1 //===- SourceManager.h - Track and cache source files -----------*- C++ -*-===//
2 //
3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4 // See https://llvm.org/LICENSE.txt for license information.
5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6 //
7 //===----------------------------------------------------------------------===//
8 //
9 /// \file
10 /// Defines the SourceManager interface.
11 ///
12 /// There are three different types of locations in a %file: a spelling
13 /// location, an expansion location, and a presumed location.
14 ///
15 /// Given an example of:
16 /// \code
17 /// #define min(x, y) x < y ? x : y
18 /// \endcode
19 ///
20 /// and then later on a use of min:
21 /// \code
22 /// #line 17
23 /// return min(a, b);
24 /// \endcode
25 ///
26 /// The expansion location is the line in the source code where the macro
27 /// was expanded (the return statement), the spelling location is the
28 /// location in the source where the macro was originally defined,
29 /// and the presumed location is where the line directive states that
30 /// the line is 17, or any other line.
31 //
32 //===----------------------------------------------------------------------===//
33
34 #ifndef LLVM_CLANG_BASIC_SOURCEMANAGER_H
35 #define LLVM_CLANG_BASIC_SOURCEMANAGER_H
36
37 #include "clang/Basic/Diagnostic.h"
38 #include "clang/Basic/FileEntry.h"
39 #include "clang/Basic/SourceLocation.h"
40 #include "llvm/ADT/ArrayRef.h"
41 #include "llvm/ADT/BitVector.h"
42 #include "llvm/ADT/DenseMap.h"
43 #include "llvm/ADT/DenseSet.h"
44 #include "llvm/ADT/IntrusiveRefCntPtr.h"
45 #include "llvm/ADT/PointerIntPair.h"
46 #include "llvm/ADT/SmallVector.h"
47 #include "llvm/ADT/StringRef.h"
48 #include "llvm/Support/Allocator.h"
49 #include "llvm/Support/Compiler.h"
50 #include "llvm/Support/MemoryBuffer.h"
51 #include <cassert>
52 #include <cstddef>
53 #include <map>
54 #include <memory>
55 #include <string>
56 #include <utility>
57 #include <vector>
58
59 namespace clang {
60
61 class ASTReader;
62 class ASTWriter;
63 class FileManager;
64 class LineTableInfo;
65 class SourceManager;
66
67 /// Public enums and private classes that are part of the
68 /// SourceManager implementation.
69 namespace SrcMgr {
70
71 /// Indicates whether a file or directory holds normal user code,
72 /// system code, or system code which is implicitly 'extern "C"' in C++ mode.
73 ///
74 /// Entire directories can be tagged with this (this is maintained by
75 /// DirectoryLookup and friends) as can specific FileInfos when a \#pragma
76 /// system_header is seen or in various other cases.
77 ///
78 enum CharacteristicKind {
79 C_User,
80 C_System,
81 C_ExternCSystem,
82 C_User_ModuleMap,
83 C_System_ModuleMap
84 };
85
86 /// Determine whether a file / directory characteristic is for system code.
isSystem(CharacteristicKind CK)87 inline bool isSystem(CharacteristicKind CK) {
88 return CK != C_User && CK != C_User_ModuleMap;
89 }
90
91 /// Determine whether a file characteristic is for a module map.
isModuleMap(CharacteristicKind CK)92 inline bool isModuleMap(CharacteristicKind CK) {
93 return CK == C_User_ModuleMap || CK == C_System_ModuleMap;
94 }
95
96 /// Mapping of line offsets into a source file. This does not own the storage
97 /// for the line numbers.
98 class LineOffsetMapping {
99 public:
100 explicit operator bool() const { return Storage; }
size()101 unsigned size() const {
102 assert(Storage);
103 return Storage[0];
104 }
getLines()105 ArrayRef<unsigned> getLines() const {
106 assert(Storage);
107 return ArrayRef<unsigned>(Storage + 1, Storage + 1 + size());
108 }
begin()109 const unsigned *begin() const { return getLines().begin(); }
end()110 const unsigned *end() const { return getLines().end(); }
111 const unsigned &operator[](int I) const { return getLines()[I]; }
112
113 static LineOffsetMapping get(llvm::MemoryBufferRef Buffer,
114 llvm::BumpPtrAllocator &Alloc);
115
116 LineOffsetMapping() = default;
117 LineOffsetMapping(ArrayRef<unsigned> LineOffsets,
118 llvm::BumpPtrAllocator &Alloc);
119
120 private:
121 /// First element is the size, followed by elements at off-by-one indexes.
122 unsigned *Storage = nullptr;
123 };
124
125 /// One instance of this struct is kept for every file loaded or used.
126 ///
127 /// This object owns the MemoryBuffer object.
128 class alignas(8) ContentCache {
129 /// The actual buffer containing the characters from the input
130 /// file.
131 mutable std::unique_ptr<llvm::MemoryBuffer> Buffer;
132
133 public:
134 /// Reference to the file entry representing this ContentCache.
135 ///
136 /// This reference does not own the FileEntry object.
137 ///
138 /// It is possible for this to be NULL if the ContentCache encapsulates
139 /// an imaginary text buffer.
140 ///
141 /// FIXME: Turn this into a FileEntryRef and remove Filename.
142 const FileEntry *OrigEntry;
143
144 /// References the file which the contents were actually loaded from.
145 ///
146 /// Can be different from 'Entry' if we overridden the contents of one file
147 /// with the contents of another file.
148 const FileEntry *ContentsEntry;
149
150 /// The filename that is used to access OrigEntry.
151 ///
152 /// FIXME: Remove this once OrigEntry is a FileEntryRef with a stable name.
153 StringRef Filename;
154
155 /// A bump pointer allocated array of offsets for each source line.
156 ///
157 /// This is lazily computed. The lines are owned by the SourceManager
158 /// BumpPointerAllocator object.
159 mutable LineOffsetMapping SourceLineCache;
160
161 /// Indicates whether the buffer itself was provided to override
162 /// the actual file contents.
163 ///
164 /// When true, the original entry may be a virtual file that does not
165 /// exist.
166 unsigned BufferOverridden : 1;
167
168 /// True if this content cache was initially created for a source file
169 /// considered to be volatile (likely to change between stat and open).
170 unsigned IsFileVolatile : 1;
171
172 /// True if this file may be transient, that is, if it might not
173 /// exist at some later point in time when this content entry is used,
174 /// after serialization and deserialization.
175 unsigned IsTransient : 1;
176
177 mutable unsigned IsBufferInvalid : 1;
178
ContentCache(Ent,Ent)179 ContentCache(const FileEntry *Ent = nullptr) : ContentCache(Ent, Ent) {}
180
ContentCache(const FileEntry * Ent,const FileEntry * contentEnt)181 ContentCache(const FileEntry *Ent, const FileEntry *contentEnt)
182 : OrigEntry(Ent), ContentsEntry(contentEnt), BufferOverridden(false),
183 IsFileVolatile(false), IsTransient(false), IsBufferInvalid(false) {}
184
185 /// The copy ctor does not allow copies where source object has either
186 /// a non-NULL Buffer or SourceLineCache. Ownership of allocated memory
187 /// is not transferred, so this is a logical error.
ContentCache(const ContentCache & RHS)188 ContentCache(const ContentCache &RHS)
189 : BufferOverridden(false), IsFileVolatile(false), IsTransient(false),
190 IsBufferInvalid(false) {
191 OrigEntry = RHS.OrigEntry;
192 ContentsEntry = RHS.ContentsEntry;
193
194 assert(!RHS.Buffer && !RHS.SourceLineCache &&
195 "Passed ContentCache object cannot own a buffer.");
196 }
197
198 ContentCache &operator=(const ContentCache &RHS) = delete;
199
200 /// Returns the memory buffer for the associated content.
201 ///
202 /// \param Diag Object through which diagnostics will be emitted if the
203 /// buffer cannot be retrieved.
204 ///
205 /// \param Loc If specified, is the location that invalid file diagnostics
206 /// will be emitted at.
207 llvm::Optional<llvm::MemoryBufferRef>
208 getBufferOrNone(DiagnosticsEngine &Diag, FileManager &FM,
209 SourceLocation Loc = SourceLocation()) const;
210
211 /// Returns the size of the content encapsulated by this
212 /// ContentCache.
213 ///
214 /// This can be the size of the source file or the size of an
215 /// arbitrary scratch buffer. If the ContentCache encapsulates a source
216 /// file this size is retrieved from the file's FileEntry.
217 unsigned getSize() const;
218
219 /// Returns the number of bytes actually mapped for this
220 /// ContentCache.
221 ///
222 /// This can be 0 if the MemBuffer was not actually expanded.
223 unsigned getSizeBytesMapped() const;
224
225 /// Returns the kind of memory used to back the memory buffer for
226 /// this content cache. This is used for performance analysis.
227 llvm::MemoryBuffer::BufferKind getMemoryBufferKind() const;
228
229 /// Return the buffer, only if it has been loaded.
getBufferIfLoaded()230 llvm::Optional<llvm::MemoryBufferRef> getBufferIfLoaded() const {
231 if (Buffer)
232 return Buffer->getMemBufferRef();
233 return None;
234 }
235
236 /// Return a StringRef to the source buffer data, only if it has already
237 /// been loaded.
getBufferDataIfLoaded()238 llvm::Optional<StringRef> getBufferDataIfLoaded() const {
239 if (Buffer)
240 return Buffer->getBuffer();
241 return None;
242 }
243
244 /// Set the buffer.
setBuffer(std::unique_ptr<llvm::MemoryBuffer> B)245 void setBuffer(std::unique_ptr<llvm::MemoryBuffer> B) {
246 IsBufferInvalid = false;
247 Buffer = std::move(B);
248 }
249
250 /// Set the buffer to one that's not owned (or to nullptr).
251 ///
252 /// \pre Buffer cannot already be set.
setUnownedBuffer(llvm::Optional<llvm::MemoryBufferRef> B)253 void setUnownedBuffer(llvm::Optional<llvm::MemoryBufferRef> B) {
254 assert(!Buffer && "Expected to be called right after construction");
255 if (B)
256 setBuffer(llvm::MemoryBuffer::getMemBuffer(*B));
257 }
258
259 // If BufStr has an invalid BOM, returns the BOM name; otherwise, returns
260 // nullptr
261 static const char *getInvalidBOM(StringRef BufStr);
262 };
263
264 // Assert that the \c ContentCache objects will always be 8-byte aligned so
265 // that we can pack 3 bits of integer into pointers to such objects.
266 static_assert(alignof(ContentCache) >= 8,
267 "ContentCache must be 8-byte aligned.");
268
269 /// Information about a FileID, basically just the logical file
270 /// that it represents and include stack information.
271 ///
272 /// Each FileInfo has include stack information, indicating where it came
273 /// from. This information encodes the \#include chain that a token was
274 /// expanded from. The main include file has an invalid IncludeLoc.
275 ///
276 /// FileInfo should not grow larger than ExpansionInfo. Doing so will
277 /// cause memory to bloat in compilations with many unloaded macro
278 /// expansions, since the two data structurs are stored in a union in
279 /// SLocEntry. Extra fields should instead go in "ContentCache *", which
280 /// stores file contents and other bits on the side.
281 ///
282 class FileInfo {
283 friend class clang::SourceManager;
284 friend class clang::ASTWriter;
285 friend class clang::ASTReader;
286
287 /// The location of the \#include that brought in this file.
288 ///
289 /// This is an invalid SLOC for the main file (top of the \#include chain).
290 SourceLocation IncludeLoc;
291
292 /// Number of FileIDs (files and macros) that were created during
293 /// preprocessing of this \#include, including this SLocEntry.
294 ///
295 /// Zero means the preprocessor didn't provide such info for this SLocEntry.
296 unsigned NumCreatedFIDs : 31;
297
298 /// Whether this FileInfo has any \#line directives.
299 unsigned HasLineDirectives : 1;
300
301 /// The content cache and the characteristic of the file.
302 llvm::PointerIntPair<const ContentCache *, 3, CharacteristicKind>
303 ContentAndKind;
304
305 public:
306 /// Return a FileInfo object.
get(SourceLocation IL,ContentCache & Con,CharacteristicKind FileCharacter,StringRef Filename)307 static FileInfo get(SourceLocation IL, ContentCache &Con,
308 CharacteristicKind FileCharacter, StringRef Filename) {
309 FileInfo X;
310 X.IncludeLoc = IL;
311 X.NumCreatedFIDs = 0;
312 X.HasLineDirectives = false;
313 X.ContentAndKind.setPointer(&Con);
314 X.ContentAndKind.setInt(FileCharacter);
315 Con.Filename = Filename;
316 return X;
317 }
318
getIncludeLoc()319 SourceLocation getIncludeLoc() const {
320 return IncludeLoc;
321 }
322
getContentCache()323 const ContentCache &getContentCache() const {
324 return *ContentAndKind.getPointer();
325 }
326
327 /// Return whether this is a system header or not.
getFileCharacteristic()328 CharacteristicKind getFileCharacteristic() const {
329 return ContentAndKind.getInt();
330 }
331
332 /// Return true if this FileID has \#line directives in it.
hasLineDirectives()333 bool hasLineDirectives() const { return HasLineDirectives; }
334
335 /// Set the flag that indicates that this FileID has
336 /// line table entries associated with it.
setHasLineDirectives()337 void setHasLineDirectives() { HasLineDirectives = true; }
338
339 /// Returns the name of the file that was used when the file was loaded from
340 /// the underlying file system.
getName()341 StringRef getName() const { return getContentCache().Filename; }
342 };
343
344 /// Each ExpansionInfo encodes the expansion location - where
345 /// the token was ultimately expanded, and the SpellingLoc - where the actual
346 /// character data for the token came from.
347 class ExpansionInfo {
348 // Really these are all SourceLocations.
349
350 /// Where the spelling for the token can be found.
351 SourceLocation SpellingLoc;
352
353 /// In a macro expansion, ExpansionLocStart and ExpansionLocEnd
354 /// indicate the start and end of the expansion. In object-like macros,
355 /// they will be the same. In a function-like macro expansion, the start
356 /// will be the identifier and the end will be the ')'. Finally, in
357 /// macro-argument instantiations, the end will be 'SourceLocation()', an
358 /// invalid location.
359 SourceLocation ExpansionLocStart, ExpansionLocEnd;
360
361 /// Whether the expansion range is a token range.
362 bool ExpansionIsTokenRange;
363
364 public:
getSpellingLoc()365 SourceLocation getSpellingLoc() const {
366 return SpellingLoc.isInvalid() ? getExpansionLocStart() : SpellingLoc;
367 }
368
getExpansionLocStart()369 SourceLocation getExpansionLocStart() const {
370 return ExpansionLocStart;
371 }
372
getExpansionLocEnd()373 SourceLocation getExpansionLocEnd() const {
374 return ExpansionLocEnd.isInvalid() ? getExpansionLocStart()
375 : ExpansionLocEnd;
376 }
377
isExpansionTokenRange()378 bool isExpansionTokenRange() const { return ExpansionIsTokenRange; }
379
getExpansionLocRange()380 CharSourceRange getExpansionLocRange() const {
381 return CharSourceRange(
382 SourceRange(getExpansionLocStart(), getExpansionLocEnd()),
383 isExpansionTokenRange());
384 }
385
isMacroArgExpansion()386 bool isMacroArgExpansion() const {
387 // Note that this needs to return false for default constructed objects.
388 return getExpansionLocStart().isValid() && ExpansionLocEnd.isInvalid();
389 }
390
isMacroBodyExpansion()391 bool isMacroBodyExpansion() const {
392 return getExpansionLocStart().isValid() && ExpansionLocEnd.isValid();
393 }
394
isFunctionMacroExpansion()395 bool isFunctionMacroExpansion() const {
396 return getExpansionLocStart().isValid() &&
397 getExpansionLocStart() != getExpansionLocEnd();
398 }
399
400 /// Return a ExpansionInfo for an expansion.
401 ///
402 /// Start and End specify the expansion range (where the macro is
403 /// expanded), and SpellingLoc specifies the spelling location (where
404 /// the characters from the token come from). All three can refer to
405 /// normal File SLocs or expansion locations.
406 static ExpansionInfo create(SourceLocation SpellingLoc, SourceLocation Start,
407 SourceLocation End,
408 bool ExpansionIsTokenRange = true) {
409 ExpansionInfo X;
410 X.SpellingLoc = SpellingLoc;
411 X.ExpansionLocStart = Start;
412 X.ExpansionLocEnd = End;
413 X.ExpansionIsTokenRange = ExpansionIsTokenRange;
414 return X;
415 }
416
417 /// Return a special ExpansionInfo for the expansion of
418 /// a macro argument into a function-like macro's body.
419 ///
420 /// ExpansionLoc specifies the expansion location (where the macro is
421 /// expanded). This doesn't need to be a range because a macro is always
422 /// expanded at a macro parameter reference, and macro parameters are
423 /// always exactly one token. SpellingLoc specifies the spelling location
424 /// (where the characters from the token come from). ExpansionLoc and
425 /// SpellingLoc can both refer to normal File SLocs or expansion locations.
426 ///
427 /// Given the code:
428 /// \code
429 /// #define F(x) f(x)
430 /// F(42);
431 /// \endcode
432 ///
433 /// When expanding '\c F(42)', the '\c x' would call this with an
434 /// SpellingLoc pointing at '\c 42' and an ExpansionLoc pointing at its
435 /// location in the definition of '\c F'.
createForMacroArg(SourceLocation SpellingLoc,SourceLocation ExpansionLoc)436 static ExpansionInfo createForMacroArg(SourceLocation SpellingLoc,
437 SourceLocation ExpansionLoc) {
438 // We store an intentionally invalid source location for the end of the
439 // expansion range to mark that this is a macro argument location rather
440 // than a normal one.
441 return create(SpellingLoc, ExpansionLoc, SourceLocation());
442 }
443
444 /// Return a special ExpansionInfo representing a token that ends
445 /// prematurely. This is used to model a '>>' token that has been split
446 /// into '>' tokens and similar cases. Unlike for the other forms of
447 /// expansion, the expansion range in this case is a character range, not
448 /// a token range.
createForTokenSplit(SourceLocation SpellingLoc,SourceLocation Start,SourceLocation End)449 static ExpansionInfo createForTokenSplit(SourceLocation SpellingLoc,
450 SourceLocation Start,
451 SourceLocation End) {
452 return create(SpellingLoc, Start, End, false);
453 }
454 };
455
456 // Assert that the \c FileInfo objects are no bigger than \c ExpansionInfo
457 // objects. This controls the size of \c SLocEntry, of which we have one for
458 // each macro expansion. The number of (unloaded) macro expansions can be
459 // very large. Any other fields needed in FileInfo should go in ContentCache.
460 static_assert(sizeof(FileInfo) <= sizeof(ExpansionInfo),
461 "FileInfo must be no larger than ExpansionInfo.");
462
463 /// This is a discriminated union of FileInfo and ExpansionInfo.
464 ///
465 /// SourceManager keeps an array of these objects, and they are uniquely
466 /// identified by the FileID datatype.
467 class SLocEntry {
468 unsigned Offset : 31;
469 unsigned IsExpansion : 1;
470 union {
471 FileInfo File;
472 ExpansionInfo Expansion;
473 };
474
475 public:
SLocEntry()476 SLocEntry() : Offset(), IsExpansion(), File() {}
477
getOffset()478 unsigned getOffset() const { return Offset; }
479
isExpansion()480 bool isExpansion() const { return IsExpansion; }
isFile()481 bool isFile() const { return !isExpansion(); }
482
getFile()483 const FileInfo &getFile() const {
484 assert(isFile() && "Not a file SLocEntry!");
485 return File;
486 }
487
getExpansion()488 const ExpansionInfo &getExpansion() const {
489 assert(isExpansion() && "Not a macro expansion SLocEntry!");
490 return Expansion;
491 }
492
get(unsigned Offset,const FileInfo & FI)493 static SLocEntry get(unsigned Offset, const FileInfo &FI) {
494 assert(!(Offset & (1u << 31)) && "Offset is too large");
495 SLocEntry E;
496 E.Offset = Offset;
497 E.IsExpansion = false;
498 E.File = FI;
499 return E;
500 }
501
get(unsigned Offset,const ExpansionInfo & Expansion)502 static SLocEntry get(unsigned Offset, const ExpansionInfo &Expansion) {
503 assert(!(Offset & (1u << 31)) && "Offset is too large");
504 SLocEntry E;
505 E.Offset = Offset;
506 E.IsExpansion = true;
507 new (&E.Expansion) ExpansionInfo(Expansion);
508 return E;
509 }
510 };
511
512 } // namespace SrcMgr
513
514 /// External source of source location entries.
515 class ExternalSLocEntrySource {
516 public:
517 virtual ~ExternalSLocEntrySource();
518
519 /// Read the source location entry with index ID, which will always be
520 /// less than -1.
521 ///
522 /// \returns true if an error occurred that prevented the source-location
523 /// entry from being loaded.
524 virtual bool ReadSLocEntry(int ID) = 0;
525
526 /// Retrieve the module import location and name for the given ID, if
527 /// in fact it was loaded from a module (rather than, say, a precompiled
528 /// header).
529 virtual std::pair<SourceLocation, StringRef> getModuleImportLoc(int ID) = 0;
530 };
531
532 /// Holds the cache used by isBeforeInTranslationUnit.
533 ///
534 /// The cache structure is complex enough to be worth breaking out of
535 /// SourceManager.
536 class InBeforeInTUCacheEntry {
537 /// The FileID's of the cached query.
538 ///
539 /// If these match up with a subsequent query, the result can be reused.
540 FileID LQueryFID, RQueryFID;
541
542 /// True if LQueryFID was created before RQueryFID.
543 ///
544 /// This is used to compare macro expansion locations.
545 bool IsLQFIDBeforeRQFID;
546
547 /// The file found in common between the two \#include traces, i.e.,
548 /// the nearest common ancestor of the \#include tree.
549 FileID CommonFID;
550
551 /// The offset of the previous query in CommonFID.
552 ///
553 /// Usually, this represents the location of the \#include for QueryFID, but
554 /// if LQueryFID is a parent of RQueryFID (or vice versa) then these can be a
555 /// random token in the parent.
556 unsigned LCommonOffset, RCommonOffset;
557
558 public:
559 /// Return true if the currently cached values match up with
560 /// the specified LHS/RHS query.
561 ///
562 /// If not, we can't use the cache.
isCacheValid(FileID LHS,FileID RHS)563 bool isCacheValid(FileID LHS, FileID RHS) const {
564 return LQueryFID == LHS && RQueryFID == RHS;
565 }
566
567 /// If the cache is valid, compute the result given the
568 /// specified offsets in the LHS/RHS FileID's.
getCachedResult(unsigned LOffset,unsigned ROffset)569 bool getCachedResult(unsigned LOffset, unsigned ROffset) const {
570 // If one of the query files is the common file, use the offset. Otherwise,
571 // use the #include loc in the common file.
572 if (LQueryFID != CommonFID) LOffset = LCommonOffset;
573 if (RQueryFID != CommonFID) ROffset = RCommonOffset;
574
575 // It is common for multiple macro expansions to be "included" from the same
576 // location (expansion location), in which case use the order of the FileIDs
577 // to determine which came first. This will also take care the case where
578 // one of the locations points at the inclusion/expansion point of the other
579 // in which case its FileID will come before the other.
580 if (LOffset == ROffset)
581 return IsLQFIDBeforeRQFID;
582
583 return LOffset < ROffset;
584 }
585
586 /// Set up a new query.
setQueryFIDs(FileID LHS,FileID RHS,bool isLFIDBeforeRFID)587 void setQueryFIDs(FileID LHS, FileID RHS, bool isLFIDBeforeRFID) {
588 assert(LHS != RHS);
589 LQueryFID = LHS;
590 RQueryFID = RHS;
591 IsLQFIDBeforeRQFID = isLFIDBeforeRFID;
592 }
593
clear()594 void clear() {
595 LQueryFID = RQueryFID = FileID();
596 IsLQFIDBeforeRQFID = false;
597 }
598
setCommonLoc(FileID commonFID,unsigned lCommonOffset,unsigned rCommonOffset)599 void setCommonLoc(FileID commonFID, unsigned lCommonOffset,
600 unsigned rCommonOffset) {
601 CommonFID = commonFID;
602 LCommonOffset = lCommonOffset;
603 RCommonOffset = rCommonOffset;
604 }
605 };
606
607 /// The stack used when building modules on demand, which is used
608 /// to provide a link between the source managers of the different compiler
609 /// instances.
610 using ModuleBuildStack = ArrayRef<std::pair<std::string, FullSourceLoc>>;
611
612 /// This class handles loading and caching of source files into memory.
613 ///
614 /// This object owns the MemoryBuffer objects for all of the loaded
615 /// files and assigns unique FileID's for each unique \#include chain.
616 ///
617 /// The SourceManager can be queried for information about SourceLocation
618 /// objects, turning them into either spelling or expansion locations. Spelling
619 /// locations represent where the bytes corresponding to a token came from and
620 /// expansion locations represent where the location is in the user's view. In
621 /// the case of a macro expansion, for example, the spelling location indicates
622 /// where the expanded token came from and the expansion location specifies
623 /// where it was expanded.
624 class SourceManager : public RefCountedBase<SourceManager> {
625 /// DiagnosticsEngine object.
626 DiagnosticsEngine &Diag;
627
628 FileManager &FileMgr;
629
630 mutable llvm::BumpPtrAllocator ContentCacheAlloc;
631
632 /// Memoized information about all of the files tracked by this
633 /// SourceManager.
634 ///
635 /// This map allows us to merge ContentCache entries based
636 /// on their FileEntry*. All ContentCache objects will thus have unique,
637 /// non-null, FileEntry pointers.
638 llvm::DenseMap<const FileEntry*, SrcMgr::ContentCache*> FileInfos;
639
640 /// True if the ContentCache for files that are overridden by other
641 /// files, should report the original file name. Defaults to true.
642 bool OverridenFilesKeepOriginalName = true;
643
644 /// True if non-system source files should be treated as volatile
645 /// (likely to change while trying to use them). Defaults to false.
646 bool UserFilesAreVolatile;
647
648 /// True if all files read during this compilation should be treated
649 /// as transient (may not be present in later compilations using a module
650 /// file created from this compilation). Defaults to false.
651 bool FilesAreTransient = false;
652
653 struct OverriddenFilesInfoTy {
654 /// Files that have been overridden with the contents from another
655 /// file.
656 llvm::DenseMap<const FileEntry *, const FileEntry *> OverriddenFiles;
657
658 /// Files that were overridden with a memory buffer.
659 llvm::DenseSet<const FileEntry *> OverriddenFilesWithBuffer;
660 };
661
662 /// Lazily create the object keeping overridden files info, since
663 /// it is uncommonly used.
664 std::unique_ptr<OverriddenFilesInfoTy> OverriddenFilesInfo;
665
getOverriddenFilesInfo()666 OverriddenFilesInfoTy &getOverriddenFilesInfo() {
667 if (!OverriddenFilesInfo)
668 OverriddenFilesInfo.reset(new OverriddenFilesInfoTy);
669 return *OverriddenFilesInfo;
670 }
671
672 /// Information about various memory buffers that we have read in.
673 ///
674 /// All FileEntry* within the stored ContentCache objects are NULL,
675 /// as they do not refer to a file.
676 std::vector<SrcMgr::ContentCache*> MemBufferInfos;
677
678 /// The table of SLocEntries that are local to this module.
679 ///
680 /// Positive FileIDs are indexes into this table. Entry 0 indicates an invalid
681 /// expansion.
682 SmallVector<SrcMgr::SLocEntry, 0> LocalSLocEntryTable;
683
684 /// The table of SLocEntries that are loaded from other modules.
685 ///
686 /// Negative FileIDs are indexes into this table. To get from ID to an index,
687 /// use (-ID - 2).
688 SmallVector<SrcMgr::SLocEntry, 0> LoadedSLocEntryTable;
689
690 /// The starting offset of the next local SLocEntry.
691 ///
692 /// This is LocalSLocEntryTable.back().Offset + the size of that entry.
693 unsigned NextLocalOffset;
694
695 /// The starting offset of the latest batch of loaded SLocEntries.
696 ///
697 /// This is LoadedSLocEntryTable.back().Offset, except that that entry might
698 /// not have been loaded, so that value would be unknown.
699 unsigned CurrentLoadedOffset;
700
701 /// The highest possible offset is 2^31-1, so CurrentLoadedOffset
702 /// starts at 2^31.
703 static const unsigned MaxLoadedOffset = 1U << 31U;
704
705 /// A bitmap that indicates whether the entries of LoadedSLocEntryTable
706 /// have already been loaded from the external source.
707 ///
708 /// Same indexing as LoadedSLocEntryTable.
709 llvm::BitVector SLocEntryLoaded;
710
711 /// An external source for source location entries.
712 ExternalSLocEntrySource *ExternalSLocEntries = nullptr;
713
714 /// A one-entry cache to speed up getFileID.
715 ///
716 /// LastFileIDLookup records the last FileID looked up or created, because it
717 /// is very common to look up many tokens from the same file.
718 mutable FileID LastFileIDLookup;
719
720 /// Holds information for \#line directives.
721 ///
722 /// This is referenced by indices from SLocEntryTable.
723 std::unique_ptr<LineTableInfo> LineTable;
724
725 /// These ivars serve as a cache used in the getLineNumber
726 /// method which is used to speedup getLineNumber calls to nearby locations.
727 mutable FileID LastLineNoFileIDQuery;
728 mutable const SrcMgr::ContentCache *LastLineNoContentCache;
729 mutable unsigned LastLineNoFilePos;
730 mutable unsigned LastLineNoResult;
731
732 /// The file ID for the main source file of the translation unit.
733 FileID MainFileID;
734
735 /// The file ID for the precompiled preamble there is one.
736 FileID PreambleFileID;
737
738 // Statistics for -print-stats.
739 mutable unsigned NumLinearScans = 0;
740 mutable unsigned NumBinaryProbes = 0;
741
742 /// Associates a FileID with its "included/expanded in" decomposed
743 /// location.
744 ///
745 /// Used to cache results from and speed-up \c getDecomposedIncludedLoc
746 /// function.
747 mutable llvm::DenseMap<FileID, std::pair<FileID, unsigned>> IncludedLocMap;
748
749 /// The key value into the IsBeforeInTUCache table.
750 using IsBeforeInTUCacheKey = std::pair<FileID, FileID>;
751
752 /// The IsBeforeInTranslationUnitCache is a mapping from FileID pairs
753 /// to cache results.
754 using InBeforeInTUCache =
755 llvm::DenseMap<IsBeforeInTUCacheKey, InBeforeInTUCacheEntry>;
756
757 /// Cache results for the isBeforeInTranslationUnit method.
758 mutable InBeforeInTUCache IBTUCache;
759 mutable InBeforeInTUCacheEntry IBTUCacheOverflow;
760
761 /// Return the cache entry for comparing the given file IDs
762 /// for isBeforeInTranslationUnit.
763 InBeforeInTUCacheEntry &getInBeforeInTUCache(FileID LFID, FileID RFID) const;
764
765 // Cache for the "fake" buffer used for error-recovery purposes.
766 mutable std::unique_ptr<llvm::MemoryBuffer> FakeBufferForRecovery;
767
768 mutable std::unique_ptr<SrcMgr::ContentCache> FakeContentCacheForRecovery;
769
770 mutable std::unique_ptr<SrcMgr::SLocEntry> FakeSLocEntryForRecovery;
771
772 /// Lazily computed map of macro argument chunks to their expanded
773 /// source location.
774 using MacroArgsMap = std::map<unsigned, SourceLocation>;
775
776 mutable llvm::DenseMap<FileID, std::unique_ptr<MacroArgsMap>>
777 MacroArgsCacheMap;
778
779 /// The stack of modules being built, which is used to detect
780 /// cycles in the module dependency graph as modules are being built, as
781 /// well as to describe why we're rebuilding a particular module.
782 ///
783 /// There is no way to set this value from the command line. If we ever need
784 /// to do so (e.g., if on-demand module construction moves out-of-process),
785 /// we can add a cc1-level option to do so.
786 SmallVector<std::pair<std::string, FullSourceLoc>, 2> StoredModuleBuildStack;
787
788 public:
789 SourceManager(DiagnosticsEngine &Diag, FileManager &FileMgr,
790 bool UserFilesAreVolatile = false);
791 explicit SourceManager(const SourceManager &) = delete;
792 SourceManager &operator=(const SourceManager &) = delete;
793 ~SourceManager();
794
795 void clearIDTables();
796
797 /// Initialize this source manager suitably to replay the compilation
798 /// described by \p Old. Requires that \p Old outlive \p *this.
799 void initializeForReplay(const SourceManager &Old);
800
getDiagnostics()801 DiagnosticsEngine &getDiagnostics() const { return Diag; }
802
getFileManager()803 FileManager &getFileManager() const { return FileMgr; }
804
805 /// Set true if the SourceManager should report the original file name
806 /// for contents of files that were overridden by other files. Defaults to
807 /// true.
setOverridenFilesKeepOriginalName(bool value)808 void setOverridenFilesKeepOriginalName(bool value) {
809 OverridenFilesKeepOriginalName = value;
810 }
811
812 /// True if non-system source files should be treated as volatile
813 /// (likely to change while trying to use them).
userFilesAreVolatile()814 bool userFilesAreVolatile() const { return UserFilesAreVolatile; }
815
816 /// Retrieve the module build stack.
getModuleBuildStack()817 ModuleBuildStack getModuleBuildStack() const {
818 return StoredModuleBuildStack;
819 }
820
821 /// Set the module build stack.
setModuleBuildStack(ModuleBuildStack stack)822 void setModuleBuildStack(ModuleBuildStack stack) {
823 StoredModuleBuildStack.clear();
824 StoredModuleBuildStack.append(stack.begin(), stack.end());
825 }
826
827 /// Push an entry to the module build stack.
pushModuleBuildStack(StringRef moduleName,FullSourceLoc importLoc)828 void pushModuleBuildStack(StringRef moduleName, FullSourceLoc importLoc) {
829 StoredModuleBuildStack.push_back(std::make_pair(moduleName.str(),importLoc));
830 }
831
832 //===--------------------------------------------------------------------===//
833 // MainFileID creation and querying methods.
834 //===--------------------------------------------------------------------===//
835
836 /// Returns the FileID of the main source file.
getMainFileID()837 FileID getMainFileID() const { return MainFileID; }
838
839 /// Set the file ID for the main source file.
setMainFileID(FileID FID)840 void setMainFileID(FileID FID) {
841 MainFileID = FID;
842 }
843
844 /// Returns true when the given FileEntry corresponds to the main file.
845 ///
846 /// The main file should be set prior to calling this function.
847 bool isMainFile(const FileEntry &SourceFile);
848
849 /// Set the file ID for the precompiled preamble.
setPreambleFileID(FileID Preamble)850 void setPreambleFileID(FileID Preamble) {
851 assert(PreambleFileID.isInvalid() && "PreambleFileID already set!");
852 PreambleFileID = Preamble;
853 }
854
855 /// Get the file ID for the precompiled preamble if there is one.
getPreambleFileID()856 FileID getPreambleFileID() const { return PreambleFileID; }
857
858 //===--------------------------------------------------------------------===//
859 // Methods to create new FileID's and macro expansions.
860 //===--------------------------------------------------------------------===//
861
862 /// Create a new FileID that represents the specified file
863 /// being \#included from the specified IncludePosition.
864 ///
865 /// This translates NULL into standard input.
866 FileID createFileID(const FileEntry *SourceFile, SourceLocation IncludePos,
867 SrcMgr::CharacteristicKind FileCharacter,
868 int LoadedID = 0, unsigned LoadedOffset = 0);
869
870 FileID createFileID(FileEntryRef SourceFile, SourceLocation IncludePos,
871 SrcMgr::CharacteristicKind FileCharacter,
872 int LoadedID = 0, unsigned LoadedOffset = 0);
873
874 /// Create a new FileID that represents the specified memory buffer.
875 ///
876 /// This does no caching of the buffer and takes ownership of the
877 /// MemoryBuffer, so only pass a MemoryBuffer to this once.
878 FileID createFileID(std::unique_ptr<llvm::MemoryBuffer> Buffer,
879 SrcMgr::CharacteristicKind FileCharacter = SrcMgr::C_User,
880 int LoadedID = 0, unsigned LoadedOffset = 0,
881 SourceLocation IncludeLoc = SourceLocation());
882
883 /// Create a new FileID that represents the specified memory buffer.
884 ///
885 /// This does not take ownership of the MemoryBuffer. The memory buffer must
886 /// outlive the SourceManager.
887 FileID createFileID(const llvm::MemoryBufferRef &Buffer,
888 SrcMgr::CharacteristicKind FileCharacter = SrcMgr::C_User,
889 int LoadedID = 0, unsigned LoadedOffset = 0,
890 SourceLocation IncludeLoc = SourceLocation());
891
892 /// Get the FileID for \p SourceFile if it exists. Otherwise, create a
893 /// new FileID for the \p SourceFile.
894 FileID getOrCreateFileID(const FileEntry *SourceFile,
895 SrcMgr::CharacteristicKind FileCharacter);
896
897 /// Return a new SourceLocation that encodes the
898 /// fact that a token from SpellingLoc should actually be referenced from
899 /// ExpansionLoc, and that it represents the expansion of a macro argument
900 /// into the function-like macro body.
901 SourceLocation createMacroArgExpansionLoc(SourceLocation Loc,
902 SourceLocation ExpansionLoc,
903 unsigned TokLength);
904
905 /// Return a new SourceLocation that encodes the fact
906 /// that a token from SpellingLoc should actually be referenced from
907 /// ExpansionLoc.
908 SourceLocation createExpansionLoc(SourceLocation Loc,
909 SourceLocation ExpansionLocStart,
910 SourceLocation ExpansionLocEnd,
911 unsigned TokLength,
912 bool ExpansionIsTokenRange = true,
913 int LoadedID = 0,
914 unsigned LoadedOffset = 0);
915
916 /// Return a new SourceLocation that encodes that the token starting
917 /// at \p TokenStart ends prematurely at \p TokenEnd.
918 SourceLocation createTokenSplitLoc(SourceLocation SpellingLoc,
919 SourceLocation TokenStart,
920 SourceLocation TokenEnd);
921
922 /// Retrieve the memory buffer associated with the given file.
923 ///
924 /// Returns None if the buffer is not valid.
925 llvm::Optional<llvm::MemoryBufferRef>
926 getMemoryBufferForFileOrNone(const FileEntry *File);
927
928 /// Retrieve the memory buffer associated with the given file.
929 ///
930 /// Returns a fake buffer if there isn't a real one.
getMemoryBufferForFileOrFake(const FileEntry * File)931 llvm::MemoryBufferRef getMemoryBufferForFileOrFake(const FileEntry *File) {
932 if (auto B = getMemoryBufferForFileOrNone(File))
933 return *B;
934 return getFakeBufferForRecovery();
935 }
936
937 /// Override the contents of the given source file by providing an
938 /// already-allocated buffer.
939 ///
940 /// \param SourceFile the source file whose contents will be overridden.
941 ///
942 /// \param Buffer the memory buffer whose contents will be used as the
943 /// data in the given source file.
overrideFileContents(const FileEntry * SourceFile,const llvm::MemoryBufferRef & Buffer)944 void overrideFileContents(const FileEntry *SourceFile,
945 const llvm::MemoryBufferRef &Buffer) {
946 overrideFileContents(SourceFile, llvm::MemoryBuffer::getMemBuffer(Buffer));
947 }
948
949 /// Override the contents of the given source file by providing an
950 /// already-allocated buffer.
951 ///
952 /// \param SourceFile the source file whose contents will be overridden.
953 ///
954 /// \param Buffer the memory buffer whose contents will be used as the
955 /// data in the given source file.
956 void overrideFileContents(const FileEntry *SourceFile,
957 std::unique_ptr<llvm::MemoryBuffer> Buffer);
overrideFileContents(FileEntryRef SourceFile,std::unique_ptr<llvm::MemoryBuffer> Buffer)958 void overrideFileContents(FileEntryRef SourceFile,
959 std::unique_ptr<llvm::MemoryBuffer> Buffer) {
960 overrideFileContents(&SourceFile.getFileEntry(), std::move(Buffer));
961 }
962
963 /// Override the given source file with another one.
964 ///
965 /// \param SourceFile the source file which will be overridden.
966 ///
967 /// \param NewFile the file whose contents will be used as the
968 /// data instead of the contents of the given source file.
969 void overrideFileContents(const FileEntry *SourceFile,
970 const FileEntry *NewFile);
971
972 /// Returns true if the file contents have been overridden.
isFileOverridden(const FileEntry * File)973 bool isFileOverridden(const FileEntry *File) const {
974 if (OverriddenFilesInfo) {
975 if (OverriddenFilesInfo->OverriddenFilesWithBuffer.count(File))
976 return true;
977 if (OverriddenFilesInfo->OverriddenFiles.find(File) !=
978 OverriddenFilesInfo->OverriddenFiles.end())
979 return true;
980 }
981 return false;
982 }
983
984 /// Bypass the overridden contents of a file. This creates a new FileEntry
985 /// and initializes the content cache for it. Returns None if there is no
986 /// such file in the filesystem.
987 ///
988 /// This should be called before parsing has begun.
989 Optional<FileEntryRef> bypassFileContentsOverride(FileEntryRef File);
990
991 /// Specify that a file is transient.
992 void setFileIsTransient(const FileEntry *SourceFile);
993
994 /// Specify that all files that are read during this compilation are
995 /// transient.
setAllFilesAreTransient(bool Transient)996 void setAllFilesAreTransient(bool Transient) {
997 FilesAreTransient = Transient;
998 }
999
1000 //===--------------------------------------------------------------------===//
1001 // FileID manipulation methods.
1002 //===--------------------------------------------------------------------===//
1003
1004 /// Return the buffer for the specified FileID.
1005 ///
1006 /// If there is an error opening this buffer the first time, return None.
1007 llvm::Optional<llvm::MemoryBufferRef>
1008 getBufferOrNone(FileID FID, SourceLocation Loc = SourceLocation()) const {
1009 if (auto *Entry = getSLocEntryForFile(FID))
1010 return Entry->getFile().getContentCache().getBufferOrNone(
1011 Diag, getFileManager(), Loc);
1012 return None;
1013 }
1014
1015 /// Return the buffer for the specified FileID.
1016 ///
1017 /// If there is an error opening this buffer the first time, this
1018 /// manufactures a temporary buffer and returns it.
1019 llvm::MemoryBufferRef
1020 getBufferOrFake(FileID FID, SourceLocation Loc = SourceLocation()) const {
1021 if (auto B = getBufferOrNone(FID, Loc))
1022 return *B;
1023 return getFakeBufferForRecovery();
1024 }
1025
1026 /// Returns the FileEntry record for the provided FileID.
getFileEntryForID(FileID FID)1027 const FileEntry *getFileEntryForID(FileID FID) const {
1028 if (auto *Entry = getSLocEntryForFile(FID))
1029 return Entry->getFile().getContentCache().OrigEntry;
1030 return nullptr;
1031 }
1032
1033 /// Returns the FileEntryRef for the provided FileID.
getFileEntryRefForID(FileID FID)1034 Optional<FileEntryRef> getFileEntryRefForID(FileID FID) const {
1035 if (auto *Entry = getFileEntryForID(FID))
1036 return Entry->getLastRef();
1037 return None;
1038 }
1039
1040 /// Returns the filename for the provided FileID, unless it's a built-in
1041 /// buffer that's not represented by a filename.
1042 ///
1043 /// Returns None for non-files and built-in files.
1044 Optional<StringRef> getNonBuiltinFilenameForID(FileID FID) const;
1045
1046 /// Returns the FileEntry record for the provided SLocEntry.
getFileEntryForSLocEntry(const SrcMgr::SLocEntry & sloc)1047 const FileEntry *getFileEntryForSLocEntry(const SrcMgr::SLocEntry &sloc) const
1048 {
1049 return sloc.getFile().getContentCache().OrigEntry;
1050 }
1051
1052 /// Return a StringRef to the source buffer data for the
1053 /// specified FileID.
1054 ///
1055 /// \param FID The file ID whose contents will be returned.
1056 /// \param Invalid If non-NULL, will be set true if an error occurred.
1057 StringRef getBufferData(FileID FID, bool *Invalid = nullptr) const;
1058
1059 /// Return a StringRef to the source buffer data for the
1060 /// specified FileID, returning None if invalid.
1061 ///
1062 /// \param FID The file ID whose contents will be returned.
1063 llvm::Optional<StringRef> getBufferDataOrNone(FileID FID) const;
1064
1065 /// Return a StringRef to the source buffer data for the
1066 /// specified FileID, returning None if it's not yet loaded.
1067 ///
1068 /// \param FID The file ID whose contents will be returned.
1069 llvm::Optional<StringRef> getBufferDataIfLoaded(FileID FID) const;
1070
1071 /// Get the number of FileIDs (files and macros) that were created
1072 /// during preprocessing of \p FID, including it.
getNumCreatedFIDsForFileID(FileID FID)1073 unsigned getNumCreatedFIDsForFileID(FileID FID) const {
1074 if (auto *Entry = getSLocEntryForFile(FID))
1075 return Entry->getFile().NumCreatedFIDs;
1076 return 0;
1077 }
1078
1079 /// Set the number of FileIDs (files and macros) that were created
1080 /// during preprocessing of \p FID, including it.
1081 void setNumCreatedFIDsForFileID(FileID FID, unsigned NumFIDs,
1082 bool Force = false) const {
1083 auto *Entry = getSLocEntryForFile(FID);
1084 if (!Entry)
1085 return;
1086 assert((Force || Entry->getFile().NumCreatedFIDs == 0) && "Already set!");
1087 const_cast<SrcMgr::FileInfo &>(Entry->getFile()).NumCreatedFIDs = NumFIDs;
1088 }
1089
1090 //===--------------------------------------------------------------------===//
1091 // SourceLocation manipulation methods.
1092 //===--------------------------------------------------------------------===//
1093
1094 /// Return the FileID for a SourceLocation.
1095 ///
1096 /// This is a very hot method that is used for all SourceManager queries
1097 /// that start with a SourceLocation object. It is responsible for finding
1098 /// the entry in SLocEntryTable which contains the specified location.
1099 ///
getFileID(SourceLocation SpellingLoc)1100 FileID getFileID(SourceLocation SpellingLoc) const {
1101 unsigned SLocOffset = SpellingLoc.getOffset();
1102
1103 // If our one-entry cache covers this offset, just return it.
1104 if (isOffsetInFileID(LastFileIDLookup, SLocOffset))
1105 return LastFileIDLookup;
1106
1107 return getFileIDSlow(SLocOffset);
1108 }
1109
1110 /// Return the filename of the file containing a SourceLocation.
1111 StringRef getFilename(SourceLocation SpellingLoc) const;
1112
1113 /// Return the source location corresponding to the first byte of
1114 /// the specified file.
getLocForStartOfFile(FileID FID)1115 SourceLocation getLocForStartOfFile(FileID FID) const {
1116 if (auto *Entry = getSLocEntryForFile(FID))
1117 return SourceLocation::getFileLoc(Entry->getOffset());
1118 return SourceLocation();
1119 }
1120
1121 /// Return the source location corresponding to the last byte of the
1122 /// specified file.
getLocForEndOfFile(FileID FID)1123 SourceLocation getLocForEndOfFile(FileID FID) const {
1124 if (auto *Entry = getSLocEntryForFile(FID))
1125 return SourceLocation::getFileLoc(Entry->getOffset() +
1126 getFileIDSize(FID));
1127 return SourceLocation();
1128 }
1129
1130 /// Returns the include location if \p FID is a \#include'd file
1131 /// otherwise it returns an invalid location.
getIncludeLoc(FileID FID)1132 SourceLocation getIncludeLoc(FileID FID) const {
1133 if (auto *Entry = getSLocEntryForFile(FID))
1134 return Entry->getFile().getIncludeLoc();
1135 return SourceLocation();
1136 }
1137
1138 // Returns the import location if the given source location is
1139 // located within a module, or an invalid location if the source location
1140 // is within the current translation unit.
1141 std::pair<SourceLocation, StringRef>
getModuleImportLoc(SourceLocation Loc)1142 getModuleImportLoc(SourceLocation Loc) const {
1143 FileID FID = getFileID(Loc);
1144
1145 // Positive file IDs are in the current translation unit, and -1 is a
1146 // placeholder.
1147 if (FID.ID >= -1)
1148 return std::make_pair(SourceLocation(), "");
1149
1150 return ExternalSLocEntries->getModuleImportLoc(FID.ID);
1151 }
1152
1153 /// Given a SourceLocation object \p Loc, return the expansion
1154 /// location referenced by the ID.
getExpansionLoc(SourceLocation Loc)1155 SourceLocation getExpansionLoc(SourceLocation Loc) const {
1156 // Handle the non-mapped case inline, defer to out of line code to handle
1157 // expansions.
1158 if (Loc.isFileID()) return Loc;
1159 return getExpansionLocSlowCase(Loc);
1160 }
1161
1162 /// Given \p Loc, if it is a macro location return the expansion
1163 /// location or the spelling location, depending on if it comes from a
1164 /// macro argument or not.
getFileLoc(SourceLocation Loc)1165 SourceLocation getFileLoc(SourceLocation Loc) const {
1166 if (Loc.isFileID()) return Loc;
1167 return getFileLocSlowCase(Loc);
1168 }
1169
1170 /// Return the start/end of the expansion information for an
1171 /// expansion location.
1172 ///
1173 /// \pre \p Loc is required to be an expansion location.
1174 CharSourceRange getImmediateExpansionRange(SourceLocation Loc) const;
1175
1176 /// Given a SourceLocation object, return the range of
1177 /// tokens covered by the expansion in the ultimate file.
1178 CharSourceRange getExpansionRange(SourceLocation Loc) const;
1179
1180 /// Given a SourceRange object, return the range of
1181 /// tokens or characters covered by the expansion in the ultimate file.
getExpansionRange(SourceRange Range)1182 CharSourceRange getExpansionRange(SourceRange Range) const {
1183 SourceLocation Begin = getExpansionRange(Range.getBegin()).getBegin();
1184 CharSourceRange End = getExpansionRange(Range.getEnd());
1185 return CharSourceRange(SourceRange(Begin, End.getEnd()),
1186 End.isTokenRange());
1187 }
1188
1189 /// Given a CharSourceRange object, return the range of
1190 /// tokens or characters covered by the expansion in the ultimate file.
getExpansionRange(CharSourceRange Range)1191 CharSourceRange getExpansionRange(CharSourceRange Range) const {
1192 CharSourceRange Expansion = getExpansionRange(Range.getAsRange());
1193 if (Expansion.getEnd() == Range.getEnd())
1194 Expansion.setTokenRange(Range.isTokenRange());
1195 return Expansion;
1196 }
1197
1198 /// Given a SourceLocation object, return the spelling
1199 /// location referenced by the ID.
1200 ///
1201 /// This is the place where the characters that make up the lexed token
1202 /// can be found.
getSpellingLoc(SourceLocation Loc)1203 SourceLocation getSpellingLoc(SourceLocation Loc) const {
1204 // Handle the non-mapped case inline, defer to out of line code to handle
1205 // expansions.
1206 if (Loc.isFileID()) return Loc;
1207 return getSpellingLocSlowCase(Loc);
1208 }
1209
1210 /// Given a SourceLocation object, return the spelling location
1211 /// referenced by the ID.
1212 ///
1213 /// This is the first level down towards the place where the characters
1214 /// that make up the lexed token can be found. This should not generally
1215 /// be used by clients.
1216 SourceLocation getImmediateSpellingLoc(SourceLocation Loc) const;
1217
1218 /// Form a SourceLocation from a FileID and Offset pair.
getComposedLoc(FileID FID,unsigned Offset)1219 SourceLocation getComposedLoc(FileID FID, unsigned Offset) const {
1220 auto *Entry = getSLocEntryOrNull(FID);
1221 if (!Entry)
1222 return SourceLocation();
1223
1224 unsigned GlobalOffset = Entry->getOffset() + Offset;
1225 return Entry->isFile() ? SourceLocation::getFileLoc(GlobalOffset)
1226 : SourceLocation::getMacroLoc(GlobalOffset);
1227 }
1228
1229 /// Decompose the specified location into a raw FileID + Offset pair.
1230 ///
1231 /// The first element is the FileID, the second is the offset from the
1232 /// start of the buffer of the location.
getDecomposedLoc(SourceLocation Loc)1233 std::pair<FileID, unsigned> getDecomposedLoc(SourceLocation Loc) const {
1234 FileID FID = getFileID(Loc);
1235 auto *Entry = getSLocEntryOrNull(FID);
1236 if (!Entry)
1237 return std::make_pair(FileID(), 0);
1238 return std::make_pair(FID, Loc.getOffset() - Entry->getOffset());
1239 }
1240
1241 /// Decompose the specified location into a raw FileID + Offset pair.
1242 ///
1243 /// If the location is an expansion record, walk through it until we find
1244 /// the final location expanded.
1245 std::pair<FileID, unsigned>
getDecomposedExpansionLoc(SourceLocation Loc)1246 getDecomposedExpansionLoc(SourceLocation Loc) const {
1247 FileID FID = getFileID(Loc);
1248 auto *E = getSLocEntryOrNull(FID);
1249 if (!E)
1250 return std::make_pair(FileID(), 0);
1251
1252 unsigned Offset = Loc.getOffset()-E->getOffset();
1253 if (Loc.isFileID())
1254 return std::make_pair(FID, Offset);
1255
1256 return getDecomposedExpansionLocSlowCase(E);
1257 }
1258
1259 /// Decompose the specified location into a raw FileID + Offset pair.
1260 ///
1261 /// If the location is an expansion record, walk through it until we find
1262 /// its spelling record.
1263 std::pair<FileID, unsigned>
getDecomposedSpellingLoc(SourceLocation Loc)1264 getDecomposedSpellingLoc(SourceLocation Loc) const {
1265 FileID FID = getFileID(Loc);
1266 auto *E = getSLocEntryOrNull(FID);
1267 if (!E)
1268 return std::make_pair(FileID(), 0);
1269
1270 unsigned Offset = Loc.getOffset()-E->getOffset();
1271 if (Loc.isFileID())
1272 return std::make_pair(FID, Offset);
1273 return getDecomposedSpellingLocSlowCase(E, Offset);
1274 }
1275
1276 /// Returns the "included/expanded in" decomposed location of the given
1277 /// FileID.
1278 std::pair<FileID, unsigned> getDecomposedIncludedLoc(FileID FID) const;
1279
1280 /// Returns the offset from the start of the file that the
1281 /// specified SourceLocation represents.
1282 ///
1283 /// This is not very meaningful for a macro ID.
getFileOffset(SourceLocation SpellingLoc)1284 unsigned getFileOffset(SourceLocation SpellingLoc) const {
1285 return getDecomposedLoc(SpellingLoc).second;
1286 }
1287
1288 /// Tests whether the given source location represents a macro
1289 /// argument's expansion into the function-like macro definition.
1290 ///
1291 /// \param StartLoc If non-null and function returns true, it is set to the
1292 /// start location of the macro argument expansion.
1293 ///
1294 /// Such source locations only appear inside of the expansion
1295 /// locations representing where a particular function-like macro was
1296 /// expanded.
1297 bool isMacroArgExpansion(SourceLocation Loc,
1298 SourceLocation *StartLoc = nullptr) const;
1299
1300 /// Tests whether the given source location represents the expansion of
1301 /// a macro body.
1302 ///
1303 /// This is equivalent to testing whether the location is part of a macro
1304 /// expansion but not the expansion of an argument to a function-like macro.
1305 bool isMacroBodyExpansion(SourceLocation Loc) const;
1306
1307 /// Returns true if the given MacroID location points at the beginning
1308 /// of the immediate macro expansion.
1309 ///
1310 /// \param MacroBegin If non-null and function returns true, it is set to the
1311 /// begin location of the immediate macro expansion.
1312 bool isAtStartOfImmediateMacroExpansion(SourceLocation Loc,
1313 SourceLocation *MacroBegin = nullptr) const;
1314
1315 /// Returns true if the given MacroID location points at the character
1316 /// end of the immediate macro expansion.
1317 ///
1318 /// \param MacroEnd If non-null and function returns true, it is set to the
1319 /// character end location of the immediate macro expansion.
1320 bool
1321 isAtEndOfImmediateMacroExpansion(SourceLocation Loc,
1322 SourceLocation *MacroEnd = nullptr) const;
1323
1324 /// Returns true if \p Loc is inside the [\p Start, +\p Length)
1325 /// chunk of the source location address space.
1326 ///
1327 /// If it's true and \p RelativeOffset is non-null, it will be set to the
1328 /// relative offset of \p Loc inside the chunk.
1329 bool isInSLocAddrSpace(SourceLocation Loc,
1330 SourceLocation Start, unsigned Length,
1331 unsigned *RelativeOffset = nullptr) const {
1332 assert(((Start.getOffset() < NextLocalOffset &&
1333 Start.getOffset()+Length <= NextLocalOffset) ||
1334 (Start.getOffset() >= CurrentLoadedOffset &&
1335 Start.getOffset()+Length < MaxLoadedOffset)) &&
1336 "Chunk is not valid SLoc address space");
1337 unsigned LocOffs = Loc.getOffset();
1338 unsigned BeginOffs = Start.getOffset();
1339 unsigned EndOffs = BeginOffs + Length;
1340 if (LocOffs >= BeginOffs && LocOffs < EndOffs) {
1341 if (RelativeOffset)
1342 *RelativeOffset = LocOffs - BeginOffs;
1343 return true;
1344 }
1345
1346 return false;
1347 }
1348
1349 /// Return true if both \p LHS and \p RHS are in the local source
1350 /// location address space or the loaded one.
1351 ///
1352 /// If it's true and \p RelativeOffset is non-null, it will be set to the
1353 /// offset of \p RHS relative to \p LHS.
isInSameSLocAddrSpace(SourceLocation LHS,SourceLocation RHS,int * RelativeOffset)1354 bool isInSameSLocAddrSpace(SourceLocation LHS, SourceLocation RHS,
1355 int *RelativeOffset) const {
1356 unsigned LHSOffs = LHS.getOffset(), RHSOffs = RHS.getOffset();
1357 bool LHSLoaded = LHSOffs >= CurrentLoadedOffset;
1358 bool RHSLoaded = RHSOffs >= CurrentLoadedOffset;
1359
1360 if (LHSLoaded == RHSLoaded) {
1361 if (RelativeOffset)
1362 *RelativeOffset = RHSOffs - LHSOffs;
1363 return true;
1364 }
1365
1366 return false;
1367 }
1368
1369 //===--------------------------------------------------------------------===//
1370 // Queries about the code at a SourceLocation.
1371 //===--------------------------------------------------------------------===//
1372
1373 /// Return a pointer to the start of the specified location
1374 /// in the appropriate spelling MemoryBuffer.
1375 ///
1376 /// \param Invalid If non-NULL, will be set \c true if an error occurs.
1377 const char *getCharacterData(SourceLocation SL,
1378 bool *Invalid = nullptr) const;
1379
1380 /// Return the column # for the specified file position.
1381 ///
1382 /// This is significantly cheaper to compute than the line number. This
1383 /// returns zero if the column number isn't known. This may only be called
1384 /// on a file sloc, so you must choose a spelling or expansion location
1385 /// before calling this method.
1386 unsigned getColumnNumber(FileID FID, unsigned FilePos,
1387 bool *Invalid = nullptr) const;
1388 unsigned getSpellingColumnNumber(SourceLocation Loc,
1389 bool *Invalid = nullptr) const;
1390 unsigned getExpansionColumnNumber(SourceLocation Loc,
1391 bool *Invalid = nullptr) const;
1392 unsigned getPresumedColumnNumber(SourceLocation Loc,
1393 bool *Invalid = nullptr) const;
1394
1395 /// Given a SourceLocation, return the spelling line number
1396 /// for the position indicated.
1397 ///
1398 /// This requires building and caching a table of line offsets for the
1399 /// MemoryBuffer, so this is not cheap: use only when about to emit a
1400 /// diagnostic.
1401 unsigned getLineNumber(FileID FID, unsigned FilePos, bool *Invalid = nullptr) const;
1402 unsigned getSpellingLineNumber(SourceLocation Loc, bool *Invalid = nullptr) const;
1403 unsigned getExpansionLineNumber(SourceLocation Loc, bool *Invalid = nullptr) const;
1404 unsigned getPresumedLineNumber(SourceLocation Loc, bool *Invalid = nullptr) const;
1405
1406 /// Return the filename or buffer identifier of the buffer the
1407 /// location is in.
1408 ///
1409 /// Note that this name does not respect \#line directives. Use
1410 /// getPresumedLoc for normal clients.
1411 StringRef getBufferName(SourceLocation Loc, bool *Invalid = nullptr) const;
1412
1413 /// Return the file characteristic of the specified source
1414 /// location, indicating whether this is a normal file, a system
1415 /// header, or an "implicit extern C" system header.
1416 ///
1417 /// This state can be modified with flags on GNU linemarker directives like:
1418 /// \code
1419 /// # 4 "foo.h" 3
1420 /// \endcode
1421 /// which changes all source locations in the current file after that to be
1422 /// considered to be from a system header.
1423 SrcMgr::CharacteristicKind getFileCharacteristic(SourceLocation Loc) const;
1424
1425 /// Returns the "presumed" location of a SourceLocation specifies.
1426 ///
1427 /// A "presumed location" can be modified by \#line or GNU line marker
1428 /// directives. This provides a view on the data that a user should see
1429 /// in diagnostics, for example.
1430 ///
1431 /// Note that a presumed location is always given as the expansion point of
1432 /// an expansion location, not at the spelling location.
1433 ///
1434 /// \returns The presumed location of the specified SourceLocation. If the
1435 /// presumed location cannot be calculated (e.g., because \p Loc is invalid
1436 /// or the file containing \p Loc has changed on disk), returns an invalid
1437 /// presumed location.
1438 PresumedLoc getPresumedLoc(SourceLocation Loc,
1439 bool UseLineDirectives = true) const;
1440
1441 /// Returns whether the PresumedLoc for a given SourceLocation is
1442 /// in the main file.
1443 ///
1444 /// This computes the "presumed" location for a SourceLocation, then checks
1445 /// whether it came from a file other than the main file. This is different
1446 /// from isWrittenInMainFile() because it takes line marker directives into
1447 /// account.
1448 bool isInMainFile(SourceLocation Loc) const;
1449
1450 /// Returns true if the spelling locations for both SourceLocations
1451 /// are part of the same file buffer.
1452 ///
1453 /// This check ignores line marker directives.
isWrittenInSameFile(SourceLocation Loc1,SourceLocation Loc2)1454 bool isWrittenInSameFile(SourceLocation Loc1, SourceLocation Loc2) const {
1455 return getFileID(Loc1) == getFileID(Loc2);
1456 }
1457
1458 /// Returns true if the spelling location for the given location
1459 /// is in the main file buffer.
1460 ///
1461 /// This check ignores line marker directives.
isWrittenInMainFile(SourceLocation Loc)1462 bool isWrittenInMainFile(SourceLocation Loc) const {
1463 return getFileID(Loc) == getMainFileID();
1464 }
1465
1466 /// Returns whether \p Loc is located in a <built-in> file.
isWrittenInBuiltinFile(SourceLocation Loc)1467 bool isWrittenInBuiltinFile(SourceLocation Loc) const {
1468 StringRef Filename(getPresumedLoc(Loc).getFilename());
1469 return Filename.equals("<built-in>");
1470 }
1471
1472 /// Returns whether \p Loc is located in a <command line> file.
isWrittenInCommandLineFile(SourceLocation Loc)1473 bool isWrittenInCommandLineFile(SourceLocation Loc) const {
1474 StringRef Filename(getPresumedLoc(Loc).getFilename());
1475 return Filename.equals("<command line>");
1476 }
1477
1478 /// Returns whether \p Loc is located in a <scratch space> file.
isWrittenInScratchSpace(SourceLocation Loc)1479 bool isWrittenInScratchSpace(SourceLocation Loc) const {
1480 StringRef Filename(getPresumedLoc(Loc).getFilename());
1481 return Filename.equals("<scratch space>");
1482 }
1483
1484 /// Returns if a SourceLocation is in a system header.
isInSystemHeader(SourceLocation Loc)1485 bool isInSystemHeader(SourceLocation Loc) const {
1486 return isSystem(getFileCharacteristic(Loc));
1487 }
1488
1489 /// Returns if a SourceLocation is in an "extern C" system header.
isInExternCSystemHeader(SourceLocation Loc)1490 bool isInExternCSystemHeader(SourceLocation Loc) const {
1491 return getFileCharacteristic(Loc) == SrcMgr::C_ExternCSystem;
1492 }
1493
1494 /// Returns whether \p Loc is expanded from a macro in a system header.
isInSystemMacro(SourceLocation loc)1495 bool isInSystemMacro(SourceLocation loc) const {
1496 if (!loc.isMacroID())
1497 return false;
1498
1499 // This happens when the macro is the result of a paste, in that case
1500 // its spelling is the scratch memory, so we take the parent context.
1501 // There can be several level of token pasting.
1502 if (isWrittenInScratchSpace(getSpellingLoc(loc))) {
1503 do {
1504 loc = getImmediateMacroCallerLoc(loc);
1505 } while (isWrittenInScratchSpace(getSpellingLoc(loc)));
1506 return isInSystemMacro(loc);
1507 }
1508
1509 return isInSystemHeader(getSpellingLoc(loc));
1510 }
1511
1512 /// The size of the SLocEntry that \p FID represents.
1513 unsigned getFileIDSize(FileID FID) const;
1514
1515 /// Given a specific FileID, returns true if \p Loc is inside that
1516 /// FileID chunk and sets relative offset (offset of \p Loc from beginning
1517 /// of FileID) to \p relativeOffset.
1518 bool isInFileID(SourceLocation Loc, FileID FID,
1519 unsigned *RelativeOffset = nullptr) const {
1520 unsigned Offs = Loc.getOffset();
1521 if (isOffsetInFileID(FID, Offs)) {
1522 if (RelativeOffset)
1523 *RelativeOffset = Offs - getSLocEntry(FID).getOffset();
1524 return true;
1525 }
1526
1527 return false;
1528 }
1529
1530 //===--------------------------------------------------------------------===//
1531 // Line Table Manipulation Routines
1532 //===--------------------------------------------------------------------===//
1533
1534 /// Return the uniqued ID for the specified filename.
1535 unsigned getLineTableFilenameID(StringRef Str);
1536
1537 /// Add a line note to the line table for the FileID and offset
1538 /// specified by Loc.
1539 ///
1540 /// If FilenameID is -1, it is considered to be unspecified.
1541 void AddLineNote(SourceLocation Loc, unsigned LineNo, int FilenameID,
1542 bool IsFileEntry, bool IsFileExit,
1543 SrcMgr::CharacteristicKind FileKind);
1544
1545 /// Determine if the source manager has a line table.
hasLineTable()1546 bool hasLineTable() const { return LineTable != nullptr; }
1547
1548 /// Retrieve the stored line table.
1549 LineTableInfo &getLineTable();
1550
1551 //===--------------------------------------------------------------------===//
1552 // Queries for performance analysis.
1553 //===--------------------------------------------------------------------===//
1554
1555 /// Return the total amount of physical memory allocated by the
1556 /// ContentCache allocator.
getContentCacheSize()1557 size_t getContentCacheSize() const {
1558 return ContentCacheAlloc.getTotalMemory();
1559 }
1560
1561 struct MemoryBufferSizes {
1562 const size_t malloc_bytes;
1563 const size_t mmap_bytes;
1564
MemoryBufferSizesMemoryBufferSizes1565 MemoryBufferSizes(size_t malloc_bytes, size_t mmap_bytes)
1566 : malloc_bytes(malloc_bytes), mmap_bytes(mmap_bytes) {}
1567 };
1568
1569 /// Return the amount of memory used by memory buffers, breaking down
1570 /// by heap-backed versus mmap'ed memory.
1571 MemoryBufferSizes getMemoryBufferSizes() const;
1572
1573 /// Return the amount of memory used for various side tables and
1574 /// data structures in the SourceManager.
1575 size_t getDataStructureSizes() const;
1576
1577 //===--------------------------------------------------------------------===//
1578 // Other miscellaneous methods.
1579 //===--------------------------------------------------------------------===//
1580
1581 /// Get the source location for the given file:line:col triplet.
1582 ///
1583 /// If the source file is included multiple times, the source location will
1584 /// be based upon the first inclusion.
1585 SourceLocation translateFileLineCol(const FileEntry *SourceFile,
1586 unsigned Line, unsigned Col) const;
1587
1588 /// Get the FileID for the given file.
1589 ///
1590 /// If the source file is included multiple times, the FileID will be the
1591 /// first inclusion.
1592 FileID translateFile(const FileEntry *SourceFile) const;
translateFile(FileEntryRef SourceFile)1593 FileID translateFile(FileEntryRef SourceFile) const {
1594 return translateFile(&SourceFile.getFileEntry());
1595 }
1596
1597 /// Get the source location in \p FID for the given line:col.
1598 /// Returns null location if \p FID is not a file SLocEntry.
1599 SourceLocation translateLineCol(FileID FID,
1600 unsigned Line, unsigned Col) const;
1601
1602 /// If \p Loc points inside a function macro argument, the returned
1603 /// location will be the macro location in which the argument was expanded.
1604 /// If a macro argument is used multiple times, the expanded location will
1605 /// be at the first expansion of the argument.
1606 /// e.g.
1607 /// MY_MACRO(foo);
1608 /// ^
1609 /// Passing a file location pointing at 'foo', will yield a macro location
1610 /// where 'foo' was expanded into.
1611 SourceLocation getMacroArgExpandedLocation(SourceLocation Loc) const;
1612
1613 /// Determines the order of 2 source locations in the translation unit.
1614 ///
1615 /// \returns true if LHS source location comes before RHS, false otherwise.
1616 bool isBeforeInTranslationUnit(SourceLocation LHS, SourceLocation RHS) const;
1617
1618 /// Determines whether the two decomposed source location is in the
1619 /// same translation unit. As a byproduct, it also calculates the order
1620 /// of the source locations in case they are in the same TU.
1621 ///
1622 /// \returns Pair of bools the first component is true if the two locations
1623 /// are in the same TU. The second bool is true if the first is true
1624 /// and \p LOffs is before \p ROffs.
1625 std::pair<bool, bool>
1626 isInTheSameTranslationUnit(std::pair<FileID, unsigned> &LOffs,
1627 std::pair<FileID, unsigned> &ROffs) const;
1628
1629 /// Determines the order of 2 source locations in the "source location
1630 /// address space".
isBeforeInSLocAddrSpace(SourceLocation LHS,SourceLocation RHS)1631 bool isBeforeInSLocAddrSpace(SourceLocation LHS, SourceLocation RHS) const {
1632 return isBeforeInSLocAddrSpace(LHS, RHS.getOffset());
1633 }
1634
1635 /// Determines the order of a source location and a source location
1636 /// offset in the "source location address space".
1637 ///
1638 /// Note that we always consider source locations loaded from
isBeforeInSLocAddrSpace(SourceLocation LHS,unsigned RHS)1639 bool isBeforeInSLocAddrSpace(SourceLocation LHS, unsigned RHS) const {
1640 unsigned LHSOffset = LHS.getOffset();
1641 bool LHSLoaded = LHSOffset >= CurrentLoadedOffset;
1642 bool RHSLoaded = RHS >= CurrentLoadedOffset;
1643 if (LHSLoaded == RHSLoaded)
1644 return LHSOffset < RHS;
1645
1646 return LHSLoaded;
1647 }
1648
1649 /// Return true if the Point is within Start and End.
isPointWithin(SourceLocation Location,SourceLocation Start,SourceLocation End)1650 bool isPointWithin(SourceLocation Location, SourceLocation Start,
1651 SourceLocation End) const {
1652 return Location == Start || Location == End ||
1653 (isBeforeInTranslationUnit(Start, Location) &&
1654 isBeforeInTranslationUnit(Location, End));
1655 }
1656
1657 // Iterators over FileInfos.
1658 using fileinfo_iterator =
1659 llvm::DenseMap<const FileEntry*, SrcMgr::ContentCache*>::const_iterator;
1660
fileinfo_begin()1661 fileinfo_iterator fileinfo_begin() const { return FileInfos.begin(); }
fileinfo_end()1662 fileinfo_iterator fileinfo_end() const { return FileInfos.end(); }
hasFileInfo(const FileEntry * File)1663 bool hasFileInfo(const FileEntry *File) const {
1664 return FileInfos.find(File) != FileInfos.end();
1665 }
1666
1667 /// Print statistics to stderr.
1668 void PrintStats() const;
1669
1670 void dump() const;
1671
1672 /// Get the number of local SLocEntries we have.
local_sloc_entry_size()1673 unsigned local_sloc_entry_size() const { return LocalSLocEntryTable.size(); }
1674
1675 /// Get a local SLocEntry. This is exposed for indexing.
getLocalSLocEntry(unsigned Index)1676 const SrcMgr::SLocEntry &getLocalSLocEntry(unsigned Index) const {
1677 assert(Index < LocalSLocEntryTable.size() && "Invalid index");
1678 return LocalSLocEntryTable[Index];
1679 }
1680
1681 /// Get the number of loaded SLocEntries we have.
loaded_sloc_entry_size()1682 unsigned loaded_sloc_entry_size() const { return LoadedSLocEntryTable.size();}
1683
1684 /// Get a loaded SLocEntry. This is exposed for indexing.
1685 const SrcMgr::SLocEntry &getLoadedSLocEntry(unsigned Index,
1686 bool *Invalid = nullptr) const {
1687 assert(Index < LoadedSLocEntryTable.size() && "Invalid index");
1688 if (SLocEntryLoaded[Index])
1689 return LoadedSLocEntryTable[Index];
1690 return loadSLocEntry(Index, Invalid);
1691 }
1692
1693 const SrcMgr::SLocEntry &getSLocEntry(FileID FID,
1694 bool *Invalid = nullptr) const {
1695 if (FID.ID == 0 || FID.ID == -1) {
1696 if (Invalid) *Invalid = true;
1697 return LocalSLocEntryTable[0];
1698 }
1699 return getSLocEntryByID(FID.ID, Invalid);
1700 }
1701
getNextLocalOffset()1702 unsigned getNextLocalOffset() const { return NextLocalOffset; }
1703
setExternalSLocEntrySource(ExternalSLocEntrySource * Source)1704 void setExternalSLocEntrySource(ExternalSLocEntrySource *Source) {
1705 assert(LoadedSLocEntryTable.empty() &&
1706 "Invalidating existing loaded entries");
1707 ExternalSLocEntries = Source;
1708 }
1709
1710 /// Allocate a number of loaded SLocEntries, which will be actually
1711 /// loaded on demand from the external source.
1712 ///
1713 /// NumSLocEntries will be allocated, which occupy a total of TotalSize space
1714 /// in the global source view. The lowest ID and the base offset of the
1715 /// entries will be returned.
1716 std::pair<int, unsigned>
1717 AllocateLoadedSLocEntries(unsigned NumSLocEntries, unsigned TotalSize);
1718
1719 /// Returns true if \p Loc came from a PCH/Module.
isLoadedSourceLocation(SourceLocation Loc)1720 bool isLoadedSourceLocation(SourceLocation Loc) const {
1721 return Loc.getOffset() >= CurrentLoadedOffset;
1722 }
1723
1724 /// Returns true if \p Loc did not come from a PCH/Module.
isLocalSourceLocation(SourceLocation Loc)1725 bool isLocalSourceLocation(SourceLocation Loc) const {
1726 return Loc.getOffset() < NextLocalOffset;
1727 }
1728
1729 /// Returns true if \p FID came from a PCH/Module.
isLoadedFileID(FileID FID)1730 bool isLoadedFileID(FileID FID) const {
1731 assert(FID.ID != -1 && "Using FileID sentinel value");
1732 return FID.ID < 0;
1733 }
1734
1735 /// Returns true if \p FID did not come from a PCH/Module.
isLocalFileID(FileID FID)1736 bool isLocalFileID(FileID FID) const {
1737 return !isLoadedFileID(FID);
1738 }
1739
1740 /// Gets the location of the immediate macro caller, one level up the stack
1741 /// toward the initial macro typed into the source.
getImmediateMacroCallerLoc(SourceLocation Loc)1742 SourceLocation getImmediateMacroCallerLoc(SourceLocation Loc) const {
1743 if (!Loc.isMacroID()) return Loc;
1744
1745 // When we have the location of (part of) an expanded parameter, its
1746 // spelling location points to the argument as expanded in the macro call,
1747 // and therefore is used to locate the macro caller.
1748 if (isMacroArgExpansion(Loc))
1749 return getImmediateSpellingLoc(Loc);
1750
1751 // Otherwise, the caller of the macro is located where this macro is
1752 // expanded (while the spelling is part of the macro definition).
1753 return getImmediateExpansionRange(Loc).getBegin();
1754 }
1755
1756 /// \return Location of the top-level macro caller.
1757 SourceLocation getTopMacroCallerLoc(SourceLocation Loc) const;
1758
1759 private:
1760 friend class ASTReader;
1761 friend class ASTWriter;
1762
1763 llvm::MemoryBufferRef getFakeBufferForRecovery() const;
1764 SrcMgr::ContentCache &getFakeContentCacheForRecovery() const;
1765
1766 const SrcMgr::SLocEntry &loadSLocEntry(unsigned Index, bool *Invalid) const;
1767
getSLocEntryOrNull(FileID FID)1768 const SrcMgr::SLocEntry *getSLocEntryOrNull(FileID FID) const {
1769 bool Invalid = false;
1770 const SrcMgr::SLocEntry &Entry = getSLocEntry(FID, &Invalid);
1771 return Invalid ? nullptr : &Entry;
1772 }
1773
getSLocEntryForFile(FileID FID)1774 const SrcMgr::SLocEntry *getSLocEntryForFile(FileID FID) const {
1775 if (auto *Entry = getSLocEntryOrNull(FID))
1776 if (Entry->isFile())
1777 return Entry;
1778 return nullptr;
1779 }
1780
1781 /// Get the entry with the given unwrapped FileID.
1782 /// Invalid will not be modified for Local IDs.
1783 const SrcMgr::SLocEntry &getSLocEntryByID(int ID,
1784 bool *Invalid = nullptr) const {
1785 assert(ID != -1 && "Using FileID sentinel value");
1786 if (ID < 0)
1787 return getLoadedSLocEntryByID(ID, Invalid);
1788 return getLocalSLocEntry(static_cast<unsigned>(ID));
1789 }
1790
1791 const SrcMgr::SLocEntry &
1792 getLoadedSLocEntryByID(int ID, bool *Invalid = nullptr) const {
1793 return getLoadedSLocEntry(static_cast<unsigned>(-ID - 2), Invalid);
1794 }
1795
1796 /// Implements the common elements of storing an expansion info struct into
1797 /// the SLocEntry table and producing a source location that refers to it.
1798 SourceLocation createExpansionLocImpl(const SrcMgr::ExpansionInfo &Expansion,
1799 unsigned TokLength,
1800 int LoadedID = 0,
1801 unsigned LoadedOffset = 0);
1802
1803 /// Return true if the specified FileID contains the
1804 /// specified SourceLocation offset. This is a very hot method.
isOffsetInFileID(FileID FID,unsigned SLocOffset)1805 inline bool isOffsetInFileID(FileID FID, unsigned SLocOffset) const {
1806 const SrcMgr::SLocEntry &Entry = getSLocEntry(FID);
1807 // If the entry is after the offset, it can't contain it.
1808 if (SLocOffset < Entry.getOffset()) return false;
1809
1810 // If this is the very last entry then it does.
1811 if (FID.ID == -2)
1812 return true;
1813
1814 // If it is the last local entry, then it does if the location is local.
1815 if (FID.ID+1 == static_cast<int>(LocalSLocEntryTable.size()))
1816 return SLocOffset < NextLocalOffset;
1817
1818 // Otherwise, the entry after it has to not include it. This works for both
1819 // local and loaded entries.
1820 return SLocOffset < getSLocEntryByID(FID.ID+1).getOffset();
1821 }
1822
1823 /// Returns the previous in-order FileID or an invalid FileID if there
1824 /// is no previous one.
1825 FileID getPreviousFileID(FileID FID) const;
1826
1827 /// Returns the next in-order FileID or an invalid FileID if there is
1828 /// no next one.
1829 FileID getNextFileID(FileID FID) const;
1830
1831 /// Create a new fileID for the specified ContentCache and
1832 /// include position.
1833 ///
1834 /// This works regardless of whether the ContentCache corresponds to a
1835 /// file or some other input source.
1836 FileID createFileIDImpl(SrcMgr::ContentCache &File, StringRef Filename,
1837 SourceLocation IncludePos,
1838 SrcMgr::CharacteristicKind DirCharacter, int LoadedID,
1839 unsigned LoadedOffset);
1840
1841 SrcMgr::ContentCache &getOrCreateContentCache(FileEntryRef SourceFile,
1842 bool isSystemFile = false);
1843
1844 /// Create a new ContentCache for the specified memory buffer.
1845 SrcMgr::ContentCache &
1846 createMemBufferContentCache(std::unique_ptr<llvm::MemoryBuffer> Buf);
1847
1848 FileID getFileIDSlow(unsigned SLocOffset) const;
1849 FileID getFileIDLocal(unsigned SLocOffset) const;
1850 FileID getFileIDLoaded(unsigned SLocOffset) const;
1851
1852 SourceLocation getExpansionLocSlowCase(SourceLocation Loc) const;
1853 SourceLocation getSpellingLocSlowCase(SourceLocation Loc) const;
1854 SourceLocation getFileLocSlowCase(SourceLocation Loc) const;
1855
1856 std::pair<FileID, unsigned>
1857 getDecomposedExpansionLocSlowCase(const SrcMgr::SLocEntry *E) const;
1858 std::pair<FileID, unsigned>
1859 getDecomposedSpellingLocSlowCase(const SrcMgr::SLocEntry *E,
1860 unsigned Offset) const;
1861 void computeMacroArgsCache(MacroArgsMap &MacroArgsCache, FileID FID) const;
1862 void associateFileChunkWithMacroArgExp(MacroArgsMap &MacroArgsCache,
1863 FileID FID,
1864 SourceLocation SpellLoc,
1865 SourceLocation ExpansionLoc,
1866 unsigned ExpansionLength) const;
1867 };
1868
1869 /// Comparison function object.
1870 template<typename T>
1871 class BeforeThanCompare;
1872
1873 /// Compare two source locations.
1874 template<>
1875 class BeforeThanCompare<SourceLocation> {
1876 SourceManager &SM;
1877
1878 public:
BeforeThanCompare(SourceManager & SM)1879 explicit BeforeThanCompare(SourceManager &SM) : SM(SM) {}
1880
operator()1881 bool operator()(SourceLocation LHS, SourceLocation RHS) const {
1882 return SM.isBeforeInTranslationUnit(LHS, RHS);
1883 }
1884 };
1885
1886 /// Compare two non-overlapping source ranges.
1887 template<>
1888 class BeforeThanCompare<SourceRange> {
1889 SourceManager &SM;
1890
1891 public:
BeforeThanCompare(SourceManager & SM)1892 explicit BeforeThanCompare(SourceManager &SM) : SM(SM) {}
1893
operator()1894 bool operator()(SourceRange LHS, SourceRange RHS) const {
1895 return SM.isBeforeInTranslationUnit(LHS.getBegin(), RHS.getBegin());
1896 }
1897 };
1898
1899 /// SourceManager and necessary depdencies (e.g. VFS, FileManager) for a single
1900 /// in-memorty file.
1901 class SourceManagerForFile {
1902 public:
1903 /// Creates SourceManager and necessary depdencies (e.g. VFS, FileManager).
1904 /// The main file in the SourceManager will be \p FileName with \p Content.
1905 SourceManagerForFile(StringRef FileName, StringRef Content);
1906
get()1907 SourceManager &get() {
1908 assert(SourceMgr);
1909 return *SourceMgr;
1910 }
1911
1912 private:
1913 // The order of these fields are important - they should be in the same order
1914 // as they are created in `createSourceManagerForFile` so that they can be
1915 // deleted in the reverse order as they are created.
1916 std::unique_ptr<FileManager> FileMgr;
1917 std::unique_ptr<DiagnosticsEngine> Diagnostics;
1918 std::unique_ptr<SourceManager> SourceMgr;
1919 };
1920
1921 } // namespace clang
1922
1923 #endif // LLVM_CLANG_BASIC_SOURCEMANAGER_H
1924