diff options
Diffstat (limited to 'vendor/pem-rfc7468/src')
| -rw-r--r-- | vendor/pem-rfc7468/src/decoder.rs | 270 | ||||
| -rw-r--r-- | vendor/pem-rfc7468/src/encoder.rs | 299 | ||||
| -rw-r--r-- | vendor/pem-rfc7468/src/error.rs | 107 | ||||
| -rw-r--r-- | vendor/pem-rfc7468/src/grammar.rs | 232 | ||||
| -rw-r--r-- | vendor/pem-rfc7468/src/lib.rs | 122 |
5 files changed, 1030 insertions, 0 deletions
diff --git a/vendor/pem-rfc7468/src/decoder.rs b/vendor/pem-rfc7468/src/decoder.rs new file mode 100644 index 0000000..16e97fb --- /dev/null +++ b/vendor/pem-rfc7468/src/decoder.rs @@ -0,0 +1,270 @@ +//! Decoder for PEM encapsulated data. +//! +//! From RFC 7468 Section 2: +//! +//! > Textual encoding begins with a line comprising "-----BEGIN ", a +//! > label, and "-----", and ends with a line comprising "-----END ", a +//! > label, and "-----". Between these lines, or "encapsulation +//! > boundaries", are base64-encoded data according to Section 4 of +//! > [RFC 4648]. +//! +//! [RFC 4648]: https://datatracker.ietf.org/doc/html/rfc4648 + +use crate::{ + grammar, Base64Decoder, Error, Result, BASE64_WRAP_WIDTH, POST_ENCAPSULATION_BOUNDARY, + PRE_ENCAPSULATION_BOUNDARY, +}; +use core::str; + +#[cfg(feature = "alloc")] +use alloc::vec::Vec; + +#[cfg(feature = "std")] +use std::io; + +/// Decode a PEM document according to RFC 7468's "Strict" grammar. +/// +/// On success, writes the decoded document into the provided buffer, returning +/// the decoded label and the portion of the provided buffer containing the +/// decoded message. +pub fn decode<'i, 'o>(pem: &'i [u8], buf: &'o mut [u8]) -> Result<(&'i str, &'o [u8])> { + let mut decoder = Decoder::new(pem).map_err(|e| check_for_headers(pem, e))?; + let type_label = decoder.type_label(); + let buf = buf + .get_mut(..decoder.remaining_len()) + .ok_or(Error::Length)?; + let decoded = decoder.decode(buf).map_err(|e| check_for_headers(pem, e))?; + + if decoder.base64.is_finished() { + Ok((type_label, decoded)) + } else { + Err(Error::Length) + } +} + +/// Decode a PEM document according to RFC 7468's "Strict" grammar, returning +/// the result as a [`Vec`] upon success. +#[cfg(feature = "alloc")] +pub fn decode_vec(pem: &[u8]) -> Result<(&str, Vec<u8>)> { + let mut decoder = Decoder::new(pem).map_err(|e| check_for_headers(pem, e))?; + let type_label = decoder.type_label(); + let mut buf = Vec::new(); + decoder + .decode_to_end(&mut buf) + .map_err(|e| check_for_headers(pem, e))?; + Ok((type_label, buf)) +} + +/// Decode the encapsulation boundaries of a PEM document according to RFC 7468's "Strict" grammar. +/// +/// On success, returning the decoded label. +pub fn decode_label(pem: &[u8]) -> Result<&str> { + Ok(Encapsulation::try_from(pem)?.label()) +} + +/// Buffered PEM decoder. +/// +/// Stateful buffered decoder type which decodes an input PEM document according +/// to RFC 7468's "Strict" grammar. +#[derive(Clone)] +pub struct Decoder<'i> { + /// PEM type label. + type_label: &'i str, + + /// Buffered Base64 decoder. + base64: Base64Decoder<'i>, +} + +impl<'i> Decoder<'i> { + /// Create a new PEM [`Decoder`] with the default options. + /// + /// Uses the default 64-character line wrapping. + pub fn new(pem: &'i [u8]) -> Result<Self> { + Self::new_wrapped(pem, BASE64_WRAP_WIDTH) + } + + /// Create a new PEM [`Decoder`] which wraps at the given line width. + pub fn new_wrapped(pem: &'i [u8], line_width: usize) -> Result<Self> { + let encapsulation = Encapsulation::try_from(pem)?; + let type_label = encapsulation.label(); + let base64 = Base64Decoder::new_wrapped(encapsulation.encapsulated_text, line_width)?; + Ok(Self { type_label, base64 }) + } + + /// Get the PEM type label for the input document. + pub fn type_label(&self) -> &'i str { + self.type_label + } + + /// Decode data into the provided output buffer. + /// + /// There must be at least as much remaining Base64 input to be decoded + /// in order to completely fill `buf`. + pub fn decode<'o>(&mut self, buf: &'o mut [u8]) -> Result<&'o [u8]> { + Ok(self.base64.decode(buf)?) + } + + /// Decode all of the remaining data in the input buffer into `buf`. + #[cfg(feature = "alloc")] + pub fn decode_to_end<'o>(&mut self, buf: &'o mut Vec<u8>) -> Result<&'o [u8]> { + Ok(self.base64.decode_to_end(buf)?) + } + + /// Get the decoded length of the remaining PEM data after Base64 decoding. + pub fn remaining_len(&self) -> usize { + self.base64.remaining_len() + } + + /// Are we finished decoding the PEM input? + pub fn is_finished(&self) -> bool { + self.base64.is_finished() + } +} + +impl<'i> From<Decoder<'i>> for Base64Decoder<'i> { + fn from(decoder: Decoder<'i>) -> Base64Decoder<'i> { + decoder.base64 + } +} + +#[cfg(feature = "std")] +impl<'i> io::Read for Decoder<'i> { + fn read(&mut self, buf: &mut [u8]) -> io::Result<usize> { + self.base64.read(buf) + } + + fn read_to_end(&mut self, buf: &mut Vec<u8>) -> io::Result<usize> { + self.base64.read_to_end(buf) + } + + fn read_exact(&mut self, buf: &mut [u8]) -> io::Result<()> { + self.base64.read_exact(buf) + } +} + +/// PEM encapsulation parser. +/// +/// This parser performs an initial pass over the data, locating the +/// pre-encapsulation (`---BEGIN [...]---`) and post-encapsulation +/// (`---END [...]`) boundaries while attempting to avoid branching +/// on the potentially secret Base64-encoded data encapsulated between +/// the two boundaries. +/// +/// It only supports a single encapsulated message at present. Future work +/// could potentially include extending it provide an iterator over a series +/// of encapsulated messages. +#[derive(Copy, Clone, Debug)] +struct Encapsulation<'a> { + /// Type label extracted from the pre/post-encapsulation boundaries. + /// + /// From RFC 7468 Section 2: + /// + /// > The type of data encoded is labeled depending on the type label in + /// > the "-----BEGIN " line (pre-encapsulation boundary). For example, + /// > the line may be "-----BEGIN CERTIFICATE-----" to indicate that the + /// > content is a PKIX certificate (see further below). Generators MUST + /// > put the same label on the "-----END " line (post-encapsulation + /// > boundary) as the corresponding "-----BEGIN " line. Labels are + /// > formally case-sensitive, uppercase, and comprised of zero or more + /// > characters; they do not contain consecutive spaces or hyphen-minuses, + /// > nor do they contain spaces or hyphen-minuses at either end. Parsers + /// > MAY disregard the label in the post-encapsulation boundary instead of + /// > signaling an error if there is a label mismatch: some extant + /// > implementations require the labels to match; others do not. + label: &'a str, + + /// Encapsulated text portion contained between the boundaries. + /// + /// This data should be encoded as Base64, however this type performs no + /// validation of it so it can be handled in constant-time. + encapsulated_text: &'a [u8], +} + +impl<'a> Encapsulation<'a> { + /// Parse the type label and encapsulated text from between the + /// pre/post-encapsulation boundaries. + pub fn parse(data: &'a [u8]) -> Result<Self> { + // Strip the "preamble": optional text occurring before the pre-encapsulation boundary + let data = grammar::strip_preamble(data)?; + + // Parse pre-encapsulation boundary (including label) + let data = data + .strip_prefix(PRE_ENCAPSULATION_BOUNDARY) + .ok_or(Error::PreEncapsulationBoundary)?; + + let (label, body) = grammar::split_label(data).ok_or(Error::Label)?; + + let mut body = match grammar::strip_trailing_eol(body).unwrap_or(body) { + [head @ .., b'-', b'-', b'-', b'-', b'-'] => head, + _ => return Err(Error::PreEncapsulationBoundary), + }; + + // Ensure body ends with a properly labeled post-encapsulation boundary + for &slice in [POST_ENCAPSULATION_BOUNDARY, label.as_bytes()].iter().rev() { + // Ensure the input ends with the post encapsulation boundary as + // well as a matching label + if !body.ends_with(slice) { + return Err(Error::PostEncapsulationBoundary); + } + + let len = body.len().checked_sub(slice.len()).ok_or(Error::Length)?; + body = body.get(..len).ok_or(Error::PostEncapsulationBoundary)?; + } + + let encapsulated_text = + grammar::strip_trailing_eol(body).ok_or(Error::PostEncapsulationBoundary)?; + + Ok(Self { + label, + encapsulated_text, + }) + } + + /// Get the label parsed from the encapsulation boundaries. + pub fn label(self) -> &'a str { + self.label + } +} + +impl<'a> TryFrom<&'a [u8]> for Encapsulation<'a> { + type Error = Error; + + fn try_from(bytes: &'a [u8]) -> Result<Self> { + Self::parse(bytes) + } +} + +/// Check for PEM headers in the input, as they are disallowed by RFC7468. +/// +/// Returns `Error::HeaderDisallowed` if headers are encountered. +fn check_for_headers(pem: &[u8], err: Error) -> Error { + if err == Error::Base64(base64ct::Error::InvalidEncoding) + && pem.iter().any(|&b| b == grammar::CHAR_COLON) + { + Error::HeaderDisallowed + } else { + err + } +} + +#[cfg(test)] +mod tests { + use super::Encapsulation; + + #[test] + fn pkcs8_example() { + let pem = include_bytes!("../tests/examples/pkcs8.pem"); + let encapsulation = Encapsulation::parse(pem).unwrap(); + assert_eq!(encapsulation.label, "PRIVATE KEY"); + + assert_eq!( + encapsulation.encapsulated_text, + &[ + 77, 67, 52, 67, 65, 81, 65, 119, 66, 81, 89, 68, 75, 50, 86, 119, 66, 67, 73, 69, + 73, 66, 102, 116, 110, 72, 80, 112, 50, 50, 83, 101, 119, 89, 109, 109, 69, 111, + 77, 99, 88, 56, 86, 119, 73, 52, 73, 72, 119, 97, 113, 100, 43, 57, 76, 70, 80, + 106, 47, 49, 53, 101, 113, 70 + ] + ); + } +} diff --git a/vendor/pem-rfc7468/src/encoder.rs b/vendor/pem-rfc7468/src/encoder.rs new file mode 100644 index 0000000..c48df1f --- /dev/null +++ b/vendor/pem-rfc7468/src/encoder.rs @@ -0,0 +1,299 @@ +//! PEM encoder. + +use crate::{ + grammar, Base64Encoder, Error, LineEnding, Result, BASE64_WRAP_WIDTH, + ENCAPSULATION_BOUNDARY_DELIMITER, POST_ENCAPSULATION_BOUNDARY, PRE_ENCAPSULATION_BOUNDARY, +}; +use base64ct::{Base64, Encoding}; +use core::str; + +#[cfg(feature = "alloc")] +use alloc::string::String; + +#[cfg(feature = "std")] +use std::io; + +/// Compute the length of a PEM encoded document which encapsulates a +/// Base64-encoded body including line endings every 64 characters. +/// +/// The `input_len` parameter specifies the length of the raw input +/// bytes prior to Base64 encoding. +/// +/// Note that the current implementation of this function computes an upper +/// bound of the length and the actual encoded document may be slightly shorter +/// (typically 1-byte). Downstream consumers of this function should check the +/// actual encoded length and potentially truncate buffers allocated using this +/// function to estimate the encapsulated size. +/// +/// Use [`encoded_len`] (when possible) to obtain a precise length. +/// +/// ## Returns +/// - `Ok(len)` on success +/// - `Err(Error::Length)` on length overflow +pub fn encapsulated_len(label: &str, line_ending: LineEnding, input_len: usize) -> Result<usize> { + encapsulated_len_wrapped(label, BASE64_WRAP_WIDTH, line_ending, input_len) +} + +/// Compute the length of a PEM encoded document with the Base64 body +/// line wrapped at the specified `width`. +/// +/// This is the same as [`encapsulated_len`], which defaults to a width of 64. +/// +/// Note that per [RFC7468 § 2] encoding PEM with any other wrap width besides +/// 64 is technically non-compliant: +/// +/// > Generators MUST wrap the base64-encoded lines so that each line +/// > consists of exactly 64 characters except for the final line, which +/// > will encode the remainder of the data (within the 64-character line +/// > boundary) +/// +/// [RFC7468 § 2]: https://datatracker.ietf.org/doc/html/rfc7468#section-2 +pub fn encapsulated_len_wrapped( + label: &str, + line_width: usize, + line_ending: LineEnding, + input_len: usize, +) -> Result<usize> { + if line_width < 4 { + return Err(Error::Length); + } + + let base64_len = input_len + .checked_mul(4) + .and_then(|n| n.checked_div(3)) + .and_then(|n| n.checked_add(3)) + .ok_or(Error::Length)? + & !3; + + let base64_len_wrapped = base64_len_wrapped(base64_len, line_width, line_ending)?; + encapsulated_len_inner(label, line_ending, base64_len_wrapped) +} + +/// Get the length of a PEM encoded document with the given bytes and label. +/// +/// This function computes a precise length of the PEM encoding of the given +/// `input` data. +/// +/// ## Returns +/// - `Ok(len)` on success +/// - `Err(Error::Length)` on length overflow +pub fn encoded_len(label: &str, line_ending: LineEnding, input: &[u8]) -> Result<usize> { + let base64_len = Base64::encoded_len(input); + let base64_len_wrapped = base64_len_wrapped(base64_len, BASE64_WRAP_WIDTH, line_ending)?; + encapsulated_len_inner(label, line_ending, base64_len_wrapped) +} + +/// Encode a PEM document according to RFC 7468's "Strict" grammar. +pub fn encode<'o>( + type_label: &str, + line_ending: LineEnding, + input: &[u8], + buf: &'o mut [u8], +) -> Result<&'o str> { + let mut encoder = Encoder::new(type_label, line_ending, buf)?; + encoder.encode(input)?; + let encoded_len = encoder.finish()?; + let output = &buf[..encoded_len]; + + // Sanity check + debug_assert!(str::from_utf8(output).is_ok()); + + // Ensure `output` contains characters from the lower 7-bit ASCII set + if output.iter().fold(0u8, |acc, &byte| acc | (byte & 0x80)) == 0 { + // Use unchecked conversion to avoid applying UTF-8 checks to potentially + // secret PEM documents (and therefore introducing a potential timing + // sidechannel) + // + // SAFETY: contents of this buffer are controlled entirely by the encoder, + // which ensures the contents are always a valid (ASCII) subset of UTF-8. + // It's also additionally sanity checked by two assertions above to ensure + // the validity (with the always-on runtime check implemented in a + // constant time-ish manner. + #[allow(unsafe_code)] + Ok(unsafe { str::from_utf8_unchecked(output) }) + } else { + Err(Error::CharacterEncoding) + } +} + +/// Encode a PEM document according to RFC 7468's "Strict" grammar, returning +/// the result as a [`String`]. +#[cfg(feature = "alloc")] +pub fn encode_string(label: &str, line_ending: LineEnding, input: &[u8]) -> Result<String> { + let expected_len = encoded_len(label, line_ending, input)?; + let mut buf = vec![0u8; expected_len]; + let actual_len = encode(label, line_ending, input, &mut buf)?.len(); + debug_assert_eq!(expected_len, actual_len); + String::from_utf8(buf).map_err(|_| Error::CharacterEncoding) +} + +/// Compute the encapsulated length of Base64 data of the given length. +fn encapsulated_len_inner( + label: &str, + line_ending: LineEnding, + base64_len: usize, +) -> Result<usize> { + [ + PRE_ENCAPSULATION_BOUNDARY.len(), + label.as_bytes().len(), + ENCAPSULATION_BOUNDARY_DELIMITER.len(), + line_ending.len(), + base64_len, + line_ending.len(), + POST_ENCAPSULATION_BOUNDARY.len(), + label.as_bytes().len(), + ENCAPSULATION_BOUNDARY_DELIMITER.len(), + line_ending.len(), + ] + .into_iter() + .try_fold(0usize, |acc, len| acc.checked_add(len)) + .ok_or(Error::Length) +} + +/// Compute Base64 length line-wrapped at the specified width with the given +/// line ending. +fn base64_len_wrapped( + base64_len: usize, + line_width: usize, + line_ending: LineEnding, +) -> Result<usize> { + base64_len + .saturating_sub(1) + .checked_div(line_width) + .and_then(|lines| lines.checked_mul(line_ending.len())) + .and_then(|len| len.checked_add(base64_len)) + .ok_or(Error::Length) +} + +/// Buffered PEM encoder. +/// +/// Stateful buffered encoder type which encodes an input PEM document according +/// to RFC 7468's "Strict" grammar. +pub struct Encoder<'l, 'o> { + /// PEM type label. + type_label: &'l str, + + /// Line ending used to wrap Base64. + line_ending: LineEnding, + + /// Buffered Base64 encoder. + base64: Base64Encoder<'o>, +} + +impl<'l, 'o> Encoder<'l, 'o> { + /// Create a new PEM [`Encoder`] with the default options which + /// writes output into the provided buffer. + /// + /// Uses the default 64-character line wrapping. + pub fn new(type_label: &'l str, line_ending: LineEnding, out: &'o mut [u8]) -> Result<Self> { + Self::new_wrapped(type_label, BASE64_WRAP_WIDTH, line_ending, out) + } + + /// Create a new PEM [`Encoder`] which wraps at the given line width. + /// + /// Note that per [RFC7468 § 2] encoding PEM with any other wrap width besides + /// 64 is technically non-compliant: + /// + /// > Generators MUST wrap the base64-encoded lines so that each line + /// > consists of exactly 64 characters except for the final line, which + /// > will encode the remainder of the data (within the 64-character line + /// > boundary) + /// + /// This method is provided with the intended purpose of implementing the + /// OpenSSH private key format, which uses a non-standard wrap width of 70. + /// + /// [RFC7468 § 2]: https://datatracker.ietf.org/doc/html/rfc7468#section-2 + pub fn new_wrapped( + type_label: &'l str, + line_width: usize, + line_ending: LineEnding, + mut out: &'o mut [u8], + ) -> Result<Self> { + grammar::validate_label(type_label.as_bytes())?; + + for boundary_part in [ + PRE_ENCAPSULATION_BOUNDARY, + type_label.as_bytes(), + ENCAPSULATION_BOUNDARY_DELIMITER, + line_ending.as_bytes(), + ] { + if out.len() < boundary_part.len() { + return Err(Error::Length); + } + + let (part, rest) = out.split_at_mut(boundary_part.len()); + out = rest; + + part.copy_from_slice(boundary_part); + } + + let base64 = Base64Encoder::new_wrapped(out, line_width, line_ending)?; + + Ok(Self { + type_label, + line_ending, + base64, + }) + } + + /// Get the PEM type label used for this document. + pub fn type_label(&self) -> &'l str { + self.type_label + } + + /// Encode the provided input data. + /// + /// This method can be called as many times as needed with any sized input + /// to write data encoded data into the output buffer, so long as there is + /// sufficient space in the buffer to handle the resulting Base64 encoded + /// data. + pub fn encode(&mut self, input: &[u8]) -> Result<()> { + self.base64.encode(input)?; + Ok(()) + } + + /// Borrow the inner [`Base64Encoder`]. + pub fn base64_encoder(&mut self) -> &mut Base64Encoder<'o> { + &mut self.base64 + } + + /// Finish encoding PEM, writing the post-encapsulation boundary. + /// + /// On success, returns the total number of bytes written to the output + /// buffer. + pub fn finish(self) -> Result<usize> { + let (base64, mut out) = self.base64.finish_with_remaining()?; + + for boundary_part in [ + self.line_ending.as_bytes(), + POST_ENCAPSULATION_BOUNDARY, + self.type_label.as_bytes(), + ENCAPSULATION_BOUNDARY_DELIMITER, + self.line_ending.as_bytes(), + ] { + if out.len() < boundary_part.len() { + return Err(Error::Length); + } + + let (part, rest) = out.split_at_mut(boundary_part.len()); + out = rest; + + part.copy_from_slice(boundary_part); + } + + encapsulated_len_inner(self.type_label, self.line_ending, base64.len()) + } +} + +#[cfg(feature = "std")] +impl<'l, 'o> io::Write for Encoder<'l, 'o> { + fn write(&mut self, buf: &[u8]) -> io::Result<usize> { + self.encode(buf)?; + Ok(buf.len()) + } + + fn flush(&mut self) -> io::Result<()> { + // TODO(tarcieri): return an error if there's still data remaining in the buffer? + Ok(()) + } +} diff --git a/vendor/pem-rfc7468/src/error.rs b/vendor/pem-rfc7468/src/error.rs new file mode 100644 index 0000000..6a93465 --- /dev/null +++ b/vendor/pem-rfc7468/src/error.rs @@ -0,0 +1,107 @@ +//! Error types + +use core::fmt; + +/// Result type with the `pem-rfc7468` crate's [`Error`] type. +pub type Result<T> = core::result::Result<T, Error>; + +/// PEM errors. +#[derive(Copy, Clone, Debug, Eq, PartialEq)] +#[non_exhaustive] +pub enum Error { + /// Base64-related errors. + Base64(base64ct::Error), + + /// Character encoding-related errors. + CharacterEncoding, + + /// Errors in the encapsulated text (which aren't specifically Base64-related). + EncapsulatedText, + + /// Header detected in the encapsulated text. + HeaderDisallowed, + + /// Invalid label. + Label, + + /// Invalid length. + Length, + + /// "Preamble" (text before pre-encapsulation boundary) contains invalid data. + Preamble, + + /// Errors in the pre-encapsulation boundary. + PreEncapsulationBoundary, + + /// Errors in the post-encapsulation boundary. + PostEncapsulationBoundary, + + /// Unexpected PEM type label. + UnexpectedTypeLabel { + /// Type label that was expected. + expected: &'static str, + }, +} + +impl fmt::Display for Error { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + match self { + Error::Base64(err) => write!(f, "PEM Base64 error: {}", err), + Error::CharacterEncoding => f.write_str("PEM character encoding error"), + Error::EncapsulatedText => f.write_str("PEM error in encapsulated text"), + Error::HeaderDisallowed => f.write_str("PEM headers disallowed by RFC7468"), + Error::Label => f.write_str("PEM type label invalid"), + Error::Length => f.write_str("PEM length invalid"), + Error::Preamble => f.write_str("PEM preamble contains invalid data (NUL byte)"), + Error::PreEncapsulationBoundary => { + f.write_str("PEM error in pre-encapsulation boundary") + } + Error::PostEncapsulationBoundary => { + f.write_str("PEM error in post-encapsulation boundary") + } + Error::UnexpectedTypeLabel { expected } => { + write!(f, "unexpected PEM type label: expecting \"{}\"", expected) + } + } + } +} + +#[cfg(feature = "std")] +impl std::error::Error for Error {} + +impl From<base64ct::Error> for Error { + fn from(err: base64ct::Error) -> Error { + Error::Base64(err) + } +} + +impl From<base64ct::InvalidLengthError> for Error { + fn from(_: base64ct::InvalidLengthError) -> Error { + Error::Length + } +} + +impl From<core::str::Utf8Error> for Error { + fn from(_: core::str::Utf8Error) -> Error { + Error::CharacterEncoding + } +} + +#[cfg(feature = "std")] +impl From<Error> for std::io::Error { + fn from(err: Error) -> std::io::Error { + let kind = match err { + Error::Base64(err) => return err.into(), // Use existing conversion + Error::CharacterEncoding + | Error::EncapsulatedText + | Error::Label + | Error::Preamble + | Error::PreEncapsulationBoundary + | Error::PostEncapsulationBoundary => std::io::ErrorKind::InvalidData, + Error::Length => std::io::ErrorKind::UnexpectedEof, + _ => std::io::ErrorKind::Other, + }; + + std::io::Error::new(kind, err) + } +} diff --git a/vendor/pem-rfc7468/src/grammar.rs b/vendor/pem-rfc7468/src/grammar.rs new file mode 100644 index 0000000..91085fe --- /dev/null +++ b/vendor/pem-rfc7468/src/grammar.rs @@ -0,0 +1,232 @@ +//! Helper functions and rules for enforcing the ABNF grammar for +//! RFC 7468-flavored PEM as described in Section 3. +//! +//! The grammar described below is intended to follow the "ABNF (Strict)" +//! subset of the grammar as described in Section 3 Figure 3. + +use crate::{Error, Result, PRE_ENCAPSULATION_BOUNDARY}; +use core::str; + +/// NUL char +pub(crate) const CHAR_NUL: u8 = 0x00; + +/// Horizontal tab +pub(crate) const CHAR_HT: u8 = 0x09; + +/// Space +pub(crate) const CHAR_SP: u8 = 0x20; + +/// Carriage return +pub(crate) const CHAR_CR: u8 = 0x0d; + +/// Line feed +pub(crate) const CHAR_LF: u8 = 0x0a; + +/// Colon ':' +pub(crate) const CHAR_COLON: u8 = 0x3A; + +/// Any printable character except hyphen-minus, as defined in the +/// 'labelchar' production in the RFC 7468 ABNF grammar +pub(crate) fn is_labelchar(char: u8) -> bool { + matches!(char, 0x21..=0x2C | 0x2E..=0x7E) +} + +/// Does the provided byte match a character allowed in a label? +// TODO: allow hyphen-minus to match the 'label' production in the ABNF grammar +pub(crate) fn is_allowed_in_label(char: u8) -> bool { + is_labelchar(char) || matches!(char, CHAR_HT | CHAR_SP) +} + +/// Does the provided byte match the "WSP" ABNF production from Section 3? +/// +/// > The common ABNF production WSP is congruent with "blank"; +/// > a new production W is used for "whitespace" +pub(crate) fn is_wsp(char: u8) -> bool { + matches!(char, CHAR_HT | CHAR_SP) +} + +/// Strip the "preamble", i.e. data that appears before the PEM +/// pre-encapsulation boundary. +/// +/// Presently no attempt is made to ensure the preamble decodes successfully +/// under any particular character encoding. The only byte which is disallowed +/// is the NUL byte. This restriction does not appear in RFC7468, but rather +/// is inspired by the OpenSSL PEM decoder. +/// +/// Returns a slice which starts at the beginning of the encapsulated text. +/// +/// From RFC7468: +/// > Data before the encapsulation boundaries are permitted, and +/// > parsers MUST NOT malfunction when processing such data. +pub(crate) fn strip_preamble(mut bytes: &[u8]) -> Result<&[u8]> { + if bytes.starts_with(PRE_ENCAPSULATION_BOUNDARY) { + return Ok(bytes); + } + + while let Some((byte, remaining)) = bytes.split_first() { + match *byte { + CHAR_NUL => { + return Err(Error::Preamble); + } + CHAR_LF if remaining.starts_with(PRE_ENCAPSULATION_BOUNDARY) => { + return Ok(remaining); + } + _ => (), + } + + bytes = remaining; + } + + Err(Error::Preamble) +} + +/// Strip a newline (`eol`) from the beginning of the provided byte slice. +/// +/// The newline is considered mandatory and a decoding error will occur if it +/// is not present. +/// +/// From RFC 7468 Section 3: +/// > lines are divided with CRLF, CR, or LF. +pub(crate) fn strip_leading_eol(bytes: &[u8]) -> Option<&[u8]> { + match bytes { + [CHAR_LF, rest @ ..] => Some(rest), + [CHAR_CR, CHAR_LF, rest @ ..] => Some(rest), + [CHAR_CR, rest @ ..] => Some(rest), + _ => None, + } +} + +/// Strip a newline (`eol`) from the end of the provided byte slice. +/// +/// The newline is considered mandatory and a decoding error will occur if it +/// is not present. +/// +/// From RFC 7468 Section 3: +/// > lines are divided with CRLF, CR, or LF. +pub(crate) fn strip_trailing_eol(bytes: &[u8]) -> Option<&[u8]> { + match bytes { + [head @ .., CHAR_CR, CHAR_LF] => Some(head), + [head @ .., CHAR_LF] => Some(head), + [head @ .., CHAR_CR] => Some(head), + _ => None, + } +} + +/// Split a slice beginning with a type label as located in an encapsulation +/// boundary. Returns the label as a `&str`, and slice beginning with the +/// encapsulated text with leading `-----` and newline removed. +/// +/// This implementation follows the rules put forth in Section 2, which are +/// stricter than those found in the ABNF grammar: +/// +/// > Labels are formally case-sensitive, uppercase, and comprised of zero or more +/// > characters; they do not contain consecutive spaces or hyphen-minuses, +/// > nor do they contain spaces or hyphen-minuses at either end. +/// +/// We apply a slightly stricter interpretation: +/// - Labels MAY be empty +/// - Non-empty labels MUST start with an upper-case letter: `'A'..='Z'` +/// - The only allowable characters subsequently are `'A'..='Z'` or WSP. +/// (NOTE: this is an overly strict initial implementation and should be relaxed) +/// - Whitespace MUST NOT contain more than one consecutive WSP character +// TODO(tarcieri): evaluate whether this is too strict; support '-' +pub(crate) fn split_label(bytes: &[u8]) -> Option<(&str, &[u8])> { + let mut n = 0usize; + + // TODO(tarcieri): handle hyphens in labels as well as spaces + let mut last_was_wsp = false; + + for &char in bytes { + // Validate character + if is_labelchar(char) { + last_was_wsp = false; + } else if char == b'-' { + // Possible start of encapsulation boundary delimiter + break; + } else if n != 0 && is_wsp(char) { + // Repeated whitespace disallowed + if last_was_wsp { + return None; + } + + last_was_wsp = true; + } else { + return None; + } + + n = n.checked_add(1)?; + } + + let (raw_label, rest) = bytes.split_at(n); + let label = str::from_utf8(raw_label).ok()?; + + match rest { + [b'-', b'-', b'-', b'-', b'-', body @ ..] => Some((label, strip_leading_eol(body)?)), + _ => None, + } +} + +/// Validate that the given bytes are allowed as a PEM type label, i.e. the +/// label encoded in the `BEGIN` and `END` encapsulation boundaries. +pub(crate) fn validate_label(label: &[u8]) -> Result<()> { + // TODO(tarcieri): handle hyphens in labels as well as spaces + let mut last_was_wsp = false; + + for &char in label { + if !is_allowed_in_label(char) { + return Err(Error::Label); + } + + if is_wsp(char) { + // Double sequential whitespace characters disallowed + if last_was_wsp { + return Err(Error::Label); + } + + last_was_wsp = true; + } else { + last_was_wsp = false; + } + } + + Ok(()) +} + +#[cfg(test)] +mod tests { + use super::*; + + /// Empty label is OK. + #[test] + fn split_label_empty() { + let (label, body) = split_label(b"-----\nBODY").unwrap(); + assert_eq!(label, ""); + assert_eq!(body, b"BODY"); + } + + /// Label containing text. + #[test] + fn split_label_with_text() { + let (label, body) = split_label(b"PRIVATE KEY-----\nBODY").unwrap(); + assert_eq!(label, "PRIVATE KEY"); + assert_eq!(body, b"BODY"); + } + + /// Reject labels containing repeated spaces + #[test] + fn split_label_with_repeat_wsp_is_err() { + assert!(split_label(b"PRIVATE KEY-----\nBODY").is_none()); + } + + /// Basic validation of a label + #[test] + fn validate_private_key_label() { + assert_eq!(validate_label(b"PRIVATE KEY"), Ok(())); + } + + /// Reject labels with double spaces + #[test] + fn validate_private_key_label_reject_double_space() { + assert_eq!(validate_label(b"PRIVATE KEY"), Err(Error::Label)); + } +} diff --git a/vendor/pem-rfc7468/src/lib.rs b/vendor/pem-rfc7468/src/lib.rs new file mode 100644 index 0000000..45f4877 --- /dev/null +++ b/vendor/pem-rfc7468/src/lib.rs @@ -0,0 +1,122 @@ +#![no_std] +#![cfg_attr(docsrs, feature(doc_auto_cfg))] +#![doc = include_str!("../README.md")] +#![doc( + html_logo_url = "https://raw.githubusercontent.com/RustCrypto/media/6ee8e381/logo.svg", + html_favicon_url = "https://raw.githubusercontent.com/RustCrypto/media/6ee8e381/logo.svg" +)] +#![deny(unsafe_code)] +#![warn( + clippy::integer_arithmetic, + clippy::mod_module_files, + clippy::panic, + clippy::panic_in_result_fn, + clippy::unwrap_used, + missing_docs, + rust_2018_idioms, + unused_lifetimes, + unused_qualifications +)] + +//! # Usage +//! +#![cfg_attr(feature = "std", doc = " ```")] +#![cfg_attr(not(feature = "std"), doc = " ```ignore")] +//! # fn main() -> Result<(), Box<dyn std::error::Error>> { +//! /// Example PEM document +//! /// NOTE: do not actually put private key literals into your source code!!! +//! let example_pem = "\ +//! -----BEGIN PRIVATE KEY----- +//! MC4CAQAwBQYDK2VwBCIEIBftnHPp22SewYmmEoMcX8VwI4IHwaqd+9LFPj/15eqF +//! -----END PRIVATE KEY----- +//! "; +//! +//! // Decode PEM +//! let (type_label, data) = pem_rfc7468::decode_vec(example_pem.as_bytes())?; +//! assert_eq!(type_label, "PRIVATE KEY"); +//! assert_eq!( +//! data, +//! &[ +//! 48, 46, 2, 1, 0, 48, 5, 6, 3, 43, 101, 112, 4, 34, 4, 32, 23, 237, 156, 115, 233, 219, +//! 100, 158, 193, 137, 166, 18, 131, 28, 95, 197, 112, 35, 130, 7, 193, 170, 157, 251, +//! 210, 197, 62, 63, 245, 229, 234, 133 +//! ] +//! ); +//! +//! // Encode PEM +//! use pem_rfc7468::LineEnding; +//! let encoded_pem = pem_rfc7468::encode_string(type_label, LineEnding::default(), &data)?; +//! assert_eq!(&encoded_pem, example_pem); +//! # Ok(()) +//! # } +//! ``` + +#[cfg(feature = "alloc")] +#[macro_use] +extern crate alloc; +#[cfg(feature = "std")] +extern crate std; + +mod decoder; +mod encoder; +mod error; +mod grammar; + +pub use crate::{ + decoder::{decode, decode_label, Decoder}, + encoder::{encapsulated_len, encapsulated_len_wrapped, encode, encoded_len, Encoder}, + error::{Error, Result}, +}; +pub use base64ct::LineEnding; + +#[cfg(feature = "alloc")] +pub use crate::{decoder::decode_vec, encoder::encode_string}; + +/// The pre-encapsulation boundary appears before the encapsulated text. +/// +/// From RFC 7468 Section 2: +/// > There are exactly five hyphen-minus (also known as dash) characters ("-") +/// > on both ends of the encapsulation boundaries, no more, no less. +const PRE_ENCAPSULATION_BOUNDARY: &[u8] = b"-----BEGIN "; + +/// The post-encapsulation boundary appears immediately after the encapsulated text. +const POST_ENCAPSULATION_BOUNDARY: &[u8] = b"-----END "; + +/// Delimiter of encapsulation boundaries. +const ENCAPSULATION_BOUNDARY_DELIMITER: &[u8] = b"-----"; + +/// Width at which the Base64 body of RFC7468-compliant PEM is wrapped. +/// +/// From [RFC7468 § 2]: +/// +/// > Generators MUST wrap the base64-encoded lines so that each line +/// > consists of exactly 64 characters except for the final line, which +/// > will encode the remainder of the data (within the 64-character line +/// > boundary), and they MUST NOT emit extraneous whitespace. Parsers MAY +/// > handle other line sizes. +/// +/// [RFC7468 § 2]: https://datatracker.ietf.org/doc/html/rfc7468#section-2 +pub const BASE64_WRAP_WIDTH: usize = 64; + +/// Buffered Base64 decoder type. +pub type Base64Decoder<'i> = base64ct::Decoder<'i, base64ct::Base64>; + +/// Buffered Base64 encoder type. +pub type Base64Encoder<'o> = base64ct::Encoder<'o, base64ct::Base64>; + +/// Marker trait for types with an associated PEM type label. +pub trait PemLabel { + /// Expected PEM type label for a given document, e.g. `"PRIVATE KEY"` + const PEM_LABEL: &'static str; + + /// Validate that a given label matches the expected label. + fn validate_pem_label(actual: &str) -> Result<()> { + if Self::PEM_LABEL == actual { + Ok(()) + } else { + Err(Error::UnexpectedTypeLabel { + expected: Self::PEM_LABEL, + }) + } + } +} |