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" |
| 29 | #include "clang/Lex/HeaderSearch.h" |
| 30 | #include "clang/Lex/MacroInfo.h" |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 31 | #include "clang/Lex/Pragma.h" |
| 32 | #include "clang/Lex/ScratchBuffer.h" |
Chris Lattner | 500d329 | 2009-01-29 05:15:15 +0000 | [diff] [blame] | 33 | #include "clang/Lex/LexDiagnostic.h" |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 34 | #include "clang/Basic/SourceManager.h" |
Ted Kremenek | 337edcd | 2009-02-12 03:26:59 +0000 | [diff] [blame] | 35 | #include "clang/Basic/FileManager.h" |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 36 | #include "clang/Basic/TargetInfo.h" |
Chris Lattner | 2db78dd | 2008-10-05 20:40:30 +0000 | [diff] [blame] | 37 | #include "llvm/ADT/APFloat.h" |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 38 | #include "llvm/ADT/SmallVector.h" |
Chris Lattner | 97ba77c | 2007-07-16 06:48:38 +0000 | [diff] [blame] | 39 | #include "llvm/Support/MemoryBuffer.h" |
Benjamin Kramer | 6cb7c1a | 2009-08-23 12:08:50 +0000 | [diff] [blame] | 40 | #include "llvm/Support/raw_ostream.h" |
Chris Lattner | 3daed52 | 2009-03-02 22:20:04 +0000 | [diff] [blame] | 41 | #include <cstdio> |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 42 | using namespace clang; |
| 43 | |
| 44 | //===----------------------------------------------------------------------===// |
| 45 | |
Ted Kremenek | ec6c574 | 2008-04-17 21:23:07 +0000 | [diff] [blame] | 46 | PreprocessorFactory::~PreprocessorFactory() {} |
| 47 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 48 | Preprocessor::Preprocessor(Diagnostic &diags, const LangOptions &opts, |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 49 | TargetInfo &target, SourceManager &SM, |
Ted Kremenek | 72b1b15 | 2009-01-15 18:47:46 +0000 | [diff] [blame] | 50 | HeaderSearch &Headers, |
| 51 | IdentifierInfoLookup* IILookup) |
Chris Lattner | 836040f | 2009-03-13 21:17:43 +0000 | [diff] [blame] | 52 | : Diags(&diags), Features(opts), Target(target),FileMgr(Headers.getFileMgr()), |
Ted Kremenek | 72b1b15 | 2009-01-15 18:47:46 +0000 | [diff] [blame] | 53 | SourceMgr(SM), HeaderInfo(Headers), Identifiers(opts, IILookup), |
Chris Lattner | 030e8fe | 2009-06-16 16:18:48 +0000 | [diff] [blame] | 54 | BuiltinInfo(Target), CurPPLexer(0), CurDirLookup(0), Callbacks(0) { |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 55 | ScratchBuf = new ScratchBuffer(SourceMgr); |
Chris Lattner | c1f9d82 | 2009-04-13 01:29:17 +0000 | [diff] [blame] | 56 | CounterValue = 0; // __COUNTER__ starts at 0. |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 57 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 58 | // Clear stats. |
| 59 | NumDirectives = NumDefined = NumUndefined = NumPragma = 0; |
| 60 | NumIf = NumElse = NumEndif = 0; |
| 61 | NumEnteredSourceFiles = 0; |
| 62 | NumMacroExpanded = NumFnMacroExpanded = NumBuiltinMacroExpanded = 0; |
| 63 | NumFastMacroExpanded = NumTokenPaste = NumFastTokenPaste = 0; |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 64 | MaxIncludeStackDepth = 0; |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 65 | NumSkipped = 0; |
| 66 | |
| 67 | // Default to discarding comments. |
| 68 | KeepComments = false; |
| 69 | KeepMacroComments = false; |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 70 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 71 | // Macro expansion is enabled. |
| 72 | DisableMacroExpansion = false; |
| 73 | InMacroArgs = false; |
Chris Lattner | 6cfe759 | 2008-03-09 02:26:03 +0000 | [diff] [blame] | 74 | NumCachedTokenLexers = 0; |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 75 | |
Argyrios Kyrtzidis | 03db1b3 | 2008-08-10 13:15:22 +0000 | [diff] [blame] | 76 | CachedLexPos = 0; |
| 77 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 78 | // "Poison" __VA_ARGS__, which can only appear in the expansion of a macro. |
| 79 | // This gets unpoisoned where it is allowed. |
| 80 | (Ident__VA_ARGS__ = getIdentifierInfo("__VA_ARGS__"))->setIsPoisoned(); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 81 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 82 | // Initialize the pragma handlers. |
| 83 | PragmaHandlers = new PragmaNamespace(0); |
| 84 | RegisterBuiltinPragmas(); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 85 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 86 | // Initialize builtin macros like __LINE__ and friends. |
| 87 | RegisterBuiltinMacros(); |
| 88 | } |
| 89 | |
| 90 | Preprocessor::~Preprocessor() { |
Argyrios Kyrtzidis | 2174a4f | 2008-08-23 12:12:06 +0000 | [diff] [blame] | 91 | assert(BacktrackPositions.empty() && "EnableBacktrack/Backtrack imbalance!"); |
| 92 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 93 | while (!IncludeMacroStack.empty()) { |
| 94 | delete IncludeMacroStack.back().TheLexer; |
Chris Lattner | 6cfe759 | 2008-03-09 02:26:03 +0000 | [diff] [blame] | 95 | delete IncludeMacroStack.back().TheTokenLexer; |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 96 | IncludeMacroStack.pop_back(); |
| 97 | } |
Chris Lattner | cc1a875 | 2007-10-07 08:44:20 +0000 | [diff] [blame] | 98 | |
| 99 | // Free any macro definitions. |
| 100 | for (llvm::DenseMap<IdentifierInfo*, MacroInfo*>::iterator I = |
| 101 | Macros.begin(), E = Macros.end(); I != E; ++I) { |
Ted Kremenek | 0ea7672 | 2008-12-15 19:56:42 +0000 | [diff] [blame] | 102 | // We don't need to free the MacroInfo objects directly. These |
| 103 | // will be released when the BumpPtrAllocator 'BP' object gets |
Ted Kremenek | 9ee7d45 | 2009-01-19 07:45:44 +0000 | [diff] [blame] | 104 | // destroyed. We still need to run the dstor, however, to free |
| 105 | // memory alocated by MacroInfo. |
Chris Lattner | 685befe | 2009-02-20 22:46:43 +0000 | [diff] [blame] | 106 | I->second->Destroy(BP); |
Chris Lattner | cc1a875 | 2007-10-07 08:44:20 +0000 | [diff] [blame] | 107 | I->first->setHasMacroDefinition(false); |
| 108 | } |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 109 | |
Chris Lattner | 9594acf | 2007-07-15 00:25:26 +0000 | [diff] [blame] | 110 | // Free any cached macro expanders. |
Chris Lattner | 6cfe759 | 2008-03-09 02:26:03 +0000 | [diff] [blame] | 111 | for (unsigned i = 0, e = NumCachedTokenLexers; i != e; ++i) |
| 112 | delete TokenLexerCache[i]; |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 113 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 114 | // Release pragma information. |
| 115 | delete PragmaHandlers; |
| 116 | |
| 117 | // Delete the scratch buffer info. |
| 118 | delete ScratchBuf; |
Chris Lattner | eb50ed8 | 2008-03-14 06:07:05 +0000 | [diff] [blame] | 119 | |
| 120 | delete Callbacks; |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 121 | } |
| 122 | |
Ted Kremenek | 337edcd | 2009-02-12 03:26:59 +0000 | [diff] [blame] | 123 | void Preprocessor::setPTHManager(PTHManager* pm) { |
| 124 | PTH.reset(pm); |
Douglas Gregor | 52e7108 | 2009-10-16 18:18:30 +0000 | [diff] [blame] | 125 | FileMgr.addStatCache(PTH->createStatCache()); |
Ted Kremenek | 337edcd | 2009-02-12 03:26:59 +0000 | [diff] [blame] | 126 | } |
| 127 | |
Chris Lattner | d217773 | 2007-07-20 16:59:19 +0000 | [diff] [blame] | 128 | void Preprocessor::DumpToken(const Token &Tok, bool DumpFlags) const { |
Benjamin Kramer | 6cb7c1a | 2009-08-23 12:08:50 +0000 | [diff] [blame] | 129 | llvm::errs() << tok::getTokenName(Tok.getKind()) << " '" |
| 130 | << getSpelling(Tok) << "'"; |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 131 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 132 | if (!DumpFlags) return; |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 133 | |
Benjamin Kramer | 6cb7c1a | 2009-08-23 12:08:50 +0000 | [diff] [blame] | 134 | llvm::errs() << "\t"; |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 135 | if (Tok.isAtStartOfLine()) |
Benjamin Kramer | 6cb7c1a | 2009-08-23 12:08:50 +0000 | [diff] [blame] | 136 | llvm::errs() << " [StartOfLine]"; |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 137 | if (Tok.hasLeadingSpace()) |
Benjamin Kramer | 6cb7c1a | 2009-08-23 12:08:50 +0000 | [diff] [blame] | 138 | llvm::errs() << " [LeadingSpace]"; |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 139 | if (Tok.isExpandDisabled()) |
Benjamin Kramer | 6cb7c1a | 2009-08-23 12:08:50 +0000 | [diff] [blame] | 140 | llvm::errs() << " [ExpandDisabled]"; |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 141 | if (Tok.needsCleaning()) { |
| 142 | const char *Start = SourceMgr.getCharacterData(Tok.getLocation()); |
Benjamin Kramer | 6cb7c1a | 2009-08-23 12:08:50 +0000 | [diff] [blame] | 143 | llvm::errs() << " [UnClean='" << std::string(Start, Start+Tok.getLength()) |
| 144 | << "']"; |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 145 | } |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 146 | |
Benjamin Kramer | 6cb7c1a | 2009-08-23 12:08:50 +0000 | [diff] [blame] | 147 | llvm::errs() << "\tLoc=<"; |
Chris Lattner | c3d8d57 | 2007-12-09 20:31:55 +0000 | [diff] [blame] | 148 | DumpLocation(Tok.getLocation()); |
Benjamin Kramer | 6cb7c1a | 2009-08-23 12:08:50 +0000 | [diff] [blame] | 149 | llvm::errs() << ">"; |
Chris Lattner | c3d8d57 | 2007-12-09 20:31:55 +0000 | [diff] [blame] | 150 | } |
| 151 | |
| 152 | void Preprocessor::DumpLocation(SourceLocation Loc) const { |
Chris Lattner | b9c3f96 | 2009-01-27 07:57:44 +0000 | [diff] [blame] | 153 | Loc.dump(SourceMgr); |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 154 | } |
| 155 | |
| 156 | void Preprocessor::DumpMacro(const MacroInfo &MI) const { |
Benjamin Kramer | 6cb7c1a | 2009-08-23 12:08:50 +0000 | [diff] [blame] | 157 | llvm::errs() << "MACRO: "; |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 158 | for (unsigned i = 0, e = MI.getNumTokens(); i != e; ++i) { |
| 159 | DumpToken(MI.getReplacementToken(i)); |
Benjamin Kramer | 6cb7c1a | 2009-08-23 12:08:50 +0000 | [diff] [blame] | 160 | llvm::errs() << " "; |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 161 | } |
Benjamin Kramer | 6cb7c1a | 2009-08-23 12:08:50 +0000 | [diff] [blame] | 162 | llvm::errs() << "\n"; |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 163 | } |
| 164 | |
| 165 | void Preprocessor::PrintStats() { |
Benjamin Kramer | 6cb7c1a | 2009-08-23 12:08:50 +0000 | [diff] [blame] | 166 | llvm::errs() << "\n*** Preprocessor Stats:\n"; |
| 167 | llvm::errs() << NumDirectives << " directives found:\n"; |
| 168 | llvm::errs() << " " << NumDefined << " #define.\n"; |
| 169 | llvm::errs() << " " << NumUndefined << " #undef.\n"; |
| 170 | llvm::errs() << " #include/#include_next/#import:\n"; |
| 171 | llvm::errs() << " " << NumEnteredSourceFiles << " source files entered.\n"; |
| 172 | llvm::errs() << " " << MaxIncludeStackDepth << " max include stack depth\n"; |
| 173 | llvm::errs() << " " << NumIf << " #if/#ifndef/#ifdef.\n"; |
| 174 | llvm::errs() << " " << NumElse << " #else/#elif.\n"; |
| 175 | llvm::errs() << " " << NumEndif << " #endif.\n"; |
| 176 | llvm::errs() << " " << NumPragma << " #pragma.\n"; |
| 177 | llvm::errs() << NumSkipped << " #if/#ifndef#ifdef regions skipped\n"; |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 178 | |
Benjamin Kramer | 6cb7c1a | 2009-08-23 12:08:50 +0000 | [diff] [blame] | 179 | llvm::errs() << NumMacroExpanded << "/" << NumFnMacroExpanded << "/" |
Ted Kremenek | bdd30c2 | 2008-01-14 16:44:48 +0000 | [diff] [blame] | 180 | << NumBuiltinMacroExpanded << " obj/fn/builtin macros expanded, " |
| 181 | << NumFastMacroExpanded << " on the fast path.\n"; |
Benjamin Kramer | 6cb7c1a | 2009-08-23 12:08:50 +0000 | [diff] [blame] | 182 | llvm::errs() << (NumFastTokenPaste+NumTokenPaste) |
Ted Kremenek | bdd30c2 | 2008-01-14 16:44:48 +0000 | [diff] [blame] | 183 | << " token paste (##) operations performed, " |
| 184 | << NumFastTokenPaste << " on the fast path.\n"; |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 185 | } |
| 186 | |
| 187 | //===----------------------------------------------------------------------===// |
| 188 | // Token Spelling |
| 189 | //===----------------------------------------------------------------------===// |
| 190 | |
| 191 | |
| 192 | /// getSpelling() - Return the 'spelling' of this token. The spelling of a |
| 193 | /// token are the characters used to represent the token in the source file |
| 194 | /// after trigraph expansion and escaped-newline folding. In particular, this |
| 195 | /// wants to get the true, uncanonicalized, spelling of things like digraphs |
| 196 | /// UCNs, etc. |
Chris Lattner | d217773 | 2007-07-20 16:59:19 +0000 | [diff] [blame] | 197 | std::string Preprocessor::getSpelling(const Token &Tok) const { |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 198 | assert((int)Tok.getLength() >= 0 && "Token character range is bogus!"); |
Ted Kremenek | 277faca | 2009-01-27 00:01:05 +0000 | [diff] [blame] | 199 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 200 | // If this token contains nothing interesting, return it directly. |
Ted Kremenek | 277faca | 2009-01-27 00:01:05 +0000 | [diff] [blame] | 201 | const char* TokStart = SourceMgr.getCharacterData(Tok.getLocation()); |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 202 | if (!Tok.needsCleaning()) |
| 203 | return std::string(TokStart, TokStart+Tok.getLength()); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 204 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 205 | std::string Result; |
| 206 | Result.reserve(Tok.getLength()); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 207 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 208 | // Otherwise, hard case, relex the characters into the string. |
| 209 | for (const char *Ptr = TokStart, *End = TokStart+Tok.getLength(); |
| 210 | Ptr != End; ) { |
| 211 | unsigned CharSize; |
| 212 | Result.push_back(Lexer::getCharAndSizeNoWarn(Ptr, CharSize, Features)); |
| 213 | Ptr += CharSize; |
| 214 | } |
| 215 | assert(Result.size() != unsigned(Tok.getLength()) && |
| 216 | "NeedsCleaning flag set on something that didn't need cleaning!"); |
| 217 | return Result; |
| 218 | } |
| 219 | |
| 220 | /// getSpelling - This method is used to get the spelling of a token into a |
| 221 | /// preallocated buffer, instead of as an std::string. The caller is required |
| 222 | /// to allocate enough space for the token, which is guaranteed to be at least |
| 223 | /// Tok.getLength() bytes long. The actual length of the token is returned. |
| 224 | /// |
| 225 | /// Note that this method may do two possible things: it may either fill in |
| 226 | /// the buffer specified with characters, or it may *change the input pointer* |
| 227 | /// to point to a constant buffer with the data already in it (avoiding a |
| 228 | /// copy). The caller is not allowed to modify the returned buffer pointer |
| 229 | /// if an internal buffer is returned. |
Chris Lattner | d217773 | 2007-07-20 16:59:19 +0000 | [diff] [blame] | 230 | unsigned Preprocessor::getSpelling(const Token &Tok, |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 231 | const char *&Buffer) const { |
| 232 | assert((int)Tok.getLength() >= 0 && "Token character range is bogus!"); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 233 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 234 | // If this token is an identifier, just return the string from the identifier |
| 235 | // table, which is very quick. |
| 236 | if (const IdentifierInfo *II = Tok.getIdentifierInfo()) { |
Daniel Dunbar | e013d68 | 2009-10-18 20:26:12 +0000 | [diff] [blame] | 237 | Buffer = II->getNameStart(); |
Chris Lattner | e1dccae | 2009-01-05 19:44:41 +0000 | [diff] [blame] | 238 | return II->getLength(); |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 239 | } |
Ted Kremenek | b70e3da | 2009-01-08 02:47:16 +0000 | [diff] [blame] | 240 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 241 | // Otherwise, compute the start of the token in the input lexer buffer. |
Chris Lattner | 47246be | 2009-01-26 19:29:26 +0000 | [diff] [blame] | 242 | const char *TokStart = 0; |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 243 | |
Chris Lattner | 47246be | 2009-01-26 19:29:26 +0000 | [diff] [blame] | 244 | if (Tok.isLiteral()) |
| 245 | TokStart = Tok.getLiteralData(); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 246 | |
Chris Lattner | 47246be | 2009-01-26 19:29:26 +0000 | [diff] [blame] | 247 | if (TokStart == 0) |
| 248 | TokStart = SourceMgr.getCharacterData(Tok.getLocation()); |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 249 | |
| 250 | // If this token contains nothing interesting, return it directly. |
| 251 | if (!Tok.needsCleaning()) { |
| 252 | Buffer = TokStart; |
| 253 | return Tok.getLength(); |
| 254 | } |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 255 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 256 | // Otherwise, hard case, relex the characters into the string. |
| 257 | char *OutBuf = const_cast<char*>(Buffer); |
| 258 | for (const char *Ptr = TokStart, *End = TokStart+Tok.getLength(); |
| 259 | Ptr != End; ) { |
| 260 | unsigned CharSize; |
| 261 | *OutBuf++ = Lexer::getCharAndSizeNoWarn(Ptr, CharSize, Features); |
| 262 | Ptr += CharSize; |
| 263 | } |
| 264 | assert(unsigned(OutBuf-Buffer) != Tok.getLength() && |
| 265 | "NeedsCleaning flag set on something that didn't need cleaning!"); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 266 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 267 | return OutBuf-Buffer; |
| 268 | } |
| 269 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 270 | /// CreateString - Plop the specified string into a scratch buffer and return a |
| 271 | /// location for it. If specified, the source location provides a source |
| 272 | /// location for the token. |
Chris Lattner | 47246be | 2009-01-26 19:29:26 +0000 | [diff] [blame] | 273 | void Preprocessor::CreateString(const char *Buf, unsigned Len, Token &Tok, |
| 274 | SourceLocation InstantiationLoc) { |
| 275 | Tok.setLength(Len); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 276 | |
Chris Lattner | 47246be | 2009-01-26 19:29:26 +0000 | [diff] [blame] | 277 | const char *DestPtr; |
| 278 | SourceLocation Loc = ScratchBuf->getToken(Buf, Len, DestPtr); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 279 | |
Chris Lattner | 47246be | 2009-01-26 19:29:26 +0000 | [diff] [blame] | 280 | if (InstantiationLoc.isValid()) |
Chris Lattner | e7fb484 | 2009-02-15 20:52:18 +0000 | [diff] [blame] | 281 | Loc = SourceMgr.createInstantiationLoc(Loc, InstantiationLoc, |
| 282 | InstantiationLoc, Len); |
Chris Lattner | 47246be | 2009-01-26 19:29:26 +0000 | [diff] [blame] | 283 | Tok.setLocation(Loc); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 284 | |
Chris Lattner | 47246be | 2009-01-26 19:29:26 +0000 | [diff] [blame] | 285 | // If this is a literal token, set the pointer data. |
| 286 | if (Tok.isLiteral()) |
| 287 | Tok.setLiteralData(DestPtr); |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 288 | } |
| 289 | |
| 290 | |
Chris Lattner | 97ba77c | 2007-07-16 06:48:38 +0000 | [diff] [blame] | 291 | /// AdvanceToTokenCharacter - Given a location that specifies the start of a |
| 292 | /// token, return a new location that specifies a character within the token. |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 293 | SourceLocation Preprocessor::AdvanceToTokenCharacter(SourceLocation TokStart, |
Chris Lattner | 97ba77c | 2007-07-16 06:48:38 +0000 | [diff] [blame] | 294 | unsigned CharNo) { |
Chris Lattner | f7cf85b | 2009-01-16 07:36:28 +0000 | [diff] [blame] | 295 | // Figure out how many physical characters away the specified instantiation |
Chris Lattner | 97ba77c | 2007-07-16 06:48:38 +0000 | [diff] [blame] | 296 | // character is. This needs to take into consideration newlines and |
| 297 | // trigraphs. |
Chris Lattner | 9dc1f53 | 2007-07-20 16:37:10 +0000 | [diff] [blame] | 298 | const char *TokPtr = SourceMgr.getCharacterData(TokStart); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 299 | |
Chris Lattner | 88e2524 | 2009-04-18 22:28:58 +0000 | [diff] [blame] | 300 | // If they request the first char of the token, we're trivially done. |
| 301 | if (CharNo == 0 && Lexer::isObviouslySimpleCharacter(*TokPtr)) |
| 302 | return TokStart; |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 303 | |
Chris Lattner | 9dc1f53 | 2007-07-20 16:37:10 +0000 | [diff] [blame] | 304 | unsigned PhysOffset = 0; |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 305 | |
Chris Lattner | 97ba77c | 2007-07-16 06:48:38 +0000 | [diff] [blame] | 306 | // The usual case is that tokens don't contain anything interesting. Skip |
| 307 | // over the uninteresting characters. If a token only consists of simple |
| 308 | // chars, this method is extremely fast. |
Chris Lattner | 88e2524 | 2009-04-18 22:28:58 +0000 | [diff] [blame] | 309 | while (Lexer::isObviouslySimpleCharacter(*TokPtr)) { |
| 310 | if (CharNo == 0) |
| 311 | return TokStart.getFileLocWithOffset(PhysOffset); |
Chris Lattner | 9dc1f53 | 2007-07-20 16:37:10 +0000 | [diff] [blame] | 312 | ++TokPtr, --CharNo, ++PhysOffset; |
Chris Lattner | 88e2524 | 2009-04-18 22:28:58 +0000 | [diff] [blame] | 313 | } |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 314 | |
Chris Lattner | 28c90ad | 2009-01-17 07:57:25 +0000 | [diff] [blame] | 315 | // 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] | 316 | // lexer to parse it correctly. |
Chris Lattner | 88e2524 | 2009-04-18 22:28:58 +0000 | [diff] [blame] | 317 | for (; CharNo; --CharNo) { |
| 318 | unsigned Size; |
| 319 | Lexer::getCharAndSizeNoWarn(TokPtr, Size, Features); |
| 320 | TokPtr += Size; |
| 321 | PhysOffset += Size; |
Chris Lattner | 97ba77c | 2007-07-16 06:48:38 +0000 | [diff] [blame] | 322 | } |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 323 | |
Chris Lattner | 88e2524 | 2009-04-18 22:28:58 +0000 | [diff] [blame] | 324 | // Final detail: if we end up on an escaped newline, we want to return the |
| 325 | // location of the actual byte of the token. For example foo\<newline>bar |
| 326 | // advanced by 3 should return the location of b, not of \\. One compounding |
| 327 | // detail of this is that the escape may be made by a trigraph. |
| 328 | if (!Lexer::isObviouslySimpleCharacter(*TokPtr)) |
| 329 | PhysOffset = Lexer::SkipEscapedNewLines(TokPtr)-TokPtr; |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 330 | |
Chris Lattner | 9dc1f53 | 2007-07-20 16:37:10 +0000 | [diff] [blame] | 331 | return TokStart.getFileLocWithOffset(PhysOffset); |
Chris Lattner | 97ba77c | 2007-07-16 06:48:38 +0000 | [diff] [blame] | 332 | } |
| 333 | |
Douglas Gregor | b2fb6de | 2009-02-27 17:53:17 +0000 | [diff] [blame] | 334 | /// \brief Computes the source location just past the end of the |
| 335 | /// token at this source location. |
| 336 | /// |
| 337 | /// This routine can be used to produce a source location that |
| 338 | /// points just past the end of the token referenced by \p Loc, and |
| 339 | /// is generally used when a diagnostic needs to point just after a |
| 340 | /// token where it expected something different that it received. If |
| 341 | /// the returned source location would not be meaningful (e.g., if |
| 342 | /// it points into a macro), this routine returns an invalid |
| 343 | /// source location. |
| 344 | SourceLocation Preprocessor::getLocForEndOfToken(SourceLocation Loc) { |
| 345 | if (Loc.isInvalid() || !Loc.isFileID()) |
| 346 | return SourceLocation(); |
| 347 | |
Chris Lattner | 2c78b87 | 2009-04-14 23:22:57 +0000 | [diff] [blame] | 348 | unsigned Len = Lexer::MeasureTokenLength(Loc, getSourceManager(), Features); |
Douglas Gregor | b2fb6de | 2009-02-27 17:53:17 +0000 | [diff] [blame] | 349 | return AdvanceToTokenCharacter(Loc, Len); |
| 350 | } |
| 351 | |
| 352 | |
Chris Lattner | 97ba77c | 2007-07-16 06:48:38 +0000 | [diff] [blame] | 353 | |
Chris Lattner | 53b0dab | 2007-10-09 22:10:18 +0000 | [diff] [blame] | 354 | //===----------------------------------------------------------------------===// |
| 355 | // Preprocessor Initialization Methods |
| 356 | //===----------------------------------------------------------------------===// |
| 357 | |
Chris Lattner | 53b0dab | 2007-10-09 22:10:18 +0000 | [diff] [blame] | 358 | |
| 359 | /// EnterMainSourceFile - Enter the specified FileID as the main source file, |
Nate Begeman | 6b61602 | 2008-01-07 04:01:26 +0000 | [diff] [blame] | 360 | /// which implicitly adds the builtin defines etc. |
Ted Kremenek | 95041a2 | 2007-12-19 22:51:13 +0000 | [diff] [blame] | 361 | void Preprocessor::EnterMainSourceFile() { |
Chris Lattner | 05db427 | 2009-02-13 19:33:24 +0000 | [diff] [blame] | 362 | // We do not allow the preprocessor to reenter the main file. Doing so will |
| 363 | // cause FileID's to accumulate information from both runs (e.g. #line |
| 364 | // information) and predefined macros aren't guaranteed to be set properly. |
| 365 | assert(NumEnteredSourceFiles == 0 && "Cannot reenter the main file!"); |
Chris Lattner | 2b2453a | 2009-01-17 06:22:33 +0000 | [diff] [blame] | 366 | FileID MainFileID = SourceMgr.getMainFileID(); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 367 | |
Chris Lattner | 53b0dab | 2007-10-09 22:10:18 +0000 | [diff] [blame] | 368 | // Enter the main file source buffer. |
| 369 | EnterSourceFile(MainFileID, 0); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 370 | |
Chris Lattner | b283298 | 2007-11-15 19:07:47 +0000 | [diff] [blame] | 371 | // Tell the header info that the main file was entered. If the file is later |
| 372 | // #imported, it won't be re-entered. |
Chris Lattner | 2b2453a | 2009-01-17 06:22:33 +0000 | [diff] [blame] | 373 | if (const FileEntry *FE = SourceMgr.getFileEntryForID(MainFileID)) |
Chris Lattner | b283298 | 2007-11-15 19:07:47 +0000 | [diff] [blame] | 374 | HeaderInfo.IncrementIncludeCount(FE); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 375 | |
Chris Lattner | 53b0dab | 2007-10-09 22:10:18 +0000 | [diff] [blame] | 376 | std::vector<char> PrologFile; |
| 377 | PrologFile.reserve(4080); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 378 | |
Chris Lattner | 6bf0f4c | 2009-04-10 21:58:23 +0000 | [diff] [blame] | 379 | // FIXME: Don't make a copy. |
Chris Lattner | aa39197 | 2008-04-19 23:09:31 +0000 | [diff] [blame] | 380 | PrologFile.insert(PrologFile.end(), Predefines.begin(), Predefines.end()); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 381 | |
Chris Lattner | 53b0dab | 2007-10-09 22:10:18 +0000 | [diff] [blame] | 382 | // Memory buffer must end with a null byte! |
| 383 | PrologFile.push_back(0); |
| 384 | |
| 385 | // Now that we have emitted the predefined macros, #includes, etc into |
| 386 | // PrologFile, preprocess it to populate the initial preprocessor state. |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 387 | llvm::MemoryBuffer *SB = |
Chris Lattner | 53b0dab | 2007-10-09 22:10:18 +0000 | [diff] [blame] | 388 | llvm::MemoryBuffer::getMemBufferCopy(&PrologFile.front(),&PrologFile.back(), |
Chris Lattner | f2e880a | 2009-03-20 20:16:10 +0000 | [diff] [blame] | 389 | "<built-in>"); |
Chris Lattner | 53b0dab | 2007-10-09 22:10:18 +0000 | [diff] [blame] | 390 | assert(SB && "Cannot fail to create predefined source buffer"); |
Chris Lattner | 2b2453a | 2009-01-17 06:22:33 +0000 | [diff] [blame] | 391 | FileID FID = SourceMgr.createFileIDForMemBuffer(SB); |
| 392 | assert(!FID.isInvalid() && "Could not create FileID for predefines?"); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 393 | |
Chris Lattner | 53b0dab | 2007-10-09 22:10:18 +0000 | [diff] [blame] | 394 | // Start parsing the predefines. |
Chris Lattner | 2b2453a | 2009-01-17 06:22:33 +0000 | [diff] [blame] | 395 | EnterSourceFile(FID, 0); |
Chris Lattner | 53b0dab | 2007-10-09 22:10:18 +0000 | [diff] [blame] | 396 | } |
Chris Lattner | 97ba77c | 2007-07-16 06:48:38 +0000 | [diff] [blame] | 397 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 398 | |
| 399 | //===----------------------------------------------------------------------===// |
| 400 | // Lexer Event Handling. |
| 401 | //===----------------------------------------------------------------------===// |
| 402 | |
| 403 | /// LookUpIdentifierInfo - Given a tok::identifier token, look up the |
| 404 | /// identifier information for the token and install it into the token. |
Chris Lattner | d217773 | 2007-07-20 16:59:19 +0000 | [diff] [blame] | 405 | IdentifierInfo *Preprocessor::LookUpIdentifierInfo(Token &Identifier, |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 406 | const char *BufPtr) { |
Chris Lattner | 22f6bbc | 2007-10-09 18:02:16 +0000 | [diff] [blame] | 407 | assert(Identifier.is(tok::identifier) && "Not an identifier!"); |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 408 | assert(Identifier.getIdentifierInfo() == 0 && "Identinfo already exists!"); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 409 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 410 | // Look up this token, see if it is a macro, or if it is a language keyword. |
| 411 | IdentifierInfo *II; |
| 412 | if (BufPtr && !Identifier.needsCleaning()) { |
| 413 | // No cleaning needed, just use the characters from the lexed buffer. |
| 414 | II = getIdentifierInfo(BufPtr, BufPtr+Identifier.getLength()); |
| 415 | } else { |
| 416 | // Cleaning needed, alloca a buffer, clean into it, then use the buffer. |
Chris Lattner | c35717a | 2007-07-13 17:10:38 +0000 | [diff] [blame] | 417 | llvm::SmallVector<char, 64> IdentifierBuffer; |
| 418 | IdentifierBuffer.resize(Identifier.getLength()); |
| 419 | const char *TmpBuf = &IdentifierBuffer[0]; |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 420 | unsigned Size = getSpelling(Identifier, TmpBuf); |
| 421 | II = getIdentifierInfo(TmpBuf, TmpBuf+Size); |
| 422 | } |
| 423 | Identifier.setIdentifierInfo(II); |
| 424 | return II; |
| 425 | } |
| 426 | |
| 427 | |
| 428 | /// HandleIdentifier - This callback is invoked when the lexer reads an |
| 429 | /// identifier. This callback looks up the identifier in the map and/or |
| 430 | /// 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] | 431 | /// |
| 432 | /// Note that callers of this method are guarded by checking the |
| 433 | /// IdentifierInfo's 'isHandleIdentifierCase' bit. If this method changes, the |
| 434 | /// IdentifierInfo methods that compute these properties will need to change to |
| 435 | /// match. |
Chris Lattner | d217773 | 2007-07-20 16:59:19 +0000 | [diff] [blame] | 436 | void Preprocessor::HandleIdentifier(Token &Identifier) { |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 437 | assert(Identifier.getIdentifierInfo() && |
| 438 | "Can't handle identifiers without identifier info!"); |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 439 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 440 | IdentifierInfo &II = *Identifier.getIdentifierInfo(); |
| 441 | |
| 442 | // If this identifier was poisoned, and if it was not produced from a macro |
| 443 | // expansion, emit an error. |
Ted Kremenek | 1a53157 | 2008-11-19 22:43:49 +0000 | [diff] [blame] | 444 | if (II.isPoisoned() && CurPPLexer) { |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 445 | if (&II != Ident__VA_ARGS__) // We warn about __VA_ARGS__ with poisoning. |
| 446 | Diag(Identifier, diag::err_pp_used_poisoned_id); |
| 447 | else |
| 448 | Diag(Identifier, diag::ext_pp_bad_vaargs_use); |
| 449 | } |
Mike Stump | 1eb4433 | 2009-09-09 15:08:12 +0000 | [diff] [blame] | 450 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 451 | // If this is a macro to be expanded, do it. |
Chris Lattner | cc1a875 | 2007-10-07 08:44:20 +0000 | [diff] [blame] | 452 | if (MacroInfo *MI = getMacroInfo(&II)) { |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 453 | if (!DisableMacroExpansion && !Identifier.isExpandDisabled()) { |
| 454 | if (MI->isEnabled()) { |
| 455 | if (!HandleMacroExpandedIdentifier(Identifier, MI)) |
| 456 | return; |
| 457 | } else { |
| 458 | // C99 6.10.3.4p2 says that a disabled macro may never again be |
| 459 | // expanded, even if it's in a context where it could be expanded in the |
| 460 | // future. |
Chris Lattner | d217773 | 2007-07-20 16:59:19 +0000 | [diff] [blame] | 461 | Identifier.setFlag(Token::DisableExpand); |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 462 | } |
| 463 | } |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 464 | } |
| 465 | |
| 466 | // C++ 2.11p2: If this is an alternative representation of a C++ operator, |
| 467 | // then we act as if it is the actual operator and not the textual |
| 468 | // representation of it. |
| 469 | if (II.isCPlusPlusOperatorKeyword()) |
| 470 | Identifier.setIdentifierInfo(0); |
| 471 | |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 472 | // If this is an extension token, diagnose its use. |
Steve Naroff | b4eaf9c | 2008-09-02 18:50:17 +0000 | [diff] [blame] | 473 | // We avoid diagnosing tokens that originate from macro definitions. |
Eli Friedman | 2962f4d | 2009-04-28 03:59:15 +0000 | [diff] [blame] | 474 | // FIXME: This warning is disabled in cases where it shouldn't be, |
| 475 | // like "#define TY typeof", "TY(1) x". |
| 476 | if (II.isExtensionToken() && !DisableMacroExpansion) |
Reid Spencer | 5f016e2 | 2007-07-11 17:01:13 +0000 | [diff] [blame] | 477 | Diag(Identifier, diag::ext_token_used); |
| 478 | } |
Douglas Gregor | 2e22253 | 2009-07-02 17:08:52 +0000 | [diff] [blame] | 479 | |
| 480 | void Preprocessor::AddCommentHandler(CommentHandler *Handler) { |
| 481 | assert(Handler && "NULL comment handler"); |
| 482 | assert(std::find(CommentHandlers.begin(), CommentHandlers.end(), Handler) == |
| 483 | CommentHandlers.end() && "Comment handler already registered"); |
| 484 | CommentHandlers.push_back(Handler); |
| 485 | } |
| 486 | |
| 487 | void Preprocessor::RemoveCommentHandler(CommentHandler *Handler) { |
| 488 | std::vector<CommentHandler *>::iterator Pos |
| 489 | = std::find(CommentHandlers.begin(), CommentHandlers.end(), Handler); |
| 490 | assert(Pos != CommentHandlers.end() && "Comment handler not registered"); |
| 491 | CommentHandlers.erase(Pos); |
| 492 | } |
| 493 | |
| 494 | void Preprocessor::HandleComment(SourceRange Comment) { |
| 495 | for (std::vector<CommentHandler *>::iterator H = CommentHandlers.begin(), |
| 496 | HEnd = CommentHandlers.end(); |
| 497 | H != HEnd; ++H) |
| 498 | (*H)->HandleComment(*this, Comment); |
| 499 | } |
| 500 | |
| 501 | CommentHandler::~CommentHandler() { } |