SourceManager.h revision b1c86492f9a9bef01a4567408c22f961bbd604fe
1//===--- SourceManager.h - Track and cache source files ---------*- C++ -*-===// 2// 3// The LLVM Compiler Infrastructure 4// 5// This file is distributed under the University of Illinois Open Source 6// License. See LICENSE.TXT for details. 7// 8//===----------------------------------------------------------------------===// 9// 10// This file defines the SourceManager interface. 11// 12//===----------------------------------------------------------------------===// 13 14#ifndef LLVM_CLANG_SOURCEMANAGER_H 15#define LLVM_CLANG_SOURCEMANAGER_H 16 17#include "clang/Basic/SourceLocation.h" 18#include "llvm/Support/Allocator.h" 19#include "llvm/Support/DataTypes.h" 20#include "llvm/ADT/PointerIntPair.h" 21#include "llvm/ADT/PointerUnion.h" 22#include "llvm/ADT/DenseMap.h" 23#include <vector> 24#include <cassert> 25 26namespace llvm { 27class MemoryBuffer; 28class StringRef; 29} 30 31namespace clang { 32 33class Diagnostic; 34class SourceManager; 35class FileManager; 36class FileEntry; 37class LineTableInfo; 38 39/// SrcMgr - Public enums and private classes that are part of the 40/// SourceManager implementation. 41/// 42namespace SrcMgr { 43 /// CharacteristicKind - This is used to represent whether a file or directory 44 /// holds normal user code, system code, or system code which is implicitly 45 /// 'extern "C"' in C++ mode. Entire directories can be tagged with this 46 /// (this is maintained by DirectoryLookup and friends) as can specific 47 /// FileIDInfos when a #pragma system_header is seen or various other cases. 48 /// 49 enum CharacteristicKind { 50 C_User, C_System, C_ExternCSystem 51 }; 52 53 /// ContentCache - One instance of this struct is kept for every file 54 /// loaded or used. This object owns the MemoryBuffer object. 55 class ContentCache { 56 enum CCFlags { 57 /// \brief Whether the buffer is invalid. 58 InvalidFlag = 0x01, 59 /// \brief Whether the buffer should not be freed on destruction. 60 DoNotFreeFlag = 0x02 61 }; 62 63 /// Buffer - The actual buffer containing the characters from the input 64 /// file. This is owned by the ContentCache object. 65 /// The bits indicate indicates whether the buffer is invalid. 66 mutable llvm::PointerIntPair<const llvm::MemoryBuffer *, 2> Buffer; 67 68 public: 69 /// Reference to the file entry representing this ContentCache. 70 /// This reference does not own the FileEntry object. 71 /// It is possible for this to be NULL if 72 /// the ContentCache encapsulates an imaginary text buffer. 73 const FileEntry *OrigEntry; 74 75 /// \brief References the file which the contents were actually loaded from. 76 /// Can be different from 'Entry' if we overridden the contents of one file 77 /// with the contents of another file. 78 const FileEntry *ContentsEntry; 79 80 /// SourceLineCache - A bump pointer allocated array of offsets for each 81 /// source line. This is lazily computed. This is owned by the 82 /// SourceManager BumpPointerAllocator object. 83 unsigned *SourceLineCache; 84 85 /// NumLines - The number of lines in this ContentCache. This is only valid 86 /// if SourceLineCache is non-null. 87 unsigned NumLines; 88 89 /// getBuffer - Returns the memory buffer for the associated content. 90 /// 91 /// \param Diag Object through which diagnostics will be emitted it the 92 /// buffer cannot be retrieved. 93 /// 94 /// \param Loc If specified, is the location that invalid file diagnostics 95 /// will be emitted at. 96 /// 97 /// \param Invalid If non-NULL, will be set \c true if an error occurred. 98 const llvm::MemoryBuffer *getBuffer(Diagnostic &Diag, 99 const SourceManager &SM, 100 SourceLocation Loc = SourceLocation(), 101 bool *Invalid = 0) const; 102 103 /// getSize - Returns the size of the content encapsulated by this 104 /// ContentCache. This can be the size of the source file or the size of an 105 /// arbitrary scratch buffer. If the ContentCache encapsulates a source 106 /// file this size is retrieved from the file's FileEntry. 107 unsigned getSize() const; 108 109 /// getSizeBytesMapped - Returns the number of bytes actually mapped for 110 /// this ContentCache. This can be 0 if the MemBuffer was not actually 111 /// instantiated. 112 unsigned getSizeBytesMapped() const; 113 114 void setBuffer(const llvm::MemoryBuffer *B) { 115 assert(!Buffer.getPointer() && "MemoryBuffer already set."); 116 Buffer.setPointer(B); 117 Buffer.setInt(false); 118 } 119 120 /// \brief Get the underlying buffer, returning NULL if the buffer is not 121 /// yet available. 122 const llvm::MemoryBuffer *getRawBuffer() const { 123 return Buffer.getPointer(); 124 } 125 126 /// \brief Replace the existing buffer (which will be deleted) 127 /// with the given buffer. 128 void replaceBuffer(const llvm::MemoryBuffer *B, bool DoNotFree = false); 129 130 /// \brief Determine whether the buffer itself is invalid. 131 bool isBufferInvalid() const { 132 return Buffer.getInt() & InvalidFlag; 133 } 134 135 /// \brief Determine whether the buffer should be freed. 136 bool shouldFreeBuffer() const { 137 return (Buffer.getInt() & DoNotFreeFlag) == 0; 138 } 139 140 ContentCache(const FileEntry *Ent = 0) 141 : Buffer(0, false), OrigEntry(Ent), ContentsEntry(Ent), 142 SourceLineCache(0), NumLines(0) {} 143 144 ContentCache(const FileEntry *Ent, const FileEntry *contentEnt) 145 : Buffer(0, false), OrigEntry(Ent), ContentsEntry(contentEnt), 146 SourceLineCache(0), NumLines(0) {} 147 148 ~ContentCache(); 149 150 /// The copy ctor does not allow copies where source object has either 151 /// a non-NULL Buffer or SourceLineCache. Ownership of allocated memory 152 /// is not transfered, so this is a logical error. 153 ContentCache(const ContentCache &RHS) 154 : Buffer(0, false), SourceLineCache(0) 155 { 156 OrigEntry = RHS.OrigEntry; 157 ContentsEntry = RHS.ContentsEntry; 158 159 assert (RHS.Buffer.getPointer() == 0 && RHS.SourceLineCache == 0 160 && "Passed ContentCache object cannot own a buffer."); 161 162 NumLines = RHS.NumLines; 163 } 164 165 private: 166 // Disable assignments. 167 ContentCache &operator=(const ContentCache& RHS); 168 }; 169 170 /// FileInfo - Information about a FileID, basically just the logical file 171 /// that it represents and include stack information. 172 /// 173 /// Each FileInfo has include stack information, indicating where it came 174 /// from. This information encodes the #include chain that a token was 175 /// instantiated from. The main include file has an invalid IncludeLoc. 176 /// 177 /// FileInfos contain a "ContentCache *", with the contents of the file. 178 /// 179 class FileInfo { 180 /// IncludeLoc - The location of the #include that brought in this file. 181 /// This is an invalid SLOC for the main file (top of the #include chain). 182 unsigned IncludeLoc; // Really a SourceLocation 183 184 /// Data - This contains the ContentCache* and the bits indicating the 185 /// characteristic of the file and whether it has #line info, all bitmangled 186 /// together. 187 uintptr_t Data; 188 public: 189 /// get - Return a FileInfo object. 190 static FileInfo get(SourceLocation IL, const ContentCache *Con, 191 CharacteristicKind FileCharacter) { 192 FileInfo X; 193 X.IncludeLoc = IL.getRawEncoding(); 194 X.Data = (uintptr_t)Con; 195 assert((X.Data & 7) == 0 &&"ContentCache pointer insufficiently aligned"); 196 assert((unsigned)FileCharacter < 4 && "invalid file character"); 197 X.Data |= (unsigned)FileCharacter; 198 return X; 199 } 200 201 SourceLocation getIncludeLoc() const { 202 return SourceLocation::getFromRawEncoding(IncludeLoc); 203 } 204 const ContentCache* getContentCache() const { 205 return reinterpret_cast<const ContentCache*>(Data & ~7UL); 206 } 207 208 /// getCharacteristic - Return whether this is a system header or not. 209 CharacteristicKind getFileCharacteristic() const { 210 return (CharacteristicKind)(Data & 3); 211 } 212 213 /// hasLineDirectives - Return true if this FileID has #line directives in 214 /// it. 215 bool hasLineDirectives() const { return (Data & 4) != 0; } 216 217 /// setHasLineDirectives - Set the flag that indicates that this FileID has 218 /// line table entries associated with it. 219 void setHasLineDirectives() { 220 Data |= 4; 221 } 222 }; 223 224 /// InstantiationInfo - Each InstantiationInfo encodes the Instantiation 225 /// location - where the token was ultimately instantiated, and the 226 /// SpellingLoc - where the actual character data for the token came from. 227 class InstantiationInfo { 228 // Really these are all SourceLocations. 229 230 /// SpellingLoc - Where the spelling for the token can be found. 231 unsigned SpellingLoc; 232 233 /// InstantiationLocStart/InstantiationLocEnd - In a macro expansion, these 234 /// indicate the start and end of the instantiation. In object-like macros, 235 /// these will be the same. In a function-like macro instantiation, the 236 /// start will be the identifier and the end will be the ')'. 237 unsigned InstantiationLocStart, InstantiationLocEnd; 238 public: 239 SourceLocation getSpellingLoc() const { 240 return SourceLocation::getFromRawEncoding(SpellingLoc); 241 } 242 SourceLocation getInstantiationLocStart() const { 243 return SourceLocation::getFromRawEncoding(InstantiationLocStart); 244 } 245 SourceLocation getInstantiationLocEnd() const { 246 return SourceLocation::getFromRawEncoding(InstantiationLocEnd); 247 } 248 249 std::pair<SourceLocation,SourceLocation> getInstantiationLocRange() const { 250 return std::make_pair(getInstantiationLocStart(), 251 getInstantiationLocEnd()); 252 } 253 254 /// get - Return a InstantiationInfo for an expansion. IL specifies 255 /// the instantiation location (where the macro is expanded), and SL 256 /// specifies the spelling location (where the characters from the token 257 /// come from). IL and PL can both refer to normal File SLocs or 258 /// instantiation locations. 259 static InstantiationInfo get(SourceLocation ILStart, SourceLocation ILEnd, 260 SourceLocation SL) { 261 InstantiationInfo X; 262 X.SpellingLoc = SL.getRawEncoding(); 263 X.InstantiationLocStart = ILStart.getRawEncoding(); 264 X.InstantiationLocEnd = ILEnd.getRawEncoding(); 265 return X; 266 } 267 }; 268 269 /// SLocEntry - This is a discriminated union of FileInfo and 270 /// InstantiationInfo. SourceManager keeps an array of these objects, and 271 /// they are uniquely identified by the FileID datatype. 272 class SLocEntry { 273 unsigned Offset; // low bit is set for instantiation info. 274 union { 275 FileInfo File; 276 InstantiationInfo Instantiation; 277 }; 278 public: 279 unsigned getOffset() const { return Offset >> 1; } 280 281 bool isInstantiation() const { return Offset & 1; } 282 bool isFile() const { return !isInstantiation(); } 283 284 const FileInfo &getFile() const { 285 assert(isFile() && "Not a file SLocEntry!"); 286 return File; 287 } 288 289 const InstantiationInfo &getInstantiation() const { 290 assert(isInstantiation() && "Not an instantiation SLocEntry!"); 291 return Instantiation; 292 } 293 294 static SLocEntry get(unsigned Offset, const FileInfo &FI) { 295 SLocEntry E; 296 E.Offset = Offset << 1; 297 E.File = FI; 298 return E; 299 } 300 301 static SLocEntry get(unsigned Offset, const InstantiationInfo &II) { 302 SLocEntry E; 303 E.Offset = (Offset << 1) | 1; 304 E.Instantiation = II; 305 return E; 306 } 307 }; 308} // end SrcMgr namespace. 309 310/// \brief External source of source location entries. 311class ExternalSLocEntrySource { 312public: 313 virtual ~ExternalSLocEntrySource(); 314 315 /// \brief Read the source location entry with index ID. 316 virtual void ReadSLocEntry(unsigned ID) = 0; 317}; 318 319 320/// IsBeforeInTranslationUnitCache - This class holds the cache used by 321/// isBeforeInTranslationUnit. The cache structure is complex enough to be 322/// worth breaking out of SourceManager. 323class IsBeforeInTranslationUnitCache { 324 /// L/R QueryFID - These are the FID's of the cached query. If these match up 325 /// with a subsequent query, the result can be reused. 326 FileID LQueryFID, RQueryFID; 327 328 /// CommonFID - This is the file found in common between the two #include 329 /// traces. It is the nearest common ancestor of the #include tree. 330 FileID CommonFID; 331 332 /// L/R CommonOffset - This is the offset of the previous query in CommonFID. 333 /// Usually, this represents the location of the #include for QueryFID, but if 334 /// LQueryFID is a parent of RQueryFID (or vise versa) then these can be a 335 /// random token in the parent. 336 unsigned LCommonOffset, RCommonOffset; 337public: 338 339 /// isCacheValid - Return true if the currently cached values match up with 340 /// the specified LHS/RHS query. If not, we can't use the cache. 341 bool isCacheValid(FileID LHS, FileID RHS) const { 342 return LQueryFID == LHS && RQueryFID == RHS; 343 } 344 345 /// getCachedResult - If the cache is valid, compute the result given the 346 /// specified offsets in the LHS/RHS FID's. 347 bool getCachedResult(unsigned LOffset, unsigned ROffset) const { 348 // If one of the query files is the common file, use the offset. Otherwise, 349 // use the #include loc in the common file. 350 if (LQueryFID != CommonFID) LOffset = LCommonOffset; 351 if (RQueryFID != CommonFID) ROffset = RCommonOffset; 352 return LOffset < ROffset; 353 } 354 355 // Set up a new query. 356 void setQueryFIDs(FileID LHS, FileID RHS) { 357 LQueryFID = LHS; 358 RQueryFID = RHS; 359 } 360 361 void setCommonLoc(FileID commonFID, unsigned lCommonOffset, 362 unsigned rCommonOffset) { 363 CommonFID = commonFID; 364 LCommonOffset = lCommonOffset; 365 RCommonOffset = rCommonOffset; 366 } 367 368}; 369 370/// SourceManager - This file handles loading and caching of source files into 371/// memory. This object owns the MemoryBuffer objects for all of the loaded 372/// files and assigns unique FileID's for each unique #include chain. 373/// 374/// The SourceManager can be queried for information about SourceLocation 375/// objects, turning them into either spelling or instantiation locations. 376/// Spelling locations represent where the bytes corresponding to a token came 377/// from and instantiation locations represent where the location is in the 378/// user's view. In the case of a macro expansion, for example, the spelling 379/// location indicates where the expanded token came from and the instantiation 380/// location specifies where it was expanded. 381class SourceManager { 382 /// \brief Diagnostic object. 383 Diagnostic &Diag; 384 385 FileManager &FileMgr; 386 387 mutable llvm::BumpPtrAllocator ContentCacheAlloc; 388 389 /// FileInfos - Memoized information about all of the files tracked by this 390 /// SourceManager. This set allows us to merge ContentCache entries based 391 /// on their FileEntry*. All ContentCache objects will thus have unique, 392 /// non-null, FileEntry pointers. 393 llvm::DenseMap<const FileEntry*, SrcMgr::ContentCache*> FileInfos; 394 395 /// \brief Files that have been overriden with the contents from another file. 396 llvm::DenseMap<const FileEntry *, const FileEntry *> OverriddenFiles; 397 398 /// MemBufferInfos - Information about various memory buffers that we have 399 /// read in. All FileEntry* within the stored ContentCache objects are NULL, 400 /// as they do not refer to a file. 401 std::vector<SrcMgr::ContentCache*> MemBufferInfos; 402 403 /// SLocEntryTable - This is an array of SLocEntry's that we have created. 404 /// FileID is an index into this vector. This array is sorted by the offset. 405 std::vector<SrcMgr::SLocEntry> SLocEntryTable; 406 /// NextOffset - This is the next available offset that a new SLocEntry can 407 /// start at. It is SLocEntryTable.back().getOffset()+size of back() entry. 408 unsigned NextOffset; 409 410 /// \brief If source location entries are being lazily loaded from 411 /// an external source, this vector indicates whether the Ith source 412 /// location entry has already been loaded from the external storage. 413 std::vector<bool> SLocEntryLoaded; 414 415 /// \brief An external source for source location entries. 416 ExternalSLocEntrySource *ExternalSLocEntries; 417 418 /// LastFileIDLookup - This is a one-entry cache to speed up getFileID. 419 /// LastFileIDLookup records the last FileID looked up or created, because it 420 /// is very common to look up many tokens from the same file. 421 mutable FileID LastFileIDLookup; 422 423 /// LineTable - This holds information for #line directives. It is referenced 424 /// by indices from SLocEntryTable. 425 LineTableInfo *LineTable; 426 427 /// LastLineNo - These ivars serve as a cache used in the getLineNumber 428 /// method which is used to speedup getLineNumber calls to nearby locations. 429 mutable FileID LastLineNoFileIDQuery; 430 mutable SrcMgr::ContentCache *LastLineNoContentCache; 431 mutable unsigned LastLineNoFilePos; 432 mutable unsigned LastLineNoResult; 433 434 /// MainFileID - The file ID for the main source file of the translation unit. 435 FileID MainFileID; 436 437 // Statistics for -print-stats. 438 mutable unsigned NumLinearScans, NumBinaryProbes; 439 440 // Cache results for the isBeforeInTranslationUnit method. 441 mutable IsBeforeInTranslationUnitCache IsBeforeInTUCache; 442 443 // SourceManager doesn't support copy construction. 444 explicit SourceManager(const SourceManager&); 445 void operator=(const SourceManager&); 446public: 447 SourceManager(Diagnostic &Diag, FileManager &FileMgr); 448 ~SourceManager(); 449 450 void clearIDTables(); 451 452 Diagnostic &getDiagnostics() const { return Diag; } 453 454 FileManager &getFileManager() const { return FileMgr; } 455 456 //===--------------------------------------------------------------------===// 457 // MainFileID creation and querying methods. 458 //===--------------------------------------------------------------------===// 459 460 /// getMainFileID - Returns the FileID of the main source file. 461 FileID getMainFileID() const { return MainFileID; } 462 463 /// createMainFileID - Create the FileID for the main source file. 464 FileID createMainFileID(const FileEntry *SourceFile) { 465 assert(MainFileID.isInvalid() && "MainFileID already set!"); 466 MainFileID = createFileID(SourceFile, SourceLocation(), SrcMgr::C_User); 467 return MainFileID; 468 } 469 470 /// \brief Set the file ID for the precompiled preamble, which is also the 471 /// main file. 472 void SetPreambleFileID(FileID Preamble) { 473 assert(MainFileID.isInvalid() && "MainFileID already set!"); 474 MainFileID = Preamble; 475 } 476 477 //===--------------------------------------------------------------------===// 478 // Methods to create new FileID's and instantiations. 479 //===--------------------------------------------------------------------===// 480 481 /// createFileID - Create a new FileID that represents the specified file 482 /// being #included from the specified IncludePosition. This returns 0 on 483 /// error and translates NULL into standard input. 484 /// PreallocateID should be non-zero to specify which pre-allocated, 485 /// lazily computed source location is being filled in by this operation. 486 FileID createFileID(const FileEntry *SourceFile, SourceLocation IncludePos, 487 SrcMgr::CharacteristicKind FileCharacter, 488 unsigned PreallocatedID = 0, 489 unsigned Offset = 0) { 490 const SrcMgr::ContentCache *IR = getOrCreateContentCache(SourceFile); 491 assert(IR && "getOrCreateContentCache() cannot return NULL"); 492 return createFileID(IR, IncludePos, FileCharacter, PreallocatedID, Offset); 493 } 494 495 /// createFileIDForMemBuffer - Create a new FileID that represents the 496 /// specified memory buffer. This does no caching of the buffer and takes 497 /// ownership of the MemoryBuffer, so only pass a MemoryBuffer to this once. 498 FileID createFileIDForMemBuffer(const llvm::MemoryBuffer *Buffer, 499 unsigned PreallocatedID = 0, 500 unsigned Offset = 0) { 501 return createFileID(createMemBufferContentCache(Buffer), SourceLocation(), 502 SrcMgr::C_User, PreallocatedID, Offset); 503 } 504 505 /// createMainFileIDForMembuffer - Create the FileID for a memory buffer 506 /// that will represent the FileID for the main source. One example 507 /// of when this would be used is when the main source is read from STDIN. 508 FileID createMainFileIDForMemBuffer(const llvm::MemoryBuffer *Buffer) { 509 assert(MainFileID.isInvalid() && "MainFileID already set!"); 510 MainFileID = createFileIDForMemBuffer(Buffer); 511 return MainFileID; 512 } 513 514 /// createInstantiationLoc - Return a new SourceLocation that encodes the fact 515 /// that a token at Loc should actually be referenced from InstantiationLoc. 516 /// TokLength is the length of the token being instantiated. 517 SourceLocation createInstantiationLoc(SourceLocation Loc, 518 SourceLocation InstantiationLocStart, 519 SourceLocation InstantiationLocEnd, 520 unsigned TokLength, 521 unsigned PreallocatedID = 0, 522 unsigned Offset = 0); 523 524 /// \brief Retrieve the memory buffer associated with the given file. 525 /// 526 /// \param Invalid If non-NULL, will be set \c true if an error 527 /// occurs while retrieving the memory buffer. 528 const llvm::MemoryBuffer *getMemoryBufferForFile(const FileEntry *File, 529 bool *Invalid = 0); 530 531 /// \brief Override the contents of the given source file by providing an 532 /// already-allocated buffer. 533 /// 534 /// \param SourceFile the source file whose contents will be overriden. 535 /// 536 /// \param Buffer the memory buffer whose contents will be used as the 537 /// data in the given source file. 538 /// 539 /// \param DoNotFree If true, then the buffer will not be freed when the 540 /// source manager is destroyed. 541 void overrideFileContents(const FileEntry *SourceFile, 542 const llvm::MemoryBuffer *Buffer, 543 bool DoNotFree = false); 544 545 /// \brief Override the the given source file with another one. 546 /// 547 /// \param SourceFile the source file which will be overriden. 548 /// 549 /// \param NewFile the file whose contents will be used as the 550 /// data instead of the contents of the given source file. 551 void overrideFileContents(const FileEntry *SourceFile, 552 const FileEntry *NewFile); 553 554 //===--------------------------------------------------------------------===// 555 // FileID manipulation methods. 556 //===--------------------------------------------------------------------===// 557 558 /// getBuffer - Return the buffer for the specified FileID. If there is an 559 /// error opening this buffer the first time, this manufactures a temporary 560 /// buffer and returns a non-empty error string. 561 const llvm::MemoryBuffer *getBuffer(FileID FID, SourceLocation Loc, 562 bool *Invalid = 0) const { 563 return getSLocEntry(FID).getFile().getContentCache() 564 ->getBuffer(Diag, *this, Loc, Invalid); 565 } 566 567 const llvm::MemoryBuffer *getBuffer(FileID FID, bool *Invalid = 0) const { 568 return getSLocEntry(FID).getFile().getContentCache() 569 ->getBuffer(Diag, *this, SourceLocation(), Invalid); 570 } 571 572 /// getFileEntryForID - Returns the FileEntry record for the provided FileID. 573 const FileEntry *getFileEntryForID(FileID FID) const { 574 return getSLocEntry(FID).getFile().getContentCache()->OrigEntry; 575 } 576 577 /// getBufferData - Return a StringRef to the source buffer data for the 578 /// specified FileID. 579 /// 580 /// \param FID The file ID whose contents will be returned. 581 /// \param Invalid If non-NULL, will be set true if an error occurred. 582 llvm::StringRef getBufferData(FileID FID, bool *Invalid = 0) const; 583 584 585 //===--------------------------------------------------------------------===// 586 // SourceLocation manipulation methods. 587 //===--------------------------------------------------------------------===// 588 589 /// getFileID - Return the FileID for a SourceLocation. This is a very 590 /// hot method that is used for all SourceManager queries that start with a 591 /// SourceLocation object. It is responsible for finding the entry in 592 /// SLocEntryTable which contains the specified location. 593 /// 594 FileID getFileID(SourceLocation SpellingLoc) const { 595 unsigned SLocOffset = SpellingLoc.getOffset(); 596 597 // If our one-entry cache covers this offset, just return it. 598 if (isOffsetInFileID(LastFileIDLookup, SLocOffset)) 599 return LastFileIDLookup; 600 601 return getFileIDSlow(SLocOffset); 602 } 603 604 /// getLocForStartOfFile - Return the source location corresponding to the 605 /// first byte of the specified file. 606 SourceLocation getLocForStartOfFile(FileID FID) const { 607 assert(FID.ID < SLocEntryTable.size() && "FileID out of range"); 608 assert(getSLocEntry(FID).isFile() && "FileID is not a file"); 609 unsigned FileOffset = getSLocEntry(FID).getOffset(); 610 return SourceLocation::getFileLoc(FileOffset); 611 } 612 613 /// getInstantiationLoc - Given a SourceLocation object, return the 614 /// instantiation location referenced by the ID. 615 SourceLocation getInstantiationLoc(SourceLocation Loc) const { 616 // Handle the non-mapped case inline, defer to out of line code to handle 617 // instantiations. 618 if (Loc.isFileID()) return Loc; 619 return getInstantiationLocSlowCase(Loc); 620 } 621 622 /// getImmediateInstantiationRange - Loc is required to be an instantiation 623 /// location. Return the start/end of the instantiation information. 624 std::pair<SourceLocation,SourceLocation> 625 getImmediateInstantiationRange(SourceLocation Loc) const; 626 627 /// getInstantiationRange - Given a SourceLocation object, return the 628 /// range of tokens covered by the instantiation in the ultimate file. 629 std::pair<SourceLocation,SourceLocation> 630 getInstantiationRange(SourceLocation Loc) const; 631 632 633 /// getSpellingLoc - Given a SourceLocation object, return the spelling 634 /// location referenced by the ID. This is the place where the characters 635 /// that make up the lexed token can be found. 636 SourceLocation getSpellingLoc(SourceLocation Loc) const { 637 // Handle the non-mapped case inline, defer to out of line code to handle 638 // instantiations. 639 if (Loc.isFileID()) return Loc; 640 return getSpellingLocSlowCase(Loc); 641 } 642 643 /// getImmediateSpellingLoc - Given a SourceLocation object, return the 644 /// spelling location referenced by the ID. This is the first level down 645 /// towards the place where the characters that make up the lexed token can be 646 /// found. This should not generally be used by clients. 647 SourceLocation getImmediateSpellingLoc(SourceLocation Loc) const; 648 649 /// getDecomposedLoc - Decompose the specified location into a raw FileID + 650 /// Offset pair. The first element is the FileID, the second is the 651 /// offset from the start of the buffer of the location. 652 std::pair<FileID, unsigned> getDecomposedLoc(SourceLocation Loc) const { 653 FileID FID = getFileID(Loc); 654 return std::make_pair(FID, Loc.getOffset()-getSLocEntry(FID).getOffset()); 655 } 656 657 /// getDecomposedInstantiationLoc - Decompose the specified location into a 658 /// raw FileID + Offset pair. If the location is an instantiation record, 659 /// walk through it until we find the final location instantiated. 660 std::pair<FileID, unsigned> 661 getDecomposedInstantiationLoc(SourceLocation Loc) const { 662 FileID FID = getFileID(Loc); 663 const SrcMgr::SLocEntry *E = &getSLocEntry(FID); 664 665 unsigned Offset = Loc.getOffset()-E->getOffset(); 666 if (Loc.isFileID()) 667 return std::make_pair(FID, Offset); 668 669 return getDecomposedInstantiationLocSlowCase(E, Offset); 670 } 671 672 /// getDecomposedSpellingLoc - Decompose the specified location into a raw 673 /// FileID + Offset pair. If the location is an instantiation record, walk 674 /// through it until we find its spelling record. 675 std::pair<FileID, unsigned> 676 getDecomposedSpellingLoc(SourceLocation Loc) const { 677 FileID FID = getFileID(Loc); 678 const SrcMgr::SLocEntry *E = &getSLocEntry(FID); 679 680 unsigned Offset = Loc.getOffset()-E->getOffset(); 681 if (Loc.isFileID()) 682 return std::make_pair(FID, Offset); 683 return getDecomposedSpellingLocSlowCase(E, Offset); 684 } 685 686 /// getFileOffset - This method returns the offset from the start 687 /// of the file that the specified SourceLocation represents. This is not very 688 /// meaningful for a macro ID. 689 unsigned getFileOffset(SourceLocation SpellingLoc) const { 690 return getDecomposedLoc(SpellingLoc).second; 691 } 692 693 694 //===--------------------------------------------------------------------===// 695 // Queries about the code at a SourceLocation. 696 //===--------------------------------------------------------------------===// 697 698 /// getCharacterData - Return a pointer to the start of the specified location 699 /// in the appropriate spelling MemoryBuffer. 700 /// 701 /// \param Invalid If non-NULL, will be set \c true if an error occurs. 702 const char *getCharacterData(SourceLocation SL, bool *Invalid = 0) const; 703 704 /// getColumnNumber - Return the column # for the specified file position. 705 /// This is significantly cheaper to compute than the line number. This 706 /// returns zero if the column number isn't known. This may only be called on 707 /// a file sloc, so you must choose a spelling or instantiation location 708 /// before calling this method. 709 unsigned getColumnNumber(FileID FID, unsigned FilePos, 710 bool *Invalid = 0) const; 711 unsigned getSpellingColumnNumber(SourceLocation Loc, bool *Invalid = 0) const; 712 unsigned getInstantiationColumnNumber(SourceLocation Loc, 713 bool *Invalid = 0) const; 714 unsigned getPresumedColumnNumber(SourceLocation Loc, bool *Invalid = 0) const; 715 716 717 /// getLineNumber - Given a SourceLocation, return the spelling line number 718 /// for the position indicated. This requires building and caching a table of 719 /// line offsets for the MemoryBuffer, so this is not cheap: use only when 720 /// about to emit a diagnostic. 721 unsigned getLineNumber(FileID FID, unsigned FilePos, bool *Invalid = 0) const; 722 unsigned getSpellingLineNumber(SourceLocation Loc, bool *Invalid = 0) const; 723 unsigned getInstantiationLineNumber(SourceLocation Loc, 724 bool *Invalid = 0) const; 725 unsigned getPresumedLineNumber(SourceLocation Loc, bool *Invalid = 0) const; 726 727 /// Return the filename or buffer identifier of the buffer the location is in. 728 /// Note that this name does not respect #line directives. Use getPresumedLoc 729 /// for normal clients. 730 const char *getBufferName(SourceLocation Loc, bool *Invalid = 0) const; 731 732 /// getFileCharacteristic - return the file characteristic of the specified 733 /// source location, indicating whether this is a normal file, a system 734 /// header, or an "implicit extern C" system header. 735 /// 736 /// This state can be modified with flags on GNU linemarker directives like: 737 /// # 4 "foo.h" 3 738 /// which changes all source locations in the current file after that to be 739 /// considered to be from a system header. 740 SrcMgr::CharacteristicKind getFileCharacteristic(SourceLocation Loc) const; 741 742 /// getPresumedLoc - This method returns the "presumed" location of a 743 /// SourceLocation specifies. A "presumed location" can be modified by #line 744 /// or GNU line marker directives. This provides a view on the data that a 745 /// user should see in diagnostics, for example. 746 /// 747 /// Note that a presumed location is always given as the instantiation point 748 /// of an instantiation location, not at the spelling location. 749 /// 750 /// \returns The presumed location of the specified SourceLocation. If the 751 /// presumed location cannot be calculate (e.g., because \p Loc is invalid 752 /// or the file containing \p Loc has changed on disk), returns an invalid 753 /// presumed location. 754 PresumedLoc getPresumedLoc(SourceLocation Loc) const; 755 756 /// isFromSameFile - Returns true if both SourceLocations correspond to 757 /// the same file. 758 bool isFromSameFile(SourceLocation Loc1, SourceLocation Loc2) const { 759 return getFileID(Loc1) == getFileID(Loc2); 760 } 761 762 /// isFromMainFile - Returns true if the file of provided SourceLocation is 763 /// the main file. 764 bool isFromMainFile(SourceLocation Loc) const { 765 return getFileID(Loc) == getMainFileID(); 766 } 767 768 /// isInSystemHeader - Returns if a SourceLocation is in a system header. 769 bool isInSystemHeader(SourceLocation Loc) const { 770 return getFileCharacteristic(Loc) != SrcMgr::C_User; 771 } 772 773 /// isInExternCSystemHeader - Returns if a SourceLocation is in an "extern C" 774 /// system header. 775 bool isInExternCSystemHeader(SourceLocation Loc) const { 776 return getFileCharacteristic(Loc) == SrcMgr::C_ExternCSystem; 777 } 778 779 //===--------------------------------------------------------------------===// 780 // Line Table Manipulation Routines 781 //===--------------------------------------------------------------------===// 782 783 /// getLineTableFilenameID - Return the uniqued ID for the specified filename. 784 /// 785 unsigned getLineTableFilenameID(const char *Ptr, unsigned Len); 786 787 /// AddLineNote - Add a line note to the line table for the FileID and offset 788 /// specified by Loc. If FilenameID is -1, it is considered to be 789 /// unspecified. 790 void AddLineNote(SourceLocation Loc, unsigned LineNo, int FilenameID); 791 void AddLineNote(SourceLocation Loc, unsigned LineNo, int FilenameID, 792 bool IsFileEntry, bool IsFileExit, 793 bool IsSystemHeader, bool IsExternCHeader); 794 795 /// \brief Determine if the source manager has a line table. 796 bool hasLineTable() const { return LineTable != 0; } 797 798 /// \brief Retrieve the stored line table. 799 LineTableInfo &getLineTable(); 800 801 //===--------------------------------------------------------------------===// 802 // Other miscellaneous methods. 803 //===--------------------------------------------------------------------===// 804 805 /// \brief Get the source location for the given file:line:col triplet. 806 /// 807 /// If the source file is included multiple times, the source location will 808 /// be based upon the first inclusion. 809 SourceLocation getLocation(const FileEntry *SourceFile, 810 unsigned Line, unsigned Col); 811 812 /// \brief Determines the order of 2 source locations in the translation unit. 813 /// 814 /// \returns true if LHS source location comes before RHS, false otherwise. 815 bool isBeforeInTranslationUnit(SourceLocation LHS, SourceLocation RHS) const; 816 817 // Iterators over FileInfos. 818 typedef llvm::DenseMap<const FileEntry*, SrcMgr::ContentCache*> 819 ::const_iterator fileinfo_iterator; 820 fileinfo_iterator fileinfo_begin() const { return FileInfos.begin(); } 821 fileinfo_iterator fileinfo_end() const { return FileInfos.end(); } 822 bool hasFileInfo(const FileEntry *File) const { 823 return FileInfos.find(File) != FileInfos.end(); 824 } 825 826 /// PrintStats - Print statistics to stderr. 827 /// 828 void PrintStats() const; 829 830 unsigned sloc_entry_size() const { return SLocEntryTable.size(); } 831 832 // FIXME: Exposing this is a little gross; what we want is a good way 833 // to iterate the entries that were not defined in an AST file (or 834 // any other external source). 835 unsigned sloc_loaded_entry_size() const { return SLocEntryLoaded.size(); } 836 837 const SrcMgr::SLocEntry &getSLocEntry(unsigned ID) const { 838 assert(ID < SLocEntryTable.size() && "Invalid id"); 839 if (ExternalSLocEntries && 840 ID < SLocEntryLoaded.size() && 841 !SLocEntryLoaded[ID]) 842 ExternalSLocEntries->ReadSLocEntry(ID); 843 return SLocEntryTable[ID]; 844 } 845 846 const SrcMgr::SLocEntry &getSLocEntry(FileID FID) const { 847 return getSLocEntry(FID.ID); 848 } 849 850 unsigned getNextOffset() const { return NextOffset; } 851 852 /// \brief Preallocate some number of source location entries, which 853 /// will be loaded as needed from the given external source. 854 void PreallocateSLocEntries(ExternalSLocEntrySource *Source, 855 unsigned NumSLocEntries, 856 unsigned NextOffset); 857 858 /// \brief Clear out any preallocated source location entries that 859 /// haven't already been loaded. 860 void ClearPreallocatedSLocEntries(); 861 862private: 863 /// isOffsetInFileID - Return true if the specified FileID contains the 864 /// specified SourceLocation offset. This is a very hot method. 865 inline bool isOffsetInFileID(FileID FID, unsigned SLocOffset) const { 866 const SrcMgr::SLocEntry &Entry = getSLocEntry(FID); 867 // If the entry is after the offset, it can't contain it. 868 if (SLocOffset < Entry.getOffset()) return false; 869 870 // If this is the last entry than it does. Otherwise, the entry after it 871 // has to not include it. 872 if (FID.ID+1 == SLocEntryTable.size()) return true; 873 874 return SLocOffset < getSLocEntry(FileID::get(FID.ID+1)).getOffset(); 875 } 876 877 /// createFileID - Create a new fileID for the specified ContentCache and 878 /// include position. This works regardless of whether the ContentCache 879 /// corresponds to a file or some other input source. 880 FileID createFileID(const SrcMgr::ContentCache* File, 881 SourceLocation IncludePos, 882 SrcMgr::CharacteristicKind DirCharacter, 883 unsigned PreallocatedID = 0, 884 unsigned Offset = 0); 885 886 const SrcMgr::ContentCache * 887 getOrCreateContentCache(const FileEntry *SourceFile); 888 889 /// createMemBufferContentCache - Create a new ContentCache for the specified 890 /// memory buffer. 891 const SrcMgr::ContentCache* 892 createMemBufferContentCache(const llvm::MemoryBuffer *Buf); 893 894 FileID getFileIDSlow(unsigned SLocOffset) const; 895 896 SourceLocation getInstantiationLocSlowCase(SourceLocation Loc) const; 897 SourceLocation getSpellingLocSlowCase(SourceLocation Loc) const; 898 899 std::pair<FileID, unsigned> 900 getDecomposedInstantiationLocSlowCase(const SrcMgr::SLocEntry *E, 901 unsigned Offset) const; 902 std::pair<FileID, unsigned> 903 getDecomposedSpellingLocSlowCase(const SrcMgr::SLocEntry *E, 904 unsigned Offset) const; 905}; 906 907 908} // end namespace clang 909 910#endif 911