From c23a457e72abe608715ac76f076f47dc42af07a5 Mon Sep 17 00:00:00 2001 From: Daniel Baumann Date: Thu, 30 May 2024 20:31:44 +0200 Subject: Merging upstream version 1.74.1+dfsg1. Signed-off-by: Daniel Baumann --- compiler/rustc_span/src/lib.rs | 512 +++++++++++++++++++++-------------------- 1 file changed, 268 insertions(+), 244 deletions(-) (limited to 'compiler/rustc_span/src/lib.rs') diff --git a/compiler/rustc_span/src/lib.rs b/compiler/rustc_span/src/lib.rs index c24b8d9ec..772e09291 100644 --- a/compiler/rustc_span/src/lib.rs +++ b/compiler/rustc_span/src/lib.rs @@ -21,9 +21,11 @@ #![feature(rustc_attrs)] #![feature(let_chains)] #![feature(round_char_boundary)] +#![feature(read_buf)] +#![feature(new_uninit)] #![deny(rustc::untranslatable_diagnostic)] #![deny(rustc::diagnostic_outside_of_impl)] -#![cfg_attr(not(bootstrap), allow(internal_features))] +#![allow(internal_features)] #[macro_use] extern crate rustc_macros; @@ -31,7 +33,7 @@ extern crate rustc_macros; #[macro_use] extern crate tracing; -use rustc_data_structures::AtomicRef; +use rustc_data_structures::{outline, AtomicRef}; use rustc_macros::HashStable_Generic; use rustc_serialize::{Decodable, Decoder, Encodable, Encoder}; @@ -62,7 +64,7 @@ pub mod fatal_error; pub mod profiling; use rustc_data_structures::stable_hasher::{Hash128, Hash64, HashStable, StableHasher}; -use rustc_data_structures::sync::{Lock, Lrc}; +use rustc_data_structures::sync::{FreezeLock, FreezeWriteGuard, Lock, Lrc}; use std::borrow::Cow; use std::cmp::{self, Ordering}; @@ -508,10 +510,6 @@ impl SpanData { pub fn is_dummy(self) -> bool { self.lo.0 == 0 && self.hi.0 == 0 } - #[inline] - pub fn is_visible(self, sm: &SourceMap) -> bool { - !self.is_dummy() && sm.is_span_accessible(self.span()) - } /// Returns `true` if `self` fully encloses `other`. pub fn contains(self, other: Self) -> bool { self.lo <= other.lo && other.hi <= self.hi @@ -571,15 +569,9 @@ impl Span { self.data().with_parent(ctxt) } - /// Returns `true` if this is a dummy span with any hygienic context. - #[inline] - pub fn is_dummy(self) -> bool { - self.data_untracked().is_dummy() - } - #[inline] pub fn is_visible(self, sm: &SourceMap) -> bool { - self.data_untracked().is_visible(sm) + !self.is_dummy() && sm.is_span_accessible(self) } /// Returns `true` if this span comes from any kind of macro, desugaring or inlining. @@ -1105,27 +1097,27 @@ impl fmt::Debug for SpanData { } /// Identifies an offset of a multi-byte character in a `SourceFile`. -#[derive(Copy, Clone, Encodable, Decodable, Eq, PartialEq, Debug)] +#[derive(Copy, Clone, Encodable, Decodable, Eq, PartialEq, Debug, HashStable_Generic)] pub struct MultiByteChar { - /// The absolute offset of the character in the `SourceMap`. - pub pos: BytePos, + /// The relative offset of the character in the `SourceFile`. + pub pos: RelativeBytePos, /// The number of bytes, `>= 2`. pub bytes: u8, } /// Identifies an offset of a non-narrow character in a `SourceFile`. -#[derive(Copy, Clone, Encodable, Decodable, Eq, PartialEq, Debug)] +#[derive(Copy, Clone, Encodable, Decodable, Eq, PartialEq, Debug, HashStable_Generic)] pub enum NonNarrowChar { /// Represents a zero-width character. - ZeroWidth(BytePos), + ZeroWidth(RelativeBytePos), /// Represents a wide (full-width) character. - Wide(BytePos), + Wide(RelativeBytePos), /// Represents a tab character, represented visually with a width of 4 characters. - Tab(BytePos), + Tab(RelativeBytePos), } impl NonNarrowChar { - fn new(pos: BytePos, width: usize) -> Self { + fn new(pos: RelativeBytePos, width: usize) -> Self { match width { 0 => NonNarrowChar::ZeroWidth(pos), 2 => NonNarrowChar::Wide(pos), @@ -1134,8 +1126,8 @@ impl NonNarrowChar { } } - /// Returns the absolute offset of the character in the `SourceMap`. - pub fn pos(&self) -> BytePos { + /// Returns the relative offset of the character in the `SourceFile`. + pub fn pos(&self) -> RelativeBytePos { match *self { NonNarrowChar::ZeroWidth(p) | NonNarrowChar::Wide(p) | NonNarrowChar::Tab(p) => p, } @@ -1151,10 +1143,10 @@ impl NonNarrowChar { } } -impl Add for NonNarrowChar { +impl Add for NonNarrowChar { type Output = Self; - fn add(self, rhs: BytePos) -> Self { + fn add(self, rhs: RelativeBytePos) -> Self { match self { NonNarrowChar::ZeroWidth(pos) => NonNarrowChar::ZeroWidth(pos + rhs), NonNarrowChar::Wide(pos) => NonNarrowChar::Wide(pos + rhs), @@ -1163,10 +1155,10 @@ impl Add for NonNarrowChar { } } -impl Sub for NonNarrowChar { +impl Sub for NonNarrowChar { type Output = Self; - fn sub(self, rhs: BytePos) -> Self { + fn sub(self, rhs: RelativeBytePos) -> Self { match self { NonNarrowChar::ZeroWidth(pos) => NonNarrowChar::ZeroWidth(pos - rhs), NonNarrowChar::Wide(pos) => NonNarrowChar::Wide(pos - rhs), @@ -1176,10 +1168,10 @@ impl Sub for NonNarrowChar { } /// Identifies an offset of a character that was normalized away from `SourceFile`. -#[derive(Copy, Clone, Encodable, Decodable, Eq, PartialEq, Debug)] +#[derive(Copy, Clone, Encodable, Decodable, Eq, PartialEq, Debug, HashStable_Generic)] pub struct NormalizedPos { - /// The absolute offset of the character in the `SourceMap`. - pub pos: BytePos, + /// The relative offset of the character in the `SourceFile`. + pub pos: RelativeBytePos, /// The difference between original and normalized string at position. pub diff: u32, } @@ -1204,7 +1196,6 @@ pub enum ExternalSourceKind { AbsentOk, /// A failed attempt has been made to load the external source. AbsentErr, - Unneeded, } impl ExternalSource { @@ -1291,7 +1282,7 @@ impl SourceFileHash { #[derive(Clone)] pub enum SourceFileLines { /// The source file lines, in decoded (random-access) form. - Lines(Vec), + Lines(Vec), /// The source file lines, in undecoded difference list form. Diffs(SourceFileDiffs), @@ -1312,11 +1303,6 @@ impl SourceFileLines { /// small crates where very little of `std`'s metadata is used. #[derive(Clone)] pub struct SourceFileDiffs { - /// Position of the first line. Note that this is always encoded as a - /// `BytePos` because it is often much larger than any of the - /// differences. - line_start: BytePos, - /// Always 1, 2, or 4. Always as small as possible, while being big /// enough to hold the length of the longest line in the source file. /// The 1 case is by far the most common. @@ -1346,13 +1332,13 @@ pub struct SourceFile { pub src_hash: SourceFileHash, /// The external source code (used for external crates, which will have a `None` /// value as `self.src`. - pub external_src: Lock, + pub external_src: FreezeLock, /// The start position of this source in the `SourceMap`. pub start_pos: BytePos, - /// The end position of this source in the `SourceMap`. - pub end_pos: BytePos, + /// The byte length of this source. + pub source_len: RelativeBytePos, /// Locations of lines beginnings in the source code. - pub lines: Lock, + pub lines: FreezeLock, /// Locations of multi-byte characters in the source code. pub multibyte_chars: Vec, /// Width of characters that are not narrow in the source code. @@ -1371,10 +1357,10 @@ impl Clone for SourceFile { name: self.name.clone(), src: self.src.clone(), src_hash: self.src_hash, - external_src: Lock::new(self.external_src.borrow().clone()), + external_src: self.external_src.clone(), start_pos: self.start_pos, - end_pos: self.end_pos, - lines: Lock::new(self.lines.borrow().clone()), + source_len: self.source_len, + lines: self.lines.clone(), multibyte_chars: self.multibyte_chars.clone(), non_narrow_chars: self.non_narrow_chars.clone(), normalized_pos: self.normalized_pos.clone(), @@ -1388,68 +1374,67 @@ impl Encodable for SourceFile { fn encode(&self, s: &mut S) { self.name.encode(s); self.src_hash.encode(s); - self.start_pos.encode(s); - self.end_pos.encode(s); + // Do not encode `start_pos` as it's global state for this session. + self.source_len.encode(s); // We are always in `Lines` form by the time we reach here. - assert!(self.lines.borrow().is_lines()); - self.lines(|lines| { - // Store the length. - s.emit_u32(lines.len() as u32); - - // Compute and store the difference list. - if lines.len() != 0 { - let max_line_length = if lines.len() == 1 { - 0 - } else { - lines - .array_windows() - .map(|&[fst, snd]| snd - fst) - .map(|bp| bp.to_usize()) - .max() - .unwrap() - }; - - let bytes_per_diff: usize = match max_line_length { - 0..=0xFF => 1, - 0x100..=0xFFFF => 2, - _ => 4, - }; - - // Encode the number of bytes used per diff. - s.emit_u8(bytes_per_diff as u8); - - // Encode the first element. - lines[0].encode(s); - - // Encode the difference list. - let diff_iter = lines.array_windows().map(|&[fst, snd]| snd - fst); - let num_diffs = lines.len() - 1; - let mut raw_diffs; - match bytes_per_diff { - 1 => { - raw_diffs = Vec::with_capacity(num_diffs); - for diff in diff_iter { - raw_diffs.push(diff.0 as u8); - } + assert!(self.lines.read().is_lines()); + let lines = self.lines(); + // Store the length. + s.emit_u32(lines.len() as u32); + + // Compute and store the difference list. + if lines.len() != 0 { + let max_line_length = if lines.len() == 1 { + 0 + } else { + lines + .array_windows() + .map(|&[fst, snd]| snd - fst) + .map(|bp| bp.to_usize()) + .max() + .unwrap() + }; + + let bytes_per_diff: usize = match max_line_length { + 0..=0xFF => 1, + 0x100..=0xFFFF => 2, + _ => 4, + }; + + // Encode the number of bytes used per diff. + s.emit_u8(bytes_per_diff as u8); + + // Encode the first element. + assert_eq!(lines[0], RelativeBytePos(0)); + + // Encode the difference list. + let diff_iter = lines.array_windows().map(|&[fst, snd]| snd - fst); + let num_diffs = lines.len() - 1; + let mut raw_diffs; + match bytes_per_diff { + 1 => { + raw_diffs = Vec::with_capacity(num_diffs); + for diff in diff_iter { + raw_diffs.push(diff.0 as u8); } - 2 => { - raw_diffs = Vec::with_capacity(bytes_per_diff * num_diffs); - for diff in diff_iter { - raw_diffs.extend_from_slice(&(diff.0 as u16).to_le_bytes()); - } + } + 2 => { + raw_diffs = Vec::with_capacity(bytes_per_diff * num_diffs); + for diff in diff_iter { + raw_diffs.extend_from_slice(&(diff.0 as u16).to_le_bytes()); } - 4 => { - raw_diffs = Vec::with_capacity(bytes_per_diff * num_diffs); - for diff in diff_iter { - raw_diffs.extend_from_slice(&(diff.0).to_le_bytes()); - } + } + 4 => { + raw_diffs = Vec::with_capacity(bytes_per_diff * num_diffs); + for diff in diff_iter { + raw_diffs.extend_from_slice(&(diff.0).to_le_bytes()); } - _ => unreachable!(), } - s.emit_raw_bytes(&raw_diffs); + _ => unreachable!(), } - }); + s.emit_raw_bytes(&raw_diffs); + } self.multibyte_chars.encode(s); self.non_narrow_chars.encode(s); @@ -1463,26 +1448,17 @@ impl Decodable for SourceFile { fn decode(d: &mut D) -> SourceFile { let name: FileName = Decodable::decode(d); let src_hash: SourceFileHash = Decodable::decode(d); - let start_pos: BytePos = Decodable::decode(d); - let end_pos: BytePos = Decodable::decode(d); + let source_len: RelativeBytePos = Decodable::decode(d); let lines = { let num_lines: u32 = Decodable::decode(d); if num_lines > 0 { // Read the number of bytes used per diff. let bytes_per_diff = d.read_u8() as usize; - // Read the first element. - let line_start: BytePos = Decodable::decode(d); - // Read the difference list. let num_diffs = num_lines as usize - 1; let raw_diffs = d.read_raw_bytes(bytes_per_diff * num_diffs).to_vec(); - SourceFileLines::Diffs(SourceFileDiffs { - line_start, - bytes_per_diff, - num_diffs, - raw_diffs, - }) + SourceFileLines::Diffs(SourceFileDiffs { bytes_per_diff, num_diffs, raw_diffs }) } else { SourceFileLines::Lines(vec![]) } @@ -1494,14 +1470,14 @@ impl Decodable for SourceFile { let cnum: CrateNum = Decodable::decode(d); SourceFile { name, - start_pos, - end_pos, + start_pos: BytePos::from_u32(0), + source_len, src: None, src_hash, // Unused - the metadata decoder will construct // a new SourceFile, filling in `external_src` properly - external_src: Lock::new(ExternalSource::Unneeded), - lines: Lock::new(lines), + external_src: FreezeLock::frozen(ExternalSource::Unneeded), + lines: FreezeLock::new(lines), multibyte_chars, non_narrow_chars, normalized_pos, @@ -1521,102 +1497,116 @@ impl SourceFile { pub fn new( name: FileName, mut src: String, - start_pos: BytePos, hash_kind: SourceFileHashAlgorithm, - ) -> Self { + ) -> Result { // Compute the file hash before any normalization. let src_hash = SourceFileHash::new(hash_kind, &src); - let normalized_pos = normalize_src(&mut src, start_pos); + let normalized_pos = normalize_src(&mut src); let name_hash = { let mut hasher: StableHasher = StableHasher::new(); name.hash(&mut hasher); hasher.finish() }; - let end_pos = start_pos.to_usize() + src.len(); - assert!(end_pos <= u32::MAX as usize); + let source_len = src.len(); + let source_len = u32::try_from(source_len).map_err(|_| OffsetOverflowError)?; let (lines, multibyte_chars, non_narrow_chars) = - analyze_source_file::analyze_source_file(&src, start_pos); + analyze_source_file::analyze_source_file(&src); - SourceFile { + Ok(SourceFile { name, src: Some(Lrc::new(src)), src_hash, - external_src: Lock::new(ExternalSource::Unneeded), - start_pos, - end_pos: Pos::from_usize(end_pos), - lines: Lock::new(SourceFileLines::Lines(lines)), + external_src: FreezeLock::frozen(ExternalSource::Unneeded), + start_pos: BytePos::from_u32(0), + source_len: RelativeBytePos::from_u32(source_len), + lines: FreezeLock::frozen(SourceFileLines::Lines(lines)), multibyte_chars, non_narrow_chars, normalized_pos, name_hash, cnum: LOCAL_CRATE, - } + }) } - pub fn lines(&self, f: F) -> R - where - F: FnOnce(&[BytePos]) -> R, - { - let mut guard = self.lines.borrow_mut(); - match &*guard { - SourceFileLines::Lines(lines) => f(lines), - SourceFileLines::Diffs(SourceFileDiffs { - mut line_start, - bytes_per_diff, - num_diffs, - raw_diffs, - }) => { - // Convert from "diffs" form to "lines" form. - let num_lines = num_diffs + 1; - let mut lines = Vec::with_capacity(num_lines); - lines.push(line_start); - - assert_eq!(*num_diffs, raw_diffs.len() / bytes_per_diff); - match bytes_per_diff { - 1 => { - lines.extend(raw_diffs.into_iter().map(|&diff| { - line_start = line_start + BytePos(diff as u32); - line_start - })); - } - 2 => { - lines.extend((0..*num_diffs).map(|i| { - let pos = bytes_per_diff * i; - let bytes = [raw_diffs[pos], raw_diffs[pos + 1]]; - let diff = u16::from_le_bytes(bytes); - line_start = line_start + BytePos(diff as u32); - line_start - })); - } - 4 => { - lines.extend((0..*num_diffs).map(|i| { - let pos = bytes_per_diff * i; - let bytes = [ - raw_diffs[pos], - raw_diffs[pos + 1], - raw_diffs[pos + 2], - raw_diffs[pos + 3], - ]; - let diff = u32::from_le_bytes(bytes); - line_start = line_start + BytePos(diff); - line_start - })); - } - _ => unreachable!(), - } - let res = f(&lines); - *guard = SourceFileLines::Lines(lines); - res + /// This converts the `lines` field to contain `SourceFileLines::Lines` if needed and freezes it. + fn convert_diffs_to_lines_frozen(&self) { + let mut guard = if let Some(guard) = self.lines.try_write() { guard } else { return }; + + let SourceFileDiffs { bytes_per_diff, num_diffs, raw_diffs } = match &*guard { + SourceFileLines::Diffs(diffs) => diffs, + SourceFileLines::Lines(..) => { + FreezeWriteGuard::freeze(guard); + return; + } + }; + + // Convert from "diffs" form to "lines" form. + let num_lines = num_diffs + 1; + let mut lines = Vec::with_capacity(num_lines); + let mut line_start = RelativeBytePos(0); + lines.push(line_start); + + assert_eq!(*num_diffs, raw_diffs.len() / bytes_per_diff); + match bytes_per_diff { + 1 => { + lines.extend(raw_diffs.into_iter().map(|&diff| { + line_start = line_start + RelativeBytePos(diff as u32); + line_start + })); + } + 2 => { + lines.extend((0..*num_diffs).map(|i| { + let pos = bytes_per_diff * i; + let bytes = [raw_diffs[pos], raw_diffs[pos + 1]]; + let diff = u16::from_le_bytes(bytes); + line_start = line_start + RelativeBytePos(diff as u32); + line_start + })); } + 4 => { + lines.extend((0..*num_diffs).map(|i| { + let pos = bytes_per_diff * i; + let bytes = [ + raw_diffs[pos], + raw_diffs[pos + 1], + raw_diffs[pos + 2], + raw_diffs[pos + 3], + ]; + let diff = u32::from_le_bytes(bytes); + line_start = line_start + RelativeBytePos(diff); + line_start + })); + } + _ => unreachable!(), } + + *guard = SourceFileLines::Lines(lines); + + FreezeWriteGuard::freeze(guard); + } + + pub fn lines(&self) -> &[RelativeBytePos] { + if let Some(SourceFileLines::Lines(lines)) = self.lines.get() { + return &lines[..]; + } + + outline(|| { + self.convert_diffs_to_lines_frozen(); + if let Some(SourceFileLines::Lines(lines)) = self.lines.get() { + return &lines[..]; + } + unreachable!() + }) } /// Returns the `BytePos` of the beginning of the current line. pub fn line_begin_pos(&self, pos: BytePos) -> BytePos { + let pos = self.relative_position(pos); let line_index = self.lookup_line(pos).unwrap(); - self.lines(|lines| lines[line_index]) + let line_start_pos = self.lines()[line_index]; + self.absolute_position(line_start_pos) } /// Add externally loaded source. @@ -1627,35 +1617,37 @@ impl SourceFile { where F: FnOnce() -> Option, { - if matches!( - *self.external_src.borrow(), - ExternalSource::Foreign { kind: ExternalSourceKind::AbsentOk, .. } - ) { + if !self.external_src.is_frozen() { let src = get_src(); - let mut external_src = self.external_src.borrow_mut(); - // Check that no-one else have provided the source while we were getting it - if let ExternalSource::Foreign { - kind: src_kind @ ExternalSourceKind::AbsentOk, .. - } = &mut *external_src - { - if let Some(mut src) = src { - // The src_hash needs to be computed on the pre-normalized src. - if self.src_hash.matches(&src) { - normalize_src(&mut src, BytePos::from_usize(0)); - *src_kind = ExternalSourceKind::Present(Lrc::new(src)); - return true; - } + let src = src.and_then(|mut src| { + // The src_hash needs to be computed on the pre-normalized src. + self.src_hash.matches(&src).then(|| { + normalize_src(&mut src); + src + }) + }); + + self.external_src.try_write().map(|mut external_src| { + if let ExternalSource::Foreign { + kind: src_kind @ ExternalSourceKind::AbsentOk, + .. + } = &mut *external_src + { + *src_kind = if let Some(src) = src { + ExternalSourceKind::Present(Lrc::new(src)) + } else { + ExternalSourceKind::AbsentErr + }; } else { - *src_kind = ExternalSourceKind::AbsentErr; + panic!("unexpected state {:?}", *external_src) } - false - } else { - self.src.is_some() || external_src.get_source().is_some() - } - } else { - self.src.is_some() || self.external_src.borrow().get_source().is_some() + // Freeze this so we don't try to load the source again. + FreezeWriteGuard::freeze(external_src) + }); } + + self.src.is_some() || self.external_src.read().get_source().is_some() } /// Gets a line from the list of pre-computed line-beginnings. @@ -1673,9 +1665,8 @@ impl SourceFile { } let begin = { - let line = self.lines(|lines| lines.get(line_number).copied())?; - let begin: BytePos = line - self.start_pos; - begin.to_usize() + let line = self.lines().get(line_number).copied()?; + line.to_usize() }; if let Some(ref src) = self.src { @@ -1698,30 +1689,44 @@ impl SourceFile { } pub fn count_lines(&self) -> usize { - self.lines(|lines| lines.len()) + self.lines().len() + } + + #[inline] + pub fn absolute_position(&self, pos: RelativeBytePos) -> BytePos { + BytePos::from_u32(pos.to_u32() + self.start_pos.to_u32()) + } + + #[inline] + pub fn relative_position(&self, pos: BytePos) -> RelativeBytePos { + RelativeBytePos::from_u32(pos.to_u32() - self.start_pos.to_u32()) + } + + #[inline] + pub fn end_position(&self) -> BytePos { + self.absolute_position(self.source_len) } /// Finds the line containing the given position. The return value is the /// index into the `lines` array of this `SourceFile`, not the 1-based line /// number. If the source_file is empty or the position is located before the /// first line, `None` is returned. - pub fn lookup_line(&self, pos: BytePos) -> Option { - self.lines(|lines| lines.partition_point(|x| x <= &pos).checked_sub(1)) + pub fn lookup_line(&self, pos: RelativeBytePos) -> Option { + self.lines().partition_point(|x| x <= &pos).checked_sub(1) } pub fn line_bounds(&self, line_index: usize) -> Range { if self.is_empty() { - return self.start_pos..self.end_pos; + return self.start_pos..self.start_pos; } - self.lines(|lines| { - assert!(line_index < lines.len()); - if line_index == (lines.len() - 1) { - lines[line_index]..self.end_pos - } else { - lines[line_index]..lines[line_index + 1] - } - }) + let lines = self.lines(); + assert!(line_index < lines.len()); + if line_index == (lines.len() - 1) { + self.absolute_position(lines[line_index])..self.end_position() + } else { + self.absolute_position(lines[line_index])..self.absolute_position(lines[line_index + 1]) + } } /// Returns whether or not the file contains the given `SourceMap` byte @@ -1730,27 +1735,29 @@ impl SourceFile { /// returns true still contain one byte position according to this function. #[inline] pub fn contains(&self, byte_pos: BytePos) -> bool { - byte_pos >= self.start_pos && byte_pos <= self.end_pos + byte_pos >= self.start_pos && byte_pos <= self.end_position() } #[inline] pub fn is_empty(&self) -> bool { - self.start_pos == self.end_pos + self.source_len.to_u32() == 0 } /// Calculates the original byte position relative to the start of the file /// based on the given byte position. - pub fn original_relative_byte_pos(&self, pos: BytePos) -> BytePos { + pub fn original_relative_byte_pos(&self, pos: BytePos) -> RelativeBytePos { + let pos = self.relative_position(pos); + // Diff before any records is 0. Otherwise use the previously recorded // diff as that applies to the following characters until a new diff // is recorded. let diff = match self.normalized_pos.binary_search_by(|np| np.pos.cmp(&pos)) { Ok(i) => self.normalized_pos[i].diff, - Err(i) if i == 0 => 0, + Err(0) => 0, Err(i) => self.normalized_pos[i - 1].diff, }; - BytePos::from_u32(pos.0 - self.start_pos.0 + diff) + RelativeBytePos::from_u32(pos.0 + diff) } /// Calculates a normalized byte position from a byte offset relative to the @@ -1768,15 +1775,15 @@ impl SourceFile { .binary_search_by(|np| (np.pos.0 + np.diff).cmp(&(self.start_pos.0 + offset))) { Ok(i) => self.normalized_pos[i].diff, - Err(i) if i == 0 => 0, + Err(0) => 0, Err(i) => self.normalized_pos[i - 1].diff, }; BytePos::from_u32(self.start_pos.0 + offset - diff) } - /// Converts an absolute `BytePos` to a `CharPos` relative to the `SourceFile`. - pub fn bytepos_to_file_charpos(&self, bpos: BytePos) -> CharPos { + /// Converts an relative `RelativeBytePos` to a `CharPos` relative to the `SourceFile`. + fn bytepos_to_file_charpos(&self, bpos: RelativeBytePos) -> CharPos { // The number of extra bytes due to multibyte chars in the `SourceFile`. let mut total_extra_bytes = 0; @@ -1794,18 +1801,18 @@ impl SourceFile { } } - assert!(self.start_pos.to_u32() + total_extra_bytes <= bpos.to_u32()); - CharPos(bpos.to_usize() - self.start_pos.to_usize() - total_extra_bytes as usize) + assert!(total_extra_bytes <= bpos.to_u32()); + CharPos(bpos.to_usize() - total_extra_bytes as usize) } /// Looks up the file's (1-based) line number and (0-based `CharPos`) column offset, for a - /// given `BytePos`. - pub fn lookup_file_pos(&self, pos: BytePos) -> (usize, CharPos) { + /// given `RelativeBytePos`. + fn lookup_file_pos(&self, pos: RelativeBytePos) -> (usize, CharPos) { let chpos = self.bytepos_to_file_charpos(pos); match self.lookup_line(pos) { Some(a) => { let line = a + 1; // Line numbers start at 1 - let linebpos = self.lines(|lines| lines[a]); + let linebpos = self.lines()[a]; let linechpos = self.bytepos_to_file_charpos(linebpos); let col = chpos - linechpos; debug!("byte pos {:?} is on the line at byte pos {:?}", pos, linebpos); @@ -1821,10 +1828,11 @@ impl SourceFile { /// Looks up the file's (1-based) line number, (0-based `CharPos`) column offset, and (0-based) /// column offset when displayed, for a given `BytePos`. pub fn lookup_file_pos_with_col_display(&self, pos: BytePos) -> (usize, CharPos, usize) { + let pos = self.relative_position(pos); let (line, col_or_chpos) = self.lookup_file_pos(pos); if line > 0 { let col = col_or_chpos; - let linebpos = self.lines(|lines| lines[line - 1]); + let linebpos = self.lines()[line - 1]; let col_display = { let start_width_idx = self .non_narrow_chars @@ -1859,16 +1867,10 @@ impl SourceFile { } /// Normalizes the source code and records the normalizations. -fn normalize_src(src: &mut String, start_pos: BytePos) -> Vec { +fn normalize_src(src: &mut String) -> Vec { let mut normalized_pos = vec![]; remove_bom(src, &mut normalized_pos); normalize_newlines(src, &mut normalized_pos); - - // Offset all the positions by start_pos to match the final file positions. - for np in &mut normalized_pos { - np.pos.0 += start_pos.0; - } - normalized_pos } @@ -1876,7 +1878,7 @@ fn normalize_src(src: &mut String, start_pos: BytePos) -> Vec { fn remove_bom(src: &mut String, normalized_pos: &mut Vec) { if src.starts_with('\u{feff}') { src.drain(..3); - normalized_pos.push(NormalizedPos { pos: BytePos(0), diff: 3 }); + normalized_pos.push(NormalizedPos { pos: RelativeBytePos(0), diff: 3 }); } } @@ -1911,7 +1913,7 @@ fn normalize_newlines(src: &mut String, normalized_pos: &mut Vec) cursor += idx - gap_len; gap_len += 1; normalized_pos.push(NormalizedPos { - pos: BytePos::from_usize(cursor + 1), + pos: RelativeBytePos::from_usize(cursor + 1), diff: original_gap + gap_len as u32, }); } @@ -2013,6 +2015,10 @@ impl_pos! { #[derive(Clone, Copy, PartialEq, Eq, Hash, PartialOrd, Ord, Debug)] pub struct BytePos(pub u32); + /// A byte offset relative to file beginning. + #[derive(Clone, Copy, PartialEq, Eq, Hash, PartialOrd, Ord, Debug)] + pub struct RelativeBytePos(pub u32); + /// A character offset. /// /// Because of multibyte UTF-8 characters, a byte offset @@ -2034,6 +2040,24 @@ impl Decodable for BytePos { } } +impl HashStable for RelativeBytePos { + fn hash_stable(&self, hcx: &mut H, hasher: &mut StableHasher) { + self.0.hash_stable(hcx, hasher); + } +} + +impl Encodable for RelativeBytePos { + fn encode(&self, s: &mut S) { + s.emit_u32(self.0); + } +} + +impl Decodable for RelativeBytePos { + fn decode(d: &mut D) -> RelativeBytePos { + RelativeBytePos(d.read_u32()) + } +} + // _____________________________________________________________________________ // Loc, SourceFileAndLine, SourceFileAndBytePos // -- cgit v1.2.3