline] +pub fn nonzero<T: Integer>(x: &[T], rindex: usize) -> bool { + let len = x.len(); + let slc = &x[..len - rindex]; + slc.iter().rev().any(|&x| x != T::ZERO) +} + +/// Shift 64-bit integer to high 64-bits. +#[inline] +fn u64_to_hi64_1(r0: u64) -> (u64, bool) { + debug_assert!(r0 != 0); + let ls = r0.leading_zeros(); + (r0 << ls, false) +} + +/// Shift 2 64-bit integers to high 64-bits. +#[inline] +fn u64_to_hi64_2(r0: u64, r1: u64) -> (u64, bool) { + debug_assert!(r0 != 0); + let ls = r0.leading_zeros(); + let rs = 64 - ls; + let v = match ls { + 0 => r0, + _ => (r0 << ls) | (r1 >> rs), + }; + let n = r1 << ls != 0; + (v, n) +} + +/// Trait to export the high 64-bits from a little-endian slice. +trait Hi64<T>: AsRef<[T]> { + /// Get the hi64 bits from a 1-limb slice. + fn hi64_1(&self) -> (u64, bool); + + /// Get the hi64 bits from a 2-limb slice. + fn hi64_2(&self) -> (u64, bool); + + /// Get the hi64 bits from a 3-limb slice. + fn hi64_3(&self) -> (u64, bool); + + /// High-level exporter to extract the high 64 bits from a little-endian slice. + #[inline] + fn hi64(&self) -> (u64, bool) { + match self.as_ref().len() { + 0 => (0, false), + 1 => self.hi64_1(), + 2 => self.hi64_2(), + _ => self.hi64_3(), + } + } +} + +impl Hi64<u32> for [u32] { + #[inline] + fn hi64_1(&self) -> (u64, bool) { + debug_assert!(self.len() == 1); + let r0 = self[0] as u64; + u64_to_hi64_1(r0) + } + + #[inline] + fn hi64_2(&self) -> (u64, bool) { + debug_assert!(self.len() == 2); + let r0 = (self[1] as u64) << 32; + let r1 = self[0] as u64; + u64_to_hi64_1(r0 | r1) + } + + #[inline] + fn hi64_3(&self) -> (u64, bool) { + debug_assert!(self.len() >= 3); + let r0 = self[self.len() - 1] as u64; + let r1 = (self[self.len() - 2] as u64) << 32; + let r2 = self[self.len() - 3] as u64; + let (v, n) = u64_to_hi64_2(r0, r1 | r2); + (v, n || nonzero(self, 3)) + } +} + +impl Hi64<u64> for [u64] { + #[inline] + fn hi64_1(&self) -> (u64, bool) { + debug_assert!(self.len() == 1); + let r0 = self[0]; + u64_to_hi64_1(r0) + } + + #[inline] + fn hi64_2(&self) -> (u64, bool) { + debug_assert!(self.len() >= 2); + let r0 = self[self.len() - 1]; + let r1 = self[self.len() - 2]; + let (v, n) = u64_to_hi64_2(r0, r1); + (v, n || nonzero(self, 2)) + } + + #[inline] + fn hi64_3(&self) -> (u64, bool) { + self.hi64_2() + } +} + +// SCALAR +// ------ + +// Scalar-to-scalar operations, for building-blocks for arbitrary-precision +// operations. + +mod scalar { + use super::*; + + // ADDITION + + /// Add two small integers and return the resulting value and if overflow happens. + #[inline] + pub fn add(x: Limb, y: Limb) -> (Limb, bool) { + x.overflowing_add(y) + } + + /// AddAssign two small integers and return if overflow happens. + #[inline] + pub fn iadd(x: &mut Limb, y: Limb) -> bool { + let t = add(*x, y); + *x = t.0; + t.1 + } + + // SUBTRACTION + + /// Subtract two small integers and return the resulting value and if overflow happens. + #[inline] + pub fn sub(x: Limb, y: Limb) -> (Limb, bool) { + x.overflowing_sub(y) + } + + /// SubAssign two small integers and return if overflow happens. + #[inline] + pub fn isub(x: &mut Limb, y: Limb) -> bool { + let t = sub(*x, y); + *x = t.0; + t.1 + } + + // MULTIPLICATION + + /// Multiply two small integers (with carry) (and return the overflow contribution). + /// + /// Returns the (low, high) components. + #[inline] + pub fn mul(x: Limb, y: Limb, carry: Limb) -> (Limb, Limb) { + // Cannot overflow, as long as wide is 2x as wide. This is because + // the following is always true: + // `Wide::max_value() - (Narrow::max_value() * Narrow::max_value()) >= Narrow::max_value()` + let z: Wide = as_wide(x) * as_wide(y) + as_wide(carry); + let bits = mem::size_of::<Limb>() * 8; + (as_limb(z), as_limb(z >> bits)) + } + + /// Multiply two small integers (with carry) (and return if overflow happens). + #[inline] + pub fn imul(x: &mut Limb, y: Limb, carry: Limb) -> Limb { + let t = mul(*x, y, carry); + *x = t.0; + t.1 + } +} // scalar + +// SMALL +// ----- + +// Large-to-small operations, to modify a big integer from a native scalar. + +mod small { + use super::*; + + // MULTIPLICATIION + + /// ADDITION + + /// Implied AddAssign implementation for adding a small integer to bigint. + /// + /// Allows us to choose a start-index in x to store, to allow incrementing + /// from a non-zero start. + #[inline] + pub fn iadd_impl(x: &mut Vec<Limb>, y: Limb, xstart: usize) { + if x.len() <= xstart { + x.push(y); + } else { + // Initial add + let mut carry = scalar::iadd(&mut x[xstart], y); + + // Increment until overflow stops occurring. + let mut size = xstart + 1; + while carry && size < x.len() { + carry = scalar::iadd(&mut x[size], 1); + size += 1; + } + + // If we overflowed the buffer entirely, need to add 1 to the end + // of the buffer. + if carry { + x.push(1); + } + } + } + + /// AddAssign small integer to bigint. + #[inline] + pub fn iadd(x: &mut Vec<Limb>, y: Limb) { + iadd_impl(x, y, 0); + } + + // SUBTRACTION + + /// SubAssign small integer to bigint. + /// Does not do overflowing subtraction. + #[inline] + pub fn isub_impl(x: &mut Vec<Limb>, y: Limb, xstart: usize) { + debug_assert!(x.len() > xstart && (x[xstart] >= y || x.len() > xstart + 1)); + + // Initial subtraction + let mut carry = scalar::isub(&mut x[xstart], y); + + // Increment until overflow stops occurring. + let mut size = xstart + 1; + while carry && size < x.len() { + carry = scalar::isub(&mut x[size], 1); + size += 1; + } + normalize(x); + } + + // MULTIPLICATION + + /// MulAssign small integer to bigint. + #[inline] + pub fn imul(x: &mut Vec<Limb>, y: Limb) { + // Multiply iteratively over all elements, adding the carry each time. + let mut carry: Limb = 0; + for xi in x.iter_mut() { + carry = scalar::imul(xi, y, carry); + } + + // Overflow of value, add to end. + if carry != 0 { + x.push(carry); + } + } + + /// Mul small integer to bigint. + #[inline] + pub fn mul(x: &[Limb], y: Limb) -> Vec<Limb> { + let mut z = Vec::<Limb>::default(); + z.extend_from_slice(x); + imul(&mut z, y); + z + } + + /// MulAssign by a power. + /// + /// Theoretically... + /// + /// Use an exponentiation by squaring method, since it reduces the time + /// complexity of the multiplication to ~`O(log(n))` for the squaring, + /// and `O(n*m)` for the result. Since `m` is typically a lower-order + /// factor, this significantly reduces the number of multiplications + /// we need to do. Iteratively multiplying by small powers follows + /// the nth triangular number series, which scales as `O(p^2)`, but + /// where `p` is `n+m`. In short, it scales very poorly. + /// + /// Practically.... + /// + /// Exponentiation by Squaring: + /// running 2 tests + /// test bigcomp_f32_lexical ... bench: 1,018 ns/iter (+/- 78) + /// test bigcomp_f64_lexical ... bench: 3,639 ns/iter (+/- 1,007) + /// + /// Exponentiation by Iterative Small Powers: + /// running 2 tests + /// test bigcomp_f32_lexical ... bench: 518 ns/iter (+/- 31) + /// test bigcomp_f64_lexical ... bench: 583 ns/iter (+/- 47) + /// + /// Exponentiation by Iterative Large Powers (of 2): + /// running 2 tests + /// test bigcomp_f32_lexical ... bench: 671 ns/iter (+/- 31) + /// test bigcomp_f64_lexical ... bench: 1,394 ns/iter (+/- 47) + /// + /// Even using worst-case scenarios, exponentiation by squaring is + /// significantly slower for our workloads. Just multiply by small powers, + /// in simple cases, and use precalculated large powers in other cases. + pub fn imul_pow5(x: &mut Vec<Limb>, n: u32) { + use super::large::KARATSUBA_CUTOFF; + + let small_powers = POW5_LIMB; + let large_powers = large_powers::POW5; + + if n == 0 { + // No exponent, just return. + // The 0-index of the large powers is `2^0`, which is 1, so we want + // to make sure we don't take that path with a literal 0. + return; + } + + // We want to use the asymptotically faster algorithm if we're going + // to be using Karabatsu multiplication sometime during the result, + // otherwise, just use exponentiation by squaring. + let bit_length = 32 - n.leading_zeros() as usize; + debug_assert!(bit_length != 0 && bit_length <= large_powers.len()); + if x.len() + large_powers[bit_length - 1].len() < 2 * KARATSUBA_CUTOFF { + // We can use iterative small powers to make this faster for the + // easy cases. + + // Multiply by the largest small power until n < step. + let step = small_powers.len() - 1; + let power = small_powers[step]; + let mut n = n as usize; + while n >= step { + imul(x, power); + n -= step; + } + + // Multiply by the remainder. + imul(x, small_powers[n]); + } else { + // In theory, this code should be asymptotically a lot faster, + // in practice, our small::imul seems to be the limiting step, + // and large imul is slow as well. + + // Multiply by higher order powers. + let mut idx: usize = 0; + let mut bit: usize = 1; + let mut n = n as usize; + while n != 0 { + if n & bit != 0 { + debug_assert!(idx < large_powers.len()); + large::imul(x, large_powers[idx]); + n ^= bit; + } + idx += 1; + bit <<= 1; + } + } + } + + // BIT LENGTH + + /// Get number of leading zero bits in the storage. + #[inline] + pub fn leading_zeros(x: &[Limb]) -> usize { + x.last().map_or(0, |x| x.leading_zeros() as usize) + } + + /// Calculate the bit-length of the big-integer. + #[inline] + pub fn bit_length(x: &[Limb]) -> usize { + let bits = mem::size_of::<Limb>() * 8; + // Avoid overflowing, calculate via total number of bits + // minus leading zero bits. + let nlz = leading_zeros(x); + bits.checked_mul(x.len()) + .map_or_else(usize::max_value, |v| v - nlz) + } + + // SHL + + /// Shift-left bits inside a buffer. + /// + /// Assumes `n < Limb::BITS`, IE, internally shifting bits. + #[inline] + pub fn ishl_bits(x: &mut Vec<Limb>, n: usize) { + // Need to shift by the number of `bits % Limb::BITS)`. + let bits = mem::size_of::<Limb>() * 8; + debug_assert!(n < bits); + if n == 0 { + return; + } + + // Internally, for each item, we shift left by n, and add the previous + // right shifted limb-bits. + // For example, we transform (for u8) shifted left 2, to: + // b10100100 b01000010 + // b10 b10010001 b00001000 + let rshift = bits - n; + let lshift = n; + let mut prev: Limb = 0; + for xi in x.iter_mut() { + let tmp = *xi; + *xi <<= lshift; + *xi |= prev >> rshift; + prev = tmp; + } + + // Always push the carry, even if it creates a non-normal result. + let carry = prev >> rshift; + if carry != 0 { + x.push(carry); + } + } + + /// Shift-left `n` digits inside a buffer. + /// + /// Assumes `n` is not 0. + #[inline] + pub fn ishl_limbs(x: &mut Vec<Limb>, n: usize) { + debug_assert!(n != 0); + if !x.is_empty() { + x.reserve(n); + x.splice(..0, iter::repeat(0).take(n)); + } + } + + /// Shift-left buffer by n bits. + #[inline] + pub fn ishl(x: &mut Vec<Limb>, n: usize) { + let bits = mem::size_of::<Limb>() * 8; + // Need to pad with zeros for the number of `bits / Limb::BITS`, + // and shift-left with carry for `bits % Limb::BITS`. + let rem = n % bits; + let div = n / bits; + ishl_bits(x, rem); + if div != 0 { + ishl_limbs(x, div); + } + } + + // NORMALIZE + + /// Normalize the container by popping any leading zeros. + #[inline] + pub fn normalize(x: &mut Vec<Limb>) { + // Remove leading zero if we cause underflow. Since we're dividing + // by a small power, we have at max 1 int removed. + while x.last() == Some(&0) { + x.pop(); + } + } +} // small + +// LARGE +// ----- + +// Large-to-large operations, to modify a big integer from a native scalar. + +mod large { + use super::*; + + // RELATIVE OPERATORS + + /// Compare `x` to `y`, in little-endian order. + #[inline] + pub fn compare(x: &[Limb], y: &[Limb]) -> cmp::Ordering { + if x.len() > y.len() { + cmp::Ordering::Greater + } else if x.len() < y.len() { + cmp::Ordering::Less + } else { + let iter = x.iter().rev().zip(y.iter().rev()); + for (&xi, &yi) in iter { + if xi > yi { + return cmp::Ordering::Greater; + } else if xi < yi { + return cmp::Ordering::Less; + } + } + // Equal case. + cmp::Ordering::Equal + } + } + + /// Check if x is less than y. + #[inline] + pub fn less(x: &[Limb], y: &[Limb]) -> bool { + compare(x, y) == cmp::Ordering::Less + } + + /// Check if x is greater than or equal to y. + #[inline] + pub fn greater_equal(x: &[Limb], y: &[Limb]) -> bool { + !less(x, y) + } + + // ADDITION + + /// Implied AddAssign implementation for bigints. + /// + /// Allows us to choose a start-index in x to store, so we can avoid + /// padding the buffer with zeros when not needed, optimized for vectors. + pub fn iadd_impl(x: &mut Vec<Limb>, y: &[Limb], xstart: usize) { + // The effective x buffer is from `xstart..x.len()`, so we need to treat + // that as the current range. If the effective y buffer is longer, need + // to resize to that, + the start index. + if y.len() > x.len() - xstart { + x.resize(y.len() + xstart, 0); + } + + // Iteratively add elements from y to x. + let mut carry = false; + for (xi, yi) in x[xstart..].iter_mut().zip(y.iter()) { + // Only one op of the two can overflow, since we added at max + // Limb::max_value() + Limb::max_value(). Add the previous carry, + // and store the current carry for the next. + let mut tmp = scalar::iadd(xi, *yi); + if carry { + tmp |= scalar::iadd(xi, 1); + } + carry = tmp; + } + + // Overflow from the previous bit. + if carry { + small::iadd_impl(x, 1, y.len() + xstart); + } + } + + /// AddAssign bigint to bigint. + #[inline] + pub fn iadd(x: &mut Vec<Limb>, y: &[Limb]) { + iadd_impl(x, y, 0); + } + + /// Add bigint to bigint. + #[inline] + pub fn add(x: &[Limb], y: &[Limb]) -> Vec<Limb> { + let mut z = Vec::<Limb>::default(); + z.extend_from_slice(x); + iadd(&mut z, y); + z + } + + // SUBTRACTION + + /// SubAssign bigint to bigint. + pub fn isub(x: &mut Vec<Limb>, y: &[Limb]) { + // Basic underflow checks. + debug_assert!(greater_equal(x, y)); + + // Iteratively add elements from y to x. + let mut carry = false; + for (xi, yi) in x.iter_mut().zip(y.iter()) { + // Only one op of the two can overflow, since we added at max + // Limb::max_value() + Limb::max_value(). Add the previous carry, + // and store the current carry for the next. + let mut tmp = scalar::isub(xi, *yi); + if carry { + tmp |= scalar::isub(xi, 1); + } + carry = tmp; + } + + if carry { + small::isub_impl(x, 1, y.len()); + } else { + small::normalize(x); + } + } + + // MULTIPLICATION + + /// Number of digits to bottom-out to asymptotically slow algorithms. + /// + /// Karatsuba tends to out-perform long-multiplication at ~320-640 bits, + /// so we go halfway, while Newton division tends to out-perform + /// Algorithm D at ~1024 bits. We can toggle this for optimal performance. + pub const KARATSUBA_CUTOFF: usize = 32; + + /// Grade-school multiplication algorithm. + /// + /// Slow, naive algorithm, using limb-bit bases and just shifting left for + /// each iteration. This could be optimized with numerous other algorithms, + /// but it's extremely simple, and works in O(n*m) time, which is fine + /// by me. Each iteration, of which there are `m` iterations, requires + /// `n` multiplications, and `n` additions, or grade-school multiplication. + fn long_mul(x: &[Limb], y: &[Limb]) -> Vec<Limb> { + // Using the immutable value, multiply by all the scalars in y, using + // the algorithm defined above. Use a single buffer to avoid + // frequent reallocations. Handle the first case to avoid a redundant + // addition, since we know y.len() >= 1. + let mut z: Vec<Limb> = small::mul(x, y[0]); + z.resize(x.len() + y.len(), 0); + + // Handle the iterative cases. + for (i, &yi) in y[1..].iter().enumerate() { + let zi: Vec<Limb> = small::mul(x, yi); + iadd_impl(&mut z, &zi, i + 1); + } + + small::normalize(&mut z); + + z + } + + /// Split two buffers into halfway, into (lo, hi). + #[inline] + pub fn karatsuba_split(z: &[Limb], m: usize) -> (&[Limb], &[Limb]) { + (&z[..m], &z[m..]) + } + + /// Karatsuba multiplication algorithm with roughly equal input sizes. + /// + /// Assumes `y.len() >= x.len()`. + fn karatsuba_mul(x: &[Limb], y: &[Limb]) -> Vec<Limb> { + if y.len() <= KARATSUBA_CUTOFF { + // Bottom-out to long division for small cases. + long_mul(x, y) + } else if x.len() < y.len() / 2 { + karatsuba_uneven_mul(x, y) + } else { + // Do our 3 multiplications. + let m = y.len() / 2; + let (xl, xh) = karatsuba_split(x, m); + let (yl, yh) = karatsuba_split(y, m); + let sumx = add(xl, xh); + let sumy = add(yl, yh); + let z0 = karatsuba_mul(xl, yl); + let mut z1 = karatsuba_mul(&sumx, &sumy); + let z2 = karatsuba_mul(xh, yh); + // Properly scale z1, which is `z1 - z2 - zo`. + isub(&mut z1, &z2); + isub(&mut z1, &z0); + + // Create our result, which is equal to, in little-endian order: + // [z0, z1 - z2 - z0, z2] + // z1 must be shifted m digits (2^(32m)) over. + // z2 must be shifted 2*m digits (2^(64m)) over. + let len = z0.len().max(m + z1.len()).max(2 * m + z2.len()); + let mut result = z0; + result.reserve_exact(len - result.len()); + iadd_impl(&mut result, &z1, m); + iadd_impl(&mut result, &z2, 2 * m); + + result + } + } + + /// Karatsuba multiplication algorithm where y is substantially larger than x. + /// + /// Assumes `y.len() >= x.len()`. + fn karatsuba_uneven_mul(x: &[Limb], mut y: &[Limb]) -> Vec<Limb> { + let mut result = Vec::<Limb>::default(); + result.resize(x.len() + y.len(), 0); + + // This effectively is like grade-school multiplication between + // two numbers, except we're using splits on `y`, and the intermediate + // step is a Karatsuba multiplication. + let mut start = 0; + while !y.is_empty() { + let m = x.len().min(y.len()); + let (yl, yh) = karatsuba_split(y, m); + let prod = karatsuba_mul(x, yl); + iadd_impl(&mut result, &prod, start); + y = yh; + start += m; + } + small::normalize(&mut result); + + result + } + + /// Forwarder to the proper Karatsuba algorithm. + #[inline] + fn karatsuba_mul_fwd(x: &[Limb], y: &[Limb]) -> Vec<Limb> { + if x.len() < y.len() { + karatsuba_mul(x, y) + } else { + karatsuba_mul(y, x) + } + } + + /// MulAssign bigint to bigint. + #[inline] + pub fn imul(x: &mut Vec<Limb>, y: &[Limb]) { + if y.len() == 1 { + small::imul(x, y[0]); + } else { + // We're not really in a condition where using Karatsuba + // multiplication makes sense, so we're just going to use long + // division. ~20% speedup compared to: + // *x = karatsuba_mul_fwd(x, y); + *x = karatsuba_mul_fwd(x, y); + } + } +} // large + +// TRAITS +// ------ + +/// Traits for shared operations for big integers. +/// +/// None of these are implemented using normal traits, since these +/// are very expensive operations, and we want to deliberately +/// and explicitly use these functions. +pub(crate) trait Math: Clone + Sized + Default { + // DATA + + /// Get access to the underlying data + fn data(&self) -> &Vec<Limb>; + + /// Get access to the underlying data + fn data_mut(&mut self) -> &mut Vec<Limb>; + + // RELATIVE OPERATIONS + + /// Compare self to y. + #[inline] + fn compare(&self, y: &Self) -> cmp::Ordering { + large::compare(self.data(), y.data()) + } + + // PROPERTIES + + /// Get the high 64-bits from the bigint and if there are remaining bits. + #[inline] + fn hi64(&self) -> (u64, bool) { + self.data().as_slice().hi64() + } + + /// Calculate the bit-length of the big-integer. + /// Returns usize::max_value() if the value overflows, + /// IE, if `self.data().len() > usize::max_value() / 8`. + #[inline] + fn bit_length(&self) -> usize { + small::bit_length(self.data()) + } + + // INTEGER CONVERSIONS + + /// Create new big integer from u64. + #[inline] + fn from_u64(x: u64) -> Self { + let mut v = Self::default(); + let slc = split_u64(x); + v.data_mut().extend_from_slice(&slc); + v.normalize(); + v + } + + // NORMALIZE + + /// Normalize the integer, so any leading zero values are removed. + #[inline] + fn normalize(&mut self) { + small::normalize(self.data_mut()); + } + + // ADDITION + + /// AddAssign small integer. + #[inline] + fn iadd_small(&mut self, y: Limb) { + small::iadd(self.data_mut(), y); + } + + // MULTIPLICATION + + /// MulAssign small integer. + #[inline] + fn imul_small(&mut self, y: Limb) { + small::imul(self.data_mut(), y); + } + + /// Multiply by a power of 2. + #[inline] + fn imul_pow2(&mut self, n: u32) { + self.ishl(n as usize); + } + + /// Multiply by a power of 5. + #[inline] + fn imul_pow5(&mut self, n: u32) { + small::imul_pow5(self.data_mut(), n); + } + + /// MulAssign by a power of 10. + #[inline] + fn imul_pow10(&mut self, n: u32) { + self.imul_pow5(n); + self.imul_pow2(n); + } + + // SHIFTS + + /// Shift-left the entire buffer n bits. + #[inline] + fn ishl(&mut self, n: usize) { + small::ishl(self.data_mut(), n); + } +} diff --git a/third_party/rust/serde_json/src/lexical/mod.rs b/third_party/rust/serde_json/src/lexical/mod.rs new file mode 100644 index 0000000000..b1a45e218d --- /dev/null +++ b/third_party/rust/serde_json/src/lexical/mod.rs @@ -0,0 +1,38 @@ +// The code in this module is derived from the `lexical` crate by @Alexhuszagh +// which the author condensed into this minimal subset for use in serde_json. +// For the serde_json use case we care more about reliably round tripping all +// possible floating point values than about parsing any arbitrarily long string +// of digits with perfect accuracy, as the latter would take a high cost in +// compile time and performance. +// +// Dual licensed as MIT and Apache 2.0 just like the rest of serde_json, but +// copyright Alexander Huszagh. + +//! Fast, minimal float-parsing algorithm. + +// MODULES +pub(crate) mod algorithm; +mod bhcomp; +mod bignum; +mod cached; +mod cached_float80; +mod digit; +mod errors; +pub(crate) mod exponent; +pub(crate) mod float; +mod large_powers; +pub(crate) mod math; +pub(crate) mod num; +pub(crate) mod parse; +pub(crate) mod rounding; +mod shift; +mod small_powers; + +#[cfg(limb_width_32)] +mod large_powers32; + +#[cfg(limb_width_64)] +mod large_powers64; + +// API +pub use self::parse::{parse_concise_float, parse_truncated_float}; diff --git a/third_party/rust/serde_json/src/lexical/num.rs b/third_party/rust/serde_json/src/lexical/num.rs new file mode 100644 index 0000000000..e47e003419 --- /dev/null +++ b/third_party/rust/serde_json/src/lexical/num.rs @@ -0,0 +1,440 @@ +// Adapted from https://github.com/Alexhuszagh/rust-lexical. + +//! Utilities for Rust numbers. + +use core::ops; + +/// Precalculated values of radix**i for i in range [0, arr.len()-1]. +/// Each value can be **exactly** represented as that type. +const F32_POW10: [f32; 11] = [ + 1.0, + 10.0, + 100.0, + 1000.0, + 10000.0, + 100000.0, + 1000000.0, + 10000000.0, + 100000000.0, + 1000000000.0, + 10000000000.0, +]; + +/// Precalculated values of radix**i for i in range [0, arr.len()-1]. +/// Each value can be **exactly** represented as that type. +const F64_POW10: [f64; 23] = [ + 1.0, + 10.0, + 100.0, + 1000.0, + 10000.0, + 100000.0, + 1000000.0, + 10000000.0, + 100000000.0, + 1000000000.0, + 10000000000.0, + 100000000000.0, + 1000000000000.0, + 10000000000000.0, + 100000000000000.0, + 1000000000000000.0, + 10000000000000000.0, + 100000000000000000.0, + 1000000000000000000.0, + 10000000000000000000.0, + 100000000000000000000.0, + 1000000000000000000000.0, + 10000000000000000000000.0, +]; + +/// Type that can be converted to primitive with `as`. +pub trait AsPrimitive: Sized + Copy + PartialOrd { + fn as_u32(self) -> u32; + fn as_u64(self) -> u64; + fn as_u128(self) -> u128; + fn as_usize(self) -> usize; + fn as_f32(self) -> f32; + fn as_f64(self) -> f64; +} + +macro_rules! as_primitive_impl { + ($($ty:ident)*) => { + $( + impl AsPrimitive for $ty { + #[inline] + fn as_u32(self) -> u32 { + self as u32 + } + + #[inline] + fn as_u64(self) -> u64 { + self as u64 + } + + #[inline] + fn as_u128(self) -> u128 { + self as u128 + } + + #[inline] + fn as_usize(self) -> usize { + self as usize + } + + #[inline] + fn as_f32(self) -> f32 { + self as f32 + } + + #[inline] + fn as_f64(self) -> f64 { + self as f64 + } + } + )* + }; +} + +as_primitive_impl! { u32 u64 u128 usize f32 f64 } + +/// An interface for casting between machine scalars. +pub trait AsCast: AsPrimitive { + /// Creates a number from another value that can be converted into + /// a primitive via the `AsPrimitive` trait. + fn as_cast<N: AsPrimitive>(n: N) -> Self; +} + +macro_rules! as_cast_impl { + ($ty:ident, $method:ident) => { + impl AsCast for $ty { + #[inline] + fn as_cast<N: AsPrimitive>(n: N) -> Self { + n.$method() + } + } + }; +} + +as_cast_impl!(u32, as_u32); +as_cast_impl!(u64, as_u64); +as_cast_impl!(u128, as_u128); +as_cast_impl!(usize, as_usize); +as_cast_impl!(f32, as_f32); +as_cast_impl!(f64, as_f64); + +/// Numerical type trait. +pub trait Number: AsCast + ops::Add<Output = Self> {} + +macro_rules! number_impl { + ($($ty:ident)*) => { + $( + impl Number for $ty {} + )* + }; +} + +number_impl! { u32 u64 u128 usize f32 f64 } + +/// Defines a trait that supports integral operations. +pub trait Integer: Number + ops::BitAnd<Output = Self> + ops::Shr<i32, Output = Self> { + const ZERO: Self; +} + +macro_rules! integer_impl { + ($($ty:tt)*) => { + $( + impl Integer for $ty { + const ZERO: Self = 0; + } + )* + }; +} + +integer_impl! { u32 u64 u128 usize } + +/// Type trait for the mantissa type. +pub trait Mantissa: Integer { + /// Mask to extract the high bits from the integer. + const HIMASK: Self; + /// Mask to extract the low bits from the integer. + const LOMASK: Self; + /// Full size of the integer, in bits. + const FULL: i32; + /// Half size of the integer, in bits. + const HALF: i32 = Self::FULL / 2; +} + +impl Mantissa for u64 { + const HIMASK: u64 = 0xFFFFFFFF00000000; + const LOMASK: u64 = 0x00000000FFFFFFFF; + const FULL: i32 = 64; +} + +/// Get exact exponent limit for radix. +pub trait Float: Number { + /// Unsigned type of the same size. + type Unsigned: Integer; + + /// Literal zero. + const ZERO: Self; + /// Maximum number of digits that can contribute in the mantissa. + /// + /// We can exactly represent a float in radix `b` from radix 2 if + /// `b` is divisible by 2. This function calculates the exact number of + /// digits required to exactly represent that float. + /// + /// According to the "Handbook of Floating Point Arithmetic", + /// for IEEE754, with emin being the min exponent, p2 being the + /// precision, and b being the radix, the number of digits follows as: + /// + /// `−emin + p2 + ⌊(emin + 1) log(2, b) − log(1 − 2^(−p2), b)⌋` + /// + /// For f32, this follows as: + /// emin = -126 + /// p2 = 24 + /// + /// For f64, this follows as: + /// emin = -1022 + /// p2 = 53 + /// + /// In Python: + /// `-emin + p2 + math.floor((emin+1)*math.log(2, b) - math.log(1-2**(-p2), b))` + /// + /// This was used to calculate the maximum number of digits for [2, 36]. + const MAX_DIGITS: usize; + + // MASKS + + /// Bitmask for the sign bit. + const SIGN_MASK: Self::Unsigned; + /// Bitmask for the exponent, including the hidden bit. + const EXPONENT_MASK: Self::Unsigned; + /// Bitmask for the hidden bit in exponent, which is an implicit 1 in the fraction. + const HIDDEN_BIT_MASK: Self::Unsigned; + /// Bitmask for the mantissa (fraction), excluding the hidden bit. + const MANTISSA_MASK: Self::Unsigned; + + // PROPERTIES + + /// Positive infinity as bits. + const INFINITY_BITS: Self::Unsigned; + /// Positive infinity as bits. + const NEGATIVE_INFINITY_BITS: Self::Unsigned; + /// Size of the significand (mantissa) without hidden bit. + const MANTISSA_SIZE: i32; + /// Bias of the exponet + const EXPONENT_BIAS: i32; + /// Exponent portion of a denormal float. + const DENORMAL_EXPONENT: i32; + /// Maximum exponent value in float. + const MAX_EXPONENT: i32; + + // ROUNDING + + /// Default number of bits to shift (or 64 - mantissa size - 1). + const DEFAULT_SHIFT: i32; + /// Mask to determine if a full-carry occurred (1 in bit above hidden bit). + const CARRY_MASK: u64; + + /// Get min and max exponent limits (exact) from radix. + fn exponent_limit() -> (i32, i32); + + /// Get the number of digits that can be shifted from exponent to mantissa. + fn mantissa_limit() -> i32; + + // Re-exported methods from std. + fn pow10(self, n: i32) -> Self; + fn from_bits(u: Self::Unsigned) -> Self; + fn to_bits(self) -> Self::Unsigned; + fn is_sign_positive(self) -> bool; + fn is_sign_negative(self) -> bool; + + /// Returns true if the float is a denormal. + #[inline] + fn is_denormal(self) -> bool { + self.to_bits() & Self::EXPONENT_MASK == Self::Unsigned::ZERO + } + + /// Returns true if the float is a NaN or Infinite. + #[inline] + fn is_special(self) -> bool { + self.to_bits() & Self::EXPONENT_MASK == Self::EXPONENT_MASK + } + + /// Returns true if the float is infinite. + #[inline] + fn is_inf(self) -> bool { + self.is_special() && (self.to_bits() & Self::MANTISSA_MASK) == Self::Unsigned::ZERO + } + + /// Get exponent component from the float. + #[inline] + fn exponent(self) -> i32 { + if self.is_denormal() { + return Self::DENORMAL_EXPONENT; + } + + let bits = self.to_bits(); + let biased_e = ((bits & Self::EXPONENT_MASK) >> Self::MANTISSA_SIZE).as_u32(); + biased_e as i32 - Self::EXPONENT_BIAS + } + + /// Get mantissa (significand) component from float. + #[inline] + fn mantissa(self) -> Self::Unsigned { + let bits = self.to_bits(); + let s = bits & Self::MANTISSA_MASK; + if !self.is_denormal() { + s + Self::HIDDEN_BIT_MASK + } else { + s + } + } + + /// Get next greater float for a positive float. + /// Value must be >= 0.0 and < INFINITY. + #[inline] + fn next_positive(self) -> Self { + debug_assert!(self.is_sign_positive() && !self.is_inf()); + Self::from_bits(self.to_bits() + Self::Unsigned::as_cast(1u32)) + } + + /// Round a positive number to even. + #[inline] + fn round_positive_even(self) -> Self { + if self.mantissa() & Self::Unsigned::as_cast(1u32) == Self::Unsigned::as_cast(1u32) { + self.next_positive() + } else { + self + } + } +} + +impl Float for f32 { + type Unsigned = u32; + + const ZERO: f32 = 0.0; + const MAX_DIGITS: usize = 114; + const SIGN_MASK: u32 = 0x80000000; + const EXPONENT_MASK: u32 = 0x7F800000; + const HIDDEN_BIT_MASK: u32 = 0x00800000; + const MANTISSA_MASK: u32 = 0x007FFFFF; + const INFINITY_BITS: u32 = 0x7F800000; + const NEGATIVE_INFINITY_BITS: u32 = Self::INFINITY_BITS | Self::SIGN_MASK; + const MANTISSA_SIZE: i32 = 23; + const EXPONENT_BIAS: i32 = 127 + Self::MANTISSA_SIZE; + const DENORMAL_EXPONENT: i32 = 1 - Self::EXPONENT_BIAS; + const MAX_EXPONENT: i32 = 0xFF - Self::EXPONENT_BIAS; + const DEFAULT_SHIFT: i32 = u64::FULL - f32::MANTISSA_SIZE - 1; + const CARRY_MASK: u64 = 0x1000000; + + #[inline] + fn exponent_limit() -> (i32, i32) { + (-10, 10) + } + + #[inline] + fn mantissa_limit() -> i32 { + 7 + } + + #[inline] + fn pow10(self, n: i32) -> f32 { + // Check the exponent is within bounds in debug builds. + debug_assert!({ + let (min, max) = Self::exponent_limit(); + n >= min && n <= max + }); + + if n > 0 { + self * F32_POW10[n as usize] + } else { + self / F32_POW10[-n as usize] + } + } + + #[inline] + fn from_bits(u: u32) -> f32 { + f32::from_bits(u) + } + + #[inline] + fn to_bits(self) -> u32 { + f32::to_bits(self) + } + + #[inline] + fn is_sign_positive(self) -> bool { + f32::is_sign_positive(self) + } + + #[inline] + fn is_sign_negative(self) -> bool { + f32::is_sign_negative(self) + } +} + +impl Float for f64 { + type Unsigned = u64; + + const ZERO: f64 = 0.0; + const MAX_DIGITS: usize = 769; + const SIGN_MASK: u64 = 0x8000000000000000; + const EXPONENT_MASK: u64 = 0x7FF0000000000000; + const HIDDEN_BIT_MASK: u64 = 0x0010000000000000; + const MANTISSA_MASK: u64 = 0x000FFFFFFFFFFFFF; + const INFINITY_BITS: u64 = 0x7FF0000000000000; + const NEGATIVE_INFINITY_BITS: u64 = Self::INFINITY_BITS | Self::SIGN_MASK; + const MANTISSA_SIZE: i32 = 52; + const EXPONENT_BIAS: i32 = 1023 + Self::MANTISSA_SIZE; + const DENORMAL_EXPONENT: i32 = 1 - Self::EXPONENT_BIAS; + const MAX_EXPONENT: i32 = 0x7FF - Self::EXPONENT_BIAS; + const DEFAULT_SHIFT: i32 = u64::FULL - f64::MANTISSA_SIZE - 1; + const CARRY_MASK: u64 = 0x20000000000000; + + #[inline] + fn exponent_limit() -> (i32, i32) { + (-22, 22) + } + + #[inline] + fn mantissa_limit() -> i32 { + 15 + } + + #[inline] + fn pow10(self, n: i32) -> f64 { + // Check the exponent is within bounds in debug builds. + debug_assert!({ + let (min, max) = Self::exponent_limit(); + n >= min && n <= max + }); + + if n > 0 { + self * F64_POW10[n as usize] + } else { + self / F64_POW10[-n as usize] + } + } + + #[inline] + fn from_bits(u: u64) -> f64 { + f64::from_bits(u) + } + + #[inline] + fn to_bits(self) -> u64 { + f64::to_bits(self) + } + + #[inline] + fn is_sign_positive(self) -> bool { + f64::is_sign_positive(self) + } + + #[inline] + fn is_sign_negative(self) -> bool { + f64::is_sign_negative(self) + } +} diff --git a/third_party/rust/serde_json/src/lexical/parse.rs b/third_party/rust/serde_json/src/lexical/parse.rs new file mode 100644 index 0000000000..e3d7f1e871 --- /dev/null +++ b/third_party/rust/serde_json/src/lexical/parse.rs @@ -0,0 +1,83 @@ +// Adapted from https://github.com/Alexhuszagh/rust-lexical. + +use super::algorithm::*; +use super::bhcomp::*; +use super::digit::*; +use super::exponent::*; +use super::num::*; + +// PARSERS +// ------- + +/// Parse float for which the entire integer and fraction parts fit into a 64 +/// bit mantissa. +pub fn parse_concise_float<F>(mantissa: u64, mant_exp: i32) -> F +where + F: Float, +{ + if let Some(float) = fast_path(mantissa, mant_exp) { + return float; + } + + // Moderate path (use an extended 80-bit representation). + let truncated = false; + let (fp, valid) = moderate_path::<F>(mantissa, mant_exp, truncated); + if valid { + return fp.into_float::<F>(); + } + + let b = fp.into_downward_float::<F>(); + if b.is_special() { + // We have a non-finite number, we get to leave early. + return b; + } + + // Slow path, fast path didn't work. + let mut buffer = itoa::Buffer::new(); + let integer = buffer.format(mantissa).as_bytes(); + let fraction = &[]; + bhcomp(b, integer, fraction, mant_exp) +} + +/// Parse float from extracted float components. +/// +/// * `integer` - Slice containing the integer digits. +/// * `fraction` - Slice containing the fraction digits. +/// * `exponent` - Parsed, 32-bit exponent. +/// +/// Precondition: The integer must not have leading zeros. +pub fn parse_truncated_float<F>(integer: &[u8], mut fraction: &[u8], exponent: i32) -> F +where + F: Float, +{ + // Trim trailing zeroes from the fraction part. + while fraction.last() == Some(&b'0') { + fraction = &fraction[..fraction.len() - 1]; + } + + // Calculate the number of truncated digits. + let mut truncated = 0; + let mut mantissa: u64 = 0; + let mut iter = integer.iter().chain(fraction); + for &c in &mut iter { + mantissa = match add_digit(mantissa, to_digit(c).unwrap()) { + Some(v) => v, + None => { + truncated = 1 + iter.count(); + break; + } + }; + } + + let mant_exp = mantissa_exponent(exponent, fraction.len(), truncated); + let is_truncated = true; + + fallback_path( + integer, + fraction, + mantissa, + exponent, + mant_exp, + is_truncated, + ) +} diff --git a/third_party/rust/serde_json/src/lexical/rounding.rs b/third_party/rust/serde_json/src/lexical/rounding.rs new file mode 100644 index 0000000000..6ec1292aa5 --- /dev/null +++ b/third_party/rust/serde_json/src/lexical/rounding.rs @@ -0,0 +1,231 @@ +// Adapted from https://github.com/Alexhuszagh/rust-lexical. + +//! Defines rounding schemes for floating-point numbers. + +use super::float::ExtendedFloat; +use super::num::*; +use super::shift::*; +use core::mem; + +// MASKS + +/// Calculate a scalar factor of 2 above the halfway point. +#[inline] +pub(crate) fn nth_bit(n: u64) -> u64 { + let bits: u64 = mem::size_of::<u64>() as u64 * 8; + debug_assert!(n < bits, "nth_bit() overflow in shl."); + + 1 << n +} + +/// Generate a bitwise mask for the lower `n` bits. +#[inline] +pub(crate) fn lower_n_mask(n: u64) -> u64 { + let bits: u64 = mem::size_of::<u64>() as u64 * 8; + debug_assert!(n <= bits, "lower_n_mask() overflow in shl."); + + if n == bits { + u64::max_value() + } else { + (1 << n) - 1 + } +} + +/// Calculate the halfway point for the lower `n` bits. +#[inline] +pub(crate) fn lower_n_halfway(n: u64) -> u64 { + let bits: u64 = mem::size_of::<u64>() as u64 * 8; + debug_assert!(n <= bits, "lower_n_halfway() overflow in shl."); + + if n == 0 { + 0 + } else { + nth_bit(n - 1) + } +} + +/// Calculate a bitwise mask with `n` 1 bits starting at the `bit` position. +#[inline] +pub(crate) fn internal_n_mask(bit: u64, n: u64) -> u64 { + let bits: u64 = mem::size_of::<u64>() as u64 * 8; + debug_assert!(bit <= bits, "internal_n_halfway() overflow in shl."); + debug_assert!(n <= bits, "internal_n_halfway() overflow in shl."); + debug_assert!(bit >= n, "internal_n_halfway() overflow in sub."); + + lower_n_mask(bit) ^ lower_n_mask(bit - n) +} + +// NEAREST ROUNDING + +// Shift right N-bytes and round to the nearest. +// +// Return if we are above halfway and if we are halfway. +#[inline] +pub(crate) fn round_nearest(fp: &mut ExtendedFloat, shift: i32) -> (bool, bool) { + // Extract the truncated bits using mask. + // Calculate if the value of the truncated bits are either above + // the mid-way point, or equal to it. + // + // For example, for 4 truncated bytes, the mask would be b1111 + // and the midway point would be b1000. + let mask: u64 = lower_n_mask(shift as u64); + let halfway: u64 = lower_n_halfway(shift as u64); + + let truncated_bits = fp.mant & mask; + let is_above = truncated_bits > halfway; + let is_halfway = truncated_bits == halfway; + + // Bit shift so the leading bit is in the hidden bit. + overflowing_shr(fp, shift); + + (is_above, is_halfway) +} + +// Tie rounded floating point to event. +#[inline] +pub(crate) fn tie_even(fp: &mut ExtendedFloat, is_above: bool, is_halfway: bool) { + // Extract the last bit after shifting (and determine if it is odd). + let is_odd = fp.mant & 1 == 1; + + // Calculate if we need to roundup. + // We need to roundup if we are above halfway, or if we are odd + // and at half-way (need to tie-to-even). + if is_above || (is_odd && is_halfway) { + fp.mant += 1; + } +} + +// Shift right N-bytes and round nearest, tie-to-even. +// +// Floating-point arithmetic uses round to nearest, ties to even, +// which rounds to the nearest value, if the value is halfway in between, +// round to an even value. +#[inline] +pub(crate) fn round_nearest_tie_even(fp: &mut ExtendedFloat, shift: i32) { + let (is_above, is_halfway) = round_nearest(fp, shift); + tie_even(fp, is_above, is_halfway); +} + +// DIRECTED ROUNDING + +// Shift right N-bytes and round towards a direction. +// +// Return if we have any truncated bytes. +#[inline] +fn round_toward(fp: &mut ExtendedFloat, shift: i32) -> bool { + let mask: u64 = lower_n_mask(shift as u64); + let truncated_bits = fp.mant & mask; + + // Bit shift so the leading bit is in the hidden bit. + overflowing_shr(fp, shift); + + truncated_bits != 0 +} + +// Round down. +#[inline] +fn downard(_: &mut ExtendedFloat, _: bool) {} + +// Shift right N-bytes and round toward zero. +// +// Floating-point arithmetic defines round toward zero, which rounds +// towards positive zero. +#[inline] +pub(crate) fn round_downward(fp: &mut ExtendedFloat, shift: i32) { + // Bit shift so the leading bit is in the hidden bit. + // No rounding schemes, so we just ignore everything else. + let is_truncated = round_toward(fp, shift); + downard(fp, is_truncated); +} + +// ROUND TO FLOAT + +// Shift the ExtendedFloat fraction to the fraction bits in a native float. +// +// Floating-point arithmetic uses round to nearest, ties to even, +// which rounds to the nearest value, if the value is halfway in between, +// round to an even value. +#[inline] +pub(crate) fn round_to_float<F, Algorithm>(fp: &mut ExtendedFloat, algorithm: Algorithm) +where + F: Float, + Algorithm: FnOnce(&mut ExtendedFloat, i32), +{ + // Calculate the difference to allow a single calculation + // rather than a loop, to minimize the number of ops required. + // This does underflow detection. + let final_exp = fp.exp + F::DEFAULT_SHIFT; + if final_exp < F::DENORMAL_EXPONENT { + // We would end up with a denormal exponent, try to round to more + // digits. Only shift right if we can avoid zeroing out the value, + // which requires the exponent diff to be < M::BITS. The value + // is already normalized, so we shouldn't have any issue zeroing + // out the value. + let diff = F::DENORMAL_EXPONENT - fp.exp; + if diff <= u64::FULL { + // We can avoid underflow, can get a valid representation. + algorithm(fp, diff); + } else { + // Certain underflow, assign literal 0s. + fp.mant = 0; + fp.exp = 0; + } + } else { + algorithm(fp, F::DEFAULT_SHIFT); + } + + if fp.mant & F::CARRY_MASK == F::CARRY_MASK { + // Roundup carried over to 1 past the hidden bit. + shr(fp, 1); + } +} + +// AVOID OVERFLOW/UNDERFLOW + +// Avoid overflow for large values, shift left as needed. +// +// Shift until a 1-bit is in the hidden bit, if the mantissa is not 0. +#[inline] +pub(crate) fn avoid_overflow<F>(fp: &mut ExtendedFloat) +where + F: Float, +{ + // Calculate the difference to allow a single calculation + // rather than a loop, minimizing the number of ops required. + if fp.exp >= F::MAX_EXPONENT { + let diff = fp.exp - F::MAX_EXPONENT; + if diff <= F::MANTISSA_SIZE { + // Our overflow mask needs to start at the hidden bit, or at + // `F::MANTISSA_SIZE+1`, and needs to have `diff+1` bits set, + // to see if our value overflows. + let bit = (F::MANTISSA_SIZE + 1) as u64; + let n = (diff + 1) as u64; + let mask = internal_n_mask(bit, n); + if (fp.mant & mask) == 0 { + // If we have no 1-bit in the hidden-bit position, + // which is index 0, we need to shift 1. + let shift = diff + 1; + shl(fp, shift); + } + } + } +} + +// ROUND TO NATIVE + +// Round an extended-precision float to a native float representation. +#[inline] +pub(crate) fn round_to_native<F, Algorithm>(fp: &mut ExtendedFloat, algorithm: Algorithm) +where + F: Float, + Algorithm: FnOnce(&mut ExtendedFloat, i32), +{ + // Shift all the way left, to ensure a consistent representation. + // The following right-shifts do not work for a non-normalized number. + fp.normalize(); + + // Round so the fraction is in a native mantissa representation, + // and avoid overflow/underflow. + round_to_float::<F, _>(fp, algorithm); + avoid_overflow::<F>(fp); +} diff --git a/third_party/rust/serde_json/src/lexical/shift.rs b/third_party/rust/serde_json/src/lexical/shift.rs new file mode 100644 index 0000000000..a0bae01e0f --- /dev/null +++ b/third_party/rust/serde_json/src/lexical/shift.rs @@ -0,0 +1,46 @@ +// Adapted from https://github.com/Alexhuszagh/rust-lexical. + +//! Bit-shift helpers. + +use super::float::ExtendedFloat; +use core::mem; + +// Shift extended-precision float right `shift` bytes. +#[inline] +pub(crate) fn shr(fp: &mut ExtendedFloat, shift: i32) { + let bits: u64 = mem::size_of::<u64>() as u64 * 8; + debug_assert!((shift as u64) < bits, "shr() overflow in shift right."); + + fp.mant >>= shift; + fp.exp += shift; +} + +// Shift extended-precision float right `shift` bytes. +// +// Accepts when the shift is the same as the type size, and +// sets the value to 0. +#[inline] +pub(crate) fn overflowing_shr(fp: &mut ExtendedFloat, shift: i32) { + let bits: u64 = mem::size_of::<u64>() as u64 * 8; + debug_assert!( + (shift as u64) <= bits, + "overflowing_shr() overflow in shift right." + ); + + fp.mant = if shift as u64 == bits { + 0 + } else { + fp.mant >> shift + }; + fp.exp += shift; +} + +// Shift extended-precision float left `shift` bytes. +#[inline] +pub(crate) fn shl(fp: &mut ExtendedFloat, shift: i32) { + let bits: u64 = mem::size_of::<u64>() as u64 * 8; + debug_assert!((shift as u64) < bits, "shl() overflow in shift left."); + + fp.mant <<= shift; + fp.exp -= shift; +} diff --git a/third_party/rust/serde_json/src/lexical/small_powers.rs b/third_party/rust/serde_json/src/lexical/small_powers.rs new file mode 100644 index 0000000000..219d826116 --- /dev/null +++ b/third_party/rust/serde_json/src/lexical/small_powers.rs @@ -0,0 +1,70 @@ +// Adapted from https://github.com/Alexhuszagh/rust-lexical. + +//! Pre-computed small powers. + +// 32 BIT +#[cfg(limb_width_32)] +pub(crate) const POW5_32: [u32; 14] = [ + 1, 5, 25, 125, 625, 3125, 15625, 78125, 390625, 1953125, 9765625, 48828125, 244140625, + 1220703125, +]; + +#[cfg(limb_width_32)] +pub(crate) const POW10_32: [u32; 10] = [ + 1, 10, 100, 1000, 10000, 100000, 1000000, 10000000, 100000000, 1000000000, +]; + +// 64 BIT +#[cfg(limb_width_64)] +pub(crate) const POW5_64: [u64; 28] = [ + 1, + 5, + 25, + 125, + 625, + 3125, + 15625, + 78125, + 390625, + 1953125, + 9765625, + 48828125, + 244140625, + 1220703125, + 6103515625, + 30517578125, + 152587890625, + 762939453125, + 3814697265625, + 19073486328125, + 95367431640625, + 476837158203125, + 2384185791015625, + 11920928955078125, + 59604644775390625, + 298023223876953125, + 1490116119384765625, + 7450580596923828125, +]; +pub(crate) const POW10_64: [u64; 20] = [ + 1, + 10, + 100, + 1000, + 10000, + 100000, + 1000000, + 10000000, + 100000000, + 1000000000, + 10000000000, + 100000000000, + 1000000000000, + 10000000000000, + 100000000000000, + 1000000000000000, + 10000000000000000, + 100000000000000000, + 1000000000000000000, + 10000000000000000000, +]; diff --git a/third_party/rust/serde_json/src/lib.rs b/third_party/rust/serde_json/src/lib.rs new file mode 100644 index 0000000000..48d0fe2191 --- /dev/null +++ b/third_party/rust/serde_json/src/lib.rs @@ -0,0 +1,421 @@ +//! # Serde JSON +//! +//! JSON is a ubiquitous open-standard format that uses human-readable text to +//! transmit data objects consisting of key-value pairs. +//! +//! ```json +//! { +//! "name": "John Doe", +//! "age": 43, +//! "address": { +//! "street": "10 Downing Street", +//! "city": "London" +//! }, +//! "phones": [ +//! "+44 1234567", +//! "+44 2345678" +//! ] +//! } +//! ``` +//! +//! There are three common ways that you might find yourself needing to work +//! with JSON data in Rust. +//! +//! - **As text data.** An unprocessed string of JSON data that you receive on +//! an HTTP endpoint, read from a file, or prepare to send to a remote +//! server. +//! - **As an untyped or loosely typed representation.** Maybe you want to +//! check that some JSON data is valid before passing it on, but without +//! knowing the structure of what it contains. Or you want to do very basic +//! manipulations like insert a key in a particular spot. +//! - **As a strongly typed Rust data structure.** When you expect all or most +//! of your data to conform to a particular structure and want to get real +//! work done without JSON's loosey-goosey nature tripping you up. +//! +//! Serde JSON provides efficient, flexible, safe ways of converting data +//! between each of these representations. +//! +//! # Operating on untyped JSON values +//! +//! Any valid JSON data can be manipulated in the following recursive enum +//! representation. This data structure is [`serde_json::Value`][value]. +//! +//! ``` +//! # use serde_json::{Number, Map}; +//! # +//! # #[allow(dead_code)] +//! enum Value { +//! Null, +//! Bool(bool), +//! Number(Number), +//! String(String), +//! Array(Vec<Value>), +//! Object(Map<String, Value>), +//! } +//! ``` +//! +//! A string of JSON data can be parsed into a `serde_json::Value` by the +//! [`serde_json::from_str`][from_str] function. There is also +//! [`from_slice`][from_slice] for parsing from a byte slice &[u8] and +//! [`from_reader`][from_reader] for parsing from any `io::Read` like a File or +//! a TCP stream. +//! +//! ``` +//! use serde_json::{Result, Value}; +//! +//! fn untyped_example() -> Result<()> { +//! // Some JSON input data as a &str. Maybe this comes from the user. +//! let data = r#" +//! { +//! "name": "John Doe", +//! "age": 43, +//! "phones": [ +//! "+44 1234567", +//! "+44 2345678" +//! ] +//! }"#; +//! +//! // Parse the string of data into serde_json::Value. +//! let v: Value = serde_json::from_str(data)?; +//! +//! // Access parts of the data by indexing with square brackets. +//! println!("Please call {} at the number {}", v["name"], v["phones"][0]); +//! +//! Ok(()) +//! } +//! # +//! # fn main() { +//! # untyped_example().unwrap(); +//! # } +//! ``` +//! +//! The result of square bracket indexing like `v["name"]` is a borrow of the +//! data at that index, so the type is `&Value`. A JSON map can be indexed with +//! string keys, while a JSON array can be indexed with integer keys. If the +//! type of the data is not right for the type with which it is being indexed, +//! or if a map does not contain the key being indexed, or if the index into a +//! vector is out of bounds, the returned element is `Value::Null`. +//! +//! When a `Value` is printed, it is printed as a JSON string. So in the code +//! above, the output looks like `Please call "John Doe" at the number "+44 +//! 1234567"`. The quotation marks appear because `v["name"]` is a `&Value` +//! containing a JSON string and its JSON representation is `"John Doe"`. +//! Printing as a plain string without quotation marks involves converting from +//! a JSON string to a Rust string with [`as_str()`] or avoiding the use of +//! `Value` as described in the following section. +//! +//! [`as_str()`]: crate::Value::as_str +//! +//! The `Value` representation is sufficient for very basic tasks but can be +//! tedious to work with for anything more significant. Error handling is +//! verbose to implement correctly, for example imagine trying to detect the +//! presence of unrecognized fields in the input data. The compiler is powerless +//! to help you when you make a mistake, for example imagine typoing `v["name"]` +//! as `v["nmae"]` in one of the dozens of places it is used in your code. +//! +//! # Parsing JSON as strongly typed data structures +//! +//! Serde provides a powerful way of mapping JSON data into Rust data structures +//! largely automatically. +//! +//! ``` +//! use serde::{Deserialize, Serialize}; +//! use serde_json::Result; +//! +//! #[derive(Serialize, Deserialize)] +//! struct Person { +//! name: String, +//! age: u8, +//! phones: Vec<String>, +//! } +//! +//! fn typed_example() -> Result<()> { +//! // Some JSON input data as a &str. Maybe this comes from the user. +//! let data = r#" +//! { +//! "name": "John Doe", +//! "age": 43, +//! "phones": [ +//! "+44 1234567", +//! "+44 2345678" +//! ] +//! }"#; +//! +//! // Parse the string of data into a Person object. This is exactly the +//! // same function as the one that produced serde_json::Value above, but +//! // now we are asking it for a Person as output. +//! let p: Person = serde_json::from_str(data)?; +//! +//! // Do things just like with any other Rust data structure. +//! println!("Please call {} at the number {}", p.name, p.phones[0]); +//! +//! Ok(()) +//! } +//! # +//! # fn main() { +//! # typed_example().unwrap(); +//! # } +//! ``` +//! +//! This is the same `serde_json::from_str` function as before, but this time we +//! assign the return value to a variable of type `Person` so Serde will +//! automatically interpret the input data as a `Person` and produce informative +//! error messages if the layout does not conform to what a `Person` is expected +//! to look like. +//! +//! Any type that implements Serde's `Deserialize` trait can be deserialized +//! this way. This includes built-in Rust standard library types like `Vec<T>` +//! and `HashMap<K, V>`, as well as any structs or enums annotated with +//! `#[derive(Deserialize)]`. +//! +//! Once we have `p` of type `Person`, our IDE and the Rust compiler can help us +//! use it correctly like they do for any other Rust code. The IDE can +//! autocomplete field names to prevent typos, which was impossible in the +//! `serde_json::Value` representation. And the Rust compiler can check that +//! when we write `p.phones[0]`, then `p.phones` is guaranteed to be a +//! `Vec<String>` so indexing into it makes sense and produces a `String`. +//! +//! # Constructing JSON values +//! +//! Serde JSON provides a [`json!` macro][macro] to build `serde_json::Value` +//! objects with very natural JSON syntax. +//! +//! ``` +//! use serde_json::json; +//! +//! fn main() { +//! // The type of `john` is `serde_json::Value` +//! let john = json!({ +//! "name": "John Doe", +//! "age": 43, +//! "phones": [ +//! "+44 1234567", +//! "+44 2345678" +//! ] +//! }); +//! +//! println!("first phone number: {}", john["phones"][0]); +//! +//! // Convert to a string of JSON and print it out +//! println!("{}", john.to_string()); +//! } +//! ``` +//! +//! The `Value::to_string()` function converts a `serde_json::Value` into a +//! `String` of JSON text. +//! +//! One neat thing about the `json!` macro is that variables and expressions can +//! be interpolated directly into the JSON value as you are building it. Serde +//! will check at compile time that the value you are interpolating is able to +//! be represented as JSON. +//! +//! ``` +//! # use serde_json::json; +//! # +//! # fn random_phone() -> u16 { 0 } +//! # +//! let full_name = "John Doe"; +//! let age_last_year = 42; +//! +//! // The type of `john` is `serde_json::Value` +//! let john = json!({ +//! "name": full_name, +//! "age": age_last_year + 1, +//! "phones": [ +//! format!("+44 {}", random_phone()) +//! ] +//! }); +//! ``` +//! +//! This is amazingly convenient, but we have the problem we had before with +//! `Value`: the IDE and Rust compiler cannot help us if we get it wrong. Serde +//! JSON provides a better way of serializing strongly-typed data structures +//! into JSON text. +//! +//! # Creating JSON by serializing data structures +//! +//! A data structure can be converted to a JSON string by +//! [`serde_json::to_string`][to_string]. There is also +//! [`serde_json::to_vec`][to_vec] which serializes to a `Vec<u8>` and +//! [`serde_json::to_writer`][to_writer] which serializes to any `io::Write` +//! such as a File or a TCP stream. +//! +//! ``` +//! use serde::{Deserialize, Serialize}; +//! use serde_json::Result; +//! +//! #[derive(Serialize, Deserialize)] +//! struct Address { +//! street: String, +//! city: String, +//! } +//! +//! fn print_an_address() -> Result<()> { +//! // Some data structure. +//! let address = Address { +//! street: "10 Downing Street".to_owned(), +//! city: "London".to_owned(), +//! }; +//! +//! // Serialize it to a JSON string. +//! let j = serde_json::to_string(&address)?; +//! +//! // Print, write to a file, or send to an HTTP server. +//! println!("{}", j); +//! +//! Ok(()) +//! } +//! # +//! # fn main() { +//! # print_an_address().unwrap(); +//! # } +//! ``` +//! +//! Any type that implements Serde's `Serialize` trait can be serialized this +//! way. This includes built-in Rust standard library types like `Vec<T>` and +//! `HashMap<K, V>`, as well as any structs or enums annotated with +//! `#[derive(Serialize)]`. +//! +//! # No-std support +//! +//! As long as there is a memory allocator, it is possible to use serde_json +//! without the rest of the Rust standard library. Disable the default "std" +//! feature and enable the "alloc" feature: +//! +//! ```toml +//! [dependencies] +//! serde_json = { version = "1.0", default-features = false, features = ["alloc"] } +//! ``` +//! +//! For JSON support in Serde without a memory allocator, please see the +//! [`serde-json-core`] crate. +//! +//! [value]: crate::value::Value +//! [from_str]: crate::de::from_str +//! [from_slice]: crate::de::from_slice +//! [from_reader]: crate::de::from_reader +//! [to_string]: crate::ser::to_string +//! [to_vec]: crate::ser::to_vec +//! [to_writer]: crate::ser::to_writer +//! [macro]: crate::json +//! [`serde-json-core`]: https://github.com/rust-embedded-community/serde-json-core + +#![doc(html_root_url = "https://docs.rs/serde_json/1.0.93")] +// Ignored clippy lints +#![allow( + clippy::collapsible_else_if, + clippy::comparison_chain, + clippy::deprecated_cfg_attr, + clippy::doc_markdown, + clippy::excessive_precision, + clippy::explicit_auto_deref, + clippy::float_cmp, + clippy::manual_range_contains, + clippy::match_like_matches_macro, + clippy::match_single_binding, + clippy::needless_doctest_main, + clippy::needless_late_init, + // clippy bug: https://github.com/rust-lang/rust-clippy/issues/8366 + clippy::ptr_arg, + clippy::return_self_not_must_use, + clippy::transmute_ptr_to_ptr, + clippy::unnecessary_wraps, + // clippy bug: https://github.com/rust-lang/rust-clippy/issues/5704 + clippy::unnested_or_patterns, +)] +// Ignored clippy_pedantic lints +#![allow( + // buggy + clippy::iter_not_returning_iterator, // https://github.com/rust-lang/rust-clippy/issues/8285 + // Deserializer::from_str, into_iter + clippy::should_implement_trait, + // integer and float ser/de requires these sorts of casts + clippy::cast_possible_truncation, + clippy::cast_possible_wrap, + clippy::cast_precision_loss, + clippy::cast_sign_loss, + // correctly used + clippy::enum_glob_use, + clippy::if_not_else, + clippy::integer_division, + clippy::map_err_ignore, + clippy::match_same_arms, + clippy::similar_names, + clippy::unused_self, + clippy::wildcard_imports, + // things are often more readable this way + clippy::cast_lossless, + clippy::module_name_repetitions, + clippy::redundant_else, + clippy::shadow_unrelated, + clippy::single_match_else, + clippy::too_many_lines, + clippy::unreadable_literal, + clippy::unseparated_literal_suffix, + clippy::use_self, + clippy::zero_prefixed_literal, + // we support older compilers + clippy::checked_conversions, + clippy::mem_replace_with_default, + // noisy + clippy::missing_errors_doc, + clippy::must_use_candidate, +)] +#![allow(non_upper_case_globals)] +#![deny(missing_docs)] +#![cfg_attr(not(feature = "std"), no_std)] +#![cfg_attr(docsrs, feature(doc_cfg))] + +extern crate alloc; + +#[cfg(feature = "std")] +#[doc(inline)] +pub use crate::de::from_reader; +#[doc(inline)] +pub use crate::de::{from_slice, from_str, Deserializer, StreamDeserializer}; +#[doc(inline)] +pub use crate::error::{Error, Result}; +#[doc(inline)] +pub use crate::ser::{to_string, to_string_pretty, to_vec, to_vec_pretty}; +#[cfg(feature = "std")] +#[doc(inline)] +pub use crate::ser::{to_writer, to_writer_pretty, Serializer}; +#[doc(inline)] +pub use crate::value::{from_value, to_value, Map, Number, Value}; + +// We only use our own error type; no need for From conversions provided by the +// standard library's try! macro. This reduces lines of LLVM IR by 4%. +macro_rules! tri { + ($e:expr $(,)?) => { + match $e { + core::result::Result::Ok(val) => val, + core::result::Result::Err(err) => return core::result::Result::Err(err), + } + }; +} + +#[macro_use] +mod macros; + +pub mod de; +pub mod error; +pub mod map; +#[cfg(feature = "std")] +#[cfg_attr(docsrs, doc(cfg(feature = "std")))] +pub mod ser; +#[cfg(not(feature = "std"))] +mod ser; +pub mod value; + +mod features_check; + +mod io; +#[cfg(feature = "std")] +mod iter; +#[cfg(feature = "float_roundtrip")] +mod lexical; +mod number; +mod read; + +#[cfg(feature = "raw_value")] +mod raw; diff --git a/third_party/rust/serde_json/src/macros.rs b/third_party/rust/serde_json/src/macros.rs new file mode 100644 index 0000000000..5287998b46 --- /dev/null +++ b/third_party/rust/serde_json/src/macros.rs @@ -0,0 +1,303 @@ +/// Construct a `serde_json::Value` from a JSON literal. +/// +/// ``` +/// # use serde_json::json; +/// # +/// let value = json!({ +/// "code": 200, +/// "success": true, +/// "payload": { +/// "features": [ +/// "serde", +/// "json" +/// ] +/// } +/// }); +/// ``` +/// +/// Variables or expressions can be interpolated into the JSON literal. Any type +/// interpolated into an array element or object value must implement Serde's +/// `Serialize` trait, while any type interpolated into a object key must +/// implement `Into<String>`. If the `Serialize` implementation of the +/// interpolated type decides to fail, or if the interpolated type contains a +/// map with non-string keys, the `json!` macro will panic. +/// +/// ``` +/// # use serde_json::json; +/// # +/// let code = 200; +/// let features = vec!["serde", "json"]; +/// +/// let value = json!({ +/// "code": code, +/// "success": code == 200, +/// "payload": { +/// features[0]: features[1] +/// } +/// }); +/// ``` +/// +/// Trailing commas are allowed inside both arrays and objects. +/// +/// ``` +/// # use serde_json::json; +/// # +/// let value = json!([ +/// "notice", +/// "the", +/// "trailing", +/// "comma -->", +/// ]); +/// ``` +#[macro_export(local_inner_macros)] +macro_rules! json { + // Hide distracting implementation details from the generated rustdoc. + ($($json:tt)+) => { + json_internal!($($json)+) + }; +} + +// Rocket relies on this because they export their own `json!` with a different +// doc comment than ours, and various Rust bugs prevent them from calling our +// `json!` from their `json!` so they call `json_internal!` directly. Check with +// @SergioBenitez before making breaking changes to this macro. +// +// Changes are fine as long as `json_internal!` does not call any new helper +// macros and can still be invoked as `json_internal!($($json)+)`. +#[macro_export(local_inner_macros)] +#[doc(hidden)] +macro_rules! json_internal { + ////////////////////////////////////////////////////////////////////////// + // TT muncher for parsing the inside of an array [...]. Produces a vec![...] + // of the elements. + // + // Must be invoked as: json_internal!(@array [] $($tt)*) + ////////////////////////////////////////////////////////////////////////// + + // Done with trailing comma. + (@array [$($elems:expr,)*]) => { + json_internal_vec![$($elems,)*] + }; + + // Done without trailing comma. + (@array [$($elems:expr),*]) => { + json_internal_vec![$($elems),*] + }; + + // Next element is `null`. + (@array [$($elems:expr,)*] null $($rest:tt)*) => { + json_internal!(@array [$($elems,)* json_internal!(null)] $($rest)*) + }; + + // Next element is `true`. + (@array [$($elems:expr,)*] true $($rest:tt)*) => { + json_internal!(@array [$($elems,)* json_internal!(true)] $($rest)*) + }; + + // Next element is `false`. + (@array [$($elems:expr,)*] false $($rest:tt)*) => { + json_internal!(@array [$($elems,)* json_internal!(false)] $($rest)*) + }; + + // Next element is an array. + (@array [$($elems:expr,)*] [$($array:tt)*] $($rest:tt)*) => { + json_internal!(@array [$($elems,)* json_internal!([$($array)*])] $($rest)*) + }; + + // Next element is a map. + (@array [$($elems:expr,)*] {$($map:tt)*} $($rest:tt)*) => { + json_internal!(@array [$($elems,)* json_internal!({$($map)*})] $($rest)*) + }; + + // Next element is an expression followed by comma. + (@array [$($elems:expr,)*] $next:expr, $($rest:tt)*) => { + json_internal!(@array [$($elems,)* json_internal!($next),] $($rest)*) + }; + + // Last element is an expression with no trailing comma. + (@array [$($elems:expr,)*] $last:expr) => { + json_internal!(@array [$($elems,)* json_internal!($last)]) + }; + + // Comma after the most recent element. + (@array [$($elems:expr),*] , $($rest:tt)*) => { + json_internal!(@array [$($elems,)*] $($rest)*) + }; + + // Unexpected token after most recent element. + (@array [$($elems:expr),*] $unexpected:tt $($rest:tt)*) => { + json_unexpected!($unexpected) + }; + + ////////////////////////////////////////////////////////////////////////// + // TT muncher for parsing the inside of an object {...}. Each entry is + // inserted into the given map variable. + // + // Must be invoked as: json_internal!(@object $map () ($($tt)*) ($($tt)*)) + // + // We require two copies of the input tokens so that we can match on one + // copy and trigger errors on the other copy. + ////////////////////////////////////////////////////////////////////////// + + // Done. + (@object $object:ident () () ()) => {}; + + // Insert the current entry followed by trailing comma. + (@object $object:ident [$($key:tt)+] ($value:expr) , $($rest:tt)*) => { + let _ = $object.insert(($($key)+).into(), $value); + json_internal!(@object $object () ($($rest)*) ($($rest)*)); + }; + + // Current entry followed by unexpected token. + (@object $object:ident [$($key:tt)+] ($value:expr) $unexpected:tt $($rest:tt)*) => { + json_unexpected!($unexpected); + }; + + // Insert the last entry without trailing comma. + (@object $object:ident [$($key:tt)+] ($value:expr)) => { + let _ = $object.insert(($($key)+).into(), $value); + }; + + // Next value is `null`. + (@object $object:ident ($($key:tt)+) (: null $($rest:tt)*) $copy:tt) => { + json_internal!(@object $object [$($key)+] (json_internal!(null)) $($rest)*); + }; + + // Next value is `true`. + (@object $object:ident ($($key:tt)+) (: true $($rest:tt)*) $copy:tt) => { + json_internal!(@object $object [$($key)+] (json_internal!(true)) $($rest)*); + }; + + // Next value is `false`. + (@object $object:ident ($($key:tt)+) (: false $($rest:tt)*) $copy:tt) => { + json_internal!(@object $object [$($key)+] (json_internal!(false)) $($rest)*); + }; + + // Next value is an array. + (@object $object:ident ($($key:tt)+) (: [$($array:tt)*] $($rest:tt)*) $copy:tt) => { + json_internal!(@object $object [$($key)+] (json_internal!([$($array)*])) $($rest)*); + }; + + // Next value is a map. + (@object $object:ident ($($key:tt)+) (: {$($map:tt)*} $($rest:tt)*) $copy:tt) => { + json_internal!(@object $object [$($key)+] (json_internal!({$($map)*})) $($rest)*); + }; + + // Next value is an expression followed by comma. + (@object $object:ident ($($key:tt)+) (: $value:expr , $($rest:tt)*) $copy:tt) => { + json_internal!(@object $object [$($key)+] (json_internal!($value)) , $($rest)*); + }; + + // Last value is an expression with no trailing comma. + (@object $object:ident ($($key:tt)+) (: $value:expr) $copy:tt) => { + json_internal!(@object $object [$($key)+] (json_internal!($value))); + }; + + // Missing value for last entry. Trigger a reasonable error message. + (@object $object:ident ($($key:tt)+) (:) $copy:tt) => { + // "unexpected end of macro invocation" + json_internal!(); + }; + + // Missing colon and value for last entry. Trigger a reasonable error + // message. + (@object $object:ident ($($key:tt)+) () $copy:tt) => { + // "unexpected end of macro invocation" + json_internal!(); + }; + + // Misplaced colon. Trigger a reasonable error message. + (@object $object:ident () (: $($rest:tt)*) ($colon:tt $($copy:tt)*)) => { + // Takes no arguments so "no rules expected the token `:`". + json_unexpected!($colon); + }; + + // Found a comma inside a key. Trigger a reasonable error message. + (@object $object:ident ($($key:tt)*) (, $($rest:tt)*) ($comma:tt $($copy:tt)*)) => { + // Takes no arguments so "no rules expected the token `,`". + json_unexpected!($comma); + }; + + // Key is fully parenthesized. This avoids clippy double_parens false + // positives because the parenthesization may be necessary here. + (@object $object:ident () (($key:expr) : $($rest:tt)*) $copy:tt) => { + json_internal!(@object $object ($key) (: $($rest)*) (: $($rest)*)); + }; + + // Refuse to absorb colon token into key expression. + (@object $object:ident ($($key:tt)*) (: $($unexpected:tt)+) $copy:tt) => { + json_expect_expr_comma!($($unexpected)+); + }; + + // Munch a token into the current key. + (@object $object:ident ($($key:tt)*) ($tt:tt $($rest:tt)*) $copy:tt) => { + json_internal!(@object $object ($($key)* $tt) ($($rest)*) ($($rest)*)); + }; + + ////////////////////////////////////////////////////////////////////////// + // The main implementation. + // + // Must be invoked as: json_internal!($($json)+) + ////////////////////////////////////////////////////////////////////////// + + (null) => { + $crate::Value::Null + }; + + (true) => { + $crate::Value::Bool(true) + }; + + (false) => { + $crate::Value::Bool(false) + }; + + ([]) => { + $crate::Value::Array(json_internal_vec![]) + }; + + ([ $($tt:tt)+ ]) => { + $crate::Value::Array(json_internal!(@array [] $($tt)+)) + }; + + ({}) => { + $crate::Value::Object($crate::Map::new()) + }; + + ({ $($tt:tt)+ }) => { + $crate::Value::Object({ + let mut object = $crate::Map::new(); + json_internal!(@object object () ($($tt)+) ($($tt)+)); + object + }) + }; + + // Any Serialize type: numbers, strings, struct literals, variables etc. + // Must be below every other rule. + ($other:expr) => { + $crate::to_value(&$other).unwrap() + }; +} + +// The json_internal macro above cannot invoke vec directly because it uses +// local_inner_macros. A vec invocation there would resolve to $crate::vec. +// Instead invoke vec here outside of local_inner_macros. +#[macro_export] +#[doc(hidden)] +macro_rules! json_internal_vec { + ($($content:tt)*) => { + vec![$($content)*] + }; +} + +#[macro_export] +#[doc(hidden)] +macro_rules! json_unexpected { + () => {}; +} + +#[macro_export] +#[doc(hidden)] +macro_rules! json_expect_expr_comma { + ($e:expr , $($tt:tt)*) => {}; +} diff --git a/third_party/rust/serde_json/src/map.rs b/third_party/rust/serde_json/src/map.rs new file mode 100644 index 0000000000..3e8a3814c9 --- /dev/null +++ b/third_party/rust/serde_json/src/map.rs @@ -0,0 +1,940 @@ +//! A map of String to serde_json::Value. +//! +//! By default the map is backed by a [`BTreeMap`]. Enable the `preserve_order` +//! feature of serde_json to use [`IndexMap`] instead. +//! +//! [`BTreeMap`]: https://doc.rust-lang.org/std/collections/struct.BTreeMap.html +//! [`IndexMap`]: https://docs.rs/indexmap/*/indexmap/map/struct.IndexMap.html + +use crate::value::Value; +use alloc::string::String; +use core::borrow::Borrow; +use core::fmt::{self, Debug}; +use core::hash::Hash; +use core::iter::{FromIterator, FusedIterator}; +#[cfg(feature = "preserve_order")] +use core::mem; +use core::ops; +use serde::de; + +#[cfg(not(feature = "preserve_order"))] +use alloc::collections::{btree_map, BTreeMap}; +#[cfg(feature = "preserve_order")] +use indexmap::{self, IndexMap}; + +/// Represents a JSON key/value type. +pub struct Map<K, V> { + map: MapImpl<K, V>, +} + +#[cfg(not(feature = "preserve_order"))] +type MapImpl<K, V> = BTreeMap<K, V>; +#[cfg(feature = "preserve_order")] +type MapImpl<K, V> = IndexMap<K, V>; + +impl Map<String, Value> { + /// Makes a new empty Map. + #[inline] + pub fn new() -> Self { + Map { + map: MapImpl::new(), + } + } + + /// Makes a new empty Map with the given initial capacity. + #[inline] + pub fn with_capacity(capacity: usize) -> Self { + Map { + #[cfg(not(feature = "preserve_order"))] + map: { + // does not support with_capacity + let _ = capacity; + BTreeMap::new() + }, + #[cfg(feature = "preserve_order")] + map: IndexMap::with_capacity(capacity), + } + } + + /// Clears the map, removing all values. + #[inline] + pub fn clear(&mut self) { + self.map.clear(); + } + + /// Returns a reference to the value corresponding to the key. + /// + /// The key may be any borrowed form of the map's key type, but the ordering + /// on the borrowed form *must* match the ordering on the key type. + #[inline] + pub fn get<Q>(&self, key: &Q) -> Option<&Value> + where + String: Borrow<Q>, + Q: ?Sized + Ord + Eq + Hash, + { + self.map.get(key) + } + + /// Returns true if the map contains a value for the specified key. + /// + /// The key may be any borrowed form of the map's key type, but the ordering + /// on the borrowed form *must* match the ordering on the key type. + #[inline] + pub fn contains_key<Q>(&self, key: &Q) -> bool + where + String: Borrow<Q>, + Q: ?Sized + Ord + Eq + Hash, + { + self.map.contains_key(key) + } + + /// Returns a mutable reference to the value corresponding to the key. + /// + /// The key may be any borrowed form of the map's key type, but the ordering + /// on the borrowed form *must* match the ordering on the key type. + #[inline] + pub fn get_mut<Q>(&mut self, key: &Q) -> Option<&mut Value> + where + String: Borrow<Q>, + Q: ?Sized + Ord + Eq + Hash, + { + self.map.get_mut(key) + } + + /// Returns the key-value pair matching the given key. + /// + /// The key may be any borrowed form of the map's key type, but the ordering + /// on the borrowed form *must* match the ordering on the key type. + #[inline] + #[cfg(any(feature = "preserve_order", not(no_btreemap_get_key_value)))] + pub fn get_key_value<Q>(&self, key: &Q) -> Option<(&String, &Value)> + where + String: Borrow<Q>, + Q: ?Sized + Ord + Eq + Hash, + { + self.map.get_key_value(key) + } + + /// Inserts a key-value pair into the map. + /// + /// If the map did not have this key present, `None` is returned. + /// + /// If the map did have this key present, the value is updated, and the old + /// value is returned. + #[inline] + pub fn insert(&mut self, k: String, v: Value) -> Option<Value> { + self.map.insert(k, v) + } + + /// Removes a key from the map, returning the value at the key if the key + /// was previously in the map. + /// + /// The key may be any borrowed form of the map's key type, but the ordering + /// on the borrowed form *must* match the ordering on the key type. + #[inline] + pub fn remove<Q>(&mut self, key: &Q) -> Option<Value> + where + String: Borrow<Q>, + Q: ?Sized + Ord + Eq + Hash, + { + #[cfg(feature = "preserve_order")] + return self.map.swap_remove(key); + #[cfg(not(feature = "preserve_order"))] + return self.map.remove(key); + } + + /// Removes a key from the map, returning the stored key and value if the + /// key was previously in the map. + /// + /// The key may be any borrowed form of the map's key type, but the ordering + /// on the borrowed form *must* match the ordering on the key type. + pub fn remove_entry<Q>(&mut self, key: &Q) -> Option<(String, Value)> + where + String: Borrow<Q>, + Q: ?Sized + Ord + Eq + Hash, + { + #[cfg(any(feature = "preserve_order", not(no_btreemap_remove_entry)))] + return self.map.remove_entry(key); + #[cfg(all( + not(feature = "preserve_order"), + no_btreemap_remove_entry, + not(no_btreemap_get_key_value), + ))] + { + let (key, _value) = self.map.get_key_value(key)?; + let key = key.clone(); + let value = self.map.remove::<String>(&key)?; + Some((key, value)) + } + #[cfg(all( + not(feature = "preserve_order"), + no_btreemap_remove_entry, + no_btreemap_get_key_value, + ))] + { + use core::ops::{Bound, RangeBounds}; + + struct Key<'a, Q: ?Sized>(&'a Q); + + impl<'a, Q: ?Sized> RangeBounds<Q> for Key<'a, Q> { + fn start_bound(&self) -> Bound<&Q> { + Bound::Included(self.0) + } + fn end_bound(&self) -> Bound<&Q> { + Bound::Included(self.0) + } + } + + let mut range = self.map.range(Key(key)); + let (key, _value) = range.next()?; + let key = key.clone(); + let value = self.map.remove::<String>(&key)?; + Some((key, value)) + } + } + + /// Moves all elements from other into self, leaving other empty. + #[inline] + pub fn append(&mut self, other: &mut Self) { + #[cfg(feature = "preserve_order")] + self.map + .extend(mem::replace(&mut other.map, MapImpl::default())); + #[cfg(not(feature = "preserve_order"))] + self.map.append(&mut other.map); + } + + /// Gets the given key's corresponding entry in the map for in-place + /// manipulation. + pub fn entry<S>(&mut self, key: S) -> Entry + where + S: Into<String>, + { + #[cfg(not(feature = "preserve_order"))] + use alloc::collections::btree_map::Entry as EntryImpl; + #[cfg(feature = "preserve_order")] + use indexmap::map::Entry as EntryImpl; + + match self.map.entry(key.into()) { + EntryImpl::Vacant(vacant) => Entry::Vacant(VacantEntry { vacant }), + EntryImpl::Occupied(occupied) => Entry::Occupied(OccupiedEntry { occupied }), + } + } + + /// Returns the number of elements in the map. + #[inline] + pub fn len(&self) -> usize { + self.map.len() + } + + /// Returns true if the map contains no elements. + #[inline] + pub fn is_empty(&self) -> bool { + self.map.is_empty() + } + + /// Gets an iterator over the entries of the map. + #[inline] + pub fn iter(&self) -> Iter { + Iter { + iter: self.map.iter(), + } + } + + /// Gets a mutable iterator over the entries of the map. + #[inline] + pub fn iter_mut(&mut self) -> IterMut { + IterMut { + iter: self.map.iter_mut(), + } + } + + /// Gets an iterator over the keys of the map. + #[inline] + pub fn keys(&self) -> Keys { + Keys { + iter: self.map.keys(), + } + } + + /// Gets an iterator over the values of the map. + #[inline] + pub fn values(&self) -> Values { + Values { + iter: self.map.values(), + } + } + + /// Gets an iterator over mutable values of the map. + #[inline] + pub fn values_mut(&mut self) -> ValuesMut { + ValuesMut { + iter: self.map.values_mut(), + } + } + + /// Retains only the elements specified by the predicate. + /// + /// In other words, remove all pairs `(k, v)` such that `f(&k, &mut v)` + /// returns `false`. + #[cfg(not(no_btreemap_retain))] + #[inline] + pub fn retain<F>(&mut self, f: F) + where + F: FnMut(&String, &mut Value) -> bool, + { + self.map.retain(f); + } +} + +#[allow(clippy::derivable_impls)] // clippy bug: https://github.com/rust-lang/rust-clippy/issues/7655 +impl Default for Map<String, Value> { + #[inline] + fn default() -> Self { + Map { + map: MapImpl::new(), + } + } +} + +impl Clone for Map<String, Value> { + #[inline] + fn clone(&self) -> Self { + Map { + map: self.map.clone(), + } + } + + #[inline] + fn clone_from(&mut self, source: &Self) { + self.map.clone_from(&source.map); + } +} + +impl PartialEq for Map<String, Value> { + #[inline] + fn eq(&self, other: &Self) -> bool { + self.map.eq(&other.map) + } +} + +impl Eq for Map<String, Value> {} + +/// Access an element of this map. Panics if the given key is not present in the +/// map. +/// +/// ``` +/// # use serde_json::Value; +/// # +/// # let val = &Value::String("".to_owned()); +/// # let _ = +/// match val { +/// Value::String(s) => Some(s.as_str()), +/// Value::Array(arr) => arr[0].as_str(), +/// Value::Object(map) => map["type"].as_str(), +/// _ => None, +/// } +/// # ; +/// ``` +impl<'a, Q> ops::Index<&'a Q> for Map<String, Value> +where + String: Borrow<Q>, + Q: ?Sized + Ord + Eq + Hash, +{ + type Output = Value; + + fn index(&self, index: &Q) -> &Value { + self.map.index(index) + } +} + +/// Mutably access an element of this map. Panics if the given key is not +/// present in the map. +/// +/// ``` +/// # use serde_json::json; +/// # +/// # let mut map = serde_json::Map::new(); +/// # map.insert("key".to_owned(), serde_json::Value::Null); +/// # +/// map["key"] = json!("value"); +/// ``` +impl<'a, Q> ops::IndexMut<&'a Q> for Map<String, Value> +where + String: Borrow<Q>, + Q: ?Sized + Ord + Eq + Hash, +{ + fn index_mut(&mut self, index: &Q) -> &mut Value { + self.map.get_mut(index).expect("no entry found for key") + } +} + +impl Debug for Map<String, Value> { + #[inline] + fn fmt(&self, formatter: &mut fmt::Formatter) -> Result<(), fmt::Error> { + self.map.fmt(formatter) + } +} + +#[cfg(any(feature = "std", feature = "alloc"))] +impl serde::ser::Serialize for Map<String, Value> { + #[inline] + fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error> + where + S: serde::ser::Serializer, + { + use serde::ser::SerializeMap; + let mut map = tri!(serializer.serialize_map(Some(self.len()))); + for (k, v) in self { + tri!(map.serialize_entry(k, v)); + } + map.end() + } +} + +impl<'de> de::Deserialize<'de> for Map<String, Value> { + #[inline] + fn deserialize<D>(deserializer: D) -> Result<Self, D::Error> + where + D: de::Deserializer<'de>, + { + struct Visitor; + + impl<'de> de::Visitor<'de> for Visitor { + type Value = Map<String, Value>; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("a map") + } + + #[inline] + fn visit_unit<E>(self) -> Result<Self::Value, E> + where + E: de::Error, + { + Ok(Map::new()) + } + + #[cfg(any(feature = "std", feature = "alloc"))] + #[inline] + fn visit_map<V>(self, mut visitor: V) -> Result<Self::Value, V::Error> + where + V: de::MapAccess<'de>, + { + let mut values = Map::new(); + + while let Some((key, value)) = tri!(visitor.next_entry()) { + values.insert(key, value); + } + + Ok(values) + } + } + + deserializer.deserialize_map(Visitor) + } +} + +impl FromIterator<(String, Value)> for Map<String, Value> { + fn from_iter<T>(iter: T) -> Self + where + T: IntoIterator<Item = (String, Value)>, + { + Map { + map: FromIterator::from_iter(iter), + } + } +} + +impl Extend<(String, Value)> for Map<String, Value> { + fn extend<T>(&mut self, iter: T) + where + T: IntoIterator<Item = (String, Value)>, + { + self.map.extend(iter); + } +} + +macro_rules! delegate_iterator { + (($name:ident $($generics:tt)*) => $item:ty) => { + impl $($generics)* Iterator for $name $($generics)* { + type Item = $item; + #[inline] + fn next(&mut self) -> Option<Self::Item> { + self.iter.next() + } + #[inline] + fn size_hint(&self) -> (usize, Option<usize>) { + self.iter.size_hint() + } + } + + impl $($generics)* DoubleEndedIterator for $name $($generics)* { + #[inline] + fn next_back(&mut self) -> Option<Self::Item> { + self.iter.next_back() + } + } + + impl $($generics)* ExactSizeIterator for $name $($generics)* { + #[inline] + fn len(&self) -> usize { + self.iter.len() + } + } + + impl $($generics)* FusedIterator for $name $($generics)* {} + } +} + +////////////////////////////////////////////////////////////////////////////// + +/// A view into a single entry in a map, which may either be vacant or occupied. +/// This enum is constructed from the [`entry`] method on [`Map`]. +/// +/// [`entry`]: struct.Map.html#method.entry +/// [`Map`]: struct.Map.html +pub enum Entry<'a> { + /// A vacant Entry. + Vacant(VacantEntry<'a>), + /// An occupied Entry. + Occupied(OccupiedEntry<'a>), +} + +/// A vacant Entry. It is part of the [`Entry`] enum. +/// +/// [`Entry`]: enum.Entry.html +pub struct VacantEntry<'a> { + vacant: VacantEntryImpl<'a>, +} + +/// An occupied Entry. It is part of the [`Entry`] enum. +/// +/// [`Entry`]: enum.Entry.html +pub struct OccupiedEntry<'a> { + occupied: OccupiedEntryImpl<'a>, +} + +#[cfg(not(feature = "preserve_order"))] +type VacantEntryImpl<'a> = btree_map::VacantEntry<'a, String, Value>; +#[cfg(feature = "preserve_order")] +type VacantEntryImpl<'a> = indexmap::map::VacantEntry<'a, String, Value>; + +#[cfg(not(feature = "preserve_order"))] +type OccupiedEntryImpl<'a> = btree_map::OccupiedEntry<'a, String, Value>; +#[cfg(feature = "preserve_order")] +type OccupiedEntryImpl<'a> = indexmap::map::OccupiedEntry<'a, String, Value>; + +impl<'a> Entry<'a> { + /// Returns a reference to this entry's key. + /// + /// # Examples + /// + /// ``` + /// let mut map = serde_json::Map::new(); + /// assert_eq!(map.entry("serde").key(), &"serde"); + /// ``` + pub fn key(&self) -> &String { + match self { + Entry::Vacant(e) => e.key(), + Entry::Occupied(e) => e.key(), + } + } + + /// Ensures a value is in the entry by inserting the default if empty, and + /// returns a mutable reference to the value in the entry. + /// + /// # Examples + /// + /// ``` + /// # use serde_json::json; + /// # + /// let mut map = serde_json::Map::new(); + /// map.entry("serde").or_insert(json!(12)); + /// + /// assert_eq!(map["serde"], 12); + /// ``` + pub fn or_insert(self, default: Value) -> &'a mut Value { + match self { + Entry::Vacant(entry) => entry.insert(default), + Entry::Occupied(entry) => entry.into_mut(), + } + } + + /// Ensures a value is in the entry by inserting the result of the default + /// function if empty, and returns a mutable reference to the value in the + /// entry. + /// + /// # Examples + /// + /// ``` + /// # use serde_json::json; + /// # + /// let mut map = serde_json::Map::new(); + /// map.entry("serde").or_insert_with(|| json!("hoho")); + /// + /// assert_eq!(map["serde"], "hoho".to_owned()); + /// ``` + pub fn or_insert_with<F>(self, default: F) -> &'a mut Value + where + F: FnOnce() -> Value, + { + match self { + Entry::Vacant(entry) => entry.insert(default()), + Entry::Occupied(entry) => entry.into_mut(), + } + } + + /// Provides in-place mutable access to an occupied entry before any + /// potential inserts into the map. + /// + /// # Examples + /// + /// ``` + /// # use serde_json::json; + /// # + /// let mut map = serde_json::Map::new(); + /// map.entry("serde") + /// .and_modify(|e| *e = json!("rust")) + /// .or_insert(json!("cpp")); + /// + /// assert_eq!(map["serde"], "cpp"); + /// + /// map.entry("serde") + /// .and_modify(|e| *e = json!("rust")) + /// .or_insert(json!("cpp")); + /// + /// assert_eq!(map["serde"], "rust"); + /// ``` + pub fn and_modify<F>(self, f: F) -> Self + where + F: FnOnce(&mut Value), + { + match self { + Entry::Occupied(mut entry) => { + f(entry.get_mut()); + Entry::Occupied(entry) + } + Entry::Vacant(entry) => Entry::Vacant(entry), + } + } +} + +impl<'a> VacantEntry<'a> { + /// Gets a reference to the key that would be used when inserting a value + /// through the VacantEntry. + /// + /// # Examples + /// + /// ``` + /// use serde_json::map::Entry; + /// + /// let mut map = serde_json::Map::new(); + /// + /// match map.entry("serde") { + /// Entry::Vacant(vacant) => { + /// assert_eq!(vacant.key(), &"serde"); + /// } + /// Entry::Occupied(_) => unimplemented!(), + /// } + /// ``` + #[inline] + pub fn key(&self) -> &String { + self.vacant.key() + } + + /// Sets the value of the entry with the VacantEntry's key, and returns a + /// mutable reference to it. + /// + /// # Examples + /// + /// ``` + /// # use serde_json::json; + /// # + /// use serde_json::map::Entry; + /// + /// let mut map = serde_json::Map::new(); + /// + /// match map.entry("serde") { + /// Entry::Vacant(vacant) => { + /// vacant.insert(json!("hoho")); + /// } + /// Entry::Occupied(_) => unimplemented!(), + /// } + /// ``` + #[inline] + pub fn insert(self, value: Value) -> &'a mut Value { + self.vacant.insert(value) + } +} + +impl<'a> OccupiedEntry<'a> { + /// Gets a reference to the key in the entry. + /// + /// # Examples + /// + /// ``` + /// # use serde_json::json; + /// # + /// use serde_json::map::Entry; + /// + /// let mut map = serde_json::Map::new(); + /// map.insert("serde".to_owned(), json!(12)); + /// + /// match map.entry("serde") { + /// Entry::Occupied(occupied) => { + /// assert_eq!(occupied.key(), &"serde"); + /// } + /// Entry::Vacant(_) => unimplemented!(), + /// } + /// ``` + #[inline] + pub fn key(&self) -> &String { + self.occupied.key() + } + + /// Gets a reference to the value in the entry. + /// + /// # Examples + /// + /// ``` + /// # use serde_json::json; + /// # + /// use serde_json::map::Entry; + /// + /// let mut map = serde_json::Map::new(); + /// map.insert("serde".to_owned(), json!(12)); + /// + /// match map.entry("serde") { + /// Entry::Occupied(occupied) => { + /// assert_eq!(occupied.get(), 12); + /// } + /// Entry::Vacant(_) => unimplemented!(), + /// } + /// ``` + #[inline] + pub fn get(&self) -> &Value { + self.occupied.get() + } + + /// Gets a mutable reference to the value in the entry. + /// + /// # Examples + /// + /// ``` + /// # use serde_json::json; + /// # + /// use serde_json::map::Entry; + /// + /// let mut map = serde_json::Map::new(); + /// map.insert("serde".to_owned(), json!([1, 2, 3])); + /// + /// match map.entry("serde") { + /// Entry::Occupied(mut occupied) => { + /// occupied.get_mut().as_array_mut().unwrap().push(json!(4)); + /// } + /// Entry::Vacant(_) => unimplemented!(), + /// } + /// + /// assert_eq!(map["serde"].as_array().unwrap().len(), 4); + /// ``` + #[inline] + pub fn get_mut(&mut self) -> &mut Value { + self.occupied.get_mut() + } + + /// Converts the entry into a mutable reference to its value. + /// + /// # Examples + /// + /// ``` + /// # use serde_json::json; + /// # + /// use serde_json::map::Entry; + /// + /// let mut map = serde_json::Map::new(); + /// map.insert("serde".to_owned(), json!([1, 2, 3])); + /// + /// match map.entry("serde") { + /// Entry::Occupied(mut occupied) => { + /// occupied.into_mut().as_array_mut().unwrap().push(json!(4)); + /// } + /// Entry::Vacant(_) => unimplemented!(), + /// } + /// + /// assert_eq!(map["serde"].as_array().unwrap().len(), 4); + /// ``` + #[inline] + pub fn into_mut(self) -> &'a mut Value { + self.occupied.into_mut() + } + + /// Sets the value of the entry with the `OccupiedEntry`'s key, and returns + /// the entry's old value. + /// + /// # Examples + /// + /// ``` + /// # use serde_json::json; + /// # + /// use serde_json::map::Entry; + /// + /// let mut map = serde_json::Map::new(); + /// map.insert("serde".to_owned(), json!(12)); + /// + /// match map.entry("serde") { + /// Entry::Occupied(mut occupied) => { + /// assert_eq!(occupied.insert(json!(13)), 12); + /// assert_eq!(occupied.get(), 13); + /// } + /// Entry::Vacant(_) => unimplemented!(), + /// } + /// ``` + #[inline] + pub fn insert(&mut self, value: Value) -> Value { + self.occupied.insert(value) + } + + /// Takes the value of the entry out of the map, and returns it. + /// + /// # Examples + /// + /// ``` + /// # use serde_json::json; + /// # + /// use serde_json::map::Entry; + /// + /// let mut map = serde_json::Map::new(); + /// map.insert("serde".to_owned(), json!(12)); + /// + /// match map.entry("serde") { + /// Entry::Occupied(occupied) => { + /// assert_eq!(occupied.remove(), 12); + /// } + /// Entry::Vacant(_) => unimplemented!(), + /// } + /// ``` + #[inline] + pub fn remove(self) -> Value { + #[cfg(feature = "preserve_order")] + return self.occupied.swap_remove(); + #[cfg(not(feature = "preserve_order"))] + return self.occupied.remove(); + } +} + +////////////////////////////////////////////////////////////////////////////// + +impl<'a> IntoIterator for &'a Map<String, Value> { + type Item = (&'a String, &'a Value); + type IntoIter = Iter<'a>; + #[inline] + fn into_iter(self) -> Self::IntoIter { + Iter { + iter: self.map.iter(), + } + } +} + +/// An iterator over a serde_json::Map's entries. +pub struct Iter<'a> { + iter: IterImpl<'a>, +} + +#[cfg(not(feature = "preserve_order"))] +type IterImpl<'a> = btree_map::Iter<'a, String, Value>; +#[cfg(feature = "preserve_order")] +type IterImpl<'a> = indexmap::map::Iter<'a, String, Value>; + +delegate_iterator!((Iter<'a>) => (&'a String, &'a Value)); + +////////////////////////////////////////////////////////////////////////////// + +impl<'a> IntoIterator for &'a mut Map<String, Value> { + type Item = (&'a String, &'a mut Value); + type IntoIter = IterMut<'a>; + #[inline] + fn into_iter(self) -> Self::IntoIter { + IterMut { + iter: self.map.iter_mut(), + } + } +} + +/// A mutable iterator over a serde_json::Map's entries. +pub struct IterMut<'a> { + iter: IterMutImpl<'a>, +} + +#[cfg(not(feature = "preserve_order"))] +type IterMutImpl<'a> = btree_map::IterMut<'a, String, Value>; +#[cfg(feature = "preserve_order")] +type IterMutImpl<'a> = indexmap::map::IterMut<'a, String, Value>; + +delegate_iterator!((IterMut<'a>) => (&'a String, &'a mut Value)); + +////////////////////////////////////////////////////////////////////////////// + +impl IntoIterator for Map<String, Value> { + type Item = (String, Value); + type IntoIter = IntoIter; + #[inline] + fn into_iter(self) -> Self::IntoIter { + IntoIter { + iter: self.map.into_iter(), + } + } +} + +/// An owning iterator over a serde_json::Map's entries. +pub struct IntoIter { + iter: IntoIterImpl, +} + +#[cfg(not(feature = "preserve_order"))] +type IntoIterImpl = btree_map::IntoIter<String, Value>; +#[cfg(feature = "preserve_order")] +type IntoIterImpl = indexmap::map::IntoIter<String, Value>; + +delegate_iterator!((IntoIter) => (String, Value)); + +////////////////////////////////////////////////////////////////////////////// + +/// An iterator over a serde_json::Map's keys. +pub struct Keys<'a> { + iter: KeysImpl<'a>, +} + +#[cfg(not(feature = "preserve_order"))] +type KeysImpl<'a> = btree_map::Keys<'a, String, Value>; +#[cfg(feature = "preserve_order")] +type KeysImpl<'a> = indexmap::map::Keys<'a, String, Value>; + +delegate_iterator!((Keys<'a>) => &'a String); + +////////////////////////////////////////////////////////////////////////////// + +/// An iterator over a serde_json::Map's values. +pub struct Values<'a> { + iter: ValuesImpl<'a>, +} + +#[cfg(not(feature = "preserve_order"))] +type ValuesImpl<'a> = btree_map::Values<'a, String, Value>; +#[cfg(feature = "preserve_order")] +type ValuesImpl<'a> = indexmap::map::Values<'a, String, Value>; + +delegate_iterator!((Values<'a>) => &'a Value); + +////////////////////////////////////////////////////////////////////////////// + +/// A mutable iterator over a serde_json::Map's values. +pub struct ValuesMut<'a> { + iter: ValuesMutImpl<'a>, +} + +#[cfg(not(feature = "preserve_order"))] +type ValuesMutImpl<'a> = btree_map::ValuesMut<'a, String, Value>; +#[cfg(feature = "preserve_order")] +type ValuesMutImpl<'a> = indexmap::map::ValuesMut<'a, String, Value>; + +delegate_iterator!((ValuesMut<'a>) => &'a mut Value); diff --git a/third_party/rust/serde_json/src/number.rs b/third_party/rust/serde_json/src/number.rs new file mode 100644 index 0000000000..21a76411cc --- /dev/null +++ b/third_party/rust/serde_json/src/number.rs @@ -0,0 +1,745 @@ +use crate::de::ParserNumber; +use crate::error::Error; +#[cfg(feature = "arbitrary_precision")] +use crate::error::ErrorCode; +#[cfg(feature = "arbitrary_precision")] +use alloc::borrow::ToOwned; +#[cfg(feature = "arbitrary_precision")] +use alloc::string::{String, ToString}; +use core::fmt::{self, Debug, Display}; +#[cfg(not(feature = "arbitrary_precision"))] +use core::hash::{Hash, Hasher}; +use serde::de::{self, Unexpected, Visitor}; +#[cfg(feature = "arbitrary_precision")] +use serde::de::{IntoDeserializer, MapAccess}; +use serde::{forward_to_deserialize_any, Deserialize, Deserializer, Serialize, Serializer}; + +#[cfg(feature = "arbitrary_precision")] +pub(crate) const TOKEN: &str = "$serde_json::private::Number"; + +/// Represents a JSON number, whether integer or floating point. +#[derive(Clone, PartialEq, Eq, Hash)] +pub struct Number { + n: N, +} + +#[cfg(not(feature = "arbitrary_precision"))] +#[derive(Copy, Clone)] +enum N { + PosInt(u64), + /// Always less than zero. + NegInt(i64), + /// Always finite. + Float(f64), +} + +#[cfg(not(feature = "arbitrary_precision"))] +impl PartialEq for N { + fn eq(&self, other: &Self) -> bool { + match (self, other) { + (N::PosInt(a), N::PosInt(b)) => a == b, + (N::NegInt(a), N::NegInt(b)) => a == b, + (N::Float(a), N::Float(b)) => a == b, + _ => false, + } + } +} + +// Implementing Eq is fine since any float values are always finite. +#[cfg(not(feature = "arbitrary_precision"))] +impl Eq for N {} + +#[cfg(not(feature = "arbitrary_precision"))] +impl Hash for N { + fn hash<H: Hasher>(&self, h: &mut H) { + match *self { + N::PosInt(i) => i.hash(h), + N::NegInt(i) => i.hash(h), + N::Float(f) => { + if f == 0.0f64 { + // There are 2 zero representations, +0 and -0, which + // compare equal but have different bits. We use the +0 hash + // for both so that hash(+0) == hash(-0). + 0.0f64.to_bits().hash(h); + } else { + f.to_bits().hash(h); + } + } + } + } +} + +#[cfg(feature = "arbitrary_precision")] +type N = String; + +impl Number { + /// Returns true if the `Number` is an integer between `i64::MIN` and + /// `i64::MAX`. + /// + /// For any Number on which `is_i64` returns true, `as_i64` is guaranteed to + /// return the integer value. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let big = i64::max_value() as u64 + 10; + /// let v = json!({ "a": 64, "b": big, "c": 256.0 }); + /// + /// assert!(v["a"].is_i64()); + /// + /// // Greater than i64::MAX. + /// assert!(!v["b"].is_i64()); + /// + /// // Numbers with a decimal point are not considered integers. + /// assert!(!v["c"].is_i64()); + /// ``` + #[inline] + pub fn is_i64(&self) -> bool { + #[cfg(not(feature = "arbitrary_precision"))] + match self.n { + N::PosInt(v) => v <= i64::max_value() as u64, + N::NegInt(_) => true, + N::Float(_) => false, + } + #[cfg(feature = "arbitrary_precision")] + self.as_i64().is_some() + } + + /// Returns true if the `Number` is an integer between zero and `u64::MAX`. + /// + /// For any Number on which `is_u64` returns true, `as_u64` is guaranteed to + /// return the integer value. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let v = json!({ "a": 64, "b": -64, "c": 256.0 }); + /// + /// assert!(v["a"].is_u64()); + /// + /// // Negative integer. + /// assert!(!v["b"].is_u64()); + /// + /// // Numbers with a decimal point are not considered integers. + /// assert!(!v["c"].is_u64()); + /// ``` + #[inline] + pub fn is_u64(&self) -> bool { + #[cfg(not(feature = "arbitrary_precision"))] + match self.n { + N::PosInt(_) => true, + N::NegInt(_) | N::Float(_) => false, + } + #[cfg(feature = "arbitrary_precision")] + self.as_u64().is_some() + } + + /// Returns true if the `Number` can be represented by f64. + /// + /// For any Number on which `is_f64` returns true, `as_f64` is guaranteed to + /// return the floating point value. + /// + /// Currently this function returns true if and only if both `is_i64` and + /// `is_u64` return false but this is not a guarantee in the future. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let v = json!({ "a": 256.0, "b": 64, "c": -64 }); + /// + /// assert!(v["a"].is_f64()); + /// + /// // Integers. + /// assert!(!v["b"].is_f64()); + /// assert!(!v["c"].is_f64()); + /// ``` + #[inline] + pub fn is_f64(&self) -> bool { + #[cfg(not(feature = "arbitrary_precision"))] + match self.n { + N::Float(_) => true, + N::PosInt(_) | N::NegInt(_) => false, + } + #[cfg(feature = "arbitrary_precision")] + { + for c in self.n.chars() { + if c == '.' || c == 'e' || c == 'E' { + return self.n.parse::<f64>().ok().map_or(false, f64::is_finite); + } + } + false + } + } + + /// If the `Number` is an integer, represent it as i64 if possible. Returns + /// None otherwise. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let big = i64::max_value() as u64 + 10; + /// let v = json!({ "a": 64, "b": big, "c": 256.0 }); + /// + /// assert_eq!(v["a"].as_i64(), Some(64)); + /// assert_eq!(v["b"].as_i64(), None); + /// assert_eq!(v["c"].as_i64(), None); + /// ``` + #[inline] + pub fn as_i64(&self) -> Option<i64> { + #[cfg(not(feature = "arbitrary_precision"))] + match self.n { + N::PosInt(n) => { + if n <= i64::max_value() as u64 { + Some(n as i64) + } else { + None + } + } + N::NegInt(n) => Some(n), + N::Float(_) => None, + } + #[cfg(feature = "arbitrary_precision")] + self.n.parse().ok() + } + + /// If the `Number` is an integer, represent it as u64 if possible. Returns + /// None otherwise. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let v = json!({ "a": 64, "b": -64, "c": 256.0 }); + /// + /// assert_eq!(v["a"].as_u64(), Some(64)); + /// assert_eq!(v["b"].as_u64(), None); + /// assert_eq!(v["c"].as_u64(), None); + /// ``` + #[inline] + pub fn as_u64(&self) -> Option<u64> { + #[cfg(not(feature = "arbitrary_precision"))] + match self.n { + N::PosInt(n) => Some(n), + N::NegInt(_) | N::Float(_) => None, + } + #[cfg(feature = "arbitrary_precision")] + self.n.parse().ok() + } + + /// Represents the number as f64 if possible. Returns None otherwise. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let v = json!({ "a": 256.0, "b": 64, "c": -64 }); + /// + /// assert_eq!(v["a"].as_f64(), Some(256.0)); + /// assert_eq!(v["b"].as_f64(), Some(64.0)); + /// assert_eq!(v["c"].as_f64(), Some(-64.0)); + /// ``` + #[inline] + pub fn as_f64(&self) -> Option<f64> { + #[cfg(not(feature = "arbitrary_precision"))] + match self.n { + N::PosInt(n) => Some(n as f64), + N::NegInt(n) => Some(n as f64), + N::Float(n) => Some(n), + } + #[cfg(feature = "arbitrary_precision")] + self.n.parse::<f64>().ok().filter(|float| float.is_finite()) + } + + /// Converts a finite `f64` to a `Number`. Infinite or NaN values are not JSON + /// numbers. + /// + /// ``` + /// # use std::f64; + /// # + /// # use serde_json::Number; + /// # + /// assert!(Number::from_f64(256.0).is_some()); + /// + /// assert!(Number::from_f64(f64::NAN).is_none()); + /// ``` + #[inline] + pub fn from_f64(f: f64) -> Option<Number> { + if f.is_finite() { + let n = { + #[cfg(not(feature = "arbitrary_precision"))] + { + N::Float(f) + } + #[cfg(feature = "arbitrary_precision")] + { + ryu::Buffer::new().format_finite(f).to_owned() + } + }; + Some(Number { n }) + } else { + None + } + } + + #[cfg(feature = "arbitrary_precision")] + /// Not public API. Only tests use this. + #[doc(hidden)] + #[inline] + pub fn from_string_unchecked(n: String) -> Self { + Number { n } + } +} + +impl Display for Number { + #[cfg(not(feature = "arbitrary_precision"))] + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + match self.n { + N::PosInt(u) => formatter.write_str(itoa::Buffer::new().format(u)), + N::NegInt(i) => formatter.write_str(itoa::Buffer::new().format(i)), + N::Float(f) => formatter.write_str(ryu::Buffer::new().format_finite(f)), + } + } + + #[cfg(feature = "arbitrary_precision")] + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + Display::fmt(&self.n, formatter) + } +} + +impl Debug for Number { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + write!(formatter, "Number({})", self) + } +} + +impl Serialize for Number { + #[cfg(not(feature = "arbitrary_precision"))] + #[inline] + fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error> + where + S: Serializer, + { + match self.n { + N::PosInt(u) => serializer.serialize_u64(u), + N::NegInt(i) => serializer.serialize_i64(i), + N::Float(f) => serializer.serialize_f64(f), + } + } + + #[cfg(feature = "arbitrary_precision")] + #[inline] + fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error> + where + S: Serializer, + { + use serde::ser::SerializeStruct; + + let mut s = serializer.serialize_struct(TOKEN, 1)?; + s.serialize_field(TOKEN, &self.n)?; + s.end() + } +} + +impl<'de> Deserialize<'de> for Number { + #[inline] + fn deserialize<D>(deserializer: D) -> Result<Number, D::Error> + where + D: Deserializer<'de>, + { + struct NumberVisitor; + + impl<'de> Visitor<'de> for NumberVisitor { + type Value = Number; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("a JSON number") + } + + #[inline] + fn visit_i64<E>(self, value: i64) -> Result<Number, E> { + Ok(value.into()) + } + + #[inline] + fn visit_u64<E>(self, value: u64) -> Result<Number, E> { + Ok(value.into()) + } + + #[inline] + fn visit_f64<E>(self, value: f64) -> Result<Number, E> + where + E: de::Error, + { + Number::from_f64(value).ok_or_else(|| de::Error::custom("not a JSON number")) + } + + #[cfg(feature = "arbitrary_precision")] + #[inline] + fn visit_map<V>(self, mut visitor: V) -> Result<Number, V::Error> + where + V: de::MapAccess<'de>, + { + let value = visitor.next_key::<NumberKey>()?; + if value.is_none() { + return Err(de::Error::invalid_type(Unexpected::Map, &self)); + } + let v: NumberFromString = visitor.next_value()?; + Ok(v.value) + } + } + + deserializer.deserialize_any(NumberVisitor) + } +} + +#[cfg(feature = "arbitrary_precision")] +struct NumberKey; + +#[cfg(feature = "arbitrary_precision")] +impl<'de> de::Deserialize<'de> for NumberKey { + fn deserialize<D>(deserializer: D) -> Result<NumberKey, D::Error> + where + D: de::Deserializer<'de>, + { + struct FieldVisitor; + + impl<'de> de::Visitor<'de> for FieldVisitor { + type Value = (); + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("a valid number field") + } + + fn visit_str<E>(self, s: &str) -> Result<(), E> + where + E: de::Error, + { + if s == TOKEN { + Ok(()) + } else { + Err(de::Error::custom("expected field with custom name")) + } + } + } + + deserializer.deserialize_identifier(FieldVisitor)?; + Ok(NumberKey) + } +} + +#[cfg(feature = "arbitrary_precision")] +pub struct NumberFromString { + pub value: Number, +} + +#[cfg(feature = "arbitrary_precision")] +impl<'de> de::Deserialize<'de> for NumberFromString { + fn deserialize<D>(deserializer: D) -> Result<NumberFromString, D::Error> + where + D: de::Deserializer<'de>, + { + struct Visitor; + + impl<'de> de::Visitor<'de> for Visitor { + type Value = NumberFromString; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("string containing a number") + } + + fn visit_str<E>(self, s: &str) -> Result<NumberFromString, E> + where + E: de::Error, + { + let n = tri!(s.parse().map_err(de::Error::custom)); + Ok(NumberFromString { value: n }) + } + } + + deserializer.deserialize_str(Visitor) + } +} + +#[cfg(feature = "arbitrary_precision")] +fn invalid_number() -> Error { + Error::syntax(ErrorCode::InvalidNumber, 0, 0) +} + +macro_rules! deserialize_any { + (@expand [$($num_string:tt)*]) => { + #[cfg(not(feature = "arbitrary_precision"))] + #[inline] + fn deserialize_any<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + match self.n { + N::PosInt(u) => visitor.visit_u64(u), + N::NegInt(i) => visitor.visit_i64(i), + N::Float(f) => visitor.visit_f64(f), + } + } + + #[cfg(feature = "arbitrary_precision")] + #[inline] + fn deserialize_any<V>(self, visitor: V) -> Result<V::Value, Error> + where V: Visitor<'de> + { + if let Some(u) = self.as_u64() { + return visitor.visit_u64(u); + } else if let Some(i) = self.as_i64() { + return visitor.visit_i64(i); + } else if let Some(f) = self.as_f64() { + if ryu::Buffer::new().format_finite(f) == self.n || f.to_string() == self.n { + return visitor.visit_f64(f); + } + } + + visitor.visit_map(NumberDeserializer { + number: Some(self.$($num_string)*), + }) + } + }; + + (owned) => { + deserialize_any!(@expand [n]); + }; + + (ref) => { + deserialize_any!(@expand [n.clone()]); + }; +} + +macro_rules! deserialize_number { + ($deserialize:ident => $visit:ident) => { + #[cfg(not(feature = "arbitrary_precision"))] + fn $deserialize<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + self.deserialize_any(visitor) + } + + #[cfg(feature = "arbitrary_precision")] + fn $deserialize<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: de::Visitor<'de>, + { + visitor.$visit(self.n.parse().map_err(|_| invalid_number())?) + } + }; +} + +impl<'de> Deserializer<'de> for Number { + type Error = Error; + + deserialize_any!(owned); + + deserialize_number!(deserialize_i8 => visit_i8); + deserialize_number!(deserialize_i16 => visit_i16); + deserialize_number!(deserialize_i32 => visit_i32); + deserialize_number!(deserialize_i64 => visit_i64); + deserialize_number!(deserialize_i128 => visit_i128); + deserialize_number!(deserialize_u8 => visit_u8); + deserialize_number!(deserialize_u16 => visit_u16); + deserialize_number!(deserialize_u32 => visit_u32); + deserialize_number!(deserialize_u64 => visit_u64); + deserialize_number!(deserialize_u128 => visit_u128); + deserialize_number!(deserialize_f32 => visit_f32); + deserialize_number!(deserialize_f64 => visit_f64); + + forward_to_deserialize_any! { + bool char str string bytes byte_buf option unit unit_struct + newtype_struct seq tuple tuple_struct map struct enum identifier + ignored_any + } +} + +impl<'de, 'a> Deserializer<'de> for &'a Number { + type Error = Error; + + deserialize_any!(ref); + + deserialize_number!(deserialize_i8 => visit_i8); + deserialize_number!(deserialize_i16 => visit_i16); + deserialize_number!(deserialize_i32 => visit_i32); + deserialize_number!(deserialize_i64 => visit_i64); + deserialize_number!(deserialize_i128 => visit_i128); + deserialize_number!(deserialize_u8 => visit_u8); + deserialize_number!(deserialize_u16 => visit_u16); + deserialize_number!(deserialize_u32 => visit_u32); + deserialize_number!(deserialize_u64 => visit_u64); + deserialize_number!(deserialize_u128 => visit_u128); + deserialize_number!(deserialize_f32 => visit_f32); + deserialize_number!(deserialize_f64 => visit_f64); + + forward_to_deserialize_any! { + bool char str string bytes byte_buf option unit unit_struct + newtype_struct seq tuple tuple_struct map struct enum identifier + ignored_any + } +} + +#[cfg(feature = "arbitrary_precision")] +pub(crate) struct NumberDeserializer { + pub number: Option<String>, +} + +#[cfg(feature = "arbitrary_precision")] +impl<'de> MapAccess<'de> for NumberDeserializer { + type Error = Error; + + fn next_key_seed<K>(&mut self, seed: K) -> Result<Option<K::Value>, Error> + where + K: de::DeserializeSeed<'de>, + { + if self.number.is_none() { + return Ok(None); + } + seed.deserialize(NumberFieldDeserializer).map(Some) + } + + fn next_value_seed<V>(&mut self, seed: V) -> Result<V::Value, Error> + where + V: de::DeserializeSeed<'de>, + { + seed.deserialize(self.number.take().unwrap().into_deserializer()) + } +} + +#[cfg(feature = "arbitrary_precision")] +struct NumberFieldDeserializer; + +#[cfg(feature = "arbitrary_precision")] +impl<'de> Deserializer<'de> for NumberFieldDeserializer { + type Error = Error; + + fn deserialize_any<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: de::Visitor<'de>, + { + visitor.visit_borrowed_str(TOKEN) + } + + forward_to_deserialize_any! { + bool u8 u16 u32 u64 u128 i8 i16 i32 i64 i128 f32 f64 char str string seq + bytes byte_buf map struct option unit newtype_struct ignored_any + unit_struct tuple_struct tuple enum identifier + } +} + +impl From<ParserNumber> for Number { + fn from(value: ParserNumber) -> Self { + let n = match value { + ParserNumber::F64(f) => { + #[cfg(not(feature = "arbitrary_precision"))] + { + N::Float(f) + } + #[cfg(feature = "arbitrary_precision")] + { + f.to_string() + } + } + ParserNumber::U64(u) => { + #[cfg(not(feature = "arbitrary_precision"))] + { + N::PosInt(u) + } + #[cfg(feature = "arbitrary_precision")] + { + u.to_string() + } + } + ParserNumber::I64(i) => { + #[cfg(not(feature = "arbitrary_precision"))] + { + N::NegInt(i) + } + #[cfg(feature = "arbitrary_precision")] + { + i.to_string() + } + } + #[cfg(feature = "arbitrary_precision")] + ParserNumber::String(s) => s, + }; + Number { n } + } +} + +macro_rules! impl_from_unsigned { + ( + $($ty:ty),* + ) => { + $( + impl From<$ty> for Number { + #[inline] + fn from(u: $ty) -> Self { + let n = { + #[cfg(not(feature = "arbitrary_precision"))] + { N::PosInt(u as u64) } + #[cfg(feature = "arbitrary_precision")] + { + itoa::Buffer::new().format(u).to_owned() + } + }; + Number { n } + } + } + )* + }; +} + +macro_rules! impl_from_signed { + ( + $($ty:ty),* + ) => { + $( + impl From<$ty> for Number { + #[inline] + fn from(i: $ty) -> Self { + let n = { + #[cfg(not(feature = "arbitrary_precision"))] + { + if i < 0 { + N::NegInt(i as i64) + } else { + N::PosInt(i as u64) + } + } + #[cfg(feature = "arbitrary_precision")] + { + itoa::Buffer::new().format(i).to_owned() + } + }; + Number { n } + } + } + )* + }; +} + +impl_from_unsigned!(u8, u16, u32, u64, usize); +impl_from_signed!(i8, i16, i32, i64, isize); + +#[cfg(feature = "arbitrary_precision")] +impl_from_unsigned!(u128); +#[cfg(feature = "arbitrary_precision")] +impl_from_signed!(i128); + +impl Number { + #[cfg(not(feature = "arbitrary_precision"))] + #[cold] + pub(crate) fn unexpected(&self) -> Unexpected { + match self.n { + N::PosInt(u) => Unexpected::Unsigned(u), + N::NegInt(i) => Unexpected::Signed(i), + N::Float(f) => Unexpected::Float(f), + } + } + + #[cfg(feature = "arbitrary_precision")] + #[cold] + pub(crate) fn unexpected(&self) -> Unexpected { + Unexpected::Other("number") + } +} diff --git a/third_party/rust/serde_json/src/raw.rs b/third_party/rust/serde_json/src/raw.rs new file mode 100644 index 0000000000..6aa4ffcb61 --- /dev/null +++ b/third_party/rust/serde_json/src/raw.rs @@ -0,0 +1,531 @@ +use crate::error::Error; +use alloc::borrow::ToOwned; +use alloc::boxed::Box; +use alloc::string::String; +use core::fmt::{self, Debug, Display}; +use core::mem; +use serde::de::value::BorrowedStrDeserializer; +use serde::de::{ + self, Deserialize, DeserializeSeed, Deserializer, IntoDeserializer, MapAccess, Unexpected, + Visitor, +}; +use serde::forward_to_deserialize_any; +use serde::ser::{Serialize, SerializeStruct, Serializer}; + +/// Reference to a range of bytes encompassing a single valid JSON value in the +/// input data. +/// +/// A `RawValue` can be used to defer parsing parts of a payload until later, +/// or to avoid parsing it at all in the case that part of the payload just +/// needs to be transferred verbatim into a different output object. +/// +/// When serializing, a value of this type will retain its original formatting +/// and will not be minified or pretty-printed. +/// +/// # Note +/// +/// `RawValue` is only available if serde\_json is built with the `"raw_value"` +/// feature. +/// +/// ```toml +/// [dependencies] +/// serde_json = { version = "1.0", features = ["raw_value"] } +/// ``` +/// +/// # Example +/// +/// ``` +/// use serde::{Deserialize, Serialize}; +/// use serde_json::{Result, value::RawValue}; +/// +/// #[derive(Deserialize)] +/// struct Input<'a> { +/// code: u32, +/// #[serde(borrow)] +/// payload: &'a RawValue, +/// } +/// +/// #[derive(Serialize)] +/// struct Output<'a> { +/// info: (u32, &'a RawValue), +/// } +/// +/// // Efficiently rearrange JSON input containing separate "code" and "payload" +/// // keys into a single "info" key holding an array of code and payload. +/// // +/// // This could be done equivalently using serde_json::Value as the type for +/// // payload, but &RawValue will perform better because it does not require +/// // memory allocation. The correct range of bytes is borrowed from the input +/// // data and pasted verbatim into the output. +/// fn rearrange(input: &str) -> Result<String> { +/// let input: Input = serde_json::from_str(input)?; +/// +/// let output = Output { +/// info: (input.code, input.payload), +/// }; +/// +/// serde_json::to_string(&output) +/// } +/// +/// fn main() -> Result<()> { +/// let out = rearrange(r#" {"code": 200, "payload": {}} "#)?; +/// +/// assert_eq!(out, r#"{"info":[200,{}]}"#); +/// +/// Ok(()) +/// } +/// ``` +/// +/// # Ownership +/// +/// The typical usage of `RawValue` will be in the borrowed form: +/// +/// ``` +/// # use serde::Deserialize; +/// # use serde_json::value::RawValue; +/// # +/// #[derive(Deserialize)] +/// struct SomeStruct<'a> { +/// #[serde(borrow)] +/// raw_value: &'a RawValue, +/// } +/// ``` +/// +/// The borrowed form is suitable when deserializing through +/// [`serde_json::from_str`] and [`serde_json::from_slice`] which support +/// borrowing from the input data without memory allocation. +/// +/// When deserializing through [`serde_json::from_reader`] you will need to use +/// the boxed form of `RawValue` instead. This is almost as efficient but +/// involves buffering the raw value from the I/O stream into memory. +/// +/// [`serde_json::from_str`]: ../fn.from_str.html +/// [`serde_json::from_slice`]: ../fn.from_slice.html +/// [`serde_json::from_reader`]: ../fn.from_reader.html +/// +/// ``` +/// # use serde::Deserialize; +/// # use serde_json::value::RawValue; +/// # +/// #[derive(Deserialize)] +/// struct SomeStruct { +/// raw_value: Box<RawValue>, +/// } +/// ``` +#[cfg_attr(not(doc), repr(transparent))] +#[cfg_attr(docsrs, doc(cfg(feature = "raw_value")))] +pub struct RawValue { + json: str, +} + +impl RawValue { + fn from_borrowed(json: &str) -> &Self { + unsafe { mem::transmute::<&str, &RawValue>(json) } + } + + fn from_owned(json: Box<str>) -> Box<Self> { + unsafe { mem::transmute::<Box<str>, Box<RawValue>>(json) } + } + + fn into_owned(raw_value: Box<Self>) -> Box<str> { + unsafe { mem::transmute::<Box<RawValue>, Box<str>>(raw_value) } + } +} + +impl Clone for Box<RawValue> { + fn clone(&self) -> Self { + (**self).to_owned() + } +} + +impl ToOwned for RawValue { + type Owned = Box<RawValue>; + + fn to_owned(&self) -> Self::Owned { + RawValue::from_owned(self.json.to_owned().into_boxed_str()) + } +} + +impl Default for Box<RawValue> { + fn default() -> Self { + RawValue::from_borrowed("null").to_owned() + } +} + +impl Debug for RawValue { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter + .debug_tuple("RawValue") + .field(&format_args!("{}", &self.json)) + .finish() + } +} + +impl Display for RawValue { + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + f.write_str(&self.json) + } +} + +impl RawValue { + /// Convert an owned `String` of JSON data to an owned `RawValue`. + /// + /// This function is equivalent to `serde_json::from_str::<Box<RawValue>>` + /// except that we avoid an allocation and memcpy if both of the following + /// are true: + /// + /// - the input has no leading or trailing whitespace, and + /// - the input has capacity equal to its length. + pub fn from_string(json: String) -> Result<Box<Self>, Error> { + { + let borrowed = crate::from_str::<&Self>(&json)?; + if borrowed.json.len() < json.len() { + return Ok(borrowed.to_owned()); + } + } + Ok(Self::from_owned(json.into_boxed_str())) + } + + /// Access the JSON text underlying a raw value. + /// + /// # Example + /// + /// ``` + /// use serde::Deserialize; + /// use serde_json::{Result, value::RawValue}; + /// + /// #[derive(Deserialize)] + /// struct Response<'a> { + /// code: u32, + /// #[serde(borrow)] + /// payload: &'a RawValue, + /// } + /// + /// fn process(input: &str) -> Result<()> { + /// let response: Response = serde_json::from_str(input)?; + /// + /// let payload = response.payload.get(); + /// if payload.starts_with('{') { + /// // handle a payload which is a JSON map + /// } else { + /// // handle any other type + /// } + /// + /// Ok(()) + /// } + /// + /// fn main() -> Result<()> { + /// process(r#" {"code": 200, "payload": {}} "#)?; + /// Ok(()) + /// } + /// ``` + pub fn get(&self) -> &str { + &self.json + } +} + +impl From<Box<RawValue>> for Box<str> { + fn from(raw_value: Box<RawValue>) -> Self { + RawValue::into_owned(raw_value) + } +} + +/// Convert a `T` into a boxed `RawValue`. +/// +/// # Example +/// +/// ``` +/// // Upstream crate +/// # #[derive(Serialize)] +/// pub struct Thing { +/// foo: String, +/// bar: Option<String>, +/// extra_data: Box<RawValue>, +/// } +/// +/// // Local crate +/// use serde::Serialize; +/// use serde_json::value::{to_raw_value, RawValue}; +/// +/// #[derive(Serialize)] +/// struct MyExtraData { +/// a: u32, +/// b: u32, +/// } +/// +/// let my_thing = Thing { +/// foo: "FooVal".into(), +/// bar: None, +/// extra_data: to_raw_value(&MyExtraData { a: 1, b: 2 }).unwrap(), +/// }; +/// # assert_eq!( +/// # serde_json::to_value(my_thing).unwrap(), +/// # serde_json::json!({ +/// # "foo": "FooVal", +/// # "bar": null, +/// # "extra_data": { "a": 1, "b": 2 } +/// # }) +/// # ); +/// ``` +/// +/// # Errors +/// +/// This conversion can fail if `T`'s implementation of `Serialize` decides to +/// fail, or if `T` contains a map with non-string keys. +/// +/// ``` +/// use std::collections::BTreeMap; +/// +/// // The keys in this map are vectors, not strings. +/// let mut map = BTreeMap::new(); +/// map.insert(vec![32, 64], "x86"); +/// +/// println!("{}", serde_json::value::to_raw_value(&map).unwrap_err()); +/// ``` +#[cfg_attr(docsrs, doc(cfg(feature = "raw_value")))] +pub fn to_raw_value<T>(value: &T) -> Result<Box<RawValue>, Error> +where + T: ?Sized + Serialize, +{ + let json_string = crate::to_string(value)?; + Ok(RawValue::from_owned(json_string.into_boxed_str())) +} + +pub const TOKEN: &str = "$serde_json::private::RawValue"; + +impl Serialize for RawValue { + fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error> + where + S: Serializer, + { + let mut s = serializer.serialize_struct(TOKEN, 1)?; + s.serialize_field(TOKEN, &self.json)?; + s.end() + } +} + +impl<'de: 'a, 'a> Deserialize<'de> for &'a RawValue { + fn deserialize<D>(deserializer: D) -> Result<Self, D::Error> + where + D: Deserializer<'de>, + { + struct ReferenceVisitor; + + impl<'de> Visitor<'de> for ReferenceVisitor { + type Value = &'de RawValue; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + write!(formatter, "any valid JSON value") + } + + fn visit_map<V>(self, mut visitor: V) -> Result<Self::Value, V::Error> + where + V: MapAccess<'de>, + { + let value = visitor.next_key::<RawKey>()?; + if value.is_none() { + return Err(de::Error::invalid_type(Unexpected::Map, &self)); + } + visitor.next_value_seed(ReferenceFromString) + } + } + + deserializer.deserialize_newtype_struct(TOKEN, ReferenceVisitor) + } +} + +impl<'de> Deserialize<'de> for Box<RawValue> { + fn deserialize<D>(deserializer: D) -> Result<Self, D::Error> + where + D: Deserializer<'de>, + { + struct BoxedVisitor; + + impl<'de> Visitor<'de> for BoxedVisitor { + type Value = Box<RawValue>; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + write!(formatter, "any valid JSON value") + } + + fn visit_map<V>(self, mut visitor: V) -> Result<Self::Value, V::Error> + where + V: MapAccess<'de>, + { + let value = visitor.next_key::<RawKey>()?; + if value.is_none() { + return Err(de::Error::invalid_type(Unexpected::Map, &self)); + } + visitor.next_value_seed(BoxedFromString) + } + } + + deserializer.deserialize_newtype_struct(TOKEN, BoxedVisitor) + } +} + +struct RawKey; + +impl<'de> Deserialize<'de> for RawKey { + fn deserialize<D>(deserializer: D) -> Result<RawKey, D::Error> + where + D: Deserializer<'de>, + { + struct FieldVisitor; + + impl<'de> Visitor<'de> for FieldVisitor { + type Value = (); + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("raw value") + } + + fn visit_str<E>(self, s: &str) -> Result<(), E> + where + E: de::Error, + { + if s == TOKEN { + Ok(()) + } else { + Err(de::Error::custom("unexpected raw value")) + } + } + } + + deserializer.deserialize_identifier(FieldVisitor)?; + Ok(RawKey) + } +} + +pub struct ReferenceFromString; + +impl<'de> DeserializeSeed<'de> for ReferenceFromString { + type Value = &'de RawValue; + + fn deserialize<D>(self, deserializer: D) -> Result<Self::Value, D::Error> + where + D: Deserializer<'de>, + { + deserializer.deserialize_str(self) + } +} + +impl<'de> Visitor<'de> for ReferenceFromString { + type Value = &'de RawValue; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("raw value") + } + + fn visit_borrowed_str<E>(self, s: &'de str) -> Result<Self::Value, E> + where + E: de::Error, + { + Ok(RawValue::from_borrowed(s)) + } +} + +pub struct BoxedFromString; + +impl<'de> DeserializeSeed<'de> for BoxedFromString { + type Value = Box<RawValue>; + + fn deserialize<D>(self, deserializer: D) -> Result<Self::Value, D::Error> + where + D: Deserializer<'de>, + { + deserializer.deserialize_str(self) + } +} + +impl<'de> Visitor<'de> for BoxedFromString { + type Value = Box<RawValue>; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("raw value") + } + + fn visit_str<E>(self, s: &str) -> Result<Self::Value, E> + where + E: de::Error, + { + Ok(RawValue::from_owned(s.to_owned().into_boxed_str())) + } + + #[cfg(any(feature = "std", feature = "alloc"))] + fn visit_string<E>(self, s: String) -> Result<Self::Value, E> + where + E: de::Error, + { + Ok(RawValue::from_owned(s.into_boxed_str())) + } +} + +struct RawKeyDeserializer; + +impl<'de> Deserializer<'de> for RawKeyDeserializer { + type Error = Error; + + fn deserialize_any<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: de::Visitor<'de>, + { + visitor.visit_borrowed_str(TOKEN) + } + + forward_to_deserialize_any! { + bool u8 u16 u32 u64 u128 i8 i16 i32 i64 i128 f32 f64 char str string seq + bytes byte_buf map struct option unit newtype_struct ignored_any + unit_struct tuple_struct tuple enum identifier + } +} + +pub struct OwnedRawDeserializer { + pub raw_value: Option<String>, +} + +impl<'de> MapAccess<'de> for OwnedRawDeserializer { + type Error = Error; + + fn next_key_seed<K>(&mut self, seed: K) -> Result<Option<K::Value>, Error> + where + K: de::DeserializeSeed<'de>, + { + if self.raw_value.is_none() { + return Ok(None); + } + seed.deserialize(RawKeyDeserializer).map(Some) + } + + fn next_value_seed<V>(&mut self, seed: V) -> Result<V::Value, Error> + where + V: de::DeserializeSeed<'de>, + { + seed.deserialize(self.raw_value.take().unwrap().into_deserializer()) + } +} + +pub struct BorrowedRawDeserializer<'de> { + pub raw_value: Option<&'de str>, +} + +impl<'de> MapAccess<'de> for BorrowedRawDeserializer<'de> { + type Error = Error; + + fn next_key_seed<K>(&mut self, seed: K) -> Result<Option<K::Value>, Error> + where + K: de::DeserializeSeed<'de>, + { + if self.raw_value.is_none() { + return Ok(None); + } + seed.deserialize(RawKeyDeserializer).map(Some) + } + + fn next_value_seed<V>(&mut self, seed: V) -> Result<V::Value, Error> + where + V: de::DeserializeSeed<'de>, + { + seed.deserialize(BorrowedStrDeserializer::new(self.raw_value.take().unwrap())) + } +} diff --git a/third_party/rust/serde_json/src/read.rs b/third_party/rust/serde_json/src/read.rs new file mode 100644 index 0000000000..fc3a3ca748 --- /dev/null +++ b/third_party/rust/serde_json/src/read.rs @@ -0,0 +1,1004 @@ +use crate::error::{Error, ErrorCode, Result}; +use alloc::vec::Vec; +use core::char; +use core::cmp; +use core::ops::Deref; +use core::str; + +#[cfg(feature = "std")] +use crate::io; +#[cfg(feature = "std")] +use crate::iter::LineColIterator; + +#[cfg(feature = "raw_value")] +use crate::raw::BorrowedRawDeserializer; +#[cfg(all(feature = "raw_value", feature = "std"))] +use crate::raw::OwnedRawDeserializer; +#[cfg(feature = "raw_value")] +use serde::de::Visitor; + +/// Trait used by the deserializer for iterating over input. This is manually +/// "specialized" for iterating over &[u8]. Once feature(specialization) is +/// stable we can use actual specialization. +/// +/// This trait is sealed and cannot be implemented for types outside of +/// `serde_json`. +pub trait Read<'de>: private::Sealed { + #[doc(hidden)] + fn next(&mut self) -> Result<Option<u8>>; + #[doc(hidden)] + fn peek(&mut self) -> Result<Option<u8>>; + + /// Only valid after a call to peek(). Discards the peeked byte. + #[doc(hidden)] + fn discard(&mut self); + + /// Position of the most recent call to next(). + /// + /// The most recent call was probably next() and not peek(), but this method + /// should try to return a sensible result if the most recent call was + /// actually peek() because we don't always know. + /// + /// Only called in case of an error, so performance is not important. + #[doc(hidden)] + fn position(&self) -> Position; + + /// Position of the most recent call to peek(). + /// + /// The most recent call was probably peek() and not next(), but this method + /// should try to return a sensible result if the most recent call was + /// actually next() because we don't always know. + /// + /// Only called in case of an error, so performance is not important. + #[doc(hidden)] + fn peek_position(&self) -> Position; + + /// Offset from the beginning of the input to the next byte that would be + /// returned by next() or peek(). + #[doc(hidden)] + fn byte_offset(&self) -> usize; + + /// Assumes the previous byte was a quotation mark. Parses a JSON-escaped + /// string until the next quotation mark using the given scratch space if + /// necessary. The scratch space is initially empty. + #[doc(hidden)] + fn parse_str<'s>(&'s mut self, scratch: &'s mut Vec<u8>) -> Result<Reference<'de, 's, str>>; + + /// Assumes the previous byte was a quotation mark. Parses a JSON-escaped + /// string until the next quotation mark using the given scratch space if + /// necessary. The scratch space is initially empty. + /// + /// This function returns the raw bytes in the string with escape sequences + /// expanded but without performing unicode validation. + #[doc(hidden)] + fn parse_str_raw<'s>( + &'s mut self, + scratch: &'s mut Vec<u8>, + ) -> Result<Reference<'de, 's, [u8]>>; + + /// Assumes the previous byte was a quotation mark. Parses a JSON-escaped + /// string until the next quotation mark but discards the data. + #[doc(hidden)] + fn ignore_str(&mut self) -> Result<()>; + + /// Assumes the previous byte was a hex escape sequnce ('\u') in a string. + /// Parses next hexadecimal sequence. + #[doc(hidden)] + fn decode_hex_escape(&mut self) -> Result<u16>; + + /// Switch raw buffering mode on. + /// + /// This is used when deserializing `RawValue`. + #[cfg(feature = "raw_value")] + #[doc(hidden)] + fn begin_raw_buffering(&mut self); + + /// Switch raw buffering mode off and provides the raw buffered data to the + /// given visitor. + #[cfg(feature = "raw_value")] + #[doc(hidden)] + fn end_raw_buffering<V>(&mut self, visitor: V) -> Result<V::Value> + where + V: Visitor<'de>; + + /// Whether StreamDeserializer::next needs to check the failed flag. True + /// for IoRead, false for StrRead and SliceRead which can track failure by + /// truncating their input slice to avoid the extra check on every next + /// call. + #[doc(hidden)] + const should_early_return_if_failed: bool; + + /// Mark a persistent failure of StreamDeserializer, either by setting the + /// flag or by truncating the input data. + #[doc(hidden)] + fn set_failed(&mut self, failed: &mut bool); +} + +pub struct Position { + pub line: usize, + pub column: usize, +} + +pub enum Reference<'b, 'c, T> +where + T: ?Sized + 'static, +{ + Borrowed(&'b T), + Copied(&'c T), +} + +impl<'b, 'c, T> Deref for Reference<'b, 'c, T> +where + T: ?Sized + 'static, +{ + type Target = T; + + fn deref(&self) -> &Self::Target { + match *self { + Reference::Borrowed(b) => b, + Reference::Copied(c) => c, + } + } +} + +/// JSON input source that reads from a std::io input stream. +#[cfg(feature = "std")] +#[cfg_attr(docsrs, doc(cfg(feature = "std")))] +pub struct IoRead<R> +where + R: io::Read, +{ + iter: LineColIterator<io::Bytes<R>>, + /// Temporary storage of peeked byte. + ch: Option<u8>, + #[cfg(feature = "raw_value")] + raw_buffer: Option<Vec<u8>>, +} + +/// JSON input source that reads from a slice of bytes. +// +// This is more efficient than other iterators because peek() can be read-only +// and we can compute line/col position only if an error happens. +pub struct SliceRead<'a> { + slice: &'a [u8], + /// Index of the *next* byte that will be returned by next() or peek(). + index: usize, + #[cfg(feature = "raw_value")] + raw_buffering_start_index: usize, +} + +/// JSON input source that reads from a UTF-8 string. +// +// Able to elide UTF-8 checks by assuming that the input is valid UTF-8. +pub struct StrRead<'a> { + delegate: SliceRead<'a>, + #[cfg(feature = "raw_value")] + data: &'a str, +} + +// Prevent users from implementing the Read trait. +mod private { + pub trait Sealed {} +} + +////////////////////////////////////////////////////////////////////////////// + +#[cfg(feature = "std")] +impl<R> IoRead<R> +where + R: io::Read, +{ + /// Create a JSON input source to read from a std::io input stream. + pub fn new(reader: R) -> Self { + IoRead { + iter: LineColIterator::new(reader.bytes()), + ch: None, + #[cfg(feature = "raw_value")] + raw_buffer: None, + } + } +} + +#[cfg(feature = "std")] +impl<R> private::Sealed for IoRead<R> where R: io::Read {} + +#[cfg(feature = "std")] +impl<R> IoRead<R> +where + R: io::Read, +{ + fn parse_str_bytes<'s, T, F>( + &'s mut self, + scratch: &'s mut Vec<u8>, + validate: bool, + result: F, + ) -> Result<T> + where + T: 's, + F: FnOnce(&'s Self, &'s [u8]) -> Result<T>, + { + loop { + let ch = tri!(next_or_eof(self)); + if !ESCAPE[ch as usize] { + scratch.push(ch); + continue; + } + match ch { + b'"' => { + return result(self, scratch); + } + b'\\' => { + tri!(parse_escape(self, validate, scratch)); + } + _ => { + if validate { + return error(self, ErrorCode::ControlCharacterWhileParsingString); + } + scratch.push(ch); + } + } + } + } +} + +#[cfg(feature = "std")] +impl<'de, R> Read<'de> for IoRead<R> +where + R: io::Read, +{ + #[inline] + fn next(&mut self) -> Result<Option<u8>> { + match self.ch.take() { + Some(ch) => { + #[cfg(feature = "raw_value")] + { + if let Some(buf) = &mut self.raw_buffer { + buf.push(ch); + } + } + Ok(Some(ch)) + } + None => match self.iter.next() { + Some(Err(err)) => Err(Error::io(err)), + Some(Ok(ch)) => { + #[cfg(feature = "raw_value")] + { + if let Some(buf) = &mut self.raw_buffer { + buf.push(ch); + } + } + Ok(Some(ch)) + } + None => Ok(None), + }, + } + } + + #[inline] + fn peek(&mut self) -> Result<Option<u8>> { + match self.ch { + Some(ch) => Ok(Some(ch)), + None => match self.iter.next() { + Some(Err(err)) => Err(Error::io(err)), + Some(Ok(ch)) => { + self.ch = Some(ch); + Ok(self.ch) + } + None => Ok(None), + }, + } + } + + #[cfg(not(feature = "raw_value"))] + #[inline] + fn discard(&mut self) { + self.ch = None; + } + + #[cfg(feature = "raw_value")] + fn discard(&mut self) { + if let Some(ch) = self.ch.take() { + if let Some(buf) = &mut self.raw_buffer { + buf.push(ch); + } + } + } + + fn position(&self) -> Position { + Position { + line: self.iter.line(), + column: self.iter.col(), + } + } + + fn peek_position(&self) -> Position { + // The LineColIterator updates its position during peek() so it has the + // right one here. + self.position() + } + + fn byte_offset(&self) -> usize { + match self.ch { + Some(_) => self.iter.byte_offset() - 1, + None => self.iter.byte_offset(), + } + } + + fn parse_str<'s>(&'s mut self, scratch: &'s mut Vec<u8>) -> Result<Reference<'de, 's, str>> { + self.parse_str_bytes(scratch, true, as_str) + .map(Reference::Copied) + } + + fn parse_str_raw<'s>( + &'s mut self, + scratch: &'s mut Vec<u8>, + ) -> Result<Reference<'de, 's, [u8]>> { + self.parse_str_bytes(scratch, false, |_, bytes| Ok(bytes)) + .map(Reference::Copied) + } + + fn ignore_str(&mut self) -> Result<()> { + loop { + let ch = tri!(next_or_eof(self)); + if !ESCAPE[ch as usize] { + continue; + } + match ch { + b'"' => { + return Ok(()); + } + b'\\' => { + tri!(ignore_escape(self)); + } + _ => { + return error(self, ErrorCode::ControlCharacterWhileParsingString); + } + } + } + } + + fn decode_hex_escape(&mut self) -> Result<u16> { + let mut n = 0; + for _ in 0..4 { + match decode_hex_val(tri!(next_or_eof(self))) { + None => return error(self, ErrorCode::InvalidEscape), + Some(val) => { + n = (n << 4) + val; + } + } + } + Ok(n) + } + + #[cfg(feature = "raw_value")] + fn begin_raw_buffering(&mut self) { + self.raw_buffer = Some(Vec::new()); + } + + #[cfg(feature = "raw_value")] + fn end_raw_buffering<V>(&mut self, visitor: V) -> Result<V::Value> + where + V: Visitor<'de>, + { + let raw = self.raw_buffer.take().unwrap(); + let raw = match String::from_utf8(raw) { + Ok(raw) => raw, + Err(_) => return error(self, ErrorCode::InvalidUnicodeCodePoint), + }; + visitor.visit_map(OwnedRawDeserializer { + raw_value: Some(raw), + }) + } + + const should_early_return_if_failed: bool = true; + + #[inline] + #[cold] + fn set_failed(&mut self, failed: &mut bool) { + *failed = true; + } +} + +////////////////////////////////////////////////////////////////////////////// + +impl<'a> SliceRead<'a> { + /// Create a JSON input source to read from a slice of bytes. + pub fn new(slice: &'a [u8]) -> Self { + SliceRead { + slice, + index: 0, + #[cfg(feature = "raw_value")] + raw_buffering_start_index: 0, + } + } + + fn position_of_index(&self, i: usize) -> Position { + let mut position = Position { line: 1, column: 0 }; + for ch in &self.slice[..i] { + match *ch { + b'\n' => { + position.line += 1; + position.column = 0; + } + _ => { + position.column += 1; + } + } + } + position + } + + /// The big optimization here over IoRead is that if the string contains no + /// backslash escape sequences, the returned &str is a slice of the raw JSON + /// data so we avoid copying into the scratch space. + fn parse_str_bytes<'s, T, F>( + &'s mut self, + scratch: &'s mut Vec<u8>, + validate: bool, + result: F, + ) -> Result<Reference<'a, 's, T>> + where + T: ?Sized + 's, + F: for<'f> FnOnce(&'s Self, &'f [u8]) -> Result<&'f T>, + { + // Index of the first byte not yet copied into the scratch space. + let mut start = self.index; + + loop { + while self.index < self.slice.len() && !ESCAPE[self.slice[self.index] as usize] { + self.index += 1; + } + if self.index == self.slice.len() { + return error(self, ErrorCode::EofWhileParsingString); + } + match self.slice[self.index] { + b'"' => { + if scratch.is_empty() { + // Fast path: return a slice of the raw JSON without any + // copying. + let borrowed = &self.slice[start..self.index]; + self.index += 1; + return result(self, borrowed).map(Reference::Borrowed); + } else { + scratch.extend_from_slice(&self.slice[start..self.index]); + self.index += 1; + return result(self, scratch).map(Reference::Copied); + } + } + b'\\' => { + scratch.extend_from_slice(&self.slice[start..self.index]); + self.index += 1; + tri!(parse_escape(self, validate, scratch)); + start = self.index; + } + _ => { + self.index += 1; + if validate { + return error(self, ErrorCode::ControlCharacterWhileParsingString); + } + } + } + } + } +} + +impl<'a> private::Sealed for SliceRead<'a> {} + +impl<'a> Read<'a> for SliceRead<'a> { + #[inline] + fn next(&mut self) -> Result<Option<u8>> { + // `Ok(self.slice.get(self.index).map(|ch| { self.index += 1; *ch }))` + // is about 10% slower. + Ok(if self.index < self.slice.len() { + let ch = self.slice[self.index]; + self.index += 1; + Some(ch) + } else { + None + }) + } + + #[inline] + fn peek(&mut self) -> Result<Option<u8>> { + // `Ok(self.slice.get(self.index).map(|ch| *ch))` is about 10% slower + // for some reason. + Ok(if self.index < self.slice.len() { + Some(self.slice[self.index]) + } else { + None + }) + } + + #[inline] + fn discard(&mut self) { + self.index += 1; + } + + fn position(&self) -> Position { + self.position_of_index(self.index) + } + + fn peek_position(&self) -> Position { + // Cap it at slice.len() just in case the most recent call was next() + // and it returned the last byte. + self.position_of_index(cmp::min(self.slice.len(), self.index + 1)) + } + + fn byte_offset(&self) -> usize { + self.index + } + + fn parse_str<'s>(&'s mut self, scratch: &'s mut Vec<u8>) -> Result<Reference<'a, 's, str>> { + self.parse_str_bytes(scratch, true, as_str) + } + + fn parse_str_raw<'s>( + &'s mut self, + scratch: &'s mut Vec<u8>, + ) -> Result<Reference<'a, 's, [u8]>> { + self.parse_str_bytes(scratch, false, |_, bytes| Ok(bytes)) + } + + fn ignore_str(&mut self) -> Result<()> { + loop { + while self.index < self.slice.len() && !ESCAPE[self.slice[self.index] as usize] { + self.index += 1; + } + if self.index == self.slice.len() { + return error(self, ErrorCode::EofWhileParsingString); + } + match self.slice[self.index] { + b'"' => { + self.index += 1; + return Ok(()); + } + b'\\' => { + self.index += 1; + tri!(ignore_escape(self)); + } + _ => { + return error(self, ErrorCode::ControlCharacterWhileParsingString); + } + } + } + } + + fn decode_hex_escape(&mut self) -> Result<u16> { + if self.index + 4 > self.slice.len() { + self.index = self.slice.len(); + return error(self, ErrorCode::EofWhileParsingString); + } + + let mut n = 0; + for _ in 0..4 { + let ch = decode_hex_val(self.slice[self.index]); + self.index += 1; + match ch { + None => return error(self, ErrorCode::InvalidEscape), + Some(val) => { + n = (n << 4) + val; + } + } + } + Ok(n) + } + + #[cfg(feature = "raw_value")] + fn begin_raw_buffering(&mut self) { + self.raw_buffering_start_index = self.index; + } + + #[cfg(feature = "raw_value")] + fn end_raw_buffering<V>(&mut self, visitor: V) -> Result<V::Value> + where + V: Visitor<'a>, + { + let raw = &self.slice[self.raw_buffering_start_index..self.index]; + let raw = match str::from_utf8(raw) { + Ok(raw) => raw, + Err(_) => return error(self, ErrorCode::InvalidUnicodeCodePoint), + }; + visitor.visit_map(BorrowedRawDeserializer { + raw_value: Some(raw), + }) + } + + const should_early_return_if_failed: bool = false; + + #[inline] + #[cold] + fn set_failed(&mut self, _failed: &mut bool) { + self.slice = &self.slice[..self.index]; + } +} + +////////////////////////////////////////////////////////////////////////////// + +impl<'a> StrRead<'a> { + /// Create a JSON input source to read from a UTF-8 string. + pub fn new(s: &'a str) -> Self { + StrRead { + delegate: SliceRead::new(s.as_bytes()), + #[cfg(feature = "raw_value")] + data: s, + } + } +} + +impl<'a> private::Sealed for StrRead<'a> {} + +impl<'a> Read<'a> for StrRead<'a> { + #[inline] + fn next(&mut self) -> Result<Option<u8>> { + self.delegate.next() + } + + #[inline] + fn peek(&mut self) -> Result<Option<u8>> { + self.delegate.peek() + } + + #[inline] + fn discard(&mut self) { + self.delegate.discard(); + } + + fn position(&self) -> Position { + self.delegate.position() + } + + fn peek_position(&self) -> Position { + self.delegate.peek_position() + } + + fn byte_offset(&self) -> usize { + self.delegate.byte_offset() + } + + fn parse_str<'s>(&'s mut self, scratch: &'s mut Vec<u8>) -> Result<Reference<'a, 's, str>> { + self.delegate.parse_str_bytes(scratch, true, |_, bytes| { + // The deserialization input came in as &str with a UTF-8 guarantee, + // and the \u-escapes are checked along the way, so don't need to + // check here. + Ok(unsafe { str::from_utf8_unchecked(bytes) }) + }) + } + + fn parse_str_raw<'s>( + &'s mut self, + scratch: &'s mut Vec<u8>, + ) -> Result<Reference<'a, 's, [u8]>> { + self.delegate.parse_str_raw(scratch) + } + + fn ignore_str(&mut self) -> Result<()> { + self.delegate.ignore_str() + } + + fn decode_hex_escape(&mut self) -> Result<u16> { + self.delegate.decode_hex_escape() + } + + #[cfg(feature = "raw_value")] + fn begin_raw_buffering(&mut self) { + self.delegate.begin_raw_buffering(); + } + + #[cfg(feature = "raw_value")] + fn end_raw_buffering<V>(&mut self, visitor: V) -> Result<V::Value> + where + V: Visitor<'a>, + { + let raw = &self.data[self.delegate.raw_buffering_start_index..self.delegate.index]; + visitor.visit_map(BorrowedRawDeserializer { + raw_value: Some(raw), + }) + } + + const should_early_return_if_failed: bool = false; + + #[inline] + #[cold] + fn set_failed(&mut self, failed: &mut bool) { + self.delegate.set_failed(failed); + } +} + +////////////////////////////////////////////////////////////////////////////// + +impl<'a, 'de, R> private::Sealed for &'a mut R where R: Read<'de> {} + +impl<'a, 'de, R> Read<'de> for &'a mut R +where + R: Read<'de>, +{ + fn next(&mut self) -> Result<Option<u8>> { + R::next(self) + } + + fn peek(&mut self) -> Result<Option<u8>> { + R::peek(self) + } + + fn discard(&mut self) { + R::discard(self); + } + + fn position(&self) -> Position { + R::position(self) + } + + fn peek_position(&self) -> Position { + R::peek_position(self) + } + + fn byte_offset(&self) -> usize { + R::byte_offset(self) + } + + fn parse_str<'s>(&'s mut self, scratch: &'s mut Vec<u8>) -> Result<Reference<'de, 's, str>> { + R::parse_str(self, scratch) + } + + fn parse_str_raw<'s>( + &'s mut self, + scratch: &'s mut Vec<u8>, + ) -> Result<Reference<'de, 's, [u8]>> { + R::parse_str_raw(self, scratch) + } + + fn ignore_str(&mut self) -> Result<()> { + R::ignore_str(self) + } + + fn decode_hex_escape(&mut self) -> Result<u16> { + R::decode_hex_escape(self) + } + + #[cfg(feature = "raw_value")] + fn begin_raw_buffering(&mut self) { + R::begin_raw_buffering(self); + } + + #[cfg(feature = "raw_value")] + fn end_raw_buffering<V>(&mut self, visitor: V) -> Result<V::Value> + where + V: Visitor<'de>, + { + R::end_raw_buffering(self, visitor) + } + + const should_early_return_if_failed: bool = R::should_early_return_if_failed; + + fn set_failed(&mut self, failed: &mut bool) { + R::set_failed(self, failed); + } +} + +////////////////////////////////////////////////////////////////////////////// + +/// Marker for whether StreamDeserializer can implement FusedIterator. +pub trait Fused: private::Sealed {} +impl<'a> Fused for SliceRead<'a> {} +impl<'a> Fused for StrRead<'a> {} + +// Lookup table of bytes that must be escaped. A value of true at index i means +// that byte i requires an escape sequence in the input. +static ESCAPE: [bool; 256] = { + const CT: bool = true; // control character \x00..=\x1F + const QU: bool = true; // quote \x22 + const BS: bool = true; // backslash \x5C + const __: bool = false; // allow unescaped + [ + // 1 2 3 4 5 6 7 8 9 A B C D E F + CT, CT, CT, CT, CT, CT, CT, CT, CT, CT, CT, CT, CT, CT, CT, CT, // 0 + CT, CT, CT, CT, CT, CT, CT, CT, CT, CT, CT, CT, CT, CT, CT, CT, // 1 + __, __, QU, __, __, __, __, __, __, __, __, __, __, __, __, __, // 2 + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // 3 + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // 4 + __, __, __, __, __, __, __, __, __, __, __, __, BS, __, __, __, // 5 + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // 6 + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // 7 + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // 8 + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // 9 + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // A + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // B + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // C + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // D + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // E + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // F + ] +}; + +fn next_or_eof<'de, R>(read: &mut R) -> Result<u8> +where + R: ?Sized + Read<'de>, +{ + match tri!(read.next()) { + Some(b) => Ok(b), + None => error(read, ErrorCode::EofWhileParsingString), + } +} + +fn peek_or_eof<'de, R>(read: &mut R) -> Result<u8> +where + R: ?Sized + Read<'de>, +{ + match tri!(read.peek()) { + Some(b) => Ok(b), + None => error(read, ErrorCode::EofWhileParsingString), + } +} + +fn error<'de, R, T>(read: &R, reason: ErrorCode) -> Result<T> +where + R: ?Sized + Read<'de>, +{ + let position = read.position(); + Err(Error::syntax(reason, position.line, position.column)) +} + +fn as_str<'de, 's, R: Read<'de>>(read: &R, slice: &'s [u8]) -> Result<&'s str> { + str::from_utf8(slice).or_else(|_| error(read, ErrorCode::InvalidUnicodeCodePoint)) +} + +/// Parses a JSON escape sequence and appends it into the scratch space. Assumes +/// the previous byte read was a backslash. +fn parse_escape<'de, R: Read<'de>>( + read: &mut R, + validate: bool, + scratch: &mut Vec<u8>, +) -> Result<()> { + let ch = tri!(next_or_eof(read)); + + match ch { + b'"' => scratch.push(b'"'), + b'\\' => scratch.push(b'\\'), + b'/' => scratch.push(b'/'), + b'b' => scratch.push(b'\x08'), + b'f' => scratch.push(b'\x0c'), + b'n' => scratch.push(b'\n'), + b'r' => scratch.push(b'\r'), + b't' => scratch.push(b'\t'), + b'u' => { + fn encode_surrogate(scratch: &mut Vec<u8>, n: u16) { + scratch.extend_from_slice(&[ + (n >> 12 & 0b0000_1111) as u8 | 0b1110_0000, + (n >> 6 & 0b0011_1111) as u8 | 0b1000_0000, + (n & 0b0011_1111) as u8 | 0b1000_0000, + ]); + } + + let c = match tri!(read.decode_hex_escape()) { + n @ 0xDC00..=0xDFFF => { + return if validate { + error(read, ErrorCode::LoneLeadingSurrogateInHexEscape) + } else { + encode_surrogate(scratch, n); + Ok(()) + }; + } + + // Non-BMP characters are encoded as a sequence of two hex + // escapes, representing UTF-16 surrogates. If deserializing a + // utf-8 string the surrogates are required to be paired, + // whereas deserializing a byte string accepts lone surrogates. + n1 @ 0xD800..=0xDBFF => { + if tri!(peek_or_eof(read)) == b'\\' { + read.discard(); + } else { + return if validate { + read.discard(); + error(read, ErrorCode::UnexpectedEndOfHexEscape) + } else { + encode_surrogate(scratch, n1); + Ok(()) + }; + } + + if tri!(peek_or_eof(read)) == b'u' { + read.discard(); + } else { + return if validate { + read.discard(); + error(read, ErrorCode::UnexpectedEndOfHexEscape) + } else { + encode_surrogate(scratch, n1); + // The \ prior to this byte started an escape sequence, + // so we need to parse that now. This recursive call + // does not blow the stack on malicious input because + // the escape is not \u, so it will be handled by one + // of the easy nonrecursive cases. + parse_escape(read, validate, scratch) + }; + } + + let n2 = tri!(read.decode_hex_escape()); + + if n2 < 0xDC00 || n2 > 0xDFFF { + return error(read, ErrorCode::LoneLeadingSurrogateInHexEscape); + } + + let n = (((n1 - 0xD800) as u32) << 10 | (n2 - 0xDC00) as u32) + 0x1_0000; + + match char::from_u32(n) { + Some(c) => c, + None => { + return error(read, ErrorCode::InvalidUnicodeCodePoint); + } + } + } + + // Every u16 outside of the surrogate ranges above is guaranteed + // to be a legal char. + n => char::from_u32(n as u32).unwrap(), + }; + + scratch.extend_from_slice(c.encode_utf8(&mut [0_u8; 4]).as_bytes()); + } + _ => { + return error(read, ErrorCode::InvalidEscape); + } + } + + Ok(()) +} + +/// Parses a JSON escape sequence and discards the value. Assumes the previous +/// byte read was a backslash. +fn ignore_escape<'de, R>(read: &mut R) -> Result<()> +where + R: ?Sized + Read<'de>, +{ + let ch = tri!(next_or_eof(read)); + + match ch { + b'"' | b'\\' | b'/' | b'b' | b'f' | b'n' | b'r' | b't' => {} + b'u' => { + // At this point we don't care if the codepoint is valid. We just + // want to consume it. We don't actually know what is valid or not + // at this point, because that depends on if this string will + // ultimately be parsed into a string or a byte buffer in the "real" + // parse. + + tri!(read.decode_hex_escape()); + } + _ => { + return error(read, ErrorCode::InvalidEscape); + } + } + + Ok(()) +} + +static HEX: [u8; 256] = { + const __: u8 = 255; // not a hex digit + [ + // 1 2 3 4 5 6 7 8 9 A B C D E F + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // 0 + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // 1 + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // 2 + 00, 01, 02, 03, 04, 05, 06, 07, 08, 09, __, __, __, __, __, __, // 3 + __, 10, 11, 12, 13, 14, 15, __, __, __, __, __, __, __, __, __, // 4 + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // 5 + __, 10, 11, 12, 13, 14, 15, __, __, __, __, __, __, __, __, __, // 6 + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // 7 + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // 8 + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // 9 + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // A + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // B + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // C + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // D + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // E + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // F + ] +}; + +fn decode_hex_val(val: u8) -> Option<u16> { + let n = HEX[val as usize] as u16; + if n == 255 { + None + } else { + Some(n) + } +} diff --git a/third_party/rust/serde_json/src/ser.rs b/third_party/rust/serde_json/src/ser.rs new file mode 100644 index 0000000000..80c2deb0c1 --- /dev/null +++ b/third_party/rust/serde_json/src/ser.rs @@ -0,0 +1,2179 @@ +//! Serialize a Rust data structure into JSON data. + +use crate::error::{Error, ErrorCode, Result}; +use crate::io; +use alloc::string::{String, ToString}; +use alloc::vec::Vec; +use core::fmt::{self, Display}; +use core::num::FpCategory; +use serde::ser::{self, Impossible, Serialize}; + +/// A structure for serializing Rust values into JSON. +#[cfg_attr(docsrs, doc(cfg(feature = "std")))] +pub struct Serializer<W, F = CompactFormatter> { + writer: W, + formatter: F, +} + +impl<W> Serializer<W> +where + W: io::Write, +{ + /// Creates a new JSON serializer. + #[inline] + pub fn new(writer: W) -> Self { + Serializer::with_formatter(writer, CompactFormatter) + } +} + +impl<'a, W> Serializer<W, PrettyFormatter<'a>> +where + W: io::Write, +{ + /// Creates a new JSON pretty print serializer. + #[inline] + pub fn pretty(writer: W) -> Self { + Serializer::with_formatter(writer, PrettyFormatter::new()) + } +} + +impl<W, F> Serializer<W, F> +where + W: io::Write, + F: Formatter, +{ + /// Creates a new JSON visitor whose output will be written to the writer + /// specified. + #[inline] + pub fn with_formatter(writer: W, formatter: F) -> Self { + Serializer { writer, formatter } + } + + /// Unwrap the `Writer` from the `Serializer`. + #[inline] + pub fn into_inner(self) -> W { + self.writer + } +} + +impl<'a, W, F> ser::Serializer for &'a mut Serializer<W, F> +where + W: io::Write, + F: Formatter, +{ + type Ok = (); + type Error = Error; + + type SerializeSeq = Compound<'a, W, F>; + type SerializeTuple = Compound<'a, W, F>; + type SerializeTupleStruct = Compound<'a, W, F>; + type SerializeTupleVariant = Compound<'a, W, F>; + type SerializeMap = Compound<'a, W, F>; + type SerializeStruct = Compound<'a, W, F>; + type SerializeStructVariant = Compound<'a, W, F>; + + #[inline] + fn serialize_bool(self, value: bool) -> Result<()> { + self.formatter + .write_bool(&mut self.writer, value) + .map_err(Error::io) + } + + #[inline] + fn serialize_i8(self, value: i8) -> Result<()> { + self.formatter + .write_i8(&mut self.writer, value) + .map_err(Error::io) + } + + #[inline] + fn serialize_i16(self, value: i16) -> Result<()> { + self.formatter + .write_i16(&mut self.writer, value) + .map_err(Error::io) + } + + #[inline] + fn serialize_i32(self, value: i32) -> Result<()> { + self.formatter + .write_i32(&mut self.writer, value) + .map_err(Error::io) + } + + #[inline] + fn serialize_i64(self, value: i64) -> Result<()> { + self.formatter + .write_i64(&mut self.writer, value) + .map_err(Error::io) + } + + fn serialize_i128(self, value: i128) -> Result<()> { + self.formatter + .write_i128(&mut self.writer, value) + .map_err(Error::io) + } + + #[inline] + fn serialize_u8(self, value: u8) -> Result<()> { + self.formatter + .write_u8(&mut self.writer, value) + .map_err(Error::io) + } + + #[inline] + fn serialize_u16(self, value: u16) -> Result<()> { + self.formatter + .write_u16(&mut self.writer, value) + .map_err(Error::io) + } + + #[inline] + fn serialize_u32(self, value: u32) -> Result<()> { + self.formatter + .write_u32(&mut self.writer, value) + .map_err(Error::io) + } + + #[inline] + fn serialize_u64(self, value: u64) -> Result<()> { + self.formatter + .write_u64(&mut self.writer, value) + .map_err(Error::io) + } + + fn serialize_u128(self, value: u128) -> Result<()> { + self.formatter + .write_u128(&mut self.writer, value) + .map_err(Error::io) + } + + #[inline] + fn serialize_f32(self, value: f32) -> Result<()> { + match value.classify() { + FpCategory::Nan | FpCategory::Infinite => self + .formatter + .write_null(&mut self.writer) + .map_err(Error::io), + _ => self + .formatter + .write_f32(&mut self.writer, value) + .map_err(Error::io), + } + } + + #[inline] + fn serialize_f64(self, value: f64) -> Result<()> { + match value.classify() { + FpCategory::Nan | FpCategory::Infinite => self + .formatter + .write_null(&mut self.writer) + .map_err(Error::io), + _ => self + .formatter + .write_f64(&mut self.writer, value) + .map_err(Error::io), + } + } + + #[inline] + fn serialize_char(self, value: char) -> Result<()> { + // A char encoded as UTF-8 takes 4 bytes at most. + let mut buf = [0; 4]; + self.serialize_str(value.encode_utf8(&mut buf)) + } + + #[inline] + fn serialize_str(self, value: &str) -> Result<()> { + format_escaped_str(&mut self.writer, &mut self.formatter, value).map_err(Error::io) + } + + #[inline] + fn serialize_bytes(self, value: &[u8]) -> Result<()> { + use serde::ser::SerializeSeq; + let mut seq = tri!(self.serialize_seq(Some(value.len()))); + for byte in value { + tri!(seq.serialize_element(byte)); + } + seq.end() + } + + #[inline] + fn serialize_unit(self) -> Result<()> { + self.formatter + .write_null(&mut self.writer) + .map_err(Error::io) + } + + #[inline] + fn serialize_unit_struct(self, _name: &'static str) -> Result<()> { + self.serialize_unit() + } + + #[inline] + fn serialize_unit_variant( + self, + _name: &'static str, + _variant_index: u32, + variant: &'static str, + ) -> Result<()> { + self.serialize_str(variant) + } + + /// Serialize newtypes without an object wrapper. + #[inline] + fn serialize_newtype_struct<T>(self, _name: &'static str, value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + value.serialize(self) + } + + #[inline] + fn serialize_newtype_variant<T>( + self, + _name: &'static str, + _variant_index: u32, + variant: &'static str, + value: &T, + ) -> Result<()> + where + T: ?Sized + Serialize, + { + tri!(self + .formatter + .begin_object(&mut self.writer) + .map_err(Error::io)); + tri!(self + .formatter + .begin_object_key(&mut self.writer, true) + .map_err(Error::io)); + tri!(self.serialize_str(variant)); + tri!(self + .formatter + .end_object_key(&mut self.writer) + .map_err(Error::io)); + tri!(self + .formatter + .begin_object_value(&mut self.writer) + .map_err(Error::io)); + tri!(value.serialize(&mut *self)); + tri!(self + .formatter + .end_object_value(&mut self.writer) + .map_err(Error::io)); + self.formatter + .end_object(&mut self.writer) + .map_err(Error::io) + } + + #[inline] + fn serialize_none(self) -> Result<()> { + self.serialize_unit() + } + + #[inline] + fn serialize_some<T>(self, value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + value.serialize(self) + } + + #[inline] + fn serialize_seq(self, len: Option<usize>) -> Result<Self::SerializeSeq> { + tri!(self + .formatter + .begin_array(&mut self.writer) + .map_err(Error::io)); + if len == Some(0) { + tri!(self + .formatter + .end_array(&mut self.writer) + .map_err(Error::io)); + Ok(Compound::Map { + ser: self, + state: State::Empty, + }) + } else { + Ok(Compound::Map { + ser: self, + state: State::First, + }) + } + } + + #[inline] + fn serialize_tuple(self, len: usize) -> Result<Self::SerializeTuple> { + self.serialize_seq(Some(len)) + } + + #[inline] + fn serialize_tuple_struct( + self, + _name: &'static str, + len: usize, + ) -> Result<Self::SerializeTupleStruct> { + self.serialize_seq(Some(len)) + } + + #[inline] + fn serialize_tuple_variant( + self, + _name: &'static str, + _variant_index: u32, + variant: &'static str, + len: usize, + ) -> Result<Self::SerializeTupleVariant> { + tri!(self + .formatter + .begin_object(&mut self.writer) + .map_err(Error::io)); + tri!(self + .formatter + .begin_object_key(&mut self.writer, true) + .map_err(Error::io)); + tri!(self.serialize_str(variant)); + tri!(self + .formatter + .end_object_key(&mut self.writer) + .map_err(Error::io)); + tri!(self + .formatter + .begin_object_value(&mut self.writer) + .map_err(Error::io)); + self.serialize_seq(Some(len)) + } + + #[inline] + fn serialize_map(self, len: Option<usize>) -> Result<Self::SerializeMap> { + tri!(self + .formatter + .begin_object(&mut self.writer) + .map_err(Error::io)); + if len == Some(0) { + tri!(self + .formatter + .end_object(&mut self.writer) + .map_err(Error::io)); + Ok(Compound::Map { + ser: self, + state: State::Empty, + }) + } else { + Ok(Compound::Map { + ser: self, + state: State::First, + }) + } + } + + #[inline] + fn serialize_struct(self, name: &'static str, len: usize) -> Result<Self::SerializeStruct> { + match name { + #[cfg(feature = "arbitrary_precision")] + crate::number::TOKEN => Ok(Compound::Number { ser: self }), + #[cfg(feature = "raw_value")] + crate::raw::TOKEN => Ok(Compound::RawValue { ser: self }), + _ => self.serialize_map(Some(len)), + } + } + + #[inline] + fn serialize_struct_variant( + self, + _name: &'static str, + _variant_index: u32, + variant: &'static str, + len: usize, + ) -> Result<Self::SerializeStructVariant> { + tri!(self + .formatter + .begin_object(&mut self.writer) + .map_err(Error::io)); + tri!(self + .formatter + .begin_object_key(&mut self.writer, true) + .map_err(Error::io)); + tri!(self.serialize_str(variant)); + tri!(self + .formatter + .end_object_key(&mut self.writer) + .map_err(Error::io)); + tri!(self + .formatter + .begin_object_value(&mut self.writer) + .map_err(Error::io)); + self.serialize_map(Some(len)) + } + + fn collect_str<T>(self, value: &T) -> Result<()> + where + T: ?Sized + Display, + { + use self::fmt::Write; + + struct Adapter<'ser, W: 'ser, F: 'ser> { + writer: &'ser mut W, + formatter: &'ser mut F, + error: Option<io::Error>, + } + + impl<'ser, W, F> Write for Adapter<'ser, W, F> + where + W: io::Write, + F: Formatter, + { + fn write_str(&mut self, s: &str) -> fmt::Result { + debug_assert!(self.error.is_none()); + match format_escaped_str_contents(self.writer, self.formatter, s) { + Ok(()) => Ok(()), + Err(err) => { + self.error = Some(err); + Err(fmt::Error) + } + } + } + } + + tri!(self + .formatter + .begin_string(&mut self.writer) + .map_err(Error::io)); + { + let mut adapter = Adapter { + writer: &mut self.writer, + formatter: &mut self.formatter, + error: None, + }; + match write!(adapter, "{}", value) { + Ok(()) => debug_assert!(adapter.error.is_none()), + Err(fmt::Error) => { + return Err(Error::io(adapter.error.expect("there should be an error"))); + } + } + } + self.formatter + .end_string(&mut self.writer) + .map_err(Error::io) + } +} + +// Not public API. Should be pub(crate). +#[doc(hidden)] +#[derive(Eq, PartialEq)] +pub enum State { + Empty, + First, + Rest, +} + +// Not public API. Should be pub(crate). +#[doc(hidden)] +pub enum Compound<'a, W: 'a, F: 'a> { + Map { + ser: &'a mut Serializer<W, F>, + state: State, + }, + #[cfg(feature = "arbitrary_precision")] + Number { ser: &'a mut Serializer<W, F> }, + #[cfg(feature = "raw_value")] + RawValue { ser: &'a mut Serializer<W, F> }, +} + +impl<'a, W, F> ser::SerializeSeq for Compound<'a, W, F> +where + W: io::Write, + F: Formatter, +{ + type Ok = (); + type Error = Error; + + #[inline] + fn serialize_element<T>(&mut self, value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + match self { + Compound::Map { ser, state } => { + tri!(ser + .formatter + .begin_array_value(&mut ser.writer, *state == State::First) + .map_err(Error::io)); + *state = State::Rest; + tri!(value.serialize(&mut **ser)); + ser.formatter + .end_array_value(&mut ser.writer) + .map_err(Error::io) + } + #[cfg(feature = "arbitrary_precision")] + Compound::Number { .. } => unreachable!(), + #[cfg(feature = "raw_value")] + Compound::RawValue { .. } => unreachable!(), + } + } + + #[inline] + fn end(self) -> Result<()> { + match self { + Compound::Map { ser, state } => match state { + State::Empty => Ok(()), + _ => ser.formatter.end_array(&mut ser.writer).map_err(Error::io), + }, + #[cfg(feature = "arbitrary_precision")] + Compound::Number { .. } => unreachable!(), + #[cfg(feature = "raw_value")] + Compound::RawValue { .. } => unreachable!(), + } + } +} + +impl<'a, W, F> ser::SerializeTuple for Compound<'a, W, F> +where + W: io::Write, + F: Formatter, +{ + type Ok = (); + type Error = Error; + + #[inline] + fn serialize_element<T>(&mut self, value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + ser::SerializeSeq::serialize_element(self, value) + } + + #[inline] + fn end(self) -> Result<()> { + ser::SerializeSeq::end(self) + } +} + +impl<'a, W, F> ser::SerializeTupleStruct for Compound<'a, W, F> +where + W: io::Write, + F: Formatter, +{ + type Ok = (); + type Error = Error; + + #[inline] + fn serialize_field<T>(&mut self, value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + ser::SerializeSeq::serialize_element(self, value) + } + + #[inline] + fn end(self) -> Result<()> { + ser::SerializeSeq::end(self) + } +} + +impl<'a, W, F> ser::SerializeTupleVariant for Compound<'a, W, F> +where + W: io::Write, + F: Formatter, +{ + type Ok = (); + type Error = Error; + + #[inline] + fn serialize_field<T>(&mut self, value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + ser::SerializeSeq::serialize_element(self, value) + } + + #[inline] + fn end(self) -> Result<()> { + match self { + Compound::Map { ser, state } => { + match state { + State::Empty => {} + _ => tri!(ser.formatter.end_array(&mut ser.writer).map_err(Error::io)), + } + tri!(ser + .formatter + .end_object_value(&mut ser.writer) + .map_err(Error::io)); + ser.formatter.end_object(&mut ser.writer).map_err(Error::io) + } + #[cfg(feature = "arbitrary_precision")] + Compound::Number { .. } => unreachable!(), + #[cfg(feature = "raw_value")] + Compound::RawValue { .. } => unreachable!(), + } + } +} + +impl<'a, W, F> ser::SerializeMap for Compound<'a, W, F> +where + W: io::Write, + F: Formatter, +{ + type Ok = (); + type Error = Error; + + #[inline] + fn serialize_key<T>(&mut self, key: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + match self { + Compound::Map { ser, state } => { + tri!(ser + .formatter + .begin_object_key(&mut ser.writer, *state == State::First) + .map_err(Error::io)); + *state = State::Rest; + + tri!(key.serialize(MapKeySerializer { ser: *ser })); + + ser.formatter + .end_object_key(&mut ser.writer) + .map_err(Error::io) + } + #[cfg(feature = "arbitrary_precision")] + Compound::Number { .. } => unreachable!(), + #[cfg(feature = "raw_value")] + Compound::RawValue { .. } => unreachable!(), + } + } + + #[inline] + fn serialize_value<T>(&mut self, value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + match self { + Compound::Map { ser, .. } => { + tri!(ser + .formatter + .begin_object_value(&mut ser.writer) + .map_err(Error::io)); + tri!(value.serialize(&mut **ser)); + ser.formatter + .end_object_value(&mut ser.writer) + .map_err(Error::io) + } + #[cfg(feature = "arbitrary_precision")] + Compound::Number { .. } => unreachable!(), + #[cfg(feature = "raw_value")] + Compound::RawValue { .. } => unreachable!(), + } + } + + #[inline] + fn end(self) -> Result<()> { + match self { + Compound::Map { ser, state } => match state { + State::Empty => Ok(()), + _ => ser.formatter.end_object(&mut ser.writer).map_err(Error::io), + }, + #[cfg(feature = "arbitrary_precision")] + Compound::Number { .. } => unreachable!(), + #[cfg(feature = "raw_value")] + Compound::RawValue { .. } => unreachable!(), + } + } +} + +impl<'a, W, F> ser::SerializeStruct for Compound<'a, W, F> +where + W: io::Write, + F: Formatter, +{ + type Ok = (); + type Error = Error; + + #[inline] + fn serialize_field<T>(&mut self, key: &'static str, value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + match self { + Compound::Map { .. } => ser::SerializeMap::serialize_entry(self, key, value), + #[cfg(feature = "arbitrary_precision")] + Compound::Number { ser, .. } => { + if key == crate::number::TOKEN { + value.serialize(NumberStrEmitter(ser)) + } else { + Err(invalid_number()) + } + } + #[cfg(feature = "raw_value")] + Compound::RawValue { ser, .. } => { + if key == crate::raw::TOKEN { + value.serialize(RawValueStrEmitter(ser)) + } else { + Err(invalid_raw_value()) + } + } + } + } + + #[inline] + fn end(self) -> Result<()> { + match self { + Compound::Map { .. } => ser::SerializeMap::end(self), + #[cfg(feature = "arbitrary_precision")] + Compound::Number { .. } => Ok(()), + #[cfg(feature = "raw_value")] + Compound::RawValue { .. } => Ok(()), + } + } +} + +impl<'a, W, F> ser::SerializeStructVariant for Compound<'a, W, F> +where + W: io::Write, + F: Formatter, +{ + type Ok = (); + type Error = Error; + + #[inline] + fn serialize_field<T>(&mut self, key: &'static str, value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + match *self { + Compound::Map { .. } => ser::SerializeStruct::serialize_field(self, key, value), + #[cfg(feature = "arbitrary_precision")] + Compound::Number { .. } => unreachable!(), + #[cfg(feature = "raw_value")] + Compound::RawValue { .. } => unreachable!(), + } + } + + #[inline] + fn end(self) -> Result<()> { + match self { + Compound::Map { ser, state } => { + match state { + State::Empty => {} + _ => tri!(ser.formatter.end_object(&mut ser.writer).map_err(Error::io)), + } + tri!(ser + .formatter + .end_object_value(&mut ser.writer) + .map_err(Error::io)); + ser.formatter.end_object(&mut ser.writer).map_err(Error::io) + } + #[cfg(feature = "arbitrary_precision")] + Compound::Number { .. } => unreachable!(), + #[cfg(feature = "raw_value")] + Compound::RawValue { .. } => unreachable!(), + } + } +} + +struct MapKeySerializer<'a, W: 'a, F: 'a> { + ser: &'a mut Serializer<W, F>, +} + +#[cfg(feature = "arbitrary_precision")] +fn invalid_number() -> Error { + Error::syntax(ErrorCode::InvalidNumber, 0, 0) +} + +#[cfg(feature = "raw_value")] +fn invalid_raw_value() -> Error { + Error::syntax(ErrorCode::ExpectedSomeValue, 0, 0) +} + +fn key_must_be_a_string() -> Error { + Error::syntax(ErrorCode::KeyMustBeAString, 0, 0) +} + +impl<'a, W, F> ser::Serializer for MapKeySerializer<'a, W, F> +where + W: io::Write, + F: Formatter, +{ + type Ok = (); + type Error = Error; + + #[inline] + fn serialize_str(self, value: &str) -> Result<()> { + self.ser.serialize_str(value) + } + + #[inline] + fn serialize_unit_variant( + self, + _name: &'static str, + _variant_index: u32, + variant: &'static str, + ) -> Result<()> { + self.ser.serialize_str(variant) + } + + #[inline] + fn serialize_newtype_struct<T>(self, _name: &'static str, value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + value.serialize(self) + } + + type SerializeSeq = Impossible<(), Error>; + type SerializeTuple = Impossible<(), Error>; + type SerializeTupleStruct = Impossible<(), Error>; + type SerializeTupleVariant = Impossible<(), Error>; + type SerializeMap = Impossible<(), Error>; + type SerializeStruct = Impossible<(), Error>; + type SerializeStructVariant = Impossible<(), Error>; + + fn serialize_bool(self, _value: bool) -> Result<()> { + Err(key_must_be_a_string()) + } + + fn serialize_i8(self, value: i8) -> Result<()> { + tri!(self + .ser + .formatter + .begin_string(&mut self.ser.writer) + .map_err(Error::io)); + tri!(self + .ser + .formatter + .write_i8(&mut self.ser.writer, value) + .map_err(Error::io)); + self.ser + .formatter + .end_string(&mut self.ser.writer) + .map_err(Error::io) + } + + fn serialize_i16(self, value: i16) -> Result<()> { + tri!(self + .ser + .formatter + .begin_string(&mut self.ser.writer) + .map_err(Error::io)); + tri!(self + .ser + .formatter + .write_i16(&mut self.ser.writer, value) + .map_err(Error::io)); + self.ser + .formatter + .end_string(&mut self.ser.writer) + .map_err(Error::io) + } + + fn serialize_i32(self, value: i32) -> Result<()> { + tri!(self + .ser + .formatter + .begin_string(&mut self.ser.writer) + .map_err(Error::io)); + tri!(self + .ser + .formatter + .write_i32(&mut self.ser.writer, value) + .map_err(Error::io)); + self.ser + .formatter + .end_string(&mut self.ser.writer) + .map_err(Error::io) + } + + fn serialize_i64(self, value: i64) -> Result<()> { + tri!(self + .ser + .formatter + .begin_string(&mut self.ser.writer) + .map_err(Error::io)); + tri!(self + .ser + .formatter + .write_i64(&mut self.ser.writer, value) + .map_err(Error::io)); + self.ser + .formatter + .end_string(&mut self.ser.writer) + .map_err(Error::io) + } + + fn serialize_i128(self, value: i128) -> Result<()> { + tri!(self + .ser + .formatter + .begin_string(&mut self.ser.writer) + .map_err(Error::io)); + tri!(self + .ser + .formatter + .write_i128(&mut self.ser.writer, value) + .map_err(Error::io)); + self.ser + .formatter + .end_string(&mut self.ser.writer) + .map_err(Error::io) + } + + fn serialize_u8(self, value: u8) -> Result<()> { + tri!(self + .ser + .formatter + .begin_string(&mut self.ser.writer) + .map_err(Error::io)); + tri!(self + .ser + .formatter + .write_u8(&mut self.ser.writer, value) + .map_err(Error::io)); + self.ser + .formatter + .end_string(&mut self.ser.writer) + .map_err(Error::io) + } + + fn serialize_u16(self, value: u16) -> Result<()> { + tri!(self + .ser + .formatter + .begin_string(&mut self.ser.writer) + .map_err(Error::io)); + tri!(self + .ser + .formatter + .write_u16(&mut self.ser.writer, value) + .map_err(Error::io)); + self.ser + .formatter + .end_string(&mut self.ser.writer) + .map_err(Error::io) + } + + fn serialize_u32(self, value: u32) -> Result<()> { + tri!(self + .ser + .formatter + .begin_string(&mut self.ser.writer) + .map_err(Error::io)); + tri!(self + .ser + .formatter + .write_u32(&mut self.ser.writer, value) + .map_err(Error::io)); + self.ser + .formatter + .end_string(&mut self.ser.writer) + .map_err(Error::io) + } + + fn serialize_u64(self, value: u64) -> Result<()> { + tri!(self + .ser + .formatter + .begin_string(&mut self.ser.writer) + .map_err(Error::io)); + tri!(self + .ser + .formatter + .write_u64(&mut self.ser.writer, value) + .map_err(Error::io)); + self.ser + .formatter + .end_string(&mut self.ser.writer) + .map_err(Error::io) + } + + fn serialize_u128(self, value: u128) -> Result<()> { + tri!(self + .ser + .formatter + .begin_string(&mut self.ser.writer) + .map_err(Error::io)); + tri!(self + .ser + .formatter + .write_u128(&mut self.ser.writer, value) + .map_err(Error::io)); + self.ser + .formatter + .end_string(&mut self.ser.writer) + .map_err(Error::io) + } + + fn serialize_f32(self, _value: f32) -> Result<()> { + Err(key_must_be_a_string()) + } + + fn serialize_f64(self, _value: f64) -> Result<()> { + Err(key_must_be_a_string()) + } + + fn serialize_char(self, value: char) -> Result<()> { + self.ser.serialize_str(&value.to_string()) + } + + fn serialize_bytes(self, _value: &[u8]) -> Result<()> { + Err(key_must_be_a_string()) + } + + fn serialize_unit(self) -> Result<()> { + Err(key_must_be_a_string()) + } + + fn serialize_unit_struct(self, _name: &'static str) -> Result<()> { + Err(key_must_be_a_string()) + } + + fn serialize_newtype_variant<T>( + self, + _name: &'static str, + _variant_index: u32, + _variant: &'static str, + _value: &T, + ) -> Result<()> + where + T: ?Sized + Serialize, + { + Err(key_must_be_a_string()) + } + + fn serialize_none(self) -> Result<()> { + Err(key_must_be_a_string()) + } + + fn serialize_some<T>(self, _value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + Err(key_must_be_a_string()) + } + + fn serialize_seq(self, _len: Option<usize>) -> Result<Self::SerializeSeq> { + Err(key_must_be_a_string()) + } + + fn serialize_tuple(self, _len: usize) -> Result<Self::SerializeTuple> { + Err(key_must_be_a_string()) + } + + fn serialize_tuple_struct( + self, + _name: &'static str, + _len: usize, + ) -> Result<Self::SerializeTupleStruct> { + Err(key_must_be_a_string()) + } + + fn serialize_tuple_variant( + self, + _name: &'static str, + _variant_index: u32, + _variant: &'static str, + _len: usize, + ) -> Result<Self::SerializeTupleVariant> { + Err(key_must_be_a_string()) + } + + fn serialize_map(self, _len: Option<usize>) -> Result<Self::SerializeMap> { + Err(key_must_be_a_string()) + } + + fn serialize_struct(self, _name: &'static str, _len: usize) -> Result<Self::SerializeStruct> { + Err(key_must_be_a_string()) + } + + fn serialize_struct_variant( + self, + _name: &'static str, + _variant_index: u32, + _variant: &'static str, + _len: usize, + ) -> Result<Self::SerializeStructVariant> { + Err(key_must_be_a_string()) + } + + fn collect_str<T>(self, value: &T) -> Result<()> + where + T: ?Sized + Display, + { + self.ser.collect_str(value) + } +} + +#[cfg(feature = "arbitrary_precision")] +struct NumberStrEmitter<'a, W: 'a + io::Write, F: 'a + Formatter>(&'a mut Serializer<W, F>); + +#[cfg(feature = "arbitrary_precision")] +impl<'a, W: io::Write, F: Formatter> ser::Serializer for NumberStrEmitter<'a, W, F> { + type Ok = (); + type Error = Error; + + type SerializeSeq = Impossible<(), Error>; + type SerializeTuple = Impossible<(), Error>; + type SerializeTupleStruct = Impossible<(), Error>; + type SerializeTupleVariant = Impossible<(), Error>; + type SerializeMap = Impossible<(), Error>; + type SerializeStruct = Impossible<(), Error>; + type SerializeStructVariant = Impossible<(), Error>; + + fn serialize_bool(self, _v: bool) -> Result<()> { + Err(invalid_number()) + } + + fn serialize_i8(self, _v: i8) -> Result<()> { + Err(invalid_number()) + } + + fn serialize_i16(self, _v: i16) -> Result<()> { + Err(invalid_number()) + } + + fn serialize_i32(self, _v: i32) -> Result<()> { + Err(invalid_number()) + } + + fn serialize_i64(self, _v: i64) -> Result<()> { + Err(invalid_number()) + } + + fn serialize_i128(self, _v: i128) -> Result<()> { + Err(invalid_number()) + } + + fn serialize_u8(self, _v: u8) -> Result<()> { + Err(invalid_number()) + } + + fn serialize_u16(self, _v: u16) -> Result<()> { + Err(invalid_number()) + } + + fn serialize_u32(self, _v: u32) -> Result<()> { + Err(invalid_number()) + } + + fn serialize_u64(self, _v: u64) -> Result<()> { + Err(invalid_number()) + } + + fn serialize_u128(self, _v: u128) -> Result<()> { + Err(invalid_number()) + } + + fn serialize_f32(self, _v: f32) -> Result<()> { + Err(invalid_number()) + } + + fn serialize_f64(self, _v: f64) -> Result<()> { + Err(invalid_number()) + } + + fn serialize_char(self, _v: char) -> Result<()> { + Err(invalid_number()) + } + + fn serialize_str(self, value: &str) -> Result<()> { + let NumberStrEmitter(serializer) = self; + serializer + .formatter + .write_number_str(&mut serializer.writer, value) + .map_err(Error::io) + } + + fn serialize_bytes(self, _value: &[u8]) -> Result<()> { + Err(invalid_number()) + } + + fn serialize_none(self) -> Result<()> { + Err(invalid_number()) + } + + fn serialize_some<T>(self, _value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + Err(invalid_number()) + } + + fn serialize_unit(self) -> Result<()> { + Err(invalid_number()) + } + + fn serialize_unit_struct(self, _name: &'static str) -> Result<()> { + Err(invalid_number()) + } + + fn serialize_unit_variant( + self, + _name: &'static str, + _variant_index: u32, + _variant: &'static str, + ) -> Result<()> { + Err(invalid_number()) + } + + fn serialize_newtype_struct<T>(self, _name: &'static str, _value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + Err(invalid_number()) + } + + fn serialize_newtype_variant<T>( + self, + _name: &'static str, + _variant_index: u32, + _variant: &'static str, + _value: &T, + ) -> Result<()> + where + T: ?Sized + Serialize, + { + Err(invalid_number()) + } + + fn serialize_seq(self, _len: Option<usize>) -> Result<Self::SerializeSeq> { + Err(invalid_number()) + } + + fn serialize_tuple(self, _len: usize) -> Result<Self::SerializeTuple> { + Err(invalid_number()) + } + + fn serialize_tuple_struct( + self, + _name: &'static str, + _len: usize, + ) -> Result<Self::SerializeTupleStruct> { + Err(invalid_number()) + } + + fn serialize_tuple_variant( + self, + _name: &'static str, + _variant_index: u32, + _variant: &'static str, + _len: usize, + ) -> Result<Self::SerializeTupleVariant> { + Err(invalid_number()) + } + + fn serialize_map(self, _len: Option<usize>) -> Result<Self::SerializeMap> { + Err(invalid_number()) + } + + fn serialize_struct(self, _name: &'static str, _len: usize) -> Result<Self::SerializeStruct> { + Err(invalid_number()) + } + + fn serialize_struct_variant( + self, + _name: &'static str, + _variant_index: u32, + _variant: &'static str, + _len: usize, + ) -> Result<Self::SerializeStructVariant> { + Err(invalid_number()) + } +} + +#[cfg(feature = "raw_value")] +struct RawValueStrEmitter<'a, W: 'a + io::Write, F: 'a + Formatter>(&'a mut Serializer<W, F>); + +#[cfg(feature = "raw_value")] +impl<'a, W: io::Write, F: Formatter> ser::Serializer for RawValueStrEmitter<'a, W, F> { + type Ok = (); + type Error = Error; + + type SerializeSeq = Impossible<(), Error>; + type SerializeTuple = Impossible<(), Error>; + type SerializeTupleStruct = Impossible<(), Error>; + type SerializeTupleVariant = Impossible<(), Error>; + type SerializeMap = Impossible<(), Error>; + type SerializeStruct = Impossible<(), Error>; + type SerializeStructVariant = Impossible<(), Error>; + + fn serialize_bool(self, _v: bool) -> Result<()> { + Err(ser::Error::custom("expected RawValue")) + } + + fn serialize_i8(self, _v: i8) -> Result<()> { + Err(ser::Error::custom("expected RawValue")) + } + + fn serialize_i16(self, _v: i16) -> Result<()> { + Err(ser::Error::custom("expected RawValue")) + } + + fn serialize_i32(self, _v: i32) -> Result<()> { + Err(ser::Error::custom("expected RawValue")) + } + + fn serialize_i64(self, _v: i64) -> Result<()> { + Err(ser::Error::custom("expected RawValue")) + } + + fn serialize_i128(self, _v: i128) -> Result<()> { + Err(ser::Error::custom("expected RawValue")) + } + + fn serialize_u8(self, _v: u8) -> Result<()> { + Err(ser::Error::custom("expected RawValue")) + } + + fn serialize_u16(self, _v: u16) -> Result<()> { + Err(ser::Error::custom("expected RawValue")) + } + + fn serialize_u32(self, _v: u32) -> Result<()> { + Err(ser::Error::custom("expected RawValue")) + } + + fn serialize_u64(self, _v: u64) -> Result<()> { + Err(ser::Error::custom("expected RawValue")) + } + + fn serialize_u128(self, _v: u128) -> Result<()> { + Err(ser::Error::custom("expected RawValue")) + } + + fn serialize_f32(self, _v: f32) -> Result<()> { + Err(ser::Error::custom("expected RawValue")) + } + + fn serialize_f64(self, _v: f64) -> Result<()> { + Err(ser::Error::custom("expected RawValue")) + } + + fn serialize_char(self, _v: char) -> Result<()> { + Err(ser::Error::custom("expected RawValue")) + } + + fn serialize_str(self, value: &str) -> Result<()> { + let RawValueStrEmitter(serializer) = self; + serializer + .formatter + .write_raw_fragment(&mut serializer.writer, value) + .map_err(Error::io) + } + + fn serialize_bytes(self, _value: &[u8]) -> Result<()> { + Err(ser::Error::custom("expected RawValue")) + } + + fn serialize_none(self) -> Result<()> { + Err(ser::Error::custom("expected RawValue")) + } + + fn serialize_some<T>(self, _value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + Err(ser::Error::custom("expected RawValue")) + } + + fn serialize_unit(self) -> Result<()> { + Err(ser::Error::custom("expected RawValue")) + } + + fn serialize_unit_struct(self, _name: &'static str) -> Result<()> { + Err(ser::Error::custom("expected RawValue")) + } + + fn serialize_unit_variant( + self, + _name: &'static str, + _variant_index: u32, + _variant: &'static str, + ) -> Result<()> { + Err(ser::Error::custom("expected RawValue")) + } + + fn serialize_newtype_struct<T>(self, _name: &'static str, _value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + Err(ser::Error::custom("expected RawValue")) + } + + fn serialize_newtype_variant<T>( + self, + _name: &'static str, + _variant_index: u32, + _variant: &'static str, + _value: &T, + ) -> Result<()> + where + T: ?Sized + Serialize, + { + Err(ser::Error::custom("expected RawValue")) + } + + fn serialize_seq(self, _len: Option<usize>) -> Result<Self::SerializeSeq> { + Err(ser::Error::custom("expected RawValue")) + } + + fn serialize_tuple(self, _len: usize) -> Result<Self::SerializeTuple> { + Err(ser::Error::custom("expected RawValue")) + } + + fn serialize_tuple_struct( + self, + _name: &'static str, + _len: usize, + ) -> Result<Self::SerializeTupleStruct> { + Err(ser::Error::custom("expected RawValue")) + } + + fn serialize_tuple_variant( + self, + _name: &'static str, + _variant_index: u32, + _variant: &'static str, + _len: usize, + ) -> Result<Self::SerializeTupleVariant> { + Err(ser::Error::custom("expected RawValue")) + } + + fn serialize_map(self, _len: Option<usize>) -> Result<Self::SerializeMap> { + Err(ser::Error::custom("expected RawValue")) + } + + fn serialize_struct(self, _name: &'static str, _len: usize) -> Result<Self::SerializeStruct> { + Err(ser::Error::custom("expected RawValue")) + } + + fn serialize_struct_variant( + self, + _name: &'static str, + _variant_index: u32, + _variant: &'static str, + _len: usize, + ) -> Result<Self::SerializeStructVariant> { + Err(ser::Error::custom("expected RawValue")) + } + + fn collect_str<T>(self, value: &T) -> Result<Self::Ok> + where + T: ?Sized + Display, + { + self.serialize_str(&value.to_string()) + } +} + +/// Represents a character escape code in a type-safe manner. +pub enum CharEscape { + /// An escaped quote `"` + Quote, + /// An escaped reverse solidus `\` + ReverseSolidus, + /// An escaped solidus `/` + Solidus, + /// An escaped backspace character (usually escaped as `\b`) + Backspace, + /// An escaped form feed character (usually escaped as `\f`) + FormFeed, + /// An escaped line feed character (usually escaped as `\n`) + LineFeed, + /// An escaped carriage return character (usually escaped as `\r`) + CarriageReturn, + /// An escaped tab character (usually escaped as `\t`) + Tab, + /// An escaped ASCII plane control character (usually escaped as + /// `\u00XX` where `XX` are two hex characters) + AsciiControl(u8), +} + +impl CharEscape { + #[inline] + fn from_escape_table(escape: u8, byte: u8) -> CharEscape { + match escape { + self::BB => CharEscape::Backspace, + self::TT => CharEscape::Tab, + self::NN => CharEscape::LineFeed, + self::FF => CharEscape::FormFeed, + self::RR => CharEscape::CarriageReturn, + self::QU => CharEscape::Quote, + self::BS => CharEscape::ReverseSolidus, + self::UU => CharEscape::AsciiControl(byte), + _ => unreachable!(), + } + } +} + +/// This trait abstracts away serializing the JSON control characters, which allows the user to +/// optionally pretty print the JSON output. +pub trait Formatter { + /// Writes a `null` value to the specified writer. + #[inline] + fn write_null<W>(&mut self, writer: &mut W) -> io::Result<()> + where + W: ?Sized + io::Write, + { + writer.write_all(b"null") + } + + /// Writes a `true` or `false` value to the specified writer. + #[inline] + fn write_bool<W>(&mut self, writer: &mut W, value: bool) -> io::Result<()> + where + W: ?Sized + io::Write, + { + let s = if value { + b"true" as &[u8] + } else { + b"false" as &[u8] + }; + writer.write_all(s) + } + + /// Writes an integer value like `-123` to the specified writer. + #[inline] + fn write_i8<W>(&mut self, writer: &mut W, value: i8) -> io::Result<()> + where + W: ?Sized + io::Write, + { + let mut buffer = itoa::Buffer::new(); + let s = buffer.format(value); + writer.write_all(s.as_bytes()) + } + + /// Writes an integer value like `-123` to the specified writer. + #[inline] + fn write_i16<W>(&mut self, writer: &mut W, value: i16) -> io::Result<()> + where + W: ?Sized + io::Write, + { + let mut buffer = itoa::Buffer::new(); + let s = buffer.format(value); + writer.write_all(s.as_bytes()) + } + + /// Writes an integer value like `-123` to the specified writer. + #[inline] + fn write_i32<W>(&mut self, writer: &mut W, value: i32) -> io::Result<()> + where + W: ?Sized + io::Write, + { + let mut buffer = itoa::Buffer::new(); + let s = buffer.format(value); + writer.write_all(s.as_bytes()) + } + + /// Writes an integer value like `-123` to the specified writer. + #[inline] + fn write_i64<W>(&mut self, writer: &mut W, value: i64) -> io::Result<()> + where + W: ?Sized + io::Write, + { + let mut buffer = itoa::Buffer::new(); + let s = buffer.format(value); + writer.write_all(s.as_bytes()) + } + + /// Writes an integer value like `-123` to the specified writer. + #[inline] + fn write_i128<W>(&mut self, writer: &mut W, value: i128) -> io::Result<()> + where + W: ?Sized + io::Write, + { + let mut buffer = itoa::Buffer::new(); + let s = buffer.format(value); + writer.write_all(s.as_bytes()) + } + + /// Writes an integer value like `123` to the specified writer. + #[inline] + fn write_u8<W>(&mut self, writer: &mut W, value: u8) -> io::Result<()> + where + W: ?Sized + io::Write, + { + let mut buffer = itoa::Buffer::new(); + let s = buffer.format(value); + writer.write_all(s.as_bytes()) + } + + /// Writes an integer value like `123` to the specified writer. + #[inline] + fn write_u16<W>(&mut self, writer: &mut W, value: u16) -> io::Result<()> + where + W: ?Sized + io::Write, + { + let mut buffer = itoa::Buffer::new(); + let s = buffer.format(value); + writer.write_all(s.as_bytes()) + } + + /// Writes an integer value like `123` to the specified writer. + #[inline] + fn write_u32<W>(&mut self, writer: &mut W, value: u32) -> io::Result<()> + where + W: ?Sized + io::Write, + { + let mut buffer = itoa::Buffer::new(); + let s = buffer.format(value); + writer.write_all(s.as_bytes()) + } + + /// Writes an integer value like `123` to the specified writer. + #[inline] + fn write_u64<W>(&mut self, writer: &mut W, value: u64) -> io::Result<()> + where + W: ?Sized + io::Write, + { + let mut buffer = itoa::Buffer::new(); + let s = buffer.format(value); + writer.write_all(s.as_bytes()) + } + + /// Writes an integer value like `123` to the specified writer. + #[inline] + fn write_u128<W>(&mut self, writer: &mut W, value: u128) -> io::Result<()> + where + W: ?Sized + io::Write, + { + let mut buffer = itoa::Buffer::new(); + let s = buffer.format(value); + writer.write_all(s.as_bytes()) + } + + /// Writes a floating point value like `-31.26e+12` to the specified writer. + #[inline] + fn write_f32<W>(&mut self, writer: &mut W, value: f32) -> io::Result<()> + where + W: ?Sized + io::Write, + { + let mut buffer = ryu::Buffer::new(); + let s = buffer.format_finite(value); + writer.write_all(s.as_bytes()) + } + + /// Writes a floating point value like `-31.26e+12` to the specified writer. + #[inline] + fn write_f64<W>(&mut self, writer: &mut W, value: f64) -> io::Result<()> + where + W: ?Sized + io::Write, + { + let mut buffer = ryu::Buffer::new(); + let s = buffer.format_finite(value); + writer.write_all(s.as_bytes()) + } + + /// Writes a number that has already been rendered to a string. + #[inline] + fn write_number_str<W>(&mut self, writer: &mut W, value: &str) -> io::Result<()> + where + W: ?Sized + io::Write, + { + writer.write_all(value.as_bytes()) + } + + /// Called before each series of `write_string_fragment` and + /// `write_char_escape`. Writes a `"` to the specified writer. + #[inline] + fn begin_string<W>(&mut self, writer: &mut W) -> io::Result<()> + where + W: ?Sized + io::Write, + { + writer.write_all(b"\"") + } + + /// Called after each series of `write_string_fragment` and + /// `write_char_escape`. Writes a `"` to the specified writer. + #[inline] + fn end_string<W>(&mut self, writer: &mut W) -> io::Result<()> + where + W: ?Sized + io::Write, + { + writer.write_all(b"\"") + } + + /// Writes a string fragment that doesn't need any escaping to the + /// specified writer. + #[inline] + fn write_string_fragment<W>(&mut self, writer: &mut W, fragment: &str) -> io::Result<()> + where + W: ?Sized + io::Write, + { + writer.write_all(fragment.as_bytes()) + } + + /// Writes a character escape code to the specified writer. + #[inline] + fn write_char_escape<W>(&mut self, writer: &mut W, char_escape: CharEscape) -> io::Result<()> + where + W: ?Sized + io::Write, + { + use self::CharEscape::*; + + let s = match char_escape { + Quote => b"\\\"", + ReverseSolidus => b"\\\\", + Solidus => b"\\/", + Backspace => b"\\b", + FormFeed => b"\\f", + LineFeed => b"\\n", + CarriageReturn => b"\\r", + Tab => b"\\t", + AsciiControl(byte) => { + static HEX_DIGITS: [u8; 16] = *b"0123456789abcdef"; + let bytes = &[ + b'\\', + b'u', + b'0', + b'0', + HEX_DIGITS[(byte >> 4) as usize], + HEX_DIGITS[(byte & 0xF) as usize], + ]; + return writer.write_all(bytes); + } + }; + + writer.write_all(s) + } + + /// Called before every array. Writes a `[` to the specified + /// writer. + #[inline] + fn begin_array<W>(&mut self, writer: &mut W) -> io::Result<()> + where + W: ?Sized + io::Write, + { + writer.write_all(b"[") + } + + /// Called after every array. Writes a `]` to the specified + /// writer. + #[inline] + fn end_array<W>(&mut self, writer: &mut W) -> io::Result<()> + where + W: ?Sized + io::Write, + { + writer.write_all(b"]") + } + + /// Called before every array value. Writes a `,` if needed to + /// the specified writer. + #[inline] + fn begin_array_value<W>(&mut self, writer: &mut W, first: bool) -> io::Result<()> + where + W: ?Sized + io::Write, + { + if first { + Ok(()) + } else { + writer.write_all(b",") + } + } + + /// Called after every array value. + #[inline] + fn end_array_value<W>(&mut self, _writer: &mut W) -> io::Result<()> + where + W: ?Sized + io::Write, + { + Ok(()) + } + + /// Called before every object. Writes a `{` to the specified + /// writer. + #[inline] + fn begin_object<W>(&mut self, writer: &mut W) -> io::Result<()> + where + W: ?Sized + io::Write, + { + writer.write_all(b"{") + } + + /// Called after every object. Writes a `}` to the specified + /// writer. + #[inline] + fn end_object<W>(&mut self, writer: &mut W) -> io::Result<()> + where + W: ?Sized + io::Write, + { + writer.write_all(b"}") + } + + /// Called before every object key. + #[inline] + fn begin_object_key<W>(&mut self, writer: &mut W, first: bool) -> io::Result<()> + where + W: ?Sized + io::Write, + { + if first { + Ok(()) + } else { + writer.write_all(b",") + } + } + + /// Called after every object key. A `:` should be written to the + /// specified writer by either this method or + /// `begin_object_value`. + #[inline] + fn end_object_key<W>(&mut self, _writer: &mut W) -> io::Result<()> + where + W: ?Sized + io::Write, + { + Ok(()) + } + + /// Called before every object value. A `:` should be written to + /// the specified writer by either this method or + /// `end_object_key`. + #[inline] + fn begin_object_value<W>(&mut self, writer: &mut W) -> io::Result<()> + where + W: ?Sized + io::Write, + { + writer.write_all(b":") + } + + /// Called after every object value. + #[inline] + fn end_object_value<W>(&mut self, _writer: &mut W) -> io::Result<()> + where + W: ?Sized + io::Write, + { + Ok(()) + } + + /// Writes a raw JSON fragment that doesn't need any escaping to the + /// specified writer. + #[inline] + fn write_raw_fragment<W>(&mut self, writer: &mut W, fragment: &str) -> io::Result<()> + where + W: ?Sized + io::Write, + { + writer.write_all(fragment.as_bytes()) + } +} + +/// This structure compacts a JSON value with no extra whitespace. +#[derive(Clone, Debug)] +pub struct CompactFormatter; + +impl Formatter for CompactFormatter {} + +/// This structure pretty prints a JSON value to make it human readable. +#[derive(Clone, Debug)] +pub struct PrettyFormatter<'a> { + current_indent: usize, + has_value: bool, + indent: &'a [u8], +} + +impl<'a> PrettyFormatter<'a> { + /// Construct a pretty printer formatter that defaults to using two spaces for indentation. + pub fn new() -> Self { + PrettyFormatter::with_indent(b" ") + } + + /// Construct a pretty printer formatter that uses the `indent` string for indentation. + pub fn with_indent(indent: &'a [u8]) -> Self { + PrettyFormatter { + current_indent: 0, + has_value: false, + indent, + } + } +} + +impl<'a> Default for PrettyFormatter<'a> { + fn default() -> Self { + PrettyFormatter::new() + } +} + +impl<'a> Formatter for PrettyFormatter<'a> { + #[inline] + fn begin_array<W>(&mut self, writer: &mut W) -> io::Result<()> + where + W: ?Sized + io::Write, + { + self.current_indent += 1; + self.has_value = false; + writer.write_all(b"[") + } + + #[inline] + fn end_array<W>(&mut self, writer: &mut W) -> io::Result<()> + where + W: ?Sized + io::Write, + { + self.current_indent -= 1; + + if self.has_value { + tri!(writer.write_all(b"\n")); + tri!(indent(writer, self.current_indent, self.indent)); + } + + writer.write_all(b"]") + } + + #[inline] + fn begin_array_value<W>(&mut self, writer: &mut W, first: bool) -> io::Result<()> + where + W: ?Sized + io::Write, + { + tri!(writer.write_all(if first { b"\n" } else { b",\n" })); + indent(writer, self.current_indent, self.indent) + } + + #[inline] + fn end_array_value<W>(&mut self, _writer: &mut W) -> io::Result<()> + where + W: ?Sized + io::Write, + { + self.has_value = true; + Ok(()) + } + + #[inline] + fn begin_object<W>(&mut self, writer: &mut W) -> io::Result<()> + where + W: ?Sized + io::Write, + { + self.current_indent += 1; + self.has_value = false; + writer.write_all(b"{") + } + + #[inline] + fn end_object<W>(&mut self, writer: &mut W) -> io::Result<()> + where + W: ?Sized + io::Write, + { + self.current_indent -= 1; + + if self.has_value { + tri!(writer.write_all(b"\n")); + tri!(indent(writer, self.current_indent, self.indent)); + } + + writer.write_all(b"}") + } + + #[inline] + fn begin_object_key<W>(&mut self, writer: &mut W, first: bool) -> io::Result<()> + where + W: ?Sized + io::Write, + { + tri!(writer.write_all(if first { b"\n" } else { b",\n" })); + indent(writer, self.current_indent, self.indent) + } + + #[inline] + fn begin_object_value<W>(&mut self, writer: &mut W) -> io::Result<()> + where + W: ?Sized + io::Write, + { + writer.write_all(b": ") + } + + #[inline] + fn end_object_value<W>(&mut self, _writer: &mut W) -> io::Result<()> + where + W: ?Sized + io::Write, + { + self.has_value = true; + Ok(()) + } +} + +fn format_escaped_str<W, F>(writer: &mut W, formatter: &mut F, value: &str) -> io::Result<()> +where + W: ?Sized + io::Write, + F: ?Sized + Formatter, +{ + tri!(formatter.begin_string(writer)); + tri!(format_escaped_str_contents(writer, formatter, value)); + formatter.end_string(writer) +} + +fn format_escaped_str_contents<W, F>( + writer: &mut W, + formatter: &mut F, + value: &str, +) -> io::Result<()> +where + W: ?Sized + io::Write, + F: ?Sized + Formatter, +{ + let bytes = value.as_bytes(); + + let mut start = 0; + + for (i, &byte) in bytes.iter().enumerate() { + let escape = ESCAPE[byte as usize]; + if escape == 0 { + continue; + } + + if start < i { + tri!(formatter.write_string_fragment(writer, &value[start..i])); + } + + let char_escape = CharEscape::from_escape_table(escape, byte); + tri!(formatter.write_char_escape(writer, char_escape)); + + start = i + 1; + } + + if start == bytes.len() { + return Ok(()); + } + + formatter.write_string_fragment(writer, &value[start..]) +} + +const BB: u8 = b'b'; // \x08 +const TT: u8 = b't'; // \x09 +const NN: u8 = b'n'; // \x0A +const FF: u8 = b'f'; // \x0C +const RR: u8 = b'r'; // \x0D +const QU: u8 = b'"'; // \x22 +const BS: u8 = b'\\'; // \x5C +const UU: u8 = b'u'; // \x00...\x1F except the ones above +const __: u8 = 0; + +// Lookup table of escape sequences. A value of b'x' at index i means that byte +// i is escaped as "\x" in JSON. A value of 0 means that byte i is not escaped. +static ESCAPE: [u8; 256] = [ + // 1 2 3 4 5 6 7 8 9 A B C D E F + UU, UU, UU, UU, UU, UU, UU, UU, BB, TT, NN, UU, FF, RR, UU, UU, // 0 + UU, UU, UU, UU, UU, UU, UU, UU, UU, UU, UU, UU, UU, UU, UU, UU, // 1 + __, __, QU, __, __, __, __, __, __, __, __, __, __, __, __, __, // 2 + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // 3 + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // 4 + __, __, __, __, __, __, __, __, __, __, __, __, BS, __, __, __, // 5 + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // 6 + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // 7 + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // 8 + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // 9 + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // A + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // B + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // C + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // D + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // E + __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, __, // F +]; + +/// Serialize the given data structure as JSON into the IO stream. +/// +/// # Errors +/// +/// Serialization can fail if `T`'s implementation of `Serialize` decides to +/// fail, or if `T` contains a map with non-string keys. +#[inline] +#[cfg_attr(docsrs, doc(cfg(feature = "std")))] +pub fn to_writer<W, T>(writer: W, value: &T) -> Result<()> +where + W: io::Write, + T: ?Sized + Serialize, +{ + let mut ser = Serializer::new(writer); + value.serialize(&mut ser) +} + +/// Serialize the given data structure as pretty-printed JSON into the IO +/// stream. +/// +/// # Errors +/// +/// Serialization can fail if `T`'s implementation of `Serialize` decides to +/// fail, or if `T` contains a map with non-string keys. +#[inline] +#[cfg_attr(docsrs, doc(cfg(feature = "std")))] +pub fn to_writer_pretty<W, T>(writer: W, value: &T) -> Result<()> +where + W: io::Write, + T: ?Sized + Serialize, +{ + let mut ser = Serializer::pretty(writer); + value.serialize(&mut ser) +} + +/// Serialize the given data structure as a JSON byte vector. +/// +/// # Errors +/// +/// Serialization can fail if `T`'s implementation of `Serialize` decides to +/// fail, or if `T` contains a map with non-string keys. +#[inline] +pub fn to_vec<T>(value: &T) -> Result<Vec<u8>> +where + T: ?Sized + Serialize, +{ + let mut writer = Vec::with_capacity(128); + tri!(to_writer(&mut writer, value)); + Ok(writer) +} + +/// Serialize the given data structure as a pretty-printed JSON byte vector. +/// +/// # Errors +/// +/// Serialization can fail if `T`'s implementation of `Serialize` decides to +/// fail, or if `T` contains a map with non-string keys. +#[inline] +pub fn to_vec_pretty<T>(value: &T) -> Result<Vec<u8>> +where + T: ?Sized + Serialize, +{ + let mut writer = Vec::with_capacity(128); + tri!(to_writer_pretty(&mut writer, value)); + Ok(writer) +} + +/// Serialize the given data structure as a String of JSON. +/// +/// # Errors +/// +/// Serialization can fail if `T`'s implementation of `Serialize` decides to +/// fail, or if `T` contains a map with non-string keys. +#[inline] +pub fn to_string<T>(value: &T) -> Result<String> +where + T: ?Sized + Serialize, +{ + let vec = tri!(to_vec(value)); + let string = unsafe { + // We do not emit invalid UTF-8. + String::from_utf8_unchecked(vec) + }; + Ok(string) +} + +/// Serialize the given data structure as a pretty-printed String of JSON. +/// +/// # Errors +/// +/// Serialization can fail if `T`'s implementation of `Serialize` decides to +/// fail, or if `T` contains a map with non-string keys. +#[inline] +pub fn to_string_pretty<T>(value: &T) -> Result<String> +where + T: ?Sized + Serialize, +{ + let vec = tri!(to_vec_pretty(value)); + let string = unsafe { + // We do not emit invalid UTF-8. + String::from_utf8_unchecked(vec) + }; + Ok(string) +} + +fn indent<W>(wr: &mut W, n: usize, s: &[u8]) -> io::Result<()> +where + W: ?Sized + io::Write, +{ + for _ in 0..n { + tri!(wr.write_all(s)); + } + + Ok(()) +} diff --git a/third_party/rust/serde_json/src/value/de.rs b/third_party/rust/serde_json/src/value/de.rs new file mode 100644 index 0000000000..9c266d08ae --- /dev/null +++ b/third_party/rust/serde_json/src/value/de.rs @@ -0,0 +1,1377 @@ +use crate::error::Error; +use crate::map::Map; +use crate::number::Number; +use crate::value::Value; +use alloc::borrow::{Cow, ToOwned}; +use alloc::string::String; +#[cfg(feature = "raw_value")] +use alloc::string::ToString; +use alloc::vec::{self, Vec}; +use core::fmt; +use core::slice; +use core::str::FromStr; +use serde::de::{ + self, Deserialize, DeserializeSeed, EnumAccess, Expected, IntoDeserializer, MapAccess, + SeqAccess, Unexpected, VariantAccess, Visitor, +}; +use serde::forward_to_deserialize_any; + +#[cfg(feature = "arbitrary_precision")] +use crate::number::NumberFromString; + +impl<'de> Deserialize<'de> for Value { + #[inline] + fn deserialize<D>(deserializer: D) -> Result<Value, D::Error> + where + D: serde::Deserializer<'de>, + { + struct ValueVisitor; + + impl<'de> Visitor<'de> for ValueVisitor { + type Value = Value; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("any valid JSON value") + } + + #[inline] + fn visit_bool<E>(self, value: bool) -> Result<Value, E> { + Ok(Value::Bool(value)) + } + + #[inline] + fn visit_i64<E>(self, value: i64) -> Result<Value, E> { + Ok(Value::Number(value.into())) + } + + #[inline] + fn visit_u64<E>(self, value: u64) -> Result<Value, E> { + Ok(Value::Number(value.into())) + } + + #[inline] + fn visit_f64<E>(self, value: f64) -> Result<Value, E> { + Ok(Number::from_f64(value).map_or(Value::Null, Value::Number)) + } + + #[cfg(any(feature = "std", feature = "alloc"))] + #[inline] + fn visit_str<E>(self, value: &str) -> Result<Value, E> + where + E: serde::de::Error, + { + self.visit_string(String::from(value)) + } + + #[cfg(any(feature = "std", feature = "alloc"))] + #[inline] + fn visit_string<E>(self, value: String) -> Result<Value, E> { + Ok(Value::String(value)) + } + + #[inline] + fn visit_none<E>(self) -> Result<Value, E> { + Ok(Value::Null) + } + + #[inline] + fn visit_some<D>(self, deserializer: D) -> Result<Value, D::Error> + where + D: serde::Deserializer<'de>, + { + Deserialize::deserialize(deserializer) + } + + #[inline] + fn visit_unit<E>(self) -> Result<Value, E> { + Ok(Value::Null) + } + + #[inline] + fn visit_seq<V>(self, mut visitor: V) -> Result<Value, V::Error> + where + V: SeqAccess<'de>, + { + let mut vec = Vec::new(); + + while let Some(elem) = tri!(visitor.next_element()) { + vec.push(elem); + } + + Ok(Value::Array(vec)) + } + + #[cfg(any(feature = "std", feature = "alloc"))] + fn visit_map<V>(self, mut visitor: V) -> Result<Value, V::Error> + where + V: MapAccess<'de>, + { + match visitor.next_key_seed(KeyClassifier)? { + #[cfg(feature = "arbitrary_precision")] + Some(KeyClass::Number) => { + let number: NumberFromString = visitor.next_value()?; + Ok(Value::Number(number.value)) + } + #[cfg(feature = "raw_value")] + Some(KeyClass::RawValue) => { + let value = visitor.next_value_seed(crate::raw::BoxedFromString)?; + crate::from_str(value.get()).map_err(de::Error::custom) + } + Some(KeyClass::Map(first_key)) => { + let mut values = Map::new(); + + values.insert(first_key, tri!(visitor.next_value())); + while let Some((key, value)) = tri!(visitor.next_entry()) { + values.insert(key, value); + } + + Ok(Value::Object(values)) + } + None => Ok(Value::Object(Map::new())), + } + } + } + + deserializer.deserialize_any(ValueVisitor) + } +} + +impl FromStr for Value { + type Err = Error; + fn from_str(s: &str) -> Result<Value, Error> { + super::super::de::from_str(s) + } +} + +macro_rules! deserialize_number { + ($method:ident) => { + #[cfg(not(feature = "arbitrary_precision"))] + fn $method<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + match self { + Value::Number(n) => n.deserialize_any(visitor), + _ => Err(self.invalid_type(&visitor)), + } + } + + #[cfg(feature = "arbitrary_precision")] + fn $method<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + match self { + Value::Number(n) => n.$method(visitor), + _ => self.deserialize_any(visitor), + } + } + }; +} + +fn visit_array<'de, V>(array: Vec<Value>, visitor: V) -> Result<V::Value, Error> +where + V: Visitor<'de>, +{ + let len = array.len(); + let mut deserializer = SeqDeserializer::new(array); + let seq = tri!(visitor.visit_seq(&mut deserializer)); + let remaining = deserializer.iter.len(); + if remaining == 0 { + Ok(seq) + } else { + Err(serde::de::Error::invalid_length( + len, + &"fewer elements in array", + )) + } +} + +fn visit_object<'de, V>(object: Map<String, Value>, visitor: V) -> Result<V::Value, Error> +where + V: Visitor<'de>, +{ + let len = object.len(); + let mut deserializer = MapDeserializer::new(object); + let map = tri!(visitor.visit_map(&mut deserializer)); + let remaining = deserializer.iter.len(); + if remaining == 0 { + Ok(map) + } else { + Err(serde::de::Error::invalid_length( + len, + &"fewer elements in map", + )) + } +} + +impl<'de> serde::Deserializer<'de> for Value { + type Error = Error; + + #[inline] + fn deserialize_any<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + match self { + Value::Null => visitor.visit_unit(), + Value::Bool(v) => visitor.visit_bool(v), + Value::Number(n) => n.deserialize_any(visitor), + #[cfg(any(feature = "std", feature = "alloc"))] + Value::String(v) => visitor.visit_string(v), + Value::Array(v) => visit_array(v, visitor), + Value::Object(v) => visit_object(v, visitor), + } + } + + deserialize_number!(deserialize_i8); + deserialize_number!(deserialize_i16); + deserialize_number!(deserialize_i32); + deserialize_number!(deserialize_i64); + deserialize_number!(deserialize_i128); + deserialize_number!(deserialize_u8); + deserialize_number!(deserialize_u16); + deserialize_number!(deserialize_u32); + deserialize_number!(deserialize_u64); + deserialize_number!(deserialize_u128); + deserialize_number!(deserialize_f32); + deserialize_number!(deserialize_f64); + + #[inline] + fn deserialize_option<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + match self { + Value::Null => visitor.visit_none(), + _ => visitor.visit_some(self), + } + } + + #[inline] + fn deserialize_enum<V>( + self, + _name: &str, + _variants: &'static [&'static str], + visitor: V, + ) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + let (variant, value) = match self { + Value::Object(value) => { + let mut iter = value.into_iter(); + let (variant, value) = match iter.next() { + Some(v) => v, + None => { + return Err(serde::de::Error::invalid_value( + Unexpected::Map, + &"map with a single key", + )); + } + }; + // enums are encoded in json as maps with a single key:value pair + if iter.next().is_some() { + return Err(serde::de::Error::invalid_value( + Unexpected::Map, + &"map with a single key", + )); + } + (variant, Some(value)) + } + Value::String(variant) => (variant, None), + other => { + return Err(serde::de::Error::invalid_type( + other.unexpected(), + &"string or map", + )); + } + }; + + visitor.visit_enum(EnumDeserializer { variant, value }) + } + + #[inline] + fn deserialize_newtype_struct<V>( + self, + name: &'static str, + visitor: V, + ) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + #[cfg(feature = "raw_value")] + { + if name == crate::raw::TOKEN { + return visitor.visit_map(crate::raw::OwnedRawDeserializer { + raw_value: Some(self.to_string()), + }); + } + } + + let _ = name; + visitor.visit_newtype_struct(self) + } + + fn deserialize_bool<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + match self { + Value::Bool(v) => visitor.visit_bool(v), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_char<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + self.deserialize_string(visitor) + } + + fn deserialize_str<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + self.deserialize_string(visitor) + } + + fn deserialize_string<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + match self { + #[cfg(any(feature = "std", feature = "alloc"))] + Value::String(v) => visitor.visit_string(v), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_bytes<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + self.deserialize_byte_buf(visitor) + } + + fn deserialize_byte_buf<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + match self { + #[cfg(any(feature = "std", feature = "alloc"))] + Value::String(v) => visitor.visit_string(v), + Value::Array(v) => visit_array(v, visitor), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_unit<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + match self { + Value::Null => visitor.visit_unit(), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_unit_struct<V>(self, _name: &'static str, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + self.deserialize_unit(visitor) + } + + fn deserialize_seq<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + match self { + Value::Array(v) => visit_array(v, visitor), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_tuple<V>(self, _len: usize, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + self.deserialize_seq(visitor) + } + + fn deserialize_tuple_struct<V>( + self, + _name: &'static str, + _len: usize, + visitor: V, + ) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + self.deserialize_seq(visitor) + } + + fn deserialize_map<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + match self { + Value::Object(v) => visit_object(v, visitor), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_struct<V>( + self, + _name: &'static str, + _fields: &'static [&'static str], + visitor: V, + ) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + match self { + Value::Array(v) => visit_array(v, visitor), + Value::Object(v) => visit_object(v, visitor), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_identifier<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + self.deserialize_string(visitor) + } + + fn deserialize_ignored_any<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + drop(self); + visitor.visit_unit() + } +} + +struct EnumDeserializer { + variant: String, + value: Option<Value>, +} + +impl<'de> EnumAccess<'de> for EnumDeserializer { + type Error = Error; + type Variant = VariantDeserializer; + + fn variant_seed<V>(self, seed: V) -> Result<(V::Value, VariantDeserializer), Error> + where + V: DeserializeSeed<'de>, + { + let variant = self.variant.into_deserializer(); + let visitor = VariantDeserializer { value: self.value }; + seed.deserialize(variant).map(|v| (v, visitor)) + } +} + +impl<'de> IntoDeserializer<'de, Error> for Value { + type Deserializer = Self; + + fn into_deserializer(self) -> Self::Deserializer { + self + } +} + +struct VariantDeserializer { + value: Option<Value>, +} + +impl<'de> VariantAccess<'de> for VariantDeserializer { + type Error = Error; + + fn unit_variant(self) -> Result<(), Error> { + match self.value { + Some(value) => Deserialize::deserialize(value), + None => Ok(()), + } + } + + fn newtype_variant_seed<T>(self, seed: T) -> Result<T::Value, Error> + where + T: DeserializeSeed<'de>, + { + match self.value { + Some(value) => seed.deserialize(value), + None => Err(serde::de::Error::invalid_type( + Unexpected::UnitVariant, + &"newtype variant", + )), + } + } + + fn tuple_variant<V>(self, _len: usize, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + match self.value { + Some(Value::Array(v)) => { + if v.is_empty() { + visitor.visit_unit() + } else { + visit_array(v, visitor) + } + } + Some(other) => Err(serde::de::Error::invalid_type( + other.unexpected(), + &"tuple variant", + )), + None => Err(serde::de::Error::invalid_type( + Unexpected::UnitVariant, + &"tuple variant", + )), + } + } + + fn struct_variant<V>( + self, + _fields: &'static [&'static str], + visitor: V, + ) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + match self.value { + Some(Value::Object(v)) => visit_object(v, visitor), + Some(other) => Err(serde::de::Error::invalid_type( + other.unexpected(), + &"struct variant", + )), + None => Err(serde::de::Error::invalid_type( + Unexpected::UnitVariant, + &"struct variant", + )), + } + } +} + +struct SeqDeserializer { + iter: vec::IntoIter<Value>, +} + +impl SeqDeserializer { + fn new(vec: Vec<Value>) -> Self { + SeqDeserializer { + iter: vec.into_iter(), + } + } +} + +impl<'de> SeqAccess<'de> for SeqDeserializer { + type Error = Error; + + fn next_element_seed<T>(&mut self, seed: T) -> Result<Option<T::Value>, Error> + where + T: DeserializeSeed<'de>, + { + match self.iter.next() { + Some(value) => seed.deserialize(value).map(Some), + None => Ok(None), + } + } + + fn size_hint(&self) -> Option<usize> { + match self.iter.size_hint() { + (lower, Some(upper)) if lower == upper => Some(upper), + _ => None, + } + } +} + +struct MapDeserializer { + iter: <Map<String, Value> as IntoIterator>::IntoIter, + value: Option<Value>, +} + +impl MapDeserializer { + fn new(map: Map<String, Value>) -> Self { + MapDeserializer { + iter: map.into_iter(), + value: None, + } + } +} + +impl<'de> MapAccess<'de> for MapDeserializer { + type Error = Error; + + fn next_key_seed<T>(&mut self, seed: T) -> Result<Option<T::Value>, Error> + where + T: DeserializeSeed<'de>, + { + match self.iter.next() { + Some((key, value)) => { + self.value = Some(value); + let key_de = MapKeyDeserializer { + key: Cow::Owned(key), + }; + seed.deserialize(key_de).map(Some) + } + None => Ok(None), + } + } + + fn next_value_seed<T>(&mut self, seed: T) -> Result<T::Value, Error> + where + T: DeserializeSeed<'de>, + { + match self.value.take() { + Some(value) => seed.deserialize(value), + None => Err(serde::de::Error::custom("value is missing")), + } + } + + fn size_hint(&self) -> Option<usize> { + match self.iter.size_hint() { + (lower, Some(upper)) if lower == upper => Some(upper), + _ => None, + } + } +} + +macro_rules! deserialize_value_ref_number { + ($method:ident) => { + #[cfg(not(feature = "arbitrary_precision"))] + fn $method<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + match self { + Value::Number(n) => n.deserialize_any(visitor), + _ => Err(self.invalid_type(&visitor)), + } + } + + #[cfg(feature = "arbitrary_precision")] + fn $method<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + match self { + Value::Number(n) => n.$method(visitor), + _ => self.deserialize_any(visitor), + } + } + }; +} + +fn visit_array_ref<'de, V>(array: &'de [Value], visitor: V) -> Result<V::Value, Error> +where + V: Visitor<'de>, +{ + let len = array.len(); + let mut deserializer = SeqRefDeserializer::new(array); + let seq = tri!(visitor.visit_seq(&mut deserializer)); + let remaining = deserializer.iter.len(); + if remaining == 0 { + Ok(seq) + } else { + Err(serde::de::Error::invalid_length( + len, + &"fewer elements in array", + )) + } +} + +fn visit_object_ref<'de, V>(object: &'de Map<String, Value>, visitor: V) -> Result<V::Value, Error> +where + V: Visitor<'de>, +{ + let len = object.len(); + let mut deserializer = MapRefDeserializer::new(object); + let map = tri!(visitor.visit_map(&mut deserializer)); + let remaining = deserializer.iter.len(); + if remaining == 0 { + Ok(map) + } else { + Err(serde::de::Error::invalid_length( + len, + &"fewer elements in map", + )) + } +} + +impl<'de> serde::Deserializer<'de> for &'de Value { + type Error = Error; + + fn deserialize_any<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + match self { + Value::Null => visitor.visit_unit(), + Value::Bool(v) => visitor.visit_bool(*v), + Value::Number(n) => n.deserialize_any(visitor), + Value::String(v) => visitor.visit_borrowed_str(v), + Value::Array(v) => visit_array_ref(v, visitor), + Value::Object(v) => visit_object_ref(v, visitor), + } + } + + deserialize_value_ref_number!(deserialize_i8); + deserialize_value_ref_number!(deserialize_i16); + deserialize_value_ref_number!(deserialize_i32); + deserialize_value_ref_number!(deserialize_i64); + deserialize_number!(deserialize_i128); + deserialize_value_ref_number!(deserialize_u8); + deserialize_value_ref_number!(deserialize_u16); + deserialize_value_ref_number!(deserialize_u32); + deserialize_value_ref_number!(deserialize_u64); + deserialize_number!(deserialize_u128); + deserialize_value_ref_number!(deserialize_f32); + deserialize_value_ref_number!(deserialize_f64); + + fn deserialize_option<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + match *self { + Value::Null => visitor.visit_none(), + _ => visitor.visit_some(self), + } + } + + fn deserialize_enum<V>( + self, + _name: &str, + _variants: &'static [&'static str], + visitor: V, + ) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + let (variant, value) = match self { + Value::Object(value) => { + let mut iter = value.into_iter(); + let (variant, value) = match iter.next() { + Some(v) => v, + None => { + return Err(serde::de::Error::invalid_value( + Unexpected::Map, + &"map with a single key", + )); + } + }; + // enums are encoded in json as maps with a single key:value pair + if iter.next().is_some() { + return Err(serde::de::Error::invalid_value( + Unexpected::Map, + &"map with a single key", + )); + } + (variant, Some(value)) + } + Value::String(variant) => (variant, None), + other => { + return Err(serde::de::Error::invalid_type( + other.unexpected(), + &"string or map", + )); + } + }; + + visitor.visit_enum(EnumRefDeserializer { variant, value }) + } + + #[inline] + fn deserialize_newtype_struct<V>( + self, + name: &'static str, + visitor: V, + ) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + #[cfg(feature = "raw_value")] + { + if name == crate::raw::TOKEN { + return visitor.visit_map(crate::raw::OwnedRawDeserializer { + raw_value: Some(self.to_string()), + }); + } + } + + let _ = name; + visitor.visit_newtype_struct(self) + } + + fn deserialize_bool<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + match *self { + Value::Bool(v) => visitor.visit_bool(v), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_char<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + self.deserialize_str(visitor) + } + + fn deserialize_str<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + match self { + Value::String(v) => visitor.visit_borrowed_str(v), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_string<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + self.deserialize_str(visitor) + } + + fn deserialize_bytes<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + match self { + Value::String(v) => visitor.visit_borrowed_str(v), + Value::Array(v) => visit_array_ref(v, visitor), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_byte_buf<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + self.deserialize_bytes(visitor) + } + + fn deserialize_unit<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + match *self { + Value::Null => visitor.visit_unit(), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_unit_struct<V>(self, _name: &'static str, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + self.deserialize_unit(visitor) + } + + fn deserialize_seq<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + match self { + Value::Array(v) => visit_array_ref(v, visitor), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_tuple<V>(self, _len: usize, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + self.deserialize_seq(visitor) + } + + fn deserialize_tuple_struct<V>( + self, + _name: &'static str, + _len: usize, + visitor: V, + ) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + self.deserialize_seq(visitor) + } + + fn deserialize_map<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + match self { + Value::Object(v) => visit_object_ref(v, visitor), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_struct<V>( + self, + _name: &'static str, + _fields: &'static [&'static str], + visitor: V, + ) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + match self { + Value::Array(v) => visit_array_ref(v, visitor), + Value::Object(v) => visit_object_ref(v, visitor), + _ => Err(self.invalid_type(&visitor)), + } + } + + fn deserialize_identifier<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + self.deserialize_str(visitor) + } + + fn deserialize_ignored_any<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + visitor.visit_unit() + } +} + +struct EnumRefDeserializer<'de> { + variant: &'de str, + value: Option<&'de Value>, +} + +impl<'de> EnumAccess<'de> for EnumRefDeserializer<'de> { + type Error = Error; + type Variant = VariantRefDeserializer<'de>; + + fn variant_seed<V>(self, seed: V) -> Result<(V::Value, Self::Variant), Error> + where + V: DeserializeSeed<'de>, + { + let variant = self.variant.into_deserializer(); + let visitor = VariantRefDeserializer { value: self.value }; + seed.deserialize(variant).map(|v| (v, visitor)) + } +} + +struct VariantRefDeserializer<'de> { + value: Option<&'de Value>, +} + +impl<'de> VariantAccess<'de> for VariantRefDeserializer<'de> { + type Error = Error; + + fn unit_variant(self) -> Result<(), Error> { + match self.value { + Some(value) => Deserialize::deserialize(value), + None => Ok(()), + } + } + + fn newtype_variant_seed<T>(self, seed: T) -> Result<T::Value, Error> + where + T: DeserializeSeed<'de>, + { + match self.value { + Some(value) => seed.deserialize(value), + None => Err(serde::de::Error::invalid_type( + Unexpected::UnitVariant, + &"newtype variant", + )), + } + } + + fn tuple_variant<V>(self, _len: usize, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + match self.value { + Some(Value::Array(v)) => { + if v.is_empty() { + visitor.visit_unit() + } else { + visit_array_ref(v, visitor) + } + } + Some(other) => Err(serde::de::Error::invalid_type( + other.unexpected(), + &"tuple variant", + )), + None => Err(serde::de::Error::invalid_type( + Unexpected::UnitVariant, + &"tuple variant", + )), + } + } + + fn struct_variant<V>( + self, + _fields: &'static [&'static str], + visitor: V, + ) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + match self.value { + Some(Value::Object(v)) => visit_object_ref(v, visitor), + Some(other) => Err(serde::de::Error::invalid_type( + other.unexpected(), + &"struct variant", + )), + None => Err(serde::de::Error::invalid_type( + Unexpected::UnitVariant, + &"struct variant", + )), + } + } +} + +struct SeqRefDeserializer<'de> { + iter: slice::Iter<'de, Value>, +} + +impl<'de> SeqRefDeserializer<'de> { + fn new(slice: &'de [Value]) -> Self { + SeqRefDeserializer { iter: slice.iter() } + } +} + +impl<'de> SeqAccess<'de> for SeqRefDeserializer<'de> { + type Error = Error; + + fn next_element_seed<T>(&mut self, seed: T) -> Result<Option<T::Value>, Error> + where + T: DeserializeSeed<'de>, + { + match self.iter.next() { + Some(value) => seed.deserialize(value).map(Some), + None => Ok(None), + } + } + + fn size_hint(&self) -> Option<usize> { + match self.iter.size_hint() { + (lower, Some(upper)) if lower == upper => Some(upper), + _ => None, + } + } +} + +struct MapRefDeserializer<'de> { + iter: <&'de Map<String, Value> as IntoIterator>::IntoIter, + value: Option<&'de Value>, +} + +impl<'de> MapRefDeserializer<'de> { + fn new(map: &'de Map<String, Value>) -> Self { + MapRefDeserializer { + iter: map.into_iter(), + value: None, + } + } +} + +impl<'de> MapAccess<'de> for MapRefDeserializer<'de> { + type Error = Error; + + fn next_key_seed<T>(&mut self, seed: T) -> Result<Option<T::Value>, Error> + where + T: DeserializeSeed<'de>, + { + match self.iter.next() { + Some((key, value)) => { + self.value = Some(value); + let key_de = MapKeyDeserializer { + key: Cow::Borrowed(&**key), + }; + seed.deserialize(key_de).map(Some) + } + None => Ok(None), + } + } + + fn next_value_seed<T>(&mut self, seed: T) -> Result<T::Value, Error> + where + T: DeserializeSeed<'de>, + { + match self.value.take() { + Some(value) => seed.deserialize(value), + None => Err(serde::de::Error::custom("value is missing")), + } + } + + fn size_hint(&self) -> Option<usize> { + match self.iter.size_hint() { + (lower, Some(upper)) if lower == upper => Some(upper), + _ => None, + } + } +} + +struct MapKeyDeserializer<'de> { + key: Cow<'de, str>, +} + +macro_rules! deserialize_integer_key { + ($method:ident => $visit:ident) => { + fn $method<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + match (self.key.parse(), self.key) { + (Ok(integer), _) => visitor.$visit(integer), + (Err(_), Cow::Borrowed(s)) => visitor.visit_borrowed_str(s), + #[cfg(any(feature = "std", feature = "alloc"))] + (Err(_), Cow::Owned(s)) => visitor.visit_string(s), + } + } + }; +} + +impl<'de> serde::Deserializer<'de> for MapKeyDeserializer<'de> { + type Error = Error; + + fn deserialize_any<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + BorrowedCowStrDeserializer::new(self.key).deserialize_any(visitor) + } + + deserialize_integer_key!(deserialize_i8 => visit_i8); + deserialize_integer_key!(deserialize_i16 => visit_i16); + deserialize_integer_key!(deserialize_i32 => visit_i32); + deserialize_integer_key!(deserialize_i64 => visit_i64); + deserialize_integer_key!(deserialize_i128 => visit_i128); + deserialize_integer_key!(deserialize_u8 => visit_u8); + deserialize_integer_key!(deserialize_u16 => visit_u16); + deserialize_integer_key!(deserialize_u32 => visit_u32); + deserialize_integer_key!(deserialize_u64 => visit_u64); + deserialize_integer_key!(deserialize_u128 => visit_u128); + + #[inline] + fn deserialize_option<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + // Map keys cannot be null. + visitor.visit_some(self) + } + + #[inline] + fn deserialize_newtype_struct<V>( + self, + _name: &'static str, + visitor: V, + ) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + visitor.visit_newtype_struct(self) + } + + fn deserialize_enum<V>( + self, + name: &'static str, + variants: &'static [&'static str], + visitor: V, + ) -> Result<V::Value, Error> + where + V: Visitor<'de>, + { + self.key + .into_deserializer() + .deserialize_enum(name, variants, visitor) + } + + forward_to_deserialize_any! { + bool f32 f64 char str string bytes byte_buf unit unit_struct seq tuple + tuple_struct map struct identifier ignored_any + } +} + +struct KeyClassifier; + +enum KeyClass { + Map(String), + #[cfg(feature = "arbitrary_precision")] + Number, + #[cfg(feature = "raw_value")] + RawValue, +} + +impl<'de> DeserializeSeed<'de> for KeyClassifier { + type Value = KeyClass; + + fn deserialize<D>(self, deserializer: D) -> Result<Self::Value, D::Error> + where + D: serde::Deserializer<'de>, + { + deserializer.deserialize_str(self) + } +} + +impl<'de> Visitor<'de> for KeyClassifier { + type Value = KeyClass; + + fn expecting(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + formatter.write_str("a string key") + } + + fn visit_str<E>(self, s: &str) -> Result<Self::Value, E> + where + E: de::Error, + { + match s { + #[cfg(feature = "arbitrary_precision")] + crate::number::TOKEN => Ok(KeyClass::Number), + #[cfg(feature = "raw_value")] + crate::raw::TOKEN => Ok(KeyClass::RawValue), + _ => Ok(KeyClass::Map(s.to_owned())), + } + } + + #[cfg(any(feature = "std", feature = "alloc"))] + fn visit_string<E>(self, s: String) -> Result<Self::Value, E> + where + E: de::Error, + { + match s.as_str() { + #[cfg(feature = "arbitrary_precision")] + crate::number::TOKEN => Ok(KeyClass::Number), + #[cfg(feature = "raw_value")] + crate::raw::TOKEN => Ok(KeyClass::RawValue), + _ => Ok(KeyClass::Map(s)), + } + } +} + +impl Value { + #[cold] + fn invalid_type<E>(&self, exp: &dyn Expected) -> E + where + E: serde::de::Error, + { + serde::de::Error::invalid_type(self.unexpected(), exp) + } + + #[cold] + fn unexpected(&self) -> Unexpected { + match self { + Value::Null => Unexpected::Unit, + Value::Bool(b) => Unexpected::Bool(*b), + Value::Number(n) => n.unexpected(), + Value::String(s) => Unexpected::Str(s), + Value::Array(_) => Unexpected::Seq, + Value::Object(_) => Unexpected::Map, + } + } +} + +struct BorrowedCowStrDeserializer<'de> { + value: Cow<'de, str>, +} + +impl<'de> BorrowedCowStrDeserializer<'de> { + fn new(value: Cow<'de, str>) -> Self { + BorrowedCowStrDeserializer { value } + } +} + +impl<'de> de::Deserializer<'de> for BorrowedCowStrDeserializer<'de> { + type Error = Error; + + fn deserialize_any<V>(self, visitor: V) -> Result<V::Value, Error> + where + V: de::Visitor<'de>, + { + match self.value { + Cow::Borrowed(string) => visitor.visit_borrowed_str(string), + #[cfg(any(feature = "std", feature = "alloc"))] + Cow::Owned(string) => visitor.visit_string(string), + } + } + + fn deserialize_enum<V>( + self, + _name: &str, + _variants: &'static [&'static str], + visitor: V, + ) -> Result<V::Value, Error> + where + V: de::Visitor<'de>, + { + visitor.visit_enum(self) + } + + forward_to_deserialize_any! { + bool i8 i16 i32 i64 i128 u8 u16 u32 u64 u128 f32 f64 char str string + bytes byte_buf option unit unit_struct newtype_struct seq tuple + tuple_struct map struct identifier ignored_any + } +} + +impl<'de> de::EnumAccess<'de> for BorrowedCowStrDeserializer<'de> { + type Error = Error; + type Variant = UnitOnly; + + fn variant_seed<T>(self, seed: T) -> Result<(T::Value, Self::Variant), Error> + where + T: de::DeserializeSeed<'de>, + { + let value = seed.deserialize(self)?; + Ok((value, UnitOnly)) + } +} + +struct UnitOnly; + +impl<'de> de::VariantAccess<'de> for UnitOnly { + type Error = Error; + + fn unit_variant(self) -> Result<(), Error> { + Ok(()) + } + + fn newtype_variant_seed<T>(self, _seed: T) -> Result<T::Value, Error> + where + T: de::DeserializeSeed<'de>, + { + Err(de::Error::invalid_type( + Unexpected::UnitVariant, + &"newtype variant", + )) + } + + fn tuple_variant<V>(self, _len: usize, _visitor: V) -> Result<V::Value, Error> + where + V: de::Visitor<'de>, + { + Err(de::Error::invalid_type( + Unexpected::UnitVariant, + &"tuple variant", + )) + } + + fn struct_variant<V>( + self, + _fields: &'static [&'static str], + _visitor: V, + ) -> Result<V::Value, Error> + where + V: de::Visitor<'de>, + { + Err(de::Error::invalid_type( + Unexpected::UnitVariant, + &"struct variant", + )) + } +} diff --git a/third_party/rust/serde_json/src/value/from.rs b/third_party/rust/serde_json/src/value/from.rs new file mode 100644 index 0000000000..c5a6a3960b --- /dev/null +++ b/third_party/rust/serde_json/src/value/from.rs @@ -0,0 +1,277 @@ +use super::Value; +use crate::map::Map; +use crate::number::Number; +use alloc::borrow::Cow; +use alloc::string::{String, ToString}; +use alloc::vec::Vec; +use core::iter::FromIterator; + +macro_rules! from_integer { + ($($ty:ident)*) => { + $( + impl From<$ty> for Value { + fn from(n: $ty) -> Self { + Value::Number(n.into()) + } + } + )* + }; +} + +from_integer! { + i8 i16 i32 i64 isize + u8 u16 u32 u64 usize +} + +#[cfg(feature = "arbitrary_precision")] +from_integer! { + i128 u128 +} + +impl From<f32> for Value { + /// Convert 32-bit floating point number to `Value` + /// + /// # Examples + /// + /// ``` + /// use serde_json::Value; + /// + /// let f: f32 = 13.37; + /// let x: Value = f.into(); + /// ``` + fn from(f: f32) -> Self { + From::from(f as f64) + } +} + +impl From<f64> for Value { + /// Convert 64-bit floating point number to `Value` + /// + /// # Examples + /// + /// ``` + /// use serde_json::Value; + /// + /// let f: f64 = 13.37; + /// let x: Value = f.into(); + /// ``` + fn from(f: f64) -> Self { + Number::from_f64(f).map_or(Value::Null, Value::Number) + } +} + +impl From<bool> for Value { + /// Convert boolean to `Value` + /// + /// # Examples + /// + /// ``` + /// use serde_json::Value; + /// + /// let b = false; + /// let x: Value = b.into(); + /// ``` + fn from(f: bool) -> Self { + Value::Bool(f) + } +} + +impl From<String> for Value { + /// Convert `String` to `Value` + /// + /// # Examples + /// + /// ``` + /// use serde_json::Value; + /// + /// let s: String = "lorem".to_string(); + /// let x: Value = s.into(); + /// ``` + fn from(f: String) -> Self { + Value::String(f) + } +} + +impl<'a> From<&'a str> for Value { + /// Convert string slice to `Value` + /// + /// # Examples + /// + /// ``` + /// use serde_json::Value; + /// + /// let s: &str = "lorem"; + /// let x: Value = s.into(); + /// ``` + fn from(f: &str) -> Self { + Value::String(f.to_string()) + } +} + +impl<'a> From<Cow<'a, str>> for Value { + /// Convert copy-on-write string to `Value` + /// + /// # Examples + /// + /// ``` + /// use serde_json::Value; + /// use std::borrow::Cow; + /// + /// let s: Cow<str> = Cow::Borrowed("lorem"); + /// let x: Value = s.into(); + /// ``` + /// + /// ``` + /// use serde_json::Value; + /// use std::borrow::Cow; + /// + /// let s: Cow<str> = Cow::Owned("lorem".to_string()); + /// let x: Value = s.into(); + /// ``` + fn from(f: Cow<'a, str>) -> Self { + Value::String(f.into_owned()) + } +} + +impl From<Number> for Value { + /// Convert `Number` to `Value` + /// + /// # Examples + /// + /// ``` + /// use serde_json::{Number, Value}; + /// + /// let n = Number::from(7); + /// let x: Value = n.into(); + /// ``` + fn from(f: Number) -> Self { + Value::Number(f) + } +} + +impl From<Map<String, Value>> for Value { + /// Convert map (with string keys) to `Value` + /// + /// # Examples + /// + /// ``` + /// use serde_json::{Map, Value}; + /// + /// let mut m = Map::new(); + /// m.insert("Lorem".to_string(), "ipsum".into()); + /// let x: Value = m.into(); + /// ``` + fn from(f: Map<String, Value>) -> Self { + Value::Object(f) + } +} + +impl<T: Into<Value>> From<Vec<T>> for Value { + /// Convert a `Vec` to `Value` + /// + /// # Examples + /// + /// ``` + /// use serde_json::Value; + /// + /// let v = vec!["lorem", "ipsum", "dolor"]; + /// let x: Value = v.into(); + /// ``` + fn from(f: Vec<T>) -> Self { + Value::Array(f.into_iter().map(Into::into).collect()) + } +} + +impl<'a, T: Clone + Into<Value>> From<&'a [T]> for Value { + /// Convert a slice to `Value` + /// + /// # Examples + /// + /// ``` + /// use serde_json::Value; + /// + /// let v: &[&str] = &["lorem", "ipsum", "dolor"]; + /// let x: Value = v.into(); + /// ``` + fn from(f: &'a [T]) -> Self { + Value::Array(f.iter().cloned().map(Into::into).collect()) + } +} + +impl<T: Into<Value>> FromIterator<T> for Value { + /// Convert an iteratable type to a `Value` + /// + /// # Examples + /// + /// ``` + /// use serde_json::Value; + /// + /// let v = std::iter::repeat(42).take(5); + /// let x: Value = v.collect(); + /// ``` + /// + /// ``` + /// use serde_json::Value; + /// + /// let v: Vec<_> = vec!["lorem", "ipsum", "dolor"]; + /// let x: Value = v.into_iter().collect(); + /// ``` + /// + /// ``` + /// use std::iter::FromIterator; + /// use serde_json::Value; + /// + /// let x: Value = Value::from_iter(vec!["lorem", "ipsum", "dolor"]); + /// ``` + fn from_iter<I: IntoIterator<Item = T>>(iter: I) -> Self { + Value::Array(iter.into_iter().map(Into::into).collect()) + } +} + +impl<K: Into<String>, V: Into<Value>> FromIterator<(K, V)> for Value { + /// Convert an iteratable type to a `Value` + /// + /// # Examples + /// + /// ``` + /// use serde_json::Value; + /// + /// let v: Vec<_> = vec![("lorem", 40), ("ipsum", 2)]; + /// let x: Value = v.into_iter().collect(); + /// ``` + fn from_iter<I: IntoIterator<Item = (K, V)>>(iter: I) -> Self { + Value::Object( + iter.into_iter() + .map(|(k, v)| (k.into(), v.into())) + .collect(), + ) + } +} + +impl From<()> for Value { + /// Convert `()` to `Value` + /// + /// # Examples + /// + /// ``` + /// use serde_json::Value; + /// + /// let u = (); + /// let x: Value = u.into(); + /// ``` + fn from((): ()) -> Self { + Value::Null + } +} + +impl<T> From<Option<T>> for Value +where + T: Into<Value>, +{ + fn from(opt: Option<T>) -> Self { + match opt { + None => Value::Null, + Some(value) => Into::into(value), + } + } +} diff --git a/third_party/rust/serde_json/src/value/index.rs b/third_party/rust/serde_json/src/value/index.rs new file mode 100644 index 0000000000..c74042b75a --- /dev/null +++ b/third_party/rust/serde_json/src/value/index.rs @@ -0,0 +1,258 @@ +use super::Value; +use crate::map::Map; +use alloc::borrow::ToOwned; +use alloc::string::String; +use core::fmt::{self, Display}; +use core::ops; + +/// A type that can be used to index into a `serde_json::Value`. +/// +/// The [`get`] and [`get_mut`] methods of `Value` accept any type that +/// implements `Index`, as does the [square-bracket indexing operator]. This +/// trait is implemented for strings which are used as the index into a JSON +/// map, and for `usize` which is used as the index into a JSON array. +/// +/// [`get`]: ../enum.Value.html#method.get +/// [`get_mut`]: ../enum.Value.html#method.get_mut +/// [square-bracket indexing operator]: ../enum.Value.html#impl-Index%3CI%3E +/// +/// This trait is sealed and cannot be implemented for types outside of +/// `serde_json`. +/// +/// # Examples +/// +/// ``` +/// # use serde_json::json; +/// # +/// let data = json!({ "inner": [1, 2, 3] }); +/// +/// // Data is a JSON map so it can be indexed with a string. +/// let inner = &data["inner"]; +/// +/// // Inner is a JSON array so it can be indexed with an integer. +/// let first = &inner[0]; +/// +/// assert_eq!(first, 1); +/// ``` +pub trait Index: private::Sealed { + /// Return None if the key is not already in the array or object. + #[doc(hidden)] + fn index_into<'v>(&self, v: &'v Value) -> Option<&'v Value>; + + /// Return None if the key is not already in the array or object. + #[doc(hidden)] + fn index_into_mut<'v>(&self, v: &'v mut Value) -> Option<&'v mut Value>; + + /// Panic if array index out of bounds. If key is not already in the object, + /// insert it with a value of null. Panic if Value is a type that cannot be + /// indexed into, except if Value is null then it can be treated as an empty + /// object. + #[doc(hidden)] + fn index_or_insert<'v>(&self, v: &'v mut Value) -> &'v mut Value; +} + +impl Index for usize { + fn index_into<'v>(&self, v: &'v Value) -> Option<&'v Value> { + match v { + Value::Array(vec) => vec.get(*self), + _ => None, + } + } + fn index_into_mut<'v>(&self, v: &'v mut Value) -> Option<&'v mut Value> { + match v { + Value::Array(vec) => vec.get_mut(*self), + _ => None, + } + } + fn index_or_insert<'v>(&self, v: &'v mut Value) -> &'v mut Value { + match v { + Value::Array(vec) => { + let len = vec.len(); + vec.get_mut(*self).unwrap_or_else(|| { + panic!( + "cannot access index {} of JSON array of length {}", + self, len + ) + }) + } + _ => panic!("cannot access index {} of JSON {}", self, Type(v)), + } + } +} + +impl Index for str { + fn index_into<'v>(&self, v: &'v Value) -> Option<&'v Value> { + match v { + Value::Object(map) => map.get(self), + _ => None, + } + } + fn index_into_mut<'v>(&self, v: &'v mut Value) -> Option<&'v mut Value> { + match v { + Value::Object(map) => map.get_mut(self), + _ => None, + } + } + fn index_or_insert<'v>(&self, v: &'v mut Value) -> &'v mut Value { + if let Value::Null = v { + *v = Value::Object(Map::new()); + } + match v { + Value::Object(map) => map.entry(self.to_owned()).or_insert(Value::Null), + _ => panic!("cannot access key {:?} in JSON {}", self, Type(v)), + } + } +} + +impl Index for String { + fn index_into<'v>(&self, v: &'v Value) -> Option<&'v Value> { + self[..].index_into(v) + } + fn index_into_mut<'v>(&self, v: &'v mut Value) -> Option<&'v mut Value> { + self[..].index_into_mut(v) + } + fn index_or_insert<'v>(&self, v: &'v mut Value) -> &'v mut Value { + self[..].index_or_insert(v) + } +} + +impl<'a, T> Index for &'a T +where + T: ?Sized + Index, +{ + fn index_into<'v>(&self, v: &'v Value) -> Option<&'v Value> { + (**self).index_into(v) + } + fn index_into_mut<'v>(&self, v: &'v mut Value) -> Option<&'v mut Value> { + (**self).index_into_mut(v) + } + fn index_or_insert<'v>(&self, v: &'v mut Value) -> &'v mut Value { + (**self).index_or_insert(v) + } +} + +// Prevent users from implementing the Index trait. +mod private { + pub trait Sealed {} + impl Sealed for usize {} + impl Sealed for str {} + impl Sealed for alloc::string::String {} + impl<'a, T> Sealed for &'a T where T: ?Sized + Sealed {} +} + +/// Used in panic messages. +struct Type<'a>(&'a Value); + +impl<'a> Display for Type<'a> { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + match *self.0 { + Value::Null => formatter.write_str("null"), + Value::Bool(_) => formatter.write_str("boolean"), + Value::Number(_) => formatter.write_str("number"), + Value::String(_) => formatter.write_str("string"), + Value::Array(_) => formatter.write_str("array"), + Value::Object(_) => formatter.write_str("object"), + } + } +} + +// The usual semantics of Index is to panic on invalid indexing. +// +// That said, the usual semantics are for things like Vec and BTreeMap which +// have different use cases than Value. If you are working with a Vec, you know +// that you are working with a Vec and you can get the len of the Vec and make +// sure your indices are within bounds. The Value use cases are more +// loosey-goosey. You got some JSON from an endpoint and you want to pull values +// out of it. Outside of this Index impl, you already have the option of using +// value.as_array() and working with the Vec directly, or matching on +// Value::Array and getting the Vec directly. The Index impl means you can skip +// that and index directly into the thing using a concise syntax. You don't have +// to check the type, you don't have to check the len, it is all about what you +// expect the Value to look like. +// +// Basically the use cases that would be well served by panicking here are +// better served by using one of the other approaches: get and get_mut, +// as_array, or match. The value of this impl is that it adds a way of working +// with Value that is not well served by the existing approaches: concise and +// careless and sometimes that is exactly what you want. +impl<I> ops::Index<I> for Value +where + I: Index, +{ + type Output = Value; + + /// Index into a `serde_json::Value` using the syntax `value[0]` or + /// `value["k"]`. + /// + /// Returns `Value::Null` if the type of `self` does not match the type of + /// the index, for example if the index is a string and `self` is an array + /// or a number. Also returns `Value::Null` if the given key does not exist + /// in the map or the given index is not within the bounds of the array. + /// + /// For retrieving deeply nested values, you should have a look at the + /// `Value::pointer` method. + /// + /// # Examples + /// + /// ``` + /// # use serde_json::json; + /// # + /// let data = json!({ + /// "x": { + /// "y": ["z", "zz"] + /// } + /// }); + /// + /// assert_eq!(data["x"]["y"], json!(["z", "zz"])); + /// assert_eq!(data["x"]["y"][0], json!("z")); + /// + /// assert_eq!(data["a"], json!(null)); // returns null for undefined values + /// assert_eq!(data["a"]["b"], json!(null)); // does not panic + /// ``` + fn index(&self, index: I) -> &Value { + static NULL: Value = Value::Null; + index.index_into(self).unwrap_or(&NULL) + } +} + +impl<I> ops::IndexMut<I> for Value +where + I: Index, +{ + /// Write into a `serde_json::Value` using the syntax `value[0] = ...` or + /// `value["k"] = ...`. + /// + /// If the index is a number, the value must be an array of length bigger + /// than the index. Indexing into a value that is not an array or an array + /// that is too small will panic. + /// + /// If the index is a string, the value must be an object or null which is + /// treated like an empty object. If the key is not already present in the + /// object, it will be inserted with a value of null. Indexing into a value + /// that is neither an object nor null will panic. + /// + /// # Examples + /// + /// ``` + /// # use serde_json::json; + /// # + /// let mut data = json!({ "x": 0 }); + /// + /// // replace an existing key + /// data["x"] = json!(1); + /// + /// // insert a new key + /// data["y"] = json!([false, false, false]); + /// + /// // replace an array value + /// data["y"][0] = json!(true); + /// + /// // inserted a deeply nested key + /// data["a"]["b"]["c"]["d"] = json!(true); + /// + /// println!("{}", data); + /// ``` + fn index_mut(&mut self, index: I) -> &mut Value { + index.index_or_insert(self) + } +} diff --git a/third_party/rust/serde_json/src/value/mod.rs b/third_party/rust/serde_json/src/value/mod.rs new file mode 100644 index 0000000000..470b6b24da --- /dev/null +++ b/third_party/rust/serde_json/src/value/mod.rs @@ -0,0 +1,987 @@ +//! The Value enum, a loosely typed way of representing any valid JSON value. +//! +//! # Constructing JSON +//! +//! Serde JSON provides a [`json!` macro][macro] to build `serde_json::Value` +//! objects with very natural JSON syntax. +//! +//! ``` +//! use serde_json::json; +//! +//! fn main() { +//! // The type of `john` is `serde_json::Value` +//! let john = json!({ +//! "name": "John Doe", +//! "age": 43, +//! "phones": [ +//! "+44 1234567", +//! "+44 2345678" +//! ] +//! }); +//! +//! println!("first phone number: {}", john["phones"][0]); +//! +//! // Convert to a string of JSON and print it out +//! println!("{}", john.to_string()); +//! } +//! ``` +//! +//! The `Value::to_string()` function converts a `serde_json::Value` into a +//! `String` of JSON text. +//! +//! One neat thing about the `json!` macro is that variables and expressions can +//! be interpolated directly into the JSON value as you are building it. Serde +//! will check at compile time that the value you are interpolating is able to +//! be represented as JSON. +//! +//! ``` +//! # use serde_json::json; +//! # +//! # fn random_phone() -> u16 { 0 } +//! # +//! let full_name = "John Doe"; +//! let age_last_year = 42; +//! +//! // The type of `john` is `serde_json::Value` +//! let john = json!({ +//! "name": full_name, +//! "age": age_last_year + 1, +//! "phones": [ +//! format!("+44 {}", random_phone()) +//! ] +//! }); +//! ``` +//! +//! A string of JSON data can be parsed into a `serde_json::Value` by the +//! [`serde_json::from_str`][from_str] function. There is also +//! [`from_slice`][from_slice] for parsing from a byte slice `&[u8]` and +//! [`from_reader`][from_reader] for parsing from any `io::Read` like a File or +//! a TCP stream. +//! +//! ``` +//! use serde_json::{json, Value, Error}; +//! +//! fn untyped_example() -> Result<(), Error> { +//! // Some JSON input data as a &str. Maybe this comes from the user. +//! let data = r#" +//! { +//! "name": "John Doe", +//! "age": 43, +//! "phones": [ +//! "+44 1234567", +//! "+44 2345678" +//! ] +//! }"#; +//! +//! // Parse the string of data into serde_json::Value. +//! let v: Value = serde_json::from_str(data)?; +//! +//! // Access parts of the data by indexing with square brackets. +//! println!("Please call {} at the number {}", v["name"], v["phones"][0]); +//! +//! Ok(()) +//! } +//! # +//! # untyped_example().unwrap(); +//! ``` +//! +//! [macro]: crate::json +//! [from_str]: crate::de::from_str +//! [from_slice]: crate::de::from_slice +//! [from_reader]: crate::de::from_reader + +use crate::error::Error; +use crate::io; +use alloc::string::String; +use alloc::vec::Vec; +use core::fmt::{self, Debug, Display}; +use core::mem; +use core::str; +use serde::de::DeserializeOwned; +use serde::ser::Serialize; + +pub use self::index::Index; +pub use self::ser::Serializer; +pub use crate::map::Map; +pub use crate::number::Number; + +#[cfg(feature = "raw_value")] +pub use crate::raw::{to_raw_value, RawValue}; + +/// Represents any valid JSON value. +/// +/// See the [`serde_json::value` module documentation](self) for usage examples. +#[derive(Clone, Eq, PartialEq)] +pub enum Value { + /// Represents a JSON null value. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let v = json!(null); + /// ``` + Null, + + /// Represents a JSON boolean. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let v = json!(true); + /// ``` + Bool(bool), + + /// Represents a JSON number, whether integer or floating point. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let v = json!(12.5); + /// ``` + Number(Number), + + /// Represents a JSON string. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let v = json!("a string"); + /// ``` + String(String), + + /// Represents a JSON array. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let v = json!(["an", "array"]); + /// ``` + Array(Vec<Value>), + + /// Represents a JSON object. + /// + /// By default the map is backed by a BTreeMap. Enable the `preserve_order` + /// feature of serde_json to use IndexMap instead, which preserves + /// entries in the order they are inserted into the map. In particular, this + /// allows JSON data to be deserialized into a Value and serialized to a + /// string while retaining the order of map keys in the input. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let v = json!({ "an": "object" }); + /// ``` + Object(Map<String, Value>), +} + +impl Debug for Value { + fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result { + match self { + Value::Null => formatter.write_str("Null"), + Value::Bool(boolean) => write!(formatter, "Bool({})", boolean), + Value::Number(number) => Debug::fmt(number, formatter), + Value::String(string) => write!(formatter, "String({:?})", string), + Value::Array(vec) => { + formatter.write_str("Array ")?; + Debug::fmt(vec, formatter) + } + Value::Object(map) => { + formatter.write_str("Object ")?; + Debug::fmt(map, formatter) + } + } + } +} + +impl Display for Value { + /// Display a JSON value as a string. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let json = json!({ "city": "London", "street": "10 Downing Street" }); + /// + /// // Compact format: + /// // + /// // {"city":"London","street":"10 Downing Street"} + /// let compact = format!("{}", json); + /// assert_eq!(compact, + /// "{\"city\":\"London\",\"street\":\"10 Downing Street\"}"); + /// + /// // Pretty format: + /// // + /// // { + /// // "city": "London", + /// // "street": "10 Downing Street" + /// // } + /// let pretty = format!("{:#}", json); + /// assert_eq!(pretty, + /// "{\n \"city\": \"London\",\n \"street\": \"10 Downing Street\"\n}"); + /// ``` + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { + struct WriterFormatter<'a, 'b: 'a> { + inner: &'a mut fmt::Formatter<'b>, + } + + impl<'a, 'b> io::Write for WriterFormatter<'a, 'b> { + fn write(&mut self, buf: &[u8]) -> io::Result<usize> { + // Safety: the serializer below only emits valid utf8 when using + // the default formatter. + let s = unsafe { str::from_utf8_unchecked(buf) }; + tri!(self.inner.write_str(s).map_err(io_error)); + Ok(buf.len()) + } + + fn flush(&mut self) -> io::Result<()> { + Ok(()) + } + } + + fn io_error(_: fmt::Error) -> io::Error { + // Error value does not matter because Display impl just maps it + // back to fmt::Error. + io::Error::new(io::ErrorKind::Other, "fmt error") + } + + let alternate = f.alternate(); + let mut wr = WriterFormatter { inner: f }; + if alternate { + // {:#} + super::ser::to_writer_pretty(&mut wr, self).map_err(|_| fmt::Error) + } else { + // {} + super::ser::to_writer(&mut wr, self).map_err(|_| fmt::Error) + } + } +} + +fn parse_index(s: &str) -> Option<usize> { + if s.starts_with('+') || (s.starts_with('0') && s.len() != 1) { + return None; + } + s.parse().ok() +} + +impl Value { + /// Index into a JSON array or map. A string index can be used to access a + /// value in a map, and a usize index can be used to access an element of an + /// array. + /// + /// Returns `None` if the type of `self` does not match the type of the + /// index, for example if the index is a string and `self` is an array or a + /// number. Also returns `None` if the given key does not exist in the map + /// or the given index is not within the bounds of the array. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let object = json!({ "A": 65, "B": 66, "C": 67 }); + /// assert_eq!(*object.get("A").unwrap(), json!(65)); + /// + /// let array = json!([ "A", "B", "C" ]); + /// assert_eq!(*array.get(2).unwrap(), json!("C")); + /// + /// assert_eq!(array.get("A"), None); + /// ``` + /// + /// Square brackets can also be used to index into a value in a more concise + /// way. This returns `Value::Null` in cases where `get` would have returned + /// `None`. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let object = json!({ + /// "A": ["a", "á", "à"], + /// "B": ["b", "b́"], + /// "C": ["c", "ć", "ć̣", "ḉ"], + /// }); + /// assert_eq!(object["B"][0], json!("b")); + /// + /// assert_eq!(object["D"], json!(null)); + /// assert_eq!(object[0]["x"]["y"]["z"], json!(null)); + /// ``` + pub fn get<I: Index>(&self, index: I) -> Option<&Value> { + index.index_into(self) + } + + /// Mutably index into a JSON array or map. A string index can be used to + /// access a value in a map, and a usize index can be used to access an + /// element of an array. + /// + /// Returns `None` if the type of `self` does not match the type of the + /// index, for example if the index is a string and `self` is an array or a + /// number. Also returns `None` if the given key does not exist in the map + /// or the given index is not within the bounds of the array. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let mut object = json!({ "A": 65, "B": 66, "C": 67 }); + /// *object.get_mut("A").unwrap() = json!(69); + /// + /// let mut array = json!([ "A", "B", "C" ]); + /// *array.get_mut(2).unwrap() = json!("D"); + /// ``` + pub fn get_mut<I: Index>(&mut self, index: I) -> Option<&mut Value> { + index.index_into_mut(self) + } + + /// Returns true if the `Value` is an Object. Returns false otherwise. + /// + /// For any Value on which `is_object` returns true, `as_object` and + /// `as_object_mut` are guaranteed to return the map representation of the + /// object. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let obj = json!({ "a": { "nested": true }, "b": ["an", "array"] }); + /// + /// assert!(obj.is_object()); + /// assert!(obj["a"].is_object()); + /// + /// // array, not an object + /// assert!(!obj["b"].is_object()); + /// ``` + pub fn is_object(&self) -> bool { + self.as_object().is_some() + } + + /// If the `Value` is an Object, returns the associated Map. Returns None + /// otherwise. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let v = json!({ "a": { "nested": true }, "b": ["an", "array"] }); + /// + /// // The length of `{"nested": true}` is 1 entry. + /// assert_eq!(v["a"].as_object().unwrap().len(), 1); + /// + /// // The array `["an", "array"]` is not an object. + /// assert_eq!(v["b"].as_object(), None); + /// ``` + pub fn as_object(&self) -> Option<&Map<String, Value>> { + match self { + Value::Object(map) => Some(map), + _ => None, + } + } + + /// If the `Value` is an Object, returns the associated mutable Map. + /// Returns None otherwise. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let mut v = json!({ "a": { "nested": true } }); + /// + /// v["a"].as_object_mut().unwrap().clear(); + /// assert_eq!(v, json!({ "a": {} })); + /// ``` + pub fn as_object_mut(&mut self) -> Option<&mut Map<String, Value>> { + match self { + Value::Object(map) => Some(map), + _ => None, + } + } + + /// Returns true if the `Value` is an Array. Returns false otherwise. + /// + /// For any Value on which `is_array` returns true, `as_array` and + /// `as_array_mut` are guaranteed to return the vector representing the + /// array. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let obj = json!({ "a": ["an", "array"], "b": { "an": "object" } }); + /// + /// assert!(obj["a"].is_array()); + /// + /// // an object, not an array + /// assert!(!obj["b"].is_array()); + /// ``` + pub fn is_array(&self) -> bool { + self.as_array().is_some() + } + + /// If the `Value` is an Array, returns the associated vector. Returns None + /// otherwise. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let v = json!({ "a": ["an", "array"], "b": { "an": "object" } }); + /// + /// // The length of `["an", "array"]` is 2 elements. + /// assert_eq!(v["a"].as_array().unwrap().len(), 2); + /// + /// // The object `{"an": "object"}` is not an array. + /// assert_eq!(v["b"].as_array(), None); + /// ``` + pub fn as_array(&self) -> Option<&Vec<Value>> { + match self { + Value::Array(array) => Some(array), + _ => None, + } + } + + /// If the `Value` is an Array, returns the associated mutable vector. + /// Returns None otherwise. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let mut v = json!({ "a": ["an", "array"] }); + /// + /// v["a"].as_array_mut().unwrap().clear(); + /// assert_eq!(v, json!({ "a": [] })); + /// ``` + pub fn as_array_mut(&mut self) -> Option<&mut Vec<Value>> { + match self { + Value::Array(list) => Some(list), + _ => None, + } + } + + /// Returns true if the `Value` is a String. Returns false otherwise. + /// + /// For any Value on which `is_string` returns true, `as_str` is guaranteed + /// to return the string slice. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let v = json!({ "a": "some string", "b": false }); + /// + /// assert!(v["a"].is_string()); + /// + /// // The boolean `false` is not a string. + /// assert!(!v["b"].is_string()); + /// ``` + pub fn is_string(&self) -> bool { + self.as_str().is_some() + } + + /// If the `Value` is a String, returns the associated str. Returns None + /// otherwise. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let v = json!({ "a": "some string", "b": false }); + /// + /// assert_eq!(v["a"].as_str(), Some("some string")); + /// + /// // The boolean `false` is not a string. + /// assert_eq!(v["b"].as_str(), None); + /// + /// // JSON values are printed in JSON representation, so strings are in quotes. + /// // + /// // The value is: "some string" + /// println!("The value is: {}", v["a"]); + /// + /// // Rust strings are printed without quotes. + /// // + /// // The value is: some string + /// println!("The value is: {}", v["a"].as_str().unwrap()); + /// ``` + pub fn as_str(&self) -> Option<&str> { + match self { + Value::String(s) => Some(s), + _ => None, + } + } + + /// Returns true if the `Value` is a Number. Returns false otherwise. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let v = json!({ "a": 1, "b": "2" }); + /// + /// assert!(v["a"].is_number()); + /// + /// // The string `"2"` is a string, not a number. + /// assert!(!v["b"].is_number()); + /// ``` + pub fn is_number(&self) -> bool { + match *self { + Value::Number(_) => true, + _ => false, + } + } + + /// Returns true if the `Value` is an integer between `i64::MIN` and + /// `i64::MAX`. + /// + /// For any Value on which `is_i64` returns true, `as_i64` is guaranteed to + /// return the integer value. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let big = i64::max_value() as u64 + 10; + /// let v = json!({ "a": 64, "b": big, "c": 256.0 }); + /// + /// assert!(v["a"].is_i64()); + /// + /// // Greater than i64::MAX. + /// assert!(!v["b"].is_i64()); + /// + /// // Numbers with a decimal point are not considered integers. + /// assert!(!v["c"].is_i64()); + /// ``` + pub fn is_i64(&self) -> bool { + match self { + Value::Number(n) => n.is_i64(), + _ => false, + } + } + + /// Returns true if the `Value` is an integer between zero and `u64::MAX`. + /// + /// For any Value on which `is_u64` returns true, `as_u64` is guaranteed to + /// return the integer value. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let v = json!({ "a": 64, "b": -64, "c": 256.0 }); + /// + /// assert!(v["a"].is_u64()); + /// + /// // Negative integer. + /// assert!(!v["b"].is_u64()); + /// + /// // Numbers with a decimal point are not considered integers. + /// assert!(!v["c"].is_u64()); + /// ``` + pub fn is_u64(&self) -> bool { + match self { + Value::Number(n) => n.is_u64(), + _ => false, + } + } + + /// Returns true if the `Value` is a number that can be represented by f64. + /// + /// For any Value on which `is_f64` returns true, `as_f64` is guaranteed to + /// return the floating point value. + /// + /// Currently this function returns true if and only if both `is_i64` and + /// `is_u64` return false but this is not a guarantee in the future. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let v = json!({ "a": 256.0, "b": 64, "c": -64 }); + /// + /// assert!(v["a"].is_f64()); + /// + /// // Integers. + /// assert!(!v["b"].is_f64()); + /// assert!(!v["c"].is_f64()); + /// ``` + pub fn is_f64(&self) -> bool { + match self { + Value::Number(n) => n.is_f64(), + _ => false, + } + } + + /// If the `Value` is an integer, represent it as i64 if possible. Returns + /// None otherwise. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let big = i64::max_value() as u64 + 10; + /// let v = json!({ "a": 64, "b": big, "c": 256.0 }); + /// + /// assert_eq!(v["a"].as_i64(), Some(64)); + /// assert_eq!(v["b"].as_i64(), None); + /// assert_eq!(v["c"].as_i64(), None); + /// ``` + pub fn as_i64(&self) -> Option<i64> { + match self { + Value::Number(n) => n.as_i64(), + _ => None, + } + } + + /// If the `Value` is an integer, represent it as u64 if possible. Returns + /// None otherwise. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let v = json!({ "a": 64, "b": -64, "c": 256.0 }); + /// + /// assert_eq!(v["a"].as_u64(), Some(64)); + /// assert_eq!(v["b"].as_u64(), None); + /// assert_eq!(v["c"].as_u64(), None); + /// ``` + pub fn as_u64(&self) -> Option<u64> { + match self { + Value::Number(n) => n.as_u64(), + _ => None, + } + } + + /// If the `Value` is a number, represent it as f64 if possible. Returns + /// None otherwise. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let v = json!({ "a": 256.0, "b": 64, "c": -64 }); + /// + /// assert_eq!(v["a"].as_f64(), Some(256.0)); + /// assert_eq!(v["b"].as_f64(), Some(64.0)); + /// assert_eq!(v["c"].as_f64(), Some(-64.0)); + /// ``` + pub fn as_f64(&self) -> Option<f64> { + match self { + Value::Number(n) => n.as_f64(), + _ => None, + } + } + + /// Returns true if the `Value` is a Boolean. Returns false otherwise. + /// + /// For any Value on which `is_boolean` returns true, `as_bool` is + /// guaranteed to return the boolean value. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let v = json!({ "a": false, "b": "false" }); + /// + /// assert!(v["a"].is_boolean()); + /// + /// // The string `"false"` is a string, not a boolean. + /// assert!(!v["b"].is_boolean()); + /// ``` + pub fn is_boolean(&self) -> bool { + self.as_bool().is_some() + } + + /// If the `Value` is a Boolean, returns the associated bool. Returns None + /// otherwise. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let v = json!({ "a": false, "b": "false" }); + /// + /// assert_eq!(v["a"].as_bool(), Some(false)); + /// + /// // The string `"false"` is a string, not a boolean. + /// assert_eq!(v["b"].as_bool(), None); + /// ``` + pub fn as_bool(&self) -> Option<bool> { + match *self { + Value::Bool(b) => Some(b), + _ => None, + } + } + + /// Returns true if the `Value` is a Null. Returns false otherwise. + /// + /// For any Value on which `is_null` returns true, `as_null` is guaranteed + /// to return `Some(())`. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let v = json!({ "a": null, "b": false }); + /// + /// assert!(v["a"].is_null()); + /// + /// // The boolean `false` is not null. + /// assert!(!v["b"].is_null()); + /// ``` + pub fn is_null(&self) -> bool { + self.as_null().is_some() + } + + /// If the `Value` is a Null, returns (). Returns None otherwise. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let v = json!({ "a": null, "b": false }); + /// + /// assert_eq!(v["a"].as_null(), Some(())); + /// + /// // The boolean `false` is not null. + /// assert_eq!(v["b"].as_null(), None); + /// ``` + pub fn as_null(&self) -> Option<()> { + match *self { + Value::Null => Some(()), + _ => None, + } + } + + /// Looks up a value by a JSON Pointer. + /// + /// JSON Pointer defines a string syntax for identifying a specific value + /// within a JavaScript Object Notation (JSON) document. + /// + /// A Pointer is a Unicode string with the reference tokens separated by `/`. + /// Inside tokens `/` is replaced by `~1` and `~` is replaced by `~0`. The + /// addressed value is returned and if there is no such value `None` is + /// returned. + /// + /// For more information read [RFC6901](https://tools.ietf.org/html/rfc6901). + /// + /// # Examples + /// + /// ``` + /// # use serde_json::json; + /// # + /// let data = json!({ + /// "x": { + /// "y": ["z", "zz"] + /// } + /// }); + /// + /// assert_eq!(data.pointer("/x/y/1").unwrap(), &json!("zz")); + /// assert_eq!(data.pointer("/a/b/c"), None); + /// ``` + pub fn pointer(&self, pointer: &str) -> Option<&Value> { + if pointer.is_empty() { + return Some(self); + } + if !pointer.starts_with('/') { + return None; + } + pointer + .split('/') + .skip(1) + .map(|x| x.replace("~1", "/").replace("~0", "~")) + .try_fold(self, |target, token| match target { + Value::Object(map) => map.get(&token), + Value::Array(list) => parse_index(&token).and_then(|x| list.get(x)), + _ => None, + }) + } + + /// Looks up a value by a JSON Pointer and returns a mutable reference to + /// that value. + /// + /// JSON Pointer defines a string syntax for identifying a specific value + /// within a JavaScript Object Notation (JSON) document. + /// + /// A Pointer is a Unicode string with the reference tokens separated by `/`. + /// Inside tokens `/` is replaced by `~1` and `~` is replaced by `~0`. The + /// addressed value is returned and if there is no such value `None` is + /// returned. + /// + /// For more information read [RFC6901](https://tools.ietf.org/html/rfc6901). + /// + /// # Example of Use + /// + /// ``` + /// use serde_json::Value; + /// + /// fn main() { + /// let s = r#"{"x": 1.0, "y": 2.0}"#; + /// let mut value: Value = serde_json::from_str(s).unwrap(); + /// + /// // Check value using read-only pointer + /// assert_eq!(value.pointer("/x"), Some(&1.0.into())); + /// // Change value with direct assignment + /// *value.pointer_mut("/x").unwrap() = 1.5.into(); + /// // Check that new value was written + /// assert_eq!(value.pointer("/x"), Some(&1.5.into())); + /// // Or change the value only if it exists + /// value.pointer_mut("/x").map(|v| *v = 1.5.into()); + /// + /// // "Steal" ownership of a value. Can replace with any valid Value. + /// let old_x = value.pointer_mut("/x").map(Value::take).unwrap(); + /// assert_eq!(old_x, 1.5); + /// assert_eq!(value.pointer("/x").unwrap(), &Value::Null); + /// } + /// ``` + pub fn pointer_mut(&mut self, pointer: &str) -> Option<&mut Value> { + if pointer.is_empty() { + return Some(self); + } + if !pointer.starts_with('/') { + return None; + } + pointer + .split('/') + .skip(1) + .map(|x| x.replace("~1", "/").replace("~0", "~")) + .try_fold(self, |target, token| match target { + Value::Object(map) => map.get_mut(&token), + Value::Array(list) => parse_index(&token).and_then(move |x| list.get_mut(x)), + _ => None, + }) + } + + /// Takes the value out of the `Value`, leaving a `Null` in its place. + /// + /// ``` + /// # use serde_json::json; + /// # + /// let mut v = json!({ "x": "y" }); + /// assert_eq!(v["x"].take(), json!("y")); + /// assert_eq!(v, json!({ "x": null })); + /// ``` + pub fn take(&mut self) -> Value { + mem::replace(self, Value::Null) + } +} + +/// The default value is `Value::Null`. +/// +/// This is useful for handling omitted `Value` fields when deserializing. +/// +/// # Examples +/// +/// ``` +/// # use serde::Deserialize; +/// use serde_json::Value; +/// +/// #[derive(Deserialize)] +/// struct Settings { +/// level: i32, +/// #[serde(default)] +/// extras: Value, +/// } +/// +/// # fn try_main() -> Result<(), serde_json::Error> { +/// let data = r#" { "level": 42 } "#; +/// let s: Settings = serde_json::from_str(data)?; +/// +/// assert_eq!(s.level, 42); +/// assert_eq!(s.extras, Value::Null); +/// # +/// # Ok(()) +/// # } +/// # +/// # try_main().unwrap() +/// ``` +impl Default for Value { + fn default() -> Value { + Value::Null + } +} + +mod de; +mod from; +mod index; +mod partial_eq; +mod ser; + +/// Convert a `T` into `serde_json::Value` which is an enum that can represent +/// any valid JSON data. +/// +/// # Example +/// +/// ``` +/// use serde::Serialize; +/// use serde_json::json; +/// +/// use std::error::Error; +/// +/// #[derive(Serialize)] +/// struct User { +/// fingerprint: String, +/// location: String, +/// } +/// +/// fn compare_json_values() -> Result<(), Box<Error>> { +/// let u = User { +/// fingerprint: "0xF9BA143B95FF6D82".to_owned(), +/// location: "Menlo Park, CA".to_owned(), +/// }; +/// +/// // The type of `expected` is `serde_json::Value` +/// let expected = json!({ +/// "fingerprint": "0xF9BA143B95FF6D82", +/// "location": "Menlo Park, CA", +/// }); +/// +/// let v = serde_json::to_value(u).unwrap(); +/// assert_eq!(v, expected); +/// +/// Ok(()) +/// } +/// # +/// # compare_json_values().unwrap(); +/// ``` +/// +/// # Errors +/// +/// This conversion can fail if `T`'s implementation of `Serialize` decides to +/// fail, or if `T` contains a map with non-string keys. +/// +/// ``` +/// use std::collections::BTreeMap; +/// +/// fn main() { +/// // The keys in this map are vectors, not strings. +/// let mut map = BTreeMap::new(); +/// map.insert(vec![32, 64], "x86"); +/// +/// println!("{}", serde_json::to_value(map).unwrap_err()); +/// } +/// ``` +// Taking by value is more friendly to iterator adapters, option and result +// consumers, etc. See https://github.com/serde-rs/json/pull/149. +pub fn to_value<T>(value: T) -> Result<Value, Error> +where + T: Serialize, +{ + value.serialize(Serializer) +} + +/// Interpret a `serde_json::Value` as an instance of type `T`. +/// +/// # Example +/// +/// ``` +/// use serde::Deserialize; +/// use serde_json::json; +/// +/// #[derive(Deserialize, Debug)] +/// struct User { +/// fingerprint: String, +/// location: String, +/// } +/// +/// fn main() { +/// // The type of `j` is `serde_json::Value` +/// let j = json!({ +/// "fingerprint": "0xF9BA143B95FF6D82", +/// "location": "Menlo Park, CA" +/// }); +/// +/// let u: User = serde_json::from_value(j).unwrap(); +/// println!("{:#?}", u); +/// } +/// ``` +/// +/// # Errors +/// +/// This conversion can fail if the structure of the Value does not match the +/// structure expected by `T`, for example if `T` is a struct type but the Value +/// contains something other than a JSON map. It can also fail if the structure +/// is correct but `T`'s implementation of `Deserialize` decides that something +/// is wrong with the data, for example required struct fields are missing from +/// the JSON map or some number is too big to fit in the expected primitive +/// type. +pub fn from_value<T>(value: Value) -> Result<T, Error> +where + T: DeserializeOwned, +{ + T::deserialize(value) +} diff --git a/third_party/rust/serde_json/src/value/partial_eq.rs b/third_party/rust/serde_json/src/value/partial_eq.rs new file mode 100644 index 0000000000..b4ef84c4f3 --- /dev/null +++ b/third_party/rust/serde_json/src/value/partial_eq.rs @@ -0,0 +1,95 @@ +use super::Value; +use alloc::string::String; + +fn eq_i64(value: &Value, other: i64) -> bool { + value.as_i64().map_or(false, |i| i == other) +} + +fn eq_u64(value: &Value, other: u64) -> bool { + value.as_u64().map_or(false, |i| i == other) +} + +fn eq_f64(value: &Value, other: f64) -> bool { + value.as_f64().map_or(false, |i| i == other) +} + +fn eq_bool(value: &Value, other: bool) -> bool { + value.as_bool().map_or(false, |i| i == other) +} + +fn eq_str(value: &Value, other: &str) -> bool { + value.as_str().map_or(false, |i| i == other) +} + +impl PartialEq<str> for Value { + fn eq(&self, other: &str) -> bool { + eq_str(self, other) + } +} + +impl<'a> PartialEq<&'a str> for Value { + fn eq(&self, other: &&str) -> bool { + eq_str(self, *other) + } +} + +impl PartialEq<Value> for str { + fn eq(&self, other: &Value) -> bool { + eq_str(other, self) + } +} + +impl<'a> PartialEq<Value> for &'a str { + fn eq(&self, other: &Value) -> bool { + eq_str(other, *self) + } +} + +impl PartialEq<String> for Value { + fn eq(&self, other: &String) -> bool { + eq_str(self, other.as_str()) + } +} + +impl PartialEq<Value> for String { + fn eq(&self, other: &Value) -> bool { + eq_str(other, self.as_str()) + } +} + +macro_rules! partialeq_numeric { + ($($eq:ident [$($ty:ty)*])*) => { + $($( + impl PartialEq<$ty> for Value { + fn eq(&self, other: &$ty) -> bool { + $eq(self, *other as _) + } + } + + impl PartialEq<Value> for $ty { + fn eq(&self, other: &Value) -> bool { + $eq(other, *self as _) + } + } + + impl<'a> PartialEq<$ty> for &'a Value { + fn eq(&self, other: &$ty) -> bool { + $eq(*self, *other as _) + } + } + + impl<'a> PartialEq<$ty> for &'a mut Value { + fn eq(&self, other: &$ty) -> bool { + $eq(*self, *other as _) + } + } + )*)* + } +} + +partialeq_numeric! { + eq_i64[i8 i16 i32 i64 isize] + eq_u64[u8 u16 u32 u64 usize] + eq_f64[f32 f64] + eq_bool[bool] +} diff --git a/third_party/rust/serde_json/src/value/ser.rs b/third_party/rust/serde_json/src/value/ser.rs new file mode 100644 index 0000000000..a29814e92a --- /dev/null +++ b/third_party/rust/serde_json/src/value/ser.rs @@ -0,0 +1,1044 @@ +use crate::error::{Error, ErrorCode, Result}; +use crate::map::Map; +use crate::number::Number; +use crate::value::{to_value, Value}; +use alloc::borrow::ToOwned; +use alloc::string::{String, ToString}; +use alloc::vec::Vec; +#[cfg(not(feature = "arbitrary_precision"))] +use core::convert::TryFrom; +use core::fmt::Display; +use core::result; +use serde::ser::{Impossible, Serialize}; + +impl Serialize for Value { + #[inline] + fn serialize<S>(&self, serializer: S) -> result::Result<S::Ok, S::Error> + where + S: ::serde::Serializer, + { + match self { + Value::Null => serializer.serialize_unit(), + Value::Bool(b) => serializer.serialize_bool(*b), + Value::Number(n) => n.serialize(serializer), + Value::String(s) => serializer.serialize_str(s), + Value::Array(v) => v.serialize(serializer), + #[cfg(any(feature = "std", feature = "alloc"))] + Value::Object(m) => { + use serde::ser::SerializeMap; + let mut map = tri!(serializer.serialize_map(Some(m.len()))); + for (k, v) in m { + tri!(map.serialize_entry(k, v)); + } + map.end() + } + } + } +} + +/// Serializer whose output is a `Value`. +/// +/// This is the serializer that backs [`serde_json::to_value`][crate::to_value]. +/// Unlike the main serde_json serializer which goes from some serializable +/// value of type `T` to JSON text, this one goes from `T` to +/// `serde_json::Value`. +/// +/// The `to_value` function is implementable as: +/// +/// ``` +/// use serde::Serialize; +/// use serde_json::{Error, Value}; +/// +/// pub fn to_value<T>(input: T) -> Result<Value, Error> +/// where +/// T: Serialize, +/// { +/// input.serialize(serde_json::value::Serializer) +/// } +/// ``` +pub struct Serializer; + +impl serde::Serializer for Serializer { + type Ok = Value; + type Error = Error; + + type SerializeSeq = SerializeVec; + type SerializeTuple = SerializeVec; + type SerializeTupleStruct = SerializeVec; + type SerializeTupleVariant = SerializeTupleVariant; + type SerializeMap = SerializeMap; + type SerializeStruct = SerializeMap; + type SerializeStructVariant = SerializeStructVariant; + + #[inline] + fn serialize_bool(self, value: bool) -> Result<Value> { + Ok(Value::Bool(value)) + } + + #[inline] + fn serialize_i8(self, value: i8) -> Result<Value> { + self.serialize_i64(value as i64) + } + + #[inline] + fn serialize_i16(self, value: i16) -> Result<Value> { + self.serialize_i64(value as i64) + } + + #[inline] + fn serialize_i32(self, value: i32) -> Result<Value> { + self.serialize_i64(value as i64) + } + + fn serialize_i64(self, value: i64) -> Result<Value> { + Ok(Value::Number(value.into())) + } + + fn serialize_i128(self, value: i128) -> Result<Value> { + #[cfg(feature = "arbitrary_precision")] + { + Ok(Value::Number(value.into())) + } + + #[cfg(not(feature = "arbitrary_precision"))] + { + if let Ok(value) = u64::try_from(value) { + Ok(Value::Number(value.into())) + } else if let Ok(value) = i64::try_from(value) { + Ok(Value::Number(value.into())) + } else { + Err(Error::syntax(ErrorCode::NumberOutOfRange, 0, 0)) + } + } + } + + #[inline] + fn serialize_u8(self, value: u8) -> Result<Value> { + self.serialize_u64(value as u64) + } + + #[inline] + fn serialize_u16(self, value: u16) -> Result<Value> { + self.serialize_u64(value as u64) + } + + #[inline] + fn serialize_u32(self, value: u32) -> Result<Value> { + self.serialize_u64(value as u64) + } + + #[inline] + fn serialize_u64(self, value: u64) -> Result<Value> { + Ok(Value::Number(value.into())) + } + + fn serialize_u128(self, value: u128) -> Result<Value> { + #[cfg(feature = "arbitrary_precision")] + { + Ok(Value::Number(value.into())) + } + + #[cfg(not(feature = "arbitrary_precision"))] + { + if let Ok(value) = u64::try_from(value) { + Ok(Value::Number(value.into())) + } else { + Err(Error::syntax(ErrorCode::NumberOutOfRange, 0, 0)) + } + } + } + + #[inline] + fn serialize_f32(self, value: f32) -> Result<Value> { + self.serialize_f64(value as f64) + } + + #[inline] + fn serialize_f64(self, value: f64) -> Result<Value> { + Ok(Number::from_f64(value).map_or(Value::Null, Value::Number)) + } + + #[inline] + fn serialize_char(self, value: char) -> Result<Value> { + let mut s = String::new(); + s.push(value); + Ok(Value::String(s)) + } + + #[inline] + fn serialize_str(self, value: &str) -> Result<Value> { + Ok(Value::String(value.to_owned())) + } + + fn serialize_bytes(self, value: &[u8]) -> Result<Value> { + let vec = value.iter().map(|&b| Value::Number(b.into())).collect(); + Ok(Value::Array(vec)) + } + + #[inline] + fn serialize_unit(self) -> Result<Value> { + Ok(Value::Null) + } + + #[inline] + fn serialize_unit_struct(self, _name: &'static str) -> Result<Value> { + self.serialize_unit() + } + + #[inline] + fn serialize_unit_variant( + self, + _name: &'static str, + _variant_index: u32, + variant: &'static str, + ) -> Result<Value> { + self.serialize_str(variant) + } + + #[inline] + fn serialize_newtype_struct<T>(self, _name: &'static str, value: &T) -> Result<Value> + where + T: ?Sized + Serialize, + { + value.serialize(self) + } + + fn serialize_newtype_variant<T>( + self, + _name: &'static str, + _variant_index: u32, + variant: &'static str, + value: &T, + ) -> Result<Value> + where + T: ?Sized + Serialize, + { + let mut values = Map::new(); + values.insert(String::from(variant), tri!(to_value(value))); + Ok(Value::Object(values)) + } + + #[inline] + fn serialize_none(self) -> Result<Value> { + self.serialize_unit() + } + + #[inline] + fn serialize_some<T>(self, value: &T) -> Result<Value> + where + T: ?Sized + Serialize, + { + value.serialize(self) + } + + fn serialize_seq(self, len: Option<usize>) -> Result<Self::SerializeSeq> { + Ok(SerializeVec { + vec: Vec::with_capacity(len.unwrap_or(0)), + }) + } + + fn serialize_tuple(self, len: usize) -> Result<Self::SerializeTuple> { + self.serialize_seq(Some(len)) + } + + fn serialize_tuple_struct( + self, + _name: &'static str, + len: usize, + ) -> Result<Self::SerializeTupleStruct> { + self.serialize_seq(Some(len)) + } + + fn serialize_tuple_variant( + self, + _name: &'static str, + _variant_index: u32, + variant: &'static str, + len: usize, + ) -> Result<Self::SerializeTupleVariant> { + Ok(SerializeTupleVariant { + name: String::from(variant), + vec: Vec::with_capacity(len), + }) + } + + fn serialize_map(self, _len: Option<usize>) -> Result<Self::SerializeMap> { + Ok(SerializeMap::Map { + map: Map::new(), + next_key: None, + }) + } + + fn serialize_struct(self, name: &'static str, len: usize) -> Result<Self::SerializeStruct> { + match name { + #[cfg(feature = "arbitrary_precision")] + crate::number::TOKEN => Ok(SerializeMap::Number { out_value: None }), + #[cfg(feature = "raw_value")] + crate::raw::TOKEN => Ok(SerializeMap::RawValue { out_value: None }), + _ => self.serialize_map(Some(len)), + } + } + + fn serialize_struct_variant( + self, + _name: &'static str, + _variant_index: u32, + variant: &'static str, + _len: usize, + ) -> Result<Self::SerializeStructVariant> { + Ok(SerializeStructVariant { + name: String::from(variant), + map: Map::new(), + }) + } + + fn collect_str<T>(self, value: &T) -> Result<Value> + where + T: ?Sized + Display, + { + Ok(Value::String(value.to_string())) + } +} + +pub struct SerializeVec { + vec: Vec<Value>, +} + +pub struct SerializeTupleVariant { + name: String, + vec: Vec<Value>, +} + +pub enum SerializeMap { + Map { + map: Map<String, Value>, + next_key: Option<String>, + }, + #[cfg(feature = "arbitrary_precision")] + Number { out_value: Option<Value> }, + #[cfg(feature = "raw_value")] + RawValue { out_value: Option<Value> }, +} + +pub struct SerializeStructVariant { + name: String, + map: Map<String, Value>, +} + +impl serde::ser::SerializeSeq for SerializeVec { + type Ok = Value; + type Error = Error; + + fn serialize_element<T>(&mut self, value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + self.vec.push(tri!(to_value(value))); + Ok(()) + } + + fn end(self) -> Result<Value> { + Ok(Value::Array(self.vec)) + } +} + +impl serde::ser::SerializeTuple for SerializeVec { + type Ok = Value; + type Error = Error; + + fn serialize_element<T>(&mut self, value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + serde::ser::SerializeSeq::serialize_element(self, value) + } + + fn end(self) -> Result<Value> { + serde::ser::SerializeSeq::end(self) + } +} + +impl serde::ser::SerializeTupleStruct for SerializeVec { + type Ok = Value; + type Error = Error; + + fn serialize_field<T>(&mut self, value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + serde::ser::SerializeSeq::serialize_element(self, value) + } + + fn end(self) -> Result<Value> { + serde::ser::SerializeSeq::end(self) + } +} + +impl serde::ser::SerializeTupleVariant for SerializeTupleVariant { + type Ok = Value; + type Error = Error; + + fn serialize_field<T>(&mut self, value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + self.vec.push(tri!(to_value(value))); + Ok(()) + } + + fn end(self) -> Result<Value> { + let mut object = Map::new(); + + object.insert(self.name, Value::Array(self.vec)); + + Ok(Value::Object(object)) + } +} + +impl serde::ser::SerializeMap for SerializeMap { + type Ok = Value; + type Error = Error; + + fn serialize_key<T>(&mut self, key: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + match self { + SerializeMap::Map { next_key, .. } => { + *next_key = Some(tri!(key.serialize(MapKeySerializer))); + Ok(()) + } + #[cfg(feature = "arbitrary_precision")] + SerializeMap::Number { .. } => unreachable!(), + #[cfg(feature = "raw_value")] + SerializeMap::RawValue { .. } => unreachable!(), + } + } + + fn serialize_value<T>(&mut self, value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + match self { + SerializeMap::Map { map, next_key } => { + let key = next_key.take(); + // Panic because this indicates a bug in the program rather than an + // expected failure. + let key = key.expect("serialize_value called before serialize_key"); + map.insert(key, tri!(to_value(value))); + Ok(()) + } + #[cfg(feature = "arbitrary_precision")] + SerializeMap::Number { .. } => unreachable!(), + #[cfg(feature = "raw_value")] + SerializeMap::RawValue { .. } => unreachable!(), + } + } + + fn end(self) -> Result<Value> { + match self { + SerializeMap::Map { map, .. } => Ok(Value::Object(map)), + #[cfg(feature = "arbitrary_precision")] + SerializeMap::Number { .. } => unreachable!(), + #[cfg(feature = "raw_value")] + SerializeMap::RawValue { .. } => unreachable!(), + } + } +} + +struct MapKeySerializer; + +fn key_must_be_a_string() -> Error { + Error::syntax(ErrorCode::KeyMustBeAString, 0, 0) +} + +impl serde::Serializer for MapKeySerializer { + type Ok = String; + type Error = Error; + + type SerializeSeq = Impossible<String, Error>; + type SerializeTuple = Impossible<String, Error>; + type SerializeTupleStruct = Impossible<String, Error>; + type SerializeTupleVariant = Impossible<String, Error>; + type SerializeMap = Impossible<String, Error>; + type SerializeStruct = Impossible<String, Error>; + type SerializeStructVariant = Impossible<String, Error>; + + #[inline] + fn serialize_unit_variant( + self, + _name: &'static str, + _variant_index: u32, + variant: &'static str, + ) -> Result<String> { + Ok(variant.to_owned()) + } + + #[inline] + fn serialize_newtype_struct<T>(self, _name: &'static str, value: &T) -> Result<String> + where + T: ?Sized + Serialize, + { + value.serialize(self) + } + + fn serialize_bool(self, _value: bool) -> Result<String> { + Err(key_must_be_a_string()) + } + + fn serialize_i8(self, value: i8) -> Result<String> { + Ok(value.to_string()) + } + + fn serialize_i16(self, value: i16) -> Result<String> { + Ok(value.to_string()) + } + + fn serialize_i32(self, value: i32) -> Result<String> { + Ok(value.to_string()) + } + + fn serialize_i64(self, value: i64) -> Result<String> { + Ok(value.to_string()) + } + + fn serialize_u8(self, value: u8) -> Result<String> { + Ok(value.to_string()) + } + + fn serialize_u16(self, value: u16) -> Result<String> { + Ok(value.to_string()) + } + + fn serialize_u32(self, value: u32) -> Result<String> { + Ok(value.to_string()) + } + + fn serialize_u64(self, value: u64) -> Result<String> { + Ok(value.to_string()) + } + + fn serialize_f32(self, _value: f32) -> Result<String> { + Err(key_must_be_a_string()) + } + + fn serialize_f64(self, _value: f64) -> Result<String> { + Err(key_must_be_a_string()) + } + + #[inline] + fn serialize_char(self, value: char) -> Result<String> { + Ok({ + let mut s = String::new(); + s.push(value); + s + }) + } + + #[inline] + fn serialize_str(self, value: &str) -> Result<String> { + Ok(value.to_owned()) + } + + fn serialize_bytes(self, _value: &[u8]) -> Result<String> { + Err(key_must_be_a_string()) + } + + fn serialize_unit(self) -> Result<String> { + Err(key_must_be_a_string()) + } + + fn serialize_unit_struct(self, _name: &'static str) -> Result<String> { + Err(key_must_be_a_string()) + } + + fn serialize_newtype_variant<T>( + self, + _name: &'static str, + _variant_index: u32, + _variant: &'static str, + _value: &T, + ) -> Result<String> + where + T: ?Sized + Serialize, + { + Err(key_must_be_a_string()) + } + + fn serialize_none(self) -> Result<String> { + Err(key_must_be_a_string()) + } + + fn serialize_some<T>(self, _value: &T) -> Result<String> + where + T: ?Sized + Serialize, + { + Err(key_must_be_a_string()) + } + + fn serialize_seq(self, _len: Option<usize>) -> Result<Self::SerializeSeq> { + Err(key_must_be_a_string()) + } + + fn serialize_tuple(self, _len: usize) -> Result<Self::SerializeTuple> { + Err(key_must_be_a_string()) + } + + fn serialize_tuple_struct( + self, + _name: &'static str, + _len: usize, + ) -> Result<Self::SerializeTupleStruct> { + Err(key_must_be_a_string()) + } + + fn serialize_tuple_variant( + self, + _name: &'static str, + _variant_index: u32, + _variant: &'static str, + _len: usize, + ) -> Result<Self::SerializeTupleVariant> { + Err(key_must_be_a_string()) + } + + fn serialize_map(self, _len: Option<usize>) -> Result<Self::SerializeMap> { + Err(key_must_be_a_string()) + } + + fn serialize_struct(self, _name: &'static str, _len: usize) -> Result<Self::SerializeStruct> { + Err(key_must_be_a_string()) + } + + fn serialize_struct_variant( + self, + _name: &'static str, + _variant_index: u32, + _variant: &'static str, + _len: usize, + ) -> Result<Self::SerializeStructVariant> { + Err(key_must_be_a_string()) + } + + fn collect_str<T>(self, value: &T) -> Result<String> + where + T: ?Sized + Display, + { + Ok(value.to_string()) + } +} + +impl serde::ser::SerializeStruct for SerializeMap { + type Ok = Value; + type Error = Error; + + fn serialize_field<T>(&mut self, key: &'static str, value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + match self { + SerializeMap::Map { .. } => serde::ser::SerializeMap::serialize_entry(self, key, value), + #[cfg(feature = "arbitrary_precision")] + SerializeMap::Number { out_value } => { + if key == crate::number::TOKEN { + *out_value = Some(value.serialize(NumberValueEmitter)?); + Ok(()) + } else { + Err(invalid_number()) + } + } + #[cfg(feature = "raw_value")] + SerializeMap::RawValue { out_value } => { + if key == crate::raw::TOKEN { + *out_value = Some(value.serialize(RawValueEmitter)?); + Ok(()) + } else { + Err(invalid_raw_value()) + } + } + } + } + + fn end(self) -> Result<Value> { + match self { + SerializeMap::Map { .. } => serde::ser::SerializeMap::end(self), + #[cfg(feature = "arbitrary_precision")] + SerializeMap::Number { out_value, .. } => { + Ok(out_value.expect("number value was not emitted")) + } + #[cfg(feature = "raw_value")] + SerializeMap::RawValue { out_value, .. } => { + Ok(out_value.expect("raw value was not emitted")) + } + } + } +} + +impl serde::ser::SerializeStructVariant for SerializeStructVariant { + type Ok = Value; + type Error = Error; + + fn serialize_field<T>(&mut self, key: &'static str, value: &T) -> Result<()> + where + T: ?Sized + Serialize, + { + self.map.insert(String::from(key), tri!(to_value(value))); + Ok(()) + } + + fn end(self) -> Result<Value> { + let mut object = Map::new(); + + object.insert(self.name, Value::Object(self.map)); + + Ok(Value::Object(object)) + } +} + +#[cfg(feature = "arbitrary_precision")] +struct NumberValueEmitter; + +#[cfg(feature = "arbitrary_precision")] +fn invalid_number() -> Error { + Error::syntax(ErrorCode::InvalidNumber, 0, 0) +} + +#[cfg(feature = "arbitrary_precision")] +impl serde::ser::Serializer for NumberValueEmitter { + type Ok = Value; + type Error = Error; + + type SerializeSeq = Impossible<Value, Error>; + type SerializeTuple = Impossible<Value, Error>; + type SerializeTupleStruct = Impossible<Value, Error>; + type SerializeTupleVariant = Impossible<Value, Error>; + type SerializeMap = Impossible<Value, Error>; + type SerializeStruct = Impossible<Value, Error>; + type SerializeStructVariant = Impossible<Value, Error>; + + fn serialize_bool(self, _v: bool) -> Result<Value> { + Err(invalid_number()) + } + + fn serialize_i8(self, _v: i8) -> Result<Value> { + Err(invalid_number()) + } + + fn serialize_i16(self, _v: i16) -> Result<Value> { + Err(invalid_number()) + } + + fn serialize_i32(self, _v: i32) -> Result<Value> { + Err(invalid_number()) + } + + fn serialize_i64(self, _v: i64) -> Result<Value> { + Err(invalid_number()) + } + + fn serialize_u8(self, _v: u8) -> Result<Value> { + Err(invalid_number()) + } + + fn serialize_u16(self, _v: u16) -> Result<Value> { + Err(invalid_number()) + } + + fn serialize_u32(self, _v: u32) -> Result<Value> { + Err(invalid_number()) + } + + fn serialize_u64(self, _v: u64) -> Result<Value> { + Err(invalid_number()) + } + + fn serialize_f32(self, _v: f32) -> Result<Value> { + Err(invalid_number()) + } + + fn serialize_f64(self, _v: f64) -> Result<Value> { + Err(invalid_number()) + } + + fn serialize_char(self, _v: char) -> Result<Value> { + Err(invalid_number()) + } + + fn serialize_str(self, value: &str) -> Result<Value> { + let n = tri!(value.to_owned().parse()); + Ok(Value::Number(n)) + } + + fn serialize_bytes(self, _value: &[u8]) -> Result<Value> { + Err(invalid_number()) + } + + fn serialize_none(self) -> Result<Value> { + Err(invalid_number()) + } + + fn serialize_some<T>(self, _value: &T) -> Result<Value> + where + T: ?Sized + Serialize, + { + Err(invalid_number()) + } + + fn serialize_unit(self) -> Result<Value> { + Err(invalid_number()) + } + + fn serialize_unit_struct(self, _name: &'static str) -> Result<Value> { + Err(invalid_number()) + } + + fn serialize_unit_variant( + self, + _name: &'static str, + _variant_index: u32, + _variant: &'static str, + ) -> Result<Value> { + Err(invalid_number()) + } + + fn serialize_newtype_struct<T>(self, _name: &'static str, _value: &T) -> Result<Value> + where + T: ?Sized + Serialize, + { + Err(invalid_number()) + } + + fn serialize_newtype_variant<T>( + self, + _name: &'static str, + _variant_index: u32, + _variant: &'static str, + _value: &T, + ) -> Result<Value> + where + T: ?Sized + Serialize, + { + Err(invalid_number()) + } + + fn serialize_seq(self, _len: Option<usize>) -> Result<Self::SerializeSeq> { + Err(invalid_number()) + } + + fn serialize_tuple(self, _len: usize) -> Result<Self::SerializeTuple> { + Err(invalid_number()) + } + + fn serialize_tuple_struct( + self, + _name: &'static str, + _len: usize, + ) -> Result<Self::SerializeTupleStruct> { + Err(invalid_number()) + } + + fn serialize_tuple_variant( + self, + _name: &'static str, + _variant_index: u32, + _variant: &'static str, + _len: usize, + ) -> Result<Self::SerializeTupleVariant> { + Err(invalid_number()) + } + + fn serialize_map(self, _len: Option<usize>) -> Result<Self::SerializeMap> { + Err(invalid_number()) + } + + fn serialize_struct(self, _name: &'static str, _len: usize) -> Result<Self::SerializeStruct> { + Err(invalid_number()) + } + + fn serialize_struct_variant( + self, + _name: &'static str, + _variant_index: u32, + _variant: &'static str, + _len: usize, + ) -> Result<Self::SerializeStructVariant> { + Err(invalid_number()) + } +} + +#[cfg(feature = "raw_value")] +struct RawValueEmitter; + +#[cfg(feature = "raw_value")] +fn invalid_raw_value() -> Error { + Error::syntax(ErrorCode::ExpectedSomeValue, 0, 0) +} + +#[cfg(feature = "raw_value")] +impl serde::ser::Serializer for RawValueEmitter { + type Ok = Value; + type Error = Error; + + type SerializeSeq = Impossible<Value, Error>; + type SerializeTuple = Impossible<Value, Error>; + type SerializeTupleStruct = Impossible<Value, Error>; + type SerializeTupleVariant = Impossible<Value, Error>; + type SerializeMap = Impossible<Value, Error>; + type SerializeStruct = Impossible<Value, Error>; + type SerializeStructVariant = Impossible<Value, Error>; + + fn serialize_bool(self, _v: bool) -> Result<Value> { + Err(invalid_raw_value()) + } + + fn serialize_i8(self, _v: i8) -> Result<Value> { + Err(invalid_raw_value()) + } + + fn serialize_i16(self, _v: i16) -> Result<Value> { + Err(invalid_raw_value()) + } + + fn serialize_i32(self, _v: i32) -> Result<Value> { + Err(invalid_raw_value()) + } + + fn serialize_i64(self, _v: i64) -> Result<Value> { + Err(invalid_raw_value()) + } + + fn serialize_u8(self, _v: u8) -> Result<Value> { + Err(invalid_raw_value()) + } + + fn serialize_u16(self, _v: u16) -> Result<Value> { + Err(invalid_raw_value()) + } + + fn serialize_u32(self, _v: u32) -> Result<Value> { + Err(invalid_raw_value()) + } + + fn serialize_u64(self, _v: u64) -> Result<Value> { + Err(invalid_raw_value()) + } + + fn serialize_f32(self, _v: f32) -> Result<Value> { + Err(invalid_raw_value()) + } + + fn serialize_f64(self, _v: f64) -> Result<Value> { + Err(invalid_raw_value()) + } + + fn serialize_char(self, _v: char) -> Result<Value> { + Err(invalid_raw_value()) + } + + fn serialize_str(self, value: &str) -> Result<Value> { + crate::from_str(value) + } + + fn serialize_bytes(self, _value: &[u8]) -> Result<Value> { + Err(invalid_raw_value()) + } + + fn serialize_none(self) -> Result<Value> { + Err(invalid_raw_value()) + } + + fn serialize_some<T>(self, _value: &T) -> Result<Value> + where + T: ?Sized + Serialize, + { + Err(invalid_raw_value()) + } + + fn serialize_unit(self) -> Result<Value> { + Err(invalid_raw_value()) + } + + fn serialize_unit_struct(self, _name: &'static str) -> Result<Value> { + Err(invalid_raw_value()) + } + + fn serialize_unit_variant( + self, + _name: &'static str, + _variant_index: u32, + _variant: &'static str, + ) -> Result<Value> { + Err(invalid_raw_value()) + } + + fn serialize_newtype_struct<T>(self, _name: &'static str, _value: &T) -> Result<Value> + where + T: ?Sized + Serialize, + { + Err(invalid_raw_value()) + } + + fn serialize_newtype_variant<T>( + self, + _name: &'static str, + _variant_index: u32, + _variant: &'static str, + _value: &T, + ) -> Result<Value> + where + T: ?Sized + Serialize, + { + Err(invalid_raw_value()) + } + + fn serialize_seq(self, _len: Option<usize>) -> Result<Self::SerializeSeq> { + Err(invalid_raw_value()) + } + + fn serialize_tuple(self, _len: usize) -> Result<Self::SerializeTuple> { + Err(invalid_raw_value()) + } + + fn serialize_tuple_struct( + self, + _name: &'static str, + _len: usize, + ) -> Result<Self::SerializeTupleStruct> { + Err(invalid_raw_value()) + } + + fn serialize_tuple_variant( + self, + _name: &'static str, + _variant_index: u32, + _variant: &'static str, + _len: usize, + ) -> Result<Self::SerializeTupleVariant> { + Err(invalid_raw_value()) + } + + fn serialize_map(self, _len: Option<usize>) -> Result<Self::SerializeMap> { + Err(invalid_raw_value()) + } + + fn serialize_struct(self, _name: &'static str, _len: usize) -> Result<Self::SerializeStruct> { + Err(invalid_raw_value()) + } + + fn serialize_struct_variant( + self, + _name: &'static str, + _variant_index: u32, + _variant: &'static str, + _len: usize, + ) -> Result<Self::SerializeStructVariant> { + Err(invalid_raw_value()) + } + + fn collect_str<T>(self, value: &T) -> Result<Self::Ok> + where + T: ?Sized + Display, + { + self.serialize_str(&value.to_string()) + } +} |