mir_analyzer/session/queries.rs
1use std::sync::Arc;
2use std::time::Instant;
3
4use super::*;
5
6impl AnalysisSession {
7 /// Open-file diagnostics entrypoint for issue-only consumers.
8 ///
9 /// Updates the session's current text for `file`, reuses the tracked parse
10 /// and definition queries for syntax/collector diagnostics, then runs the
11 /// targeted open-file body-analysis path without retaining whole-file
12 /// `ResolvedSymbol` payloads.
13 pub fn analyze_file_diagnostics(&mut self, file: &str, source: &str) -> crate::FileAnalysis {
14 let file: Arc<str> = Arc::from(file);
15 self.ingest_file(file.clone(), Arc::from(source));
16
17 let db = self.snapshot_db();
18 let Some(sf) = db.lookup_source_file(file.as_ref()) else {
19 return crate::FileAnalysis {
20 issues: Vec::new(),
21 symbols: Vec::new(),
22 };
23 };
24 let defs = crate::db::collect_file_definitions(&db, sf);
25 let prepared = crate::db::prepare_analysis_file(&db, sf);
26 let mut issues = Arc::unwrap_or_clone(defs.issues.clone());
27
28 if !prepared.has_hard_parse_errors {
29 let parsed = prepared.parse_result();
30 let analysis = crate::FileAnalyzer::new(self).analyze_diagnostics_only(
31 file.clone(),
32 prepared.text.as_ref(),
33 &parsed.program,
34 &parsed.source_map,
35 );
36 issues.extend(analysis.issues);
37 }
38
39 self.apply_suppressions_and_emit_unused(&mut issues, std::slice::from_ref(&file));
40 crate::FileAnalysis {
41 issues,
42 symbols: Vec::new(),
43 }
44 }
45
46 fn codebase_name_at_via_resolve(
47 &mut self,
48 file: &str,
49 byte_offset: u32,
50 ) -> Result<crate::Name, crate::SymbolLookupError> {
51 self.resolve_at(file, byte_offset)
52 .ok_or(crate::SymbolLookupError::NotFound)?
53 .to_symbol()
54 .ok_or(crate::SymbolLookupError::NotFound)
55 }
56
57 /// Resolve the codebase-level symbol name at `byte_offset` in `file`.
58 ///
59 /// This is the compact cursor-navigation helper for consumers that only
60 /// need symbol identity (for example, references queries) and not the full
61 /// `ResolvedSymbol` payload.
62 pub fn name_at(&mut self, file: &str, byte_offset: u32) -> Option<crate::Name> {
63 let started = Instant::now();
64 let name = crate::FileAnalyzer::new(self).resolve_name_at(Arc::from(file), byte_offset);
65 crate::metrics::record_name_at(started.elapsed().as_micros() as u64);
66 name
67 }
68
69 /// Resolve the symbol at `byte_offset` in `file`'s current ingested text.
70 ///
71 /// Canonical open-file navigation entrypoint: unlike
72 /// [`crate::FileAnalysis::symbol_at`], this does not require the caller to
73 /// retain a whole-file symbol list from diagnostics.
74 pub fn symbol_at(&mut self, file: &str, byte_offset: u32) -> Option<crate::ResolvedSymbol> {
75 self.resolve_at(file, byte_offset)
76 }
77
78 /// Resolve the symbol at `byte_offset` in `file`'s current ingested text.
79 ///
80 /// This powers hover / go-to-definition without requiring the caller to
81 /// retain a whole-file `ResolvedSymbol` list: the containing scope is
82 /// re-analyzed on demand and only its symbol payload is searched.
83 ///
84 /// **Side effects:** like [`Self::definition_of`] and [`Self::hover`], this
85 /// may fault in direct dependencies of `file` by running the open-file
86 /// warm-up path (`prepare_file_for_analysis`) before snapshotting.
87 pub fn resolve_at(&mut self, file: &str, byte_offset: u32) -> Option<crate::ResolvedSymbol> {
88 let started = Instant::now();
89 let sym = crate::FileAnalyzer::new(self).resolve_at(Arc::from(file), byte_offset);
90 crate::metrics::record_resolve_at(started.elapsed().as_micros() as u64);
91 sym
92 }
93
94 /// Hover information for the symbol at `byte_offset` in `file`.
95 ///
96 /// Uses the targeted [`Self::resolve_at`] navigation path, then resolves
97 /// the resulting symbol's hover payload.
98 pub fn hover_at(
99 &mut self,
100 file: &str,
101 byte_offset: u32,
102 ) -> Result<crate::HoverInfo, crate::SymbolLookupError> {
103 let started = Instant::now();
104 let hover = (|| {
105 let resolved = self
106 .resolve_at(file, byte_offset)
107 .ok_or(crate::SymbolLookupError::NotFound)?;
108 let Some(name) = resolved.to_symbol() else {
109 return Ok(crate::HoverInfo {
110 ty: resolved.resolved_type,
111 docstring: None,
112 definition: None,
113 });
114 };
115
116 let mut hover = self.hover(&name)?;
117 if !matches!(
118 resolved.kind,
119 crate::ReferenceKind::ClassReference(_) | crate::ReferenceKind::UseImport(_)
120 ) {
121 hover.ty = resolved.resolved_type;
122 }
123 Ok(hover)
124 })();
125 crate::metrics::record_hover_at(started.elapsed().as_micros() as u64);
126 hover
127 }
128
129 /// Definition location for the symbol at `byte_offset` in `file`.
130 ///
131 /// Uses the targeted [`Self::resolve_at`] navigation path, then resolves
132 /// the resulting symbol to its declaration site.
133 pub fn definition_at(
134 &mut self,
135 file: &str,
136 byte_offset: u32,
137 ) -> Result<mir_types::Location, crate::SymbolLookupError> {
138 let started = Instant::now();
139 let definition = (|| {
140 let name = self.codebase_name_at_via_resolve(file, byte_offset)?;
141 self.definition_of(&name)
142 })();
143 crate::metrics::record_definition_at(started.elapsed().as_micros() as u64);
144 definition
145 }
146
147 /// Reference locations for the symbol at `byte_offset` in `file`.
148 ///
149 /// This is the compact cursor-navigation entrypoint for find-references:
150 /// resolve a typed symbol identity with [`Self::name_at`], then answer the
151 /// query from the maintained reference index.
152 pub fn references_at(
153 &mut self,
154 file: &str,
155 byte_offset: u32,
156 files: &[Arc<str>],
157 include_declaration: bool,
158 includes: crate::ReferenceIncludes,
159 ) -> Result<Vec<(Arc<str>, crate::Range)>, crate::SymbolLookupError> {
160 self.references_at_cancellable(
161 file,
162 byte_offset,
163 files,
164 include_declaration,
165 includes,
166 &|| false,
167 )
168 .map(|refs| refs.expect("uncancelled references_at query should not return None"))
169 }
170
171 /// Cancellable variant of [`Self::references_at`].
172 ///
173 /// Returns `Err(NotFound)` when no symbol exists at the cursor. Returns
174 /// `Ok(None)` when `should_cancel` aborts the underlying indexed query.
175 #[allow(clippy::type_complexity)]
176 pub fn references_at_cancellable(
177 &mut self,
178 file: &str,
179 byte_offset: u32,
180 files: &[Arc<str>],
181 include_declaration: bool,
182 includes: crate::ReferenceIncludes,
183 should_cancel: &(dyn Fn() -> bool + Sync),
184 ) -> Result<Option<Vec<(Arc<str>, crate::Range)>>, crate::SymbolLookupError> {
185 let name = self
186 .name_at(file, byte_offset)
187 .ok_or(crate::SymbolLookupError::NotFound)?;
188 Ok(self.indexed_references_to(&name, files, include_declaration, includes, should_cancel))
189 }
190
191 /// Resolve a top-level symbol (class or function) to its declaration
192 /// location. Powers go-to-definition.
193 ///
194 /// **Side effects:** if the symbol isn't yet known, this may invoke the
195 /// configured [`crate::SourceProvider`] to fault in additional files and
196 /// mutate the salsa input set. Use [`Self::definition_of_cached`] for a
197 /// pure variant that only consults already-loaded state.
198 ///
199 /// Returns:
200 /// - `Ok(Location)` — symbol found with a source location
201 /// - `Err(NotFound)` — no such symbol in the codebase
202 /// - `Err(NoSourceLocation)` — symbol exists but has no recorded span
203 /// (e.g. some stub-only declarations)
204 pub fn definition_of(
205 &mut self,
206 symbol: &crate::Name,
207 ) -> Result<mir_types::Location, crate::SymbolLookupError> {
208 // Trigger any necessary lazy-load mutations before snapshotting.
209 match symbol {
210 crate::Name::Class(fqcn) => {
211 let _ = self.load_class(fqcn.as_ref());
212 }
213 crate::Name::Function(fqn) => {
214 let _ = self.load_class(fqn.as_ref());
215 }
216 crate::Name::Method { class, .. }
217 | crate::Name::Property { class, .. }
218 | crate::Name::ClassConstant { class, .. } => {
219 let _ = self.load_class(class.as_ref());
220 }
221 _ => {}
222 }
223 self.definition_of_cached(symbol)
224 }
225
226 /// Pure variant of [`Self::definition_of`]. Never invokes the
227 /// [`crate::SourceProvider`] and never mutates salsa inputs; resolves
228 /// only against state already loaded by `set_file_text` / `ingest_file`.
229 /// Returns `Err(NotFound)` when the symbol isn't in the loaded set, even
230 /// if a resolver could in principle map it.
231 pub fn definition_of_cached(
232 &self,
233 symbol: &crate::Name,
234 ) -> Result<mir_types::Location, crate::SymbolLookupError> {
235 let db = self.snapshot_db();
236 match symbol {
237 crate::Name::Class(fqcn) => {
238 let here = crate::db::Fqcn::from_str(&db, fqcn.as_ref());
239 let class = crate::db::find_class_like(&db, here)
240 .ok_or(crate::SymbolLookupError::NotFound)?;
241 class
242 .location()
243 .cloned()
244 .ok_or(crate::SymbolLookupError::NoSourceLocation)
245 }
246 crate::Name::Function(fqn) => {
247 let here = crate::db::Fqcn::from_str(&db, fqn.as_ref());
248 let f = crate::db::find_function(&db, here)
249 .ok_or(crate::SymbolLookupError::NotFound)?;
250 f.location
251 .clone()
252 .ok_or(crate::SymbolLookupError::NoSourceLocation)
253 }
254 crate::Name::Method { class, name }
255 | crate::Name::Property { class, name }
256 | crate::Name::ClassConstant { class, name } => {
257 crate::db::member_location(&db, class, name)
258 .ok_or(crate::SymbolLookupError::NotFound)
259 }
260 crate::Name::GlobalConstant(_) => Err(crate::SymbolLookupError::NoSourceLocation),
261 }
262 }
263
264 /// Hover information for a symbol: type, docstring, and definition location.
265 ///
266 /// For cursor-based editor navigation, prefer [`Self::hover_at`], which
267 /// uses the targeted [`Self::resolve_at`] path instead of requiring a
268 /// retained whole-file symbol list. This method assembles hover data once
269 /// the caller already has a typed [`crate::Name`].
270 ///
271 /// **Side effects:** when `symbol`'s owning class isn't yet loaded, this
272 /// may invoke the configured [`crate::SourceProvider`] to fault in
273 /// dependencies. Use [`Self::hover_cached`] for a pure variant.
274 ///
275 /// Returns `Err(NotFound)` if the symbol doesn't exist. May still return
276 /// `Ok` with `docstring: None` or `definition: None` if those specific
277 /// pieces aren't available.
278 pub fn hover(
279 &mut self,
280 symbol: &crate::Name,
281 ) -> Result<crate::HoverInfo, crate::SymbolLookupError> {
282 // Trigger lazy loading for class-rooted symbols before snapshotting.
283 // No-op when the class is already known; ensures inherited member
284 // lookups have the chain present.
285 match symbol {
286 crate::Name::Class(fqcn) => {
287 self.load_class(fqcn.as_ref());
288 }
289 crate::Name::Method { class, .. }
290 | crate::Name::Property { class, .. }
291 | crate::Name::ClassConstant { class, .. } => {
292 // Fault in the owning class for navigation if the background
293 // indexer hasn't reached it yet. Its inheritance ancestors
294 // resolve through the (eagerly-built) workspace symbol index.
295 self.load_class(class.as_ref());
296 }
297 _ => {}
298 }
299 self.hover_cached(symbol)
300 }
301
302 /// Pure variant of [`Self::hover`]. Never invokes the
303 /// [`crate::SourceProvider`]; consults only the already-loaded db.
304 pub fn hover_cached(
305 &self,
306 symbol: &crate::Name,
307 ) -> Result<crate::HoverInfo, crate::SymbolLookupError> {
308 use mir_types::{Atomic, Type};
309 let db = self.snapshot_db();
310 match symbol {
311 crate::Name::Function(fqn) => {
312 let here = crate::db::Fqcn::from_str(&db, fqn.as_ref());
313 let f = crate::db::find_function(&db, here)
314 .ok_or(crate::SymbolLookupError::NotFound)?;
315 let ty = f
316 .return_type
317 .as_deref()
318 .cloned()
319 .unwrap_or_else(Type::mixed);
320 let docstring = f.docstring.as_ref().map(|s| s.to_string());
321 Ok(crate::HoverInfo {
322 ty,
323 docstring,
324 definition: f.location.clone(),
325 })
326 }
327 crate::Name::Method { class, name } => {
328 let here = crate::db::Fqcn::from_str(&db, class.as_ref());
329 let (_, m) = crate::db::find_method_in_chain(&db, here, name)
330 .ok_or(crate::SymbolLookupError::NotFound)?;
331 let ty = m
332 .return_type
333 .as_deref()
334 .cloned()
335 .unwrap_or_else(Type::mixed);
336 let docstring = m.docstring.as_ref().map(|s| s.to_string());
337 Ok(crate::HoverInfo {
338 ty,
339 docstring,
340 definition: m.location.clone(),
341 })
342 }
343 crate::Name::Class(fqcn) => {
344 let here = crate::db::Fqcn::from_str(&db, fqcn.as_ref());
345 let class = crate::db::find_class_like(&db, here)
346 .ok_or(crate::SymbolLookupError::NotFound)?;
347 let ty = Type::single(Atomic::TNamedObject {
348 fqcn: mir_types::Name::from(fqcn.as_ref()),
349 type_params: mir_types::union::empty_type_params(),
350 });
351 Ok(crate::HoverInfo {
352 ty,
353 docstring: None,
354 definition: class.location().cloned(),
355 })
356 }
357 crate::Name::Property { class, name } => {
358 let here = crate::db::Fqcn::from_str(&db, class.as_ref());
359 let (_, p) = crate::db::find_property_in_chain(&db, here, name)
360 .ok_or(crate::SymbolLookupError::NotFound)?;
361 let ty = p.ty.as_deref().cloned().unwrap_or_else(Type::mixed);
362 Ok(crate::HoverInfo {
363 ty,
364 docstring: None,
365 definition: p.location.clone(),
366 })
367 }
368 crate::Name::ClassConstant { class, name } => {
369 let here = crate::db::Fqcn::from_str(&db, class.as_ref());
370 let (_, c) = crate::db::find_class_constant_in_chain(&db, here, name)
371 .ok_or(crate::SymbolLookupError::NotFound)?;
372 Ok(crate::HoverInfo {
373 ty: c.ty.clone(),
374 docstring: None,
375 definition: c.location.clone(),
376 })
377 }
378 crate::Name::GlobalConstant(fqn) => {
379 let here = crate::db::Fqcn::from_str(&db, fqn.as_ref());
380 let ty = crate::db::find_global_constant(&db, here)
381 .ok_or(crate::SymbolLookupError::NotFound)?;
382 Ok(crate::HoverInfo {
383 ty: (*ty).clone(),
384 docstring: None,
385 definition: None,
386 })
387 }
388 }
389 }
390
391 /// Raw reference locations indexed by string symbol key, kept for tests
392 /// that use the legacy stringly-typed API. Prefer [`Self::indexed_references_to`]
393 /// with a typed [`crate::Name`].
394 #[doc(hidden)]
395 pub fn reference_locations(&self, symbol: &str) -> Vec<(Arc<str>, u32, u16, u16)> {
396 use crate::db::MirDatabase;
397 let db = self.snapshot_db();
398 db.reference_locations(symbol)
399 }
400
401 /// Files declaring transitive subclasses of `class_fqn`, backed by the
402 /// maintained subtype index (see [`Self::indexed_subtype_classes`]).
403 /// Excludes `class_fqn`'s own declaring file — the caller adds it.
404 ///
405 /// Lets a reference-search caller scope a `protected` member to its class
406 /// hierarchy without reconstructing that hierarchy from declaration text:
407 /// subclasses are matched by resolved FQCN, so `extends \Ns\Base` and
408 /// aliased `use` forms are all found. Read-only from the caller's
409 /// perspective; may trigger an on-demand commit of stale/uncommitted
410 /// candidates' class edges (same self-heal `indexed_subtype_classes` uses).
411 pub fn subtype_files(&mut self, class_fqn: &str) -> Vec<Arc<str>> {
412 self.settle_workspace_index();
413 let files = self.snapshot_db().source_file_paths();
414 let mut out: Vec<Arc<str>> = self
415 .indexed_subtype_classes(class_fqn, &files, false)
416 .into_iter()
417 .map(|s| s.file)
418 .collect();
419 out.sort();
420 out.dedup();
421 out
422 }
423
424 /// Compatibility wrapper for callers that only want `use` import items.
425 /// Delegates to [`Self::indexed_references_to`] so import references share
426 /// the same freshness/self-heal path, scope filtering, and memoization
427 /// boundary as every other reference query.
428 pub fn indexed_use_import_locations(
429 &mut self,
430 symbol: &crate::Name,
431 files: &[Arc<str>],
432 ) -> Vec<(Arc<str>, crate::Range)> {
433 self.indexed_references_to(
434 symbol,
435 files,
436 false,
437 crate::ReferenceIncludes::UseImports,
438 &|| false,
439 )
440 .unwrap_or_default()
441 }
442
443 /// Inverted-index find-references: posting-list lookup plus an on-demand
444 /// freshness/completeness pass over `files` (the host's candidate scope
445 /// — passing the whole workspace is fine; see the gate below).
446 ///
447 /// A candidate whose postings were committed from its current input text
448 /// (Arc identity) is answered from the index with no salsa work at all.
449 /// Stale or never-committed candidates are analyzed via the memoized
450 /// `analyze_file` query and committed, so each file pays that cost once
451 /// per text change — after a background warm sweep the steady state is a
452 /// pure lookup, O(results) instead of O(candidates). Never-committed
453 /// candidates are additionally gated on their raw text mentioning the
454 /// symbol's name (whole-identifier, ASCII-case-insensitive), so hosts
455 /// need no text prefilter of their own — and must not use one, since a
456 /// host-side filter cannot know these matching semantics.
457 ///
458 /// Results are filtered to `files` (the host controls scope — e.g.
459 /// workspace files only, excluding stubs/vendor). With
460 /// `include_declaration`, the symbol's declaration name span is appended
461 /// when it lies inside the scope.
462 ///
463 /// Returned ranges use mir's native coordinates: 1-based lines and
464 /// 0-based Unicode code-point columns (UTF-32/LSP `positionEncoding`
465 /// `"utf-32"`), not UTF-8 byte offsets or UTF-16 code units.
466 ///
467 /// `should_cancel` is polled at phase boundaries and between
468 /// cancellation retries; `true` aborts with `None`.
469 ///
470 /// Memoized per `(symbol, files, include_declaration, includes,
471 /// text_revision)` — see [`RefQueryCacheKey`]. The freshness scan below
472 /// still costs O(candidates) even when every candidate is already
473 /// committed (it has to check), so a caller re-running the same query
474 /// against unchanged state (e.g. a host recomputing reference counts for
475 /// a code-lens refresh) would otherwise re-pay that scan on every call;
476 /// this makes the repeat a single hashmap lookup instead.
477 pub fn indexed_references_to(
478 &mut self,
479 symbol: &crate::Name,
480 files: &[Arc<str>],
481 include_declaration: bool,
482 includes: crate::ReferenceIncludes,
483 should_cancel: &(dyn Fn() -> bool + Sync),
484 ) -> Option<Vec<(Arc<str>, crate::Range)>> {
485 // No `should_cancel()` check here: a cache hit does no analysis
486 // work, so there's nothing to cancel, and calling it would consume
487 // one of the caller's cancellation-probe invocations before the
488 // uncached path's own checks ever run (some callers, e.g.
489 // `session_sweep_persists_postings_for_next_launch`, count these to
490 // prove a warm query needed no re-analysis).
491 let cache_key = RefQueryCacheKey {
492 symbol: symbol.codebase_key(),
493 include_declaration,
494 includes,
495 generation: self.query_cache_generation(),
496 files_hash: hash_files(files),
497 };
498 if let Some(cached) = self.ref_query_cache.read().get(&cache_key) {
499 self.ref_query_cache_hits
500 .fetch_add(1, std::sync::atomic::Ordering::Relaxed);
501 return Some((**cached).clone());
502 }
503 let result = self.indexed_references_to_uncached(
504 symbol,
505 files,
506 include_declaration,
507 includes,
508 should_cancel,
509 )?;
510 if self.query_cache_generation() != cache_key.generation {
511 // The generation moved while computing (an edit, or a defs
512 // commit growing the subtype index — possibly this query's own).
513 // The key can never be looked up again, so don't cache it; the
514 // next identical query recomputes once against settled state
515 // and caches then.
516 return Some(result);
517 }
518 let mut cache = self.ref_query_cache.write();
519 if !cache.advance_to(cache_key.generation, || {
520 self.ref_query_cache_locations
521 .store(0, std::sync::atomic::Ordering::Relaxed);
522 }) {
523 return Some(result);
524 }
525 let new_len = result.len();
526 let prior = self
527 .ref_query_cache_locations
528 .fetch_add(new_len, std::sync::atomic::Ordering::Relaxed);
529 if prior + new_len > REF_QUERY_CACHE_LOCATION_CAP {
530 cache.map.clear();
531 self.ref_query_cache_locations
532 .store(new_len, std::sync::atomic::Ordering::Relaxed);
533 }
534 cache.map.insert(cache_key, Arc::new(result.clone()));
535 Some(result)
536 }
537
538 /// Uncached implementation of [`Self::indexed_references_to`]. Callers
539 /// should use the memoizing wrapper; this is split out only so the cache
540 /// check/populate logic doesn't have to interleave with the retry loops
541 /// below.
542 fn indexed_references_to_uncached(
543 &mut self,
544 symbol: &crate::Name,
545 files: &[Arc<str>],
546 include_declaration: bool,
547 includes: crate::ReferenceIncludes,
548 should_cancel: &(dyn Fn() -> bool + Sync),
549 ) -> Option<Vec<(Arc<str>, crate::Range)>> {
550 use std::panic::AssertUnwindSafe;
551
552 let key = symbol.codebase_key();
553
554 // Freshness pass: candidates whose postings are not exact for their
555 // current text. Files not registered as `SourceFile` inputs are
556 // skipped. Never-committed files — no commit mark, hence no postings
557 // at all (every mark drop accompanies a posting clear) — are further
558 // gated on their text mentioning the symbol's name: such a file can
559 // neither hold stale postings nor produce new ones, so a cold query
560 // on a common name skips the bulk of the workspace instead of
561 // analyzing it. A LIVE-analyzed file stale only by generation
562 // (unresolved-name commit, text unchanged since) gets the same gate:
563 // the current text is exactly what this session's own analysis
564 // already scanned, so a needle miss is just as conclusive as for a
565 // never-committed file. Everything else — a genuinely edited file, or
566 // a commit seeded by an unverified disk-cache replay
567 // (`warm_start_files`, never scanned by this session) — re-analyzes
568 // unconditionally: an edited file's old postings are for different
569 // text, and a replayed commit's postings are only as trustworthy as
570 // the cache entry, so neither can be cleared by a textual gate alone.
571 // Same discipline as `commit_defs_for_matching` on the defs index.
572 //
573 let gate = self.reference_gate(symbol);
574 // The whole gate — identifier needles and raw call tokens alike —
575 // answers from the persistent mention index: every needle is admitted
576 // (a declared class-like short name is already in the universe;
577 // member/function names and the two `__construct` call tokens enter
578 // verbatim), so a recorded mention set answers with lookups and only
579 // a never-scanned or since-edited file pays one scan against the
580 // whole universe, recorded for every later consumer (including the
581 // subtype-BFS gate). A needle new to the universe epoch-invalidates
582 // older recordings for itself only — the first query on it rescans
583 // uncovered files once, the same cost the per-query scan paid every
584 // time before.
585 let has_needles = !gate.idents.is_empty() || !gate.raw.is_empty();
586 let (mention_queries, mention_scanner) = if has_needles {
587 let guard = &self.db.salsa;
588 guard.add_literal_mention_names(gate.idents.iter().map(|s| s.as_str()));
589 guard.add_raw_mention_needles(gate.raw.iter().map(|s| s.as_str()));
590 let queries: Vec<_> = gate
591 .idents
592 .iter()
593 .chain(gate.raw.iter())
594 .filter_map(|s| guard.prepare_class_mention_query(s))
595 .collect();
596 (queries, guard.class_mention_scanner())
597 } else {
598 (Vec::new(), None)
599 };
600 // Admission guarantees a query per needle and a non-empty universe;
601 // anything else is defensive — the gate then admits every candidate
602 // (analyze rather than skip, the conservative direction).
603 let gate_complete = mention_queries.len() == gate.idents.len() + gate.raw.len()
604 && mention_scanner.is_some();
605 let committed_any: rustc_hash::FxHashSet<Arc<str>> =
606 self.ref_committed_keys().into_iter().collect();
607 type MentionScanRec = (Arc<str>, Arc<str>, Box<[mir_types::Name]>);
608 let (stale, scanned): (Vec<Arc<str>>, Vec<MentionScanRec>) = loop {
609 if should_cancel() {
610 return None;
611 }
612 let attempt = salsa::Cancelled::catch(AssertUnwindSafe(|| {
613 let current_gen = self.index_generation();
614 let db_main = self.snapshot_db();
615 let mut stale = Vec::new();
616 let mut scanned = Vec::new();
617 // This pass is intentionally serial. Each request already runs
618 // on its caller thread; putting every concurrent request back
619 // onto the shared Rayon pool lets an index batch monopolize the
620 // workers while readers wait without polling `should_cancel`.
621 // The per-file work is mostly cache lookups, and keeping it on
622 // the caller thread makes cancellation responsive under write
623 // contention instead of turning a short LSP deadline into an
624 // unbounded pool wait.
625 for f in files {
626 let Some(sf) = db_main.lookup_source_file(f.as_ref()) else {
627 continue;
628 };
629 let text = sf.text(&db_main as &dyn MirDatabase);
630 if self.is_ref_committed(f.as_ref(), text, current_gen) {
631 continue;
632 }
633 if committed_any.contains(f.as_ref())
634 && !self.ref_commit_stale_by_generation_only(f.as_ref(), text)
635 {
636 stale.push(f.clone());
637 continue;
638 }
639 if has_needles && gate_complete {
640 // Any needle answering `true` admits the file; an
641 // unanswerable one forces the single recorded scan,
642 // which settles every needle at once.
643 let mut answer = Some(false);
644 for q in &mention_queries {
645 match db_main.class_mention_answer(f.as_ref(), q, text) {
646 Some(true) => {
647 answer = Some(true);
648 break;
649 }
650 Some(false) => {}
651 None => answer = None,
652 }
653 }
654 match answer {
655 Some(true) => {}
656 Some(false) => continue,
657 None => {
658 let scanner = mention_scanner
659 .as_ref()
660 .expect("gate_complete implies a scanner");
661 let names = scanner.scan(text);
662 let hit = mention_queries
663 .iter()
664 .any(|q| names.binary_search(&q.name).is_ok());
665 scanned.push((f.clone(), text.clone(), names));
666 if hit {
667 stale.push(f.clone());
668 }
669 continue;
670 }
671 }
672 }
673 stale.push(f.clone());
674 }
675 Some((stale, scanned))
676 }));
677 match attempt {
678 Ok(Some(v)) => break v,
679 Ok(None) => return None,
680 Err(_) if should_cancel() => return None,
681 Err(_) => {}
682 }
683 };
684
685 // Record the fallback scans regardless of how the query proceeds:
686 // each is a complete, current mention set for its file.
687 if let Some(scanner) = &mention_scanner {
688 if !scanned.is_empty() {
689 let guard = &self.db.salsa;
690 for (file, text, names) in scanned {
691 guard.set_file_class_mentions(&file, &text, scanner.epoch(), names);
692 }
693 }
694 }
695
696 if !stale.is_empty() {
697 // Replayed postings which are already current answer directly
698 // below. Do not make that read-only path wait for (or consume the
699 // cancellation budget on) unrelated pending symbol-index work:
700 // warm start has already seeded the index needed to interpret its
701 // postings. A stale candidate, on the other hand, is about to
702 // analyze against the workspace, so reconcile first and retain
703 // the cancellable behavior for that potentially expensive work.
704 if !self.settle_workspace_index_cancellable(should_cancel) {
705 return None;
706 }
707 // Phase 1 (serial, no live snapshot held): warm up stale
708 // candidates, retrying a cancelled file in place rather than
709 // restarting the whole query. Parallel variants deadlocked under
710 // `concurrent_reference_cancel`; keep it serial. The bump scope
711 // closes before Phase 2 reads `index_generation`.
712 {
713 let mut session = self.defer_revision_bumps();
714 for path in &stale {
715 loop {
716 if should_cancel() {
717 return None;
718 }
719 match salsa::Cancelled::catch(AssertUnwindSafe(|| {
720 session.prepare_file_for_analysis(path)
721 })) {
722 Ok(()) => break,
723 Err(_) if should_cancel() => return None,
724 Err(_) => {}
725 }
726 }
727 }
728 }
729
730 // Phase 2 (parallel, pure) under a cancellation retry loop, then
731 // a serial commit into both inverted indexes.
732 let (commit_gen, analyzed) = loop {
733 if should_cancel() {
734 return None;
735 }
736 // Generation before the snapshot: a file add racing the
737 // analysis leaves these commits stale (self-healing on the
738 // next query), never wrongly fresh.
739 let gen = self.index_generation();
740 let attempt = salsa::Cancelled::catch(AssertUnwindSafe(|| {
741 // Freeze on the pass-scoped snapshot (borrow-only symbol
742 // lookups + pass-shared subtype cache): all lazy-loading
743 // finished in Phase 1, and a concurrent index write
744 // cancels this attempt, so the frozen view is never
745 // stale. Same discipline as the batch body pass.
746 let mut db_main = self.snapshot_db();
747 db_main.freeze_workspace_index();
748 stale
749 .iter()
750 .filter_map(|path| {
751 let sf = db_main.lookup_source_file(path.as_ref())?;
752 let text = sf.text(&db_main as &dyn MirDatabase).clone();
753 let out =
754 crate::db::analyze_file(&db_main as &dyn MirDatabase, sf).clone();
755 let defs = crate::db::collect_file_definitions(
756 &db_main as &dyn MirDatabase,
757 sf,
758 );
759 let entries = crate::db::subtype_index::entries_from_slice(&defs.slice);
760 // Stage the disk-cache write only when the commit
761 // below will rewrite postings (see the sweep in
762 // `reanalyze_file_set` for the cost rationale).
763 let put = if self.ref_commit_is_current(path.as_ref(), &text, &out) {
764 None
765 } else {
766 self.stage_ref_cache_put(
767 &db_main as &dyn MirDatabase,
768 sf,
769 path.as_ref(),
770 &text,
771 &out,
772 )
773 };
774 // Mention scan piggybacks on the analysis pass
775 // (pure; committed serially below), skipped when
776 // the file already holds a current scan.
777 let mentions = mention_scanner.as_ref().and_then(|s| {
778 (!db_main.class_mentions_current(path.as_ref(), &text, s.epoch()))
779 .then(|| s.scan(&text))
780 });
781 Some((path.clone(), text, out, entries, put, mentions))
782 })
783 .collect::<Vec<_>>()
784 }));
785 match attempt {
786 Ok(v) => break (gen, v),
787 Err(_) if should_cancel() => return None,
788 Err(_) => {}
789 }
790 };
791 let mut analyzed = analyzed;
792 let guard = &self.db.salsa;
793 for (file, text, out, entries, put, mentions) in analyzed.iter_mut() {
794 // Pointer-identical memo ⇒ identical postings: skip the
795 // index rewrite and only re-stamp the freshness mark.
796 if !self.ref_commit_is_current(file.as_ref(), text, out) {
797 let file_no = guard.locked_ref_index().intern_path(file);
798 guard.set_file_reference_locations(file_no, out.ref_locs.to_vec());
799 }
800 if let (Some(s), Some(m)) = (&mention_scanner, mentions.take()) {
801 guard.set_file_class_mentions(file, text, s.epoch(), m);
802 }
803 if let Some(put) = put.take() {
804 self.apply_ref_cache_put(file.as_ref(), out, put);
805 }
806 self.mark_ref_committed(
807 file,
808 text,
809 Some(out),
810 commit_gen,
811 !out.has_unresolved_names(),
812 );
813 if !self.is_defs_committed(file.as_ref(), text) {
814 let file_no = guard.locked_ref_index().intern_path(file);
815 guard.set_file_class_edges(file_no, entries.clone());
816 self.mark_defs_committed(file, text);
817 }
818 }
819 }
820
821 // Posting lookup, filtered to the candidate scope.
822 //
823 // Member symbols resolve against the queried class plus its hierarchy
824 // (mir records member refs under the *declaring* class, so a query on
825 // an interface method must include implementor keys and vice versa).
826 // Name-only fallback postings — receivers whose type couldn't be
827 // resolved — are consulted only when the typed keys produce nothing,
828 // mirroring the pre-index two-tier behavior: exact results when
829 // resolution succeeds, by-name matches when nothing resolves.
830 // `__construct` stays exact: `new Sub()` invokes `Sub::__construct`
831 // even when only a parent declares one, so hierarchy fan-out would
832 // wrongly return subtype instantiation sites for a parent query.
833 let hierarchy: Vec<String> = match symbol {
834 crate::Name::Method { class, name } => {
835 if name.as_ref() == "__construct" || class.is_empty() {
836 if class.is_empty() {
837 Vec::new()
838 } else {
839 vec![class.trim_start_matches('\\').to_string()]
840 }
841 } else {
842 self.member_hierarchy_classes(class.as_ref())
843 }
844 }
845 crate::Name::Property { class, .. } | crate::Name::ClassConstant { class, .. } => {
846 if class.is_empty() {
847 Vec::new()
848 } else {
849 self.member_hierarchy_classes(class.as_ref())
850 }
851 }
852 _ => Vec::new(),
853 };
854 let scope: rustc_hash::FxHashSet<&str> = files.iter().map(|f| f.as_ref()).collect();
855 let read_symbol_key = |symbol_key: &str| -> Vec<(Arc<str>, crate::Range)> {
856 let guard = &self.db.salsa;
857 guard
858 .reference_locations(symbol_key)
859 .into_iter()
860 .filter(|(file, ..)| scope.contains(file.as_ref()))
861 .map(|(file, line, col_start, col_end)| {
862 (file, span_range(line, col_start as u32, col_end as u32))
863 })
864 .collect()
865 };
866 let mut scratch_key = String::new();
867 let mut read_composed_key = |prefix: &str, middle: &str, separator: &str, suffix: &str| {
868 scratch_key.clear();
869 scratch_key.reserve(prefix.len() + middle.len() + separator.len() + suffix.len());
870 scratch_key.push_str(prefix);
871 scratch_key.push_str(middle);
872 scratch_key.push_str(separator);
873 scratch_key.push_str(suffix);
874 read_symbol_key(&scratch_key)
875 };
876 let mut out: Vec<(Arc<str>, crate::Range)> = Vec::new();
877 let mut append_plain = || match symbol {
878 crate::Name::Method { name, .. } => {
879 for class in &hierarchy {
880 out.extend(read_composed_key(
881 "meth:",
882 class.as_ref(),
883 "::",
884 name.as_ref(),
885 ));
886 }
887 }
888 crate::Name::Property { name, .. } => {
889 for class in &hierarchy {
890 out.extend(read_composed_key(
891 "prop:",
892 class.as_ref(),
893 "::",
894 name.as_ref(),
895 ));
896 }
897 }
898 crate::Name::ClassConstant { name, .. } => {
899 for class in &hierarchy {
900 out.extend(read_composed_key(
901 "cnst:",
902 class.as_ref(),
903 "::",
904 name.as_ref(),
905 ));
906 }
907 }
908 _ => out.extend(read_symbol_key(key.as_str())),
909 };
910 match includes {
911 crate::ReferenceIncludes::Plain => append_plain(),
912 crate::ReferenceIncludes::UseImports => {
913 out.extend(read_composed_key("use:", key.as_str(), "", ""));
914 }
915 crate::ReferenceIncludes::PlainAndUseImports => {
916 append_plain();
917 out.extend(read_composed_key("use:", key.as_str(), "", ""));
918 }
919 }
920 if matches!(includes, crate::ReferenceIncludes::Plain) && out.is_empty() {
921 match symbol {
922 crate::Name::Method { name, .. } => {
923 out = read_composed_key("methname:", name.as_ref(), "", "");
924 }
925 crate::Name::Property { name, .. } => {
926 out = read_composed_key("propname:", name.as_ref(), "", "");
927 }
928 _ => {}
929 }
930 }
931 out.sort_by(|a, b| {
932 a.0.cmp(&b.0)
933 .then(a.1.start.line.cmp(&b.1.start.line))
934 .then(a.1.start.column.cmp(&b.1.start.column))
935 });
936 out.dedup_by(|a, b| a.0 == b.0 && a.1 == b.1);
937
938 if include_declaration {
939 // Declaration lookup runs salsa queries (and may lazy-load); a
940 // concurrent write cancels it — declarations are then simply
941 // omitted rather than failing the whole request.
942 let decls: Vec<(Arc<str>, crate::Range)> = match symbol {
943 crate::Name::Method { class, .. }
944 | crate::Name::Property { class, .. }
945 | crate::Name::ClassConstant { class, .. } => {
946 if class.is_empty() {
947 // Unknown owner: declarations by name, recorded as
948 // `methdecl:`/`propdecl:`/`cnstdecl:` postings during
949 // class/trait/interface/enum analysis.
950 match symbol {
951 crate::Name::Method { name, .. } => {
952 read_composed_key("methdecl:", name.as_ref(), "", "")
953 }
954 crate::Name::Property { name, .. } => {
955 read_composed_key("propdecl:", name.as_ref(), "", "")
956 }
957 crate::Name::ClassConstant { name, .. } => {
958 read_composed_key("cnstdecl:", name.as_ref(), "", "")
959 }
960 _ => Vec::new(),
961 }
962 } else {
963 salsa::Cancelled::catch(AssertUnwindSafe(|| {
964 self.member_decl_sites(&hierarchy, symbol)
965 }))
966 .unwrap_or_default()
967 }
968 }
969 _ => salsa::Cancelled::catch(AssertUnwindSafe(|| {
970 self.declaration_name_range(symbol).into_iter().collect()
971 }))
972 .unwrap_or_default(),
973 };
974 for (file, range) in decls {
975 if scope.contains(file.as_ref())
976 && !out.iter().any(|(f, r)| *f == file && *r == range)
977 {
978 out.push((file, range));
979 }
980 }
981 }
982 Some(out)
983 }
984
985 /// The queried class plus every class its members' references could be
986 /// keyed under: resolved ancestors (a call on a subtype instance records
987 /// the declaring ancestor) and transitive subtypes including trait users
988 /// (a call on a subtype that overrides records the subtype). Display-form
989 /// FQCNs, deduplicated case-insensitively.
990 fn member_hierarchy_classes(&self, class_fqn: &str) -> Vec<String> {
991 use std::panic::AssertUnwindSafe;
992 let target = class_fqn.trim_start_matches('\\').to_string();
993 let mut out: Vec<String> = vec![target.clone()];
994 let ancestors = salsa::Cancelled::catch(AssertUnwindSafe(|| {
995 let db = self.snapshot_db();
996 let here = crate::db::Fqcn::from_str(&db, &target);
997 crate::db::class_ancestors_by_fqcn(&db, here)
998 .iter()
999 .skip(1)
1000 .map(|a| a.trim_start_matches('\\').to_string())
1001 .collect::<Vec<_>>()
1002 }))
1003 .unwrap_or_default();
1004 out.extend(ancestors);
1005 let subs = {
1006 let guard = &self.db.salsa;
1007 guard.subtype_sites_of(&target, true)
1008 };
1009 out.extend(
1010 subs.into_iter()
1011 .map(|s| s.fqcn.trim_start_matches('\\').to_string()),
1012 );
1013 let mut seen: rustc_hash::FxHashSet<String> = rustc_hash::FxHashSet::default();
1014 out.retain(|c| seen.insert(c.to_ascii_lowercase()));
1015 out
1016 }
1017
1018 /// Own-member declaration sites for `symbol` across `classes`: each class
1019 /// that itself declares the member (not inherited) contributes its name
1020 /// token. Kind-specific lookups — a class often declares a property and a
1021 /// method with the same short name, and `member_location` can't tell them
1022 /// apart.
1023 fn member_decl_sites(
1024 &self,
1025 classes: &[String],
1026 symbol: &crate::Name,
1027 ) -> Vec<(Arc<str>, crate::Range)> {
1028 let mut out: Vec<(Arc<str>, crate::Range)> = Vec::new();
1029 let db = self.snapshot_db();
1030 for class in classes {
1031 let here = crate::db::Fqcn::from_str(&db, class);
1032 let (loc, needle) = match symbol {
1033 crate::Name::Method { name, .. } => {
1034 let Some(m) = crate::db::find_method_in_class(&db, here, name) else {
1035 continue;
1036 };
1037 (m.location.clone(), name.to_string())
1038 }
1039 crate::Name::Property { name, .. } => {
1040 let Some(p) = crate::db::find_property_in_class(&db, here, name) else {
1041 continue;
1042 };
1043 (p.location.clone(), name.to_string())
1044 }
1045 crate::Name::ClassConstant { name, .. } => {
1046 let Some(c) = crate::db::find_class_constant_in_class(&db, here, name) else {
1047 continue;
1048 };
1049 (c.location.clone(), name.to_string())
1050 }
1051 _ => continue,
1052 };
1053 let Some(loc) = loc else { continue };
1054 let range = self.refine_location_to_name(&loc, &needle);
1055 out.push((loc.file.clone(), range));
1056 }
1057 out
1058 }
1059
1060 /// The symbol's declaration site, narrowed from the collector's
1061 /// whole-declaration span to the declared name's own token (matching the
1062 /// span shape of recorded references).
1063 pub fn declaration_name_range(
1064 &mut self,
1065 symbol: &crate::Name,
1066 ) -> Option<(Arc<str>, crate::Range)> {
1067 if let crate::Name::GlobalConstant(fqn) = symbol {
1068 return self.global_constant_decl_range(fqn);
1069 }
1070 let loc = self.definition_of(symbol).ok()?;
1071 let short = match symbol {
1072 crate::Name::Class(f) | crate::Name::Function(f) | crate::Name::GlobalConstant(f) => {
1073 crate::db::subtype_index::short_name_of(f)
1074 }
1075 crate::Name::Method { name, .. }
1076 | crate::Name::Property { name, .. }
1077 | crate::Name::ClassConstant { name, .. } => name.as_ref(),
1078 };
1079 // Property declarations carry a `$` sigil in source, but reference
1080 // ranges cover the bare name; the word-boundary search below lands on
1081 // the name right after the sigil.
1082 let file = loc.file.clone();
1083 let range = self.refine_location_to_name(&loc, short);
1084 Some((file, range))
1085 }
1086
1087 /// Narrow a whole-declaration [`mir_types::Location`] to the first
1088 /// word-boundary occurrence of `needle` inside its line span. Falls back
1089 /// to the location's own coordinates when the text is unavailable or the
1090 /// name doesn't appear (e.g. stub-only declarations).
1091 fn refine_location_to_name(&self, loc: &mir_types::Location, needle: &str) -> crate::Range {
1092 let fallback = span_range(loc.line, loc.col_start as u32, loc.col_end as u32);
1093 let text = {
1094 let db = self.snapshot_db();
1095 db.lookup_source_file(loc.file.as_ref())
1096 .map(|sf| sf.text(&db as &dyn MirDatabase).clone())
1097 };
1098 let Some(text) = text else {
1099 return fallback;
1100 };
1101 let needle_chars = needle.chars().count() as u32;
1102 let first_line = loc.line.saturating_sub(1) as usize;
1103 // Exact-case first: PHP property/constant names are case-sensitive
1104 // and an early case-insensitive hit can land on an unrelated token
1105 // (a type hint sharing the name). Case-insensitive second, for
1106 // method/class needles that arrive lowercase-normalized.
1107 for case_insensitive in [false, true] {
1108 for (idx, line_text) in text.lines().enumerate().skip(first_line) {
1109 let line_no = idx as u32 + 1;
1110 if line_no > loc.line_end {
1111 break;
1112 }
1113 let min_col = if line_no == loc.line {
1114 loc.col_start as usize
1115 } else {
1116 0
1117 };
1118 if let Some(col) = identifier_char_col(line_text, needle, min_col, case_insensitive)
1119 {
1120 return span_range(line_no, col, col + needle_chars);
1121 }
1122 }
1123 }
1124 fallback
1125 }
1126
1127 /// Transitive subtypes of `class_fqn` (classes/interfaces/enums whose
1128 /// resolved ancestor chain reaches it), answered from the maintained
1129 /// subtype edge index.
1130 ///
1131 /// `files` is the host's candidate scope for the on-demand completeness
1132 /// pass: per BFS round, not-yet-committed files whose text mentions a
1133 /// frontier short name get their definitions committed, so results are
1134 /// complete even before a background sweep has covered the workspace.
1135 /// That short-name gate is only candidate discovery; subtype identity is
1136 /// still resolved from the edge index below using the exact canonical
1137 /// FQCN.
1138 /// Committed files answer from the index with no parsing at all.
1139 ///
1140 /// `include_trait_users` also counts `use Trait;` composition as a
1141 /// subtype edge (visibility-scoping semantics); leave it off for
1142 /// goto-implementation semantics (extends/implements only).
1143 ///
1144 /// Memoized per `(class_fqn, include_trait_users, files, text_revision)`
1145 /// — same shape and rationale as [`Self::indexed_references_to`]'s
1146 /// cache: `commit_defs_for_matching`'s freshness pass costs O(candidates)
1147 /// on every call regardless of outcome, so a repeat query (e.g. a host
1148 /// resolving a protected/static method's reference scope on every
1149 /// code-lens refresh) would otherwise re-pay it every time.
1150 pub fn indexed_subtype_classes(
1151 &mut self,
1152 class_fqn: &str,
1153 files: &[Arc<str>],
1154 include_trait_users: bool,
1155 ) -> Vec<SubtypeClassSite> {
1156 let cache_key = SubtypeQueryCacheKey {
1157 class_fqn: class_fqn.trim_start_matches('\\').to_ascii_lowercase(),
1158 include_trait_users,
1159 generation: self.query_cache_generation(),
1160 files_hash: hash_files(files),
1161 };
1162 if let Some(cached) = self.subtype_query_cache.read().get(&cache_key) {
1163 self.subtype_query_cache_hits
1164 .fetch_add(1, std::sync::atomic::Ordering::Relaxed);
1165 return (**cached).clone();
1166 }
1167 let result = self.indexed_subtype_classes_uncached(class_fqn, files, include_trait_users);
1168 if self.query_cache_generation() != cache_key.generation {
1169 // The generation moved while computing (an edit, or a defs
1170 // commit growing the subtype index — possibly this query's own).
1171 // The key can never be looked up again, so don't cache it; the
1172 // next identical query recomputes once against settled state
1173 // and caches then.
1174 return result;
1175 }
1176 let mut cache = self.subtype_query_cache.write();
1177 if !cache.advance_to(cache_key.generation, || {
1178 self.subtype_query_cache_sites
1179 .store(0, std::sync::atomic::Ordering::Relaxed);
1180 }) {
1181 return result;
1182 }
1183 let new_len = result.len();
1184 let prior = self
1185 .subtype_query_cache_sites
1186 .fetch_add(new_len, std::sync::atomic::Ordering::Relaxed);
1187 if prior + new_len > SUBTYPE_QUERY_CACHE_SITE_CAP {
1188 cache.map.clear();
1189 self.subtype_query_cache_sites
1190 .store(new_len, std::sync::atomic::Ordering::Relaxed);
1191 }
1192 cache.map.insert(cache_key, Arc::new(result.clone()));
1193 result
1194 }
1195
1196 /// Uncached implementation of [`Self::indexed_subtype_classes`]. Callers
1197 /// should use the memoizing wrapper.
1198 fn indexed_subtype_classes_uncached(
1199 &mut self,
1200 class_fqn: &str,
1201 files: &[Arc<str>],
1202 include_trait_users: bool,
1203 ) -> Vec<SubtypeClassSite> {
1204 self.settle_workspace_index();
1205 let mut scanned: rustc_hash::FxHashSet<String> = rustc_hash::FxHashSet::default();
1206 let mut pending: Vec<String> = vec![class_fqn.trim_start_matches('\\').to_string()];
1207 let mut sites: Vec<crate::db::SubtypeSite> = Vec::new();
1208 while !pending.is_empty() {
1209 // Use short names only to discover stale/uncommitted files worth
1210 // collecting. The query result itself comes from the subtype
1211 // edge index (`subtype_sites_of`) below.
1212 let needles: Vec<String> = pending
1213 .drain(..)
1214 .filter(|f| scanned.insert(f.clone()))
1215 .map(|f| crate::db::subtype_index::short_name_of(&f).to_string())
1216 .collect();
1217 if !needles.is_empty() {
1218 self.commit_defs_for_matching(files, &needles);
1219 }
1220 sites = {
1221 let guard = &self.db.salsa;
1222 guard.subtype_sites_of(class_fqn, include_trait_users)
1223 };
1224 pending = sites
1225 .iter()
1226 .map(|s| s.fqcn.trim_start_matches('\\').to_string())
1227 .filter(|f| !scanned.contains(f))
1228 .collect();
1229 }
1230 let mut out: Vec<SubtypeClassSite> = sites
1231 .into_iter()
1232 .filter_map(|s| {
1233 let loc = s.location.as_ref()?;
1234 let short = crate::db::subtype_index::short_name_of(&s.fqcn).to_string();
1235 let range = self.refine_location_to_name(loc, &short);
1236 Some(SubtypeClassSite {
1237 fqcn: s.fqcn,
1238 kind: s.kind,
1239 is_abstract: s.is_abstract,
1240 file: s.file,
1241 range,
1242 })
1243 })
1244 .collect();
1245 // Anonymous classes never reach the definition collector; their
1246 // `new class implements X {}` sites are recorded under the resolved
1247 // canonical parent FQCN during body analysis.
1248 let root_lc = class_fqn.trim_start_matches('\\').to_ascii_lowercase();
1249 let scope: rustc_hash::FxHashSet<&str> = files.iter().map(|f| f.as_ref()).collect();
1250 let anon: Vec<(Arc<str>, u32, u16, u16)> = {
1251 let guard = &self.db.salsa;
1252 let mut key = String::with_capacity("impl:".len() + root_lc.len());
1253 key.push_str("impl:");
1254 key.push_str(&root_lc);
1255 guard.reference_locations(&key)
1256 };
1257 for (file, line, cs, ce) in anon {
1258 if !scope.contains(file.as_ref()) {
1259 continue;
1260 }
1261 let range = span_range(line, cs as u32, ce as u32);
1262 if out.iter().any(|s| s.file == file && s.range == range) {
1263 continue;
1264 }
1265 out.push(SubtypeClassSite {
1266 fqcn: Arc::from("class@anonymous"),
1267 kind: crate::db::ClassLikeKind::Class,
1268 is_abstract: false,
1269 file,
1270 range,
1271 });
1272 }
1273 out
1274 }
1275
1276 /// Concrete implementations of `class_fqn::method` across its transitive
1277 /// subtypes: the same-named non-abstract method available to each subtype
1278 /// (its own declaration, or one inherited/composed from a parent, trait,
1279 /// or mixin), as `(subtype fqcn, file, name range)`. Subtypes resolving to
1280 /// the same declaring location collapse to a single entry.
1281 pub fn indexed_method_implementations(
1282 &mut self,
1283 class_fqn: &str,
1284 method: &str,
1285 files: &[Arc<str>],
1286 ) -> Vec<(Arc<str>, Arc<str>, crate::Range)> {
1287 use std::panic::AssertUnwindSafe;
1288 let subs = self.indexed_subtype_classes(class_fqn, files, false);
1289 if subs.is_empty() {
1290 return Vec::new();
1291 }
1292 loop {
1293 let attempt = salsa::Cancelled::catch(AssertUnwindSafe(|| {
1294 let db = self.snapshot_db();
1295 let mut out: Vec<(Arc<str>, Arc<str>, crate::Range)> = Vec::new();
1296 for sub in &subs {
1297 let here = crate::db::Fqcn::from_str(&db, sub.fqcn.as_ref());
1298 let Some((_, m)) = crate::db::find_method_in_chain(&db, here, method) else {
1299 continue;
1300 };
1301 if m.is_abstract {
1302 continue;
1303 }
1304 let Some(loc) = m.location.as_ref() else {
1305 continue;
1306 };
1307 let range = self.refine_location_to_name(loc, method);
1308 out.push((sub.fqcn.clone(), loc.file.clone(), range));
1309 }
1310 out
1311 }));
1312 if let Ok(mut out) = attempt {
1313 out.sort_by(|a, b| a.1.cmp(&b.1).then(a.2.start.line.cmp(&b.2.start.line)));
1314 out.dedup_by(|a, b| a.1 == b.1 && a.2 == b.2);
1315 return out;
1316 }
1317 }
1318 }
1319
1320 /// Commit definitions (class edges + freshness) for every file in `files`
1321 /// that is stale (committed against older text) or that has never been
1322 /// committed and mentions one of `shorts` as a whole identifier.
1323 ///
1324 /// The textual gate answers from the shared per-file mention cache — the
1325 /// same one `indexed_references_to`'s gate populates — so a file scanned
1326 /// by either consumer answers the other with a set lookup instead of an
1327 /// O(text) rescan per BFS round. A file the cache can't answer for is
1328 /// scanned once against the whole name universe and recorded.
1329 fn commit_defs_for_matching(&self, files: &[Arc<str>], shorts: &[String]) {
1330 use std::panic::AssertUnwindSafe;
1331
1332 use rayon::prelude::*;
1333
1334 let committed_any: rustc_hash::FxHashSet<Arc<str>> = {
1335 let guard = self.defs_committed_keys();
1336 guard.into_iter().collect()
1337 };
1338 // Admit the frontier names before preparing, so every needle gets a
1339 // real query (a declared class's short name is already in the
1340 // universe from indexing — admission then changes nothing).
1341 let (queries, mention_scanner) = {
1342 let guard = &self.db.salsa;
1343 guard.add_literal_mention_names(shorts.iter().map(|s| s.as_str()));
1344 let queries: Vec<_> = shorts
1345 .iter()
1346 .filter_map(|s| guard.prepare_class_mention_query(s))
1347 .collect();
1348 (queries, guard.class_mention_scanner())
1349 };
1350 // Admission guarantees a query per needle and a non-empty universe;
1351 // anything else is defensive — the gate then admits every candidate
1352 // (recommit rather than skip, the conservative direction).
1353 let use_mentions = queries.len() == shorts.len() && mention_scanner.is_some();
1354 type Work = (Arc<str>, Arc<str>, Vec<crate::db::SubtypeEntry>);
1355 type MentionScanRec = (Arc<str>, Arc<str>, Box<[mir_types::Name]>);
1356 // Cloned out: the rayon closure can't capture the non-`Sync` session.
1357 let defs_committed = Arc::clone(&self.defs_committed);
1358 let (work, scanned): (Vec<Work>, Vec<MentionScanRec>) = loop {
1359 let attempt = salsa::Cancelled::catch(AssertUnwindSafe(|| {
1360 let db_main = self.snapshot_db();
1361 files
1362 .par_iter()
1363 .map_with(db_main, |db, path| {
1364 let Some(sf) = db.lookup_source_file(path.as_ref()) else {
1365 return (None, None);
1366 };
1367 let text = sf.text(&*db as &dyn MirDatabase).clone();
1368 let already_committed = defs_committed
1369 .read()
1370 .get(path.as_ref())
1371 .is_some_and(|t| Arc::ptr_eq(t, &text));
1372 if already_committed {
1373 return (None, None);
1374 }
1375 // Never-committed files must mention a frontier name;
1376 // stale (previously committed) files recommit
1377 // unconditionally — their classes may have re-parented.
1378 let mut scan_rec: Option<MentionScanRec> = None;
1379 if use_mentions && !committed_any.contains(path.as_ref()) {
1380 let mut answer = Some(false);
1381 for q in &queries {
1382 match db.class_mention_answer(path.as_ref(), q, &text) {
1383 Some(true) => {
1384 answer = Some(true);
1385 break;
1386 }
1387 Some(false) => {}
1388 None => answer = None,
1389 }
1390 }
1391 let hit = match answer {
1392 Some(hit) => hit,
1393 None => {
1394 // Uncoverable entry: one scan answers
1395 // every query and is recorded below.
1396 let scanner = mention_scanner.as_ref().unwrap();
1397 let names = scanner.scan(&text);
1398 let hit = queries
1399 .iter()
1400 .any(|q| names.binary_search(&q.name).is_ok());
1401 scan_rec = Some((path.clone(), text.clone(), names));
1402 hit
1403 }
1404 };
1405 if !hit {
1406 return (None, scan_rec);
1407 }
1408 }
1409 let defs =
1410 crate::db::collect_file_definitions(&*db as &dyn MirDatabase, sf);
1411 let entries = crate::db::subtype_index::entries_from_slice(&defs.slice);
1412 (Some((path.clone(), text, entries)), scan_rec)
1413 })
1414 .collect::<Vec<_>>()
1415 }));
1416 if let Ok(v) = attempt {
1417 let mut work = Vec::new();
1418 let mut scanned = Vec::new();
1419 for (w, rec) in v {
1420 if let Some(w) = w {
1421 work.push(w);
1422 }
1423 if let Some(rec) = rec {
1424 scanned.push(rec);
1425 }
1426 }
1427 break (work, scanned);
1428 }
1429 };
1430 // Record the fallback scans regardless of hit/miss: each is a
1431 // complete, current mention set for its file, so the next round's
1432 // (and the references gate's) checks become set lookups.
1433 if let Some(scanner) = &mention_scanner {
1434 if !scanned.is_empty() {
1435 let guard = &self.db.salsa;
1436 for (file, text, names) in scanned {
1437 guard.set_file_class_mentions(&file, &text, scanner.epoch(), names);
1438 }
1439 }
1440 }
1441 if work.is_empty() {
1442 return;
1443 }
1444 let guard = &self.db.salsa;
1445 for (file, text, entries) in &work {
1446 let file_no = guard.locked_ref_index().intern_path(file);
1447 guard.set_file_class_edges(file_no, entries.clone());
1448 self.mark_defs_committed(file, text);
1449 }
1450 }
1451
1452 /// Declaration name span for a global constant. Constant slices carry no
1453 /// stored location, so this finds the declaring file via the workspace
1454 /// constants index and locates the `const NAME` / `define('NAME'` token
1455 /// textually.
1456 fn global_constant_decl_range(&self, fqn: &str) -> Option<(Arc<str>, crate::Range)> {
1457 use std::panic::AssertUnwindSafe;
1458 let short = crate::db::subtype_index::short_name_of(fqn).to_string();
1459 salsa::Cancelled::catch(AssertUnwindSafe(|| {
1460 let db = self.snapshot_db();
1461 let index = crate::db::workspace_index(&db);
1462 let loc = index.constant_loc(mir_types::Name::from(fqn.trim_start_matches('\\')))?;
1463 let file = loc.file().path(&db).clone();
1464 let sf = db.lookup_source_file(file.as_ref())?;
1465 let text = sf.text(&db as &dyn MirDatabase);
1466 for (idx, line) in text.lines().enumerate() {
1467 let trimmed = line.trim_start();
1468 let is_decl_line = trimmed.starts_with("const ")
1469 || trimmed.contains("define(")
1470 || trimmed.contains("define (");
1471 if !is_decl_line {
1472 continue;
1473 }
1474 if let Some(col) = identifier_char_col(line, &short, 0, false) {
1475 let n = short.chars().count() as u32;
1476 return Some((file, span_range(idx as u32 + 1, col, col + n)));
1477 }
1478 }
1479 None
1480 }))
1481 .ok()
1482 .flatten()
1483 }
1484
1485 /// Class-level issues (inheritance violations, abstract-method gaps, override
1486 /// incompatibilities) for the given set of files.
1487 ///
1488 /// These checks are cross-file by nature and are not emitted by
1489 /// [`crate::FileAnalyzer::analyze`]. Call this after ingesting or
1490 /// re-analyzing a file and its dependents to get the full diagnostic picture.
1491 ///
1492 /// Circular-inheritance checks always run against the full workspace graph
1493 /// regardless of the `files` filter — a cycle is a workspace-wide problem.
1494 pub fn class_issues(&mut self, files: &[Arc<str>]) -> Vec<crate::Issue> {
1495 self.settle_workspace_index();
1496 let db = self.snapshot_db();
1497 let file_set: HashSet<Arc<str>> = files.iter().cloned().collect();
1498 // Read source texts through the snapshot already in hand.
1499 let file_data: Vec<(Arc<str>, Arc<str>)> = files
1500 .iter()
1501 .filter_map(|f| {
1502 let sf = db.lookup_source_file(f)?;
1503 Some((
1504 f.clone(),
1505 sf.text(&db as &dyn crate::db::MirDatabase).clone(),
1506 ))
1507 })
1508 .collect();
1509 crate::class::ClassAnalyzer::with_files(&db, file_set, &file_data).analyze_all()
1510 }
1511
1512 /// Collector-phase issues (e.g. `BackedEnumCaseTypeMismatch`,
1513 /// `InvalidReadonlyPropertyDeclaration`, `InvalidDocblock`, and raw parse
1514 /// errors) for the given files.
1515 ///
1516 /// These are found while building a file's declaration slice
1517 /// ([`crate::db::collect_file_definitions`]), before body analysis or
1518 /// cross-file class checks ever run — neither [`crate::FileAnalyzer::analyze`]
1519 /// nor [`Self::class_issues`] reads them, so a caller merging just those
1520 /// two sources silently drops every collector-time diagnostic. Call this
1521 /// alongside them to get the full picture.
1522 ///
1523 /// A plain snapshot read through [`crate::db::collect_file_definitions`],
1524 /// same as [`Self::document_symbols`] — correct regardless of which path
1525 /// put the file's text into the db (`ingest_file`, `set_file_text`, lazy
1526 /// vendor load) or how many times. The parse-cache fast paths behind that
1527 /// query used to zero out `issues` on any hit, including re-collecting
1528 /// the *same* file after its salsa memo was invalidated; they now
1529 /// preserve (and, for a genuinely different file sharing content,
1530 /// re-point) the originally-computed issues instead.
1531 pub fn collector_issues(&self, files: &[Arc<str>]) -> Vec<crate::Issue> {
1532 let db = self.snapshot_db();
1533 files
1534 .iter()
1535 .filter_map(|f| db.lookup_source_file(f))
1536 .flat_map(|sf| {
1537 crate::db::collect_file_definitions(&db, sf)
1538 .issues
1539 .as_ref()
1540 .clone()
1541 })
1542 .collect()
1543 }
1544
1545 /// All declarations defined in `file` as a **hierarchical tree**.
1546 ///
1547 /// Classes/interfaces/traits/enums are returned with their methods,
1548 /// properties, and constants nested in `children`. Top-level functions
1549 /// and constants are returned with empty `children`.
1550 pub fn document_symbols(&self, file: &str) -> Vec<crate::symbol::DocumentSymbol> {
1551 use crate::symbol::{DeclarationKind, DocumentSymbol};
1552
1553 let db = self.snapshot_db();
1554 let Some(sf) = db.lookup_source_file(file) else {
1555 return Vec::new();
1556 };
1557 let defs = crate::db::collect_file_definitions(&db, sf);
1558 let mut out: Vec<DocumentSymbol> = Vec::new();
1559
1560 let class_children = |methods: &mir_codebase::definitions::MemberMap<
1561 Arc<mir_codebase::definitions::MethodDef>,
1562 >,
1563 props: Option<
1564 &mir_codebase::definitions::MemberMap<mir_codebase::definitions::PropertyDef>,
1565 >,
1566 consts: &mir_codebase::definitions::MemberMap<
1567 mir_codebase::definitions::ConstantDef,
1568 >,
1569 is_enum: bool|
1570 -> Vec<DocumentSymbol> {
1571 let mut out: Vec<DocumentSymbol> = Vec::new();
1572 for (_, m) in methods.iter() {
1573 out.push(DocumentSymbol {
1574 name: m.name.clone(),
1575 kind: DeclarationKind::Method,
1576 location: m.location.clone(),
1577 children: Vec::new(),
1578 });
1579 }
1580 if let Some(props) = props {
1581 for (_, p) in props.iter() {
1582 out.push(DocumentSymbol {
1583 name: p.name.clone(),
1584 kind: DeclarationKind::Property,
1585 location: p.location.clone(),
1586 children: Vec::new(),
1587 });
1588 }
1589 }
1590 let const_kind = if is_enum {
1591 DeclarationKind::EnumCase
1592 } else {
1593 DeclarationKind::Constant
1594 };
1595 for (_, c) in consts.iter() {
1596 out.push(DocumentSymbol {
1597 name: c.name.clone(),
1598 kind: const_kind,
1599 location: c.location.clone(),
1600 children: Vec::new(),
1601 });
1602 }
1603 out
1604 };
1605
1606 for c in defs.slice.classes.iter() {
1607 out.push(DocumentSymbol {
1608 name: c.fqcn.clone(),
1609 kind: DeclarationKind::Class,
1610 location: c.location.clone(),
1611 children: class_children(
1612 &c.own_methods,
1613 Some(&c.own_properties),
1614 &c.own_constants,
1615 false,
1616 ),
1617 });
1618 }
1619 for i in defs.slice.interfaces.iter() {
1620 out.push(DocumentSymbol {
1621 name: i.fqcn.clone(),
1622 kind: DeclarationKind::Interface,
1623 location: i.location.clone(),
1624 children: class_children(&i.own_methods, None, &i.own_constants, false),
1625 });
1626 }
1627 for t in defs.slice.traits.iter() {
1628 out.push(DocumentSymbol {
1629 name: t.fqcn.clone(),
1630 kind: DeclarationKind::Trait,
1631 location: t.location.clone(),
1632 children: class_children(
1633 &t.own_methods,
1634 Some(&t.own_properties),
1635 &t.own_constants,
1636 false,
1637 ),
1638 });
1639 }
1640 for e in defs.slice.enums.iter() {
1641 let mut children = class_children(&e.own_methods, None, &e.own_constants, true);
1642 for (_, case) in e.cases.iter() {
1643 children.push(DocumentSymbol {
1644 name: case.name.clone(),
1645 kind: DeclarationKind::EnumCase,
1646 location: case.location.clone(),
1647 children: Vec::new(),
1648 });
1649 }
1650 out.push(DocumentSymbol {
1651 name: e.fqcn.clone(),
1652 kind: DeclarationKind::Enum,
1653 location: e.location.clone(),
1654 children,
1655 });
1656 }
1657 for f in defs.slice.functions.iter() {
1658 out.push(DocumentSymbol {
1659 name: f.fqn.clone(),
1660 kind: DeclarationKind::Function,
1661 location: f.location.clone(),
1662 children: Vec::new(),
1663 });
1664 }
1665 for (name, _) in defs.slice.constants.iter() {
1666 out.push(DocumentSymbol {
1667 name: name.clone(),
1668 kind: DeclarationKind::Constant,
1669 location: None,
1670 children: Vec::new(),
1671 });
1672 }
1673 out
1674 }
1675
1676 /// Choose the candidate-admission gate for `symbol`.
1677 ///
1678 /// For any known non-constructor/non-`__invoke` method, the member name
1679 /// alone is the sound gate. Every posting-producing reference spells that
1680 /// token: `$obj->m()`, `Owner::m()`, inherited `Sub::m()`,
1681 /// `self::`/`static::`/`parent::m()`, callable strings/arrays, and trait
1682 /// aliases (`orig as alias` records `orig`, alias calls record `alias`
1683 /// plus the origin key). Dropping the owner short name matters on common
1684 /// owner names (`User`, `Model`, `Widget`): otherwise a cold reference
1685 /// query analyzes files that only type-hint the owner and cannot contain a
1686 /// reference to the queried method. Dynamic member names (`$obj->$m()`)
1687 /// produce no posting, so nothing is lost there.
1688 ///
1689 /// For `__construct` with a known owner, the identifier needle is the
1690 /// owner's short name (`new Cls(` sites never spell the member name and
1691 /// the bare word `__construct` would admit every file *declaring* a
1692 /// constructor), complemented by the raw call tokens `->__construct` /
1693 /// `::__construct`: an explicit re-init `$obj->__construct()` is a real
1694 /// recorded reference whose file may never name the class.
1695 ///
1696 /// `__invoke` keeps the general owner/name gate because `$obj()` call sites
1697 /// do not spell `__invoke`.
1698 ///
1699 /// Everything else uses the general OR needles
1700 /// ([`reference_gate_needles`]).
1701 fn reference_gate(&self, symbol: &crate::Name) -> ReferenceGate {
1702 if let crate::Name::Method { class, name } = symbol {
1703 if name.as_ref() == "__construct" && !class.is_empty() {
1704 return ReferenceGate {
1705 idents: reference_gate_needles(symbol),
1706 raw: vec!["->__construct".to_string(), "::__construct".to_string()],
1707 };
1708 }
1709 if name.as_ref() != "__construct" && name.as_ref() != "__invoke" && !class.is_empty() {
1710 let db = self.snapshot_db();
1711 let here = crate::db::Fqcn::from_str(&db, class.as_ref());
1712 let is_static = crate::db::find_method_in_chain(&db, here, name)
1713 .map(|(_, m)| m.is_static)
1714 .unwrap_or(false);
1715 if is_static || crate::db::class_exists(&db, class.as_ref()) {
1716 return ReferenceGate {
1717 idents: vec![name.to_string()],
1718 raw: Vec::new(),
1719 };
1720 }
1721 }
1722 }
1723 ReferenceGate {
1724 idents: reference_gate_needles(symbol),
1725 raw: Vec::new(),
1726 }
1727 }
1728}
1729
1730/// A transitive subtype hit with its declaration name span, as returned by
1731/// [`AnalysisSession::indexed_subtype_classes`].
1732#[derive(Debug, Clone)]
1733pub struct SubtypeClassSite {
1734 /// Display-form FQCN (no leading `\`).
1735 pub fqcn: Arc<str>,
1736 pub kind: crate::db::ClassLikeKind,
1737 pub is_abstract: bool,
1738 pub file: Arc<str>,
1739 /// The declared name's own token (1-based line, 0-based char columns).
1740 pub range: crate::Range,
1741}
1742
1743/// Build a [`crate::Range`] on one line from mir's native coordinates
1744/// (1-based line, 0-based columns).
1745fn span_range(line: u32, col_start: u32, col_end: u32) -> crate::Range {
1746 crate::Range {
1747 start: crate::Position {
1748 line,
1749 column: col_start,
1750 },
1751 end: crate::Position {
1752 line,
1753 column: col_end,
1754 },
1755 }
1756}
1757
1758/// Char column of the first word-boundary occurrence of `needle` in `line`
1759/// at or after char column `min_col`. Columns are code points, matching the
1760/// collector's `Location` convention.
1761fn identifier_char_col(
1762 line: &str,
1763 needle: &str,
1764 min_col: usize,
1765 case_insensitive: bool,
1766) -> Option<u32> {
1767 if needle.is_empty() {
1768 return None;
1769 }
1770 let is_ident = |c: char| c.is_ascii_alphanumeric() || c == '_';
1771 let chars: Vec<char> = line.chars().collect();
1772 let needle_chars: Vec<char> = needle.chars().collect();
1773 let n = needle_chars.len();
1774 if chars.len() < n {
1775 return None;
1776 }
1777 for start in min_col..=chars.len().saturating_sub(n) {
1778 let matches = chars[start..start + n]
1779 .iter()
1780 .zip(needle_chars.iter())
1781 .all(|(a, b)| {
1782 if case_insensitive {
1783 a.eq_ignore_ascii_case(b)
1784 } else {
1785 a == b
1786 }
1787 });
1788 if !matches {
1789 continue;
1790 }
1791 let before_ok = start == 0 || !is_ident(chars[start - 1]);
1792 let after = start + n;
1793 let after_ok = after >= chars.len() || !is_ident(chars[after]);
1794 if before_ok && after_ok {
1795 return Some(start as u32);
1796 }
1797 }
1798 None
1799}
1800
1801/// Whether `hay` mentions `needle` as a whole identifier (ASCII word
1802/// boundaries; conservative near multibyte text). ASCII-case-insensitive:
1803/// PHP class, function, and method names are case-insensitive, so `new
1804/// COLOR()` must count as mentioning `Color`; for the case-sensitive kinds
1805/// (constants, properties) folding only widens the candidate superset.
1806///
1807/// Test-only semantic oracle: the production gates (references freshness,
1808/// subtype-BFS defs commit) answer this predicate through the persistent
1809/// `ClassMentionIndex`; the parity test below pins the scanner to these
1810/// exact boundary and case semantics.
1811#[cfg(test)]
1812fn mentions_identifier(hay: &str, needle: &str) -> bool {
1813 let hay = hay.as_bytes();
1814 let needle = needle.as_bytes();
1815 let n = needle.len();
1816 if n == 0 || hay.len() < n {
1817 return false;
1818 }
1819 let is_ident = |b: u8| b.is_ascii_alphanumeric() || b == b'_';
1820 let first = needle[0].to_ascii_lowercase();
1821 for i in 0..=(hay.len() - n) {
1822 if hay[i].to_ascii_lowercase() != first || !hay[i..i + n].eq_ignore_ascii_case(needle) {
1823 continue;
1824 }
1825 if (i == 0 || !is_ident(hay[i - 1])) && (i + n == hay.len() || !is_ident(hay[i + n])) {
1826 return true;
1827 }
1828 }
1829 false
1830}
1831
1832fn short(fqn: &str) -> &str {
1833 fqn.rsplit('\\').next().unwrap_or(fqn)
1834}
1835
1836/// A candidate-file admission predicate for `indexed_references_to`'s
1837/// freshness pass, chosen by [`AnalysisSession::reference_gate`]. A file is
1838/// admitted when its text mentions any of `idents` as a whole identifier
1839/// (word-bounded, ASCII-case-insensitive) OR contains any of `raw` as a
1840/// plain substring (ASCII-case-insensitive, no word bounds — used for
1841/// call-shaped tokens like `->__construct`). A file matching neither can
1842/// hold no posting for the symbol.
1843struct ReferenceGate {
1844 idents: Vec<String>,
1845 raw: Vec<String>,
1846}
1847
1848/// Identifier words whose whole-word presence in a file's text is necessary
1849/// for the file to hold any posting [`AnalysisSession::indexed_references_to`]
1850/// can return for `symbol`. Member symbols include the owner class's short
1851/// name alongside the member name: `__construct` postings are recorded at
1852/// `new Cls(` sites, which never spell the member name.
1853fn reference_gate_needles(symbol: &crate::Name) -> Vec<String> {
1854 let mut needles = match symbol {
1855 crate::Name::Class(f) | crate::Name::Function(f) | crate::Name::GlobalConstant(f) => {
1856 vec![short(f).to_string()]
1857 }
1858 // `__construct` is invoked only as `new Cls(...)`, `parent::__construct()`,
1859 // or `self::__construct()`/`static::__construct()` from inside a
1860 // subclass — every real call site textually names the class itself
1861 // (directly, or via the enclosing subclass's own `extends`/`use`),
1862 // never the bare word `__construct`. Gating on the class's short name
1863 // alone is exact (no lost call sites) and, unlike the general member
1864 // case, dropping the method-name needle here doesn't reintroduce a
1865 // false negative. This matters: `__construct` is one of the most
1866 // common tokens in any real codebase, so OR-ing it in as a needle
1867 // admits nearly every file as a "must re-analyze" candidate on a
1868 // cold query, defeating the gate's entire purpose for constructors.
1869 crate::Name::Method { class, name } if name.as_ref() == "__construct" => {
1870 if class.is_empty() {
1871 // No class to scope to (owner unknown) — fall back to gating
1872 // on the bare name, same as the general member case below.
1873 vec![name.to_string()]
1874 } else {
1875 vec![short(class).to_string()]
1876 }
1877 }
1878 crate::Name::Method { class, name }
1879 | crate::Name::Property { class, name }
1880 | crate::Name::ClassConstant { class, name } => {
1881 let mut v = vec![name.to_string()];
1882 if !class.is_empty() {
1883 v.push(short(class).to_string());
1884 }
1885 v
1886 }
1887 };
1888 // An empty needle can never match; dropping it keeps the "empty needle
1889 // set disables the gate" contract at the call site conservative.
1890 needles.retain(|n| !n.is_empty());
1891 needles
1892}
1893
1894#[cfg(test)]
1895mod tests {
1896 use super::*;
1897
1898 #[test]
1899 fn mentions_identifier_is_case_insensitive_and_word_bounded() {
1900 assert!(mentions_identifier("$this->save();", "save"));
1901 assert!(mentions_identifier("new COLOR()", "Color"));
1902 assert!(mentions_identifier("use App\\Color as Paint;", "color"));
1903 assert!(!mentions_identifier("$this->saveAll();", "save"));
1904 assert!(!mentions_identifier("return $unsaved;", "save"));
1905 assert!(!mentions_identifier("no occurrence", "save"));
1906 assert!(!mentions_identifier("anything", ""));
1907 // Multibyte neighbors are conservatively treated as boundaries, and
1908 // substring scans must not split codepoints.
1909 assert!(!mentions_identifier("function xÉclairFoo() {}", "Éclair"));
1910 assert!(mentions_identifier("implements Éclair {}", "Éclair"));
1911 }
1912
1913 #[test]
1914 fn mention_scanner_membership_equals_per_needle_scans() {
1915 // The mention index is the sole implementation of the gates' textual
1916 // predicate, so scanner membership must equal the reference
1917 // per-needle predicate for every (hay, needle) pair — same boundary
1918 // and case semantics.
1919 use crate::db::MentionScanner;
1920 use std::sync::Arc;
1921 let universe = ["Color", "save", "ColorPicker", "Éclair", "C1", "_Wrap"];
1922 let names: Vec<(mir_types::Name, bool)> = universe
1923 .iter()
1924 .map(|s| (mir_types::Name::new(s).ascii_lowercase(), false))
1925 .collect();
1926 let scanner = Arc::new(MentionScanner::build(1, names).unwrap());
1927 let hays = [
1928 "$this->save();",
1929 "new COLOR()",
1930 "use App\\Color as Paint;",
1931 "$this->saveAll();",
1932 "return $unsaved;",
1933 "new ColorPicker(); Color::save();",
1934 "function xÉclairFoo() {}",
1935 "implements Éclair {}",
1936 "colorsave savecolor color_save",
1937 "class C1 extends _Wrap {}",
1938 "",
1939 ];
1940 for hay in hays {
1941 let scanned = scanner.scan(hay);
1942 for needle in universe {
1943 let expected = mentions_identifier(hay, needle);
1944 let name = mir_types::Name::new(needle).ascii_lowercase();
1945 assert_eq!(
1946 scanned.binary_search(&name).is_ok(),
1947 expected,
1948 "needle {needle:?} on {hay:?}"
1949 );
1950 }
1951 }
1952 }
1953
1954 #[test]
1955 fn gate_needles_cover_member_and_owner_class() {
1956 // A regular member (non-constructor) gates on both the member name
1957 // and the owner's short name — a call site may name only one.
1958 let n = reference_gate_needles(&crate::Name::method("App\\Job", "run"));
1959 assert!(n.contains(&"run".to_string()) && n.contains(&"Job".to_string()));
1960 let n = reference_gate_needles(&crate::Name::class("App\\Ui\\Color"));
1961 assert_eq!(n, vec!["Color".to_string()]);
1962 // Unknown-owner member symbols still gate on the member name alone.
1963 let n = reference_gate_needles(&crate::Name::method("", "run"));
1964 assert_eq!(n, vec!["run".to_string()]);
1965 }
1966
1967 #[test]
1968 fn gate_needles_for_constructor_scope_to_owner_class_only() {
1969 // `__construct` is only ever spelled at `new Cls(`/`parent::__construct()`
1970 // sites, which always name the class — the bare method-name needle is
1971 // dropped so a cold constructor query doesn't admit nearly every file
1972 // in the workspace (every class defines *some* `__construct`).
1973 let n = reference_gate_needles(&crate::Name::method("App\\Job", "__construct"));
1974 assert_eq!(n, vec!["Job".to_string()]);
1975 // Unknown owner: nothing to scope to, fall back to the bare name.
1976 let n = reference_gate_needles(&crate::Name::method("", "__construct"));
1977 assert_eq!(n, vec!["__construct".to_string()]);
1978 }
1979
1980 fn session_with(files: &[(&str, &str)]) -> crate::AnalysisSession {
1981 let mut session = crate::AnalysisSession::new(crate::PhpVersion::LATEST);
1982 for (path, text) in files {
1983 session.set_file_text(Arc::from(*path), Arc::from(*text));
1984 }
1985 session
1986 }
1987
1988 #[test]
1989 fn gate_static_method_is_member_name_only() {
1990 // Regardless of subtypes: the member token alone is the sound gate
1991 // (an instance receiver `$obj::m()` never names the owner), and it
1992 // is also the selective part — the owner short name would only
1993 // widen the admitted set.
1994 let session = session_with(&[
1995 (
1996 "owner.php",
1997 "<?php\nclass Owner { public static function m(): void {} }\n",
1998 ),
1999 ("sub.php", "<?php\nclass Sub extends Owner {}\n"),
2000 ]);
2001 let gate = session.reference_gate(&crate::Name::method("Owner", "m"));
2002 assert_eq!(gate.idents, vec!["m".to_string()]);
2003 assert!(gate.raw.is_empty());
2004 }
2005
2006 #[test]
2007 fn gate_instance_method_is_member_name_only() {
2008 let session = session_with(&[(
2009 "owner.php",
2010 "<?php\nclass Owner { public function m(): void {} }\n",
2011 )]);
2012 let gate = session.reference_gate(&crate::Name::method("Owner", "m"));
2013 assert_eq!(gate.idents, vec!["m".to_string()]);
2014 assert!(gate.raw.is_empty());
2015 }
2016
2017 #[test]
2018 fn gate_invoke_keeps_owner_needle() {
2019 let session = session_with(&[(
2020 "owner.php",
2021 "<?php\nclass Owner { public function __invoke(): void {} }\n",
2022 )]);
2023 let gate = session.reference_gate(&crate::Name::method("Owner", "__invoke"));
2024 assert_eq!(
2025 gate.idents,
2026 reference_gate_needles(&crate::Name::method("Owner", "__invoke"))
2027 );
2028 assert!(gate.raw.is_empty());
2029 }
2030
2031 #[test]
2032 fn gate_constructor_adds_raw_call_tokens() {
2033 // Owner short name for `new Cls(` sites, plus the raw call tokens
2034 // for explicit re-init (`$obj->__construct()`) whose file may never
2035 // name the class. The bare identifier `__construct` must NOT be a
2036 // needle — it would admit every file declaring a constructor.
2037 let session = session_with(&[(
2038 "owner.php",
2039 "<?php\nclass Owner { public function __construct() {} }\n",
2040 )]);
2041 let gate = session.reference_gate(&crate::Name::method("Owner", "__construct"));
2042 assert_eq!(
2043 gate.idents,
2044 reference_gate_needles(&crate::Name::method("Owner", "__construct"))
2045 );
2046 assert_eq!(
2047 gate.raw,
2048 vec!["->__construct".to_string(), "::__construct".to_string()]
2049 );
2050 }
2051
2052 #[test]
2053 fn gate_unresolvable_owner_falls_back_to_general_needles() {
2054 let session = session_with(&[(
2055 "owner.php",
2056 "<?php\nclass Owner { public static function m(): void {} }\n",
2057 )]);
2058 let gate = session.reference_gate(&crate::Name::method("Nonexistent\\Missing", "m"));
2059 assert_eq!(
2060 gate.idents,
2061 reference_gate_needles(&crate::Name::method("Nonexistent\\Missing", "m"))
2062 );
2063 assert!(gate.raw.is_empty());
2064 }
2065}