Mercurial > hg > CbC > CbC_llvm
annotate clang/lib/Lex/Preprocessor.cpp @ 209:dd44ba33042e
merged...
| author | Shinji KONO <kono@ie.u-ryukyu.ac.jp> |
|---|---|
| date | Tue, 08 Jun 2021 06:36:09 +0900 |
| parents | 2e18cbf3894f b7591485f4cd |
| children | 50b3abffaea6 |
| rev | line source |
|---|---|
| 150 | 1 //===- Preprocessor.cpp - C Language Family Preprocessor Implementation ---===// |
| 2 // | |
| 3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. | |
| 4 // See https://llvm.org/LICENSE.txt for license information. | |
| 5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception | |
| 6 // | |
| 7 //===----------------------------------------------------------------------===// | |
| 8 // | |
| 9 // This file implements the Preprocessor interface. | |
| 10 // | |
| 11 //===----------------------------------------------------------------------===// | |
| 12 // | |
| 13 // Options to support: | |
| 14 // -H - Print the name of each header file used. | |
| 15 // -d[DNI] - Dump various things. | |
| 16 // -fworking-directory - #line's with preprocessor's working dir. | |
| 17 // -fpreprocessed | |
| 18 // -dependency-file,-M,-MM,-MF,-MG,-MP,-MT,-MQ,-MD,-MMD | |
| 19 // -W* | |
| 20 // -w | |
| 21 // | |
| 22 // Messages to emit: | |
| 23 // "Multiple include guards may be useful for:\n" | |
| 24 // | |
| 25 //===----------------------------------------------------------------------===// | |
| 26 | |
| 27 #include "clang/Lex/Preprocessor.h" | |
| 28 #include "clang/Basic/Builtins.h" | |
| 29 #include "clang/Basic/FileManager.h" | |
| 30 #include "clang/Basic/FileSystemStatCache.h" | |
| 31 #include "clang/Basic/IdentifierTable.h" | |
| 32 #include "clang/Basic/LLVM.h" | |
| 33 #include "clang/Basic/LangOptions.h" | |
| 34 #include "clang/Basic/Module.h" | |
| 35 #include "clang/Basic/SourceLocation.h" | |
| 36 #include "clang/Basic/SourceManager.h" | |
| 37 #include "clang/Basic/TargetInfo.h" | |
| 38 #include "clang/Lex/CodeCompletionHandler.h" | |
| 39 #include "clang/Lex/ExternalPreprocessorSource.h" | |
| 40 #include "clang/Lex/HeaderSearch.h" | |
| 41 #include "clang/Lex/LexDiagnostic.h" | |
| 42 #include "clang/Lex/Lexer.h" | |
| 43 #include "clang/Lex/LiteralSupport.h" | |
| 44 #include "clang/Lex/MacroArgs.h" | |
| 45 #include "clang/Lex/MacroInfo.h" | |
| 46 #include "clang/Lex/ModuleLoader.h" | |
| 47 #include "clang/Lex/Pragma.h" | |
| 48 #include "clang/Lex/PreprocessingRecord.h" | |
| 49 #include "clang/Lex/PreprocessorLexer.h" | |
| 50 #include "clang/Lex/PreprocessorOptions.h" | |
| 51 #include "clang/Lex/ScratchBuffer.h" | |
| 52 #include "clang/Lex/Token.h" | |
| 53 #include "clang/Lex/TokenLexer.h" | |
| 54 #include "llvm/ADT/APInt.h" | |
| 55 #include "llvm/ADT/ArrayRef.h" | |
| 56 #include "llvm/ADT/DenseMap.h" | |
| 57 #include "llvm/ADT/STLExtras.h" | |
| 58 #include "llvm/ADT/SmallString.h" | |
| 59 #include "llvm/ADT/SmallVector.h" | |
| 60 #include "llvm/ADT/StringRef.h" | |
| 61 #include "llvm/ADT/StringSwitch.h" | |
| 62 #include "llvm/Support/Capacity.h" | |
| 63 #include "llvm/Support/ErrorHandling.h" | |
| 64 #include "llvm/Support/MemoryBuffer.h" | |
| 65 #include "llvm/Support/raw_ostream.h" | |
| 66 #include <algorithm> | |
| 67 #include <cassert> | |
| 68 #include <memory> | |
| 69 #include <string> | |
| 70 #include <utility> | |
| 71 #include <vector> | |
| 72 | |
| 73 using namespace clang; | |
| 74 | |
| 75 LLVM_INSTANTIATE_REGISTRY(PragmaHandlerRegistry) | |
| 76 | |
| 77 ExternalPreprocessorSource::~ExternalPreprocessorSource() = default; | |
| 78 | |
| 79 Preprocessor::Preprocessor(std::shared_ptr<PreprocessorOptions> PPOpts, | |
| 80 DiagnosticsEngine &diags, LangOptions &opts, | |
| 81 SourceManager &SM, HeaderSearch &Headers, | |
| 82 ModuleLoader &TheModuleLoader, | |
| 83 IdentifierInfoLookup *IILookup, bool OwnsHeaders, | |
| 84 TranslationUnitKind TUKind) | |
| 85 : PPOpts(std::move(PPOpts)), Diags(&diags), LangOpts(opts), | |
| 86 FileMgr(Headers.getFileMgr()), SourceMgr(SM), | |
| 87 ScratchBuf(new ScratchBuffer(SourceMgr)), HeaderInfo(Headers), | |
| 88 TheModuleLoader(TheModuleLoader), ExternalSource(nullptr), | |
| 89 // As the language options may have not been loaded yet (when | |
| 90 // deserializing an ASTUnit), adding keywords to the identifier table is | |
| 91 // deferred to Preprocessor::Initialize(). | |
| 92 Identifiers(IILookup), PragmaHandlers(new PragmaNamespace(StringRef())), | |
| 93 TUKind(TUKind), SkipMainFilePreamble(0, true), | |
| 94 CurSubmoduleState(&NullSubmoduleState) { | |
| 95 OwnsHeaderSearch = OwnsHeaders; | |
| 96 | |
| 97 // Default to discarding comments. | |
| 98 KeepComments = false; | |
| 99 KeepMacroComments = false; | |
| 100 SuppressIncludeNotFoundError = false; | |
| 101 | |
| 102 // Macro expansion is enabled. | |
| 103 DisableMacroExpansion = false; | |
| 104 MacroExpansionInDirectivesOverride = false; | |
| 105 InMacroArgs = false; | |
| 106 ArgMacro = nullptr; | |
| 107 InMacroArgPreExpansion = false; | |
| 108 NumCachedTokenLexers = 0; | |
| 109 PragmasEnabled = true; | |
| 110 ParsingIfOrElifDirective = false; | |
| 111 PreprocessedOutput = false; | |
| 112 | |
| 113 // We haven't read anything from the external source. | |
| 114 ReadMacrosFromExternalSource = false; | |
| 115 | |
| 116 BuiltinInfo = std::make_unique<Builtin::Context>(); | |
| 117 | |
| 118 // "Poison" __VA_ARGS__, __VA_OPT__ which can only appear in the expansion of | |
| 119 // a macro. They get unpoisoned where it is allowed. | |
| 120 (Ident__VA_ARGS__ = getIdentifierInfo("__VA_ARGS__"))->setIsPoisoned(); | |
| 121 SetPoisonReason(Ident__VA_ARGS__,diag::ext_pp_bad_vaargs_use); | |
| 207 | 122 (Ident__VA_OPT__ = getIdentifierInfo("__VA_OPT__"))->setIsPoisoned(); |
| 123 SetPoisonReason(Ident__VA_OPT__,diag::ext_pp_bad_vaopt_use); | |
| 150 | 124 |
| 125 // Initialize the pragma handlers. | |
| 126 RegisterBuiltinPragmas(); | |
| 127 | |
| 128 // Initialize builtin macros like __LINE__ and friends. | |
| 129 RegisterBuiltinMacros(); | |
| 130 | |
| 131 if(LangOpts.Borland) { | |
| 132 Ident__exception_info = getIdentifierInfo("_exception_info"); | |
| 133 Ident___exception_info = getIdentifierInfo("__exception_info"); | |
| 134 Ident_GetExceptionInfo = getIdentifierInfo("GetExceptionInformation"); | |
| 135 Ident__exception_code = getIdentifierInfo("_exception_code"); | |
| 136 Ident___exception_code = getIdentifierInfo("__exception_code"); | |
| 137 Ident_GetExceptionCode = getIdentifierInfo("GetExceptionCode"); | |
| 138 Ident__abnormal_termination = getIdentifierInfo("_abnormal_termination"); | |
| 139 Ident___abnormal_termination = getIdentifierInfo("__abnormal_termination"); | |
| 140 Ident_AbnormalTermination = getIdentifierInfo("AbnormalTermination"); | |
| 141 } else { | |
| 142 Ident__exception_info = Ident__exception_code = nullptr; | |
| 143 Ident__abnormal_termination = Ident___exception_info = nullptr; | |
| 144 Ident___exception_code = Ident___abnormal_termination = nullptr; | |
| 145 Ident_GetExceptionInfo = Ident_GetExceptionCode = nullptr; | |
| 146 Ident_AbnormalTermination = nullptr; | |
| 147 } | |
| 148 | |
| 152 | 149 |
| 150 | 150 // If using a PCH where a #pragma hdrstop is expected, start skipping tokens. |
| 151 if (usingPCHWithPragmaHdrStop()) | |
| 152 SkippingUntilPragmaHdrStop = true; | |
| 153 | |
| 154 // If using a PCH with a through header, start skipping tokens. | |
| 155 if (!this->PPOpts->PCHThroughHeader.empty() && | |
| 156 !this->PPOpts->ImplicitPCHInclude.empty()) | |
| 157 SkippingUntilPCHThroughHeader = true; | |
| 158 | |
| 152 | 159 #ifndef noCbC |
| 160 SavedDepth = 0; | |
| 161 SavedTokenFlag = false; | |
| 162 #endif | |
| 163 | |
| 150 | 164 if (this->PPOpts->GeneratePreamble) |
| 165 PreambleConditionalStack.startRecording(); | |
| 166 | |
| 167 ExcludedConditionalDirectiveSkipMappings = | |
| 168 this->PPOpts->ExcludedConditionalDirectiveSkipMappings; | |
| 169 if (ExcludedConditionalDirectiveSkipMappings) | |
| 170 ExcludedConditionalDirectiveSkipMappings->clear(); | |
| 171 | |
| 172 MaxTokens = LangOpts.MaxTokens; | |
| 173 } | |
| 174 | |
| 175 Preprocessor::~Preprocessor() { | |
| 176 assert(BacktrackPositions.empty() && "EnableBacktrack/Backtrack imbalance!"); | |
| 177 | |
| 178 IncludeMacroStack.clear(); | |
| 179 | |
| 180 // Destroy any macro definitions. | |
| 181 while (MacroInfoChain *I = MIChainHead) { | |
| 182 MIChainHead = I->Next; | |
| 183 I->~MacroInfoChain(); | |
| 184 } | |
| 185 | |
| 186 // Free any cached macro expanders. | |
| 187 // This populates MacroArgCache, so all TokenLexers need to be destroyed | |
| 188 // before the code below that frees up the MacroArgCache list. | |
| 189 std::fill(TokenLexerCache, TokenLexerCache + NumCachedTokenLexers, nullptr); | |
| 190 CurTokenLexer.reset(); | |
| 191 | |
| 192 // Free any cached MacroArgs. | |
| 193 for (MacroArgs *ArgList = MacroArgCache; ArgList;) | |
| 194 ArgList = ArgList->deallocate(); | |
| 195 | |
| 196 // Delete the header search info, if we own it. | |
| 197 if (OwnsHeaderSearch) | |
| 198 delete &HeaderInfo; | |
| 199 } | |
| 200 | |
| 201 void Preprocessor::Initialize(const TargetInfo &Target, | |
| 202 const TargetInfo *AuxTarget) { | |
| 203 assert((!this->Target || this->Target == &Target) && | |
| 204 "Invalid override of target information"); | |
| 205 this->Target = &Target; | |
| 206 | |
| 207 assert((!this->AuxTarget || this->AuxTarget == AuxTarget) && | |
| 208 "Invalid override of aux target information."); | |
| 209 this->AuxTarget = AuxTarget; | |
| 210 | |
| 211 // Initialize information about built-ins. | |
| 212 BuiltinInfo->InitializeTarget(Target, AuxTarget); | |
| 213 HeaderInfo.setTarget(Target); | |
| 214 | |
| 215 // Populate the identifier table with info about keywords for the current language. | |
| 216 Identifiers.AddKeywords(LangOpts); | |
| 217 } | |
| 218 | |
| 219 void Preprocessor::InitializeForModelFile() { | |
| 220 NumEnteredSourceFiles = 0; | |
| 221 | |
| 222 // Reset pragmas | |
| 223 PragmaHandlersBackup = std::move(PragmaHandlers); | |
| 224 PragmaHandlers = std::make_unique<PragmaNamespace>(StringRef()); | |
| 225 RegisterBuiltinPragmas(); | |
| 226 | |
| 227 // Reset PredefinesFileID | |
| 228 PredefinesFileID = FileID(); | |
| 229 } | |
| 230 | |
| 231 void Preprocessor::FinalizeForModelFile() { | |
| 232 NumEnteredSourceFiles = 1; | |
| 233 | |
| 234 PragmaHandlers = std::move(PragmaHandlersBackup); | |
| 235 } | |
| 236 | |
| 237 void Preprocessor::DumpToken(const Token &Tok, bool DumpFlags) const { | |
| 238 llvm::errs() << tok::getTokenName(Tok.getKind()) << " '" | |
| 239 << getSpelling(Tok) << "'"; | |
| 240 | |
| 241 if (!DumpFlags) return; | |
| 242 | |
| 243 llvm::errs() << "\t"; | |
| 244 if (Tok.isAtStartOfLine()) | |
| 245 llvm::errs() << " [StartOfLine]"; | |
| 246 if (Tok.hasLeadingSpace()) | |
| 247 llvm::errs() << " [LeadingSpace]"; | |
| 248 if (Tok.isExpandDisabled()) | |
| 249 llvm::errs() << " [ExpandDisabled]"; | |
| 250 if (Tok.needsCleaning()) { | |
| 251 const char *Start = SourceMgr.getCharacterData(Tok.getLocation()); | |
| 252 llvm::errs() << " [UnClean='" << StringRef(Start, Tok.getLength()) | |
| 253 << "']"; | |
| 254 } | |
| 255 | |
| 256 llvm::errs() << "\tLoc=<"; | |
| 257 DumpLocation(Tok.getLocation()); | |
| 258 llvm::errs() << ">"; | |
| 259 } | |
| 260 | |
| 261 void Preprocessor::DumpLocation(SourceLocation Loc) const { | |
| 262 Loc.print(llvm::errs(), SourceMgr); | |
| 263 } | |
| 264 | |
| 265 void Preprocessor::DumpMacro(const MacroInfo &MI) const { | |
| 266 llvm::errs() << "MACRO: "; | |
| 267 for (unsigned i = 0, e = MI.getNumTokens(); i != e; ++i) { | |
| 268 DumpToken(MI.getReplacementToken(i)); | |
| 269 llvm::errs() << " "; | |
| 270 } | |
| 271 llvm::errs() << "\n"; | |
| 272 } | |
| 273 | |
| 274 void Preprocessor::PrintStats() { | |
| 275 llvm::errs() << "\n*** Preprocessor Stats:\n"; | |
| 276 llvm::errs() << NumDirectives << " directives found:\n"; | |
| 277 llvm::errs() << " " << NumDefined << " #define.\n"; | |
| 278 llvm::errs() << " " << NumUndefined << " #undef.\n"; | |
| 279 llvm::errs() << " #include/#include_next/#import:\n"; | |
| 280 llvm::errs() << " " << NumEnteredSourceFiles << " source files entered.\n"; | |
| 281 llvm::errs() << " " << MaxIncludeStackDepth << " max include stack depth\n"; | |
| 282 llvm::errs() << " " << NumIf << " #if/#ifndef/#ifdef.\n"; | |
| 207 | 283 llvm::errs() << " " << NumElse << " #else/#elif/#elifdef/#elifndef.\n"; |
| 150 | 284 llvm::errs() << " " << NumEndif << " #endif.\n"; |
| 285 llvm::errs() << " " << NumPragma << " #pragma.\n"; | |
| 286 llvm::errs() << NumSkipped << " #if/#ifndef#ifdef regions skipped\n"; | |
| 287 | |
| 288 llvm::errs() << NumMacroExpanded << "/" << NumFnMacroExpanded << "/" | |
| 289 << NumBuiltinMacroExpanded << " obj/fn/builtin macros expanded, " | |
| 290 << NumFastMacroExpanded << " on the fast path.\n"; | |
| 291 llvm::errs() << (NumFastTokenPaste+NumTokenPaste) | |
| 292 << " token paste (##) operations performed, " | |
| 293 << NumFastTokenPaste << " on the fast path.\n"; | |
| 294 | |
| 295 llvm::errs() << "\nPreprocessor Memory: " << getTotalMemory() << "B total"; | |
| 296 | |
| 297 llvm::errs() << "\n BumpPtr: " << BP.getTotalMemory(); | |
| 298 llvm::errs() << "\n Macro Expanded Tokens: " | |
| 299 << llvm::capacity_in_bytes(MacroExpandedTokens); | |
| 300 llvm::errs() << "\n Predefines Buffer: " << Predefines.capacity(); | |
| 301 // FIXME: List information for all submodules. | |
| 302 llvm::errs() << "\n Macros: " | |
| 303 << llvm::capacity_in_bytes(CurSubmoduleState->Macros); | |
| 304 llvm::errs() << "\n #pragma push_macro Info: " | |
| 305 << llvm::capacity_in_bytes(PragmaPushMacroInfo); | |
| 306 llvm::errs() << "\n Poison Reasons: " | |
| 307 << llvm::capacity_in_bytes(PoisonReasons); | |
| 308 llvm::errs() << "\n Comment Handlers: " | |
| 309 << llvm::capacity_in_bytes(CommentHandlers) << "\n"; | |
| 310 } | |
| 311 | |
| 312 Preprocessor::macro_iterator | |
| 313 Preprocessor::macro_begin(bool IncludeExternalMacros) const { | |
| 314 if (IncludeExternalMacros && ExternalSource && | |
| 315 !ReadMacrosFromExternalSource) { | |
| 316 ReadMacrosFromExternalSource = true; | |
| 317 ExternalSource->ReadDefinedMacros(); | |
| 318 } | |
| 319 | |
| 320 // Make sure we cover all macros in visible modules. | |
| 321 for (const ModuleMacro &Macro : ModuleMacros) | |
| 322 CurSubmoduleState->Macros.insert(std::make_pair(Macro.II, MacroState())); | |
| 323 | |
| 324 return CurSubmoduleState->Macros.begin(); | |
| 325 } | |
| 326 | |
| 327 size_t Preprocessor::getTotalMemory() const { | |
| 328 return BP.getTotalMemory() | |
| 329 + llvm::capacity_in_bytes(MacroExpandedTokens) | |
| 330 + Predefines.capacity() /* Predefines buffer. */ | |
| 331 // FIXME: Include sizes from all submodules, and include MacroInfo sizes, | |
| 332 // and ModuleMacros. | |
| 333 + llvm::capacity_in_bytes(CurSubmoduleState->Macros) | |
| 334 + llvm::capacity_in_bytes(PragmaPushMacroInfo) | |
| 335 + llvm::capacity_in_bytes(PoisonReasons) | |
| 336 + llvm::capacity_in_bytes(CommentHandlers); | |
| 337 } | |
| 338 | |
| 339 Preprocessor::macro_iterator | |
| 340 Preprocessor::macro_end(bool IncludeExternalMacros) const { | |
| 341 if (IncludeExternalMacros && ExternalSource && | |
| 342 !ReadMacrosFromExternalSource) { | |
| 343 ReadMacrosFromExternalSource = true; | |
| 344 ExternalSource->ReadDefinedMacros(); | |
| 345 } | |
| 346 | |
| 347 return CurSubmoduleState->Macros.end(); | |
| 348 } | |
| 349 | |
| 350 /// Compares macro tokens with a specified token value sequence. | |
| 351 static bool MacroDefinitionEquals(const MacroInfo *MI, | |
| 352 ArrayRef<TokenValue> Tokens) { | |
| 353 return Tokens.size() == MI->getNumTokens() && | |
| 354 std::equal(Tokens.begin(), Tokens.end(), MI->tokens_begin()); | |
| 355 } | |
| 356 | |
| 357 StringRef Preprocessor::getLastMacroWithSpelling( | |
| 358 SourceLocation Loc, | |
| 359 ArrayRef<TokenValue> Tokens) const { | |
| 360 SourceLocation BestLocation; | |
| 361 StringRef BestSpelling; | |
| 362 for (Preprocessor::macro_iterator I = macro_begin(), E = macro_end(); | |
| 363 I != E; ++I) { | |
| 364 const MacroDirective::DefInfo | |
| 365 Def = I->second.findDirectiveAtLoc(Loc, SourceMgr); | |
| 366 if (!Def || !Def.getMacroInfo()) | |
| 367 continue; | |
| 368 if (!Def.getMacroInfo()->isObjectLike()) | |
| 369 continue; | |
| 370 if (!MacroDefinitionEquals(Def.getMacroInfo(), Tokens)) | |
| 371 continue; | |
| 372 SourceLocation Location = Def.getLocation(); | |
| 373 // Choose the macro defined latest. | |
| 374 if (BestLocation.isInvalid() || | |
| 375 (Location.isValid() && | |
| 376 SourceMgr.isBeforeInTranslationUnit(BestLocation, Location))) { | |
| 377 BestLocation = Location; | |
| 378 BestSpelling = I->first->getName(); | |
| 379 } | |
| 380 } | |
| 381 return BestSpelling; | |
| 382 } | |
| 383 | |
| 384 void Preprocessor::recomputeCurLexerKind() { | |
| 385 if (CurLexer) | |
| 386 CurLexerKind = CLK_Lexer; | |
| 387 else if (CurTokenLexer) | |
| 388 CurLexerKind = CLK_TokenLexer; | |
| 389 else | |
| 390 CurLexerKind = CLK_CachingLexer; | |
| 391 } | |
| 392 | |
| 393 bool Preprocessor::SetCodeCompletionPoint(const FileEntry *File, | |
| 394 unsigned CompleteLine, | |
| 395 unsigned CompleteColumn) { | |
| 396 assert(File); | |
| 397 assert(CompleteLine && CompleteColumn && "Starts from 1:1"); | |
| 398 assert(!CodeCompletionFile && "Already set"); | |
| 399 | |
| 400 // Load the actual file's contents. | |
| 207 | 401 Optional<llvm::MemoryBufferRef> Buffer = |
| 402 SourceMgr.getMemoryBufferForFileOrNone(File); | |
| 403 if (!Buffer) | |
| 150 | 404 return true; |
| 405 | |
| 406 // Find the byte position of the truncation point. | |
| 407 const char *Position = Buffer->getBufferStart(); | |
| 408 for (unsigned Line = 1; Line < CompleteLine; ++Line) { | |
| 409 for (; *Position; ++Position) { | |
| 410 if (*Position != '\r' && *Position != '\n') | |
| 411 continue; | |
| 412 | |
| 413 // Eat \r\n or \n\r as a single line. | |
| 414 if ((Position[1] == '\r' || Position[1] == '\n') && | |
| 415 Position[0] != Position[1]) | |
| 416 ++Position; | |
| 417 ++Position; | |
| 418 break; | |
| 419 } | |
| 420 } | |
| 421 | |
| 422 Position += CompleteColumn - 1; | |
| 423 | |
| 424 // If pointing inside the preamble, adjust the position at the beginning of | |
| 425 // the file after the preamble. | |
| 426 if (SkipMainFilePreamble.first && | |
| 427 SourceMgr.getFileEntryForID(SourceMgr.getMainFileID()) == File) { | |
| 428 if (Position - Buffer->getBufferStart() < SkipMainFilePreamble.first) | |
| 429 Position = Buffer->getBufferStart() + SkipMainFilePreamble.first; | |
| 430 } | |
| 431 | |
| 432 if (Position > Buffer->getBufferEnd()) | |
| 433 Position = Buffer->getBufferEnd(); | |
| 434 | |
| 435 CodeCompletionFile = File; | |
| 436 CodeCompletionOffset = Position - Buffer->getBufferStart(); | |
| 437 | |
| 438 auto NewBuffer = llvm::WritableMemoryBuffer::getNewUninitMemBuffer( | |
| 439 Buffer->getBufferSize() + 1, Buffer->getBufferIdentifier()); | |
| 440 char *NewBuf = NewBuffer->getBufferStart(); | |
| 441 char *NewPos = std::copy(Buffer->getBufferStart(), Position, NewBuf); | |
| 442 *NewPos = '\0'; | |
| 443 std::copy(Position, Buffer->getBufferEnd(), NewPos+1); | |
| 444 SourceMgr.overrideFileContents(File, std::move(NewBuffer)); | |
| 445 | |
| 446 return false; | |
| 447 } | |
| 448 | |
| 449 void Preprocessor::CodeCompleteIncludedFile(llvm::StringRef Dir, | |
| 450 bool IsAngled) { | |
| 207 | 451 setCodeCompletionReached(); |
| 150 | 452 if (CodeComplete) |
| 453 CodeComplete->CodeCompleteIncludedFile(Dir, IsAngled); | |
| 454 } | |
| 455 | |
| 456 void Preprocessor::CodeCompleteNaturalLanguage() { | |
| 207 | 457 setCodeCompletionReached(); |
| 150 | 458 if (CodeComplete) |
| 459 CodeComplete->CodeCompleteNaturalLanguage(); | |
| 460 } | |
| 461 | |
| 462 /// getSpelling - This method is used to get the spelling of a token into a | |
| 463 /// SmallVector. Note that the returned StringRef may not point to the | |
| 464 /// supplied buffer if a copy can be avoided. | |
| 465 StringRef Preprocessor::getSpelling(const Token &Tok, | |
| 466 SmallVectorImpl<char> &Buffer, | |
| 467 bool *Invalid) const { | |
| 468 // NOTE: this has to be checked *before* testing for an IdentifierInfo. | |
| 469 if (Tok.isNot(tok::raw_identifier) && !Tok.hasUCN()) { | |
| 470 // Try the fast path. | |
| 471 if (const IdentifierInfo *II = Tok.getIdentifierInfo()) | |
| 472 return II->getName(); | |
| 473 } | |
| 474 | |
| 475 // Resize the buffer if we need to copy into it. | |
| 476 if (Tok.needsCleaning()) | |
| 477 Buffer.resize(Tok.getLength()); | |
| 478 | |
| 479 const char *Ptr = Buffer.data(); | |
| 480 unsigned Len = getSpelling(Tok, Ptr, Invalid); | |
| 481 return StringRef(Ptr, Len); | |
| 482 } | |
| 483 | |
| 484 /// CreateString - Plop the specified string into a scratch buffer and return a | |
| 485 /// location for it. If specified, the source location provides a source | |
| 486 /// location for the token. | |
| 487 void Preprocessor::CreateString(StringRef Str, Token &Tok, | |
| 488 SourceLocation ExpansionLocStart, | |
| 489 SourceLocation ExpansionLocEnd) { | |
| 490 Tok.setLength(Str.size()); | |
| 491 | |
| 492 const char *DestPtr; | |
| 493 SourceLocation Loc = ScratchBuf->getToken(Str.data(), Str.size(), DestPtr); | |
| 494 | |
| 495 if (ExpansionLocStart.isValid()) | |
| 496 Loc = SourceMgr.createExpansionLoc(Loc, ExpansionLocStart, | |
| 497 ExpansionLocEnd, Str.size()); | |
| 498 Tok.setLocation(Loc); | |
| 499 | |
| 500 // If this is a raw identifier or a literal token, set the pointer data. | |
| 501 if (Tok.is(tok::raw_identifier)) | |
| 502 Tok.setRawIdentifierData(DestPtr); | |
| 503 else if (Tok.isLiteral()) | |
| 504 Tok.setLiteralData(DestPtr); | |
| 505 } | |
| 506 | |
| 507 SourceLocation Preprocessor::SplitToken(SourceLocation Loc, unsigned Length) { | |
| 508 auto &SM = getSourceManager(); | |
| 509 SourceLocation SpellingLoc = SM.getSpellingLoc(Loc); | |
| 510 std::pair<FileID, unsigned> LocInfo = SM.getDecomposedLoc(SpellingLoc); | |
| 511 bool Invalid = false; | |
| 512 StringRef Buffer = SM.getBufferData(LocInfo.first, &Invalid); | |
| 513 if (Invalid) | |
| 514 return SourceLocation(); | |
| 515 | |
| 516 // FIXME: We could consider re-using spelling for tokens we see repeatedly. | |
| 517 const char *DestPtr; | |
| 518 SourceLocation Spelling = | |
| 519 ScratchBuf->getToken(Buffer.data() + LocInfo.second, Length, DestPtr); | |
| 520 return SM.createTokenSplitLoc(Spelling, Loc, Loc.getLocWithOffset(Length)); | |
| 521 } | |
| 522 | |
| 523 Module *Preprocessor::getCurrentModule() { | |
| 524 if (!getLangOpts().isCompilingModule()) | |
| 525 return nullptr; | |
| 526 | |
| 527 return getHeaderSearchInfo().lookupModule(getLangOpts().CurrentModule); | |
| 528 } | |
| 529 | |
| 530 //===----------------------------------------------------------------------===// | |
| 531 // Preprocessor Initialization Methods | |
| 532 //===----------------------------------------------------------------------===// | |
| 533 | |
| 534 /// EnterMainSourceFile - Enter the specified FileID as the main source file, | |
| 535 /// which implicitly adds the builtin defines etc. | |
| 536 void Preprocessor::EnterMainSourceFile() { | |
| 537 // We do not allow the preprocessor to reenter the main file. Doing so will | |
| 538 // cause FileID's to accumulate information from both runs (e.g. #line | |
| 539 // information) and predefined macros aren't guaranteed to be set properly. | |
| 540 assert(NumEnteredSourceFiles == 0 && "Cannot reenter the main file!"); | |
| 541 FileID MainFileID = SourceMgr.getMainFileID(); | |
| 542 | |
| 543 // If MainFileID is loaded it means we loaded an AST file, no need to enter | |
| 544 // a main file. | |
| 545 if (!SourceMgr.isLoadedFileID(MainFileID)) { | |
| 546 // Enter the main file source buffer. | |
| 547 EnterSourceFile(MainFileID, nullptr, SourceLocation()); | |
| 548 | |
| 549 // If we've been asked to skip bytes in the main file (e.g., as part of a | |
| 550 // precompiled preamble), do so now. | |
| 551 if (SkipMainFilePreamble.first > 0) | |
| 552 CurLexer->SetByteOffset(SkipMainFilePreamble.first, | |
| 553 SkipMainFilePreamble.second); | |
| 554 | |
| 555 // Tell the header info that the main file was entered. If the file is later | |
| 556 // #imported, it won't be re-entered. | |
| 557 if (const FileEntry *FE = SourceMgr.getFileEntryForID(MainFileID)) | |
| 558 HeaderInfo.IncrementIncludeCount(FE); | |
| 559 } | |
| 560 | |
| 561 // Preprocess Predefines to populate the initial preprocessor state. | |
| 562 std::unique_ptr<llvm::MemoryBuffer> SB = | |
| 563 llvm::MemoryBuffer::getMemBufferCopy(Predefines, "<built-in>"); | |
| 564 assert(SB && "Cannot create predefined source buffer"); | |
| 565 FileID FID = SourceMgr.createFileID(std::move(SB)); | |
| 566 assert(FID.isValid() && "Could not create FileID for predefines?"); | |
| 567 setPredefinesFileID(FID); | |
| 568 | |
| 569 // Start parsing the predefines. | |
| 570 EnterSourceFile(FID, nullptr, SourceLocation()); | |
| 571 | |
| 572 if (!PPOpts->PCHThroughHeader.empty()) { | |
| 573 // Lookup and save the FileID for the through header. If it isn't found | |
| 574 // in the search path, it's a fatal error. | |
| 575 const DirectoryLookup *CurDir; | |
| 576 Optional<FileEntryRef> File = LookupFile( | |
| 577 SourceLocation(), PPOpts->PCHThroughHeader, | |
| 578 /*isAngled=*/false, /*FromDir=*/nullptr, /*FromFile=*/nullptr, CurDir, | |
| 579 /*SearchPath=*/nullptr, /*RelativePath=*/nullptr, | |
| 580 /*SuggestedModule=*/nullptr, /*IsMapped=*/nullptr, | |
| 581 /*IsFrameworkFound=*/nullptr); | |
| 582 if (!File) { | |
| 583 Diag(SourceLocation(), diag::err_pp_through_header_not_found) | |
| 584 << PPOpts->PCHThroughHeader; | |
| 585 return; | |
| 586 } | |
| 587 setPCHThroughHeaderFileID( | |
| 588 SourceMgr.createFileID(*File, SourceLocation(), SrcMgr::C_User)); | |
| 589 } | |
| 590 | |
| 591 // Skip tokens from the Predefines and if needed the main file. | |
| 592 if ((usingPCHWithThroughHeader() && SkippingUntilPCHThroughHeader) || | |
| 593 (usingPCHWithPragmaHdrStop() && SkippingUntilPragmaHdrStop)) | |
| 594 SkipTokensWhileUsingPCH(); | |
| 595 } | |
| 596 | |
| 597 void Preprocessor::setPCHThroughHeaderFileID(FileID FID) { | |
| 598 assert(PCHThroughHeaderFileID.isInvalid() && | |
| 599 "PCHThroughHeaderFileID already set!"); | |
| 600 PCHThroughHeaderFileID = FID; | |
| 601 } | |
| 602 | |
| 603 bool Preprocessor::isPCHThroughHeader(const FileEntry *FE) { | |
| 604 assert(PCHThroughHeaderFileID.isValid() && | |
| 605 "Invalid PCH through header FileID"); | |
| 606 return FE == SourceMgr.getFileEntryForID(PCHThroughHeaderFileID); | |
| 607 } | |
| 608 | |
| 609 bool Preprocessor::creatingPCHWithThroughHeader() { | |
| 610 return TUKind == TU_Prefix && !PPOpts->PCHThroughHeader.empty() && | |
| 611 PCHThroughHeaderFileID.isValid(); | |
| 612 } | |
| 613 | |
| 614 bool Preprocessor::usingPCHWithThroughHeader() { | |
| 615 return TUKind != TU_Prefix && !PPOpts->PCHThroughHeader.empty() && | |
| 616 PCHThroughHeaderFileID.isValid(); | |
| 617 } | |
| 618 | |
| 619 bool Preprocessor::creatingPCHWithPragmaHdrStop() { | |
| 620 return TUKind == TU_Prefix && PPOpts->PCHWithHdrStop; | |
| 621 } | |
| 622 | |
| 623 bool Preprocessor::usingPCHWithPragmaHdrStop() { | |
| 624 return TUKind != TU_Prefix && PPOpts->PCHWithHdrStop; | |
| 625 } | |
| 626 | |
| 627 /// Skip tokens until after the #include of the through header or | |
| 628 /// until after a #pragma hdrstop is seen. Tokens in the predefines file | |
| 629 /// and the main file may be skipped. If the end of the predefines file | |
| 630 /// is reached, skipping continues into the main file. If the end of the | |
| 631 /// main file is reached, it's a fatal error. | |
| 632 void Preprocessor::SkipTokensWhileUsingPCH() { | |
| 633 bool ReachedMainFileEOF = false; | |
| 634 bool UsingPCHThroughHeader = SkippingUntilPCHThroughHeader; | |
| 635 bool UsingPragmaHdrStop = SkippingUntilPragmaHdrStop; | |
| 636 Token Tok; | |
| 637 while (true) { | |
| 638 bool InPredefines = | |
| 639 (CurLexer && CurLexer->getFileID() == getPredefinesFileID()); | |
| 640 switch (CurLexerKind) { | |
| 641 case CLK_Lexer: | |
| 642 CurLexer->Lex(Tok); | |
| 643 break; | |
| 644 case CLK_TokenLexer: | |
| 645 CurTokenLexer->Lex(Tok); | |
| 646 break; | |
| 647 case CLK_CachingLexer: | |
| 648 CachingLex(Tok); | |
| 649 break; | |
| 650 case CLK_LexAfterModuleImport: | |
| 651 LexAfterModuleImport(Tok); | |
| 652 break; | |
| 653 } | |
| 654 if (Tok.is(tok::eof) && !InPredefines) { | |
| 655 ReachedMainFileEOF = true; | |
| 656 break; | |
| 657 } | |
| 658 if (UsingPCHThroughHeader && !SkippingUntilPCHThroughHeader) | |
| 659 break; | |
| 660 if (UsingPragmaHdrStop && !SkippingUntilPragmaHdrStop) | |
| 661 break; | |
| 662 } | |
| 663 if (ReachedMainFileEOF) { | |
| 664 if (UsingPCHThroughHeader) | |
| 665 Diag(SourceLocation(), diag::err_pp_through_header_not_seen) | |
| 666 << PPOpts->PCHThroughHeader << 1; | |
| 667 else if (!PPOpts->PCHWithHdrStopCreate) | |
| 668 Diag(SourceLocation(), diag::err_pp_pragma_hdrstop_not_seen); | |
| 669 } | |
| 670 } | |
| 671 | |
| 672 void Preprocessor::replayPreambleConditionalStack() { | |
| 673 // Restore the conditional stack from the preamble, if there is one. | |
| 674 if (PreambleConditionalStack.isReplaying()) { | |
| 675 assert(CurPPLexer && | |
| 676 "CurPPLexer is null when calling replayPreambleConditionalStack."); | |
| 677 CurPPLexer->setConditionalLevels(PreambleConditionalStack.getStack()); | |
| 678 PreambleConditionalStack.doneReplaying(); | |
| 679 if (PreambleConditionalStack.reachedEOFWhileSkipping()) | |
| 680 SkipExcludedConditionalBlock( | |
| 681 PreambleConditionalStack.SkipInfo->HashTokenLoc, | |
| 682 PreambleConditionalStack.SkipInfo->IfTokenLoc, | |
| 683 PreambleConditionalStack.SkipInfo->FoundNonSkipPortion, | |
| 684 PreambleConditionalStack.SkipInfo->FoundElse, | |
| 685 PreambleConditionalStack.SkipInfo->ElseLoc); | |
| 686 } | |
| 687 } | |
| 688 | |
| 689 void Preprocessor::EndSourceFile() { | |
| 690 // Notify the client that we reached the end of the source file. | |
| 691 if (Callbacks) | |
| 692 Callbacks->EndOfMainFile(); | |
| 693 } | |
| 694 | |
| 695 //===----------------------------------------------------------------------===// | |
| 696 // Lexer Event Handling. | |
| 697 //===----------------------------------------------------------------------===// | |
| 698 | |
| 699 /// LookUpIdentifierInfo - Given a tok::raw_identifier token, look up the | |
| 700 /// identifier information for the token and install it into the token, | |
| 701 /// updating the token kind accordingly. | |
| 702 IdentifierInfo *Preprocessor::LookUpIdentifierInfo(Token &Identifier) const { | |
| 703 assert(!Identifier.getRawIdentifier().empty() && "No raw identifier data!"); | |
| 704 | |
| 705 // Look up this token, see if it is a macro, or if it is a language keyword. | |
| 706 IdentifierInfo *II; | |
| 707 if (!Identifier.needsCleaning() && !Identifier.hasUCN()) { | |
| 708 // No cleaning needed, just use the characters from the lexed buffer. | |
| 709 II = getIdentifierInfo(Identifier.getRawIdentifier()); | |
| 710 } else { | |
| 711 // Cleaning needed, alloca a buffer, clean into it, then use the buffer. | |
| 712 SmallString<64> IdentifierBuffer; | |
| 713 StringRef CleanedStr = getSpelling(Identifier, IdentifierBuffer); | |
| 714 | |
| 715 if (Identifier.hasUCN()) { | |
| 716 SmallString<64> UCNIdentifierBuffer; | |
| 717 expandUCNs(UCNIdentifierBuffer, CleanedStr); | |
| 718 II = getIdentifierInfo(UCNIdentifierBuffer); | |
| 719 } else { | |
| 720 II = getIdentifierInfo(CleanedStr); | |
| 721 } | |
| 722 } | |
| 723 | |
| 724 // Update the token info (identifier info and appropriate token kind). | |
| 725 Identifier.setIdentifierInfo(II); | |
| 726 if (getLangOpts().MSVCCompat && II->isCPlusPlusOperatorKeyword() && | |
| 727 getSourceManager().isInSystemHeader(Identifier.getLocation())) | |
| 728 Identifier.setKind(tok::identifier); | |
| 729 else | |
| 730 Identifier.setKind(II->getTokenID()); | |
| 731 | |
| 732 return II; | |
| 733 } | |
| 734 | |
| 735 void Preprocessor::SetPoisonReason(IdentifierInfo *II, unsigned DiagID) { | |
| 736 PoisonReasons[II] = DiagID; | |
| 737 } | |
| 738 | |
| 739 void Preprocessor::PoisonSEHIdentifiers(bool Poison) { | |
| 740 assert(Ident__exception_code && Ident__exception_info); | |
| 741 assert(Ident___exception_code && Ident___exception_info); | |
| 742 Ident__exception_code->setIsPoisoned(Poison); | |
| 743 Ident___exception_code->setIsPoisoned(Poison); | |
| 744 Ident_GetExceptionCode->setIsPoisoned(Poison); | |
| 745 Ident__exception_info->setIsPoisoned(Poison); | |
| 746 Ident___exception_info->setIsPoisoned(Poison); | |
| 747 Ident_GetExceptionInfo->setIsPoisoned(Poison); | |
| 748 Ident__abnormal_termination->setIsPoisoned(Poison); | |
| 749 Ident___abnormal_termination->setIsPoisoned(Poison); | |
| 750 Ident_AbnormalTermination->setIsPoisoned(Poison); | |
| 751 } | |
| 752 | |
| 753 void Preprocessor::HandlePoisonedIdentifier(Token & Identifier) { | |
| 754 assert(Identifier.getIdentifierInfo() && | |
| 755 "Can't handle identifiers without identifier info!"); | |
| 756 llvm::DenseMap<IdentifierInfo*,unsigned>::const_iterator it = | |
| 757 PoisonReasons.find(Identifier.getIdentifierInfo()); | |
| 758 if(it == PoisonReasons.end()) | |
| 759 Diag(Identifier, diag::err_pp_used_poisoned_id); | |
| 760 else | |
| 761 Diag(Identifier,it->second) << Identifier.getIdentifierInfo(); | |
| 762 } | |
| 763 | |
| 764 /// Returns a diagnostic message kind for reporting a future keyword as | |
| 765 /// appropriate for the identifier and specified language. | |
| 766 static diag::kind getFutureCompatDiagKind(const IdentifierInfo &II, | |
| 767 const LangOptions &LangOpts) { | |
| 768 assert(II.isFutureCompatKeyword() && "diagnostic should not be needed"); | |
| 769 | |
| 770 if (LangOpts.CPlusPlus) | |
| 771 return llvm::StringSwitch<diag::kind>(II.getName()) | |
| 772 #define CXX11_KEYWORD(NAME, FLAGS) \ | |
| 773 .Case(#NAME, diag::warn_cxx11_keyword) | |
| 173 | 774 #define CXX20_KEYWORD(NAME, FLAGS) \ |
| 775 .Case(#NAME, diag::warn_cxx20_keyword) | |
| 150 | 776 #include "clang/Basic/TokenKinds.def" |
| 173 | 777 // char8_t is not modeled as a CXX20_KEYWORD because it's not |
| 778 // unconditionally enabled in C++20 mode. (It can be disabled | |
| 779 // by -fno-char8_t.) | |
| 780 .Case("char8_t", diag::warn_cxx20_keyword) | |
| 150 | 781 ; |
| 782 | |
| 783 llvm_unreachable( | |
| 784 "Keyword not known to come from a newer Standard or proposed Standard"); | |
| 785 } | |
| 786 | |
| 787 void Preprocessor::updateOutOfDateIdentifier(IdentifierInfo &II) const { | |
| 788 assert(II.isOutOfDate() && "not out of date"); | |
| 789 getExternalSource()->updateOutOfDateIdentifier(II); | |
| 790 } | |
| 791 | |
| 792 /// HandleIdentifier - This callback is invoked when the lexer reads an | |
| 793 /// identifier. This callback looks up the identifier in the map and/or | |
| 794 /// potentially macro expands it or turns it into a named token (like 'for'). | |
| 795 /// | |
| 796 /// Note that callers of this method are guarded by checking the | |
| 797 /// IdentifierInfo's 'isHandleIdentifierCase' bit. If this method changes, the | |
| 798 /// IdentifierInfo methods that compute these properties will need to change to | |
| 799 /// match. | |
| 800 bool Preprocessor::HandleIdentifier(Token &Identifier) { | |
| 801 assert(Identifier.getIdentifierInfo() && | |
| 802 "Can't handle identifiers without identifier info!"); | |
| 803 | |
| 804 IdentifierInfo &II = *Identifier.getIdentifierInfo(); | |
| 805 | |
| 806 // If the information about this identifier is out of date, update it from | |
| 807 // the external source. | |
| 808 // We have to treat __VA_ARGS__ in a special way, since it gets | |
| 809 // serialized with isPoisoned = true, but our preprocessor may have | |
| 810 // unpoisoned it if we're defining a C99 macro. | |
| 811 if (II.isOutOfDate()) { | |
| 812 bool CurrentIsPoisoned = false; | |
| 813 const bool IsSpecialVariadicMacro = | |
| 814 &II == Ident__VA_ARGS__ || &II == Ident__VA_OPT__; | |
| 815 if (IsSpecialVariadicMacro) | |
| 816 CurrentIsPoisoned = II.isPoisoned(); | |
| 817 | |
| 818 updateOutOfDateIdentifier(II); | |
| 819 Identifier.setKind(II.getTokenID()); | |
| 820 | |
| 821 if (IsSpecialVariadicMacro) | |
| 822 II.setIsPoisoned(CurrentIsPoisoned); | |
| 823 } | |
| 824 | |
| 825 // If this identifier was poisoned, and if it was not produced from a macro | |
| 826 // expansion, emit an error. | |
| 827 if (II.isPoisoned() && CurPPLexer) { | |
| 828 HandlePoisonedIdentifier(Identifier); | |
| 829 } | |
| 830 | |
| 831 // If this is a macro to be expanded, do it. | |
| 832 if (MacroDefinition MD = getMacroDefinition(&II)) { | |
| 833 auto *MI = MD.getMacroInfo(); | |
| 834 assert(MI && "macro definition with no macro info?"); | |
| 835 if (!DisableMacroExpansion) { | |
| 836 if (!Identifier.isExpandDisabled() && MI->isEnabled()) { | |
| 837 // C99 6.10.3p10: If the preprocessing token immediately after the | |
| 838 // macro name isn't a '(', this macro should not be expanded. | |
| 839 if (!MI->isFunctionLike() || isNextPPTokenLParen()) | |
| 840 return HandleMacroExpandedIdentifier(Identifier, MD); | |
| 841 } else { | |
| 842 // C99 6.10.3.4p2 says that a disabled macro may never again be | |
| 843 // expanded, even if it's in a context where it could be expanded in the | |
| 844 // future. | |
| 845 Identifier.setFlag(Token::DisableExpand); | |
| 846 if (MI->isObjectLike() || isNextPPTokenLParen()) | |
| 847 Diag(Identifier, diag::pp_disabled_macro_expansion); | |
| 848 } | |
| 849 } | |
| 850 } | |
| 851 | |
| 852 // If this identifier is a keyword in a newer Standard or proposed Standard, | |
| 853 // produce a warning. Don't warn if we're not considering macro expansion, | |
| 854 // since this identifier might be the name of a macro. | |
| 855 // FIXME: This warning is disabled in cases where it shouldn't be, like | |
| 856 // "#define constexpr constexpr", "int constexpr;" | |
| 857 if (II.isFutureCompatKeyword() && !DisableMacroExpansion) { | |
| 858 Diag(Identifier, getFutureCompatDiagKind(II, getLangOpts())) | |
| 859 << II.getName(); | |
| 860 // Don't diagnose this keyword again in this translation unit. | |
| 861 II.setIsFutureCompatKeyword(false); | |
| 862 } | |
| 863 | |
| 864 // If this is an extension token, diagnose its use. | |
| 865 // We avoid diagnosing tokens that originate from macro definitions. | |
| 866 // FIXME: This warning is disabled in cases where it shouldn't be, | |
| 867 // like "#define TY typeof", "TY(1) x". | |
| 868 if (II.isExtensionToken() && !DisableMacroExpansion) | |
| 869 Diag(Identifier, diag::ext_token_used); | |
| 870 | |
| 871 // If this is the 'import' contextual keyword following an '@', note | |
| 872 // that the next token indicates a module name. | |
| 873 // | |
| 874 // Note that we do not treat 'import' as a contextual | |
| 875 // keyword when we're in a caching lexer, because caching lexers only get | |
| 876 // used in contexts where import declarations are disallowed. | |
| 877 // | |
| 878 // Likewise if this is the C++ Modules TS import keyword. | |
| 879 if (((LastTokenWasAt && II.isModulesImport()) || | |
| 880 Identifier.is(tok::kw_import)) && | |
| 881 !InMacroArgs && !DisableMacroExpansion && | |
| 882 (getLangOpts().Modules || getLangOpts().DebuggerSupport) && | |
| 883 CurLexerKind != CLK_CachingLexer) { | |
| 884 ModuleImportLoc = Identifier.getLocation(); | |
| 885 ModuleImportPath.clear(); | |
| 886 ModuleImportExpectsIdentifier = true; | |
| 887 CurLexerKind = CLK_LexAfterModuleImport; | |
| 888 } | |
| 889 return true; | |
| 890 } | |
| 891 | |
| 892 void Preprocessor::Lex(Token &Result) { | |
| 893 ++LexLevel; | |
| 894 | |
| 895 // We loop here until a lex function returns a token; this avoids recursion. | |
| 896 bool ReturnedToken; | |
| 897 do { | |
| 898 switch (CurLexerKind) { | |
| 899 case CLK_Lexer: | |
| 152 | 900 #ifndef noCbC |
| 901 ReturnedToken = CurLexer->Lex(Result, ProtoParsing); | |
| 902 #else | |
| 150 | 903 ReturnedToken = CurLexer->Lex(Result); |
| 152 | 904 #endif |
| 150 | 905 break; |
| 906 case CLK_TokenLexer: | |
| 907 ReturnedToken = CurTokenLexer->Lex(Result); | |
| 908 break; | |
| 909 case CLK_CachingLexer: | |
| 910 CachingLex(Result); | |
| 911 ReturnedToken = true; | |
| 912 break; | |
| 913 case CLK_LexAfterModuleImport: | |
| 914 ReturnedToken = LexAfterModuleImport(Result); | |
| 915 break; | |
| 916 } | |
| 917 } while (!ReturnedToken); | |
| 918 | |
| 919 if (Result.is(tok::unknown) && TheModuleLoader.HadFatalFailure) | |
| 920 return; | |
| 921 | |
| 922 if (Result.is(tok::code_completion) && Result.getIdentifierInfo()) { | |
| 923 // Remember the identifier before code completion token. | |
| 924 setCodeCompletionIdentifierInfo(Result.getIdentifierInfo()); | |
| 925 setCodeCompletionTokenRange(Result.getLocation(), Result.getEndLoc()); | |
| 926 // Set IdenfitierInfo to null to avoid confusing code that handles both | |
| 927 // identifiers and completion tokens. | |
| 928 Result.setIdentifierInfo(nullptr); | |
| 929 } | |
| 930 | |
| 931 // Update ImportSeqState to track our position within a C++20 import-seq | |
| 932 // if this token is being produced as a result of phase 4 of translation. | |
| 933 if (getLangOpts().CPlusPlusModules && LexLevel == 1 && | |
| 934 !Result.getFlag(Token::IsReinjected)) { | |
| 935 switch (Result.getKind()) { | |
| 936 case tok::l_paren: case tok::l_square: case tok::l_brace: | |
| 937 ImportSeqState.handleOpenBracket(); | |
| 938 break; | |
| 939 case tok::r_paren: case tok::r_square: | |
| 940 ImportSeqState.handleCloseBracket(); | |
| 941 break; | |
| 942 case tok::r_brace: | |
| 943 ImportSeqState.handleCloseBrace(); | |
| 944 break; | |
| 945 case tok::semi: | |
| 946 ImportSeqState.handleSemi(); | |
| 947 break; | |
| 948 case tok::header_name: | |
| 949 case tok::annot_header_unit: | |
| 950 ImportSeqState.handleHeaderName(); | |
| 951 break; | |
| 952 case tok::kw_export: | |
| 953 ImportSeqState.handleExport(); | |
| 954 break; | |
| 955 case tok::identifier: | |
| 956 if (Result.getIdentifierInfo()->isModulesImport()) { | |
| 957 ImportSeqState.handleImport(); | |
| 958 if (ImportSeqState.afterImportSeq()) { | |
| 959 ModuleImportLoc = Result.getLocation(); | |
| 960 ModuleImportPath.clear(); | |
| 961 ModuleImportExpectsIdentifier = true; | |
| 962 CurLexerKind = CLK_LexAfterModuleImport; | |
| 963 } | |
| 964 break; | |
| 965 } | |
| 966 LLVM_FALLTHROUGH; | |
| 967 default: | |
| 968 ImportSeqState.handleMisc(); | |
| 969 break; | |
| 970 } | |
| 971 } | |
| 972 | |
| 973 LastTokenWasAt = Result.is(tok::at); | |
| 974 --LexLevel; | |
| 975 | |
| 207 | 976 if ((LexLevel == 0 || PreprocessToken) && |
| 977 !Result.getFlag(Token::IsReinjected)) { | |
| 978 if (LexLevel == 0) | |
| 979 ++TokenCount; | |
| 150 | 980 if (OnToken) |
| 981 OnToken(Result); | |
| 982 } | |
| 983 } | |
| 984 | |
| 985 /// Lex a header-name token (including one formed from header-name-tokens if | |
| 986 /// \p AllowConcatenation is \c true). | |
| 987 /// | |
| 988 /// \param FilenameTok Filled in with the next token. On success, this will | |
| 989 /// be either a header_name token. On failure, it will be whatever other | |
| 990 /// token was found instead. | |
| 991 /// \param AllowMacroExpansion If \c true, allow the header name to be formed | |
| 992 /// by macro expansion (concatenating tokens as necessary if the first | |
| 993 /// token is a '<'). | |
| 994 /// \return \c true if we reached EOD or EOF while looking for a > token in | |
| 995 /// a concatenated header name and diagnosed it. \c false otherwise. | |
| 996 bool Preprocessor::LexHeaderName(Token &FilenameTok, bool AllowMacroExpansion) { | |
| 997 // Lex using header-name tokenization rules if tokens are being lexed from | |
| 998 // a file. Just grab a token normally if we're in a macro expansion. | |
| 999 if (CurPPLexer) | |
| 1000 CurPPLexer->LexIncludeFilename(FilenameTok); | |
| 1001 else | |
| 1002 Lex(FilenameTok); | |
| 1003 | |
| 1004 // This could be a <foo/bar.h> file coming from a macro expansion. In this | |
| 1005 // case, glue the tokens together into an angle_string_literal token. | |
| 1006 SmallString<128> FilenameBuffer; | |
| 1007 if (FilenameTok.is(tok::less) && AllowMacroExpansion) { | |
| 1008 bool StartOfLine = FilenameTok.isAtStartOfLine(); | |
| 1009 bool LeadingSpace = FilenameTok.hasLeadingSpace(); | |
| 1010 bool LeadingEmptyMacro = FilenameTok.hasLeadingEmptyMacro(); | |
| 1011 | |
| 1012 SourceLocation Start = FilenameTok.getLocation(); | |
| 1013 SourceLocation End; | |
| 1014 FilenameBuffer.push_back('<'); | |
| 1015 | |
| 1016 // Consume tokens until we find a '>'. | |
| 1017 // FIXME: A header-name could be formed starting or ending with an | |
| 1018 // alternative token. It's not clear whether that's ill-formed in all | |
| 1019 // cases. | |
| 1020 while (FilenameTok.isNot(tok::greater)) { | |
| 1021 Lex(FilenameTok); | |
| 1022 if (FilenameTok.isOneOf(tok::eod, tok::eof)) { | |
| 1023 Diag(FilenameTok.getLocation(), diag::err_expected) << tok::greater; | |
| 1024 Diag(Start, diag::note_matching) << tok::less; | |
| 1025 return true; | |
| 1026 } | |
| 1027 | |
| 1028 End = FilenameTok.getLocation(); | |
| 1029 | |
| 1030 // FIXME: Provide code completion for #includes. | |
| 1031 if (FilenameTok.is(tok::code_completion)) { | |
| 1032 setCodeCompletionReached(); | |
| 1033 Lex(FilenameTok); | |
| 1034 continue; | |
| 1035 } | |
| 1036 | |
| 1037 // Append the spelling of this token to the buffer. If there was a space | |
| 1038 // before it, add it now. | |
| 1039 if (FilenameTok.hasLeadingSpace()) | |
| 1040 FilenameBuffer.push_back(' '); | |
| 1041 | |
| 1042 // Get the spelling of the token, directly into FilenameBuffer if | |
| 1043 // possible. | |
| 1044 size_t PreAppendSize = FilenameBuffer.size(); | |
| 1045 FilenameBuffer.resize(PreAppendSize + FilenameTok.getLength()); | |
| 1046 | |
| 1047 const char *BufPtr = &FilenameBuffer[PreAppendSize]; | |
| 1048 unsigned ActualLen = getSpelling(FilenameTok, BufPtr); | |
| 1049 | |
| 1050 // If the token was spelled somewhere else, copy it into FilenameBuffer. | |
| 1051 if (BufPtr != &FilenameBuffer[PreAppendSize]) | |
| 1052 memcpy(&FilenameBuffer[PreAppendSize], BufPtr, ActualLen); | |
| 1053 | |
| 1054 // Resize FilenameBuffer to the correct size. | |
| 1055 if (FilenameTok.getLength() != ActualLen) | |
| 1056 FilenameBuffer.resize(PreAppendSize + ActualLen); | |
| 1057 } | |
| 1058 | |
| 1059 FilenameTok.startToken(); | |
| 1060 FilenameTok.setKind(tok::header_name); | |
| 1061 FilenameTok.setFlagValue(Token::StartOfLine, StartOfLine); | |
| 1062 FilenameTok.setFlagValue(Token::LeadingSpace, LeadingSpace); | |
| 1063 FilenameTok.setFlagValue(Token::LeadingEmptyMacro, LeadingEmptyMacro); | |
| 1064 CreateString(FilenameBuffer, FilenameTok, Start, End); | |
| 1065 } else if (FilenameTok.is(tok::string_literal) && AllowMacroExpansion) { | |
| 1066 // Convert a string-literal token of the form " h-char-sequence " | |
| 1067 // (produced by macro expansion) into a header-name token. | |
| 1068 // | |
| 1069 // The rules for header-names don't quite match the rules for | |
| 1070 // string-literals, but all the places where they differ result in | |
| 1071 // undefined behavior, so we can and do treat them the same. | |
| 1072 // | |
| 1073 // A string-literal with a prefix or suffix is not translated into a | |
| 1074 // header-name. This could theoretically be observable via the C++20 | |
| 1075 // context-sensitive header-name formation rules. | |
| 1076 StringRef Str = getSpelling(FilenameTok, FilenameBuffer); | |
| 1077 if (Str.size() >= 2 && Str.front() == '"' && Str.back() == '"') | |
| 1078 FilenameTok.setKind(tok::header_name); | |
| 1079 } | |
| 1080 | |
| 1081 return false; | |
| 1082 } | |
| 1083 | |
| 1084 /// Collect the tokens of a C++20 pp-import-suffix. | |
| 1085 void Preprocessor::CollectPpImportSuffix(SmallVectorImpl<Token> &Toks) { | |
| 1086 // FIXME: For error recovery, consider recognizing attribute syntax here | |
| 1087 // and terminating / diagnosing a missing semicolon if we find anything | |
| 1088 // else? (Can we leave that to the parser?) | |
| 1089 unsigned BracketDepth = 0; | |
| 1090 while (true) { | |
| 1091 Toks.emplace_back(); | |
| 1092 Lex(Toks.back()); | |
| 1093 | |
| 1094 switch (Toks.back().getKind()) { | |
| 1095 case tok::l_paren: case tok::l_square: case tok::l_brace: | |
| 1096 ++BracketDepth; | |
| 1097 break; | |
| 1098 | |
| 1099 case tok::r_paren: case tok::r_square: case tok::r_brace: | |
| 1100 if (BracketDepth == 0) | |
| 1101 return; | |
| 1102 --BracketDepth; | |
| 1103 break; | |
| 1104 | |
| 1105 case tok::semi: | |
| 1106 if (BracketDepth == 0) | |
| 1107 return; | |
| 1108 break; | |
| 1109 | |
| 1110 case tok::eof: | |
| 1111 return; | |
| 1112 | |
| 1113 default: | |
| 1114 break; | |
| 1115 } | |
| 1116 } | |
| 1117 } | |
| 1118 | |
| 1119 | |
| 1120 /// Lex a token following the 'import' contextual keyword. | |
| 1121 /// | |
| 1122 /// pp-import: [C++20] | |
| 1123 /// import header-name pp-import-suffix[opt] ; | |
| 1124 /// import header-name-tokens pp-import-suffix[opt] ; | |
| 1125 /// [ObjC] @ import module-name ; | |
| 1126 /// [Clang] import module-name ; | |
| 1127 /// | |
| 1128 /// header-name-tokens: | |
| 1129 /// string-literal | |
| 1130 /// < [any sequence of preprocessing-tokens other than >] > | |
| 1131 /// | |
| 1132 /// module-name: | |
| 1133 /// module-name-qualifier[opt] identifier | |
| 1134 /// | |
| 1135 /// module-name-qualifier | |
| 1136 /// module-name-qualifier[opt] identifier . | |
| 1137 /// | |
| 1138 /// We respond to a pp-import by importing macros from the named module. | |
| 1139 bool Preprocessor::LexAfterModuleImport(Token &Result) { | |
| 1140 // Figure out what kind of lexer we actually have. | |
| 1141 recomputeCurLexerKind(); | |
| 1142 | |
| 1143 // Lex the next token. The header-name lexing rules are used at the start of | |
| 1144 // a pp-import. | |
| 1145 // | |
| 1146 // For now, we only support header-name imports in C++20 mode. | |
| 1147 // FIXME: Should we allow this in all language modes that support an import | |
| 1148 // declaration as an extension? | |
| 1149 if (ModuleImportPath.empty() && getLangOpts().CPlusPlusModules) { | |
| 1150 if (LexHeaderName(Result)) | |
| 1151 return true; | |
| 1152 } else { | |
| 1153 Lex(Result); | |
| 1154 } | |
| 1155 | |
| 1156 // Allocate a holding buffer for a sequence of tokens and introduce it into | |
| 1157 // the token stream. | |
| 1158 auto EnterTokens = [this](ArrayRef<Token> Toks) { | |
| 1159 auto ToksCopy = std::make_unique<Token[]>(Toks.size()); | |
| 1160 std::copy(Toks.begin(), Toks.end(), ToksCopy.get()); | |
| 1161 EnterTokenStream(std::move(ToksCopy), Toks.size(), | |
| 1162 /*DisableMacroExpansion*/ true, /*IsReinject*/ false); | |
| 1163 }; | |
| 1164 | |
| 1165 // Check for a header-name. | |
| 1166 SmallVector<Token, 32> Suffix; | |
| 1167 if (Result.is(tok::header_name)) { | |
| 1168 // Enter the header-name token into the token stream; a Lex action cannot | |
| 1169 // both return a token and cache tokens (doing so would corrupt the token | |
| 1170 // cache if the call to Lex comes from CachingLex / PeekAhead). | |
| 1171 Suffix.push_back(Result); | |
| 1172 | |
| 1173 // Consume the pp-import-suffix and expand any macros in it now. We'll add | |
| 1174 // it back into the token stream later. | |
| 1175 CollectPpImportSuffix(Suffix); | |
| 1176 if (Suffix.back().isNot(tok::semi)) { | |
| 1177 // This is not a pp-import after all. | |
| 1178 EnterTokens(Suffix); | |
| 1179 return false; | |
| 1180 } | |
| 1181 | |
| 1182 // C++2a [cpp.module]p1: | |
| 1183 // The ';' preprocessing-token terminating a pp-import shall not have | |
| 1184 // been produced by macro replacement. | |
| 1185 SourceLocation SemiLoc = Suffix.back().getLocation(); | |
| 1186 if (SemiLoc.isMacroID()) | |
| 1187 Diag(SemiLoc, diag::err_header_import_semi_in_macro); | |
| 1188 | |
| 1189 // Reconstitute the import token. | |
| 1190 Token ImportTok; | |
| 1191 ImportTok.startToken(); | |
| 1192 ImportTok.setKind(tok::kw_import); | |
| 1193 ImportTok.setLocation(ModuleImportLoc); | |
| 1194 ImportTok.setIdentifierInfo(getIdentifierInfo("import")); | |
| 1195 ImportTok.setLength(6); | |
| 1196 | |
| 1197 auto Action = HandleHeaderIncludeOrImport( | |
| 1198 /*HashLoc*/ SourceLocation(), ImportTok, Suffix.front(), SemiLoc); | |
| 1199 switch (Action.Kind) { | |
| 1200 case ImportAction::None: | |
| 1201 break; | |
| 1202 | |
| 1203 case ImportAction::ModuleBegin: | |
| 1204 // Let the parser know we're textually entering the module. | |
| 1205 Suffix.emplace_back(); | |
| 1206 Suffix.back().startToken(); | |
| 1207 Suffix.back().setKind(tok::annot_module_begin); | |
| 1208 Suffix.back().setLocation(SemiLoc); | |
| 1209 Suffix.back().setAnnotationEndLoc(SemiLoc); | |
| 1210 Suffix.back().setAnnotationValue(Action.ModuleForHeader); | |
| 1211 LLVM_FALLTHROUGH; | |
| 1212 | |
| 1213 case ImportAction::ModuleImport: | |
| 1214 case ImportAction::SkippedModuleImport: | |
| 1215 // We chose to import (or textually enter) the file. Convert the | |
| 1216 // header-name token into a header unit annotation token. | |
| 1217 Suffix[0].setKind(tok::annot_header_unit); | |
| 1218 Suffix[0].setAnnotationEndLoc(Suffix[0].getLocation()); | |
| 1219 Suffix[0].setAnnotationValue(Action.ModuleForHeader); | |
| 1220 // FIXME: Call the moduleImport callback? | |
| 1221 break; | |
| 1222 case ImportAction::Failure: | |
| 1223 assert(TheModuleLoader.HadFatalFailure && | |
| 1224 "This should be an early exit only to a fatal error"); | |
| 1225 Result.setKind(tok::eof); | |
| 1226 CurLexer->cutOffLexing(); | |
| 1227 EnterTokens(Suffix); | |
| 1228 return true; | |
| 1229 } | |
| 1230 | |
| 1231 EnterTokens(Suffix); | |
| 1232 return false; | |
| 1233 } | |
| 1234 | |
| 1235 // The token sequence | |
| 1236 // | |
| 1237 // import identifier (. identifier)* | |
| 1238 // | |
| 1239 // indicates a module import directive. We already saw the 'import' | |
| 1240 // contextual keyword, so now we're looking for the identifiers. | |
| 1241 if (ModuleImportExpectsIdentifier && Result.getKind() == tok::identifier) { | |
| 1242 // We expected to see an identifier here, and we did; continue handling | |
| 1243 // identifiers. | |
| 1244 ModuleImportPath.push_back(std::make_pair(Result.getIdentifierInfo(), | |
| 1245 Result.getLocation())); | |
| 1246 ModuleImportExpectsIdentifier = false; | |
| 1247 CurLexerKind = CLK_LexAfterModuleImport; | |
| 1248 return true; | |
| 1249 } | |
| 1250 | |
| 1251 // If we're expecting a '.' or a ';', and we got a '.', then wait until we | |
| 1252 // see the next identifier. (We can also see a '[[' that begins an | |
| 1253 // attribute-specifier-seq here under the C++ Modules TS.) | |
| 1254 if (!ModuleImportExpectsIdentifier && Result.getKind() == tok::period) { | |
| 1255 ModuleImportExpectsIdentifier = true; | |
| 1256 CurLexerKind = CLK_LexAfterModuleImport; | |
| 1257 return true; | |
| 1258 } | |
| 1259 | |
| 1260 // If we didn't recognize a module name at all, this is not a (valid) import. | |
| 1261 if (ModuleImportPath.empty() || Result.is(tok::eof)) | |
| 1262 return true; | |
| 1263 | |
| 1264 // Consume the pp-import-suffix and expand any macros in it now, if we're not | |
| 1265 // at the semicolon already. | |
| 1266 SourceLocation SemiLoc = Result.getLocation(); | |
| 1267 if (Result.isNot(tok::semi)) { | |
| 1268 Suffix.push_back(Result); | |
| 1269 CollectPpImportSuffix(Suffix); | |
| 1270 if (Suffix.back().isNot(tok::semi)) { | |
| 1271 // This is not an import after all. | |
| 1272 EnterTokens(Suffix); | |
| 1273 return false; | |
| 1274 } | |
| 1275 SemiLoc = Suffix.back().getLocation(); | |
| 1276 } | |
| 1277 | |
| 1278 // Under the Modules TS, the dot is just part of the module name, and not | |
| 1279 // a real hierarchy separator. Flatten such module names now. | |
| 1280 // | |
| 1281 // FIXME: Is this the right level to be performing this transformation? | |
| 1282 std::string FlatModuleName; | |
| 1283 if (getLangOpts().ModulesTS || getLangOpts().CPlusPlusModules) { | |
| 1284 for (auto &Piece : ModuleImportPath) { | |
| 1285 if (!FlatModuleName.empty()) | |
| 1286 FlatModuleName += "."; | |
| 1287 FlatModuleName += Piece.first->getName(); | |
| 1288 } | |
| 1289 SourceLocation FirstPathLoc = ModuleImportPath[0].second; | |
| 1290 ModuleImportPath.clear(); | |
| 1291 ModuleImportPath.push_back( | |
| 1292 std::make_pair(getIdentifierInfo(FlatModuleName), FirstPathLoc)); | |
| 1293 } | |
| 1294 | |
| 1295 Module *Imported = nullptr; | |
| 1296 if (getLangOpts().Modules) { | |
| 1297 Imported = TheModuleLoader.loadModule(ModuleImportLoc, | |
| 1298 ModuleImportPath, | |
| 1299 Module::Hidden, | |
| 1300 /*IsInclusionDirective=*/false); | |
| 1301 if (Imported) | |
| 1302 makeModuleVisible(Imported, SemiLoc); | |
| 1303 } | |
| 1304 if (Callbacks) | |
| 1305 Callbacks->moduleImport(ModuleImportLoc, ModuleImportPath, Imported); | |
| 1306 | |
| 1307 if (!Suffix.empty()) { | |
| 1308 EnterTokens(Suffix); | |
| 1309 return false; | |
| 1310 } | |
| 1311 return true; | |
| 1312 } | |
| 1313 | |
| 1314 void Preprocessor::makeModuleVisible(Module *M, SourceLocation Loc) { | |
| 1315 CurSubmoduleState->VisibleModules.setVisible( | |
| 1316 M, Loc, [](Module *) {}, | |
| 1317 [&](ArrayRef<Module *> Path, Module *Conflict, StringRef Message) { | |
| 1318 // FIXME: Include the path in the diagnostic. | |
| 1319 // FIXME: Include the import location for the conflicting module. | |
| 1320 Diag(ModuleImportLoc, diag::warn_module_conflict) | |
| 1321 << Path[0]->getFullModuleName() | |
| 1322 << Conflict->getFullModuleName() | |
| 1323 << Message; | |
| 1324 }); | |
| 1325 | |
| 1326 // Add this module to the imports list of the currently-built submodule. | |
| 1327 if (!BuildingSubmoduleStack.empty() && M != BuildingSubmoduleStack.back().M) | |
| 1328 BuildingSubmoduleStack.back().M->Imports.insert(M); | |
| 1329 } | |
| 1330 | |
| 1331 bool Preprocessor::FinishLexStringLiteral(Token &Result, std::string &String, | |
| 1332 const char *DiagnosticTag, | |
| 1333 bool AllowMacroExpansion) { | |
| 1334 // We need at least one string literal. | |
| 1335 if (Result.isNot(tok::string_literal)) { | |
| 1336 Diag(Result, diag::err_expected_string_literal) | |
| 1337 << /*Source='in...'*/0 << DiagnosticTag; | |
| 1338 return false; | |
| 1339 } | |
| 1340 | |
| 1341 // Lex string literal tokens, optionally with macro expansion. | |
| 1342 SmallVector<Token, 4> StrToks; | |
| 1343 do { | |
| 1344 StrToks.push_back(Result); | |
| 1345 | |
| 1346 if (Result.hasUDSuffix()) | |
| 1347 Diag(Result, diag::err_invalid_string_udl); | |
| 1348 | |
| 1349 if (AllowMacroExpansion) | |
| 1350 Lex(Result); | |
| 1351 else | |
| 1352 LexUnexpandedToken(Result); | |
| 1353 } while (Result.is(tok::string_literal)); | |
| 1354 | |
| 1355 // Concatenate and parse the strings. | |
| 1356 StringLiteralParser Literal(StrToks, *this); | |
| 1357 assert(Literal.isAscii() && "Didn't allow wide strings in"); | |
| 1358 | |
| 1359 if (Literal.hadError) | |
| 1360 return false; | |
| 1361 | |
| 1362 if (Literal.Pascal) { | |
| 1363 Diag(StrToks[0].getLocation(), diag::err_expected_string_literal) | |
| 1364 << /*Source='in...'*/0 << DiagnosticTag; | |
| 1365 return false; | |
| 1366 } | |
| 1367 | |
| 1368 String = std::string(Literal.GetString()); | |
| 1369 return true; | |
| 1370 } | |
| 1371 | |
| 1372 bool Preprocessor::parseSimpleIntegerLiteral(Token &Tok, uint64_t &Value) { | |
| 1373 assert(Tok.is(tok::numeric_constant)); | |
| 1374 SmallString<8> IntegerBuffer; | |
| 1375 bool NumberInvalid = false; | |
| 1376 StringRef Spelling = getSpelling(Tok, IntegerBuffer, &NumberInvalid); | |
| 1377 if (NumberInvalid) | |
| 1378 return false; | |
| 207 | 1379 NumericLiteralParser Literal(Spelling, Tok.getLocation(), getSourceManager(), |
| 1380 getLangOpts(), getTargetInfo(), | |
| 1381 getDiagnostics()); | |
| 150 | 1382 if (Literal.hadError || !Literal.isIntegerLiteral() || Literal.hasUDSuffix()) |
| 1383 return false; | |
| 1384 llvm::APInt APVal(64, 0); | |
| 1385 if (Literal.GetIntegerValue(APVal)) | |
| 1386 return false; | |
| 1387 Lex(Tok); | |
| 1388 Value = APVal.getLimitedValue(); | |
| 1389 return true; | |
| 1390 } | |
| 1391 | |
| 1392 void Preprocessor::addCommentHandler(CommentHandler *Handler) { | |
| 1393 assert(Handler && "NULL comment handler"); | |
| 1394 assert(llvm::find(CommentHandlers, Handler) == CommentHandlers.end() && | |
| 1395 "Comment handler already registered"); | |
| 1396 CommentHandlers.push_back(Handler); | |
| 1397 } | |
| 1398 | |
| 1399 void Preprocessor::removeCommentHandler(CommentHandler *Handler) { | |
| 1400 std::vector<CommentHandler *>::iterator Pos = | |
| 1401 llvm::find(CommentHandlers, Handler); | |
| 1402 assert(Pos != CommentHandlers.end() && "Comment handler not registered"); | |
| 1403 CommentHandlers.erase(Pos); | |
| 1404 } | |
| 1405 | |
| 1406 bool Preprocessor::HandleComment(Token &result, SourceRange Comment) { | |
| 1407 bool AnyPendingTokens = false; | |
| 1408 for (std::vector<CommentHandler *>::iterator H = CommentHandlers.begin(), | |
| 1409 HEnd = CommentHandlers.end(); | |
| 1410 H != HEnd; ++H) { | |
| 1411 if ((*H)->HandleComment(*this, Comment)) | |
| 1412 AnyPendingTokens = true; | |
| 1413 } | |
| 1414 if (!AnyPendingTokens || getCommentRetentionState()) | |
| 1415 return false; | |
| 1416 Lex(result); | |
| 1417 return true; | |
| 1418 } | |
| 1419 | |
| 1420 ModuleLoader::~ModuleLoader() = default; | |
| 1421 | |
| 1422 CommentHandler::~CommentHandler() = default; | |
| 1423 | |
| 207 | 1424 EmptylineHandler::~EmptylineHandler() = default; |
| 1425 | |
| 150 | 1426 CodeCompletionHandler::~CodeCompletionHandler() = default; |
| 1427 | |
| 1428 void Preprocessor::createPreprocessingRecord() { | |
| 1429 if (Record) | |
| 1430 return; | |
| 1431 | |
| 1432 Record = new PreprocessingRecord(getSourceManager()); | |
| 1433 addPPCallbacks(std::unique_ptr<PPCallbacks>(Record)); | |
| 1434 } | |
| 197 | 1435 |
| 1436 #ifndef noCbC | |
| 1437 | |
|
198
cef006dc7fd5
CurLexer vanish after EnterMacro finish
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents:
197
diff
changeset
|
1438 Token Preprocessor::ReadFromString(const char *src , SourceLocation Loc) { |
| 197 | 1439 // Push the ( "string" ) tokens into the token stream. |
|
198
cef006dc7fd5
CurLexer vanish after EnterMacro finish
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents:
197
diff
changeset
|
1440 MacroInfo *MI = AllocateMacroInfo(Loc); |
| 205 | 1441 Token Tok; |
|
198
cef006dc7fd5
CurLexer vanish after EnterMacro finish
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents:
197
diff
changeset
|
1442 bool Invalid = false; |
| 200 | 1443 std::unique_ptr<Lexer> lx(new Lexer(CurLexer->getFileID(),getSourceManager().getBuffer(CurLexer->getFileID(), Loc, &Invalid),*this)); |
| 1444 lx->InitLexer(src,src,src + strlen(src)); | |
| 1445 lx->Lex(Tok); | |
| 1446 CurLexer.swap(lx); | |
| 205 | 1447 int i = 0; |
|
198
cef006dc7fd5
CurLexer vanish after EnterMacro finish
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents:
197
diff
changeset
|
1448 while (Tok.getKind() != tok::TokenKind::eof) { |
|
204
e348f3e5c8b2
ReadFromString worked.
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents:
201
diff
changeset
|
1449 Tok.setLocation(Loc); |
|
198
cef006dc7fd5
CurLexer vanish after EnterMacro finish
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents:
197
diff
changeset
|
1450 MI->AddTokenToBody(Tok); |
| 205 | 1451 Lex(Tok); i++; |
|
198
cef006dc7fd5
CurLexer vanish after EnterMacro finish
Shinji KONO <kono@ie.u-ryukyu.ac.jp>
parents:
197
diff
changeset
|
1452 } |
| 205 | 1453 Tok.setLocation(Loc); |
| 1454 MI->AddTokenToBody(Tok); i++; | |
| 1455 MI->DefinitionLength = i; | |
| 200 | 1456 CurLexer = std::move(lx); |
| 201 | 1457 CurPPLexer = CurLexer.get(); |
| 205 | 1458 EnterMacro(Tok, Loc, MI , 0 ); |
| 1459 CurTokenLexer->MacroDefLength = i; | |
| 1460 return Tok; | |
| 197 | 1461 } |
| 1462 | |
| 1463 #endif |