Implement -working-directory.
[clang.git] / include / clang / Serialization / ASTReader.h
blob420197c260693d70c9875e716335b5089059de66
1 //===--- ASTReader.h - AST File Reader --------------------------*- 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 ASTReader class, which reads AST files.
12 //===----------------------------------------------------------------------===//
14 #ifndef LLVM_CLANG_FRONTEND_AST_READER_H
15 #define LLVM_CLANG_FRONTEND_AST_READER_H
17 #include "clang/Serialization/ASTBitCodes.h"
18 #include "clang/Sema/ExternalSemaSource.h"
19 #include "clang/AST/DeclarationName.h"
20 #include "clang/AST/DeclObjC.h"
21 #include "clang/AST/TemplateBase.h"
22 #include "clang/Lex/ExternalPreprocessorSource.h"
23 #include "clang/Lex/PreprocessingRecord.h"
24 #include "clang/Basic/Diagnostic.h"
25 #include "clang/Basic/IdentifierTable.h"
26 #include "clang/Basic/SourceManager.h"
27 #include "llvm/ADT/APFloat.h"
28 #include "llvm/ADT/APInt.h"
29 #include "llvm/ADT/APSInt.h"
30 #include "llvm/ADT/OwningPtr.h"
31 #include "llvm/ADT/SmallVector.h"
32 #include "llvm/ADT/StringRef.h"
33 #include "llvm/Bitcode/BitstreamReader.h"
34 #include "llvm/System/DataTypes.h"
35 #include <deque>
36 #include <map>
37 #include <string>
38 #include <utility>
39 #include <vector>
41 namespace llvm {
42 class MemoryBuffer;
45 namespace clang {
47 class AddrLabelExpr;
48 class ASTConsumer;
49 class ASTContext;
50 class ASTIdentifierIterator;
51 class Attr;
52 class Decl;
53 class DeclContext;
54 class NestedNameSpecifier;
55 class CXXBaseSpecifier;
56 class CXXBaseOrMemberInitializer;
57 class GotoStmt;
58 class LabelStmt;
59 class MacroDefinition;
60 class NamedDecl;
61 class Preprocessor;
62 class Sema;
63 class SwitchCase;
64 class ASTDeserializationListener;
65 class ASTReader;
66 class ASTDeclReader;
67 class ASTStmtReader;
68 class ASTIdentifierLookupTrait;
69 class TypeLocReader;
70 class FileSystemOptions;
71 struct HeaderFileInfo;
73 struct PCHPredefinesBlock {
74 /// \brief The file ID for this predefines buffer in a PCH file.
75 FileID BufferID;
77 /// \brief This predefines buffer in a PCH file.
78 llvm::StringRef Data;
80 typedef llvm::SmallVector<PCHPredefinesBlock, 2> PCHPredefinesBlocks;
82 /// \brief Abstract interface for callback invocations by the ASTReader.
83 ///
84 /// While reading an AST file, the ASTReader will call the methods of the
85 /// listener to pass on specific information. Some of the listener methods can
86 /// return true to indicate to the ASTReader that the information (and
87 /// consequently the AST file) is invalid.
88 class ASTReaderListener {
89 public:
90 virtual ~ASTReaderListener();
92 /// \brief Receives the language options.
93 ///
94 /// \returns true to indicate the options are invalid or false otherwise.
95 virtual bool ReadLanguageOptions(const LangOptions &LangOpts) {
96 return false;
99 /// \brief Receives the target triple.
101 /// \returns true to indicate the target triple is invalid or false otherwise.
102 virtual bool ReadTargetTriple(llvm::StringRef Triple) {
103 return false;
106 /// \brief Receives the contents of the predefines buffer.
108 /// \param Buffers Information about the predefines buffers.
110 /// \param OriginalFileName The original file name for the AST file, which
111 /// will appear as an entry in the predefines buffer.
113 /// \param SuggestedPredefines If necessary, additional definitions are added
114 /// here.
116 /// \returns true to indicate the predefines are invalid or false otherwise.
117 virtual bool ReadPredefinesBuffer(const PCHPredefinesBlocks &Buffers,
118 llvm::StringRef OriginalFileName,
119 std::string &SuggestedPredefines) {
120 return false;
123 /// \brief Receives a HeaderFileInfo entry.
124 virtual void ReadHeaderFileInfo(const HeaderFileInfo &HFI, unsigned ID) {}
126 /// \brief Receives __COUNTER__ value.
127 virtual void ReadCounter(unsigned Value) {}
130 /// \brief ASTReaderListener implementation to validate the information of
131 /// the PCH file against an initialized Preprocessor.
132 class PCHValidator : public ASTReaderListener {
133 Preprocessor &PP;
134 ASTReader &Reader;
136 unsigned NumHeaderInfos;
138 public:
139 PCHValidator(Preprocessor &PP, ASTReader &Reader)
140 : PP(PP), Reader(Reader), NumHeaderInfos(0) {}
142 virtual bool ReadLanguageOptions(const LangOptions &LangOpts);
143 virtual bool ReadTargetTriple(llvm::StringRef Triple);
144 virtual bool ReadPredefinesBuffer(const PCHPredefinesBlocks &Buffers,
145 llvm::StringRef OriginalFileName,
146 std::string &SuggestedPredefines);
147 virtual void ReadHeaderFileInfo(const HeaderFileInfo &HFI, unsigned ID);
148 virtual void ReadCounter(unsigned Value);
150 private:
151 void Error(const char *Msg);
154 /// \brief Reads an AST files chain containing the contents of a translation
155 /// unit.
157 /// The ASTReader class reads bitstreams (produced by the ASTWriter
158 /// class) containing the serialized representation of a given
159 /// abstract syntax tree and its supporting data structures. An
160 /// instance of the ASTReader can be attached to an ASTContext object,
161 /// which will provide access to the contents of the AST files.
163 /// The AST reader provides lazy de-serialization of declarations, as
164 /// required when traversing the AST. Only those AST nodes that are
165 /// actually required will be de-serialized.
166 class ASTReader
167 : public ExternalPreprocessorSource,
168 public ExternalPreprocessingRecordSource,
169 public ExternalSemaSource,
170 public IdentifierInfoLookup,
171 public ExternalIdentifierLookup,
172 public ExternalSLocEntrySource {
173 public:
174 enum ASTReadResult { Success, Failure, IgnorePCH };
175 /// \brief Types of AST files.
176 enum ASTFileType {
177 Module, ///< File is a module proper.
178 PCH, ///< File is a PCH file treated as such.
179 Preamble, ///< File is a PCH file treated as the preamble.
180 MainFile ///< File is a PCH file treated as the actual main file.
182 friend class PCHValidator;
183 friend class ASTDeclReader;
184 friend class ASTStmtReader;
185 friend class ASTIdentifierIterator;
186 friend class ASTIdentifierLookupTrait;
187 friend class TypeLocReader;
188 private:
189 /// \brief The receiver of some callbacks invoked by ASTReader.
190 llvm::OwningPtr<ASTReaderListener> Listener;
192 /// \brief The receiver of deserialization events.
193 ASTDeserializationListener *DeserializationListener;
195 SourceManager &SourceMgr;
196 FileManager &FileMgr;
197 const FileSystemOptions &FileSystemOpts;
198 Diagnostic &Diags;
200 /// \brief The semantic analysis object that will be processing the
201 /// AST files and the translation unit that uses it.
202 Sema *SemaObj;
204 /// \brief The preprocessor that will be loading the source file.
205 Preprocessor *PP;
207 /// \brief The AST context into which we'll read the AST files.
208 ASTContext *Context;
210 /// \brief The AST consumer.
211 ASTConsumer *Consumer;
213 /// \brief Information that is needed for every module.
214 struct PerFileData {
215 PerFileData(ASTFileType Ty);
216 ~PerFileData();
218 // === General information ===
220 /// \brief The type of this AST file.
221 ASTFileType Type;
223 /// \brief The file name of the AST file.
224 std::string FileName;
226 /// \brief The memory buffer that stores the data associated with
227 /// this AST file.
228 llvm::OwningPtr<llvm::MemoryBuffer> Buffer;
230 /// \brief The size of this file, in bits.
231 uint64_t SizeInBits;
233 /// \brief The bitstream reader from which we'll read the AST file.
234 llvm::BitstreamReader StreamFile;
236 /// \brief The main bitstream cursor for the main block.
237 llvm::BitstreamCursor Stream;
239 // === Source Locations ===
241 /// \brief Cursor used to read source location entries.
242 llvm::BitstreamCursor SLocEntryCursor;
244 /// \brief The number of source location entries in this AST file.
245 unsigned LocalNumSLocEntries;
247 /// \brief Offsets for all of the source location entries in the
248 /// AST file.
249 const uint32_t *SLocOffsets;
251 /// \brief The entire size of this module's source location offset range.
252 unsigned LocalSLocSize;
254 // === Identifiers ===
256 /// \brief The number of identifiers in this AST file.
257 unsigned LocalNumIdentifiers;
259 /// \brief Offsets into the identifier table data.
261 /// This array is indexed by the identifier ID (-1), and provides
262 /// the offset into IdentifierTableData where the string data is
263 /// stored.
264 const uint32_t *IdentifierOffsets;
266 /// \brief Actual data for the on-disk hash table.
268 /// This pointer points into a memory buffer, where the on-disk hash
269 /// table for identifiers actually lives.
270 const char *IdentifierTableData;
272 /// \brief A pointer to an on-disk hash table of opaque type
273 /// IdentifierHashTable.
274 void *IdentifierLookupTable;
276 // === Macros ===
278 /// \brief The cursor to the start of the preprocessor block, which stores
279 /// all of the macro definitions.
280 llvm::BitstreamCursor MacroCursor;
282 /// \brief The offset of the start of the set of defined macros.
283 uint64_t MacroStartOffset;
285 /// \brief The number of macro definitions in this file.
286 unsigned LocalNumMacroDefinitions;
288 /// \brief Offsets of all of the macro definitions in the preprocessing
289 /// record in the AST file.
290 const uint32_t *MacroDefinitionOffsets;
292 // === Selectors ===
294 /// \brief The number of selectors new to this file.
296 /// This is the number of entries in SelectorOffsets.
297 unsigned LocalNumSelectors;
299 /// \brief Offsets into the selector lookup table's data array
300 /// where each selector resides.
301 const uint32_t *SelectorOffsets;
303 /// \brief A pointer to the character data that comprises the selector table
305 /// The SelectorOffsets table refers into this memory.
306 const unsigned char *SelectorLookupTableData;
308 /// \brief A pointer to an on-disk hash table of opaque type
309 /// ASTSelectorLookupTable.
311 /// This hash table provides the IDs of all selectors, and the associated
312 /// instance and factory methods.
313 void *SelectorLookupTable;
315 /// \brief Method selectors used in a @selector expression. Used for
316 /// implementation of -Wselector.
317 llvm::SmallVector<uint64_t, 64> ReferencedSelectorsData;
319 // === Declarations ===
321 /// DeclsCursor - This is a cursor to the start of the DECLS_BLOCK block. It
322 /// has read all the abbreviations at the start of the block and is ready to
323 /// jump around with these in context.
324 llvm::BitstreamCursor DeclsCursor;
326 /// \brief The number of declarations in this AST file.
327 unsigned LocalNumDecls;
329 /// \brief Offset of each declaration within the bitstream, indexed
330 /// by the declaration ID (-1).
331 const uint32_t *DeclOffsets;
333 /// \brief A snapshot of the pending instantiations in the chain.
335 /// This record tracks the instantiations that Sema has to perform at the
336 /// end of the TU. It consists of a pair of values for every pending
337 /// instantiation where the first value is the ID of the decl and the second
338 /// is the instantiation location.
339 llvm::SmallVector<uint64_t, 64> PendingInstantiations;
341 /// \brief The number of C++ base specifier sets in this AST file.
342 unsigned LocalNumCXXBaseSpecifiers;
344 /// \brief Offset of each C++ base specifier set within the bitstream,
345 /// indexed by the C++ base specifier set ID (-1).
346 const uint32_t *CXXBaseSpecifiersOffsets;
348 // === Types ===
350 /// \brief The number of types in this AST file.
351 unsigned LocalNumTypes;
353 /// \brief Offset of each type within the bitstream, indexed by the
354 /// type ID, or the representation of a Type*.
355 const uint32_t *TypeOffsets;
357 // === Miscellaneous ===
359 /// \brief The AST stat cache installed for this file, if any.
361 /// The dynamic type of this stat cache is always ASTStatCache
362 void *StatCache;
364 /// \brief The number of preallocated preprocessing entities in the
365 /// preprocessing record.
366 unsigned NumPreallocatedPreprocessingEntities;
368 /// \brief The next module in source order.
369 PerFileData *NextInSource;
371 /// \brief All the modules that loaded this one. Can contain NULL for
372 /// directly loaded modules.
373 llvm::SmallVector<PerFileData *, 1> Loaders;
376 /// \brief All loaded modules, indexed by name.
377 llvm::StringMap<PerFileData*> Modules;
379 /// \brief The first module in source order.
380 PerFileData *FirstInSource;
382 /// \brief The chain of AST files. The first entry is the one named by the
383 /// user, the last one is the one that doesn't depend on anything further.
384 /// That is, the entry I was created with -include-pch I+1.
385 llvm::SmallVector<PerFileData*, 2> Chain;
387 /// \brief SLocEntries that we're going to preload.
388 llvm::SmallVector<uint64_t, 64> PreloadSLocEntries;
390 /// \brief Types that have already been loaded from the chain.
392 /// When the pointer at index I is non-NULL, the type with
393 /// ID = (I + 1) << FastQual::Width has already been loaded
394 std::vector<QualType> TypesLoaded;
396 /// \brief Map that provides the ID numbers of each type within the
397 /// output stream, plus those deserialized from a chained PCH.
399 /// The ID numbers of types are consecutive (in order of discovery)
400 /// and start at 1. 0 is reserved for NULL. When types are actually
401 /// stored in the stream, the ID number is shifted by 2 bits to
402 /// allow for the const/volatile qualifiers.
404 /// Keys in the map never have const/volatile qualifiers.
405 serialization::TypeIdxMap TypeIdxs;
407 /// \brief Declarations that have already been loaded from the chain.
409 /// When the pointer at index I is non-NULL, the declaration with ID
410 /// = I + 1 has already been loaded.
411 std::vector<Decl *> DeclsLoaded;
413 typedef std::pair<PerFileData *, uint64_t> FileOffset;
414 typedef llvm::SmallVector<FileOffset, 2> FileOffsetsTy;
415 typedef llvm::DenseMap<serialization::DeclID, FileOffsetsTy>
416 DeclUpdateOffsetsMap;
417 /// \brief Declarations that have modifications residing in a later file
418 /// in the chain.
419 DeclUpdateOffsetsMap DeclUpdateOffsets;
421 typedef llvm::DenseMap<serialization::DeclID,
422 std::pair<PerFileData *, uint64_t> >
423 DeclReplacementMap;
424 /// \brief Declarations that have been replaced in a later file in the chain.
425 DeclReplacementMap ReplacedDecls;
427 /// \brief Information about the contents of a DeclContext.
428 struct DeclContextInfo {
429 void *NameLookupTableData; // a ASTDeclContextNameLookupTable.
430 const serialization::KindDeclIDPair *LexicalDecls;
431 unsigned NumLexicalDecls;
433 // In a full chain, there could be multiple updates to every decl context,
434 // so this is a vector. However, typically a chain is only two elements long,
435 // with only one file containing updates, so there will be only one update
436 // per decl context.
437 typedef llvm::SmallVector<DeclContextInfo, 1> DeclContextInfos;
438 typedef llvm::DenseMap<const DeclContext *, DeclContextInfos>
439 DeclContextOffsetsMap;
440 // Updates for visible decls can occur for other contexts than just the
441 // TU, and when we read those update records, the actual context will not
442 // be available yet (unless it's the TU), so have this pending map using the
443 // ID as a key. It will be realized when the context is actually loaded.
444 typedef llvm::SmallVector<void *, 1> DeclContextVisibleUpdates;
445 typedef llvm::DenseMap<serialization::DeclID, DeclContextVisibleUpdates>
446 DeclContextVisibleUpdatesPending;
448 /// \brief Offsets of the lexical and visible declarations for each
449 /// DeclContext.
450 DeclContextOffsetsMap DeclContextOffsets;
452 /// \brief Updates to the visible declarations of declaration contexts that
453 /// haven't been loaded yet.
454 DeclContextVisibleUpdatesPending PendingVisibleUpdates;
456 typedef llvm::SmallVector<CXXRecordDecl *, 4> ForwardRefs;
457 typedef llvm::DenseMap<const CXXRecordDecl *, ForwardRefs>
458 PendingForwardRefsMap;
459 /// \brief Forward references that have a definition but the definition decl
460 /// is still initializing. When the definition gets read it will update
461 /// the DefinitionData pointer of all pending references.
462 PendingForwardRefsMap PendingForwardRefs;
464 typedef llvm::DenseMap<serialization::DeclID, serialization::DeclID>
465 FirstLatestDeclIDMap;
466 /// \brief Map of first declarations from a chained PCH that point to the
467 /// most recent declarations in another AST file.
468 FirstLatestDeclIDMap FirstLatestDeclIDs;
470 /// \brief Read the records that describe the contents of declcontexts.
471 bool ReadDeclContextStorage(llvm::BitstreamCursor &Cursor,
472 const std::pair<uint64_t, uint64_t> &Offsets,
473 DeclContextInfo &Info);
475 /// \brief A vector containing identifiers that have already been
476 /// loaded.
478 /// If the pointer at index I is non-NULL, then it refers to the
479 /// IdentifierInfo for the identifier with ID=I+1 that has already
480 /// been loaded.
481 std::vector<IdentifierInfo *> IdentifiersLoaded;
483 /// \brief A vector containing selectors that have already been loaded.
485 /// This vector is indexed by the Selector ID (-1). NULL selector
486 /// entries indicate that the particular selector ID has not yet
487 /// been loaded.
488 llvm::SmallVector<Selector, 16> SelectorsLoaded;
490 /// \brief The macro definitions we have already loaded.
491 llvm::SmallVector<MacroDefinition *, 16> MacroDefinitionsLoaded;
493 /// \brief Mapping from identifiers that represent macros whose definitions
494 /// have not yet been deserialized to the global offset where the macro
495 /// record resides.
496 llvm::DenseMap<IdentifierInfo *, uint64_t> UnreadMacroRecordOffsets;
498 /// \name CodeGen-relevant special data
499 /// \brief Fields containing data that is relevant to CodeGen.
500 //@{
502 /// \brief The IDs of all declarations that fulfill the criteria of
503 /// "interesting" decls.
505 /// This contains the data loaded from all EXTERNAL_DEFINITIONS blocks in the
506 /// chain. The referenced declarations are deserialized and passed to the
507 /// consumer eagerly.
508 llvm::SmallVector<uint64_t, 16> ExternalDefinitions;
510 /// \brief The IDs of all tentative definitions stored in the the chain.
512 /// Sema keeps track of all tentative definitions in a TU because it has to
513 /// complete them and pass them on to CodeGen. Thus, tentative definitions in
514 /// the PCH chain must be eagerly deserialized.
515 llvm::SmallVector<uint64_t, 16> TentativeDefinitions;
517 /// \brief The IDs of all CXXRecordDecls stored in the chain whose VTables are
518 /// used.
520 /// CodeGen has to emit VTables for these records, so they have to be eagerly
521 /// deserialized.
522 llvm::SmallVector<uint64_t, 64> VTableUses;
524 //@}
526 /// \name Diagnostic-relevant special data
527 /// \brief Fields containing data that is used for generating diagnostics
528 //@{
530 /// \brief A snapshot of Sema's unused file-scoped variable tracking, for
531 /// generating warnings.
532 llvm::SmallVector<uint64_t, 16> UnusedFileScopedDecls;
534 /// \brief A snapshot of Sema's weak undeclared identifier tracking, for
535 /// generating warnings.
536 llvm::SmallVector<uint64_t, 64> WeakUndeclaredIdentifiers;
538 /// \brief The IDs of type aliases for ext_vectors that exist in the chain.
540 /// Used by Sema for finding sugared names for ext_vectors in diagnostics.
541 llvm::SmallVector<uint64_t, 4> ExtVectorDecls;
543 //@}
545 /// \name Sema-relevant special data
546 /// \brief Fields containing data that is used for semantic analysis
547 //@{
549 /// \brief The IDs of all locally scoped external decls in the chain.
551 /// Sema tracks these to validate that the types are consistent across all
552 /// local external declarations.
553 llvm::SmallVector<uint64_t, 16> LocallyScopedExternalDecls;
555 /// \brief The IDs of all dynamic class declarations in the chain.
557 /// Sema tracks these because it checks for the key functions being defined
558 /// at the end of the TU, in which case it directs CodeGen to emit the VTable.
559 llvm::SmallVector<uint64_t, 16> DynamicClasses;
561 /// \brief The IDs of the declarations Sema stores directly.
563 /// Sema tracks a few important decls, such as namespace std, directly.
564 llvm::SmallVector<uint64_t, 4> SemaDeclRefs;
566 /// \brief The IDs of the types ASTContext stores directly.
568 /// The AST context tracks a few important types, such as va_list, directly.
569 llvm::SmallVector<uint64_t, 16> SpecialTypes;
571 //@}
573 /// \brief The original file name that was used to build the primary AST file,
574 /// which may have been modified for relocatable-pch support.
575 std::string OriginalFileName;
577 /// \brief The actual original file name that was used to build the primary
578 /// AST file.
579 std::string ActualOriginalFileName;
581 /// \brief Whether this precompiled header is a relocatable PCH file.
582 bool RelocatablePCH;
584 /// \brief The system include root to be used when loading the
585 /// precompiled header.
586 const char *isysroot;
588 /// \brief Whether to disable the normal validation performed on precompiled
589 /// headers when they are loaded.
590 bool DisableValidation;
592 /// \brief Mapping from switch-case IDs in the chain to switch-case statements
594 /// Statements usually don't have IDs, but switch cases need them, so that the
595 /// switch statement can refer to them.
596 std::map<unsigned, SwitchCase *> SwitchCaseStmts;
598 /// \brief Mapping from label statement IDs in the chain to label statements.
600 /// Statements usually don't have IDs, but labeled statements need them, so
601 /// that goto statements and address-of-label expressions can refer to them.
602 std::map<unsigned, LabelStmt *> LabelStmts;
604 /// \brief Mapping from label IDs to the set of "goto" statements
605 /// that point to that label before the label itself has been
606 /// de-serialized.
607 std::multimap<unsigned, GotoStmt *> UnresolvedGotoStmts;
609 /// \brief Mapping from label IDs to the set of address label
610 /// expressions that point to that label before the label itself has
611 /// been de-serialized.
612 std::multimap<unsigned, AddrLabelExpr *> UnresolvedAddrLabelExprs;
614 /// \brief The number of stat() calls that hit/missed the stat
615 /// cache.
616 unsigned NumStatHits, NumStatMisses;
618 /// \brief The number of source location entries de-serialized from
619 /// the PCH file.
620 unsigned NumSLocEntriesRead;
622 /// \brief The number of source location entries in the chain.
623 unsigned TotalNumSLocEntries;
625 /// \brief The next offset for a SLocEntry after everything in this reader.
626 unsigned NextSLocOffset;
628 /// \brief The number of statements (and expressions) de-serialized
629 /// from the chain.
630 unsigned NumStatementsRead;
632 /// \brief The total number of statements (and expressions) stored
633 /// in the chain.
634 unsigned TotalNumStatements;
636 /// \brief The number of macros de-serialized from the chain.
637 unsigned NumMacrosRead;
639 /// \brief The total number of macros stored in the chain.
640 unsigned TotalNumMacros;
642 /// \brief The number of selectors that have been read.
643 unsigned NumSelectorsRead;
645 /// \brief The number of method pool entries that have been read.
646 unsigned NumMethodPoolEntriesRead;
648 /// \brief The number of times we have looked up a selector in the method
649 /// pool and not found anything interesting.
650 unsigned NumMethodPoolMisses;
652 /// \brief The total number of method pool entries in the selector table.
653 unsigned TotalNumMethodPoolEntries;
655 /// Number of lexical decl contexts read/total.
656 unsigned NumLexicalDeclContextsRead, TotalLexicalDeclContexts;
658 /// Number of visible decl contexts read/total.
659 unsigned NumVisibleDeclContextsRead, TotalVisibleDeclContexts;
661 /// \brief Number of Decl/types that are currently deserializing.
662 unsigned NumCurrentElementsDeserializing;
664 /// \brief An IdentifierInfo that has been loaded but whose top-level
665 /// declarations of the same name have not (yet) been loaded.
666 struct PendingIdentifierInfo {
667 IdentifierInfo *II;
668 llvm::SmallVector<uint32_t, 4> DeclIDs;
671 /// \brief The set of identifiers that were read while the AST reader was
672 /// (recursively) loading declarations.
674 /// The declarations on the identifier chain for these identifiers will be
675 /// loaded once the recursive loading has completed.
676 std::deque<PendingIdentifierInfo> PendingIdentifierInfos;
678 /// \brief Contains declarations and definitions that will be
679 /// "interesting" to the ASTConsumer, when we get that AST consumer.
681 /// "Interesting" declarations are those that have data that may
682 /// need to be emitted, such as inline function definitions or
683 /// Objective-C protocols.
684 std::deque<Decl *> InterestingDecls;
686 /// \brief When reading a Stmt tree, Stmt operands are placed in this stack.
687 llvm::SmallVector<Stmt *, 16> StmtStack;
689 /// \brief What kind of records we are reading.
690 enum ReadingKind {
691 Read_Decl, Read_Type, Read_Stmt
694 /// \brief What kind of records we are reading.
695 ReadingKind ReadingKind;
697 /// \brief RAII object to change the reading kind.
698 class ReadingKindTracker {
699 ASTReader &Reader;
700 enum ReadingKind PrevKind;
702 ReadingKindTracker(const ReadingKindTracker&); // do not implement
703 ReadingKindTracker &operator=(const ReadingKindTracker&);// do not implement
705 public:
706 ReadingKindTracker(enum ReadingKind newKind, ASTReader &reader)
707 : Reader(reader), PrevKind(Reader.ReadingKind) {
708 Reader.ReadingKind = newKind;
711 ~ReadingKindTracker() { Reader.ReadingKind = PrevKind; }
714 /// \brief All predefines buffers in the chain, to be treated as if
715 /// concatenated.
716 PCHPredefinesBlocks PCHPredefinesBuffers;
718 /// \brief Suggested contents of the predefines buffer, after this
719 /// PCH file has been processed.
721 /// In most cases, this string will be empty, because the predefines
722 /// buffer computed to build the PCH file will be identical to the
723 /// predefines buffer computed from the command line. However, when
724 /// there are differences that the PCH reader can work around, this
725 /// predefines buffer may contain additional definitions.
726 std::string SuggestedPredefines;
728 /// \brief Reads a statement from the specified cursor.
729 Stmt *ReadStmtFromStream(PerFileData &F);
731 void MaybeAddSystemRootToFilename(std::string &Filename);
733 ASTReadResult ReadASTCore(llvm::StringRef FileName, ASTFileType Type);
734 ASTReadResult ReadASTBlock(PerFileData &F);
735 bool CheckPredefinesBuffers();
736 bool ParseLineTable(PerFileData &F, llvm::SmallVectorImpl<uint64_t> &Record);
737 ASTReadResult ReadSourceManagerBlock(PerFileData &F);
738 ASTReadResult ReadSLocEntryRecord(unsigned ID);
739 PerFileData *SLocCursorForID(unsigned ID);
740 SourceLocation getImportLocation(PerFileData *F);
741 bool ParseLanguageOptions(const llvm::SmallVectorImpl<uint64_t> &Record);
743 struct RecordLocation {
744 RecordLocation(PerFileData *M, uint64_t O)
745 : F(M), Offset(O) {}
746 PerFileData *F;
747 uint64_t Offset;
750 QualType ReadTypeRecord(unsigned Index);
751 RecordLocation TypeCursorForIndex(unsigned Index);
752 void LoadedDecl(unsigned Index, Decl *D);
753 Decl *ReadDeclRecord(unsigned Index, serialization::DeclID ID);
754 RecordLocation DeclCursorForIndex(unsigned Index, serialization::DeclID ID);
756 void PassInterestingDeclsToConsumer();
758 /// \brief Produce an error diagnostic and return true.
760 /// This routine should only be used for fatal errors that have to
761 /// do with non-routine failures (e.g., corrupted AST file).
762 void Error(const char *Msg);
764 ASTReader(const ASTReader&); // do not implement
765 ASTReader &operator=(const ASTReader &); // do not implement
766 public:
767 typedef llvm::SmallVector<uint64_t, 64> RecordData;
769 /// \brief Load the AST file and validate its contents against the given
770 /// Preprocessor.
772 /// \param PP the preprocessor associated with the context in which this
773 /// precompiled header will be loaded.
775 /// \param Context the AST context that this precompiled header will be
776 /// loaded into.
778 /// \param isysroot If non-NULL, the system include path specified by the
779 /// user. This is only used with relocatable PCH files. If non-NULL,
780 /// a relocatable PCH file will use the default path "/".
782 /// \param DisableValidation If true, the AST reader will suppress most
783 /// of its regular consistency checking, allowing the use of precompiled
784 /// headers that cannot be determined to be compatible.
785 ASTReader(Preprocessor &PP, ASTContext *Context, const char *isysroot = 0,
786 bool DisableValidation = false);
788 /// \brief Load the AST file without using any pre-initialized Preprocessor.
790 /// The necessary information to initialize a Preprocessor later can be
791 /// obtained by setting a ASTReaderListener.
793 /// \param SourceMgr the source manager into which the AST file will be loaded
795 /// \param FileMgr the file manager into which the AST file will be loaded.
797 /// \param Diags the diagnostics system to use for reporting errors and
798 /// warnings relevant to loading the AST file.
800 /// \param isysroot If non-NULL, the system include path specified by the
801 /// user. This is only used with relocatable PCH files. If non-NULL,
802 /// a relocatable PCH file will use the default path "/".
804 /// \param DisableValidation If true, the AST reader will suppress most
805 /// of its regular consistency checking, allowing the use of precompiled
806 /// headers that cannot be determined to be compatible.
807 ASTReader(SourceManager &SourceMgr, FileManager &FileMgr,
808 const FileSystemOptions &FileSystemOpts,
809 Diagnostic &Diags, const char *isysroot = 0,
810 bool DisableValidation = false);
811 ~ASTReader();
813 /// \brief Load the precompiled header designated by the given file
814 /// name.
815 ASTReadResult ReadAST(const std::string &FileName, ASTFileType Type);
817 /// \brief Set the AST callbacks listener.
818 void setListener(ASTReaderListener *listener) {
819 Listener.reset(listener);
822 /// \brief Set the AST deserialization listener.
823 void setDeserializationListener(ASTDeserializationListener *Listener);
825 /// \brief Set the Preprocessor to use.
826 void setPreprocessor(Preprocessor &pp);
828 /// \brief Sets and initializes the given Context.
829 void InitializeContext(ASTContext &Context);
831 /// \brief Retrieve the name of the named (primary) AST file
832 const std::string &getFileName() const { return Chain[0]->FileName; }
834 /// \brief Retrieve the name of the original source file name
835 const std::string &getOriginalSourceFile() { return OriginalFileName; }
837 /// \brief Retrieve the name of the original source file name directly from
838 /// the AST file, without actually loading the AST file.
839 static std::string getOriginalSourceFile(const std::string &ASTFileName,
840 FileManager &FileMgr,
841 const FileSystemOptions &FSOpts,
842 Diagnostic &Diags);
844 /// \brief Returns the suggested contents of the predefines buffer,
845 /// which contains a (typically-empty) subset of the predefines
846 /// build prior to including the precompiled header.
847 const std::string &getSuggestedPredefines() { return SuggestedPredefines; }
849 /// \brief Read preprocessed entities into the
850 virtual void ReadPreprocessedEntities();
852 /// \brief Returns the number of source locations found in the chain.
853 unsigned getTotalNumSLocs() const {
854 return TotalNumSLocEntries;
857 /// \brief Returns the next SLocEntry offset after the chain.
858 unsigned getNextSLocOffset() const {
859 return NextSLocOffset;
862 /// \brief Returns the number of identifiers found in the chain.
863 unsigned getTotalNumIdentifiers() const {
864 return static_cast<unsigned>(IdentifiersLoaded.size());
867 /// \brief Returns the number of types found in the chain.
868 unsigned getTotalNumTypes() const {
869 return static_cast<unsigned>(TypesLoaded.size());
872 /// \brief Returns the number of declarations found in the chain.
873 unsigned getTotalNumDecls() const {
874 return static_cast<unsigned>(DeclsLoaded.size());
877 /// \brief Returns the number of selectors found in the chain.
878 unsigned getTotalNumSelectors() const {
879 return static_cast<unsigned>(SelectorsLoaded.size());
882 /// \brief Returns the number of macro definitions found in the chain.
883 unsigned getTotalNumMacroDefinitions() const {
884 return static_cast<unsigned>(MacroDefinitionsLoaded.size());
887 /// \brief Returns the number of C++ base specifiers found in the chain.
888 unsigned getTotalNumCXXBaseSpecifiers() const;
890 /// \brief Reads a TemplateArgumentLocInfo appropriate for the
891 /// given TemplateArgument kind.
892 TemplateArgumentLocInfo
893 GetTemplateArgumentLocInfo(PerFileData &F, TemplateArgument::ArgKind Kind,
894 const RecordData &Record, unsigned &Idx);
896 /// \brief Reads a TemplateArgumentLoc.
897 TemplateArgumentLoc
898 ReadTemplateArgumentLoc(PerFileData &F,
899 const RecordData &Record, unsigned &Idx);
901 /// \brief Reads a declarator info from the given record.
902 TypeSourceInfo *GetTypeSourceInfo(PerFileData &F,
903 const RecordData &Record, unsigned &Idx);
905 /// \brief Resolve and return the translation unit declaration.
906 TranslationUnitDecl *GetTranslationUnitDecl();
908 /// \brief Resolve a type ID into a type, potentially building a new
909 /// type.
910 QualType GetType(serialization::TypeID ID);
912 /// \brief Returns the type ID associated with the given type.
913 /// If the type didn't come from the AST file the ID that is returned is
914 /// marked as "doesn't exist in AST".
915 serialization::TypeID GetTypeID(QualType T) const;
917 /// \brief Returns the type index associated with the given type.
918 /// If the type didn't come from the AST file the index that is returned is
919 /// marked as "doesn't exist in AST".
920 serialization::TypeIdx GetTypeIdx(QualType T) const;
922 /// \brief Resolve a declaration ID into a declaration, potentially
923 /// building a new declaration.
924 Decl *GetDecl(serialization::DeclID ID);
925 virtual Decl *GetExternalDecl(uint32_t ID);
927 /// \brief Resolve a CXXBaseSpecifiers ID into an offset into the chain
928 /// of loaded AST files.
929 uint64_t GetCXXBaseSpecifiersOffset(serialization::CXXBaseSpecifiersID ID);
931 virtual CXXBaseSpecifier *GetExternalCXXBaseSpecifiers(uint64_t Offset);
933 /// \brief Resolve the offset of a statement into a statement.
935 /// This operation will read a new statement from the external
936 /// source each time it is called, and is meant to be used via a
937 /// LazyOffsetPtr (which is used by Decls for the body of functions, etc).
938 virtual Stmt *GetExternalDeclStmt(uint64_t Offset);
940 /// ReadBlockAbbrevs - Enter a subblock of the specified BlockID with the
941 /// specified cursor. Read the abbreviations that are at the top of the block
942 /// and then leave the cursor pointing into the block.
943 bool ReadBlockAbbrevs(llvm::BitstreamCursor &Cursor, unsigned BlockID);
945 /// \brief Finds all the visible declarations with a given name.
946 /// The current implementation of this method just loads the entire
947 /// lookup table as unmaterialized references.
948 virtual DeclContext::lookup_result
949 FindExternalVisibleDeclsByName(const DeclContext *DC,
950 DeclarationName Name);
952 virtual void MaterializeVisibleDecls(const DeclContext *DC);
954 /// \brief Read all of the declarations lexically stored in a
955 /// declaration context.
957 /// \param DC The declaration context whose declarations will be
958 /// read.
960 /// \param Decls Vector that will contain the declarations loaded
961 /// from the external source. The caller is responsible for merging
962 /// these declarations with any declarations already stored in the
963 /// declaration context.
965 /// \returns true if there was an error while reading the
966 /// declarations for this declaration context.
967 virtual bool FindExternalLexicalDecls(const DeclContext *DC,
968 bool (*isKindWeWant)(Decl::Kind),
969 llvm::SmallVectorImpl<Decl*> &Decls);
971 /// \brief Notify ASTReader that we started deserialization of
972 /// a decl or type so until FinishedDeserializing is called there may be
973 /// decls that are initializing. Must be paired with FinishedDeserializing.
974 virtual void StartedDeserializing() { ++NumCurrentElementsDeserializing; }
976 /// \brief Notify ASTReader that we finished the deserialization of
977 /// a decl or type. Must be paired with StartedDeserializing.
978 virtual void FinishedDeserializing();
980 /// \brief Function that will be invoked when we begin parsing a new
981 /// translation unit involving this external AST source.
983 /// This function will provide all of the external definitions to
984 /// the ASTConsumer.
985 virtual void StartTranslationUnit(ASTConsumer *Consumer);
987 /// \brief Print some statistics about AST usage.
988 virtual void PrintStats();
990 /// \brief Initialize the semantic source with the Sema instance
991 /// being used to perform semantic analysis on the abstract syntax
992 /// tree.
993 virtual void InitializeSema(Sema &S);
995 /// \brief Inform the semantic consumer that Sema is no longer available.
996 virtual void ForgetSema() { SemaObj = 0; }
998 /// \brief Retrieve the IdentifierInfo for the named identifier.
1000 /// This routine builds a new IdentifierInfo for the given identifier. If any
1001 /// declarations with this name are visible from translation unit scope, their
1002 /// declarations will be deserialized and introduced into the declaration
1003 /// chain of the identifier.
1004 virtual IdentifierInfo *get(const char *NameStart, const char *NameEnd);
1005 IdentifierInfo *get(llvm::StringRef Name) {
1006 return get(Name.begin(), Name.end());
1009 /// \brief Retrieve an iterator into the set of all identifiers
1010 /// in all loaded AST files.
1011 virtual IdentifierIterator *getIdentifiers() const;
1013 /// \brief Load the contents of the global method pool for a given
1014 /// selector.
1016 /// \returns a pair of Objective-C methods lists containing the
1017 /// instance and factory methods, respectively, with this selector.
1018 virtual std::pair<ObjCMethodList, ObjCMethodList>
1019 ReadMethodPool(Selector Sel);
1021 /// \brief Load a selector from disk, registering its ID if it exists.
1022 void LoadSelector(Selector Sel);
1024 void SetIdentifierInfo(unsigned ID, IdentifierInfo *II);
1025 void SetGloballyVisibleDecls(IdentifierInfo *II,
1026 const llvm::SmallVectorImpl<uint32_t> &DeclIDs,
1027 bool Nonrecursive = false);
1029 /// \brief Report a diagnostic.
1030 DiagnosticBuilder Diag(unsigned DiagID);
1032 /// \brief Report a diagnostic.
1033 DiagnosticBuilder Diag(SourceLocation Loc, unsigned DiagID);
1035 IdentifierInfo *DecodeIdentifierInfo(unsigned Idx);
1037 IdentifierInfo *GetIdentifierInfo(const RecordData &Record, unsigned &Idx) {
1038 return DecodeIdentifierInfo(Record[Idx++]);
1041 virtual IdentifierInfo *GetIdentifier(unsigned ID) {
1042 return DecodeIdentifierInfo(ID);
1045 /// \brief Read the source location entry with index ID.
1046 virtual void ReadSLocEntry(unsigned ID);
1048 Selector DecodeSelector(unsigned Idx);
1050 virtual Selector GetExternalSelector(uint32_t ID);
1051 uint32_t GetNumExternalSelectors();
1053 Selector GetSelector(const RecordData &Record, unsigned &Idx) {
1054 return DecodeSelector(Record[Idx++]);
1057 /// \brief Read a declaration name.
1058 DeclarationName ReadDeclarationName(const RecordData &Record, unsigned &Idx);
1059 void ReadDeclarationNameLoc(PerFileData &F,
1060 DeclarationNameLoc &DNLoc, DeclarationName Name,
1061 const RecordData &Record, unsigned &Idx);
1062 void ReadDeclarationNameInfo(PerFileData &F, DeclarationNameInfo &NameInfo,
1063 const RecordData &Record, unsigned &Idx);
1065 void ReadQualifierInfo(PerFileData &F, QualifierInfo &Info,
1066 const RecordData &Record, unsigned &Idx);
1068 NestedNameSpecifier *ReadNestedNameSpecifier(const RecordData &Record,
1069 unsigned &Idx);
1071 /// \brief Read a template name.
1072 TemplateName ReadTemplateName(const RecordData &Record, unsigned &Idx);
1074 /// \brief Read a template argument.
1075 TemplateArgument ReadTemplateArgument(PerFileData &F,
1076 const RecordData &Record,unsigned &Idx);
1078 /// \brief Read a template parameter list.
1079 TemplateParameterList *ReadTemplateParameterList(PerFileData &F,
1080 const RecordData &Record,
1081 unsigned &Idx);
1083 /// \brief Read a template argument array.
1084 void
1085 ReadTemplateArgumentList(llvm::SmallVector<TemplateArgument, 8> &TemplArgs,
1086 PerFileData &F, const RecordData &Record,
1087 unsigned &Idx);
1089 /// \brief Read a UnresolvedSet structure.
1090 void ReadUnresolvedSet(UnresolvedSetImpl &Set,
1091 const RecordData &Record, unsigned &Idx);
1093 /// \brief Read a C++ base specifier.
1094 CXXBaseSpecifier ReadCXXBaseSpecifier(PerFileData &F,
1095 const RecordData &Record,unsigned &Idx);
1097 /// \brief Read a CXXBaseOrMemberInitializer array.
1098 std::pair<CXXBaseOrMemberInitializer **, unsigned>
1099 ReadCXXBaseOrMemberInitializers(PerFileData &F,
1100 const RecordData &Record, unsigned &Idx);
1102 /// \brief Read a source location from raw form.
1103 SourceLocation ReadSourceLocation(PerFileData &Module, unsigned Raw) {
1104 (void)Module; // No remapping yet
1105 return SourceLocation::getFromRawEncoding(Raw);
1108 /// \brief Read a source location.
1109 SourceLocation ReadSourceLocation(PerFileData &Module,
1110 const RecordData &Record, unsigned& Idx) {
1111 return ReadSourceLocation(Module, Record[Idx++]);
1114 /// \brief Read a source range.
1115 SourceRange ReadSourceRange(PerFileData &F,
1116 const RecordData &Record, unsigned& Idx);
1118 /// \brief Read an integral value
1119 llvm::APInt ReadAPInt(const RecordData &Record, unsigned &Idx);
1121 /// \brief Read a signed integral value
1122 llvm::APSInt ReadAPSInt(const RecordData &Record, unsigned &Idx);
1124 /// \brief Read a floating-point value
1125 llvm::APFloat ReadAPFloat(const RecordData &Record, unsigned &Idx);
1127 // \brief Read a string
1128 std::string ReadString(const RecordData &Record, unsigned &Idx);
1130 CXXTemporary *ReadCXXTemporary(const RecordData &Record, unsigned &Idx);
1132 /// \brief Reads attributes from the current stream position.
1133 void ReadAttributes(PerFileData &F, AttrVec &Attrs,
1134 const RecordData &Record, unsigned &Idx);
1136 /// \brief Reads a statement.
1137 Stmt *ReadStmt(PerFileData &F);
1139 /// \brief Reads an expression.
1140 Expr *ReadExpr(PerFileData &F);
1142 /// \brief Reads a sub-statement operand during statement reading.
1143 Stmt *ReadSubStmt() {
1144 assert(ReadingKind == Read_Stmt &&
1145 "Should be called only during statement reading!");
1146 // Subexpressions are stored from last to first, so the next Stmt we need
1147 // is at the back of the stack.
1148 assert(!StmtStack.empty() && "Read too many sub statements!");
1149 return StmtStack.pop_back_val();
1152 /// \brief Reads a sub-expression operand during statement reading.
1153 Expr *ReadSubExpr();
1155 /// \brief Reads the macro record located at the given offset.
1156 void ReadMacroRecord(PerFileData &F, uint64_t Offset);
1158 /// \brief Note that the identifier is a macro whose record will be loaded
1159 /// from the given AST file at the given (file-local) offset.
1160 void SetIdentifierIsMacro(IdentifierInfo *II, PerFileData &F,
1161 uint64_t Offset);
1163 /// \brief Read the set of macros defined by this external macro source.
1164 virtual void ReadDefinedMacros();
1166 /// \brief Read the macro definition for this identifier.
1167 virtual void LoadMacroDefinition(IdentifierInfo *II);
1169 /// \brief Read the macro definition corresponding to this iterator
1170 /// into the unread macro record offsets table.
1171 void LoadMacroDefinition(
1172 llvm::DenseMap<IdentifierInfo *, uint64_t>::iterator Pos);
1174 /// \brief Retrieve the macro definition with the given ID.
1175 MacroDefinition *getMacroDefinition(serialization::MacroID ID);
1177 /// \brief Retrieve the AST context that this AST reader supplements.
1178 ASTContext *getContext() { return Context; }
1180 // \brief Contains declarations that were loaded before we have
1181 // access to a Sema object.
1182 llvm::SmallVector<NamedDecl *, 16> PreloadedDecls;
1184 /// \brief Retrieve the semantic analysis object used to analyze the
1185 /// translation unit in which the precompiled header is being
1186 /// imported.
1187 Sema *getSema() { return SemaObj; }
1189 /// \brief Retrieve the identifier table associated with the
1190 /// preprocessor.
1191 IdentifierTable &getIdentifierTable();
1193 /// \brief Record that the given ID maps to the given switch-case
1194 /// statement.
1195 void RecordSwitchCaseID(SwitchCase *SC, unsigned ID);
1197 /// \brief Retrieve the switch-case statement with the given ID.
1198 SwitchCase *getSwitchCaseWithID(unsigned ID);
1200 /// \brief Record that the given label statement has been
1201 /// deserialized and has the given ID.
1202 void RecordLabelStmt(LabelStmt *S, unsigned ID);
1204 void ClearSwitchCaseIDs();
1206 /// \brief Set the label of the given statement to the label
1207 /// identified by ID.
1209 /// Depending on the order in which the label and other statements
1210 /// referencing that label occur, this operation may complete
1211 /// immediately (updating the statement) or it may queue the
1212 /// statement to be back-patched later.
1213 void SetLabelOf(GotoStmt *S, unsigned ID);
1215 /// \brief Set the label of the given expression to the label
1216 /// identified by ID.
1218 /// Depending on the order in which the label and other statements
1219 /// referencing that label occur, this operation may complete
1220 /// immediately (updating the statement) or it may queue the
1221 /// statement to be back-patched later.
1222 void SetLabelOf(AddrLabelExpr *S, unsigned ID);
1225 /// \brief Helper class that saves the current stream position and
1226 /// then restores it when destroyed.
1227 struct SavedStreamPosition {
1228 explicit SavedStreamPosition(llvm::BitstreamCursor &Cursor)
1229 : Cursor(Cursor), Offset(Cursor.GetCurrentBitNo()) { }
1231 ~SavedStreamPosition() {
1232 Cursor.JumpToBit(Offset);
1235 private:
1236 llvm::BitstreamCursor &Cursor;
1237 uint64_t Offset;
1240 inline void PCHValidator::Error(const char *Msg) {
1241 Reader.Error(Msg);
1244 } // end namespace clang
1246 #endif