1//===- SourceLocation.h - Compact identifier for 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 clang::SourceLocation class and associated facilities.
11//
12//===----------------------------------------------------------------------===//
13
14#ifndef LLVM_CLANG_BASIC_SOURCELOCATION_H
15#define LLVM_CLANG_BASIC_SOURCELOCATION_H
16
17#include "clang/Basic/LLVM.h"
18#include "llvm/ADT/StringRef.h"
19#include "llvm/Support/PointerLikeTypeTraits.h"
20#include <cassert>
21#include <cstdint>
22#include <string>
23#include <utility>
24
25namespace llvm {
26
27template <typename T> struct DenseMapInfo;
28
29class FoldingSetNodeID;
30template <typename T> struct FoldingSetTrait;
31
32} // namespace llvm
33
34namespace clang {
35
36class SourceManager;
37
38/// An opaque identifier used by SourceManager which refers to a
39/// source file (MemoryBuffer) along with its \#include path and \#line data.
40///
41class FileID {
42 /// A mostly-opaque identifier, where 0 is "invalid", >0 is
43 /// this module, and <-1 is something loaded from another module.
44 int ID = 0;
45
46public:
47 bool isValid() const { return ID != 0; }
48 bool isInvalid() const { return ID == 0; }
49
50 bool operator==(const FileID &RHS) const { return ID == RHS.ID; }
51 bool operator<(const FileID &RHS) const { return ID < RHS.ID; }
52 bool operator<=(const FileID &RHS) const { return ID <= RHS.ID; }
53 bool operator!=(const FileID &RHS) const { return !(*this == RHS); }
54 bool operator>(const FileID &RHS) const { return RHS < *this; }
55 bool operator>=(const FileID &RHS) const { return RHS <= *this; }
56
57 static FileID getSentinel() { return get(-1); }
58 unsigned getHashValue() const { return static_cast<unsigned>(ID); }
59
60private:
61 friend class ASTWriter;
62 friend class ASTReader;
63 friend class SourceManager;
64
65 static FileID get(int V) {
66 FileID F;
67 F.ID = V;
68 return F;
69 }
70
71 int getOpaqueValue() const { return ID; }
72};
73
74/// Encodes a location in the source. The SourceManager can decode this
75/// to get at the full include stack, line and column information.
76///
77/// Technically, a source location is simply an offset into the manager's view
78/// of the input source, which is all input buffers (including macro
79/// expansions) concatenated in an effectively arbitrary order. The manager
80/// actually maintains two blocks of input buffers. One, starting at offset
81/// 0 and growing upwards, contains all buffers from this module. The other,
82/// starting at the highest possible offset and growing downwards, contains
83/// buffers of loaded modules.
84///
85/// In addition, one bit of SourceLocation is used for quick access to the
86/// information whether the location is in a file or a macro expansion.
87///
88/// It is important that this type remains small. It is currently 32 bits wide.
89class SourceLocation {
90 friend class ASTReader;
91 friend class ASTWriter;
92 friend class SourceManager;
93 friend struct llvm::FoldingSetTrait<SourceLocation>;
94
95 unsigned ID = 0;
96
97 enum : unsigned {
98 MacroIDBit = 1U << 31
99 };
100
101public:
102 bool isFileID() const { return (ID & MacroIDBit) == 0; }
103 bool isMacroID() const { return (ID & MacroIDBit) != 0; }
104
105 /// Return true if this is a valid SourceLocation object.
106 ///
107 /// Invalid SourceLocations are often used when events have no corresponding
108 /// location in the source (e.g. a diagnostic is required for a command line
109 /// option).
110 bool isValid() const { return ID != 0; }
111 bool isInvalid() const { return ID == 0; }
112
113private:
114 /// Return the offset into the manager's global input view.
115 unsigned getOffset() const {
116 return ID & ~MacroIDBit;
117 }
118
119 static SourceLocation getFileLoc(unsigned ID) {
120 assert((ID & MacroIDBit) == 0 && "Ran out of source locations!");
121 SourceLocation L;
122 L.ID = ID;
123 return L;
124 }
125
126 static SourceLocation getMacroLoc(unsigned ID) {
127 assert((ID & MacroIDBit) == 0 && "Ran out of source locations!");
128 SourceLocation L;
129 L.ID = MacroIDBit | ID;
130 return L;
131 }
132
133public:
134 /// Return a source location with the specified offset from this
135 /// SourceLocation.
136 SourceLocation getLocWithOffset(int Offset) const {
137 assert(((getOffset()+Offset) & MacroIDBit) == 0 && "offset overflow");
138 SourceLocation L;
139 L.ID = ID+Offset;
140 return L;
141 }
142
143 /// When a SourceLocation itself cannot be used, this returns
144 /// an (opaque) 32-bit integer encoding for it.
145 ///
146 /// This should only be passed to SourceLocation::getFromRawEncoding, it
147 /// should not be inspected directly.
148 unsigned getRawEncoding() const { return ID; }
149
150 /// Turn a raw encoding of a SourceLocation object into
151 /// a real SourceLocation.
152 ///
153 /// \see getRawEncoding.
154 static SourceLocation getFromRawEncoding(unsigned Encoding) {
155 SourceLocation X;
156 X.ID = Encoding;
157 return X;
158 }
159
160 /// When a SourceLocation itself cannot be used, this returns
161 /// an (opaque) pointer encoding for it.
162 ///
163 /// This should only be passed to SourceLocation::getFromPtrEncoding, it
164 /// should not be inspected directly.
165 void* getPtrEncoding() const {
166 // Double cast to avoid a warning "cast to pointer from integer of different
167 // size".
168 return (void*)(uintptr_t)getRawEncoding();
169 }
170
171 /// Turn a pointer encoding of a SourceLocation object back
172 /// into a real SourceLocation.
173 static SourceLocation getFromPtrEncoding(const void *Encoding) {
174 return getFromRawEncoding((unsigned)(uintptr_t)Encoding);
175 }
176
177 static bool isPairOfFileLocations(SourceLocation Start, SourceLocation End) {
178 return Start.isValid() && Start.isFileID() && End.isValid() &&
179 End.isFileID();
180 }
181
182 unsigned getHashValue() const;
183 void print(raw_ostream &OS, const SourceManager &SM) const;
184 std::string printToString(const SourceManager &SM) const;
185 void dump(const SourceManager &SM) const;
186};
187
188inline bool operator==(const SourceLocation &LHS, const SourceLocation &RHS) {
189 return LHS.getRawEncoding() == RHS.getRawEncoding();
190}
191
192inline bool operator!=(const SourceLocation &LHS, const SourceLocation &RHS) {
193 return !(LHS == RHS);
194}
195
196// Ordering is meaningful only if LHS and RHS have the same FileID!
197// Otherwise use SourceManager::isBeforeInTranslationUnit().
198inline bool operator<(const SourceLocation &LHS, const SourceLocation &RHS) {
199 return LHS.getRawEncoding() < RHS.getRawEncoding();
200}
201inline bool operator>(const SourceLocation &LHS, const SourceLocation &RHS) {
202 return LHS.getRawEncoding() > RHS.getRawEncoding();
203}
204inline bool operator<=(const SourceLocation &LHS, const SourceLocation &RHS) {
205 return LHS.getRawEncoding() <= RHS.getRawEncoding();
206}
207inline bool operator>=(const SourceLocation &LHS, const SourceLocation &RHS) {
208 return LHS.getRawEncoding() >= RHS.getRawEncoding();
209}
210
211/// A trivial tuple used to represent a source range.
212class SourceRange {
213 SourceLocation B;
214 SourceLocation E;
215
216public:
217 SourceRange() = default;
218 SourceRange(SourceLocation loc) : B(loc), E(loc) {}
219 SourceRange(SourceLocation begin, SourceLocation end) : B(begin), E(end) {}
220
221 SourceLocation getBegin() const { return B; }
222 SourceLocation getEnd() const { return E; }
223
224 void setBegin(SourceLocation b) { B = b; }
225 void setEnd(SourceLocation e) { E = e; }
226
227 bool isValid() const { return B.isValid() && E.isValid(); }
228 bool isInvalid() const { return !isValid(); }
229
230 bool operator==(const SourceRange &X) const {
231 return B == X.B && E == X.E;
232 }
233
234 bool operator!=(const SourceRange &X) const {
235 return B != X.B || E != X.E;
236 }
237
238 // Returns true iff other is wholly contained within this range.
239 bool fullyContains(const SourceRange &other) const {
240 return B <= other.B && E >= other.E;
241 }
242
243 void print(raw_ostream &OS, const SourceManager &SM) const;
244 std::string printToString(const SourceManager &SM) const;
245 void dump(const SourceManager &SM) const;
246};
247
248/// Represents a character-granular source range.
249///
250/// The underlying SourceRange can either specify the starting/ending character
251/// of the range, or it can specify the start of the range and the start of the
252/// last token of the range (a "token range"). In the token range case, the
253/// size of the last token must be measured to determine the actual end of the
254/// range.
255class CharSourceRange {
256 SourceRange Range;
257 bool IsTokenRange = false;
258
259public:
260 CharSourceRange() = default;
261 CharSourceRange(SourceRange R, bool ITR) : Range(R), IsTokenRange(ITR) {}
262
263 static CharSourceRange getTokenRange(SourceRange R) {
264 return CharSourceRange(R, true);
265 }
266
267 static CharSourceRange getCharRange(SourceRange R) {
268 return CharSourceRange(R, false);
269 }
270
271 static CharSourceRange getTokenRange(SourceLocation B, SourceLocation E) {
272 return getTokenRange(SourceRange(B, E));
273 }
274
275 static CharSourceRange getCharRange(SourceLocation B, SourceLocation E) {
276 return getCharRange(SourceRange(B, E));
277 }
278
279 /// Return true if the end of this range specifies the start of
280 /// the last token. Return false if the end of this range specifies the last
281 /// character in the range.
282 bool isTokenRange() const { return IsTokenRange; }
283 bool isCharRange() const { return !IsTokenRange; }
284
285 SourceLocation getBegin() const { return Range.getBegin(); }
286 SourceLocation getEnd() const { return Range.getEnd(); }
287 SourceRange getAsRange() const { return Range; }
288
289 void setBegin(SourceLocation b) { Range.setBegin(b); }
290 void setEnd(SourceLocation e) { Range.setEnd(e); }
291 void setTokenRange(bool TR) { IsTokenRange = TR; }
292
293 bool isValid() const { return Range.isValid(); }
294 bool isInvalid() const { return !isValid(); }
295};
296
297/// Represents an unpacked "presumed" location which can be presented
298/// to the user.
299///
300/// A 'presumed' location can be modified by \#line and GNU line marker
301/// directives and is always the expansion point of a normal location.
302///
303/// You can get a PresumedLoc from a SourceLocation with SourceManager.
304class PresumedLoc {
305 const char *Filename = nullptr;
306 FileID ID;
307 unsigned Line, Col;
308 SourceLocation IncludeLoc;
309
310public:
311 PresumedLoc() = default;
312 PresumedLoc(const char *FN, FileID FID, unsigned Ln, unsigned Co,
313 SourceLocation IL)
314 : Filename(FN), ID(FID), Line(Ln), Col(Co), IncludeLoc(IL) {}
315
316 /// Return true if this object is invalid or uninitialized.
317 ///
318 /// This occurs when created with invalid source locations or when walking
319 /// off the top of a \#include stack.
320 bool isInvalid() const { return Filename == nullptr; }
321 bool isValid() const { return Filename != nullptr; }
322
323 /// Return the presumed filename of this location.
324 ///
325 /// This can be affected by \#line etc.
326 const char *getFilename() const {
327 assert(isValid());
328 return Filename;
329 }
330
331 FileID getFileID() const {
332 assert(isValid());
333 return ID;
334 }
335
336 /// Return the presumed line number of this location.
337 ///
338 /// This can be affected by \#line etc.
339 unsigned getLine() const {
340 assert(isValid());
341 return Line;
342 }
343
344 /// Return the presumed column number of this location.
345 ///
346 /// This cannot be affected by \#line, but is packaged here for convenience.
347 unsigned getColumn() const {
348 assert(isValid());
349 return Col;
350 }
351
352 /// Return the presumed include location of this location.
353 ///
354 /// This can be affected by GNU linemarker directives.
355 SourceLocation getIncludeLoc() const {
356 assert(isValid());
357 return IncludeLoc;
358 }
359};
360
361class FileEntry;
362
363/// A SourceLocation and its associated SourceManager.
364///
365/// This is useful for argument passing to functions that expect both objects.
366class FullSourceLoc : public SourceLocation {
367 const SourceManager *SrcMgr = nullptr;
368
369public:
370 /// Creates a FullSourceLoc where isValid() returns \c false.
371 FullSourceLoc() = default;
372
373 explicit FullSourceLoc(SourceLocation Loc, const SourceManager &SM)
374 : SourceLocation(Loc), SrcMgr(&SM) {}
375
376 bool hasManager() const {
377 bool hasSrcMgr = SrcMgr != nullptr;
378 assert(hasSrcMgr == isValid() && "FullSourceLoc has location but no manager");
379 return hasSrcMgr;
380 }
381
382 /// \pre This FullSourceLoc has an associated SourceManager.
383 const SourceManager &getManager() const {
384 assert(SrcMgr && "SourceManager is NULL.");
385 return *SrcMgr;
386 }
387
388 FileID getFileID() const;
389
390 FullSourceLoc getExpansionLoc() const;
391 FullSourceLoc getSpellingLoc() const;
392 FullSourceLoc getFileLoc() const;
393 PresumedLoc getPresumedLoc(bool UseLineDirectives = true) const;
394 bool isMacroArgExpansion(FullSourceLoc *StartLoc = nullptr) const;
395 FullSourceLoc getImmediateMacroCallerLoc() const;
396 std::pair<FullSourceLoc, StringRef> getModuleImportLoc() const;
397 unsigned getFileOffset() const;
398
399 unsigned getExpansionLineNumber(bool *Invalid = nullptr) const;
400 unsigned getExpansionColumnNumber(bool *Invalid = nullptr) const;
401
402 unsigned getSpellingLineNumber(bool *Invalid = nullptr) const;
403 unsigned getSpellingColumnNumber(bool *Invalid = nullptr) const;
404
405 const char *getCharacterData(bool *Invalid = nullptr) const;
406
407 unsigned getLineNumber(bool *Invalid = nullptr) const;
408 unsigned getColumnNumber(bool *Invalid = nullptr) const;
409
410 const FileEntry *getFileEntry() const;
411
412 /// Return a StringRef to the source buffer data for the
413 /// specified FileID.
414 StringRef getBufferData(bool *Invalid = nullptr) const;
415
416 /// Decompose the specified location into a raw FileID + Offset pair.
417 ///
418 /// The first element is the FileID, the second is the offset from the
419 /// start of the buffer of the location.
420 std::pair<FileID, unsigned> getDecomposedLoc() const;
421
422 bool isInSystemHeader() const;
423
424 /// Determines the order of 2 source locations in the translation unit.
425 ///
426 /// \returns true if this source location comes before 'Loc', false otherwise.
427 bool isBeforeInTranslationUnitThan(SourceLocation Loc) const;
428
429 /// Determines the order of 2 source locations in the translation unit.
430 ///
431 /// \returns true if this source location comes before 'Loc', false otherwise.
432 bool isBeforeInTranslationUnitThan(FullSourceLoc Loc) const {
433 assert(Loc.isValid());
434 assert(SrcMgr == Loc.SrcMgr && "Loc comes from another SourceManager!");
435 return isBeforeInTranslationUnitThan((SourceLocation)Loc);
436 }
437
438 /// Comparison function class, useful for sorting FullSourceLocs.
439 struct BeforeThanCompare {
440 bool operator()(const FullSourceLoc& lhs, const FullSourceLoc& rhs) const {
441 return lhs.isBeforeInTranslationUnitThan(rhs);
442 }
443 };
444
445 /// Prints information about this FullSourceLoc to stderr.
446 ///
447 /// This is useful for debugging.
448 void dump() const;
449
450 friend bool
451 operator==(const FullSourceLoc &LHS, const FullSourceLoc &RHS) {
452 return LHS.getRawEncoding() == RHS.getRawEncoding() &&
453 LHS.SrcMgr == RHS.SrcMgr;
454 }
455
456 friend bool
457 operator!=(const FullSourceLoc &LHS, const FullSourceLoc &RHS) {
458 return !(LHS == RHS);
459 }
460};
461
462} // namespace clang
463
464namespace llvm {
465
466 /// Define DenseMapInfo so that FileID's can be used as keys in DenseMap and
467 /// DenseSets.
468 template <>
469 struct DenseMapInfo<clang::FileID> {
470 static clang::FileID getEmptyKey() {
471 return {};
472 }
473
474 static clang::FileID getTombstoneKey() {
475 return clang::FileID::getSentinel();
476 }
477
478 static unsigned getHashValue(clang::FileID S) {
479 return S.getHashValue();
480 }
481
482 static bool isEqual(clang::FileID LHS, clang::FileID RHS) {
483 return LHS == RHS;
484 }
485 };
486
487 /// Define DenseMapInfo so that SourceLocation's can be used as keys in
488 /// DenseMap and DenseSet. This trait class is eqivalent to
489 /// DenseMapInfo<unsigned> which uses SourceLocation::ID is used as a key.
490 template <> struct DenseMapInfo<clang::SourceLocation> {
491 static clang::SourceLocation getEmptyKey() {
492 return clang::SourceLocation::getFromRawEncoding(~0U);
493 }
494
495 static clang::SourceLocation getTombstoneKey() {
496 return clang::SourceLocation::getFromRawEncoding(~0U - 1);
497 }
498
499 static unsigned getHashValue(clang::SourceLocation Loc) {
500 return Loc.getHashValue();
501 }
502
503 static bool isEqual(clang::SourceLocation LHS, clang::SourceLocation RHS) {
504 return LHS == RHS;
505 }
506 };
507
508 // Allow calling FoldingSetNodeID::Add with SourceLocation object as parameter
509 template <> struct FoldingSetTrait<clang::SourceLocation> {
510 static void Profile(const clang::SourceLocation &X, FoldingSetNodeID &ID);
511 };
512
513 // Teach SmallPtrSet how to handle SourceLocation.
514 template<>
515 struct PointerLikeTypeTraits<clang::SourceLocation> {
516 static constexpr int NumLowBitsAvailable = 0;
517
518 static void *getAsVoidPointer(clang::SourceLocation L) {
519 return L.getPtrEncoding();
520 }
521
522 static clang::SourceLocation getFromVoidPointer(void *P) {
523 return clang::SourceLocation::getFromRawEncoding((unsigned)(uintptr_t)P);
524 }
525 };
526
527} // namespace llvm
528
529#endif // LLVM_CLANG_BASIC_SOURCELOCATION_H
530