Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 1 | //===--- Preprocess.cpp - C Language Family Preprocessor Implementation ---===// |
| 2 | // |
| 3 | // The LLVM Compiler Infrastructure |
| 4 | // |
Chris Lattner | 0bc735f | 2007-12-29 19:59:25 +0000 | [diff] [blame] | 5 | // This file is distributed under the University of Illinois Open Source |
| 6 | // License. See LICENSE.TXT for details. |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 7 | // |
| 8 | //===----------------------------------------------------------------------===// |
| 9 | // |
| 10 | // This file implements the Preprocessor interface. |
| 11 | // |
| 12 | //===----------------------------------------------------------------------===// |
| 13 | // |
| 14 | // Options to support: |
| 15 | // -H - Print the name of each header file used. |
Chris Lattner | f73903a | 2009-02-06 06:45:26 +0000 | [diff] [blame] | 16 | // -d[DNI] - Dump various things. |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 17 | // -fworking-directory - #line's with preprocessor's working dir. |
| 18 | // -fpreprocessed |
| 19 | // -dependency-file,-M,-MM,-MF,-MG,-MP,-MT,-MQ,-MD,-MMD |
| 20 | // -W* |
| 21 | // -w |
| 22 | // |
| 23 | // Messages to emit: |
| 24 | // "Multiple include guards may be useful for:\n" |
| 25 | // |
| 26 | //===----------------------------------------------------------------------===// |
| 27 | |
| 28 | #include "clang/Lex/Preprocessor.h" |
Chris Lattner | 23f77e5 | 2009-12-15 01:51:03 +0000 | [diff] [blame] | 29 | #include "MacroArgs.h" |
Douglas Gregor | 88a3586 | 2010-01-04 19:18:44 +0000 | [diff] [blame] | 30 | #include "clang/Lex/ExternalPreprocessorSource.h" |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 31 | #include "clang/Lex/HeaderSearch.h" |
| 32 | #include "clang/Lex/MacroInfo.h" |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 33 | #include "clang/Lex/Pragma.h" |
Douglas Gregor | 94dc8f6 | 2010-03-19 16:15:56 +0000 | [diff] [blame] | 34 | #include "clang/Lex/PreprocessingRecord.h" |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 35 | #include "clang/Lex/ScratchBuffer.h" |
Chris Lattner | 500d329 | 2009-01-29 05:15:15 +0000 | [diff] [blame] | 36 | #include "clang/Lex/LexDiagnostic.h" |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 37 | #include "clang/Basic/SourceManager.h" |
Ted Kremenek | 337edcd | 2009-02-12 03:26:59 +0000 | [diff] [blame] | 38 | #include "clang/Basic/FileManager.h" |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 39 | #include "clang/Basic/TargetInfo.h" |
Chris Lattner | 2db78dd | 2008-10-05 20:40:30 +0000 | [diff] [blame] | 40 | #include "llvm/ADT/APFloat.h" |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 41 | #include "llvm/ADT/SmallVector.h" |
Chris Lattner | 97ba77c | 2007-07-16 06:48:38 +0000 | [diff] [blame] | 42 | #include "llvm/Support/MemoryBuffer.h" |
Benjamin Kramer | 6cb7c1a | 2009-08-23 12:08:50 +0000 | [diff] [blame] | 43 | #include "llvm/Support/raw_ostream.h" |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 44 | using namespace clang; |
| 45 | |
| 46 | //===----------------------------------------------------------------------===// |
Douglas Gregor | 88a3586 | 2010-01-04 19:18:44 +0000 | [diff] [blame] | 47 | ExternalPreprocessorSource::~ExternalPreprocessorSource() { } |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 48 | |
| 49 | Preprocessor::Preprocessor(Diagnostic &diags, const LangOptions &opts, |
Daniel Dunbar | 444be73 | 2009-11-13 05:51:54 +0000 | [diff] [blame] | 50 | const TargetInfo &target, SourceManager &SM, |
Ted Kremenek | 72b1b15 | 2009-01-15 18:47:46 +0000 | [diff] [blame] | 51 | HeaderSearch &Headers, |
Daniel Dunbar | 5814e65 | 2009-11-11 21:44:21 +0000 | [diff] [blame] | 52 | IdentifierInfoLookup* IILookup, |
| 53 | bool OwnsHeaders) |
Chris Lattner | 836040f | 2009-03-13 21:17:43 +0000 | [diff] [blame] | 54 | : Diags(&diags), Features(opts), Target(target),FileMgr(Headers.getFileMgr()), |
Douglas Gregor | 88a3586 | 2010-01-04 19:18:44 +0000 | [diff] [blame] | 55 | SourceMgr(SM), HeaderInfo(Headers), ExternalSource(0), |
Kovarththanan Rajaratnam | 1935754 | 2010-03-13 10:17:05 +0000 | [diff] [blame] | 56 | Identifiers(opts, IILookup), BuiltinInfo(Target), CodeCompletionFile(0), |
Douglas Gregor | b9e1b75 | 2010-03-19 17:12:43 +0000 | [diff] [blame] | 57 | CurPPLexer(0), CurDirLookup(0), Callbacks(0), MacroArgCache(0), Record(0) { |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 58 | ScratchBuf = new ScratchBuffer(SourceMgr); |
Chris Lattner | c1f9d82 | 2009-04-13 01:29:17 +0000 | [diff] [blame] | 59 | CounterValue = 0; // __COUNTER__ starts at 0. |
Daniel Dunbar | 5814e65 | 2009-11-11 21:44:21 +0000 | [diff] [blame] | 60 | OwnsHeaderSearch = OwnsHeaders; |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 61 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 62 | // Clear stats. |
| 63 | NumDirectives = NumDefined = NumUndefined = NumPragma = 0; |
| 64 | NumIf = NumElse = NumEndif = 0; |
| 65 | NumEnteredSourceFiles = 0; |
| 66 | NumMacroExpanded = NumFnMacroExpanded = NumBuiltinMacroExpanded = 0; |
| 67 | NumFastMacroExpanded = NumTokenPaste = NumFastTokenPaste = 0; |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 68 | MaxIncludeStackDepth = 0; |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 69 | NumSkipped = 0; |
| 70 | |
| 71 | // Default to discarding comments. |
| 72 | KeepComments = false; |
| 73 | KeepMacroComments = false; |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 74 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 75 | // Macro expansion is enabled. |
| 76 | DisableMacroExpansion = false; |
| 77 | InMacroArgs = false; |
Chris Lattner | 6cfe759 | 2008-03-09 02:26:03 +0000 | [diff] [blame] | 78 | NumCachedTokenLexers = 0; |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 79 | |
Argyrios Kyrtzidis | 03db1b3 | 2008-08-10 13:15:22 +0000 | [diff] [blame] | 80 | CachedLexPos = 0; |
| 81 | |
Douglas Gregor | 88a3586 | 2010-01-04 19:18:44 +0000 | [diff] [blame] | 82 | // We haven't read anything from the external source. |
| 83 | ReadMacrosFromExternalSource = false; |
Kovarththanan Rajaratnam | 1935754 | 2010-03-13 10:17:05 +0000 | [diff] [blame] | 84 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 85 | // "Poison" __VA_ARGS__, which can only appear in the expansion of a macro. |
| 86 | // This gets unpoisoned where it is allowed. |
| 87 | (Ident__VA_ARGS__ = getIdentifierInfo("__VA_ARGS__"))->setIsPoisoned(); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 88 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 89 | // Initialize the pragma handlers. |
| 90 | PragmaHandlers = new PragmaNamespace(0); |
| 91 | RegisterBuiltinPragmas(); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 92 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 93 | // Initialize builtin macros like __LINE__ and friends. |
| 94 | RegisterBuiltinMacros(); |
| 95 | } |
| 96 | |
| 97 | Preprocessor::~Preprocessor() { |
Argyrios Kyrtzidis | 2174a4f | 2008-08-23 12:12:06 +0000 | [diff] [blame] | 98 | assert(BacktrackPositions.empty() && "EnableBacktrack/Backtrack imbalance!"); |
| 99 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 100 | while (!IncludeMacroStack.empty()) { |
| 101 | delete IncludeMacroStack.back().TheLexer; |
Chris Lattner | 6cfe759 | 2008-03-09 02:26:03 +0000 | [diff] [blame] | 102 | delete IncludeMacroStack.back().TheTokenLexer; |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 103 | IncludeMacroStack.pop_back(); |
| 104 | } |
Chris Lattner | cc1a875 | 2007-10-07 08:44:20 +0000 | [diff] [blame] | 105 | |
| 106 | // Free any macro definitions. |
| 107 | for (llvm::DenseMap<IdentifierInfo*, MacroInfo*>::iterator I = |
| 108 | Macros.begin(), E = Macros.end(); I != E; ++I) { |
Ted Kremenek | 0ea7672 | 2008-12-15 19:56:42 +0000 | [diff] [blame] | 109 | // We don't need to free the MacroInfo objects directly. These |
| 110 | // will be released when the BumpPtrAllocator 'BP' object gets |
Chris Lattner | 2b9d637 | 2009-12-14 22:02:43 +0000 | [diff] [blame] | 111 | // destroyed. We still need to run the dtor, however, to free |
Ted Kremenek | 9ee7d45 | 2009-01-19 07:45:44 +0000 | [diff] [blame] | 112 | // memory alocated by MacroInfo. |
Chris Lattner | 685befe | 2009-02-20 22:46:43 +0000 | [diff] [blame] | 113 | I->second->Destroy(BP); |
Chris Lattner | cc1a875 | 2007-10-07 08:44:20 +0000 | [diff] [blame] | 114 | I->first->setHasMacroDefinition(false); |
| 115 | } |
Ted Kremenek | 83c1a6f | 2010-06-08 23:00:53 +0000 | [diff] [blame^] | 116 | for (std::vector<MacroInfo*>::iterator I = MICache.begin(), |
| 117 | E = MICache.end(); I != E; ++I) { |
| 118 | // We don't need to free the MacroInfo objects directly. These |
| 119 | // will be released when the BumpPtrAllocator 'BP' object gets |
| 120 | // destroyed. We still need to run the dtor, however, to free |
| 121 | // memory alocated by MacroInfo. |
| 122 | (*I)->Destroy(BP); |
| 123 | } |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 124 | |
Chris Lattner | 9594acf | 2007-07-15 00:25:26 +0000 | [diff] [blame] | 125 | // Free any cached macro expanders. |
Chris Lattner | 6cfe759 | 2008-03-09 02:26:03 +0000 | [diff] [blame] | 126 | for (unsigned i = 0, e = NumCachedTokenLexers; i != e; ++i) |
| 127 | delete TokenLexerCache[i]; |
Kovarththanan Rajaratnam | 1935754 | 2010-03-13 10:17:05 +0000 | [diff] [blame] | 128 | |
Chris Lattner | 23f77e5 | 2009-12-15 01:51:03 +0000 | [diff] [blame] | 129 | // Free any cached MacroArgs. |
| 130 | for (MacroArgs *ArgList = MacroArgCache; ArgList; ) |
| 131 | ArgList = ArgList->deallocate(); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 132 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 133 | // Release pragma information. |
| 134 | delete PragmaHandlers; |
| 135 | |
| 136 | // Delete the scratch buffer info. |
| 137 | delete ScratchBuf; |
Chris Lattner | eb50ed8 | 2008-03-14 06:07:05 +0000 | [diff] [blame] | 138 | |
Daniel Dunbar | 5814e65 | 2009-11-11 21:44:21 +0000 | [diff] [blame] | 139 | // Delete the header search info, if we own it. |
| 140 | if (OwnsHeaderSearch) |
| 141 | delete &HeaderInfo; |
| 142 | |
Chris Lattner | eb50ed8 | 2008-03-14 06:07:05 +0000 | [diff] [blame] | 143 | delete Callbacks; |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 144 | } |
| 145 | |
Ted Kremenek | 337edcd | 2009-02-12 03:26:59 +0000 | [diff] [blame] | 146 | void Preprocessor::setPTHManager(PTHManager* pm) { |
| 147 | PTH.reset(pm); |
Douglas Gregor | 52e7108 | 2009-10-16 18:18:30 +0000 | [diff] [blame] | 148 | FileMgr.addStatCache(PTH->createStatCache()); |
Ted Kremenek | 337edcd | 2009-02-12 03:26:59 +0000 | [diff] [blame] | 149 | } |
| 150 | |
Chris Lattner | d217773 | 2007-07-20 16:59:19 +0000 | [diff] [blame] | 151 | void Preprocessor::DumpToken(const Token &Tok, bool DumpFlags) const { |
Benjamin Kramer | 6cb7c1a | 2009-08-23 12:08:50 +0000 | [diff] [blame] | 152 | llvm::errs() << tok::getTokenName(Tok.getKind()) << " '" |
| 153 | << getSpelling(Tok) << "'"; |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 154 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 155 | if (!DumpFlags) return; |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 156 | |
Benjamin Kramer | 6cb7c1a | 2009-08-23 12:08:50 +0000 | [diff] [blame] | 157 | llvm::errs() << "\t"; |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 158 | if (Tok.isAtStartOfLine()) |
Benjamin Kramer | 6cb7c1a | 2009-08-23 12:08:50 +0000 | [diff] [blame] | 159 | llvm::errs() << " [StartOfLine]"; |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 160 | if (Tok.hasLeadingSpace()) |
Benjamin Kramer | 6cb7c1a | 2009-08-23 12:08:50 +0000 | [diff] [blame] | 161 | llvm::errs() << " [LeadingSpace]"; |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 162 | if (Tok.isExpandDisabled()) |
Benjamin Kramer | 6cb7c1a | 2009-08-23 12:08:50 +0000 | [diff] [blame] | 163 | llvm::errs() << " [ExpandDisabled]"; |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 164 | if (Tok.needsCleaning()) { |
| 165 | const char *Start = SourceMgr.getCharacterData(Tok.getLocation()); |
Benjamin Kramer | 6cb7c1a | 2009-08-23 12:08:50 +0000 | [diff] [blame] | 166 | llvm::errs() << " [UnClean='" << std::string(Start, Start+Tok.getLength()) |
| 167 | << "']"; |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 168 | } |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 169 | |
Benjamin Kramer | 6cb7c1a | 2009-08-23 12:08:50 +0000 | [diff] [blame] | 170 | llvm::errs() << "\tLoc=<"; |
Chris Lattner | c3d8d57 | 2007-12-09 20:31:55 +0000 | [diff] [blame] | 171 | DumpLocation(Tok.getLocation()); |
Benjamin Kramer | 6cb7c1a | 2009-08-23 12:08:50 +0000 | [diff] [blame] | 172 | llvm::errs() << ">"; |
Chris Lattner | c3d8d57 | 2007-12-09 20:31:55 +0000 | [diff] [blame] | 173 | } |
| 174 | |
| 175 | void Preprocessor::DumpLocation(SourceLocation Loc) const { |
Chris Lattner | b9c3f96 | 2009-01-27 07:57:44 +0000 | [diff] [blame] | 176 | Loc.dump(SourceMgr); |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 177 | } |
| 178 | |
| 179 | void Preprocessor::DumpMacro(const MacroInfo &MI) const { |
Benjamin Kramer | 6cb7c1a | 2009-08-23 12:08:50 +0000 | [diff] [blame] | 180 | llvm::errs() << "MACRO: "; |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 181 | for (unsigned i = 0, e = MI.getNumTokens(); i != e; ++i) { |
| 182 | DumpToken(MI.getReplacementToken(i)); |
Benjamin Kramer | 6cb7c1a | 2009-08-23 12:08:50 +0000 | [diff] [blame] | 183 | llvm::errs() << " "; |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 184 | } |
Benjamin Kramer | 6cb7c1a | 2009-08-23 12:08:50 +0000 | [diff] [blame] | 185 | llvm::errs() << "\n"; |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 186 | } |
| 187 | |
| 188 | void Preprocessor::PrintStats() { |
Benjamin Kramer | 6cb7c1a | 2009-08-23 12:08:50 +0000 | [diff] [blame] | 189 | llvm::errs() << "\n*** Preprocessor Stats:\n"; |
| 190 | llvm::errs() << NumDirectives << " directives found:\n"; |
| 191 | llvm::errs() << " " << NumDefined << " #define.\n"; |
| 192 | llvm::errs() << " " << NumUndefined << " #undef.\n"; |
| 193 | llvm::errs() << " #include/#include_next/#import:\n"; |
| 194 | llvm::errs() << " " << NumEnteredSourceFiles << " source files entered.\n"; |
| 195 | llvm::errs() << " " << MaxIncludeStackDepth << " max include stack depth\n"; |
| 196 | llvm::errs() << " " << NumIf << " #if/#ifndef/#ifdef.\n"; |
| 197 | llvm::errs() << " " << NumElse << " #else/#elif.\n"; |
| 198 | llvm::errs() << " " << NumEndif << " #endif.\n"; |
| 199 | llvm::errs() << " " << NumPragma << " #pragma.\n"; |
| 200 | llvm::errs() << NumSkipped << " #if/#ifndef#ifdef regions skipped\n"; |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 201 | |
Benjamin Kramer | 6cb7c1a | 2009-08-23 12:08:50 +0000 | [diff] [blame] | 202 | llvm::errs() << NumMacroExpanded << "/" << NumFnMacroExpanded << "/" |
Ted Kremenek | bdd30c2 | 2008-01-14 16:44:48 +0000 | [diff] [blame] | 203 | << NumBuiltinMacroExpanded << " obj/fn/builtin macros expanded, " |
| 204 | << NumFastMacroExpanded << " on the fast path.\n"; |
Benjamin Kramer | 6cb7c1a | 2009-08-23 12:08:50 +0000 | [diff] [blame] | 205 | llvm::errs() << (NumFastTokenPaste+NumTokenPaste) |
Ted Kremenek | bdd30c2 | 2008-01-14 16:44:48 +0000 | [diff] [blame] | 206 | << " token paste (##) operations performed, " |
| 207 | << NumFastTokenPaste << " on the fast path.\n"; |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 208 | } |
| 209 | |
Kovarththanan Rajaratnam | 1935754 | 2010-03-13 10:17:05 +0000 | [diff] [blame] | 210 | Preprocessor::macro_iterator |
| 211 | Preprocessor::macro_begin(bool IncludeExternalMacros) const { |
| 212 | if (IncludeExternalMacros && ExternalSource && |
Douglas Gregor | 88a3586 | 2010-01-04 19:18:44 +0000 | [diff] [blame] | 213 | !ReadMacrosFromExternalSource) { |
| 214 | ReadMacrosFromExternalSource = true; |
| 215 | ExternalSource->ReadDefinedMacros(); |
| 216 | } |
Kovarththanan Rajaratnam | 1935754 | 2010-03-13 10:17:05 +0000 | [diff] [blame] | 217 | |
| 218 | return Macros.begin(); |
Douglas Gregor | 88a3586 | 2010-01-04 19:18:44 +0000 | [diff] [blame] | 219 | } |
| 220 | |
Kovarththanan Rajaratnam | 1935754 | 2010-03-13 10:17:05 +0000 | [diff] [blame] | 221 | Preprocessor::macro_iterator |
| 222 | Preprocessor::macro_end(bool IncludeExternalMacros) const { |
| 223 | if (IncludeExternalMacros && ExternalSource && |
Douglas Gregor | 88a3586 | 2010-01-04 19:18:44 +0000 | [diff] [blame] | 224 | !ReadMacrosFromExternalSource) { |
| 225 | ReadMacrosFromExternalSource = true; |
| 226 | ExternalSource->ReadDefinedMacros(); |
| 227 | } |
Kovarththanan Rajaratnam | 1935754 | 2010-03-13 10:17:05 +0000 | [diff] [blame] | 228 | |
| 229 | return Macros.end(); |
Douglas Gregor | 88a3586 | 2010-01-04 19:18:44 +0000 | [diff] [blame] | 230 | } |
| 231 | |
Kovarththanan Rajaratnam | 1935754 | 2010-03-13 10:17:05 +0000 | [diff] [blame] | 232 | bool Preprocessor::SetCodeCompletionPoint(const FileEntry *File, |
| 233 | unsigned TruncateAtLine, |
Douglas Gregor | 2968442 | 2009-12-02 06:49:09 +0000 | [diff] [blame] | 234 | unsigned TruncateAtColumn) { |
| 235 | using llvm::MemoryBuffer; |
| 236 | |
| 237 | CodeCompletionFile = File; |
| 238 | |
| 239 | // Okay to clear out the code-completion point by passing NULL. |
| 240 | if (!CodeCompletionFile) |
| 241 | return false; |
| 242 | |
| 243 | // Load the actual file's contents. |
Douglas Gregor | aa38c3d | 2010-03-16 19:49:24 +0000 | [diff] [blame] | 244 | bool Invalid = false; |
| 245 | const MemoryBuffer *Buffer = SourceMgr.getMemoryBufferForFile(File, &Invalid); |
| 246 | if (Invalid) |
Douglas Gregor | 2968442 | 2009-12-02 06:49:09 +0000 | [diff] [blame] | 247 | return true; |
| 248 | |
| 249 | // Find the byte position of the truncation point. |
| 250 | const char *Position = Buffer->getBufferStart(); |
| 251 | for (unsigned Line = 1; Line < TruncateAtLine; ++Line) { |
| 252 | for (; *Position; ++Position) { |
| 253 | if (*Position != '\r' && *Position != '\n') |
| 254 | continue; |
Kovarththanan Rajaratnam | 1935754 | 2010-03-13 10:17:05 +0000 | [diff] [blame] | 255 | |
Douglas Gregor | 2968442 | 2009-12-02 06:49:09 +0000 | [diff] [blame] | 256 | // Eat \r\n or \n\r as a single line. |
| 257 | if ((Position[1] == '\r' || Position[1] == '\n') && |
| 258 | Position[0] != Position[1]) |
| 259 | ++Position; |
| 260 | ++Position; |
| 261 | break; |
| 262 | } |
| 263 | } |
Kovarththanan Rajaratnam | 1935754 | 2010-03-13 10:17:05 +0000 | [diff] [blame] | 264 | |
Douglas Gregor | b760fe8 | 2009-12-08 21:45:46 +0000 | [diff] [blame] | 265 | Position += TruncateAtColumn - 1; |
Kovarththanan Rajaratnam | 1935754 | 2010-03-13 10:17:05 +0000 | [diff] [blame] | 266 | |
Douglas Gregor | 2968442 | 2009-12-02 06:49:09 +0000 | [diff] [blame] | 267 | // Truncate the buffer. |
Douglas Gregor | b760fe8 | 2009-12-08 21:45:46 +0000 | [diff] [blame] | 268 | if (Position < Buffer->getBufferEnd()) { |
Chris Lattner | a0a270c | 2010-04-05 22:42:27 +0000 | [diff] [blame] | 269 | llvm::StringRef Data(Buffer->getBufferStart(), |
| 270 | Position-Buffer->getBufferStart()); |
Kovarththanan Rajaratnam | 1935754 | 2010-03-13 10:17:05 +0000 | [diff] [blame] | 271 | MemoryBuffer *TruncatedBuffer |
Chris Lattner | a0a270c | 2010-04-05 22:42:27 +0000 | [diff] [blame] | 272 | = MemoryBuffer::getMemBufferCopy(Data, Buffer->getBufferIdentifier()); |
Douglas Gregor | 2968442 | 2009-12-02 06:49:09 +0000 | [diff] [blame] | 273 | SourceMgr.overrideFileContents(File, TruncatedBuffer); |
| 274 | } |
| 275 | |
| 276 | return false; |
| 277 | } |
| 278 | |
Douglas Gregor | 109ae73 | 2009-12-03 17:05:59 +0000 | [diff] [blame] | 279 | bool Preprocessor::isCodeCompletionFile(SourceLocation FileLoc) const { |
Douglas Gregor | 2968442 | 2009-12-02 06:49:09 +0000 | [diff] [blame] | 280 | return CodeCompletionFile && FileLoc.isFileID() && |
| 281 | SourceMgr.getFileEntryForID(SourceMgr.getFileID(FileLoc)) |
| 282 | == CodeCompletionFile; |
| 283 | } |
| 284 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 285 | //===----------------------------------------------------------------------===// |
| 286 | // Token Spelling |
| 287 | //===----------------------------------------------------------------------===// |
| 288 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 289 | /// getSpelling() - Return the 'spelling' of this token. The spelling of a |
| 290 | /// token are the characters used to represent the token in the source file |
| 291 | /// after trigraph expansion and escaped-newline folding. In particular, this |
| 292 | /// wants to get the true, uncanonicalized, spelling of things like digraphs |
| 293 | /// UCNs, etc. |
Daniel Dunbar | 0ff1042 | 2009-11-14 01:20:48 +0000 | [diff] [blame] | 294 | std::string Preprocessor::getSpelling(const Token &Tok, |
| 295 | const SourceManager &SourceMgr, |
Douglas Gregor | 50f6af7 | 2010-03-16 05:20:39 +0000 | [diff] [blame] | 296 | const LangOptions &Features, |
| 297 | bool *Invalid) { |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 298 | assert((int)Tok.getLength() >= 0 && "Token character range is bogus!"); |
Ted Kremenek | 277faca | 2009-01-27 00:01:05 +0000 | [diff] [blame] | 299 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 300 | // If this token contains nothing interesting, return it directly. |
Douglas Gregor | 50f6af7 | 2010-03-16 05:20:39 +0000 | [diff] [blame] | 301 | bool CharDataInvalid = false; |
| 302 | const char* TokStart = SourceMgr.getCharacterData(Tok.getLocation(), |
| 303 | &CharDataInvalid); |
| 304 | if (Invalid) |
| 305 | *Invalid = CharDataInvalid; |
| 306 | if (CharDataInvalid) |
| 307 | return std::string(); |
| 308 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 309 | if (!Tok.needsCleaning()) |
| 310 | return std::string(TokStart, TokStart+Tok.getLength()); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 311 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 312 | std::string Result; |
| 313 | Result.reserve(Tok.getLength()); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 314 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 315 | // Otherwise, hard case, relex the characters into the string. |
| 316 | for (const char *Ptr = TokStart, *End = TokStart+Tok.getLength(); |
| 317 | Ptr != End; ) { |
| 318 | unsigned CharSize; |
| 319 | Result.push_back(Lexer::getCharAndSizeNoWarn(Ptr, CharSize, Features)); |
| 320 | Ptr += CharSize; |
| 321 | } |
| 322 | assert(Result.size() != unsigned(Tok.getLength()) && |
| 323 | "NeedsCleaning flag set on something that didn't need cleaning!"); |
| 324 | return Result; |
| 325 | } |
| 326 | |
Daniel Dunbar | 0ff1042 | 2009-11-14 01:20:48 +0000 | [diff] [blame] | 327 | /// getSpelling() - Return the 'spelling' of this token. The spelling of a |
| 328 | /// token are the characters used to represent the token in the source file |
| 329 | /// after trigraph expansion and escaped-newline folding. In particular, this |
| 330 | /// wants to get the true, uncanonicalized, spelling of things like digraphs |
| 331 | /// UCNs, etc. |
Douglas Gregor | 50f6af7 | 2010-03-16 05:20:39 +0000 | [diff] [blame] | 332 | std::string Preprocessor::getSpelling(const Token &Tok, bool *Invalid) const { |
| 333 | return getSpelling(Tok, SourceMgr, Features, Invalid); |
Daniel Dunbar | 0ff1042 | 2009-11-14 01:20:48 +0000 | [diff] [blame] | 334 | } |
| 335 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 336 | /// getSpelling - This method is used to get the spelling of a token into a |
| 337 | /// preallocated buffer, instead of as an std::string. The caller is required |
| 338 | /// to allocate enough space for the token, which is guaranteed to be at least |
| 339 | /// Tok.getLength() bytes long. The actual length of the token is returned. |
| 340 | /// |
| 341 | /// Note that this method may do two possible things: it may either fill in |
| 342 | /// the buffer specified with characters, or it may *change the input pointer* |
| 343 | /// to point to a constant buffer with the data already in it (avoiding a |
| 344 | /// copy). The caller is not allowed to modify the returned buffer pointer |
| 345 | /// if an internal buffer is returned. |
Chris Lattner | d217773 | 2007-07-20 16:59:19 +0000 | [diff] [blame] | 346 | unsigned Preprocessor::getSpelling(const Token &Tok, |
Douglas Gregor | 50f6af7 | 2010-03-16 05:20:39 +0000 | [diff] [blame] | 347 | const char *&Buffer, bool *Invalid) const { |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 348 | assert((int)Tok.getLength() >= 0 && "Token character range is bogus!"); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 349 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 350 | // If this token is an identifier, just return the string from the identifier |
| 351 | // table, which is very quick. |
| 352 | if (const IdentifierInfo *II = Tok.getIdentifierInfo()) { |
Daniel Dunbar | e013d68 | 2009-10-18 20:26:12 +0000 | [diff] [blame] | 353 | Buffer = II->getNameStart(); |
Chris Lattner | e1dccae | 2009-01-05 19:44:41 +0000 | [diff] [blame] | 354 | return II->getLength(); |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 355 | } |
Ted Kremenek | b70e3da | 2009-01-08 02:47:16 +0000 | [diff] [blame] | 356 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 357 | // Otherwise, compute the start of the token in the input lexer buffer. |
Chris Lattner | 47246be | 2009-01-26 19:29:26 +0000 | [diff] [blame] | 358 | const char *TokStart = 0; |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 359 | |
Chris Lattner | 47246be | 2009-01-26 19:29:26 +0000 | [diff] [blame] | 360 | if (Tok.isLiteral()) |
| 361 | TokStart = Tok.getLiteralData(); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 362 | |
Douglas Gregor | 50f6af7 | 2010-03-16 05:20:39 +0000 | [diff] [blame] | 363 | if (TokStart == 0) { |
| 364 | bool CharDataInvalid = false; |
| 365 | TokStart = SourceMgr.getCharacterData(Tok.getLocation(), &CharDataInvalid); |
| 366 | if (Invalid) |
| 367 | *Invalid = CharDataInvalid; |
| 368 | if (CharDataInvalid) { |
| 369 | Buffer = ""; |
| 370 | return 0; |
| 371 | } |
| 372 | } |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 373 | |
| 374 | // If this token contains nothing interesting, return it directly. |
| 375 | if (!Tok.needsCleaning()) { |
| 376 | Buffer = TokStart; |
| 377 | return Tok.getLength(); |
| 378 | } |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 379 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 380 | // Otherwise, hard case, relex the characters into the string. |
| 381 | char *OutBuf = const_cast<char*>(Buffer); |
| 382 | for (const char *Ptr = TokStart, *End = TokStart+Tok.getLength(); |
| 383 | Ptr != End; ) { |
| 384 | unsigned CharSize; |
| 385 | *OutBuf++ = Lexer::getCharAndSizeNoWarn(Ptr, CharSize, Features); |
| 386 | Ptr += CharSize; |
| 387 | } |
| 388 | assert(unsigned(OutBuf-Buffer) != Tok.getLength() && |
| 389 | "NeedsCleaning flag set on something that didn't need cleaning!"); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 390 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 391 | return OutBuf-Buffer; |
| 392 | } |
| 393 | |
Benjamin Kramer | 51f5fe3 | 2010-02-27 17:05:45 +0000 | [diff] [blame] | 394 | /// getSpelling - This method is used to get the spelling of a token into a |
| 395 | /// SmallVector. Note that the returned StringRef may not point to the |
| 396 | /// supplied buffer if a copy can be avoided. |
| 397 | llvm::StringRef Preprocessor::getSpelling(const Token &Tok, |
Douglas Gregor | 50f6af7 | 2010-03-16 05:20:39 +0000 | [diff] [blame] | 398 | llvm::SmallVectorImpl<char> &Buffer, |
| 399 | bool *Invalid) const { |
Benjamin Kramer | 51f5fe3 | 2010-02-27 17:05:45 +0000 | [diff] [blame] | 400 | // Try the fast path. |
| 401 | if (const IdentifierInfo *II = Tok.getIdentifierInfo()) |
| 402 | return II->getName(); |
| 403 | |
| 404 | // Resize the buffer if we need to copy into it. |
| 405 | if (Tok.needsCleaning()) |
| 406 | Buffer.resize(Tok.getLength()); |
| 407 | |
| 408 | const char *Ptr = Buffer.data(); |
Douglas Gregor | 50f6af7 | 2010-03-16 05:20:39 +0000 | [diff] [blame] | 409 | unsigned Len = getSpelling(Tok, Ptr, Invalid); |
Benjamin Kramer | 51f5fe3 | 2010-02-27 17:05:45 +0000 | [diff] [blame] | 410 | return llvm::StringRef(Ptr, Len); |
| 411 | } |
| 412 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 413 | /// CreateString - Plop the specified string into a scratch buffer and return a |
| 414 | /// location for it. If specified, the source location provides a source |
| 415 | /// location for the token. |
Chris Lattner | 47246be | 2009-01-26 19:29:26 +0000 | [diff] [blame] | 416 | void Preprocessor::CreateString(const char *Buf, unsigned Len, Token &Tok, |
| 417 | SourceLocation InstantiationLoc) { |
| 418 | Tok.setLength(Len); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 419 | |
Chris Lattner | 47246be | 2009-01-26 19:29:26 +0000 | [diff] [blame] | 420 | const char *DestPtr; |
| 421 | SourceLocation Loc = ScratchBuf->getToken(Buf, Len, DestPtr); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 422 | |
Chris Lattner | 47246be | 2009-01-26 19:29:26 +0000 | [diff] [blame] | 423 | if (InstantiationLoc.isValid()) |
Chris Lattner | e7fb484 | 2009-02-15 20:52:18 +0000 | [diff] [blame] | 424 | Loc = SourceMgr.createInstantiationLoc(Loc, InstantiationLoc, |
| 425 | InstantiationLoc, Len); |
Chris Lattner | 47246be | 2009-01-26 19:29:26 +0000 | [diff] [blame] | 426 | Tok.setLocation(Loc); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 427 | |
Chris Lattner | 47246be | 2009-01-26 19:29:26 +0000 | [diff] [blame] | 428 | // If this is a literal token, set the pointer data. |
| 429 | if (Tok.isLiteral()) |
| 430 | Tok.setLiteralData(DestPtr); |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 431 | } |
| 432 | |
| 433 | |
Chris Lattner | 97ba77c | 2007-07-16 06:48:38 +0000 | [diff] [blame] | 434 | /// AdvanceToTokenCharacter - Given a location that specifies the start of a |
| 435 | /// token, return a new location that specifies a character within the token. |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 436 | SourceLocation Preprocessor::AdvanceToTokenCharacter(SourceLocation TokStart, |
Chris Lattner | 97ba77c | 2007-07-16 06:48:38 +0000 | [diff] [blame] | 437 | unsigned CharNo) { |
Chris Lattner | f7cf85b | 2009-01-16 07:36:28 +0000 | [diff] [blame] | 438 | // Figure out how many physical characters away the specified instantiation |
Chris Lattner | 97ba77c | 2007-07-16 06:48:38 +0000 | [diff] [blame] | 439 | // character is. This needs to take into consideration newlines and |
| 440 | // trigraphs. |
Douglas Gregor | a543016 | 2010-03-16 20:46:42 +0000 | [diff] [blame] | 441 | bool Invalid = false; |
| 442 | const char *TokPtr = SourceMgr.getCharacterData(TokStart, &Invalid); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 443 | |
Chris Lattner | 88e2524 | 2009-04-18 22:28:58 +0000 | [diff] [blame] | 444 | // If they request the first char of the token, we're trivially done. |
Douglas Gregor | a543016 | 2010-03-16 20:46:42 +0000 | [diff] [blame] | 445 | if (Invalid || (CharNo == 0 && Lexer::isObviouslySimpleCharacter(*TokPtr))) |
Chris Lattner | 88e2524 | 2009-04-18 22:28:58 +0000 | [diff] [blame] | 446 | return TokStart; |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 447 | |
Chris Lattner | 9dc1f53 | 2007-07-20 16:37:10 +0000 | [diff] [blame] | 448 | unsigned PhysOffset = 0; |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 449 | |
Chris Lattner | 97ba77c | 2007-07-16 06:48:38 +0000 | [diff] [blame] | 450 | // The usual case is that tokens don't contain anything interesting. Skip |
| 451 | // over the uninteresting characters. If a token only consists of simple |
| 452 | // chars, this method is extremely fast. |
Chris Lattner | 88e2524 | 2009-04-18 22:28:58 +0000 | [diff] [blame] | 453 | while (Lexer::isObviouslySimpleCharacter(*TokPtr)) { |
| 454 | if (CharNo == 0) |
| 455 | return TokStart.getFileLocWithOffset(PhysOffset); |
Chris Lattner | 9dc1f53 | 2007-07-20 16:37:10 +0000 | [diff] [blame] | 456 | ++TokPtr, --CharNo, ++PhysOffset; |
Chris Lattner | 88e2524 | 2009-04-18 22:28:58 +0000 | [diff] [blame] | 457 | } |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 458 | |
Chris Lattner | 28c90ad | 2009-01-17 07:57:25 +0000 | [diff] [blame] | 459 | // If we have a character that may be a trigraph or escaped newline, use a |
Chris Lattner | 97ba77c | 2007-07-16 06:48:38 +0000 | [diff] [blame] | 460 | // lexer to parse it correctly. |
Chris Lattner | 88e2524 | 2009-04-18 22:28:58 +0000 | [diff] [blame] | 461 | for (; CharNo; --CharNo) { |
| 462 | unsigned Size; |
| 463 | Lexer::getCharAndSizeNoWarn(TokPtr, Size, Features); |
| 464 | TokPtr += Size; |
| 465 | PhysOffset += Size; |
Chris Lattner | 97ba77c | 2007-07-16 06:48:38 +0000 | [diff] [blame] | 466 | } |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 467 | |
Chris Lattner | 88e2524 | 2009-04-18 22:28:58 +0000 | [diff] [blame] | 468 | // Final detail: if we end up on an escaped newline, we want to return the |
| 469 | // location of the actual byte of the token. For example foo\<newline>bar |
| 470 | // advanced by 3 should return the location of b, not of \\. One compounding |
| 471 | // detail of this is that the escape may be made by a trigraph. |
| 472 | if (!Lexer::isObviouslySimpleCharacter(*TokPtr)) |
Ted Kremenek | c0178e9 | 2010-01-29 19:38:24 +0000 | [diff] [blame] | 473 | PhysOffset += Lexer::SkipEscapedNewLines(TokPtr)-TokPtr; |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 474 | |
Chris Lattner | 9dc1f53 | 2007-07-20 16:37:10 +0000 | [diff] [blame] | 475 | return TokStart.getFileLocWithOffset(PhysOffset); |
Chris Lattner | 97ba77c | 2007-07-16 06:48:38 +0000 | [diff] [blame] | 476 | } |
| 477 | |
Kovarththanan Rajaratnam | 1935754 | 2010-03-13 10:17:05 +0000 | [diff] [blame] | 478 | SourceLocation Preprocessor::getLocForEndOfToken(SourceLocation Loc, |
Douglas Gregor | 33e9abd | 2010-01-22 19:49:59 +0000 | [diff] [blame] | 479 | unsigned Offset) { |
Douglas Gregor | b2fb6de | 2009-02-27 17:53:17 +0000 | [diff] [blame] | 480 | if (Loc.isInvalid() || !Loc.isFileID()) |
| 481 | return SourceLocation(); |
| 482 | |
Chris Lattner | 2c78b87 | 2009-04-14 23:22:57 +0000 | [diff] [blame] | 483 | unsigned Len = Lexer::MeasureTokenLength(Loc, getSourceManager(), Features); |
Douglas Gregor | 33e9abd | 2010-01-22 19:49:59 +0000 | [diff] [blame] | 484 | if (Len > Offset) |
| 485 | Len = Len - Offset; |
| 486 | else |
| 487 | return Loc; |
Kovarththanan Rajaratnam | 1935754 | 2010-03-13 10:17:05 +0000 | [diff] [blame] | 488 | |
Douglas Gregor | b2fb6de | 2009-02-27 17:53:17 +0000 | [diff] [blame] | 489 | return AdvanceToTokenCharacter(Loc, Len); |
| 490 | } |
| 491 | |
| 492 | |
Chris Lattner | 97ba77c | 2007-07-16 06:48:38 +0000 | [diff] [blame] | 493 | |
Chris Lattner | 53b0dab | 2007-10-09 22:10:18 +0000 | [diff] [blame] | 494 | //===----------------------------------------------------------------------===// |
| 495 | // Preprocessor Initialization Methods |
| 496 | //===----------------------------------------------------------------------===// |
| 497 | |
Chris Lattner | 53b0dab | 2007-10-09 22:10:18 +0000 | [diff] [blame] | 498 | |
| 499 | /// EnterMainSourceFile - Enter the specified FileID as the main source file, |
Nate Begeman | 6b61602 | 2008-01-07 04:01:26 +0000 | [diff] [blame] | 500 | /// which implicitly adds the builtin defines etc. |
Chris Lattner | e127a0d | 2010-04-20 20:35:58 +0000 | [diff] [blame] | 501 | void Preprocessor::EnterMainSourceFile() { |
Chris Lattner | 05db427 | 2009-02-13 19:33:24 +0000 | [diff] [blame] | 502 | // We do not allow the preprocessor to reenter the main file. Doing so will |
| 503 | // cause FileID's to accumulate information from both runs (e.g. #line |
| 504 | // information) and predefined macros aren't guaranteed to be set properly. |
| 505 | assert(NumEnteredSourceFiles == 0 && "Cannot reenter the main file!"); |
Chris Lattner | 2b2453a | 2009-01-17 06:22:33 +0000 | [diff] [blame] | 506 | FileID MainFileID = SourceMgr.getMainFileID(); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 507 | |
Chris Lattner | 53b0dab | 2007-10-09 22:10:18 +0000 | [diff] [blame] | 508 | // Enter the main file source buffer. |
Chris Lattner | e127a0d | 2010-04-20 20:35:58 +0000 | [diff] [blame] | 509 | EnterSourceFile(MainFileID, 0, SourceLocation()); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 510 | |
Chris Lattner | b283298 | 2007-11-15 19:07:47 +0000 | [diff] [blame] | 511 | // Tell the header info that the main file was entered. If the file is later |
| 512 | // #imported, it won't be re-entered. |
Chris Lattner | 2b2453a | 2009-01-17 06:22:33 +0000 | [diff] [blame] | 513 | if (const FileEntry *FE = SourceMgr.getFileEntryForID(MainFileID)) |
Chris Lattner | b283298 | 2007-11-15 19:07:47 +0000 | [diff] [blame] | 514 | HeaderInfo.IncrementIncludeCount(FE); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 515 | |
Benjamin Kramer | ffd6e39 | 2009-12-31 15:33:09 +0000 | [diff] [blame] | 516 | // Preprocess Predefines to populate the initial preprocessor state. |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 517 | llvm::MemoryBuffer *SB = |
Chris Lattner | a0a270c | 2010-04-05 22:42:27 +0000 | [diff] [blame] | 518 | llvm::MemoryBuffer::getMemBufferCopy(Predefines, "<built-in>"); |
Chris Lattner | 53b0dab | 2007-10-09 22:10:18 +0000 | [diff] [blame] | 519 | assert(SB && "Cannot fail to create predefined source buffer"); |
Chris Lattner | 2b2453a | 2009-01-17 06:22:33 +0000 | [diff] [blame] | 520 | FileID FID = SourceMgr.createFileIDForMemBuffer(SB); |
| 521 | assert(!FID.isInvalid() && "Could not create FileID for predefines?"); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 522 | |
Chris Lattner | 53b0dab | 2007-10-09 22:10:18 +0000 | [diff] [blame] | 523 | // Start parsing the predefines. |
Chris Lattner | e127a0d | 2010-04-20 20:35:58 +0000 | [diff] [blame] | 524 | EnterSourceFile(FID, 0, SourceLocation()); |
Chris Lattner | 53b0dab | 2007-10-09 22:10:18 +0000 | [diff] [blame] | 525 | } |
Chris Lattner | 97ba77c | 2007-07-16 06:48:38 +0000 | [diff] [blame] | 526 | |
Daniel Dunbar | dbd8209 | 2010-03-23 05:09:10 +0000 | [diff] [blame] | 527 | void Preprocessor::EndSourceFile() { |
| 528 | // Notify the client that we reached the end of the source file. |
| 529 | if (Callbacks) |
| 530 | Callbacks->EndOfMainFile(); |
| 531 | } |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 532 | |
| 533 | //===----------------------------------------------------------------------===// |
| 534 | // Lexer Event Handling. |
| 535 | //===----------------------------------------------------------------------===// |
| 536 | |
| 537 | /// LookUpIdentifierInfo - Given a tok::identifier token, look up the |
| 538 | /// identifier information for the token and install it into the token. |
Chris Lattner | d217773 | 2007-07-20 16:59:19 +0000 | [diff] [blame] | 539 | IdentifierInfo *Preprocessor::LookUpIdentifierInfo(Token &Identifier, |
Daniel Dunbar | c322209 | 2009-11-05 01:53:52 +0000 | [diff] [blame] | 540 | const char *BufPtr) const { |
Chris Lattner | 22f6bbc | 2007-10-09 18:02:16 +0000 | [diff] [blame] | 541 | assert(Identifier.is(tok::identifier) && "Not an identifier!"); |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 542 | assert(Identifier.getIdentifierInfo() == 0 && "Identinfo already exists!"); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 543 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 544 | // Look up this token, see if it is a macro, or if it is a language keyword. |
| 545 | IdentifierInfo *II; |
| 546 | if (BufPtr && !Identifier.needsCleaning()) { |
| 547 | // No cleaning needed, just use the characters from the lexed buffer. |
Daniel Dunbar | 3da736c | 2009-11-05 01:53:39 +0000 | [diff] [blame] | 548 | II = getIdentifierInfo(llvm::StringRef(BufPtr, Identifier.getLength())); |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 549 | } else { |
| 550 | // Cleaning needed, alloca a buffer, clean into it, then use the buffer. |
Kovarththanan Rajaratnam | 1935754 | 2010-03-13 10:17:05 +0000 | [diff] [blame] | 551 | llvm::SmallString<64> IdentifierBuffer; |
Benjamin Kramer | ddeea56 | 2010-02-27 13:44:12 +0000 | [diff] [blame] | 552 | llvm::StringRef CleanedStr = getSpelling(Identifier, IdentifierBuffer); |
| 553 | II = getIdentifierInfo(CleanedStr); |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 554 | } |
| 555 | Identifier.setIdentifierInfo(II); |
| 556 | return II; |
| 557 | } |
| 558 | |
| 559 | |
| 560 | /// HandleIdentifier - This callback is invoked when the lexer reads an |
| 561 | /// identifier. This callback looks up the identifier in the map and/or |
| 562 | /// potentially macro expands it or turns it into a named token (like 'for'). |
Chris Lattner | 6a170eb | 2009-01-21 07:43:11 +0000 | [diff] [blame] | 563 | /// |
| 564 | /// Note that callers of this method are guarded by checking the |
| 565 | /// IdentifierInfo's 'isHandleIdentifierCase' bit. If this method changes, the |
| 566 | /// IdentifierInfo methods that compute these properties will need to change to |
| 567 | /// match. |
Chris Lattner | d217773 | 2007-07-20 16:59:19 +0000 | [diff] [blame] | 568 | void Preprocessor::HandleIdentifier(Token &Identifier) { |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 569 | assert(Identifier.getIdentifierInfo() && |
| 570 | "Can't handle identifiers without identifier info!"); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 571 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 572 | IdentifierInfo &II = *Identifier.getIdentifierInfo(); |
| 573 | |
| 574 | // If this identifier was poisoned, and if it was not produced from a macro |
| 575 | // expansion, emit an error. |
Ted Kremenek | 1a53157 | 2008-11-19 22:43:49 +0000 | [diff] [blame] | 576 | if (II.isPoisoned() && CurPPLexer) { |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 577 | if (&II != Ident__VA_ARGS__) // We warn about __VA_ARGS__ with poisoning. |
| 578 | Diag(Identifier, diag::err_pp_used_poisoned_id); |
| 579 | else |
| 580 | Diag(Identifier, diag::ext_pp_bad_vaargs_use); |
| 581 | } |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 582 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 583 | // If this is a macro to be expanded, do it. |
Chris Lattner | cc1a875 | 2007-10-07 08:44:20 +0000 | [diff] [blame] | 584 | if (MacroInfo *MI = getMacroInfo(&II)) { |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 585 | if (!DisableMacroExpansion && !Identifier.isExpandDisabled()) { |
| 586 | if (MI->isEnabled()) { |
| 587 | if (!HandleMacroExpandedIdentifier(Identifier, MI)) |
| 588 | return; |
| 589 | } else { |
| 590 | // C99 6.10.3.4p2 says that a disabled macro may never again be |
| 591 | // expanded, even if it's in a context where it could be expanded in the |
| 592 | // future. |
Chris Lattner | d217773 | 2007-07-20 16:59:19 +0000 | [diff] [blame] | 593 | Identifier.setFlag(Token::DisableExpand); |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 594 | } |
| 595 | } |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 596 | } |
| 597 | |
| 598 | // C++ 2.11p2: If this is an alternative representation of a C++ operator, |
| 599 | // then we act as if it is the actual operator and not the textual |
| 600 | // representation of it. |
| 601 | if (II.isCPlusPlusOperatorKeyword()) |
| 602 | Identifier.setIdentifierInfo(0); |
| 603 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 604 | // If this is an extension token, diagnose its use. |
Steve Naroff | b4eaf9c | 2008-09-02 18:50:17 +0000 | [diff] [blame] | 605 | // We avoid diagnosing tokens that originate from macro definitions. |
Eli Friedman | 2962f4d | 2009-04-28 03:59:15 +0000 | [diff] [blame] | 606 | // FIXME: This warning is disabled in cases where it shouldn't be, |
| 607 | // like "#define TY typeof", "TY(1) x". |
| 608 | if (II.isExtensionToken() && !DisableMacroExpansion) |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 609 | Diag(Identifier, diag::ext_token_used); |
| 610 | } |
Douglas Gregor | 2e22253 | 2009-07-02 17:08:52 +0000 | [diff] [blame] | 611 | |
| 612 | void Preprocessor::AddCommentHandler(CommentHandler *Handler) { |
| 613 | assert(Handler && "NULL comment handler"); |
| 614 | assert(std::find(CommentHandlers.begin(), CommentHandlers.end(), Handler) == |
| 615 | CommentHandlers.end() && "Comment handler already registered"); |
| 616 | CommentHandlers.push_back(Handler); |
| 617 | } |
| 618 | |
| 619 | void Preprocessor::RemoveCommentHandler(CommentHandler *Handler) { |
| 620 | std::vector<CommentHandler *>::iterator Pos |
| 621 | = std::find(CommentHandlers.begin(), CommentHandlers.end(), Handler); |
| 622 | assert(Pos != CommentHandlers.end() && "Comment handler not registered"); |
| 623 | CommentHandlers.erase(Pos); |
| 624 | } |
| 625 | |
Chris Lattner | 046c227 | 2010-01-18 22:35:47 +0000 | [diff] [blame] | 626 | bool Preprocessor::HandleComment(Token &result, SourceRange Comment) { |
| 627 | bool AnyPendingTokens = false; |
Douglas Gregor | 2e22253 | 2009-07-02 17:08:52 +0000 | [diff] [blame] | 628 | for (std::vector<CommentHandler *>::iterator H = CommentHandlers.begin(), |
| 629 | HEnd = CommentHandlers.end(); |
Chris Lattner | 046c227 | 2010-01-18 22:35:47 +0000 | [diff] [blame] | 630 | H != HEnd; ++H) { |
| 631 | if ((*H)->HandleComment(*this, Comment)) |
| 632 | AnyPendingTokens = true; |
| 633 | } |
| 634 | if (!AnyPendingTokens || getCommentRetentionState()) |
| 635 | return false; |
| 636 | Lex(result); |
| 637 | return true; |
Douglas Gregor | 2e22253 | 2009-07-02 17:08:52 +0000 | [diff] [blame] | 638 | } |
| 639 | |
| 640 | CommentHandler::~CommentHandler() { } |
Douglas Gregor | 94dc8f6 | 2010-03-19 16:15:56 +0000 | [diff] [blame] | 641 | |
Douglas Gregor | 94dc8f6 | 2010-03-19 16:15:56 +0000 | [diff] [blame] | 642 | void Preprocessor::createPreprocessingRecord() { |
| 643 | if (Record) |
| 644 | return; |
| 645 | |
Douglas Gregor | b9e1b75 | 2010-03-19 17:12:43 +0000 | [diff] [blame] | 646 | Record = new PreprocessingRecord; |
| 647 | addPPCallbacks(Record); |
Douglas Gregor | 94dc8f6 | 2010-03-19 16:15:56 +0000 | [diff] [blame] | 648 | } |