]>
Commit | Line | Data |
---|---|---|
f20569fa XL |
1 | //! Source positions and related helper functions. |
2 | //! | |
3 | //! Important concepts in this module include: | |
4 | //! | |
5 | //! - the *span*, represented by [`SpanData`] and related types; | |
6 | //! - source code as represented by a [`SourceMap`]; and | |
7 | //! - interned strings, represented by [`Symbol`]s, with some common symbols available statically in the [`sym`] module. | |
8 | //! | |
9 | //! Unlike most compilers, the span contains not only the position in the source code, but also various other metadata, | |
10 | //! such as the edition and macro hygiene. This metadata is stored in [`SyntaxContext`] and [`ExpnData`]. | |
11 | //! | |
12 | //! ## Note | |
13 | //! | |
14 | //! This API is completely unstable and subject to change. | |
15 | ||
16 | #![doc(html_root_url = "https://doc.rust-lang.org/nightly/nightly-rustc/")] | |
17 | #![feature(array_windows)] | |
18 | #![feature(crate_visibility_modifier)] | |
19 | #![feature(const_fn)] | |
20 | #![feature(const_panic)] | |
21 | #![feature(negative_impls)] | |
22 | #![feature(nll)] | |
23 | #![feature(min_specialization)] | |
24 | #![feature(option_expect_none)] | |
25 | ||
26 | #[macro_use] | |
27 | extern crate rustc_macros; | |
28 | ||
29 | use rustc_data_structures::AtomicRef; | |
30 | use rustc_macros::HashStable_Generic; | |
31 | use rustc_serialize::{Decodable, Decoder, Encodable, Encoder}; | |
32 | ||
33 | mod caching_source_map_view; | |
34 | pub mod source_map; | |
35 | pub use self::caching_source_map_view::CachingSourceMapView; | |
36 | use source_map::SourceMap; | |
37 | ||
38 | pub mod edition; | |
39 | use edition::Edition; | |
40 | pub mod hygiene; | |
41 | pub use hygiene::SyntaxContext; | |
42 | use hygiene::Transparency; | |
43 | pub use hygiene::{DesugaringKind, ExpnData, ExpnId, ExpnKind, ForLoopLoc, MacroKind}; | |
44 | pub mod def_id; | |
45 | use def_id::{CrateNum, DefId, LOCAL_CRATE}; | |
46 | pub mod lev_distance; | |
47 | mod span_encoding; | |
48 | pub use span_encoding::{Span, DUMMY_SP}; | |
49 | ||
50 | pub mod symbol; | |
51 | pub use symbol::{sym, Symbol}; | |
52 | ||
53 | mod analyze_source_file; | |
54 | pub mod fatal_error; | |
55 | ||
56 | use rustc_data_structures::fingerprint::Fingerprint; | |
57 | use rustc_data_structures::stable_hasher::{HashStable, StableHasher}; | |
58 | use rustc_data_structures::sync::{Lock, Lrc}; | |
59 | ||
60 | use std::borrow::Cow; | |
61 | use std::cell::RefCell; | |
62 | use std::cmp::{self, Ordering}; | |
63 | use std::fmt; | |
64 | use std::hash::Hash; | |
65 | use std::ops::{Add, Range, Sub}; | |
66 | use std::path::{Path, PathBuf}; | |
67 | use std::str::FromStr; | |
68 | use std::thread::LocalKey; | |
69 | ||
70 | use md5::Md5; | |
71 | use sha1::Digest; | |
72 | use sha1::Sha1; | |
73 | use sha2::Sha256; | |
74 | ||
75 | use tracing::debug; | |
76 | ||
77 | #[cfg(test)] | |
78 | mod tests; | |
79 | ||
80 | // Per-session global variables: this struct is stored in thread-local storage | |
81 | // in such a way that it is accessible without any kind of handle to all | |
82 | // threads within the compilation session, but is not accessible outside the | |
83 | // session. | |
84 | pub struct SessionGlobals { | |
85 | symbol_interner: Lock<symbol::Interner>, | |
86 | span_interner: Lock<span_encoding::SpanInterner>, | |
87 | hygiene_data: Lock<hygiene::HygieneData>, | |
88 | source_map: Lock<Option<Lrc<SourceMap>>>, | |
89 | } | |
90 | ||
91 | impl SessionGlobals { | |
92 | pub fn new(edition: Edition) -> SessionGlobals { | |
93 | SessionGlobals { | |
94 | symbol_interner: Lock::new(symbol::Interner::fresh()), | |
95 | span_interner: Lock::new(span_encoding::SpanInterner::default()), | |
96 | hygiene_data: Lock::new(hygiene::HygieneData::new(edition)), | |
97 | source_map: Lock::new(None), | |
98 | } | |
99 | } | |
100 | } | |
101 | ||
102 | pub fn with_session_globals<R>(edition: Edition, f: impl FnOnce() -> R) -> R { | |
103 | let session_globals = SessionGlobals::new(edition); | |
104 | SESSION_GLOBALS.set(&session_globals, f) | |
105 | } | |
106 | ||
107 | pub fn with_default_session_globals<R>(f: impl FnOnce() -> R) -> R { | |
108 | with_session_globals(edition::DEFAULT_EDITION, f) | |
109 | } | |
110 | ||
111 | // If this ever becomes non thread-local, `decode_syntax_context` | |
112 | // and `decode_expn_id` will need to be updated to handle concurrent | |
113 | // deserialization. | |
114 | scoped_tls::scoped_thread_local!(pub static SESSION_GLOBALS: SessionGlobals); | |
115 | ||
116 | // FIXME: Perhaps this should not implement Rustc{Decodable, Encodable} | |
117 | // | |
118 | // FIXME: We should use this enum or something like it to get rid of the | |
119 | // use of magic `/rust/1.x/...` paths across the board. | |
120 | #[derive(Debug, Eq, PartialEq, Clone, Ord, PartialOrd, Hash)] | |
121 | #[derive(HashStable_Generic, Decodable, Encodable)] | |
122 | pub enum RealFileName { | |
123 | Named(PathBuf), | |
124 | /// For de-virtualized paths (namely paths into libstd that have been mapped | |
125 | /// to the appropriate spot on the local host's file system), | |
126 | Devirtualized { | |
127 | /// `local_path` is the (host-dependent) local path to the file. | |
128 | local_path: PathBuf, | |
129 | /// `virtual_name` is the stable path rustc will store internally within | |
130 | /// build artifacts. | |
131 | virtual_name: PathBuf, | |
132 | }, | |
133 | } | |
134 | ||
135 | impl RealFileName { | |
136 | /// Returns the path suitable for reading from the file system on the local host. | |
137 | /// Avoid embedding this in build artifacts; see `stable_name()` for that. | |
138 | pub fn local_path(&self) -> &Path { | |
139 | match self { | |
140 | RealFileName::Named(p) | |
141 | | RealFileName::Devirtualized { local_path: p, virtual_name: _ } => &p, | |
142 | } | |
143 | } | |
144 | ||
145 | /// Returns the path suitable for reading from the file system on the local host. | |
146 | /// Avoid embedding this in build artifacts; see `stable_name()` for that. | |
147 | pub fn into_local_path(self) -> PathBuf { | |
148 | match self { | |
149 | RealFileName::Named(p) | |
150 | | RealFileName::Devirtualized { local_path: p, virtual_name: _ } => p, | |
151 | } | |
152 | } | |
153 | ||
154 | /// Returns the path suitable for embedding into build artifacts. Note that | |
155 | /// a virtualized path will not correspond to a valid file system path; see | |
156 | /// `local_path()` for something that is more likely to return paths into the | |
157 | /// local host file system. | |
158 | pub fn stable_name(&self) -> &Path { | |
159 | match self { | |
160 | RealFileName::Named(p) | |
161 | | RealFileName::Devirtualized { local_path: _, virtual_name: p } => &p, | |
162 | } | |
163 | } | |
164 | } | |
165 | ||
166 | /// Differentiates between real files and common virtual files. | |
167 | #[derive(Debug, Eq, PartialEq, Clone, Ord, PartialOrd, Hash)] | |
168 | #[derive(HashStable_Generic, Decodable, Encodable)] | |
169 | pub enum FileName { | |
170 | Real(RealFileName), | |
171 | /// Call to `quote!`. | |
172 | QuoteExpansion(u64), | |
173 | /// Command line. | |
174 | Anon(u64), | |
175 | /// Hack in `src/librustc_ast/parse.rs`. | |
176 | // FIXME(jseyfried) | |
177 | MacroExpansion(u64), | |
178 | ProcMacroSourceCode(u64), | |
179 | /// Strings provided as `--cfg [cfgspec]` stored in a `crate_cfg`. | |
180 | CfgSpec(u64), | |
181 | /// Strings provided as crate attributes in the CLI. | |
182 | CliCrateAttr(u64), | |
183 | /// Custom sources for explicit parser calls from plugins and drivers. | |
184 | Custom(String), | |
185 | DocTest(PathBuf, isize), | |
186 | /// Post-substitution inline assembly from LLVM. | |
187 | InlineAsm(u64), | |
188 | } | |
189 | ||
190 | impl std::fmt::Display for FileName { | |
191 | fn fmt(&self, fmt: &mut std::fmt::Formatter<'_>) -> std::fmt::Result { | |
192 | use FileName::*; | |
193 | match *self { | |
194 | Real(RealFileName::Named(ref path)) => write!(fmt, "{}", path.display()), | |
195 | // FIXME: might be nice to display both components of Devirtualized. | |
196 | // But for now (to backport fix for issue #70924), best to not | |
197 | // perturb diagnostics so its obvious test suite still works. | |
198 | Real(RealFileName::Devirtualized { ref local_path, virtual_name: _ }) => { | |
199 | write!(fmt, "{}", local_path.display()) | |
200 | } | |
201 | QuoteExpansion(_) => write!(fmt, "<quote expansion>"), | |
202 | MacroExpansion(_) => write!(fmt, "<macro expansion>"), | |
203 | Anon(_) => write!(fmt, "<anon>"), | |
204 | ProcMacroSourceCode(_) => write!(fmt, "<proc-macro source code>"), | |
205 | CfgSpec(_) => write!(fmt, "<cfgspec>"), | |
206 | CliCrateAttr(_) => write!(fmt, "<crate attribute>"), | |
207 | Custom(ref s) => write!(fmt, "<{}>", s), | |
208 | DocTest(ref path, _) => write!(fmt, "{}", path.display()), | |
209 | InlineAsm(_) => write!(fmt, "<inline asm>"), | |
210 | } | |
211 | } | |
212 | } | |
213 | ||
214 | impl From<PathBuf> for FileName { | |
215 | fn from(p: PathBuf) -> Self { | |
216 | assert!(!p.to_string_lossy().ends_with('>')); | |
217 | FileName::Real(RealFileName::Named(p)) | |
218 | } | |
219 | } | |
220 | ||
221 | impl FileName { | |
222 | pub fn is_real(&self) -> bool { | |
223 | use FileName::*; | |
224 | match *self { | |
225 | Real(_) => true, | |
226 | Anon(_) | |
227 | | MacroExpansion(_) | |
228 | | ProcMacroSourceCode(_) | |
229 | | CfgSpec(_) | |
230 | | CliCrateAttr(_) | |
231 | | Custom(_) | |
232 | | QuoteExpansion(_) | |
233 | | DocTest(_, _) | |
234 | | InlineAsm(_) => false, | |
235 | } | |
236 | } | |
237 | ||
238 | pub fn macro_expansion_source_code(src: &str) -> FileName { | |
239 | let mut hasher = StableHasher::new(); | |
240 | src.hash(&mut hasher); | |
241 | FileName::MacroExpansion(hasher.finish()) | |
242 | } | |
243 | ||
244 | pub fn anon_source_code(src: &str) -> FileName { | |
245 | let mut hasher = StableHasher::new(); | |
246 | src.hash(&mut hasher); | |
247 | FileName::Anon(hasher.finish()) | |
248 | } | |
249 | ||
250 | pub fn proc_macro_source_code(src: &str) -> FileName { | |
251 | let mut hasher = StableHasher::new(); | |
252 | src.hash(&mut hasher); | |
253 | FileName::ProcMacroSourceCode(hasher.finish()) | |
254 | } | |
255 | ||
256 | pub fn cfg_spec_source_code(src: &str) -> FileName { | |
257 | let mut hasher = StableHasher::new(); | |
258 | src.hash(&mut hasher); | |
259 | FileName::QuoteExpansion(hasher.finish()) | |
260 | } | |
261 | ||
262 | pub fn cli_crate_attr_source_code(src: &str) -> FileName { | |
263 | let mut hasher = StableHasher::new(); | |
264 | src.hash(&mut hasher); | |
265 | FileName::CliCrateAttr(hasher.finish()) | |
266 | } | |
267 | ||
268 | pub fn doc_test_source_code(path: PathBuf, line: isize) -> FileName { | |
269 | FileName::DocTest(path, line) | |
270 | } | |
271 | ||
272 | pub fn inline_asm_source_code(src: &str) -> FileName { | |
273 | let mut hasher = StableHasher::new(); | |
274 | src.hash(&mut hasher); | |
275 | FileName::InlineAsm(hasher.finish()) | |
276 | } | |
277 | } | |
278 | ||
279 | /// Represents a span. | |
280 | /// | |
281 | /// Spans represent a region of code, used for error reporting. Positions in spans | |
282 | /// are *absolute* positions from the beginning of the [`SourceMap`], not positions | |
283 | /// relative to [`SourceFile`]s. Methods on the `SourceMap` can be used to relate spans back | |
284 | /// to the original source. | |
285 | /// | |
286 | /// You must be careful if the span crosses more than one file, since you will not be | |
287 | /// able to use many of the functions on spans in source_map and you cannot assume | |
288 | /// that the length of the span is equal to `span.hi - span.lo`; there may be space in the | |
289 | /// [`BytePos`] range between files. | |
290 | /// | |
291 | /// `SpanData` is public because `Span` uses a thread-local interner and can't be | |
292 | /// sent to other threads, but some pieces of performance infra run in a separate thread. | |
293 | /// Using `Span` is generally preferred. | |
294 | #[derive(Clone, Copy, Hash, PartialEq, Eq, Ord, PartialOrd)] | |
295 | pub struct SpanData { | |
296 | pub lo: BytePos, | |
297 | pub hi: BytePos, | |
298 | /// Information about where the macro came from, if this piece of | |
299 | /// code was created by a macro expansion. | |
300 | pub ctxt: SyntaxContext, | |
301 | } | |
302 | ||
303 | impl SpanData { | |
304 | #[inline] | |
305 | pub fn span(&self) -> Span { | |
306 | Span::new(self.lo, self.hi, self.ctxt) | |
307 | } | |
308 | #[inline] | |
309 | pub fn with_lo(&self, lo: BytePos) -> Span { | |
310 | Span::new(lo, self.hi, self.ctxt) | |
311 | } | |
312 | #[inline] | |
313 | pub fn with_hi(&self, hi: BytePos) -> Span { | |
314 | Span::new(self.lo, hi, self.ctxt) | |
315 | } | |
316 | #[inline] | |
317 | pub fn with_ctxt(&self, ctxt: SyntaxContext) -> Span { | |
318 | Span::new(self.lo, self.hi, ctxt) | |
319 | } | |
320 | } | |
321 | ||
322 | // The interner is pointed to by a thread local value which is only set on the main thread | |
323 | // with parallelization is disabled. So we don't allow `Span` to transfer between threads | |
324 | // to avoid panics and other errors, even though it would be memory safe to do so. | |
325 | #[cfg(not(parallel_compiler))] | |
326 | impl !Send for Span {} | |
327 | #[cfg(not(parallel_compiler))] | |
328 | impl !Sync for Span {} | |
329 | ||
330 | impl PartialOrd for Span { | |
331 | fn partial_cmp(&self, rhs: &Self) -> Option<Ordering> { | |
332 | PartialOrd::partial_cmp(&self.data(), &rhs.data()) | |
333 | } | |
334 | } | |
335 | impl Ord for Span { | |
336 | fn cmp(&self, rhs: &Self) -> Ordering { | |
337 | Ord::cmp(&self.data(), &rhs.data()) | |
338 | } | |
339 | } | |
340 | ||
341 | /// A collection of `Span`s. | |
342 | /// | |
343 | /// Spans have two orthogonal attributes: | |
344 | /// | |
345 | /// - They can be *primary spans*. In this case they are the locus of | |
346 | /// the error, and would be rendered with `^^^`. | |
347 | /// - They can have a *label*. In this case, the label is written next | |
348 | /// to the mark in the snippet when we render. | |
349 | #[derive(Clone, Debug, Hash, PartialEq, Eq, Encodable, Decodable)] | |
350 | pub struct MultiSpan { | |
351 | primary_spans: Vec<Span>, | |
352 | span_labels: Vec<(Span, String)>, | |
353 | } | |
354 | ||
355 | impl Span { | |
356 | #[inline] | |
357 | pub fn lo(self) -> BytePos { | |
358 | self.data().lo | |
359 | } | |
360 | #[inline] | |
361 | pub fn with_lo(self, lo: BytePos) -> Span { | |
362 | self.data().with_lo(lo) | |
363 | } | |
364 | #[inline] | |
365 | pub fn hi(self) -> BytePos { | |
366 | self.data().hi | |
367 | } | |
368 | #[inline] | |
369 | pub fn with_hi(self, hi: BytePos) -> Span { | |
370 | self.data().with_hi(hi) | |
371 | } | |
372 | #[inline] | |
373 | pub fn ctxt(self) -> SyntaxContext { | |
374 | self.data().ctxt | |
375 | } | |
376 | #[inline] | |
377 | pub fn with_ctxt(self, ctxt: SyntaxContext) -> Span { | |
378 | self.data().with_ctxt(ctxt) | |
379 | } | |
380 | ||
381 | /// Returns `true` if this is a dummy span with any hygienic context. | |
382 | #[inline] | |
383 | pub fn is_dummy(self) -> bool { | |
384 | let span = self.data(); | |
385 | span.lo.0 == 0 && span.hi.0 == 0 | |
386 | } | |
387 | ||
388 | /// Returns `true` if this span comes from a macro or desugaring. | |
389 | #[inline] | |
390 | pub fn from_expansion(self) -> bool { | |
391 | self.ctxt() != SyntaxContext::root() | |
392 | } | |
393 | ||
394 | /// Returns `true` if `span` originates in a derive-macro's expansion. | |
395 | pub fn in_derive_expansion(self) -> bool { | |
396 | matches!(self.ctxt().outer_expn_data().kind, ExpnKind::Macro(MacroKind::Derive, _)) | |
397 | } | |
398 | ||
399 | #[inline] | |
400 | pub fn with_root_ctxt(lo: BytePos, hi: BytePos) -> Span { | |
401 | Span::new(lo, hi, SyntaxContext::root()) | |
402 | } | |
403 | ||
404 | /// Returns a new span representing an empty span at the beginning of this span. | |
405 | #[inline] | |
406 | pub fn shrink_to_lo(self) -> Span { | |
407 | let span = self.data(); | |
408 | span.with_hi(span.lo) | |
409 | } | |
410 | /// Returns a new span representing an empty span at the end of this span. | |
411 | #[inline] | |
412 | pub fn shrink_to_hi(self) -> Span { | |
413 | let span = self.data(); | |
414 | span.with_lo(span.hi) | |
415 | } | |
416 | ||
417 | #[inline] | |
418 | /// Returns `true` if `hi == lo`. | |
419 | pub fn is_empty(&self) -> bool { | |
420 | let span = self.data(); | |
421 | span.hi == span.lo | |
422 | } | |
423 | ||
424 | /// Returns `self` if `self` is not the dummy span, and `other` otherwise. | |
425 | pub fn substitute_dummy(self, other: Span) -> Span { | |
426 | if self.is_dummy() { other } else { self } | |
427 | } | |
428 | ||
429 | /// Returns `true` if `self` fully encloses `other`. | |
430 | pub fn contains(self, other: Span) -> bool { | |
431 | let span = self.data(); | |
432 | let other = other.data(); | |
433 | span.lo <= other.lo && other.hi <= span.hi | |
434 | } | |
435 | ||
436 | /// Returns `true` if `self` touches `other`. | |
437 | pub fn overlaps(self, other: Span) -> bool { | |
438 | let span = self.data(); | |
439 | let other = other.data(); | |
440 | span.lo < other.hi && other.lo < span.hi | |
441 | } | |
442 | ||
443 | /// Returns `true` if the spans are equal with regards to the source text. | |
444 | /// | |
445 | /// Use this instead of `==` when either span could be generated code, | |
446 | /// and you only care that they point to the same bytes of source text. | |
447 | pub fn source_equal(&self, other: &Span) -> bool { | |
448 | let span = self.data(); | |
449 | let other = other.data(); | |
450 | span.lo == other.lo && span.hi == other.hi | |
451 | } | |
452 | ||
453 | /// Returns `Some(span)`, where the start is trimmed by the end of `other`. | |
454 | pub fn trim_start(self, other: Span) -> Option<Span> { | |
455 | let span = self.data(); | |
456 | let other = other.data(); | |
457 | if span.hi > other.hi { Some(span.with_lo(cmp::max(span.lo, other.hi))) } else { None } | |
458 | } | |
459 | ||
460 | /// Returns the source span -- this is either the supplied span, or the span for | |
461 | /// the macro callsite that expanded to it. | |
462 | pub fn source_callsite(self) -> Span { | |
463 | let expn_data = self.ctxt().outer_expn_data(); | |
464 | if !expn_data.is_root() { expn_data.call_site.source_callsite() } else { self } | |
465 | } | |
466 | ||
467 | /// The `Span` for the tokens in the previous macro expansion from which `self` was generated, | |
468 | /// if any. | |
469 | pub fn parent(self) -> Option<Span> { | |
470 | let expn_data = self.ctxt().outer_expn_data(); | |
471 | if !expn_data.is_root() { Some(expn_data.call_site) } else { None } | |
472 | } | |
473 | ||
474 | /// Edition of the crate from which this span came. | |
475 | pub fn edition(self) -> edition::Edition { | |
476 | self.ctxt().edition() | |
477 | } | |
478 | ||
479 | #[inline] | |
480 | pub fn rust_2015(&self) -> bool { | |
481 | self.edition() == edition::Edition::Edition2015 | |
482 | } | |
483 | ||
484 | #[inline] | |
485 | pub fn rust_2018(&self) -> bool { | |
486 | self.edition() >= edition::Edition::Edition2018 | |
487 | } | |
488 | ||
489 | #[inline] | |
490 | pub fn rust_2021(&self) -> bool { | |
491 | self.edition() >= edition::Edition::Edition2021 | |
492 | } | |
493 | ||
494 | /// Returns the source callee. | |
495 | /// | |
496 | /// Returns `None` if the supplied span has no expansion trace, | |
497 | /// else returns the `ExpnData` for the macro definition | |
498 | /// corresponding to the source callsite. | |
499 | pub fn source_callee(self) -> Option<ExpnData> { | |
500 | fn source_callee(expn_data: ExpnData) -> ExpnData { | |
501 | let next_expn_data = expn_data.call_site.ctxt().outer_expn_data(); | |
502 | if !next_expn_data.is_root() { source_callee(next_expn_data) } else { expn_data } | |
503 | } | |
504 | let expn_data = self.ctxt().outer_expn_data(); | |
505 | if !expn_data.is_root() { Some(source_callee(expn_data)) } else { None } | |
506 | } | |
507 | ||
508 | /// Checks if a span is "internal" to a macro in which `#[unstable]` | |
509 | /// items can be used (that is, a macro marked with | |
510 | /// `#[allow_internal_unstable]`). | |
511 | pub fn allows_unstable(&self, feature: Symbol) -> bool { | |
512 | self.ctxt().outer_expn_data().allow_internal_unstable.map_or(false, |features| { | |
513 | features | |
514 | .iter() | |
515 | .any(|&f| f == feature || f == sym::allow_internal_unstable_backcompat_hack) | |
516 | }) | |
517 | } | |
518 | ||
519 | /// Checks if this span arises from a compiler desugaring of kind `kind`. | |
520 | pub fn is_desugaring(&self, kind: DesugaringKind) -> bool { | |
521 | match self.ctxt().outer_expn_data().kind { | |
522 | ExpnKind::Desugaring(k) => k == kind, | |
523 | _ => false, | |
524 | } | |
525 | } | |
526 | ||
527 | /// Returns the compiler desugaring that created this span, or `None` | |
528 | /// if this span is not from a desugaring. | |
529 | pub fn desugaring_kind(&self) -> Option<DesugaringKind> { | |
530 | match self.ctxt().outer_expn_data().kind { | |
531 | ExpnKind::Desugaring(k) => Some(k), | |
532 | _ => None, | |
533 | } | |
534 | } | |
535 | ||
536 | /// Checks if a span is "internal" to a macro in which `unsafe` | |
537 | /// can be used without triggering the `unsafe_code` lint. | |
538 | // (that is, a macro marked with `#[allow_internal_unsafe]`). | |
539 | pub fn allows_unsafe(&self) -> bool { | |
540 | self.ctxt().outer_expn_data().allow_internal_unsafe | |
541 | } | |
542 | ||
543 | pub fn macro_backtrace(mut self) -> impl Iterator<Item = ExpnData> { | |
544 | let mut prev_span = DUMMY_SP; | |
545 | std::iter::from_fn(move || { | |
546 | loop { | |
547 | let expn_data = self.ctxt().outer_expn_data(); | |
548 | if expn_data.is_root() { | |
549 | return None; | |
550 | } | |
551 | ||
552 | let is_recursive = expn_data.call_site.source_equal(&prev_span); | |
553 | ||
554 | prev_span = self; | |
555 | self = expn_data.call_site; | |
556 | ||
557 | // Don't print recursive invocations. | |
558 | if !is_recursive { | |
559 | return Some(expn_data); | |
560 | } | |
561 | } | |
562 | }) | |
563 | } | |
564 | ||
565 | /// Returns a `Span` that would enclose both `self` and `end`. | |
566 | /// | |
567 | /// ```text | |
568 | /// ____ ___ | |
569 | /// self lorem ipsum end | |
570 | /// ^^^^^^^^^^^^^^^^^^^^ | |
571 | /// ``` | |
572 | pub fn to(self, end: Span) -> Span { | |
573 | let span_data = self.data(); | |
574 | let end_data = end.data(); | |
575 | // FIXME(jseyfried): `self.ctxt` should always equal `end.ctxt` here (cf. issue #23480). | |
576 | // Return the macro span on its own to avoid weird diagnostic output. It is preferable to | |
577 | // have an incomplete span than a completely nonsensical one. | |
578 | if span_data.ctxt != end_data.ctxt { | |
579 | if span_data.ctxt == SyntaxContext::root() { | |
580 | return end; | |
581 | } else if end_data.ctxt == SyntaxContext::root() { | |
582 | return self; | |
583 | } | |
584 | // Both spans fall within a macro. | |
585 | // FIXME(estebank): check if it is the *same* macro. | |
586 | } | |
587 | Span::new( | |
588 | cmp::min(span_data.lo, end_data.lo), | |
589 | cmp::max(span_data.hi, end_data.hi), | |
590 | if span_data.ctxt == SyntaxContext::root() { end_data.ctxt } else { span_data.ctxt }, | |
591 | ) | |
592 | } | |
593 | ||
594 | /// Returns a `Span` between the end of `self` to the beginning of `end`. | |
595 | /// | |
596 | /// ```text | |
597 | /// ____ ___ | |
598 | /// self lorem ipsum end | |
599 | /// ^^^^^^^^^^^^^ | |
600 | /// ``` | |
601 | pub fn between(self, end: Span) -> Span { | |
602 | let span = self.data(); | |
603 | let end = end.data(); | |
604 | Span::new( | |
605 | span.hi, | |
606 | end.lo, | |
607 | if end.ctxt == SyntaxContext::root() { end.ctxt } else { span.ctxt }, | |
608 | ) | |
609 | } | |
610 | ||
611 | /// Returns a `Span` from the beginning of `self` until the beginning of `end`. | |
612 | /// | |
613 | /// ```text | |
614 | /// ____ ___ | |
615 | /// self lorem ipsum end | |
616 | /// ^^^^^^^^^^^^^^^^^ | |
617 | /// ``` | |
618 | pub fn until(self, end: Span) -> Span { | |
619 | let span = self.data(); | |
620 | let end = end.data(); | |
621 | Span::new( | |
622 | span.lo, | |
623 | end.lo, | |
624 | if end.ctxt == SyntaxContext::root() { end.ctxt } else { span.ctxt }, | |
625 | ) | |
626 | } | |
627 | ||
628 | pub fn from_inner(self, inner: InnerSpan) -> Span { | |
629 | let span = self.data(); | |
630 | Span::new( | |
631 | span.lo + BytePos::from_usize(inner.start), | |
632 | span.lo + BytePos::from_usize(inner.end), | |
633 | span.ctxt, | |
634 | ) | |
635 | } | |
636 | ||
637 | /// Equivalent of `Span::def_site` from the proc macro API, | |
638 | /// except that the location is taken from the `self` span. | |
639 | pub fn with_def_site_ctxt(self, expn_id: ExpnId) -> Span { | |
640 | self.with_ctxt_from_mark(expn_id, Transparency::Opaque) | |
641 | } | |
642 | ||
643 | /// Equivalent of `Span::call_site` from the proc macro API, | |
644 | /// except that the location is taken from the `self` span. | |
645 | pub fn with_call_site_ctxt(&self, expn_id: ExpnId) -> Span { | |
646 | self.with_ctxt_from_mark(expn_id, Transparency::Transparent) | |
647 | } | |
648 | ||
649 | /// Equivalent of `Span::mixed_site` from the proc macro API, | |
650 | /// except that the location is taken from the `self` span. | |
651 | pub fn with_mixed_site_ctxt(&self, expn_id: ExpnId) -> Span { | |
652 | self.with_ctxt_from_mark(expn_id, Transparency::SemiTransparent) | |
653 | } | |
654 | ||
655 | /// Produces a span with the same location as `self` and context produced by a macro with the | |
656 | /// given ID and transparency, assuming that macro was defined directly and not produced by | |
657 | /// some other macro (which is the case for built-in and procedural macros). | |
658 | pub fn with_ctxt_from_mark(self, expn_id: ExpnId, transparency: Transparency) -> Span { | |
659 | self.with_ctxt(SyntaxContext::root().apply_mark(expn_id, transparency)) | |
660 | } | |
661 | ||
662 | #[inline] | |
663 | pub fn apply_mark(self, expn_id: ExpnId, transparency: Transparency) -> Span { | |
664 | let span = self.data(); | |
665 | span.with_ctxt(span.ctxt.apply_mark(expn_id, transparency)) | |
666 | } | |
667 | ||
668 | #[inline] | |
669 | pub fn remove_mark(&mut self) -> ExpnId { | |
670 | let mut span = self.data(); | |
671 | let mark = span.ctxt.remove_mark(); | |
672 | *self = Span::new(span.lo, span.hi, span.ctxt); | |
673 | mark | |
674 | } | |
675 | ||
676 | #[inline] | |
677 | pub fn adjust(&mut self, expn_id: ExpnId) -> Option<ExpnId> { | |
678 | let mut span = self.data(); | |
679 | let mark = span.ctxt.adjust(expn_id); | |
680 | *self = Span::new(span.lo, span.hi, span.ctxt); | |
681 | mark | |
682 | } | |
683 | ||
684 | #[inline] | |
685 | pub fn normalize_to_macros_2_0_and_adjust(&mut self, expn_id: ExpnId) -> Option<ExpnId> { | |
686 | let mut span = self.data(); | |
687 | let mark = span.ctxt.normalize_to_macros_2_0_and_adjust(expn_id); | |
688 | *self = Span::new(span.lo, span.hi, span.ctxt); | |
689 | mark | |
690 | } | |
691 | ||
692 | #[inline] | |
693 | pub fn glob_adjust(&mut self, expn_id: ExpnId, glob_span: Span) -> Option<Option<ExpnId>> { | |
694 | let mut span = self.data(); | |
695 | let mark = span.ctxt.glob_adjust(expn_id, glob_span); | |
696 | *self = Span::new(span.lo, span.hi, span.ctxt); | |
697 | mark | |
698 | } | |
699 | ||
700 | #[inline] | |
701 | pub fn reverse_glob_adjust( | |
702 | &mut self, | |
703 | expn_id: ExpnId, | |
704 | glob_span: Span, | |
705 | ) -> Option<Option<ExpnId>> { | |
706 | let mut span = self.data(); | |
707 | let mark = span.ctxt.reverse_glob_adjust(expn_id, glob_span); | |
708 | *self = Span::new(span.lo, span.hi, span.ctxt); | |
709 | mark | |
710 | } | |
711 | ||
712 | #[inline] | |
713 | pub fn normalize_to_macros_2_0(self) -> Span { | |
714 | let span = self.data(); | |
715 | span.with_ctxt(span.ctxt.normalize_to_macros_2_0()) | |
716 | } | |
717 | ||
718 | #[inline] | |
719 | pub fn normalize_to_macro_rules(self) -> Span { | |
720 | let span = self.data(); | |
721 | span.with_ctxt(span.ctxt.normalize_to_macro_rules()) | |
722 | } | |
723 | } | |
724 | ||
725 | /// A span together with some additional data. | |
726 | #[derive(Clone, Debug)] | |
727 | pub struct SpanLabel { | |
728 | /// The span we are going to include in the final snippet. | |
729 | pub span: Span, | |
730 | ||
731 | /// Is this a primary span? This is the "locus" of the message, | |
732 | /// and is indicated with a `^^^^` underline, versus `----`. | |
733 | pub is_primary: bool, | |
734 | ||
735 | /// What label should we attach to this span (if any)? | |
736 | pub label: Option<String>, | |
737 | } | |
738 | ||
739 | impl Default for Span { | |
740 | fn default() -> Self { | |
741 | DUMMY_SP | |
742 | } | |
743 | } | |
744 | ||
745 | impl<E: Encoder> Encodable<E> for Span { | |
746 | default fn encode(&self, s: &mut E) -> Result<(), E::Error> { | |
747 | let span = self.data(); | |
748 | s.emit_struct("Span", 2, |s| { | |
749 | s.emit_struct_field("lo", 0, |s| span.lo.encode(s))?; | |
750 | s.emit_struct_field("hi", 1, |s| span.hi.encode(s)) | |
751 | }) | |
752 | } | |
753 | } | |
754 | impl<D: Decoder> Decodable<D> for Span { | |
755 | default fn decode(s: &mut D) -> Result<Span, D::Error> { | |
756 | s.read_struct("Span", 2, |d| { | |
757 | let lo = d.read_struct_field("lo", 0, Decodable::decode)?; | |
758 | let hi = d.read_struct_field("hi", 1, Decodable::decode)?; | |
759 | ||
760 | Ok(Span::new(lo, hi, SyntaxContext::root())) | |
761 | }) | |
762 | } | |
763 | } | |
764 | ||
765 | /// Calls the provided closure, using the provided `SourceMap` to format | |
766 | /// any spans that are debug-printed during the closure's execution. | |
767 | /// | |
768 | /// Normally, the global `TyCtxt` is used to retrieve the `SourceMap` | |
769 | /// (see `rustc_interface::callbacks::span_debug1`). However, some parts | |
770 | /// of the compiler (e.g. `rustc_parse`) may debug-print `Span`s before | |
771 | /// a `TyCtxt` is available. In this case, we fall back to | |
772 | /// the `SourceMap` provided to this function. If that is not available, | |
773 | /// we fall back to printing the raw `Span` field values. | |
774 | pub fn with_source_map<T, F: FnOnce() -> T>(source_map: Lrc<SourceMap>, f: F) -> T { | |
775 | SESSION_GLOBALS.with(|session_globals| { | |
776 | *session_globals.source_map.borrow_mut() = Some(source_map); | |
777 | }); | |
778 | struct ClearSourceMap; | |
779 | impl Drop for ClearSourceMap { | |
780 | fn drop(&mut self) { | |
781 | SESSION_GLOBALS.with(|session_globals| { | |
782 | session_globals.source_map.borrow_mut().take(); | |
783 | }); | |
784 | } | |
785 | } | |
786 | ||
787 | let _guard = ClearSourceMap; | |
788 | f() | |
789 | } | |
790 | ||
791 | pub fn debug_with_source_map( | |
792 | span: Span, | |
793 | f: &mut fmt::Formatter<'_>, | |
794 | source_map: &SourceMap, | |
795 | ) -> fmt::Result { | |
796 | write!(f, "{} ({:?})", source_map.span_to_string(span), span.ctxt()) | |
797 | } | |
798 | ||
799 | pub fn default_span_debug(span: Span, f: &mut fmt::Formatter<'_>) -> fmt::Result { | |
800 | SESSION_GLOBALS.with(|session_globals| { | |
801 | if let Some(source_map) = &*session_globals.source_map.borrow() { | |
802 | debug_with_source_map(span, f, source_map) | |
803 | } else { | |
804 | f.debug_struct("Span") | |
805 | .field("lo", &span.lo()) | |
806 | .field("hi", &span.hi()) | |
807 | .field("ctxt", &span.ctxt()) | |
808 | .finish() | |
809 | } | |
810 | }) | |
811 | } | |
812 | ||
813 | impl fmt::Debug for Span { | |
814 | fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { | |
815 | (*SPAN_DEBUG)(*self, f) | |
816 | } | |
817 | } | |
818 | ||
819 | impl fmt::Debug for SpanData { | |
820 | fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { | |
821 | (*SPAN_DEBUG)(Span::new(self.lo, self.hi, self.ctxt), f) | |
822 | } | |
823 | } | |
824 | ||
825 | impl MultiSpan { | |
826 | #[inline] | |
827 | pub fn new() -> MultiSpan { | |
828 | MultiSpan { primary_spans: vec![], span_labels: vec![] } | |
829 | } | |
830 | ||
831 | pub fn from_span(primary_span: Span) -> MultiSpan { | |
832 | MultiSpan { primary_spans: vec![primary_span], span_labels: vec![] } | |
833 | } | |
834 | ||
835 | pub fn from_spans(mut vec: Vec<Span>) -> MultiSpan { | |
836 | vec.sort(); | |
837 | MultiSpan { primary_spans: vec, span_labels: vec![] } | |
838 | } | |
839 | ||
840 | pub fn push_span_label(&mut self, span: Span, label: String) { | |
841 | self.span_labels.push((span, label)); | |
842 | } | |
843 | ||
844 | /// Selects the first primary span (if any). | |
845 | pub fn primary_span(&self) -> Option<Span> { | |
846 | self.primary_spans.first().cloned() | |
847 | } | |
848 | ||
849 | /// Returns all primary spans. | |
850 | pub fn primary_spans(&self) -> &[Span] { | |
851 | &self.primary_spans | |
852 | } | |
853 | ||
854 | /// Returns `true` if any of the primary spans are displayable. | |
855 | pub fn has_primary_spans(&self) -> bool { | |
856 | self.primary_spans.iter().any(|sp| !sp.is_dummy()) | |
857 | } | |
858 | ||
859 | /// Returns `true` if this contains only a dummy primary span with any hygienic context. | |
860 | pub fn is_dummy(&self) -> bool { | |
861 | let mut is_dummy = true; | |
862 | for span in &self.primary_spans { | |
863 | if !span.is_dummy() { | |
864 | is_dummy = false; | |
865 | } | |
866 | } | |
867 | is_dummy | |
868 | } | |
869 | ||
870 | /// Replaces all occurrences of one Span with another. Used to move `Span`s in areas that don't | |
871 | /// display well (like std macros). Returns whether replacements occurred. | |
872 | pub fn replace(&mut self, before: Span, after: Span) -> bool { | |
873 | let mut replacements_occurred = false; | |
874 | for primary_span in &mut self.primary_spans { | |
875 | if *primary_span == before { | |
876 | *primary_span = after; | |
877 | replacements_occurred = true; | |
878 | } | |
879 | } | |
880 | for span_label in &mut self.span_labels { | |
881 | if span_label.0 == before { | |
882 | span_label.0 = after; | |
883 | replacements_occurred = true; | |
884 | } | |
885 | } | |
886 | replacements_occurred | |
887 | } | |
888 | ||
889 | /// Returns the strings to highlight. We always ensure that there | |
890 | /// is an entry for each of the primary spans -- for each primary | |
891 | /// span `P`, if there is at least one label with span `P`, we return | |
892 | /// those labels (marked as primary). But otherwise we return | |
893 | /// `SpanLabel` instances with empty labels. | |
894 | pub fn span_labels(&self) -> Vec<SpanLabel> { | |
895 | let is_primary = |span| self.primary_spans.contains(&span); | |
896 | ||
897 | let mut span_labels = self | |
898 | .span_labels | |
899 | .iter() | |
900 | .map(|&(span, ref label)| SpanLabel { | |
901 | span, | |
902 | is_primary: is_primary(span), | |
903 | label: Some(label.clone()), | |
904 | }) | |
905 | .collect::<Vec<_>>(); | |
906 | ||
907 | for &span in &self.primary_spans { | |
908 | if !span_labels.iter().any(|sl| sl.span == span) { | |
909 | span_labels.push(SpanLabel { span, is_primary: true, label: None }); | |
910 | } | |
911 | } | |
912 | ||
913 | span_labels | |
914 | } | |
915 | ||
916 | /// Returns `true` if any of the span labels is displayable. | |
917 | pub fn has_span_labels(&self) -> bool { | |
918 | self.span_labels.iter().any(|(sp, _)| !sp.is_dummy()) | |
919 | } | |
920 | } | |
921 | ||
922 | impl From<Span> for MultiSpan { | |
923 | fn from(span: Span) -> MultiSpan { | |
924 | MultiSpan::from_span(span) | |
925 | } | |
926 | } | |
927 | ||
928 | impl From<Vec<Span>> for MultiSpan { | |
929 | fn from(spans: Vec<Span>) -> MultiSpan { | |
930 | MultiSpan::from_spans(spans) | |
931 | } | |
932 | } | |
933 | ||
934 | /// Identifies an offset of a multi-byte character in a `SourceFile`. | |
935 | #[derive(Copy, Clone, Encodable, Decodable, Eq, PartialEq, Debug)] | |
936 | pub struct MultiByteChar { | |
937 | /// The absolute offset of the character in the `SourceMap`. | |
938 | pub pos: BytePos, | |
939 | /// The number of bytes, `>= 2`. | |
940 | pub bytes: u8, | |
941 | } | |
942 | ||
943 | /// Identifies an offset of a non-narrow character in a `SourceFile`. | |
944 | #[derive(Copy, Clone, Encodable, Decodable, Eq, PartialEq, Debug)] | |
945 | pub enum NonNarrowChar { | |
946 | /// Represents a zero-width character. | |
947 | ZeroWidth(BytePos), | |
948 | /// Represents a wide (full-width) character. | |
949 | Wide(BytePos), | |
950 | /// Represents a tab character, represented visually with a width of 4 characters. | |
951 | Tab(BytePos), | |
952 | } | |
953 | ||
954 | impl NonNarrowChar { | |
955 | fn new(pos: BytePos, width: usize) -> Self { | |
956 | match width { | |
957 | 0 => NonNarrowChar::ZeroWidth(pos), | |
958 | 2 => NonNarrowChar::Wide(pos), | |
959 | 4 => NonNarrowChar::Tab(pos), | |
960 | _ => panic!("width {} given for non-narrow character", width), | |
961 | } | |
962 | } | |
963 | ||
964 | /// Returns the absolute offset of the character in the `SourceMap`. | |
965 | pub fn pos(&self) -> BytePos { | |
966 | match *self { | |
967 | NonNarrowChar::ZeroWidth(p) | NonNarrowChar::Wide(p) | NonNarrowChar::Tab(p) => p, | |
968 | } | |
969 | } | |
970 | ||
971 | /// Returns the width of the character, 0 (zero-width) or 2 (wide). | |
972 | pub fn width(&self) -> usize { | |
973 | match *self { | |
974 | NonNarrowChar::ZeroWidth(_) => 0, | |
975 | NonNarrowChar::Wide(_) => 2, | |
976 | NonNarrowChar::Tab(_) => 4, | |
977 | } | |
978 | } | |
979 | } | |
980 | ||
981 | impl Add<BytePos> for NonNarrowChar { | |
982 | type Output = Self; | |
983 | ||
984 | fn add(self, rhs: BytePos) -> Self { | |
985 | match self { | |
986 | NonNarrowChar::ZeroWidth(pos) => NonNarrowChar::ZeroWidth(pos + rhs), | |
987 | NonNarrowChar::Wide(pos) => NonNarrowChar::Wide(pos + rhs), | |
988 | NonNarrowChar::Tab(pos) => NonNarrowChar::Tab(pos + rhs), | |
989 | } | |
990 | } | |
991 | } | |
992 | ||
993 | impl Sub<BytePos> for NonNarrowChar { | |
994 | type Output = Self; | |
995 | ||
996 | fn sub(self, rhs: BytePos) -> Self { | |
997 | match self { | |
998 | NonNarrowChar::ZeroWidth(pos) => NonNarrowChar::ZeroWidth(pos - rhs), | |
999 | NonNarrowChar::Wide(pos) => NonNarrowChar::Wide(pos - rhs), | |
1000 | NonNarrowChar::Tab(pos) => NonNarrowChar::Tab(pos - rhs), | |
1001 | } | |
1002 | } | |
1003 | } | |
1004 | ||
1005 | /// Identifies an offset of a character that was normalized away from `SourceFile`. | |
1006 | #[derive(Copy, Clone, Encodable, Decodable, Eq, PartialEq, Debug)] | |
1007 | pub struct NormalizedPos { | |
1008 | /// The absolute offset of the character in the `SourceMap`. | |
1009 | pub pos: BytePos, | |
1010 | /// The difference between original and normalized string at position. | |
1011 | pub diff: u32, | |
1012 | } | |
1013 | ||
1014 | #[derive(PartialEq, Eq, Clone, Debug)] | |
1015 | pub enum ExternalSource { | |
1016 | /// No external source has to be loaded, since the `SourceFile` represents a local crate. | |
1017 | Unneeded, | |
1018 | Foreign { | |
1019 | kind: ExternalSourceKind, | |
1020 | /// This SourceFile's byte-offset within the source_map of its original crate. | |
1021 | original_start_pos: BytePos, | |
1022 | /// The end of this SourceFile within the source_map of its original crate. | |
1023 | original_end_pos: BytePos, | |
1024 | }, | |
1025 | } | |
1026 | ||
1027 | /// The state of the lazy external source loading mechanism of a `SourceFile`. | |
1028 | #[derive(PartialEq, Eq, Clone, Debug)] | |
1029 | pub enum ExternalSourceKind { | |
1030 | /// The external source has been loaded already. | |
1031 | Present(Lrc<String>), | |
1032 | /// No attempt has been made to load the external source. | |
1033 | AbsentOk, | |
1034 | /// A failed attempt has been made to load the external source. | |
1035 | AbsentErr, | |
1036 | Unneeded, | |
1037 | } | |
1038 | ||
1039 | impl ExternalSource { | |
1040 | pub fn is_absent(&self) -> bool { | |
1041 | !matches!(self, ExternalSource::Foreign { kind: ExternalSourceKind::Present(_), .. }) | |
1042 | } | |
1043 | ||
1044 | pub fn get_source(&self) -> Option<&Lrc<String>> { | |
1045 | match self { | |
1046 | ExternalSource::Foreign { kind: ExternalSourceKind::Present(ref src), .. } => Some(src), | |
1047 | _ => None, | |
1048 | } | |
1049 | } | |
1050 | } | |
1051 | ||
1052 | #[derive(Debug)] | |
1053 | pub struct OffsetOverflowError; | |
1054 | ||
1055 | #[derive(Copy, Clone, Debug, PartialEq, Eq, PartialOrd, Ord, Hash, Encodable, Decodable)] | |
1056 | pub enum SourceFileHashAlgorithm { | |
1057 | Md5, | |
1058 | Sha1, | |
1059 | Sha256, | |
1060 | } | |
1061 | ||
1062 | impl FromStr for SourceFileHashAlgorithm { | |
1063 | type Err = (); | |
1064 | ||
1065 | fn from_str(s: &str) -> Result<SourceFileHashAlgorithm, ()> { | |
1066 | match s { | |
1067 | "md5" => Ok(SourceFileHashAlgorithm::Md5), | |
1068 | "sha1" => Ok(SourceFileHashAlgorithm::Sha1), | |
1069 | "sha256" => Ok(SourceFileHashAlgorithm::Sha256), | |
1070 | _ => Err(()), | |
1071 | } | |
1072 | } | |
1073 | } | |
1074 | ||
1075 | rustc_data_structures::impl_stable_hash_via_hash!(SourceFileHashAlgorithm); | |
1076 | ||
1077 | /// The hash of the on-disk source file used for debug info. | |
1078 | #[derive(Copy, Clone, PartialEq, Eq, Debug)] | |
1079 | #[derive(HashStable_Generic, Encodable, Decodable)] | |
1080 | pub struct SourceFileHash { | |
1081 | pub kind: SourceFileHashAlgorithm, | |
1082 | value: [u8; 32], | |
1083 | } | |
1084 | ||
1085 | impl SourceFileHash { | |
1086 | pub fn new(kind: SourceFileHashAlgorithm, src: &str) -> SourceFileHash { | |
1087 | let mut hash = SourceFileHash { kind, value: Default::default() }; | |
1088 | let len = hash.hash_len(); | |
1089 | let value = &mut hash.value[..len]; | |
1090 | let data = src.as_bytes(); | |
1091 | match kind { | |
1092 | SourceFileHashAlgorithm::Md5 => { | |
1093 | value.copy_from_slice(&Md5::digest(data)); | |
1094 | } | |
1095 | SourceFileHashAlgorithm::Sha1 => { | |
1096 | value.copy_from_slice(&Sha1::digest(data)); | |
1097 | } | |
1098 | SourceFileHashAlgorithm::Sha256 => { | |
1099 | value.copy_from_slice(&Sha256::digest(data)); | |
1100 | } | |
1101 | } | |
1102 | hash | |
1103 | } | |
1104 | ||
1105 | /// Check if the stored hash matches the hash of the string. | |
1106 | pub fn matches(&self, src: &str) -> bool { | |
1107 | Self::new(self.kind, src) == *self | |
1108 | } | |
1109 | ||
1110 | /// The bytes of the hash. | |
1111 | pub fn hash_bytes(&self) -> &[u8] { | |
1112 | let len = self.hash_len(); | |
1113 | &self.value[..len] | |
1114 | } | |
1115 | ||
1116 | fn hash_len(&self) -> usize { | |
1117 | match self.kind { | |
1118 | SourceFileHashAlgorithm::Md5 => 16, | |
1119 | SourceFileHashAlgorithm::Sha1 => 20, | |
1120 | SourceFileHashAlgorithm::Sha256 => 32, | |
1121 | } | |
1122 | } | |
1123 | } | |
1124 | ||
1125 | /// A single source in the [`SourceMap`]. | |
1126 | #[derive(Clone)] | |
1127 | pub struct SourceFile { | |
1128 | /// The name of the file that the source came from. Source that doesn't | |
1129 | /// originate from files has names between angle brackets by convention | |
1130 | /// (e.g., `<anon>`). | |
1131 | pub name: FileName, | |
1132 | /// `true` if the `name` field above has been modified by `--remap-path-prefix`. | |
1133 | pub name_was_remapped: bool, | |
1134 | /// The unmapped path of the file that the source came from. | |
1135 | /// Set to `None` if the `SourceFile` was imported from an external crate. | |
1136 | pub unmapped_path: Option<FileName>, | |
1137 | /// The complete source code. | |
1138 | pub src: Option<Lrc<String>>, | |
1139 | /// The source code's hash. | |
1140 | pub src_hash: SourceFileHash, | |
1141 | /// The external source code (used for external crates, which will have a `None` | |
1142 | /// value as `self.src`. | |
1143 | pub external_src: Lock<ExternalSource>, | |
1144 | /// The start position of this source in the `SourceMap`. | |
1145 | pub start_pos: BytePos, | |
1146 | /// The end position of this source in the `SourceMap`. | |
1147 | pub end_pos: BytePos, | |
1148 | /// Locations of lines beginnings in the source code. | |
1149 | pub lines: Vec<BytePos>, | |
1150 | /// Locations of multi-byte characters in the source code. | |
1151 | pub multibyte_chars: Vec<MultiByteChar>, | |
1152 | /// Width of characters that are not narrow in the source code. | |
1153 | pub non_narrow_chars: Vec<NonNarrowChar>, | |
1154 | /// Locations of characters removed during normalization. | |
1155 | pub normalized_pos: Vec<NormalizedPos>, | |
1156 | /// A hash of the filename, used for speeding up hashing in incremental compilation. | |
1157 | pub name_hash: u128, | |
1158 | /// Indicates which crate this `SourceFile` was imported from. | |
1159 | pub cnum: CrateNum, | |
1160 | } | |
1161 | ||
1162 | impl<S: Encoder> Encodable<S> for SourceFile { | |
1163 | fn encode(&self, s: &mut S) -> Result<(), S::Error> { | |
1164 | s.emit_struct("SourceFile", 8, |s| { | |
1165 | s.emit_struct_field("name", 0, |s| self.name.encode(s))?; | |
1166 | s.emit_struct_field("name_was_remapped", 1, |s| self.name_was_remapped.encode(s))?; | |
1167 | s.emit_struct_field("src_hash", 2, |s| self.src_hash.encode(s))?; | |
1168 | s.emit_struct_field("start_pos", 3, |s| self.start_pos.encode(s))?; | |
1169 | s.emit_struct_field("end_pos", 4, |s| self.end_pos.encode(s))?; | |
1170 | s.emit_struct_field("lines", 5, |s| { | |
1171 | let lines = &self.lines[..]; | |
1172 | // Store the length. | |
1173 | s.emit_u32(lines.len() as u32)?; | |
1174 | ||
1175 | if !lines.is_empty() { | |
1176 | // In order to preserve some space, we exploit the fact that | |
1177 | // the lines list is sorted and individual lines are | |
1178 | // probably not that long. Because of that we can store lines | |
1179 | // as a difference list, using as little space as possible | |
1180 | // for the differences. | |
1181 | let max_line_length = if lines.len() == 1 { | |
1182 | 0 | |
1183 | } else { | |
1184 | lines | |
1185 | .array_windows() | |
1186 | .map(|&[fst, snd]| snd - fst) | |
1187 | .map(|bp| bp.to_usize()) | |
1188 | .max() | |
1189 | .unwrap() | |
1190 | }; | |
1191 | ||
1192 | let bytes_per_diff: u8 = match max_line_length { | |
1193 | 0..=0xFF => 1, | |
1194 | 0x100..=0xFFFF => 2, | |
1195 | _ => 4, | |
1196 | }; | |
1197 | ||
1198 | // Encode the number of bytes used per diff. | |
1199 | bytes_per_diff.encode(s)?; | |
1200 | ||
1201 | // Encode the first element. | |
1202 | lines[0].encode(s)?; | |
1203 | ||
1204 | let diff_iter = lines[..].array_windows().map(|&[fst, snd]| snd - fst); | |
1205 | ||
1206 | match bytes_per_diff { | |
1207 | 1 => { | |
1208 | for diff in diff_iter { | |
1209 | (diff.0 as u8).encode(s)? | |
1210 | } | |
1211 | } | |
1212 | 2 => { | |
1213 | for diff in diff_iter { | |
1214 | (diff.0 as u16).encode(s)? | |
1215 | } | |
1216 | } | |
1217 | 4 => { | |
1218 | for diff in diff_iter { | |
1219 | diff.0.encode(s)? | |
1220 | } | |
1221 | } | |
1222 | _ => unreachable!(), | |
1223 | } | |
1224 | } | |
1225 | ||
1226 | Ok(()) | |
1227 | })?; | |
1228 | s.emit_struct_field("multibyte_chars", 6, |s| self.multibyte_chars.encode(s))?; | |
1229 | s.emit_struct_field("non_narrow_chars", 7, |s| self.non_narrow_chars.encode(s))?; | |
1230 | s.emit_struct_field("name_hash", 8, |s| self.name_hash.encode(s))?; | |
1231 | s.emit_struct_field("normalized_pos", 9, |s| self.normalized_pos.encode(s))?; | |
1232 | s.emit_struct_field("cnum", 10, |s| self.cnum.encode(s)) | |
1233 | }) | |
1234 | } | |
1235 | } | |
1236 | ||
1237 | impl<D: Decoder> Decodable<D> for SourceFile { | |
1238 | fn decode(d: &mut D) -> Result<SourceFile, D::Error> { | |
1239 | d.read_struct("SourceFile", 8, |d| { | |
1240 | let name: FileName = d.read_struct_field("name", 0, |d| Decodable::decode(d))?; | |
1241 | let name_was_remapped: bool = | |
1242 | d.read_struct_field("name_was_remapped", 1, |d| Decodable::decode(d))?; | |
1243 | let src_hash: SourceFileHash = | |
1244 | d.read_struct_field("src_hash", 2, |d| Decodable::decode(d))?; | |
1245 | let start_pos: BytePos = | |
1246 | d.read_struct_field("start_pos", 3, |d| Decodable::decode(d))?; | |
1247 | let end_pos: BytePos = d.read_struct_field("end_pos", 4, |d| Decodable::decode(d))?; | |
1248 | let lines: Vec<BytePos> = d.read_struct_field("lines", 5, |d| { | |
1249 | let num_lines: u32 = Decodable::decode(d)?; | |
1250 | let mut lines = Vec::with_capacity(num_lines as usize); | |
1251 | ||
1252 | if num_lines > 0 { | |
1253 | // Read the number of bytes used per diff. | |
1254 | let bytes_per_diff: u8 = Decodable::decode(d)?; | |
1255 | ||
1256 | // Read the first element. | |
1257 | let mut line_start: BytePos = Decodable::decode(d)?; | |
1258 | lines.push(line_start); | |
1259 | ||
1260 | for _ in 1..num_lines { | |
1261 | let diff = match bytes_per_diff { | |
1262 | 1 => d.read_u8()? as u32, | |
1263 | 2 => d.read_u16()? as u32, | |
1264 | 4 => d.read_u32()?, | |
1265 | _ => unreachable!(), | |
1266 | }; | |
1267 | ||
1268 | line_start = line_start + BytePos(diff); | |
1269 | ||
1270 | lines.push(line_start); | |
1271 | } | |
1272 | } | |
1273 | ||
1274 | Ok(lines) | |
1275 | })?; | |
1276 | let multibyte_chars: Vec<MultiByteChar> = | |
1277 | d.read_struct_field("multibyte_chars", 6, |d| Decodable::decode(d))?; | |
1278 | let non_narrow_chars: Vec<NonNarrowChar> = | |
1279 | d.read_struct_field("non_narrow_chars", 7, |d| Decodable::decode(d))?; | |
1280 | let name_hash: u128 = d.read_struct_field("name_hash", 8, |d| Decodable::decode(d))?; | |
1281 | let normalized_pos: Vec<NormalizedPos> = | |
1282 | d.read_struct_field("normalized_pos", 9, |d| Decodable::decode(d))?; | |
1283 | let cnum: CrateNum = d.read_struct_field("cnum", 10, |d| Decodable::decode(d))?; | |
1284 | Ok(SourceFile { | |
1285 | name, | |
1286 | name_was_remapped, | |
1287 | unmapped_path: None, | |
1288 | start_pos, | |
1289 | end_pos, | |
1290 | src: None, | |
1291 | src_hash, | |
1292 | // Unused - the metadata decoder will construct | |
1293 | // a new SourceFile, filling in `external_src` properly | |
1294 | external_src: Lock::new(ExternalSource::Unneeded), | |
1295 | lines, | |
1296 | multibyte_chars, | |
1297 | non_narrow_chars, | |
1298 | normalized_pos, | |
1299 | name_hash, | |
1300 | cnum, | |
1301 | }) | |
1302 | }) | |
1303 | } | |
1304 | } | |
1305 | ||
1306 | impl fmt::Debug for SourceFile { | |
1307 | fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { | |
1308 | write!(fmt, "SourceFile({})", self.name) | |
1309 | } | |
1310 | } | |
1311 | ||
1312 | impl SourceFile { | |
1313 | pub fn new( | |
1314 | name: FileName, | |
1315 | name_was_remapped: bool, | |
1316 | unmapped_path: FileName, | |
1317 | mut src: String, | |
1318 | start_pos: BytePos, | |
1319 | hash_kind: SourceFileHashAlgorithm, | |
1320 | ) -> Self { | |
1321 | // Compute the file hash before any normalization. | |
1322 | let src_hash = SourceFileHash::new(hash_kind, &src); | |
1323 | let normalized_pos = normalize_src(&mut src, start_pos); | |
1324 | ||
1325 | let name_hash = { | |
1326 | let mut hasher: StableHasher = StableHasher::new(); | |
1327 | name.hash(&mut hasher); | |
1328 | hasher.finish::<u128>() | |
1329 | }; | |
1330 | let end_pos = start_pos.to_usize() + src.len(); | |
1331 | assert!(end_pos <= u32::MAX as usize); | |
1332 | ||
1333 | let (lines, multibyte_chars, non_narrow_chars) = | |
1334 | analyze_source_file::analyze_source_file(&src[..], start_pos); | |
1335 | ||
1336 | SourceFile { | |
1337 | name, | |
1338 | name_was_remapped, | |
1339 | unmapped_path: Some(unmapped_path), | |
1340 | src: Some(Lrc::new(src)), | |
1341 | src_hash, | |
1342 | external_src: Lock::new(ExternalSource::Unneeded), | |
1343 | start_pos, | |
1344 | end_pos: Pos::from_usize(end_pos), | |
1345 | lines, | |
1346 | multibyte_chars, | |
1347 | non_narrow_chars, | |
1348 | normalized_pos, | |
1349 | name_hash, | |
1350 | cnum: LOCAL_CRATE, | |
1351 | } | |
1352 | } | |
1353 | ||
1354 | /// Returns the `BytePos` of the beginning of the current line. | |
1355 | pub fn line_begin_pos(&self, pos: BytePos) -> BytePos { | |
1356 | let line_index = self.lookup_line(pos).unwrap(); | |
1357 | self.lines[line_index] | |
1358 | } | |
1359 | ||
1360 | /// Add externally loaded source. | |
1361 | /// If the hash of the input doesn't match or no input is supplied via None, | |
1362 | /// it is interpreted as an error and the corresponding enum variant is set. | |
1363 | /// The return value signifies whether some kind of source is present. | |
1364 | pub fn add_external_src<F>(&self, get_src: F) -> bool | |
1365 | where | |
1366 | F: FnOnce() -> Option<String>, | |
1367 | { | |
1368 | if matches!( | |
1369 | *self.external_src.borrow(), | |
1370 | ExternalSource::Foreign { kind: ExternalSourceKind::AbsentOk, .. } | |
1371 | ) { | |
1372 | let src = get_src(); | |
1373 | let mut external_src = self.external_src.borrow_mut(); | |
1374 | // Check that no-one else have provided the source while we were getting it | |
1375 | if let ExternalSource::Foreign { | |
1376 | kind: src_kind @ ExternalSourceKind::AbsentOk, .. | |
1377 | } = &mut *external_src | |
1378 | { | |
1379 | if let Some(mut src) = src { | |
1380 | // The src_hash needs to be computed on the pre-normalized src. | |
1381 | if self.src_hash.matches(&src) { | |
1382 | normalize_src(&mut src, BytePos::from_usize(0)); | |
1383 | *src_kind = ExternalSourceKind::Present(Lrc::new(src)); | |
1384 | return true; | |
1385 | } | |
1386 | } else { | |
1387 | *src_kind = ExternalSourceKind::AbsentErr; | |
1388 | } | |
1389 | ||
1390 | false | |
1391 | } else { | |
1392 | self.src.is_some() || external_src.get_source().is_some() | |
1393 | } | |
1394 | } else { | |
1395 | self.src.is_some() || self.external_src.borrow().get_source().is_some() | |
1396 | } | |
1397 | } | |
1398 | ||
1399 | /// Gets a line from the list of pre-computed line-beginnings. | |
1400 | /// The line number here is 0-based. | |
1401 | pub fn get_line(&self, line_number: usize) -> Option<Cow<'_, str>> { | |
1402 | fn get_until_newline(src: &str, begin: usize) -> &str { | |
1403 | // We can't use `lines.get(line_number+1)` because we might | |
1404 | // be parsing when we call this function and thus the current | |
1405 | // line is the last one we have line info for. | |
1406 | let slice = &src[begin..]; | |
1407 | match slice.find('\n') { | |
1408 | Some(e) => &slice[..e], | |
1409 | None => slice, | |
1410 | } | |
1411 | } | |
1412 | ||
1413 | let begin = { | |
1414 | let line = self.lines.get(line_number)?; | |
1415 | let begin: BytePos = *line - self.start_pos; | |
1416 | begin.to_usize() | |
1417 | }; | |
1418 | ||
1419 | if let Some(ref src) = self.src { | |
1420 | Some(Cow::from(get_until_newline(src, begin))) | |
1421 | } else if let Some(src) = self.external_src.borrow().get_source() { | |
1422 | Some(Cow::Owned(String::from(get_until_newline(src, begin)))) | |
1423 | } else { | |
1424 | None | |
1425 | } | |
1426 | } | |
1427 | ||
1428 | pub fn is_real_file(&self) -> bool { | |
1429 | self.name.is_real() | |
1430 | } | |
1431 | ||
1432 | pub fn is_imported(&self) -> bool { | |
1433 | self.src.is_none() | |
1434 | } | |
1435 | ||
1436 | pub fn byte_length(&self) -> u32 { | |
1437 | self.end_pos.0 - self.start_pos.0 | |
1438 | } | |
1439 | pub fn count_lines(&self) -> usize { | |
1440 | self.lines.len() | |
1441 | } | |
1442 | ||
1443 | /// Finds the line containing the given position. The return value is the | |
1444 | /// index into the `lines` array of this `SourceFile`, not the 1-based line | |
1445 | /// number. If the source_file is empty or the position is located before the | |
1446 | /// first line, `None` is returned. | |
1447 | pub fn lookup_line(&self, pos: BytePos) -> Option<usize> { | |
1448 | if self.lines.is_empty() { | |
1449 | return None; | |
1450 | } | |
1451 | ||
1452 | let line_index = lookup_line(&self.lines[..], pos); | |
1453 | assert!(line_index < self.lines.len() as isize); | |
1454 | if line_index >= 0 { Some(line_index as usize) } else { None } | |
1455 | } | |
1456 | ||
1457 | pub fn line_bounds(&self, line_index: usize) -> Range<BytePos> { | |
1458 | if self.is_empty() { | |
1459 | return self.start_pos..self.end_pos; | |
1460 | } | |
1461 | ||
1462 | assert!(line_index < self.lines.len()); | |
1463 | if line_index == (self.lines.len() - 1) { | |
1464 | self.lines[line_index]..self.end_pos | |
1465 | } else { | |
1466 | self.lines[line_index]..self.lines[line_index + 1] | |
1467 | } | |
1468 | } | |
1469 | ||
1470 | /// Returns whether or not the file contains the given `SourceMap` byte | |
1471 | /// position. The position one past the end of the file is considered to be | |
1472 | /// contained by the file. This implies that files for which `is_empty` | |
1473 | /// returns true still contain one byte position according to this function. | |
1474 | #[inline] | |
1475 | pub fn contains(&self, byte_pos: BytePos) -> bool { | |
1476 | byte_pos >= self.start_pos && byte_pos <= self.end_pos | |
1477 | } | |
1478 | ||
1479 | #[inline] | |
1480 | pub fn is_empty(&self) -> bool { | |
1481 | self.start_pos == self.end_pos | |
1482 | } | |
1483 | ||
1484 | /// Calculates the original byte position relative to the start of the file | |
1485 | /// based on the given byte position. | |
1486 | pub fn original_relative_byte_pos(&self, pos: BytePos) -> BytePos { | |
1487 | // Diff before any records is 0. Otherwise use the previously recorded | |
1488 | // diff as that applies to the following characters until a new diff | |
1489 | // is recorded. | |
1490 | let diff = match self.normalized_pos.binary_search_by(|np| np.pos.cmp(&pos)) { | |
1491 | Ok(i) => self.normalized_pos[i].diff, | |
1492 | Err(i) if i == 0 => 0, | |
1493 | Err(i) => self.normalized_pos[i - 1].diff, | |
1494 | }; | |
1495 | ||
1496 | BytePos::from_u32(pos.0 - self.start_pos.0 + diff) | |
1497 | } | |
1498 | ||
1499 | /// Converts an absolute `BytePos` to a `CharPos` relative to the `SourceFile`. | |
1500 | pub fn bytepos_to_file_charpos(&self, bpos: BytePos) -> CharPos { | |
1501 | // The number of extra bytes due to multibyte chars in the `SourceFile`. | |
1502 | let mut total_extra_bytes = 0; | |
1503 | ||
1504 | for mbc in self.multibyte_chars.iter() { | |
1505 | debug!("{}-byte char at {:?}", mbc.bytes, mbc.pos); | |
1506 | if mbc.pos < bpos { | |
1507 | // Every character is at least one byte, so we only | |
1508 | // count the actual extra bytes. | |
1509 | total_extra_bytes += mbc.bytes as u32 - 1; | |
1510 | // We should never see a byte position in the middle of a | |
1511 | // character. | |
1512 | assert!(bpos.to_u32() >= mbc.pos.to_u32() + mbc.bytes as u32); | |
1513 | } else { | |
1514 | break; | |
1515 | } | |
1516 | } | |
1517 | ||
1518 | assert!(self.start_pos.to_u32() + total_extra_bytes <= bpos.to_u32()); | |
1519 | CharPos(bpos.to_usize() - self.start_pos.to_usize() - total_extra_bytes as usize) | |
1520 | } | |
1521 | ||
1522 | /// Looks up the file's (1-based) line number and (0-based `CharPos`) column offset, for a | |
1523 | /// given `BytePos`. | |
1524 | pub fn lookup_file_pos(&self, pos: BytePos) -> (usize, CharPos) { | |
1525 | let chpos = self.bytepos_to_file_charpos(pos); | |
1526 | match self.lookup_line(pos) { | |
1527 | Some(a) => { | |
1528 | let line = a + 1; // Line numbers start at 1 | |
1529 | let linebpos = self.lines[a]; | |
1530 | let linechpos = self.bytepos_to_file_charpos(linebpos); | |
1531 | let col = chpos - linechpos; | |
1532 | debug!("byte pos {:?} is on the line at byte pos {:?}", pos, linebpos); | |
1533 | debug!("char pos {:?} is on the line at char pos {:?}", chpos, linechpos); | |
1534 | debug!("byte is on line: {}", line); | |
1535 | assert!(chpos >= linechpos); | |
1536 | (line, col) | |
1537 | } | |
1538 | None => (0, chpos), | |
1539 | } | |
1540 | } | |
1541 | ||
1542 | /// Looks up the file's (1-based) line number, (0-based `CharPos`) column offset, and (0-based) | |
1543 | /// column offset when displayed, for a given `BytePos`. | |
1544 | pub fn lookup_file_pos_with_col_display(&self, pos: BytePos) -> (usize, CharPos, usize) { | |
1545 | let (line, col_or_chpos) = self.lookup_file_pos(pos); | |
1546 | if line > 0 { | |
1547 | let col = col_or_chpos; | |
1548 | let linebpos = self.lines[line - 1]; | |
1549 | let col_display = { | |
1550 | let start_width_idx = self | |
1551 | .non_narrow_chars | |
1552 | .binary_search_by_key(&linebpos, |x| x.pos()) | |
1553 | .unwrap_or_else(|x| x); | |
1554 | let end_width_idx = self | |
1555 | .non_narrow_chars | |
1556 | .binary_search_by_key(&pos, |x| x.pos()) | |
1557 | .unwrap_or_else(|x| x); | |
1558 | let special_chars = end_width_idx - start_width_idx; | |
1559 | let non_narrow: usize = self.non_narrow_chars[start_width_idx..end_width_idx] | |
1560 | .iter() | |
1561 | .map(|x| x.width()) | |
1562 | .sum(); | |
1563 | col.0 - special_chars + non_narrow | |
1564 | }; | |
1565 | (line, col, col_display) | |
1566 | } else { | |
1567 | let chpos = col_or_chpos; | |
1568 | let col_display = { | |
1569 | let end_width_idx = self | |
1570 | .non_narrow_chars | |
1571 | .binary_search_by_key(&pos, |x| x.pos()) | |
1572 | .unwrap_or_else(|x| x); | |
1573 | let non_narrow: usize = | |
1574 | self.non_narrow_chars[0..end_width_idx].iter().map(|x| x.width()).sum(); | |
1575 | chpos.0 - end_width_idx + non_narrow | |
1576 | }; | |
1577 | (0, chpos, col_display) | |
1578 | } | |
1579 | } | |
1580 | } | |
1581 | ||
1582 | /// Normalizes the source code and records the normalizations. | |
1583 | fn normalize_src(src: &mut String, start_pos: BytePos) -> Vec<NormalizedPos> { | |
1584 | let mut normalized_pos = vec![]; | |
1585 | remove_bom(src, &mut normalized_pos); | |
1586 | normalize_newlines(src, &mut normalized_pos); | |
1587 | ||
1588 | // Offset all the positions by start_pos to match the final file positions. | |
1589 | for np in &mut normalized_pos { | |
1590 | np.pos.0 += start_pos.0; | |
1591 | } | |
1592 | ||
1593 | normalized_pos | |
1594 | } | |
1595 | ||
1596 | /// Removes UTF-8 BOM, if any. | |
1597 | fn remove_bom(src: &mut String, normalized_pos: &mut Vec<NormalizedPos>) { | |
1598 | if src.starts_with('\u{feff}') { | |
1599 | src.drain(..3); | |
1600 | normalized_pos.push(NormalizedPos { pos: BytePos(0), diff: 3 }); | |
1601 | } | |
1602 | } | |
1603 | ||
1604 | /// Replaces `\r\n` with `\n` in-place in `src`. | |
1605 | /// | |
1606 | /// Returns error if there's a lone `\r` in the string. | |
1607 | fn normalize_newlines(src: &mut String, normalized_pos: &mut Vec<NormalizedPos>) { | |
1608 | if !src.as_bytes().contains(&b'\r') { | |
1609 | return; | |
1610 | } | |
1611 | ||
1612 | // We replace `\r\n` with `\n` in-place, which doesn't break utf-8 encoding. | |
1613 | // While we *can* call `as_mut_vec` and do surgery on the live string | |
1614 | // directly, let's rather steal the contents of `src`. This makes the code | |
1615 | // safe even if a panic occurs. | |
1616 | ||
1617 | let mut buf = std::mem::replace(src, String::new()).into_bytes(); | |
1618 | let mut gap_len = 0; | |
1619 | let mut tail = buf.as_mut_slice(); | |
1620 | let mut cursor = 0; | |
1621 | let original_gap = normalized_pos.last().map_or(0, |l| l.diff); | |
1622 | loop { | |
1623 | let idx = match find_crlf(&tail[gap_len..]) { | |
1624 | None => tail.len(), | |
1625 | Some(idx) => idx + gap_len, | |
1626 | }; | |
1627 | tail.copy_within(gap_len..idx, 0); | |
1628 | tail = &mut tail[idx - gap_len..]; | |
1629 | if tail.len() == gap_len { | |
1630 | break; | |
1631 | } | |
1632 | cursor += idx - gap_len; | |
1633 | gap_len += 1; | |
1634 | normalized_pos.push(NormalizedPos { | |
1635 | pos: BytePos::from_usize(cursor + 1), | |
1636 | diff: original_gap + gap_len as u32, | |
1637 | }); | |
1638 | } | |
1639 | ||
1640 | // Account for removed `\r`. | |
1641 | // After `set_len`, `buf` is guaranteed to contain utf-8 again. | |
1642 | let new_len = buf.len() - gap_len; | |
1643 | unsafe { | |
1644 | buf.set_len(new_len); | |
1645 | *src = String::from_utf8_unchecked(buf); | |
1646 | } | |
1647 | ||
1648 | fn find_crlf(src: &[u8]) -> Option<usize> { | |
1649 | let mut search_idx = 0; | |
1650 | while let Some(idx) = find_cr(&src[search_idx..]) { | |
1651 | if src[search_idx..].get(idx + 1) != Some(&b'\n') { | |
1652 | search_idx += idx + 1; | |
1653 | continue; | |
1654 | } | |
1655 | return Some(search_idx + idx); | |
1656 | } | |
1657 | None | |
1658 | } | |
1659 | ||
1660 | fn find_cr(src: &[u8]) -> Option<usize> { | |
1661 | src.iter().position(|&b| b == b'\r') | |
1662 | } | |
1663 | } | |
1664 | ||
1665 | // _____________________________________________________________________________ | |
1666 | // Pos, BytePos, CharPos | |
1667 | // | |
1668 | ||
1669 | pub trait Pos { | |
1670 | fn from_usize(n: usize) -> Self; | |
1671 | fn to_usize(&self) -> usize; | |
1672 | fn from_u32(n: u32) -> Self; | |
1673 | fn to_u32(&self) -> u32; | |
1674 | } | |
1675 | ||
1676 | macro_rules! impl_pos { | |
1677 | ( | |
1678 | $( | |
1679 | $(#[$attr:meta])* | |
1680 | $vis:vis struct $ident:ident($inner_vis:vis $inner_ty:ty); | |
1681 | )* | |
1682 | ) => { | |
1683 | $( | |
1684 | $(#[$attr])* | |
1685 | $vis struct $ident($inner_vis $inner_ty); | |
1686 | ||
1687 | impl Pos for $ident { | |
1688 | #[inline(always)] | |
1689 | fn from_usize(n: usize) -> $ident { | |
1690 | $ident(n as $inner_ty) | |
1691 | } | |
1692 | ||
1693 | #[inline(always)] | |
1694 | fn to_usize(&self) -> usize { | |
1695 | self.0 as usize | |
1696 | } | |
1697 | ||
1698 | #[inline(always)] | |
1699 | fn from_u32(n: u32) -> $ident { | |
1700 | $ident(n as $inner_ty) | |
1701 | } | |
1702 | ||
1703 | #[inline(always)] | |
1704 | fn to_u32(&self) -> u32 { | |
1705 | self.0 as u32 | |
1706 | } | |
1707 | } | |
1708 | ||
1709 | impl Add for $ident { | |
1710 | type Output = $ident; | |
1711 | ||
1712 | #[inline(always)] | |
1713 | fn add(self, rhs: $ident) -> $ident { | |
1714 | $ident(self.0 + rhs.0) | |
1715 | } | |
1716 | } | |
1717 | ||
1718 | impl Sub for $ident { | |
1719 | type Output = $ident; | |
1720 | ||
1721 | #[inline(always)] | |
1722 | fn sub(self, rhs: $ident) -> $ident { | |
1723 | $ident(self.0 - rhs.0) | |
1724 | } | |
1725 | } | |
1726 | )* | |
1727 | }; | |
1728 | } | |
1729 | ||
1730 | impl_pos! { | |
1731 | /// A byte offset. | |
1732 | /// | |
1733 | /// Keep this small (currently 32-bits), as AST contains a lot of them. | |
1734 | #[derive(Clone, Copy, PartialEq, Eq, Hash, PartialOrd, Ord, Debug)] | |
1735 | pub struct BytePos(pub u32); | |
1736 | ||
1737 | /// A character offset. | |
1738 | /// | |
1739 | /// Because of multibyte UTF-8 characters, a byte offset | |
1740 | /// is not equivalent to a character offset. The [`SourceMap`] will convert [`BytePos`] | |
1741 | /// values to `CharPos` values as necessary. | |
1742 | #[derive(Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Debug)] | |
1743 | pub struct CharPos(pub usize); | |
1744 | } | |
1745 | ||
1746 | impl<S: rustc_serialize::Encoder> Encodable<S> for BytePos { | |
1747 | fn encode(&self, s: &mut S) -> Result<(), S::Error> { | |
1748 | s.emit_u32(self.0) | |
1749 | } | |
1750 | } | |
1751 | ||
1752 | impl<D: rustc_serialize::Decoder> Decodable<D> for BytePos { | |
1753 | fn decode(d: &mut D) -> Result<BytePos, D::Error> { | |
1754 | Ok(BytePos(d.read_u32()?)) | |
1755 | } | |
1756 | } | |
1757 | ||
1758 | // _____________________________________________________________________________ | |
1759 | // Loc, SourceFileAndLine, SourceFileAndBytePos | |
1760 | // | |
1761 | ||
1762 | /// A source code location used for error reporting. | |
1763 | #[derive(Debug, Clone)] | |
1764 | pub struct Loc { | |
1765 | /// Information about the original source. | |
1766 | pub file: Lrc<SourceFile>, | |
1767 | /// The (1-based) line number. | |
1768 | pub line: usize, | |
1769 | /// The (0-based) column offset. | |
1770 | pub col: CharPos, | |
1771 | /// The (0-based) column offset when displayed. | |
1772 | pub col_display: usize, | |
1773 | } | |
1774 | ||
1775 | // Used to be structural records. | |
1776 | #[derive(Debug)] | |
1777 | pub struct SourceFileAndLine { | |
1778 | pub sf: Lrc<SourceFile>, | |
1779 | pub line: usize, | |
1780 | } | |
1781 | #[derive(Debug)] | |
1782 | pub struct SourceFileAndBytePos { | |
1783 | pub sf: Lrc<SourceFile>, | |
1784 | pub pos: BytePos, | |
1785 | } | |
1786 | ||
1787 | #[derive(Copy, Clone, Debug, PartialEq, Eq)] | |
1788 | pub struct LineInfo { | |
1789 | /// Index of line, starting from 0. | |
1790 | pub line_index: usize, | |
1791 | ||
1792 | /// Column in line where span begins, starting from 0. | |
1793 | pub start_col: CharPos, | |
1794 | ||
1795 | /// Column in line where span ends, starting from 0, exclusive. | |
1796 | pub end_col: CharPos, | |
1797 | } | |
1798 | ||
1799 | pub struct FileLines { | |
1800 | pub file: Lrc<SourceFile>, | |
1801 | pub lines: Vec<LineInfo>, | |
1802 | } | |
1803 | ||
1804 | pub static SPAN_DEBUG: AtomicRef<fn(Span, &mut fmt::Formatter<'_>) -> fmt::Result> = | |
1805 | AtomicRef::new(&(default_span_debug as fn(_, &mut fmt::Formatter<'_>) -> _)); | |
1806 | ||
1807 | // _____________________________________________________________________________ | |
1808 | // SpanLinesError, SpanSnippetError, DistinctSources, MalformedSourceMapPositions | |
1809 | // | |
1810 | ||
1811 | pub type FileLinesResult = Result<FileLines, SpanLinesError>; | |
1812 | ||
1813 | #[derive(Clone, PartialEq, Eq, Debug)] | |
1814 | pub enum SpanLinesError { | |
1815 | DistinctSources(DistinctSources), | |
1816 | } | |
1817 | ||
1818 | #[derive(Clone, PartialEq, Eq, Debug)] | |
1819 | pub enum SpanSnippetError { | |
1820 | IllFormedSpan(Span), | |
1821 | DistinctSources(DistinctSources), | |
1822 | MalformedForSourcemap(MalformedSourceMapPositions), | |
1823 | SourceNotAvailable { filename: FileName }, | |
1824 | } | |
1825 | ||
1826 | #[derive(Clone, PartialEq, Eq, Debug)] | |
1827 | pub struct DistinctSources { | |
1828 | pub begin: (FileName, BytePos), | |
1829 | pub end: (FileName, BytePos), | |
1830 | } | |
1831 | ||
1832 | #[derive(Clone, PartialEq, Eq, Debug)] | |
1833 | pub struct MalformedSourceMapPositions { | |
1834 | pub name: FileName, | |
1835 | pub source_len: usize, | |
1836 | pub begin_pos: BytePos, | |
1837 | pub end_pos: BytePos, | |
1838 | } | |
1839 | ||
1840 | /// Range inside of a `Span` used for diagnostics when we only have access to relative positions. | |
1841 | #[derive(Copy, Clone, PartialEq, Eq, Debug)] | |
1842 | pub struct InnerSpan { | |
1843 | pub start: usize, | |
1844 | pub end: usize, | |
1845 | } | |
1846 | ||
1847 | impl InnerSpan { | |
1848 | pub fn new(start: usize, end: usize) -> InnerSpan { | |
1849 | InnerSpan { start, end } | |
1850 | } | |
1851 | } | |
1852 | ||
1853 | // Given a slice of line start positions and a position, returns the index of | |
1854 | // the line the position is on. Returns -1 if the position is located before | |
1855 | // the first line. | |
1856 | fn lookup_line(lines: &[BytePos], pos: BytePos) -> isize { | |
1857 | match lines.binary_search(&pos) { | |
1858 | Ok(line) => line as isize, | |
1859 | Err(line) => line as isize - 1, | |
1860 | } | |
1861 | } | |
1862 | ||
1863 | /// Requirements for a `StableHashingContext` to be used in this crate. | |
1864 | /// | |
1865 | /// This is a hack to allow using the [`HashStable_Generic`] derive macro | |
1866 | /// instead of implementing everything in rustc_middle. | |
1867 | pub trait HashStableContext { | |
1868 | fn hash_def_id(&mut self, _: DefId, hasher: &mut StableHasher); | |
1869 | /// Obtains a cache for storing the `Fingerprint` of an `ExpnId`. | |
1870 | /// This method allows us to have multiple `HashStableContext` implementations | |
1871 | /// that hash things in a different way, without the results of one polluting | |
1872 | /// the cache of the other. | |
1873 | fn expn_id_cache() -> &'static LocalKey<ExpnIdCache>; | |
1874 | fn hash_crate_num(&mut self, _: CrateNum, hasher: &mut StableHasher); | |
1875 | fn hash_spans(&self) -> bool; | |
1876 | fn byte_pos_to_line_and_col( | |
1877 | &mut self, | |
1878 | byte: BytePos, | |
1879 | ) -> Option<(Lrc<SourceFile>, usize, BytePos)>; | |
1880 | fn span_data_to_lines_and_cols( | |
1881 | &mut self, | |
1882 | span: &SpanData, | |
1883 | ) -> Option<(Lrc<SourceFile>, usize, BytePos, usize, BytePos)>; | |
1884 | } | |
1885 | ||
1886 | impl<CTX> HashStable<CTX> for Span | |
1887 | where | |
1888 | CTX: HashStableContext, | |
1889 | { | |
1890 | /// Hashes a span in a stable way. We can't directly hash the span's `BytePos` | |
1891 | /// fields (that would be similar to hashing pointers, since those are just | |
1892 | /// offsets into the `SourceMap`). Instead, we hash the (file name, line, column) | |
1893 | /// triple, which stays the same even if the containing `SourceFile` has moved | |
1894 | /// within the `SourceMap`. | |
1895 | /// | |
1896 | /// Also note that we are hashing byte offsets for the column, not unicode | |
1897 | /// codepoint offsets. For the purpose of the hash that's sufficient. | |
1898 | /// Also, hashing filenames is expensive so we avoid doing it twice when the | |
1899 | /// span starts and ends in the same file, which is almost always the case. | |
1900 | fn hash_stable(&self, ctx: &mut CTX, hasher: &mut StableHasher) { | |
1901 | const TAG_VALID_SPAN: u8 = 0; | |
1902 | const TAG_INVALID_SPAN: u8 = 1; | |
1903 | ||
1904 | if !ctx.hash_spans() { | |
1905 | return; | |
1906 | } | |
1907 | ||
1908 | if self.is_dummy() { | |
1909 | Hash::hash(&TAG_INVALID_SPAN, hasher); | |
1910 | self.ctxt().hash_stable(ctx, hasher); | |
1911 | return; | |
1912 | } | |
1913 | ||
1914 | // If this is not an empty or invalid span, we want to hash the last | |
1915 | // position that belongs to it, as opposed to hashing the first | |
1916 | // position past it. | |
1917 | let span = self.data(); | |
1918 | let (file, line_lo, col_lo, line_hi, col_hi) = match ctx.span_data_to_lines_and_cols(&span) | |
1919 | { | |
1920 | Some(pos) => pos, | |
1921 | None => { | |
1922 | Hash::hash(&TAG_INVALID_SPAN, hasher); | |
1923 | span.ctxt.hash_stable(ctx, hasher); | |
1924 | return; | |
1925 | } | |
1926 | }; | |
1927 | ||
1928 | Hash::hash(&TAG_VALID_SPAN, hasher); | |
1929 | // We truncate the stable ID hash and line and column numbers. The chances | |
1930 | // of causing a collision this way should be minimal. | |
1931 | Hash::hash(&(file.name_hash as u64), hasher); | |
1932 | ||
1933 | // Hash both the length and the end location (line/column) of a span. If we | |
1934 | // hash only the length, for example, then two otherwise equal spans with | |
1935 | // different end locations will have the same hash. This can cause a problem | |
1936 | // during incremental compilation wherein a previous result for a query that | |
1937 | // depends on the end location of a span will be incorrectly reused when the | |
1938 | // end location of the span it depends on has changed (see issue #74890). A | |
1939 | // similar analysis applies if some query depends specifically on the length | |
1940 | // of the span, but we only hash the end location. So hash both. | |
1941 | ||
1942 | let col_lo_trunc = (col_lo.0 as u64) & 0xFF; | |
1943 | let line_lo_trunc = ((line_lo as u64) & 0xFF_FF_FF) << 8; | |
1944 | let col_hi_trunc = (col_hi.0 as u64) & 0xFF << 32; | |
1945 | let line_hi_trunc = ((line_hi as u64) & 0xFF_FF_FF) << 40; | |
1946 | let col_line = col_lo_trunc | line_lo_trunc | col_hi_trunc | line_hi_trunc; | |
1947 | let len = (span.hi - span.lo).0; | |
1948 | Hash::hash(&col_line, hasher); | |
1949 | Hash::hash(&len, hasher); | |
1950 | span.ctxt.hash_stable(ctx, hasher); | |
1951 | } | |
1952 | } | |
1953 | ||
1954 | impl<CTX: HashStableContext> HashStable<CTX> for SyntaxContext { | |
1955 | fn hash_stable(&self, ctx: &mut CTX, hasher: &mut StableHasher) { | |
1956 | const TAG_EXPANSION: u8 = 0; | |
1957 | const TAG_NO_EXPANSION: u8 = 1; | |
1958 | ||
1959 | if *self == SyntaxContext::root() { | |
1960 | TAG_NO_EXPANSION.hash_stable(ctx, hasher); | |
1961 | } else { | |
1962 | TAG_EXPANSION.hash_stable(ctx, hasher); | |
1963 | let (expn_id, transparency) = self.outer_mark(); | |
1964 | expn_id.hash_stable(ctx, hasher); | |
1965 | transparency.hash_stable(ctx, hasher); | |
1966 | } | |
1967 | } | |
1968 | } | |
1969 | ||
1970 | pub type ExpnIdCache = RefCell<Vec<Option<Fingerprint>>>; | |
1971 | ||
1972 | impl<CTX: HashStableContext> HashStable<CTX> for ExpnId { | |
1973 | fn hash_stable(&self, ctx: &mut CTX, hasher: &mut StableHasher) { | |
1974 | const TAG_ROOT: u8 = 0; | |
1975 | const TAG_NOT_ROOT: u8 = 1; | |
1976 | ||
1977 | if *self == ExpnId::root() { | |
1978 | TAG_ROOT.hash_stable(ctx, hasher); | |
1979 | return; | |
1980 | } | |
1981 | ||
1982 | // Since the same expansion context is usually referenced many | |
1983 | // times, we cache a stable hash of it and hash that instead of | |
1984 | // recursing every time. | |
1985 | let index = self.as_u32() as usize; | |
1986 | let res = CTX::expn_id_cache().with(|cache| cache.borrow().get(index).copied().flatten()); | |
1987 | ||
1988 | if let Some(res) = res { | |
1989 | res.hash_stable(ctx, hasher); | |
1990 | } else { | |
1991 | let new_len = index + 1; | |
1992 | ||
1993 | let mut sub_hasher = StableHasher::new(); | |
1994 | TAG_NOT_ROOT.hash_stable(ctx, &mut sub_hasher); | |
1995 | self.expn_data().hash_stable(ctx, &mut sub_hasher); | |
1996 | let sub_hash: Fingerprint = sub_hasher.finish(); | |
1997 | ||
1998 | CTX::expn_id_cache().with(|cache| { | |
1999 | let mut cache = cache.borrow_mut(); | |
2000 | if cache.len() < new_len { | |
2001 | cache.resize(new_len, None); | |
2002 | } | |
2003 | cache[index].replace(sub_hash).expect_none("Cache slot was filled"); | |
2004 | }); | |
2005 | sub_hash.hash_stable(ctx, hasher); | |
2006 | } | |
2007 | } | |
2008 | } |