1 // Copyright 2012-2013 The Rust Project Developers. See the COPYRIGHT
2 // file at the top-level directory of this distribution and at
3 // http://rust-lang.org/COPYRIGHT.
5 // Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
6 // http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
7 // <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
8 // option. This file may not be copied, modified, or distributed
9 // except according to those terms.
11 //! The source positions and related helper functions
15 //! This API is completely unstable and subject to change.
17 #![crate_name = "syntax_pos"]
18 #![crate_type = "dylib"]
19 #![crate_type = "rlib"]
20 #![doc(html_logo_url = "https://www.rust-lang.org/logos/rust-logo-128x128-blk-v2.png",
21 html_favicon_url = "https://doc.rust-lang.org/favicon.ico",
22 html_root_url = "https://doc.rust-lang.org/nightly/")]
26 #![feature(custom_attribute)]
27 #![feature(optin_builtin_traits)]
28 #![allow(unused_attributes)]
29 #![feature(specialization)]
31 #![cfg_attr(stage0, unstable(feature = "rustc_private", issue = "27812"))]
32 #![cfg_attr(stage0, feature(rustc_private))]
33 #![cfg_attr(stage0, feature(staged_api))]
35 use std::cell::{Cell, RefCell};
36 use std::ops::{Add, Sub};
42 use serialize::{Encodable, Decodable, Encoder, Decoder};
44 extern crate serialize;
45 extern crate serialize as rustc_serialize; // used by deriving
48 pub use hygiene::{SyntaxContext, ExpnInfo, ExpnFormat, NameAndSpan};
52 pub type FileName = String;
54 /// Spans represent a region of code, used for error reporting. Positions in spans
55 /// are *absolute* positions from the beginning of the codemap, not positions
56 /// relative to FileMaps. Methods on the CodeMap can be used to relate spans back
57 /// to the original source.
58 /// You must be careful if the span crosses more than one file - you will not be
59 /// able to use many of the functions on spans in codemap and you cannot assume
60 /// that the length of the span = hi - lo; there may be space in the BytePos
61 /// range between files.
62 #[derive(Clone, Copy, Hash, PartialEq, Eq, Ord, PartialOrd)]
66 /// Information about where the macro came from, if this piece of
67 /// code was created by a macro expansion.
68 pub ctxt: SyntaxContext,
71 /// A collection of spans. Spans have two orthogonal attributes:
73 /// - they can be *primary spans*. In this case they are the locus of
74 /// the error, and would be rendered with `^^^`.
75 /// - they can have a *label*. In this case, the label is written next
76 /// to the mark in the snippet when we render.
77 #[derive(Clone, Debug, Hash, PartialEq, Eq, RustcEncodable, RustcDecodable)]
78 pub struct MultiSpan {
79 primary_spans: Vec<Span>,
80 span_labels: Vec<(Span, String)>,
84 /// Returns a new span representing just the end-point of this span
85 pub fn end_point(self) -> Span {
86 let lo = cmp::max(self.hi.0 - 1, self.lo.0);
87 Span { lo: BytePos(lo), ..self }
90 /// Returns a new span representing the next character after the end-point of this span
91 pub fn next_point(self) -> Span {
92 let lo = cmp::max(self.hi.0, self.lo.0 + 1);
93 Span { lo: BytePos(lo), hi: BytePos(lo), ..self }
96 /// Returns `self` if `self` is not the dummy span, and `other` otherwise.
97 pub fn substitute_dummy(self, other: Span) -> Span {
98 if self.source_equal(&DUMMY_SP) { other } else { self }
101 pub fn contains(self, other: Span) -> bool {
102 self.lo <= other.lo && other.hi <= self.hi
105 /// Return true if the spans are equal with regards to the source text.
107 /// Use this instead of `==` when either span could be generated code,
108 /// and you only care that they point to the same bytes of source text.
109 pub fn source_equal(&self, other: &Span) -> bool {
110 self.lo == other.lo && self.hi == other.hi
113 /// Returns `Some(span)`, where the start is trimmed by the end of `other`
114 pub fn trim_start(self, other: Span) -> Option<Span> {
115 if self.hi > other.hi {
116 Some(Span { lo: cmp::max(self.lo, other.hi), .. self })
122 /// Return the source span - this is either the supplied span, or the span for
123 /// the macro callsite that expanded to it.
124 pub fn source_callsite(self) -> Span {
125 self.ctxt.outer().expn_info().map(|info| info.call_site.source_callsite()).unwrap_or(self)
128 /// Return the source callee.
130 /// Returns None if the supplied span has no expansion trace,
131 /// else returns the NameAndSpan for the macro definition
132 /// corresponding to the source callsite.
133 pub fn source_callee(self) -> Option<NameAndSpan> {
134 fn source_callee(info: ExpnInfo) -> NameAndSpan {
135 match info.call_site.ctxt.outer().expn_info() {
136 Some(info) => source_callee(info),
140 self.ctxt.outer().expn_info().map(source_callee)
143 /// Check if a span is "internal" to a macro in which #[unstable]
144 /// items can be used (that is, a macro marked with
145 /// `#[allow_internal_unstable]`).
146 pub fn allows_unstable(&self) -> bool {
147 match self.ctxt.outer().expn_info() {
148 Some(info) => info.callee.allow_internal_unstable,
153 pub fn macro_backtrace(mut self) -> Vec<MacroBacktrace> {
154 let mut prev_span = DUMMY_SP;
155 let mut result = vec![];
157 let info = match self.ctxt.outer().expn_info() {
162 let (pre, post) = match info.callee.format {
163 ExpnFormat::MacroAttribute(..) => ("#[", "]"),
164 ExpnFormat::MacroBang(..) => ("", "!"),
165 ExpnFormat::CompilerDesugaring(..) => ("desugaring of `", "`"),
167 let macro_decl_name = format!("{}{}{}", pre, info.callee.name(), post);
168 let def_site_span = info.callee.span;
170 // Don't print recursive invocations
171 if !info.call_site.source_equal(&prev_span) {
172 result.push(MacroBacktrace {
173 call_site: info.call_site,
174 macro_decl_name: macro_decl_name,
175 def_site_span: def_site_span,
180 self = info.call_site;
185 pub fn to(self, end: Span) -> Span {
186 // FIXME(jseyfried): self.ctxt should always equal end.ctxt here (c.f. issue #23480)
187 if end.ctxt == SyntaxContext::empty() {
188 Span { lo: self.lo, ..end }
190 Span { hi: end.hi, ..self }
194 pub fn between(self, end: Span) -> Span {
198 ctxt: if end.ctxt == SyntaxContext::empty() {
206 pub fn until(self, end: Span) -> Span {
210 ctxt: if end.ctxt == SyntaxContext::empty() {
219 #[derive(Clone, Debug)]
220 pub struct SpanLabel {
221 /// The span we are going to include in the final snippet.
224 /// Is this a primary span? This is the "locus" of the message,
225 /// and is indicated with a `^^^^` underline, versus `----`.
226 pub is_primary: bool,
228 /// What label should we attach to this span (if any)?
229 pub label: Option<String>,
232 impl serialize::UseSpecializedEncodable for Span {
233 fn default_encode<S: Encoder>(&self, s: &mut S) -> Result<(), S::Error> {
234 s.emit_struct("Span", 2, |s| {
235 s.emit_struct_field("lo", 0, |s| {
239 s.emit_struct_field("hi", 1, |s| {
246 impl serialize::UseSpecializedDecodable for Span {
247 fn default_decode<D: Decoder>(d: &mut D) -> Result<Span, D::Error> {
248 d.read_struct("Span", 2, |d| {
249 let lo = d.read_struct_field("lo", 0, Decodable::decode)?;
250 let hi = d.read_struct_field("hi", 1, Decodable::decode)?;
251 Ok(Span { lo: lo, hi: hi, ctxt: NO_EXPANSION })
256 fn default_span_debug(span: Span, f: &mut fmt::Formatter) -> fmt::Result {
257 write!(f, "Span {{ lo: {:?}, hi: {:?}, ctxt: {:?} }}",
258 span.lo, span.hi, span.ctxt)
261 impl fmt::Debug for Span {
262 fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
263 SPAN_DEBUG.with(|span_debug| span_debug.get()(*self, f))
267 pub const DUMMY_SP: Span = Span { lo: BytePos(0), hi: BytePos(0), ctxt: NO_EXPANSION };
270 pub fn new() -> MultiSpan {
272 primary_spans: vec![],
277 pub fn from_span(primary_span: Span) -> MultiSpan {
279 primary_spans: vec![primary_span],
284 pub fn from_spans(vec: Vec<Span>) -> MultiSpan {
291 pub fn push_span_label(&mut self, span: Span, label: String) {
292 self.span_labels.push((span, label));
295 /// Selects the first primary span (if any)
296 pub fn primary_span(&self) -> Option<Span> {
297 self.primary_spans.first().cloned()
300 /// Returns all primary spans.
301 pub fn primary_spans(&self) -> &[Span] {
305 /// Replaces all occurances of one Span with another. Used to move Spans in areas that don't
306 /// display well (like std macros). Returns true if replacements occurred.
307 pub fn replace(&mut self, before: Span, after: Span) -> bool {
308 let mut replacements_occurred = false;
309 for primary_span in &mut self.primary_spans {
310 if *primary_span == before {
311 *primary_span = after;
312 replacements_occurred = true;
315 for span_label in &mut self.span_labels {
316 if span_label.0 == before {
317 span_label.0 = after;
318 replacements_occurred = true;
321 replacements_occurred
324 /// Returns the strings to highlight. We always ensure that there
325 /// is an entry for each of the primary spans -- for each primary
326 /// span P, if there is at least one label with span P, we return
327 /// those labels (marked as primary). But otherwise we return
328 /// `SpanLabel` instances with empty labels.
329 pub fn span_labels(&self) -> Vec<SpanLabel> {
330 let is_primary = |span| self.primary_spans.contains(&span);
331 let mut span_labels = vec![];
333 for &(span, ref label) in &self.span_labels {
334 span_labels.push(SpanLabel {
336 is_primary: is_primary(span),
337 label: Some(label.clone())
341 for &span in &self.primary_spans {
342 if !span_labels.iter().any(|sl| sl.span == span) {
343 span_labels.push(SpanLabel {
355 impl From<Span> for MultiSpan {
356 fn from(span: Span) -> MultiSpan {
357 MultiSpan::from_span(span)
361 pub const NO_EXPANSION: SyntaxContext = SyntaxContext::empty();
363 /// Identifies an offset of a multi-byte character in a FileMap
364 #[derive(Copy, Clone, RustcEncodable, RustcDecodable, Eq, PartialEq)]
365 pub struct MultiByteChar {
366 /// The absolute offset of the character in the CodeMap
368 /// The number of bytes, >=2
372 /// A single source in the CodeMap.
375 /// The name of the file that the source came from, source that doesn't
376 /// originate from files has names between angle brackets by convention,
379 /// True if the `name` field above has been modified by -Zremap-path-prefix
380 pub name_was_remapped: bool,
381 /// Indicates which crate this FileMap was imported from.
382 pub crate_of_origin: u32,
383 /// The complete source code
384 pub src: Option<Rc<String>>,
385 /// The start position of this source in the CodeMap
386 pub start_pos: BytePos,
387 /// The end position of this source in the CodeMap
388 pub end_pos: BytePos,
389 /// Locations of lines beginnings in the source code
390 pub lines: RefCell<Vec<BytePos>>,
391 /// Locations of multi-byte characters in the source code
392 pub multibyte_chars: RefCell<Vec<MultiByteChar>>,
395 impl Encodable for FileMap {
396 fn encode<S: Encoder>(&self, s: &mut S) -> Result<(), S::Error> {
397 s.emit_struct("FileMap", 6, |s| {
398 s.emit_struct_field("name", 0, |s| self.name.encode(s))?;
399 s.emit_struct_field("name_was_remapped", 1, |s| self.name_was_remapped.encode(s))?;
400 s.emit_struct_field("start_pos", 2, |s| self.start_pos.encode(s))?;
401 s.emit_struct_field("end_pos", 3, |s| self.end_pos.encode(s))?;
402 s.emit_struct_field("lines", 4, |s| {
403 let lines = self.lines.borrow();
405 s.emit_u32(lines.len() as u32)?;
407 if !lines.is_empty() {
408 // In order to preserve some space, we exploit the fact that
409 // the lines list is sorted and individual lines are
410 // probably not that long. Because of that we can store lines
411 // as a difference list, using as little space as possible
412 // for the differences.
413 let max_line_length = if lines.len() == 1 {
417 .map(|w| w[1] - w[0])
418 .map(|bp| bp.to_usize())
423 let bytes_per_diff: u8 = match max_line_length {
425 0x100 ... 0xFFFF => 2,
429 // Encode the number of bytes used per diff.
430 bytes_per_diff.encode(s)?;
432 // Encode the first element.
435 let diff_iter = (&lines[..]).windows(2)
436 .map(|w| (w[1] - w[0]));
438 match bytes_per_diff {
439 1 => for diff in diff_iter { (diff.0 as u8).encode(s)? },
440 2 => for diff in diff_iter { (diff.0 as u16).encode(s)? },
441 4 => for diff in diff_iter { diff.0.encode(s)? },
448 s.emit_struct_field("multibyte_chars", 5, |s| {
449 (*self.multibyte_chars.borrow()).encode(s)
455 impl Decodable for FileMap {
456 fn decode<D: Decoder>(d: &mut D) -> Result<FileMap, D::Error> {
458 d.read_struct("FileMap", 6, |d| {
459 let name: String = d.read_struct_field("name", 0, |d| Decodable::decode(d))?;
460 let name_was_remapped: bool =
461 d.read_struct_field("name_was_remapped", 1, |d| Decodable::decode(d))?;
462 let start_pos: BytePos = d.read_struct_field("start_pos", 2, |d| Decodable::decode(d))?;
463 let end_pos: BytePos = d.read_struct_field("end_pos", 3, |d| Decodable::decode(d))?;
464 let lines: Vec<BytePos> = d.read_struct_field("lines", 4, |d| {
465 let num_lines: u32 = Decodable::decode(d)?;
466 let mut lines = Vec::with_capacity(num_lines as usize);
469 // Read the number of bytes used per diff.
470 let bytes_per_diff: u8 = Decodable::decode(d)?;
472 // Read the first element.
473 let mut line_start: BytePos = Decodable::decode(d)?;
474 lines.push(line_start);
476 for _ in 1..num_lines {
477 let diff = match bytes_per_diff {
478 1 => d.read_u8()? as u32,
479 2 => d.read_u16()? as u32,
484 line_start = line_start + BytePos(diff);
486 lines.push(line_start);
492 let multibyte_chars: Vec<MultiByteChar> =
493 d.read_struct_field("multibyte_chars", 5, |d| Decodable::decode(d))?;
496 name_was_remapped: name_was_remapped,
497 // `crate_of_origin` has to be set by the importer.
498 // This value matches up with rustc::hir::def_id::INVALID_CRATE.
499 // That constant is not available here unfortunately :(
500 crate_of_origin: ::std::u32::MAX - 1,
501 start_pos: start_pos,
504 lines: RefCell::new(lines),
505 multibyte_chars: RefCell::new(multibyte_chars)
511 impl fmt::Debug for FileMap {
512 fn fmt(&self, fmt: &mut fmt::Formatter) -> fmt::Result {
513 write!(fmt, "FileMap({})", self.name)
518 /// EFFECT: register a start-of-line offset in the
519 /// table of line-beginnings.
520 /// UNCHECKED INVARIANT: these offsets must be added in the right
521 /// order and must be in the right places; there is shared knowledge
522 /// about what ends a line between this file and parse.rs
523 /// WARNING: pos param here is the offset relative to start of CodeMap,
524 /// and CodeMap will append a newline when adding a filemap without a newline at the end,
525 /// so the safe way to call this is with value calculated as
526 /// filemap.start_pos + newline_offset_relative_to_the_start_of_filemap.
527 pub fn next_line(&self, pos: BytePos) {
528 // the new charpos must be > the last one (or it's the first one).
529 let mut lines = self.lines.borrow_mut();
530 let line_len = lines.len();
531 assert!(line_len == 0 || ((*lines)[line_len - 1] < pos));
535 /// get a line from the list of pre-computed line-beginnings.
536 /// line-number here is 0-based.
537 pub fn get_line(&self, line_number: usize) -> Option<&str> {
540 let lines = self.lines.borrow();
541 lines.get(line_number).map(|&line| {
542 let begin: BytePos = line - self.start_pos;
543 let begin = begin.to_usize();
544 // We can't use `lines.get(line_number+1)` because we might
545 // be parsing when we call this function and thus the current
546 // line is the last one we have line info for.
547 let slice = &src[begin..];
548 match slice.find('\n') {
549 Some(e) => &slice[..e],
558 pub fn record_multibyte_char(&self, pos: BytePos, bytes: usize) {
559 assert!(bytes >=2 && bytes <= 4);
560 let mbc = MultiByteChar {
564 self.multibyte_chars.borrow_mut().push(mbc);
567 pub fn is_real_file(&self) -> bool {
568 !(self.name.starts_with("<") &&
569 self.name.ends_with(">"))
572 pub fn is_imported(&self) -> bool {
576 pub fn byte_length(&self) -> u32 {
577 self.end_pos.0 - self.start_pos.0
579 pub fn count_lines(&self) -> usize {
580 self.lines.borrow().len()
583 /// Find the line containing the given position. The return value is the
584 /// index into the `lines` array of this FileMap, not the 1-based line
585 /// number. If the filemap is empty or the position is located before the
586 /// first line, None is returned.
587 pub fn lookup_line(&self, pos: BytePos) -> Option<usize> {
588 let lines = self.lines.borrow();
589 if lines.len() == 0 {
593 let line_index = lookup_line(&lines[..], pos);
594 assert!(line_index < lines.len() as isize);
596 Some(line_index as usize)
602 pub fn line_bounds(&self, line_index: usize) -> (BytePos, BytePos) {
603 if self.start_pos == self.end_pos {
604 return (self.start_pos, self.end_pos);
607 let lines = self.lines.borrow();
608 assert!(line_index < lines.len());
609 if line_index == (lines.len() - 1) {
610 (lines[line_index], self.end_pos)
612 (lines[line_index], lines[line_index + 1])
617 // _____________________________________________________________________________
618 // Pos, BytePos, CharPos
622 fn from_usize(n: usize) -> Self;
623 fn to_usize(&self) -> usize;
626 /// A byte offset. Keep this small (currently 32-bits), as AST contains
628 #[derive(Clone, Copy, PartialEq, Eq, Hash, PartialOrd, Ord, Debug)]
629 pub struct BytePos(pub u32);
631 /// A character offset. Because of multibyte utf8 characters, a byte offset
632 /// is not equivalent to a character offset. The CodeMap will convert BytePos
633 /// values to CharPos values as necessary.
634 #[derive(Copy, Clone, PartialEq, Eq, Hash, PartialOrd, Ord, Debug)]
635 pub struct CharPos(pub usize);
637 // FIXME: Lots of boilerplate in these impls, but so far my attempts to fix
638 // have been unsuccessful
640 impl Pos for BytePos {
641 fn from_usize(n: usize) -> BytePos { BytePos(n as u32) }
642 fn to_usize(&self) -> usize { let BytePos(n) = *self; n as usize }
645 impl Add for BytePos {
646 type Output = BytePos;
648 fn add(self, rhs: BytePos) -> BytePos {
649 BytePos((self.to_usize() + rhs.to_usize()) as u32)
653 impl Sub for BytePos {
654 type Output = BytePos;
656 fn sub(self, rhs: BytePos) -> BytePos {
657 BytePos((self.to_usize() - rhs.to_usize()) as u32)
661 impl Encodable for BytePos {
662 fn encode<S: Encoder>(&self, s: &mut S) -> Result<(), S::Error> {
667 impl Decodable for BytePos {
668 fn decode<D: Decoder>(d: &mut D) -> Result<BytePos, D::Error> {
669 Ok(BytePos(d.read_u32()?))
673 impl Pos for CharPos {
674 fn from_usize(n: usize) -> CharPos { CharPos(n) }
675 fn to_usize(&self) -> usize { let CharPos(n) = *self; n }
678 impl Add for CharPos {
679 type Output = CharPos;
681 fn add(self, rhs: CharPos) -> CharPos {
682 CharPos(self.to_usize() + rhs.to_usize())
686 impl Sub for CharPos {
687 type Output = CharPos;
689 fn sub(self, rhs: CharPos) -> CharPos {
690 CharPos(self.to_usize() - rhs.to_usize())
694 // _____________________________________________________________________________
695 // Loc, LocWithOpt, FileMapAndLine, FileMapAndBytePos
698 /// A source code location used for error reporting
699 #[derive(Debug, Clone)]
701 /// Information about the original source
702 pub file: Rc<FileMap>,
703 /// The (1-based) line number
705 /// The (0-based) column offset
709 /// A source code location used as the result of lookup_char_pos_adj
710 // Actually, *none* of the clients use the filename *or* file field;
711 // perhaps they should just be removed.
713 pub struct LocWithOpt {
714 pub filename: FileName,
717 pub file: Option<Rc<FileMap>>,
720 // used to be structural records. Better names, anyone?
722 pub struct FileMapAndLine { pub fm: Rc<FileMap>, pub line: usize }
724 pub struct FileMapAndBytePos { pub fm: Rc<FileMap>, pub pos: BytePos }
726 #[derive(Copy, Clone, Debug, PartialEq, Eq)]
727 pub struct LineInfo {
728 /// Index of line, starting from 0.
729 pub line_index: usize,
731 /// Column in line where span begins, starting from 0.
732 pub start_col: CharPos,
734 /// Column in line where span ends, starting from 0, exclusive.
735 pub end_col: CharPos,
738 pub struct FileLines {
739 pub file: Rc<FileMap>,
740 pub lines: Vec<LineInfo>
743 thread_local!(pub static SPAN_DEBUG: Cell<fn(Span, &mut fmt::Formatter) -> fmt::Result> =
744 Cell::new(default_span_debug));
746 pub struct MacroBacktrace {
747 /// span where macro was applied to generate this code
750 /// name of macro that was applied (e.g., "foo!" or "#[derive(Eq)]")
751 pub macro_decl_name: String,
753 /// span where macro was defined (if known)
754 pub def_site_span: Option<Span>,
757 // _____________________________________________________________________________
758 // SpanLinesError, SpanSnippetError, DistinctSources, MalformedCodemapPositions
761 pub type FileLinesResult = Result<FileLines, SpanLinesError>;
763 #[derive(Clone, PartialEq, Eq, Debug)]
764 pub enum SpanLinesError {
766 DistinctSources(DistinctSources),
769 #[derive(Clone, PartialEq, Eq, Debug)]
770 pub enum SpanSnippetError {
772 DistinctSources(DistinctSources),
773 MalformedForCodemap(MalformedCodemapPositions),
774 SourceNotAvailable { filename: String }
777 #[derive(Clone, PartialEq, Eq, Debug)]
778 pub struct DistinctSources {
779 pub begin: (String, BytePos),
780 pub end: (String, BytePos)
783 #[derive(Clone, PartialEq, Eq, Debug)]
784 pub struct MalformedCodemapPositions {
786 pub source_len: usize,
787 pub begin_pos: BytePos,
791 // Given a slice of line start positions and a position, returns the index of
792 // the line the position is on. Returns -1 if the position is located before
794 fn lookup_line(lines: &[BytePos], pos: BytePos) -> isize {
795 match lines.binary_search(&pos) {
796 Ok(line) => line as isize,
797 Err(line) => line as isize - 1
803 use super::{lookup_line, BytePos};
806 fn test_lookup_line() {
808 let lines = &[BytePos(3), BytePos(17), BytePos(28)];
810 assert_eq!(lookup_line(lines, BytePos(0)), -1);
811 assert_eq!(lookup_line(lines, BytePos(3)), 0);
812 assert_eq!(lookup_line(lines, BytePos(4)), 0);
814 assert_eq!(lookup_line(lines, BytePos(16)), 0);
815 assert_eq!(lookup_line(lines, BytePos(17)), 1);
816 assert_eq!(lookup_line(lines, BytePos(18)), 1);
818 assert_eq!(lookup_line(lines, BytePos(28)), 2);
819 assert_eq!(lookup_line(lines, BytePos(29)), 2);