xref: /freebsd/contrib/llvm-project/clang/include/clang/Basic/SourceManager.h (revision 700637cbb5e582861067a11aaca4d053546871d2)
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/FileManager.h"
40 #include "clang/Basic/SourceLocation.h"
41 #include "llvm/ADT/ArrayRef.h"
42 #include "llvm/ADT/BitVector.h"
43 #include "llvm/ADT/DenseMap.h"
44 #include "llvm/ADT/DenseSet.h"
45 #include "llvm/ADT/IntrusiveRefCntPtr.h"
46 #include "llvm/ADT/PagedVector.h"
47 #include "llvm/ADT/PointerIntPair.h"
48 #include "llvm/ADT/SmallVector.h"
49 #include "llvm/ADT/StringRef.h"
50 #include "llvm/Support/Allocator.h"
51 #include "llvm/Support/Compiler.h"
52 #include "llvm/Support/MemoryBuffer.h"
53 #include <cassert>
54 #include <cstddef>
55 #include <map>
56 #include <memory>
57 #include <optional>
58 #include <string>
59 #include <utility>
60 #include <vector>
61 
62 namespace clang {
63 
64 class ASTReader;
65 class ASTWriter;
66 class FileManager;
67 class LineTableInfo;
68 class SourceManager;
69 
70 /// Public enums and private classes that are part of the
71 /// SourceManager implementation.
72 namespace SrcMgr {
73 
74 /// Indicates whether a file or directory holds normal user code,
75 /// system code, or system code which is implicitly 'extern "C"' in C++ mode.
76 ///
77 /// Entire directories can be tagged with this (this is maintained by
78 /// DirectoryLookup and friends) as can specific FileInfos when a \#pragma
79 /// system_header is seen or in various other cases.
80 ///
81 enum CharacteristicKind {
82   C_User,
83   C_System,
84   C_ExternCSystem,
85   C_User_ModuleMap,
86   C_System_ModuleMap
87 };
88 
89 /// Determine whether a file / directory characteristic is for system code.
isSystem(CharacteristicKind CK)90 inline bool isSystem(CharacteristicKind CK) {
91   return CK != C_User && CK != C_User_ModuleMap;
92 }
93 
94 /// Determine whether a file characteristic is for a module map.
isModuleMap(CharacteristicKind CK)95 inline bool isModuleMap(CharacteristicKind CK) {
96   return CK == C_User_ModuleMap || CK == C_System_ModuleMap;
97 }
98 
99 /// Mapping of line offsets into a source file. This does not own the storage
100 /// for the line numbers.
101 class LineOffsetMapping {
102 public:
103   explicit operator bool() const { return Storage; }
size()104   unsigned size() const {
105     assert(Storage);
106     return Storage[0];
107   }
getLines()108   ArrayRef<unsigned> getLines() const {
109     assert(Storage);
110     return ArrayRef<unsigned>(Storage + 1, Storage + 1 + size());
111   }
begin()112   const unsigned *begin() const { return getLines().begin(); }
end()113   const unsigned *end() const { return getLines().end(); }
114   const unsigned &operator[](int I) const { return getLines()[I]; }
115 
116   static LineOffsetMapping get(llvm::MemoryBufferRef Buffer,
117                                llvm::BumpPtrAllocator &Alloc);
118 
119   LineOffsetMapping() = default;
120   LineOffsetMapping(ArrayRef<unsigned> LineOffsets,
121                     llvm::BumpPtrAllocator &Alloc);
122 
123 private:
124   /// First element is the size, followed by elements at off-by-one indexes.
125   unsigned *Storage = nullptr;
126 };
127 
128 /// One instance of this struct is kept for every file loaded or used.
129 ///
130 /// This object owns the MemoryBuffer object.
131 class alignas(8) ContentCache {
132   /// The actual buffer containing the characters from the input
133   /// file.
134   mutable std::unique_ptr<llvm::MemoryBuffer> Buffer;
135 
136 public:
137   /// Reference to the file entry representing this ContentCache.
138   ///
139   /// This reference does not own the FileEntry object.
140   ///
141   /// It is possible for this to be NULL if the ContentCache encapsulates
142   /// an imaginary text buffer.
143   ///
144   /// FIXME: Make non-optional using a virtual file as needed, remove \c
145   /// Filename and use \c OrigEntry.getNameAsRequested() instead.
146   OptionalFileEntryRef OrigEntry;
147 
148   /// References the file which the contents were actually loaded from.
149   ///
150   /// Can be different from 'Entry' if we overridden the contents of one file
151   /// with the contents of another file.
152   OptionalFileEntryRef ContentsEntry;
153 
154   /// The filename that is used to access OrigEntry.
155   ///
156   /// FIXME: Remove this once OrigEntry is a FileEntryRef with a stable name.
157   StringRef Filename;
158 
159   /// A bump pointer allocated array of offsets for each source line.
160   ///
161   /// This is lazily computed.  The lines are owned by the SourceManager
162   /// BumpPointerAllocator object.
163   mutable LineOffsetMapping SourceLineCache;
164 
165   /// Indicates whether the buffer itself was provided to override
166   /// the actual file contents.
167   ///
168   /// When true, the original entry may be a virtual file that does not
169   /// exist.
170   LLVM_PREFERRED_TYPE(bool)
171   unsigned BufferOverridden : 1;
172 
173   /// True if this content cache was initially created for a source file
174   /// considered to be volatile (likely to change between stat and open).
175   LLVM_PREFERRED_TYPE(bool)
176   unsigned IsFileVolatile : 1;
177 
178   /// True if this file may be transient, that is, if it might not
179   /// exist at some later point in time when this content entry is used,
180   /// after serialization and deserialization.
181   LLVM_PREFERRED_TYPE(bool)
182   unsigned IsTransient : 1;
183 
184   LLVM_PREFERRED_TYPE(bool)
185   mutable unsigned IsBufferInvalid : 1;
186 
ContentCache()187   ContentCache()
188       : OrigEntry(std::nullopt), ContentsEntry(std::nullopt),
189         BufferOverridden(false), IsFileVolatile(false), IsTransient(false),
190         IsBufferInvalid(false) {}
191 
ContentCache(FileEntryRef Ent)192   ContentCache(FileEntryRef Ent) : ContentCache(Ent, Ent) {}
193 
ContentCache(FileEntryRef Ent,FileEntryRef contentEnt)194   ContentCache(FileEntryRef Ent, FileEntryRef contentEnt)
195       : OrigEntry(Ent), ContentsEntry(contentEnt), BufferOverridden(false),
196         IsFileVolatile(false), IsTransient(false), IsBufferInvalid(false) {}
197 
198   /// The copy ctor does not allow copies where source object has either
199   /// a non-NULL Buffer or SourceLineCache.  Ownership of allocated memory
200   /// is not transferred, so this is a logical error.
ContentCache(const ContentCache & RHS)201   ContentCache(const ContentCache &RHS)
202       : BufferOverridden(false), IsFileVolatile(false), IsTransient(false),
203         IsBufferInvalid(false) {
204     OrigEntry = RHS.OrigEntry;
205     ContentsEntry = RHS.ContentsEntry;
206 
207     assert(!RHS.Buffer && !RHS.SourceLineCache &&
208            "Passed ContentCache object cannot own a buffer.");
209   }
210 
211   ContentCache &operator=(const ContentCache &RHS) = delete;
212 
213   /// Returns the memory buffer for the associated content.
214   ///
215   /// \param Diag Object through which diagnostics will be emitted if the
216   ///   buffer cannot be retrieved.
217   ///
218   /// \param Loc If specified, is the location that invalid file diagnostics
219   ///   will be emitted at.
220   std::optional<llvm::MemoryBufferRef>
221   getBufferOrNone(DiagnosticsEngine &Diag, FileManager &FM,
222                   SourceLocation Loc = SourceLocation()) const;
223 
224   /// Returns the size of the content encapsulated by this
225   /// ContentCache.
226   ///
227   /// This can be the size of the source file or the size of an
228   /// arbitrary scratch buffer.  If the ContentCache encapsulates a source
229   /// file this size is retrieved from the file's FileEntry.
230   unsigned getSize() const;
231 
232   /// Returns the number of bytes actually mapped for this
233   /// ContentCache.
234   ///
235   /// This can be 0 if the MemBuffer was not actually expanded.
236   unsigned getSizeBytesMapped() const;
237 
238   /// Returns the kind of memory used to back the memory buffer for
239   /// this content cache.  This is used for performance analysis.
240   llvm::MemoryBuffer::BufferKind getMemoryBufferKind() const;
241 
242   /// Return the buffer, only if it has been loaded.
getBufferIfLoaded()243   std::optional<llvm::MemoryBufferRef> getBufferIfLoaded() const {
244     if (Buffer)
245       return Buffer->getMemBufferRef();
246     return std::nullopt;
247   }
248 
249   /// Return a StringRef to the source buffer data, only if it has already
250   /// been loaded.
getBufferDataIfLoaded()251   std::optional<StringRef> getBufferDataIfLoaded() const {
252     if (Buffer)
253       return Buffer->getBuffer();
254     return std::nullopt;
255   }
256 
257   /// Set the buffer.
setBuffer(std::unique_ptr<llvm::MemoryBuffer> B)258   void setBuffer(std::unique_ptr<llvm::MemoryBuffer> B) {
259     IsBufferInvalid = false;
260     Buffer = std::move(B);
261   }
262 
263   /// Set the buffer to one that's not owned (or to nullptr).
264   ///
265   /// \pre Buffer cannot already be set.
setUnownedBuffer(std::optional<llvm::MemoryBufferRef> B)266   void setUnownedBuffer(std::optional<llvm::MemoryBufferRef> B) {
267     assert(!Buffer && "Expected to be called right after construction");
268     if (B)
269       setBuffer(llvm::MemoryBuffer::getMemBuffer(*B));
270   }
271 
272   // If BufStr has an invalid BOM, returns the BOM name; otherwise, returns
273   // nullptr
274   static const char *getInvalidBOM(StringRef BufStr);
275 };
276 
277 // Assert that the \c ContentCache objects will always be 8-byte aligned so
278 // that we can pack 3 bits of integer into pointers to such objects.
279 static_assert(alignof(ContentCache) >= 8,
280               "ContentCache must be 8-byte aligned.");
281 
282 /// Information about a FileID, basically just the logical file
283 /// that it represents and include stack information.
284 ///
285 /// Each FileInfo has include stack information, indicating where it came
286 /// from. This information encodes the \#include chain that a token was
287 /// expanded from. The main include file has an invalid IncludeLoc.
288 ///
289 /// FileInfo should not grow larger than ExpansionInfo. Doing so will
290 /// cause memory to bloat in compilations with many unloaded macro
291 /// expansions, since the two data structurs are stored in a union in
292 /// SLocEntry. Extra fields should instead go in "ContentCache *", which
293 /// stores file contents and other bits on the side.
294 ///
295 class FileInfo {
296   friend class clang::SourceManager;
297   friend class clang::ASTWriter;
298   friend class clang::ASTReader;
299 
300   /// The location of the \#include that brought in this file.
301   ///
302   /// This is an invalid SLOC for the main file (top of the \#include chain).
303   SourceLocation IncludeLoc;
304 
305   /// Number of FileIDs (files and macros) that were created during
306   /// preprocessing of this \#include, including this SLocEntry.
307   ///
308   /// Zero means the preprocessor didn't provide such info for this SLocEntry.
309   unsigned NumCreatedFIDs : 31;
310 
311   /// Whether this FileInfo has any \#line directives.
312   LLVM_PREFERRED_TYPE(bool)
313   unsigned HasLineDirectives : 1;
314 
315   /// The content cache and the characteristic of the file.
316   llvm::PointerIntPair<const ContentCache *, 3, CharacteristicKind>
317       ContentAndKind;
318 
319 public:
320   /// Return a FileInfo object.
get(SourceLocation IL,ContentCache & Con,CharacteristicKind FileCharacter,StringRef Filename)321   static FileInfo get(SourceLocation IL, ContentCache &Con,
322                       CharacteristicKind FileCharacter, StringRef Filename) {
323     FileInfo X;
324     X.IncludeLoc = IL;
325     X.NumCreatedFIDs = 0;
326     X.HasLineDirectives = false;
327     X.ContentAndKind.setPointer(&Con);
328     X.ContentAndKind.setInt(FileCharacter);
329     Con.Filename = Filename;
330     return X;
331   }
332 
getIncludeLoc()333   SourceLocation getIncludeLoc() const {
334     return IncludeLoc;
335   }
336 
getContentCache()337   const ContentCache &getContentCache() const {
338     return *ContentAndKind.getPointer();
339   }
340 
341   /// Return whether this is a system header or not.
getFileCharacteristic()342   CharacteristicKind getFileCharacteristic() const {
343     return ContentAndKind.getInt();
344   }
345 
346   /// Return true if this FileID has \#line directives in it.
hasLineDirectives()347   bool hasLineDirectives() const { return HasLineDirectives; }
348 
349   /// Set the flag that indicates that this FileID has
350   /// line table entries associated with it.
setHasLineDirectives()351   void setHasLineDirectives() { HasLineDirectives = true; }
352 
353   /// Returns the name of the file that was used when the file was loaded from
354   /// the underlying file system.
getName()355   StringRef getName() const { return getContentCache().Filename; }
356 };
357 
358 /// Each ExpansionInfo encodes the expansion location - where
359 /// the token was ultimately expanded, and the SpellingLoc - where the actual
360 /// character data for the token came from.
361 class ExpansionInfo {
362   // Really these are all SourceLocations.
363 
364   /// Where the spelling for the token can be found.
365   SourceLocation SpellingLoc;
366 
367   /// In a macro expansion, ExpansionLocStart and ExpansionLocEnd
368   /// indicate the start and end of the expansion. In object-like macros,
369   /// they will be the same. In a function-like macro expansion, the start
370   /// will be the identifier and the end will be the ')'. Finally, in
371   /// macro-argument instantiations, the end will be 'SourceLocation()', an
372   /// invalid location.
373   SourceLocation ExpansionLocStart, ExpansionLocEnd;
374 
375   /// Whether the expansion range is a token range.
376   bool ExpansionIsTokenRange;
377 
378 public:
getSpellingLoc()379   SourceLocation getSpellingLoc() const {
380     return SpellingLoc.isInvalid() ? getExpansionLocStart() : SpellingLoc;
381   }
382 
getExpansionLocStart()383   SourceLocation getExpansionLocStart() const {
384     return ExpansionLocStart;
385   }
386 
getExpansionLocEnd()387   SourceLocation getExpansionLocEnd() const {
388     return ExpansionLocEnd.isInvalid() ? getExpansionLocStart()
389                                        : ExpansionLocEnd;
390   }
391 
isExpansionTokenRange()392   bool isExpansionTokenRange() const { return ExpansionIsTokenRange; }
393 
getExpansionLocRange()394   CharSourceRange getExpansionLocRange() const {
395     return CharSourceRange(
396         SourceRange(getExpansionLocStart(), getExpansionLocEnd()),
397         isExpansionTokenRange());
398   }
399 
isMacroArgExpansion()400   bool isMacroArgExpansion() const {
401     // Note that this needs to return false for default constructed objects.
402     return getExpansionLocStart().isValid() && ExpansionLocEnd.isInvalid();
403   }
404 
isMacroBodyExpansion()405   bool isMacroBodyExpansion() const {
406     return getExpansionLocStart().isValid() && ExpansionLocEnd.isValid();
407   }
408 
isFunctionMacroExpansion()409   bool isFunctionMacroExpansion() const {
410     return getExpansionLocStart().isValid() &&
411            getExpansionLocStart() != getExpansionLocEnd();
412   }
413 
414   /// Return a ExpansionInfo for an expansion.
415   ///
416   /// Start and End specify the expansion range (where the macro is
417   /// expanded), and SpellingLoc specifies the spelling location (where
418   /// the characters from the token come from). All three can refer to
419   /// normal File SLocs or expansion locations.
420   static ExpansionInfo create(SourceLocation SpellingLoc, SourceLocation Start,
421                               SourceLocation End,
422                               bool ExpansionIsTokenRange = true) {
423     ExpansionInfo X;
424     X.SpellingLoc = SpellingLoc;
425     X.ExpansionLocStart = Start;
426     X.ExpansionLocEnd = End;
427     X.ExpansionIsTokenRange = ExpansionIsTokenRange;
428     return X;
429   }
430 
431   /// Return a special ExpansionInfo for the expansion of
432   /// a macro argument into a function-like macro's body.
433   ///
434   /// ExpansionLoc specifies the expansion location (where the macro is
435   /// expanded). This doesn't need to be a range because a macro is always
436   /// expanded at a macro parameter reference, and macro parameters are
437   /// always exactly one token. SpellingLoc specifies the spelling location
438   /// (where the characters from the token come from). ExpansionLoc and
439   /// SpellingLoc can both refer to normal File SLocs or expansion locations.
440   ///
441   /// Given the code:
442   /// \code
443   ///   #define F(x) f(x)
444   ///   F(42);
445   /// \endcode
446   ///
447   /// When expanding '\c F(42)', the '\c x' would call this with an
448   /// SpellingLoc pointing at '\c 42' and an ExpansionLoc pointing at its
449   /// location in the definition of '\c F'.
createForMacroArg(SourceLocation SpellingLoc,SourceLocation ExpansionLoc)450   static ExpansionInfo createForMacroArg(SourceLocation SpellingLoc,
451                                          SourceLocation ExpansionLoc) {
452     // We store an intentionally invalid source location for the end of the
453     // expansion range to mark that this is a macro argument location rather
454     // than a normal one.
455     return create(SpellingLoc, ExpansionLoc, SourceLocation());
456   }
457 
458   /// Return a special ExpansionInfo representing a token that ends
459   /// prematurely. This is used to model a '>>' token that has been split
460   /// into '>' tokens and similar cases. Unlike for the other forms of
461   /// expansion, the expansion range in this case is a character range, not
462   /// a token range.
createForTokenSplit(SourceLocation SpellingLoc,SourceLocation Start,SourceLocation End)463   static ExpansionInfo createForTokenSplit(SourceLocation SpellingLoc,
464                                            SourceLocation Start,
465                                            SourceLocation End) {
466     return create(SpellingLoc, Start, End, false);
467   }
468 };
469 
470 // Assert that the \c FileInfo objects are no bigger than \c ExpansionInfo
471 // objects. This controls the size of \c SLocEntry, of which we have one for
472 // each macro expansion. The number of (unloaded) macro expansions can be
473 // very large. Any other fields needed in FileInfo should go in ContentCache.
474 static_assert(sizeof(FileInfo) <= sizeof(ExpansionInfo),
475               "FileInfo must be no larger than ExpansionInfo.");
476 
477 /// This is a discriminated union of FileInfo and ExpansionInfo.
478 ///
479 /// SourceManager keeps an array of these objects, and they are uniquely
480 /// identified by the FileID datatype.
481 class SLocEntry {
482   static constexpr int OffsetBits = 8 * sizeof(SourceLocation::UIntTy) - 1;
483   SourceLocation::UIntTy Offset : OffsetBits;
484   LLVM_PREFERRED_TYPE(bool)
485   SourceLocation::UIntTy IsExpansion : 1;
486   union {
487     FileInfo File;
488     ExpansionInfo Expansion;
489   };
490 
491 public:
SLocEntry()492   SLocEntry() : Offset(), IsExpansion(), File() {}
493 
getOffset()494   SourceLocation::UIntTy getOffset() const { return Offset; }
495 
isExpansion()496   bool isExpansion() const { return IsExpansion; }
isFile()497   bool isFile() const { return !isExpansion(); }
498 
getFile()499   const FileInfo &getFile() const {
500     return const_cast<SLocEntry *>(this)->getFile();
501   }
502 
getFile()503   FileInfo &getFile() {
504     assert(isFile() && "Not a file SLocEntry!");
505     return File;
506   }
507 
getExpansion()508   const ExpansionInfo &getExpansion() const {
509     assert(isExpansion() && "Not a macro expansion SLocEntry!");
510     return Expansion;
511   }
512 
513   /// Creates an incomplete SLocEntry that is only able to report its offset.
getOffsetOnly(SourceLocation::UIntTy Offset)514   static SLocEntry getOffsetOnly(SourceLocation::UIntTy Offset) {
515     assert(!(Offset & (1ULL << OffsetBits)) && "Offset is too large");
516     SLocEntry E;
517     E.Offset = Offset;
518     return E;
519   }
520 
get(SourceLocation::UIntTy Offset,const FileInfo & FI)521   static SLocEntry get(SourceLocation::UIntTy Offset, const FileInfo &FI) {
522     assert(!(Offset & (1ULL << OffsetBits)) && "Offset is too large");
523     SLocEntry E;
524     E.Offset = Offset;
525     E.IsExpansion = false;
526     E.File = FI;
527     return E;
528   }
529 
get(SourceLocation::UIntTy Offset,const ExpansionInfo & Expansion)530   static SLocEntry get(SourceLocation::UIntTy Offset,
531                        const ExpansionInfo &Expansion) {
532     assert(!(Offset & (1ULL << OffsetBits)) && "Offset is too large");
533     SLocEntry E;
534     E.Offset = Offset;
535     E.IsExpansion = true;
536     new (&E.Expansion) ExpansionInfo(Expansion);
537     return E;
538   }
539 };
540 
541 } // namespace SrcMgr
542 
543 /// External source of source location entries.
544 class ExternalSLocEntrySource {
545 public:
546   virtual ~ExternalSLocEntrySource();
547 
548   /// Read the source location entry with index ID, which will always be
549   /// less than -1.
550   ///
551   /// \returns true if an error occurred that prevented the source-location
552   /// entry from being loaded.
553   virtual bool ReadSLocEntry(int ID) = 0;
554 
555   /// Get the index ID for the loaded SourceLocation offset.
556   ///
557   /// \returns Invalid index ID (0) if an error occurred that prevented the
558   /// SLocEntry  from being loaded.
559   virtual int getSLocEntryID(SourceLocation::UIntTy SLocOffset) = 0;
560 
561   /// Retrieve the module import location and name for the given ID, if
562   /// in fact it was loaded from a module (rather than, say, a precompiled
563   /// header).
564   virtual std::pair<SourceLocation, StringRef> getModuleImportLoc(int ID) = 0;
565 };
566 
567 /// Holds the cache used by isBeforeInTranslationUnit.
568 ///
569 /// The cache structure is complex enough to be worth breaking out of
570 /// SourceManager.
571 class InBeforeInTUCacheEntry {
572   /// The FileID's of the cached query.
573   ///
574   /// If these match up with a subsequent query, the result can be reused.
575   FileID LQueryFID, RQueryFID;
576 
577   /// The relative order of FileIDs that the CommonFID *immediately* includes.
578   ///
579   /// This is used to compare macro expansion locations.
580   bool LChildBeforeRChild;
581 
582   /// The file found in common between the two \#include traces, i.e.,
583   /// the nearest common ancestor of the \#include tree.
584   FileID CommonFID;
585 
586   /// The offset of the previous query in CommonFID.
587   ///
588   /// Usually, this represents the location of the \#include for QueryFID, but
589   /// if LQueryFID is a parent of RQueryFID (or vice versa) then these can be a
590   /// random token in the parent.
591   unsigned LCommonOffset, RCommonOffset;
592 
593 public:
594   InBeforeInTUCacheEntry() = default;
InBeforeInTUCacheEntry(FileID L,FileID R)595   InBeforeInTUCacheEntry(FileID L, FileID R) : LQueryFID(L), RQueryFID(R) {
596     assert(L != R);
597   }
598 
599   /// Return true if the currently cached values match up with
600   /// the specified LHS/RHS query.
601   ///
602   /// If not, we can't use the cache.
isCacheValid()603   bool isCacheValid() const {
604     return CommonFID.isValid();
605   }
606 
607   /// If the cache is valid, compute the result given the
608   /// specified offsets in the LHS/RHS FileID's.
getCachedResult(unsigned LOffset,unsigned ROffset)609   bool getCachedResult(unsigned LOffset, unsigned ROffset) const {
610     // If one of the query files is the common file, use the offset.  Otherwise,
611     // use the #include loc in the common file.
612     if (LQueryFID != CommonFID) LOffset = LCommonOffset;
613     if (RQueryFID != CommonFID) ROffset = RCommonOffset;
614 
615     // It is common for multiple macro expansions to be "included" from the same
616     // location (expansion location), in which case use the order of the FileIDs
617     // to determine which came first. This will also take care the case where
618     // one of the locations points at the inclusion/expansion point of the other
619     // in which case its FileID will come before the other.
620     if (LOffset == ROffset)
621       return LChildBeforeRChild;
622 
623     return LOffset < ROffset;
624   }
625 
626   /// Set up a new query.
627   /// If it matches the old query, we can keep the cached answer.
setQueryFIDs(FileID LHS,FileID RHS)628   void setQueryFIDs(FileID LHS, FileID RHS) {
629     assert(LHS != RHS);
630     if (LQueryFID != LHS || RQueryFID != RHS) {
631       LQueryFID = LHS;
632       RQueryFID = RHS;
633       CommonFID = FileID();
634     }
635   }
636 
setCommonLoc(FileID commonFID,unsigned lCommonOffset,unsigned rCommonOffset,bool LParentBeforeRParent)637   void setCommonLoc(FileID commonFID, unsigned lCommonOffset,
638                     unsigned rCommonOffset, bool LParentBeforeRParent) {
639     CommonFID = commonFID;
640     LCommonOffset = lCommonOffset;
641     RCommonOffset = rCommonOffset;
642     LChildBeforeRChild = LParentBeforeRParent;
643   }
644 };
645 
646 /// The stack used when building modules on demand, which is used
647 /// to provide a link between the source managers of the different compiler
648 /// instances.
649 using ModuleBuildStack = ArrayRef<std::pair<std::string, FullSourceLoc>>;
650 
651 /// This class handles loading and caching of source files into memory.
652 ///
653 /// This object owns the MemoryBuffer objects for all of the loaded
654 /// files and assigns unique FileID's for each unique \#include chain.
655 ///
656 /// The SourceManager can be queried for information about SourceLocation
657 /// objects, turning them into either spelling or expansion locations. Spelling
658 /// locations represent where the bytes corresponding to a token came from and
659 /// expansion locations represent where the location is in the user's view. In
660 /// the case of a macro expansion, for example, the spelling location indicates
661 /// where the expanded token came from and the expansion location specifies
662 /// where it was expanded.
663 class SourceManager : public RefCountedBase<SourceManager> {
664   /// DiagnosticsEngine object.
665   DiagnosticsEngine &Diag;
666 
667   FileManager &FileMgr;
668 
669   mutable llvm::BumpPtrAllocator ContentCacheAlloc;
670 
671   /// Memoized information about all of the files tracked by this
672   /// SourceManager.
673   ///
674   /// This map allows us to merge ContentCache entries based
675   /// on their FileEntry*.  All ContentCache objects will thus have unique,
676   /// non-null, FileEntry pointers.
677   llvm::DenseMap<FileEntryRef, SrcMgr::ContentCache*> FileInfos;
678 
679   /// True if the ContentCache for files that are overridden by other
680   /// files, should report the original file name. Defaults to true.
681   bool OverridenFilesKeepOriginalName = true;
682 
683   /// True if non-system source files should be treated as volatile
684   /// (likely to change while trying to use them). Defaults to false.
685   bool UserFilesAreVolatile;
686 
687   /// True if all files read during this compilation should be treated
688   /// as transient (may not be present in later compilations using a module
689   /// file created from this compilation). Defaults to false.
690   bool FilesAreTransient = false;
691 
692   struct OverriddenFilesInfoTy {
693     /// Files that have been overridden with the contents from another
694     /// file.
695     llvm::DenseMap<const FileEntry *, FileEntryRef> OverriddenFiles;
696 
697     /// Files that were overridden with a memory buffer.
698     llvm::DenseSet<const FileEntry *> OverriddenFilesWithBuffer;
699   };
700 
701   /// Lazily create the object keeping overridden files info, since
702   /// it is uncommonly used.
703   std::unique_ptr<OverriddenFilesInfoTy> OverriddenFilesInfo;
704 
getOverriddenFilesInfo()705   OverriddenFilesInfoTy &getOverriddenFilesInfo() {
706     if (!OverriddenFilesInfo)
707       OverriddenFilesInfo.reset(new OverriddenFilesInfoTy);
708     return *OverriddenFilesInfo;
709   }
710 
711   /// Information about various memory buffers that we have read in.
712   ///
713   /// All FileEntry* within the stored ContentCache objects are NULL,
714   /// as they do not refer to a file.
715   std::vector<SrcMgr::ContentCache*> MemBufferInfos;
716 
717   /// The table of SLocEntries that are local to this module.
718   ///
719   /// Positive FileIDs are indexes into this table. Entry 0 indicates an invalid
720   /// expansion.
721   SmallVector<SrcMgr::SLocEntry, 0> LocalSLocEntryTable;
722   /// An in-parallel offset table, merely used for speeding up FileID lookup.
723   SmallVector<SourceLocation::UIntTy> LocalLocOffsetTable;
724 
725   /// The table of SLocEntries that are loaded from other modules.
726   ///
727   /// Negative FileIDs are indexes into this table. To get from ID to an index,
728   /// use (-ID - 2).
729   llvm::PagedVector<SrcMgr::SLocEntry, 32> LoadedSLocEntryTable;
730 
731   /// For each allocation in LoadedSLocEntryTable, we keep the first FileID.
732   /// We assume exactly one allocation per AST file, and use that to determine
733   /// whether two FileIDs come from the same AST file.
734   SmallVector<FileID, 0> LoadedSLocEntryAllocBegin;
735 
736   /// The starting offset of the next local SLocEntry.
737   ///
738   /// This is LocalSLocEntryTable.back().Offset + the size of that entry.
739   SourceLocation::UIntTy NextLocalOffset;
740 
741   /// The starting offset of the latest batch of loaded SLocEntries.
742   ///
743   /// This is LoadedSLocEntryTable.back().Offset, except that that entry might
744   /// not have been loaded, so that value would be unknown.
745   SourceLocation::UIntTy CurrentLoadedOffset;
746 
747   /// The highest possible offset is 2^31-1 (2^63-1 for 64-bit source
748   /// locations), so CurrentLoadedOffset starts at 2^31 (2^63 resp.).
749   static const SourceLocation::UIntTy MaxLoadedOffset =
750       1ULL << (8 * sizeof(SourceLocation::UIntTy) - 1);
751 
752   /// A bitmap that indicates whether the entries of LoadedSLocEntryTable
753   /// have already been loaded from the external source.
754   ///
755   /// Same indexing as LoadedSLocEntryTable.
756   llvm::BitVector SLocEntryLoaded;
757 
758   /// A bitmap that indicates whether the entries of LoadedSLocEntryTable
759   /// have already had their offset loaded from the external source.
760   ///
761   /// Superset of SLocEntryLoaded. Same indexing as SLocEntryLoaded.
762   llvm::BitVector SLocEntryOffsetLoaded;
763 
764   /// An external source for source location entries.
765   ExternalSLocEntrySource *ExternalSLocEntries = nullptr;
766 
767   /// A one-entry cache to speed up getFileID.
768   ///
769   /// LastFileIDLookup records the last FileID looked up or created, because it
770   /// is very common to look up many tokens from the same file.
771   mutable FileID LastFileIDLookup;
772   mutable SourceLocation::UIntTy LastLookupStartOffset;
773   mutable SourceLocation::UIntTy LastLookupEndOffset; // exclude
774 
775   /// Holds information for \#line directives.
776   ///
777   /// This is referenced by indices from SLocEntryTable.
778   std::unique_ptr<LineTableInfo> LineTable;
779 
780   /// These ivars serve as a cache used in the getLineNumber
781   /// method which is used to speedup getLineNumber calls to nearby locations.
782   mutable FileID LastLineNoFileIDQuery;
783   mutable const SrcMgr::ContentCache *LastLineNoContentCache;
784   mutable unsigned LastLineNoFilePos;
785   mutable unsigned LastLineNoResult;
786 
787   /// The file ID for the main source file of the translation unit.
788   FileID MainFileID;
789 
790   /// The file ID for the precompiled preamble there is one.
791   FileID PreambleFileID;
792 
793   // Statistics for -print-stats.
794   mutable unsigned NumLinearScans = 0;
795   mutable unsigned NumBinaryProbes = 0;
796 
797   /// Associates a FileID with its "included/expanded in" decomposed
798   /// location.
799   ///
800   /// Used to cache results from and speed-up \c getDecomposedIncludedLoc
801   /// function.
802   mutable llvm::DenseMap<FileID, FileIDAndOffset> IncludedLocMap;
803 
804   /// The key value into the IsBeforeInTUCache table.
805   using IsBeforeInTUCacheKey = std::pair<FileID, FileID>;
806 
807   /// The IsBeforeInTranslationUnitCache is a mapping from FileID pairs
808   /// to cache results.
809   using InBeforeInTUCache =
810       llvm::DenseMap<IsBeforeInTUCacheKey, InBeforeInTUCacheEntry>;
811 
812   /// Cache results for the isBeforeInTranslationUnit method.
813   mutable InBeforeInTUCache IBTUCache;
814   mutable InBeforeInTUCacheEntry IBTUCacheOverflow;
815 
816   /// Return the cache entry for comparing the given file IDs
817   /// for isBeforeInTranslationUnit.
818   InBeforeInTUCacheEntry &getInBeforeInTUCache(FileID LFID, FileID RFID) const;
819 
820   // Cache for the "fake" buffer used for error-recovery purposes.
821   mutable std::unique_ptr<llvm::MemoryBuffer> FakeBufferForRecovery;
822 
823   mutable std::unique_ptr<SrcMgr::ContentCache> FakeContentCacheForRecovery;
824 
825   mutable std::unique_ptr<SrcMgr::SLocEntry> FakeSLocEntryForRecovery;
826 
827   /// Lazily computed map of macro argument chunks to their expanded
828   /// source location.
829   using MacroArgsMap = std::map<unsigned, SourceLocation>;
830 
831   mutable llvm::DenseMap<FileID, std::unique_ptr<MacroArgsMap>>
832       MacroArgsCacheMap;
833 
834   /// The stack of modules being built, which is used to detect
835   /// cycles in the module dependency graph as modules are being built, as
836   /// well as to describe why we're rebuilding a particular module.
837   ///
838   /// There is no way to set this value from the command line. If we ever need
839   /// to do so (e.g., if on-demand module construction moves out-of-process),
840   /// we can add a cc1-level option to do so.
841   SmallVector<std::pair<std::string, FullSourceLoc>, 2> StoredModuleBuildStack;
842 
843 public:
844   SourceManager(DiagnosticsEngine &Diag, FileManager &FileMgr,
845                 bool UserFilesAreVolatile = false);
846   explicit SourceManager(const SourceManager &) = delete;
847   SourceManager &operator=(const SourceManager &) = delete;
848   ~SourceManager();
849 
850   void clearIDTables();
851 
852   /// Initialize this source manager suitably to replay the compilation
853   /// described by \p Old. Requires that \p Old outlive \p *this.
854   void initializeForReplay(const SourceManager &Old);
855 
getDiagnostics()856   DiagnosticsEngine &getDiagnostics() const { return Diag; }
857 
getFileManager()858   FileManager &getFileManager() const { return FileMgr; }
859 
860   /// Set true if the SourceManager should report the original file name
861   /// for contents of files that were overridden by other files. Defaults to
862   /// true.
setOverridenFilesKeepOriginalName(bool value)863   void setOverridenFilesKeepOriginalName(bool value) {
864     OverridenFilesKeepOriginalName = value;
865   }
866 
867   /// True if non-system source files should be treated as volatile
868   /// (likely to change while trying to use them).
userFilesAreVolatile()869   bool userFilesAreVolatile() const { return UserFilesAreVolatile; }
870 
871   /// Retrieve the module build stack.
getModuleBuildStack()872   ModuleBuildStack getModuleBuildStack() const {
873     return StoredModuleBuildStack;
874   }
875 
876   /// Set the module build stack.
setModuleBuildStack(ModuleBuildStack stack)877   void setModuleBuildStack(ModuleBuildStack stack) {
878     StoredModuleBuildStack.clear();
879     StoredModuleBuildStack.append(stack.begin(), stack.end());
880   }
881 
882   /// Push an entry to the module build stack.
pushModuleBuildStack(StringRef moduleName,FullSourceLoc importLoc)883   void pushModuleBuildStack(StringRef moduleName, FullSourceLoc importLoc) {
884     StoredModuleBuildStack.push_back(std::make_pair(moduleName.str(),importLoc));
885   }
886 
887   //===--------------------------------------------------------------------===//
888   // MainFileID creation and querying methods.
889   //===--------------------------------------------------------------------===//
890 
891   /// Returns the FileID of the main source file.
getMainFileID()892   FileID getMainFileID() const { return MainFileID; }
893 
894   /// Set the file ID for the main source file.
setMainFileID(FileID FID)895   void setMainFileID(FileID FID) {
896     MainFileID = FID;
897   }
898 
899   /// Returns true when the given FileEntry corresponds to the main file.
900   ///
901   /// The main file should be set prior to calling this function.
902   bool isMainFile(const FileEntry &SourceFile);
903 
904   /// Set the file ID for the precompiled preamble.
setPreambleFileID(FileID Preamble)905   void setPreambleFileID(FileID Preamble) {
906     assert(PreambleFileID.isInvalid() && "PreambleFileID already set!");
907     PreambleFileID = Preamble;
908   }
909 
910   /// Get the file ID for the precompiled preamble if there is one.
getPreambleFileID()911   FileID getPreambleFileID() const { return PreambleFileID; }
912 
913   //===--------------------------------------------------------------------===//
914   // Methods to create new FileID's and macro expansions.
915   //===--------------------------------------------------------------------===//
916 
917   /// Create a new FileID that represents the specified file
918   /// being \#included from the specified IncludePosition.
919   FileID createFileID(FileEntryRef SourceFile, SourceLocation IncludePos,
920                       SrcMgr::CharacteristicKind FileCharacter,
921                       int LoadedID = 0,
922                       SourceLocation::UIntTy LoadedOffset = 0);
923 
924   /// Create a new FileID that represents the specified memory buffer.
925   ///
926   /// This does no caching of the buffer and takes ownership of the
927   /// MemoryBuffer, so only pass a MemoryBuffer to this once.
928   FileID createFileID(std::unique_ptr<llvm::MemoryBuffer> Buffer,
929                       SrcMgr::CharacteristicKind FileCharacter = SrcMgr::C_User,
930                       int LoadedID = 0, SourceLocation::UIntTy LoadedOffset = 0,
931                       SourceLocation IncludeLoc = SourceLocation());
932 
933   /// Create a new FileID that represents the specified memory buffer.
934   ///
935   /// This does not take ownership of the MemoryBuffer. The memory buffer must
936   /// outlive the SourceManager.
937   FileID createFileID(const llvm::MemoryBufferRef &Buffer,
938                       SrcMgr::CharacteristicKind FileCharacter = SrcMgr::C_User,
939                       int LoadedID = 0, SourceLocation::UIntTy LoadedOffset = 0,
940                       SourceLocation IncludeLoc = SourceLocation());
941 
942   /// Get the FileID for \p SourceFile if it exists. Otherwise, create a
943   /// new FileID for the \p SourceFile.
944   FileID getOrCreateFileID(FileEntryRef SourceFile,
945                            SrcMgr::CharacteristicKind FileCharacter);
946 
947   /// Creates an expansion SLocEntry for the substitution of an argument into a
948   /// function-like macro's body. Returns the start of the expansion.
949   ///
950   /// The macro argument was written at \p SpellingLoc with length \p Length.
951   /// \p ExpansionLoc is the parameter name in the (expanded) macro body.
952   SourceLocation createMacroArgExpansionLoc(SourceLocation SpellingLoc,
953                                             SourceLocation ExpansionLoc,
954                                             unsigned Length);
955 
956   /// Creates an expansion SLocEntry for a macro use. Returns its start.
957   ///
958   /// The macro body begins at \p SpellingLoc with length \p Length.
959   /// The macro use spans [ExpansionLocStart, ExpansionLocEnd].
960   SourceLocation createExpansionLoc(SourceLocation SpellingLoc,
961                                     SourceLocation ExpansionLocStart,
962                                     SourceLocation ExpansionLocEnd,
963                                     unsigned Length,
964                                     bool ExpansionIsTokenRange = true,
965                                     int LoadedID = 0,
966                                     SourceLocation::UIntTy LoadedOffset = 0);
967 
968   /// Return a new SourceLocation that encodes that the token starting
969   /// at \p TokenStart ends prematurely at \p TokenEnd.
970   SourceLocation createTokenSplitLoc(SourceLocation SpellingLoc,
971                                      SourceLocation TokenStart,
972                                      SourceLocation TokenEnd);
973 
974   /// Retrieve the memory buffer associated with the given file.
975   ///
976   /// Returns std::nullopt if the buffer is not valid.
977   std::optional<llvm::MemoryBufferRef>
978   getMemoryBufferForFileOrNone(FileEntryRef File);
979 
980   /// Retrieve the memory buffer associated with the given file.
981   ///
982   /// Returns a fake buffer if there isn't a real one.
getMemoryBufferForFileOrFake(FileEntryRef File)983   llvm::MemoryBufferRef getMemoryBufferForFileOrFake(FileEntryRef File) {
984     if (auto B = getMemoryBufferForFileOrNone(File))
985       return *B;
986     return getFakeBufferForRecovery();
987   }
988 
989   /// Override the contents of the given source file by providing an
990   /// already-allocated buffer.
991   ///
992   /// \param SourceFile the source file whose contents will be overridden.
993   ///
994   /// \param Buffer the memory buffer whose contents will be used as the
995   /// data in the given source file.
overrideFileContents(FileEntryRef SourceFile,const llvm::MemoryBufferRef & Buffer)996   void overrideFileContents(FileEntryRef SourceFile,
997                             const llvm::MemoryBufferRef &Buffer) {
998     overrideFileContents(SourceFile, llvm::MemoryBuffer::getMemBuffer(Buffer));
999   }
1000 
1001   /// Override the contents of the given source file by providing an
1002   /// already-allocated buffer.
1003   ///
1004   /// \param SourceFile the source file whose contents will be overridden.
1005   ///
1006   /// \param Buffer the memory buffer whose contents will be used as the
1007   /// data in the given source file.
1008   void overrideFileContents(FileEntryRef SourceFile,
1009                             std::unique_ptr<llvm::MemoryBuffer> Buffer);
1010 
1011   /// Override the given source file with another one.
1012   ///
1013   /// \param SourceFile the source file which will be overridden.
1014   ///
1015   /// \param NewFile the file whose contents will be used as the
1016   /// data instead of the contents of the given source file.
1017   void overrideFileContents(const FileEntry *SourceFile, FileEntryRef NewFile);
1018 
1019   /// Returns true if the file contents have been overridden.
isFileOverridden(const FileEntry * File)1020   bool isFileOverridden(const FileEntry *File) const {
1021     if (OverriddenFilesInfo) {
1022       if (OverriddenFilesInfo->OverriddenFilesWithBuffer.count(File))
1023         return true;
1024       if (OverriddenFilesInfo->OverriddenFiles.contains(File))
1025         return true;
1026     }
1027     return false;
1028   }
1029 
1030   /// Bypass the overridden contents of a file.  This creates a new FileEntry
1031   /// and initializes the content cache for it.  Returns std::nullopt if there
1032   /// is no such file in the filesystem.
1033   ///
1034   /// This should be called before parsing has begun.
1035   OptionalFileEntryRef bypassFileContentsOverride(FileEntryRef File);
1036 
1037   /// Specify that a file is transient.
1038   void setFileIsTransient(FileEntryRef SourceFile);
1039 
1040   /// Specify that all files that are read during this compilation are
1041   /// transient.
setAllFilesAreTransient(bool Transient)1042   void setAllFilesAreTransient(bool Transient) {
1043     FilesAreTransient = Transient;
1044   }
1045 
1046   //===--------------------------------------------------------------------===//
1047   // FileID manipulation methods.
1048   //===--------------------------------------------------------------------===//
1049 
1050   /// Return the buffer for the specified FileID.
1051   ///
1052   /// If there is an error opening this buffer the first time, return
1053   /// std::nullopt.
1054   std::optional<llvm::MemoryBufferRef>
1055   getBufferOrNone(FileID FID, SourceLocation Loc = SourceLocation()) const {
1056     if (auto *Entry = getSLocEntryForFile(FID))
1057       return Entry->getFile().getContentCache().getBufferOrNone(
1058           Diag, getFileManager(), Loc);
1059     return std::nullopt;
1060   }
1061 
1062   /// Return the buffer for the specified FileID.
1063   ///
1064   /// If there is an error opening this buffer the first time, this
1065   /// manufactures a temporary buffer and returns it.
1066   llvm::MemoryBufferRef
1067   getBufferOrFake(FileID FID, SourceLocation Loc = SourceLocation()) const {
1068     if (auto B = getBufferOrNone(FID, Loc))
1069       return *B;
1070     return getFakeBufferForRecovery();
1071   }
1072 
1073   /// Returns the FileEntry record for the provided FileID.
getFileEntryForID(FileID FID)1074   const FileEntry *getFileEntryForID(FileID FID) const {
1075     if (auto FE = getFileEntryRefForID(FID))
1076       return *FE;
1077     return nullptr;
1078   }
1079 
1080   /// Returns the FileEntryRef for the provided FileID.
getFileEntryRefForID(FileID FID)1081   OptionalFileEntryRef getFileEntryRefForID(FileID FID) const {
1082     if (auto *Entry = getSLocEntryForFile(FID))
1083       return Entry->getFile().getContentCache().OrigEntry;
1084     return std::nullopt;
1085   }
1086 
1087   /// Returns the filename for the provided FileID, unless it's a built-in
1088   /// buffer that's not represented by a filename.
1089   ///
1090   /// Returns std::nullopt for non-files and built-in files.
1091   std::optional<StringRef> getNonBuiltinFilenameForID(FileID FID) const;
1092 
1093   /// Returns the FileEntry record for the provided SLocEntry.
1094   const FileEntry *
getFileEntryForSLocEntry(const SrcMgr::SLocEntry & SLocEntry)1095   getFileEntryForSLocEntry(const SrcMgr::SLocEntry &SLocEntry) const {
1096     if (auto FE = SLocEntry.getFile().getContentCache().OrigEntry)
1097       return *FE;
1098     return nullptr;
1099   }
1100 
1101   /// Return a StringRef to the source buffer data for the
1102   /// specified FileID.
1103   ///
1104   /// \param FID The file ID whose contents will be returned.
1105   /// \param Invalid If non-NULL, will be set true if an error occurred.
1106   StringRef getBufferData(FileID FID, bool *Invalid = nullptr) const;
1107 
1108   /// Return a StringRef to the source buffer data for the
1109   /// specified FileID, returning std::nullopt if invalid.
1110   ///
1111   /// \param FID The file ID whose contents will be returned.
1112   std::optional<StringRef> getBufferDataOrNone(FileID FID) const;
1113 
1114   /// Return a StringRef to the source buffer data for the
1115   /// specified FileID, returning std::nullopt if it's not yet loaded.
1116   ///
1117   /// \param FID The file ID whose contents will be returned.
1118   std::optional<StringRef> getBufferDataIfLoaded(FileID FID) const;
1119 
1120   /// Get the number of FileIDs (files and macros) that were created
1121   /// during preprocessing of \p FID, including it.
getNumCreatedFIDsForFileID(FileID FID)1122   unsigned getNumCreatedFIDsForFileID(FileID FID) const {
1123     if (auto *Entry = getSLocEntryForFile(FID))
1124       return Entry->getFile().NumCreatedFIDs;
1125     return 0;
1126   }
1127 
1128   /// Set the number of FileIDs (files and macros) that were created
1129   /// during preprocessing of \p FID, including it.
1130   void setNumCreatedFIDsForFileID(FileID FID, unsigned NumFIDs,
1131                                   bool Force = false) {
1132     auto *Entry = getSLocEntryForFile(FID);
1133     if (!Entry)
1134       return;
1135     assert((Force || Entry->getFile().NumCreatedFIDs == 0) && "Already set!");
1136     Entry->getFile().NumCreatedFIDs = NumFIDs;
1137   }
1138 
1139   //===--------------------------------------------------------------------===//
1140   // SourceLocation manipulation methods.
1141   //===--------------------------------------------------------------------===//
1142 
1143   /// Return the FileID for a SourceLocation.
1144   ///
1145   /// This is a very hot method that is used for all SourceManager queries
1146   /// that start with a SourceLocation object.  It is responsible for finding
1147   /// the entry in SLocEntryTable which contains the specified location.
1148   ///
getFileID(SourceLocation SpellingLoc)1149   FileID getFileID(SourceLocation SpellingLoc) const {
1150     return getFileID(SpellingLoc.getOffset());
1151   }
1152 
1153   /// Return the filename of the file containing a SourceLocation.
1154   StringRef getFilename(SourceLocation SpellingLoc) const;
1155 
1156   /// Return the source location corresponding to the first byte of
1157   /// the specified file.
getLocForStartOfFile(FileID FID)1158   SourceLocation getLocForStartOfFile(FileID FID) const {
1159     if (auto *Entry = getSLocEntryForFile(FID))
1160       return SourceLocation::getFileLoc(Entry->getOffset());
1161     return SourceLocation();
1162   }
1163 
1164   /// Return the source location corresponding to the last byte of the
1165   /// specified file.
getLocForEndOfFile(FileID FID)1166   SourceLocation getLocForEndOfFile(FileID FID) const {
1167     if (auto *Entry = getSLocEntryForFile(FID))
1168       return SourceLocation::getFileLoc(Entry->getOffset() +
1169                                         getFileIDSize(FID));
1170     return SourceLocation();
1171   }
1172 
1173   /// Returns the include location if \p FID is a \#include'd file
1174   /// otherwise it returns an invalid location.
getIncludeLoc(FileID FID)1175   SourceLocation getIncludeLoc(FileID FID) const {
1176     if (auto *Entry = getSLocEntryForFile(FID))
1177       return Entry->getFile().getIncludeLoc();
1178     return SourceLocation();
1179   }
1180 
1181   // Returns the import location if the given source location is
1182   // located within a module, or an invalid location if the source location
1183   // is within the current translation unit.
1184   std::pair<SourceLocation, StringRef>
getModuleImportLoc(SourceLocation Loc)1185   getModuleImportLoc(SourceLocation Loc) const {
1186     FileID FID = getFileID(Loc);
1187 
1188     // Positive file IDs are in the current translation unit, and -1 is a
1189     // placeholder.
1190     if (FID.ID >= -1)
1191       return std::make_pair(SourceLocation(), "");
1192 
1193     return ExternalSLocEntries->getModuleImportLoc(FID.ID);
1194   }
1195 
1196   /// Given a SourceLocation object \p Loc, return the expansion
1197   /// location referenced by the ID.
getExpansionLoc(SourceLocation Loc)1198   SourceLocation getExpansionLoc(SourceLocation Loc) const {
1199     // Handle the non-mapped case inline, defer to out of line code to handle
1200     // expansions.
1201     if (Loc.isFileID()) return Loc;
1202     return getExpansionLocSlowCase(Loc);
1203   }
1204 
1205   /// Given \p Loc, if it is a macro location return the expansion
1206   /// location or the spelling location, depending on if it comes from a
1207   /// macro argument or not.
getFileLoc(SourceLocation Loc)1208   SourceLocation getFileLoc(SourceLocation Loc) const {
1209     if (Loc.isFileID()) return Loc;
1210     return getFileLocSlowCase(Loc);
1211   }
1212 
1213   /// Return the start/end of the expansion information for an
1214   /// expansion location.
1215   ///
1216   /// \pre \p Loc is required to be an expansion location.
1217   CharSourceRange getImmediateExpansionRange(SourceLocation Loc) const;
1218 
1219   /// Given a SourceLocation object, return the range of
1220   /// tokens covered by the expansion in the ultimate file.
1221   CharSourceRange getExpansionRange(SourceLocation Loc) const;
1222 
1223   /// Given a SourceRange object, return the range of
1224   /// tokens or characters covered by the expansion in the ultimate file.
getExpansionRange(SourceRange Range)1225   CharSourceRange getExpansionRange(SourceRange Range) const {
1226     SourceLocation Begin = getExpansionRange(Range.getBegin()).getBegin();
1227     CharSourceRange End = getExpansionRange(Range.getEnd());
1228     return CharSourceRange(SourceRange(Begin, End.getEnd()),
1229                            End.isTokenRange());
1230   }
1231 
1232   /// Given a CharSourceRange object, return the range of
1233   /// tokens or characters covered by the expansion in the ultimate file.
getExpansionRange(CharSourceRange Range)1234   CharSourceRange getExpansionRange(CharSourceRange Range) const {
1235     CharSourceRange Expansion = getExpansionRange(Range.getAsRange());
1236     if (Expansion.getEnd() == Range.getEnd())
1237       Expansion.setTokenRange(Range.isTokenRange());
1238     return Expansion;
1239   }
1240 
1241   /// Given a SourceLocation object, return the spelling
1242   /// location referenced by the ID.
1243   ///
1244   /// This is the place where the characters that make up the lexed token
1245   /// can be found.
getSpellingLoc(SourceLocation Loc)1246   SourceLocation getSpellingLoc(SourceLocation Loc) const {
1247     // Handle the non-mapped case inline, defer to out of line code to handle
1248     // expansions.
1249     if (Loc.isFileID()) return Loc;
1250     return getSpellingLocSlowCase(Loc);
1251   }
1252 
1253   /// Given a SourceLocation object, return the spelling location
1254   /// referenced by the ID.
1255   ///
1256   /// This is the first level down towards the place where the characters
1257   /// that make up the lexed token can be found.  This should not generally
1258   /// be used by clients.
1259   SourceLocation getImmediateSpellingLoc(SourceLocation Loc) const;
1260 
1261   /// Form a SourceLocation from a FileID and Offset pair.
getComposedLoc(FileID FID,unsigned Offset)1262   SourceLocation getComposedLoc(FileID FID, unsigned Offset) const {
1263     auto *Entry = getSLocEntryOrNull(FID);
1264     if (!Entry)
1265       return SourceLocation();
1266 
1267     SourceLocation::UIntTy GlobalOffset = Entry->getOffset() + Offset;
1268     return Entry->isFile() ? SourceLocation::getFileLoc(GlobalOffset)
1269                            : SourceLocation::getMacroLoc(GlobalOffset);
1270   }
1271 
1272   /// Decompose the specified location into a raw FileID + Offset pair.
1273   ///
1274   /// The first element is the FileID, the second is the offset from the
1275   /// start of the buffer of the location.
getDecomposedLoc(SourceLocation Loc)1276   FileIDAndOffset getDecomposedLoc(SourceLocation Loc) const {
1277     FileID FID = getFileID(Loc);
1278     auto *Entry = getSLocEntryOrNull(FID);
1279     if (!Entry)
1280       return std::make_pair(FileID(), 0);
1281     return std::make_pair(FID, Loc.getOffset() - Entry->getOffset());
1282   }
1283 
1284   /// Decompose the specified location into a raw FileID + Offset pair.
1285   ///
1286   /// If the location is an expansion record, walk through it until we find
1287   /// the final location expanded.
getDecomposedExpansionLoc(SourceLocation Loc)1288   FileIDAndOffset getDecomposedExpansionLoc(SourceLocation Loc) const {
1289     FileID FID = getFileID(Loc);
1290     auto *E = getSLocEntryOrNull(FID);
1291     if (!E)
1292       return std::make_pair(FileID(), 0);
1293 
1294     unsigned Offset = Loc.getOffset()-E->getOffset();
1295     if (Loc.isFileID())
1296       return std::make_pair(FID, Offset);
1297 
1298     return getDecomposedExpansionLocSlowCase(E);
1299   }
1300 
1301   /// Decompose the specified location into a raw FileID + Offset pair.
1302   ///
1303   /// If the location is an expansion record, walk through it until we find
1304   /// its spelling record.
getDecomposedSpellingLoc(SourceLocation Loc)1305   FileIDAndOffset getDecomposedSpellingLoc(SourceLocation Loc) const {
1306     FileID FID = getFileID(Loc);
1307     auto *E = getSLocEntryOrNull(FID);
1308     if (!E)
1309       return std::make_pair(FileID(), 0);
1310 
1311     unsigned Offset = Loc.getOffset()-E->getOffset();
1312     if (Loc.isFileID())
1313       return std::make_pair(FID, Offset);
1314     return getDecomposedSpellingLocSlowCase(E, Offset);
1315   }
1316 
1317   /// Returns the "included/expanded in" decomposed location of the given
1318   /// FileID.
1319   FileIDAndOffset getDecomposedIncludedLoc(FileID FID) const;
1320 
1321   /// Returns the offset from the start of the file that the
1322   /// specified SourceLocation represents.
1323   ///
1324   /// This is not very meaningful for a macro ID.
getFileOffset(SourceLocation SpellingLoc)1325   unsigned getFileOffset(SourceLocation SpellingLoc) const {
1326     return getDecomposedLoc(SpellingLoc).second;
1327   }
1328 
1329   /// Tests whether the given source location represents a macro
1330   /// argument's expansion into the function-like macro definition.
1331   ///
1332   /// \param StartLoc If non-null and function returns true, it is set to the
1333   /// start location of the macro argument expansion.
1334   ///
1335   /// Such source locations only appear inside of the expansion
1336   /// locations representing where a particular function-like macro was
1337   /// expanded.
1338   bool isMacroArgExpansion(SourceLocation Loc,
1339                            SourceLocation *StartLoc = nullptr) const;
1340 
1341   /// Tests whether the given source location represents the expansion of
1342   /// a macro body.
1343   ///
1344   /// This is equivalent to testing whether the location is part of a macro
1345   /// expansion but not the expansion of an argument to a function-like macro.
1346   bool isMacroBodyExpansion(SourceLocation Loc) const;
1347 
1348   /// Returns true if the given MacroID location points at the beginning
1349   /// of the immediate macro expansion.
1350   ///
1351   /// \param MacroBegin If non-null and function returns true, it is set to the
1352   /// begin location of the immediate macro expansion.
1353   bool isAtStartOfImmediateMacroExpansion(SourceLocation Loc,
1354                                     SourceLocation *MacroBegin = nullptr) const;
1355 
1356   /// Returns true if the given MacroID location points at the character
1357   /// end of the immediate macro expansion.
1358   ///
1359   /// \param MacroEnd If non-null and function returns true, it is set to the
1360   /// character end location of the immediate macro expansion.
1361   bool
1362   isAtEndOfImmediateMacroExpansion(SourceLocation Loc,
1363                                    SourceLocation *MacroEnd = nullptr) const;
1364 
1365   /// Returns true if \p Loc is inside the [\p Start, +\p Length)
1366   /// chunk of the source location address space.
1367   ///
1368   /// If it's true and \p RelativeOffset is non-null, it will be set to the
1369   /// relative offset of \p Loc inside the chunk.
1370   bool
1371   isInSLocAddrSpace(SourceLocation Loc, SourceLocation Start, unsigned Length,
1372                     SourceLocation::UIntTy *RelativeOffset = nullptr) const {
1373     assert(((Start.getOffset() < NextLocalOffset &&
1374                Start.getOffset()+Length <= NextLocalOffset) ||
1375             (Start.getOffset() >= CurrentLoadedOffset &&
1376                 Start.getOffset()+Length < MaxLoadedOffset)) &&
1377            "Chunk is not valid SLoc address space");
1378     SourceLocation::UIntTy LocOffs = Loc.getOffset();
1379     SourceLocation::UIntTy BeginOffs = Start.getOffset();
1380     SourceLocation::UIntTy EndOffs = BeginOffs + Length;
1381     if (LocOffs >= BeginOffs && LocOffs < EndOffs) {
1382       if (RelativeOffset)
1383         *RelativeOffset = LocOffs - BeginOffs;
1384       return true;
1385     }
1386 
1387     return false;
1388   }
1389 
1390   /// Return true if both \p LHS and \p RHS are in the local source
1391   /// location address space or the loaded one.
1392   ///
1393   /// If it's true and \p RelativeOffset is non-null, it will be set to the
1394   /// offset of \p RHS relative to \p LHS.
isInSameSLocAddrSpace(SourceLocation LHS,SourceLocation RHS,SourceLocation::IntTy * RelativeOffset)1395   bool isInSameSLocAddrSpace(SourceLocation LHS, SourceLocation RHS,
1396                              SourceLocation::IntTy *RelativeOffset) const {
1397     SourceLocation::UIntTy LHSOffs = LHS.getOffset(), RHSOffs = RHS.getOffset();
1398     bool LHSLoaded = LHSOffs >= CurrentLoadedOffset;
1399     bool RHSLoaded = RHSOffs >= CurrentLoadedOffset;
1400 
1401     if (LHSLoaded == RHSLoaded) {
1402       if (RelativeOffset)
1403         *RelativeOffset = RHSOffs - LHSOffs;
1404       return true;
1405     }
1406 
1407     return false;
1408   }
1409 
1410   //===--------------------------------------------------------------------===//
1411   // Queries about the code at a SourceLocation.
1412   //===--------------------------------------------------------------------===//
1413 
1414   /// Return a pointer to the start of the specified location
1415   /// in the appropriate spelling MemoryBuffer.
1416   ///
1417   /// \param Invalid If non-NULL, will be set \c true if an error occurs.
1418   const char *getCharacterData(SourceLocation SL,
1419                                bool *Invalid = nullptr) const;
1420 
1421   /// Return the column # for the specified file position.
1422   ///
1423   /// This is significantly cheaper to compute than the line number.  This
1424   /// returns zero if the column number isn't known.  This may only be called
1425   /// on a file sloc, so you must choose a spelling or expansion location
1426   /// before calling this method.
1427   unsigned getColumnNumber(FileID FID, unsigned FilePos,
1428                            bool *Invalid = nullptr) const;
1429   unsigned getSpellingColumnNumber(SourceLocation Loc,
1430                                    bool *Invalid = nullptr) const;
1431   unsigned getExpansionColumnNumber(SourceLocation Loc,
1432                                     bool *Invalid = nullptr) const;
1433   unsigned getPresumedColumnNumber(SourceLocation Loc,
1434                                    bool *Invalid = nullptr) const;
1435 
1436   /// Given a SourceLocation, return the spelling line number
1437   /// for the position indicated.
1438   ///
1439   /// This requires building and caching a table of line offsets for the
1440   /// MemoryBuffer, so this is not cheap: use only when about to emit a
1441   /// diagnostic.
1442   unsigned getLineNumber(FileID FID, unsigned FilePos, bool *Invalid = nullptr) const;
1443   unsigned getSpellingLineNumber(SourceLocation Loc, bool *Invalid = nullptr) const;
1444   unsigned getExpansionLineNumber(SourceLocation Loc, bool *Invalid = nullptr) const;
1445   unsigned getPresumedLineNumber(SourceLocation Loc, bool *Invalid = nullptr) const;
1446 
1447   /// Return the filename or buffer identifier of the buffer the
1448   /// location is in.
1449   ///
1450   /// Note that this name does not respect \#line directives.  Use
1451   /// getPresumedLoc for normal clients.
1452   StringRef getBufferName(SourceLocation Loc, bool *Invalid = nullptr) const;
1453 
1454   /// Return the file characteristic of the specified source
1455   /// location, indicating whether this is a normal file, a system
1456   /// header, or an "implicit extern C" system header.
1457   ///
1458   /// This state can be modified with flags on GNU linemarker directives like:
1459   /// \code
1460   ///   # 4 "foo.h" 3
1461   /// \endcode
1462   /// which changes all source locations in the current file after that to be
1463   /// considered to be from a system header.
1464   SrcMgr::CharacteristicKind getFileCharacteristic(SourceLocation Loc) const;
1465 
1466   /// Returns the "presumed" location of a SourceLocation specifies.
1467   ///
1468   /// A "presumed location" can be modified by \#line or GNU line marker
1469   /// directives.  This provides a view on the data that a user should see
1470   /// in diagnostics, for example.
1471   ///
1472   /// Note that a presumed location is always given as the expansion point of
1473   /// an expansion location, not at the spelling location.
1474   ///
1475   /// \returns The presumed location of the specified SourceLocation. If the
1476   /// presumed location cannot be calculated (e.g., because \p Loc is invalid
1477   /// or the file containing \p Loc has changed on disk), returns an invalid
1478   /// presumed location.
1479   PresumedLoc getPresumedLoc(SourceLocation Loc,
1480                              bool UseLineDirectives = true) const;
1481 
1482   /// Returns whether the PresumedLoc for a given SourceLocation is
1483   /// in the main file.
1484   ///
1485   /// This computes the "presumed" location for a SourceLocation, then checks
1486   /// whether it came from a file other than the main file. This is different
1487   /// from isWrittenInMainFile() because it takes line marker directives into
1488   /// account.
1489   bool isInMainFile(SourceLocation Loc) const;
1490 
1491   /// Returns true if the spelling locations for both SourceLocations
1492   /// are part of the same file buffer.
1493   ///
1494   /// This check ignores line marker directives.
isWrittenInSameFile(SourceLocation Loc1,SourceLocation Loc2)1495   bool isWrittenInSameFile(SourceLocation Loc1, SourceLocation Loc2) const {
1496     return getFileID(Loc1) == getFileID(Loc2);
1497   }
1498 
1499   /// Returns true if the spelling location for the given location
1500   /// is in the main file buffer.
1501   ///
1502   /// This check ignores line marker directives.
isWrittenInMainFile(SourceLocation Loc)1503   bool isWrittenInMainFile(SourceLocation Loc) const {
1504     return getFileID(Loc) == getMainFileID();
1505   }
1506 
1507   /// Returns whether \p Loc is located in a <built-in> file.
isWrittenInBuiltinFile(SourceLocation Loc)1508   bool isWrittenInBuiltinFile(SourceLocation Loc) const {
1509     PresumedLoc Presumed = getPresumedLoc(Loc);
1510     if (Presumed.isInvalid())
1511       return false;
1512     StringRef Filename(Presumed.getFilename());
1513     return Filename == "<built-in>";
1514   }
1515 
1516   /// Returns whether \p Loc is located in a <command line> file.
isWrittenInCommandLineFile(SourceLocation Loc)1517   bool isWrittenInCommandLineFile(SourceLocation Loc) const {
1518     PresumedLoc Presumed = getPresumedLoc(Loc);
1519     if (Presumed.isInvalid())
1520       return false;
1521     StringRef Filename(Presumed.getFilename());
1522     return Filename == "<command line>";
1523   }
1524 
1525   /// Returns whether \p Loc is located in a <scratch space> file.
isWrittenInScratchSpace(SourceLocation Loc)1526   bool isWrittenInScratchSpace(SourceLocation Loc) const {
1527     PresumedLoc Presumed = getPresumedLoc(Loc);
1528     if (Presumed.isInvalid())
1529       return false;
1530     StringRef Filename(Presumed.getFilename());
1531     return Filename == "<scratch space>";
1532   }
1533 
1534   /// Returns whether \p Loc is located in a built-in or command line source.
isInPredefinedFile(SourceLocation Loc)1535   bool isInPredefinedFile(SourceLocation Loc) const {
1536     PresumedLoc Presumed = getPresumedLoc(Loc);
1537     if (Presumed.isInvalid())
1538       return false;
1539     StringRef Filename(Presumed.getFilename());
1540     return Filename == "<built-in>" || Filename == "<command line>";
1541   }
1542 
1543   /// Returns if a SourceLocation is in a system header.
isInSystemHeader(SourceLocation Loc)1544   bool isInSystemHeader(SourceLocation Loc) const {
1545     if (Loc.isInvalid())
1546       return false;
1547     return isSystem(getFileCharacteristic(Loc));
1548   }
1549 
1550   /// Returns if a SourceLocation is in an "extern C" system header.
isInExternCSystemHeader(SourceLocation Loc)1551   bool isInExternCSystemHeader(SourceLocation Loc) const {
1552     return getFileCharacteristic(Loc) == SrcMgr::C_ExternCSystem;
1553   }
1554 
1555   /// Returns whether \p Loc is expanded from a macro in a system header.
isInSystemMacro(SourceLocation loc)1556   bool isInSystemMacro(SourceLocation loc) const {
1557     if (!loc.isMacroID())
1558       return false;
1559 
1560     // This happens when the macro is the result of a paste, in that case
1561     // its spelling is the scratch memory, so we take the parent context.
1562     // There can be several level of token pasting.
1563     if (isWrittenInScratchSpace(getSpellingLoc(loc))) {
1564       do {
1565         loc = getImmediateMacroCallerLoc(loc);
1566       } while (isWrittenInScratchSpace(getSpellingLoc(loc)));
1567       return isInSystemMacro(loc);
1568     }
1569 
1570     return isInSystemHeader(getSpellingLoc(loc));
1571   }
1572 
1573   /// The size of the SLocEntry that \p FID represents.
1574   unsigned getFileIDSize(FileID FID) const;
1575 
1576   /// Given a specific FileID, returns true if \p Loc is inside that
1577   /// FileID chunk and sets relative offset (offset of \p Loc from beginning
1578   /// of FileID) to \p relativeOffset.
1579   bool isInFileID(SourceLocation Loc, FileID FID,
1580                   unsigned *RelativeOffset = nullptr) const {
1581     SourceLocation::UIntTy Offs = Loc.getOffset();
1582     if (isOffsetInFileID(FID, Offs)) {
1583       if (RelativeOffset)
1584         *RelativeOffset = Offs - getSLocEntry(FID).getOffset();
1585       return true;
1586     }
1587 
1588     return false;
1589   }
1590 
1591   //===--------------------------------------------------------------------===//
1592   // Line Table Manipulation Routines
1593   //===--------------------------------------------------------------------===//
1594 
1595   /// Return the uniqued ID for the specified filename.
1596   unsigned getLineTableFilenameID(StringRef Str);
1597 
1598   /// Add a line note to the line table for the FileID and offset
1599   /// specified by Loc.
1600   ///
1601   /// If FilenameID is -1, it is considered to be unspecified.
1602   void AddLineNote(SourceLocation Loc, unsigned LineNo, int FilenameID,
1603                    bool IsFileEntry, bool IsFileExit,
1604                    SrcMgr::CharacteristicKind FileKind);
1605 
1606   /// Determine if the source manager has a line table.
hasLineTable()1607   bool hasLineTable() const { return LineTable != nullptr; }
1608 
1609   /// Retrieve the stored line table.
1610   LineTableInfo &getLineTable();
1611 
1612   //===--------------------------------------------------------------------===//
1613   // Queries for performance analysis.
1614   //===--------------------------------------------------------------------===//
1615 
1616   /// Return the total amount of physical memory allocated by the
1617   /// ContentCache allocator.
getContentCacheSize()1618   size_t getContentCacheSize() const {
1619     return ContentCacheAlloc.getTotalMemory();
1620   }
1621 
1622   struct MemoryBufferSizes {
1623     const size_t malloc_bytes;
1624     const size_t mmap_bytes;
1625 
MemoryBufferSizesMemoryBufferSizes1626     MemoryBufferSizes(size_t malloc_bytes, size_t mmap_bytes)
1627       : malloc_bytes(malloc_bytes), mmap_bytes(mmap_bytes) {}
1628   };
1629 
1630   /// Return the amount of memory used by memory buffers, breaking down
1631   /// by heap-backed versus mmap'ed memory.
1632   MemoryBufferSizes getMemoryBufferSizes() const;
1633 
1634   /// Return the amount of memory used for various side tables and
1635   /// data structures in the SourceManager.
1636   size_t getDataStructureSizes() const;
1637 
1638   //===--------------------------------------------------------------------===//
1639   // Other miscellaneous methods.
1640   //===--------------------------------------------------------------------===//
1641 
1642   /// Get the source location for the given file:line:col triplet.
1643   ///
1644   /// If the source file is included multiple times, the source location will
1645   /// be based upon the first inclusion.
1646   SourceLocation translateFileLineCol(const FileEntry *SourceFile,
1647                                       unsigned Line, unsigned Col) const;
1648 
1649   /// Get the FileID for the given file.
1650   ///
1651   /// If the source file is included multiple times, the FileID will be the
1652   /// first inclusion.
1653   FileID translateFile(const FileEntry *SourceFile) const;
translateFile(FileEntryRef SourceFile)1654   FileID translateFile(FileEntryRef SourceFile) const {
1655     return translateFile(&SourceFile.getFileEntry());
1656   }
1657 
1658   /// Get the source location in \p FID for the given line:col.
1659   /// Returns null location if \p FID is not a file SLocEntry.
1660   SourceLocation translateLineCol(FileID FID,
1661                                   unsigned Line, unsigned Col) const;
1662 
1663   /// If \p Loc points inside a function macro argument, the returned
1664   /// location will be the macro location in which the argument was expanded.
1665   /// If a macro argument is used multiple times, the expanded location will
1666   /// be at the first expansion of the argument.
1667   /// e.g.
1668   ///   MY_MACRO(foo);
1669   ///             ^
1670   /// Passing a file location pointing at 'foo', will yield a macro location
1671   /// where 'foo' was expanded into.
1672   SourceLocation getMacroArgExpandedLocation(SourceLocation Loc) const;
1673 
1674   /// Determines the order of 2 source locations in the translation unit.
1675   ///
1676   /// \returns true if LHS source location comes before RHS, false otherwise.
1677   bool isBeforeInTranslationUnit(SourceLocation LHS, SourceLocation RHS) const;
1678 
1679   /// Determines whether the two decomposed source location is in the
1680   ///        same translation unit. As a byproduct, it also calculates the order
1681   ///        of the source locations in case they are in the same TU.
1682   ///
1683   /// \returns Pair of bools the first component is true if the two locations
1684   ///          are in the same TU. The second bool is true if the first is true
1685   ///          and \p LOffs is before \p ROffs.
1686   std::pair<bool, bool>
1687   isInTheSameTranslationUnit(FileIDAndOffset &LOffs,
1688                              FileIDAndOffset &ROffs) const;
1689 
1690   /// \param Loc a source location in a loaded AST (of a PCH/Module file).
1691   /// \returns a FileID uniquely identifies the AST of a loaded
1692   /// module/PCH where `Loc` is at.
1693   FileID getUniqueLoadedASTFileID(SourceLocation Loc) const;
1694 
1695   /// Determines whether the two decomposed source location is in the same TU.
1696   bool isInTheSameTranslationUnitImpl(const FileIDAndOffset &LOffs,
1697                                       const FileIDAndOffset &ROffs) const;
1698 
1699   /// Determines the order of 2 source locations in the "source location
1700   /// address space".
isBeforeInSLocAddrSpace(SourceLocation LHS,SourceLocation RHS)1701   bool isBeforeInSLocAddrSpace(SourceLocation LHS, SourceLocation RHS) const {
1702     return isBeforeInSLocAddrSpace(LHS, RHS.getOffset());
1703   }
1704 
1705   /// Determines the order of a source location and a source location
1706   /// offset in the "source location address space".
1707   ///
1708   /// Note that we always consider source locations loaded from
isBeforeInSLocAddrSpace(SourceLocation LHS,SourceLocation::UIntTy RHS)1709   bool isBeforeInSLocAddrSpace(SourceLocation LHS,
1710                                SourceLocation::UIntTy RHS) const {
1711     SourceLocation::UIntTy LHSOffset = LHS.getOffset();
1712     bool LHSLoaded = LHSOffset >= CurrentLoadedOffset;
1713     bool RHSLoaded = RHS >= CurrentLoadedOffset;
1714     if (LHSLoaded == RHSLoaded)
1715       return LHSOffset < RHS;
1716 
1717     return LHSLoaded;
1718   }
1719 
1720   /// Return true if the Point is within Start and End.
isPointWithin(SourceLocation Location,SourceLocation Start,SourceLocation End)1721   bool isPointWithin(SourceLocation Location, SourceLocation Start,
1722                      SourceLocation End) const {
1723     return Location == Start || Location == End ||
1724            (isBeforeInTranslationUnit(Start, Location) &&
1725             isBeforeInTranslationUnit(Location, End));
1726   }
1727 
1728   // Iterators over FileInfos.
1729   using fileinfo_iterator =
1730       llvm::DenseMap<FileEntryRef, SrcMgr::ContentCache *>::const_iterator;
1731 
fileinfo_begin()1732   fileinfo_iterator fileinfo_begin() const { return FileInfos.begin(); }
fileinfo_end()1733   fileinfo_iterator fileinfo_end() const { return FileInfos.end(); }
hasFileInfo(const FileEntry * File)1734   bool hasFileInfo(const FileEntry *File) const {
1735     return FileInfos.find_as(File) != FileInfos.end();
1736   }
1737 
1738   /// Print statistics to stderr.
1739   void PrintStats() const;
1740 
1741   void dump() const;
1742 
1743   // Produce notes describing the current source location address space usage.
1744   void noteSLocAddressSpaceUsage(DiagnosticsEngine &Diag,
1745                                  std::optional<unsigned> MaxNotes = 32) const;
1746 
1747   /// Get the number of local SLocEntries we have.
local_sloc_entry_size()1748   unsigned local_sloc_entry_size() const { return LocalSLocEntryTable.size(); }
1749 
1750   /// Get a local SLocEntry. This is exposed for indexing.
getLocalSLocEntry(unsigned Index)1751   const SrcMgr::SLocEntry &getLocalSLocEntry(unsigned Index) const {
1752     return const_cast<SourceManager *>(this)->getLocalSLocEntry(Index);
1753   }
1754 
1755   /// Get a local SLocEntry. This is exposed for indexing.
getLocalSLocEntry(unsigned Index)1756   SrcMgr::SLocEntry &getLocalSLocEntry(unsigned Index) {
1757     assert(Index < LocalSLocEntryTable.size() && "Invalid index");
1758     return LocalSLocEntryTable[Index];
1759   }
1760 
1761   /// Get the number of loaded SLocEntries we have.
loaded_sloc_entry_size()1762   unsigned loaded_sloc_entry_size() const { return LoadedSLocEntryTable.size();}
1763 
1764   /// Get a loaded SLocEntry. This is exposed for indexing.
1765   const SrcMgr::SLocEntry &getLoadedSLocEntry(unsigned Index,
1766                                               bool *Invalid = nullptr) const {
1767     return const_cast<SourceManager *>(this)->getLoadedSLocEntry(Index,
1768                                                                  Invalid);
1769   }
1770 
1771   /// Get a loaded SLocEntry. This is exposed for indexing.
1772   SrcMgr::SLocEntry &getLoadedSLocEntry(unsigned Index,
1773                                         bool *Invalid = nullptr) {
1774     assert(Index < LoadedSLocEntryTable.size() && "Invalid index");
1775     if (SLocEntryLoaded[Index])
1776       return LoadedSLocEntryTable[Index];
1777     return loadSLocEntry(Index, Invalid);
1778   }
1779 
1780   const SrcMgr::SLocEntry &getSLocEntry(FileID FID,
1781                                         bool *Invalid = nullptr) const {
1782     return const_cast<SourceManager *>(this)->getSLocEntry(FID, Invalid);
1783   }
1784 
1785   SrcMgr::SLocEntry &getSLocEntry(FileID FID, bool *Invalid = nullptr) {
1786     if (FID.ID == 0 || FID.ID == -1) {
1787       if (Invalid) *Invalid = true;
1788       return LocalSLocEntryTable[0];
1789     }
1790     return getSLocEntryByID(FID.ID, Invalid);
1791   }
1792 
getNextLocalOffset()1793   SourceLocation::UIntTy getNextLocalOffset() const { return NextLocalOffset; }
1794 
setExternalSLocEntrySource(ExternalSLocEntrySource * Source)1795   void setExternalSLocEntrySource(ExternalSLocEntrySource *Source) {
1796     assert(LoadedSLocEntryTable.empty() &&
1797            "Invalidating existing loaded entries");
1798     ExternalSLocEntries = Source;
1799   }
1800 
1801   /// Allocate a number of loaded SLocEntries, which will be actually
1802   /// loaded on demand from the external source.
1803   ///
1804   /// NumSLocEntries will be allocated, which occupy a total of TotalSize space
1805   /// in the global source view. The lowest ID and the base offset of the
1806   /// entries will be returned.
1807   std::pair<int, SourceLocation::UIntTy>
1808   AllocateLoadedSLocEntries(unsigned NumSLocEntries,
1809                             SourceLocation::UIntTy TotalSize);
1810 
1811   /// Returns true if \p Loc came from a PCH/Module.
isLoadedSourceLocation(SourceLocation Loc)1812   bool isLoadedSourceLocation(SourceLocation Loc) const {
1813     return isLoadedOffset(Loc.getOffset());
1814   }
1815 
1816   /// Returns true if \p Loc did not come from a PCH/Module.
isLocalSourceLocation(SourceLocation Loc)1817   bool isLocalSourceLocation(SourceLocation Loc) const {
1818     return isLocalOffset(Loc.getOffset());
1819   }
1820 
1821   /// Returns true if \p FID came from a PCH/Module.
isLoadedFileID(FileID FID)1822   bool isLoadedFileID(FileID FID) const {
1823     assert(FID.ID != -1 && "Using FileID sentinel value");
1824     return FID.ID < 0;
1825   }
1826 
1827   /// Returns true if \p FID did not come from a PCH/Module.
isLocalFileID(FileID FID)1828   bool isLocalFileID(FileID FID) const {
1829     return !isLoadedFileID(FID);
1830   }
1831 
1832   /// Gets the location of the immediate macro caller, one level up the stack
1833   /// toward the initial macro typed into the source.
getImmediateMacroCallerLoc(SourceLocation Loc)1834   SourceLocation getImmediateMacroCallerLoc(SourceLocation Loc) const {
1835     if (!Loc.isMacroID()) return Loc;
1836 
1837     // When we have the location of (part of) an expanded parameter, its
1838     // spelling location points to the argument as expanded in the macro call,
1839     // and therefore is used to locate the macro caller.
1840     if (isMacroArgExpansion(Loc))
1841       return getImmediateSpellingLoc(Loc);
1842 
1843     // Otherwise, the caller of the macro is located where this macro is
1844     // expanded (while the spelling is part of the macro definition).
1845     return getImmediateExpansionRange(Loc).getBegin();
1846   }
1847 
1848   /// \return Location of the top-level macro caller.
1849   SourceLocation getTopMacroCallerLoc(SourceLocation Loc) const;
1850 
1851 private:
1852   friend class ASTReader;
1853   friend class ASTWriter;
1854 
1855   llvm::MemoryBufferRef getFakeBufferForRecovery() const;
1856   SrcMgr::ContentCache &getFakeContentCacheForRecovery() const;
1857 
1858   const SrcMgr::SLocEntry &loadSLocEntry(unsigned Index, bool *Invalid) const;
1859   SrcMgr::SLocEntry &loadSLocEntry(unsigned Index, bool *Invalid);
1860 
getSLocEntryOrNull(FileID FID)1861   const SrcMgr::SLocEntry *getSLocEntryOrNull(FileID FID) const {
1862     return const_cast<SourceManager *>(this)->getSLocEntryOrNull(FID);
1863   }
1864 
getSLocEntryOrNull(FileID FID)1865   SrcMgr::SLocEntry *getSLocEntryOrNull(FileID FID) {
1866     bool Invalid = false;
1867     SrcMgr::SLocEntry &Entry = getSLocEntry(FID, &Invalid);
1868     return Invalid ? nullptr : &Entry;
1869   }
1870 
getSLocEntryForFile(FileID FID)1871   const SrcMgr::SLocEntry *getSLocEntryForFile(FileID FID) const {
1872     return const_cast<SourceManager *>(this)->getSLocEntryForFile(FID);
1873   }
1874 
getSLocEntryForFile(FileID FID)1875   SrcMgr::SLocEntry *getSLocEntryForFile(FileID FID) {
1876     if (auto *Entry = getSLocEntryOrNull(FID))
1877       if (Entry->isFile())
1878         return Entry;
1879     return nullptr;
1880   }
1881 
1882   /// Get the entry with the given unwrapped FileID.
1883   /// Invalid will not be modified for Local IDs.
1884   const SrcMgr::SLocEntry &getSLocEntryByID(int ID,
1885                                             bool *Invalid = nullptr) const {
1886     return const_cast<SourceManager *>(this)->getSLocEntryByID(ID, Invalid);
1887   }
1888 
1889   SrcMgr::SLocEntry &getSLocEntryByID(int ID, bool *Invalid = nullptr) {
1890     assert(ID != -1 && "Using FileID sentinel value");
1891     if (ID < 0)
1892       return getLoadedSLocEntryByID(ID, Invalid);
1893     return getLocalSLocEntry(static_cast<unsigned>(ID));
1894   }
1895 
1896   const SrcMgr::SLocEntry &
1897   getLoadedSLocEntryByID(int ID, bool *Invalid = nullptr) const {
1898     return const_cast<SourceManager *>(this)->getLoadedSLocEntryByID(ID,
1899                                                                      Invalid);
1900   }
1901 
1902   SrcMgr::SLocEntry &getLoadedSLocEntryByID(int ID, bool *Invalid = nullptr) {
1903     return getLoadedSLocEntry(static_cast<unsigned>(-ID - 2), Invalid);
1904   }
1905 
getFileID(SourceLocation::UIntTy SLocOffset)1906   FileID getFileID(SourceLocation::UIntTy SLocOffset) const {
1907     // If our one-entry cache covers this offset, just return it.
1908     if (SLocOffset >= LastLookupStartOffset && SLocOffset < LastLookupEndOffset)
1909       return LastFileIDLookup;
1910     return getFileIDSlow(SLocOffset);
1911   }
1912 
isLocalOffset(SourceLocation::UIntTy SLocOffset)1913   bool isLocalOffset(SourceLocation::UIntTy SLocOffset) const {
1914     return SLocOffset < CurrentLoadedOffset;
1915   }
1916 
isLoadedOffset(SourceLocation::UIntTy SLocOffset)1917   bool isLoadedOffset(SourceLocation::UIntTy SLocOffset) const {
1918     return SLocOffset >= CurrentLoadedOffset;
1919   }
1920 
1921   /// Implements the common elements of storing an expansion info struct into
1922   /// the SLocEntry table and producing a source location that refers to it.
1923   SourceLocation
1924   createExpansionLocImpl(const SrcMgr::ExpansionInfo &Expansion,
1925                          unsigned Length, int LoadedID = 0,
1926                          SourceLocation::UIntTy LoadedOffset = 0);
1927 
1928   /// Return true if the specified FileID contains the
1929   /// specified SourceLocation offset.  This is a very hot method.
isOffsetInFileID(FileID FID,SourceLocation::UIntTy SLocOffset)1930   inline bool isOffsetInFileID(FileID FID,
1931                                SourceLocation::UIntTy SLocOffset) const {
1932     const SrcMgr::SLocEntry &Entry = getSLocEntry(FID);
1933     // If the entry is after the offset, it can't contain it.
1934     if (SLocOffset < Entry.getOffset()) return false;
1935 
1936     // If this is the very last entry then it does.
1937     if (FID.ID == -2)
1938       return true;
1939 
1940     // If it is the last local entry, then it does if the location is local.
1941     if (FID.ID+1 == static_cast<int>(LocalSLocEntryTable.size()))
1942       return SLocOffset < NextLocalOffset;
1943 
1944     // Otherwise, the entry after it has to not include it. This works for both
1945     // local and loaded entries.
1946     return SLocOffset < getSLocEntryByID(FID.ID+1).getOffset();
1947   }
1948 
1949   /// Returns the previous in-order FileID or an invalid FileID if there
1950   /// is no previous one.
1951   FileID getPreviousFileID(FileID FID) const;
1952 
1953   /// Returns the next in-order FileID or an invalid FileID if there is
1954   /// no next one.
1955   FileID getNextFileID(FileID FID) const;
1956 
1957   /// Create a new fileID for the specified ContentCache and
1958   /// include position.
1959   ///
1960   /// This works regardless of whether the ContentCache corresponds to a
1961   /// file or some other input source.
1962   FileID createFileIDImpl(SrcMgr::ContentCache &File, StringRef Filename,
1963                           SourceLocation IncludePos,
1964                           SrcMgr::CharacteristicKind DirCharacter, int LoadedID,
1965                           SourceLocation::UIntTy LoadedOffset);
1966 
1967   SrcMgr::ContentCache &getOrCreateContentCache(FileEntryRef SourceFile,
1968                                                 bool isSystemFile = false);
1969 
1970   /// Create a new ContentCache for the specified  memory buffer.
1971   SrcMgr::ContentCache &
1972   createMemBufferContentCache(std::unique_ptr<llvm::MemoryBuffer> Buf);
1973 
1974   FileID getFileIDSlow(SourceLocation::UIntTy SLocOffset) const;
1975   FileID getFileIDLocal(SourceLocation::UIntTy SLocOffset) const;
1976   FileID getFileIDLoaded(SourceLocation::UIntTy SLocOffset) const;
1977 
1978   SourceLocation getExpansionLocSlowCase(SourceLocation Loc) const;
1979   SourceLocation getSpellingLocSlowCase(SourceLocation Loc) const;
1980   SourceLocation getFileLocSlowCase(SourceLocation Loc) const;
1981 
1982   FileIDAndOffset
1983   getDecomposedExpansionLocSlowCase(const SrcMgr::SLocEntry *E) const;
1984   FileIDAndOffset getDecomposedSpellingLocSlowCase(const SrcMgr::SLocEntry *E,
1985                                                    unsigned Offset) const;
1986   void computeMacroArgsCache(MacroArgsMap &MacroArgsCache, FileID FID) const;
1987   void associateFileChunkWithMacroArgExp(MacroArgsMap &MacroArgsCache,
1988                                          FileID FID,
1989                                          SourceLocation SpellLoc,
1990                                          SourceLocation ExpansionLoc,
1991                                          unsigned ExpansionLength) const;
1992   void updateSlocUsageStats() const;
1993 };
1994 
1995 /// Comparison function object.
1996 template<typename T>
1997 class BeforeThanCompare;
1998 
1999 /// Compare two source locations.
2000 template<>
2001 class BeforeThanCompare<SourceLocation> {
2002   SourceManager &SM;
2003 
2004 public:
BeforeThanCompare(SourceManager & SM)2005   explicit BeforeThanCompare(SourceManager &SM) : SM(SM) {}
2006 
operator()2007   bool operator()(SourceLocation LHS, SourceLocation RHS) const {
2008     return SM.isBeforeInTranslationUnit(LHS, RHS);
2009   }
2010 };
2011 
2012 /// Compare two non-overlapping source ranges.
2013 template<>
2014 class BeforeThanCompare<SourceRange> {
2015   SourceManager &SM;
2016 
2017 public:
BeforeThanCompare(SourceManager & SM)2018   explicit BeforeThanCompare(SourceManager &SM) : SM(SM) {}
2019 
operator()2020   bool operator()(SourceRange LHS, SourceRange RHS) const {
2021     return SM.isBeforeInTranslationUnit(LHS.getBegin(), RHS.getBegin());
2022   }
2023 };
2024 
2025 /// SourceManager and necessary dependencies (e.g. VFS, FileManager) for a
2026 /// single in-memorty file.
2027 class SourceManagerForFile {
2028 public:
2029   /// Creates SourceManager and necessary dependencies (e.g. VFS, FileManager).
2030   /// The main file in the SourceManager will be \p FileName with \p Content.
2031   SourceManagerForFile(StringRef FileName, StringRef Content);
2032 
get()2033   SourceManager &get() {
2034     assert(SourceMgr);
2035     return *SourceMgr;
2036   }
2037 
2038 private:
2039   // The order of these fields are important - they should be in the same order
2040   // as they are created in `createSourceManagerForFile` so that they can be
2041   // deleted in the reverse order as they are created.
2042   std::unique_ptr<FileManager> FileMgr;
2043   std::unique_ptr<DiagnosticOptions> DiagOpts;
2044   std::unique_ptr<DiagnosticsEngine> Diagnostics;
2045   std::unique_ptr<SourceManager> SourceMgr;
2046 };
2047 
2048 } // namespace clang
2049 
2050 #endif // LLVM_CLANG_BASIC_SOURCEMANAGER_H
2051