void clang::CacheTokens(Preprocessor &PP, llvm::raw_fd_ostream* OS) { // Get the name of the main file. const SourceManager &SrcMgr = PP.getSourceManager(); const FileEntry *MainFile = SrcMgr.getFileEntryForID(SrcMgr.getMainFileID()); SmallString<128> MainFilePath(MainFile->getName()); llvm::sys::fs::make_absolute(MainFilePath); // Create the PTHWriter. PTHWriter PW(*OS, PP); // Install the 'stat' system call listener in the FileManager. StatListener *StatCache = new StatListener(PW.getPM()); PP.getFileManager().addStatCache(StatCache, /*AtBeginning=*/true); // Lex through the entire file. This will populate SourceManager with // all of the header information. Token Tok; PP.EnterMainSourceFile(); do { PP.Lex(Tok); } while (Tok.isNot(tok::eof)); // Generate the PTH file. PP.getFileManager().removeStatCache(StatCache); PW.GeneratePTH(MainFilePath.str()); }
/// InclusionRewriterInInput - Implement -frewrite-includes mode. void clang::RewriteIncludesInInput(Preprocessor &PP, raw_ostream *OS, const PreprocessorOutputOptions &Opts) { SourceManager &SM = PP.getSourceManager(); InclusionRewriter *Rewrite = new InclusionRewriter(PP, *OS, Opts.ShowLineMarkers); PP.addPPCallbacks(Rewrite); PP.IgnorePragmas(); // First let the preprocessor process the entire file and call callbacks. // Callbacks will record which #include's were actually performed. PP.EnterMainSourceFile(); Token Tok; // Only preprocessor directives matter here, so disable macro expansion // everywhere else as an optimization. // TODO: It would be even faster if the preprocessor could be switched // to a mode where it would parse only preprocessor directives and comments, // nothing else matters for parsing or processing. PP.SetMacroExpansionOnlyInDirectives(); do { PP.Lex(Tok); } while (Tok.isNot(tok::eof)); Rewrite->setPredefinesBuffer(SM.getBuffer(PP.getPredefinesFileID())); Rewrite->Process(PP.getPredefinesFileID(), SrcMgr::C_User); Rewrite->Process(SM.getMainFileID(), SrcMgr::C_User); OS->flush(); }
/// ParseAST - Parse the entire file specified, notifying the ASTConsumer as /// the file is parsed. This inserts the parsed decls into the translation unit /// held by Ctx. /// void clang::ParseAST(Preprocessor &PP, ASTConsumer *Consumer, ASTContext &Ctx, bool PrintStats, bool CompleteTranslationUnit) { // Collect global stats on Decls/Stmts (until we have a module streamer). if (PrintStats) { Decl::CollectingStats(true); Stmt::CollectingStats(true); } Sema S(PP, Ctx, *Consumer, CompleteTranslationUnit); Parser P(PP, S); PP.EnterMainSourceFile(); // Initialize the parser. P.Initialize(); Consumer->Initialize(Ctx); if (SemaConsumer *SC = dyn_cast<SemaConsumer>(Consumer)) SC->InitializeSema(S); if (ExternalASTSource *External = Ctx.getExternalSource()) { if (ExternalSemaSource *ExternalSema = dyn_cast<ExternalSemaSource>(External)) ExternalSema->InitializeSema(S); External->StartTranslationUnit(Consumer); } Parser::DeclGroupPtrTy ADecl; while (!P.ParseTopLevelDecl(ADecl)) { // Not end of file. // If we got a null return and something *was* parsed, ignore it. This // is due to a top-level semicolon, an action override, or a parse error // skipping something. if (ADecl) Consumer->HandleTopLevelDecl(ADecl.getAsVal<DeclGroupRef>()); }; Consumer->HandleTranslationUnit(Ctx); if (PrintStats) { fprintf(stderr, "\nSTATISTICS:\n"); P.getActions().PrintStats(); Ctx.PrintStats(); Decl::PrintStats(); Stmt::PrintStats(); Consumer->PrintStats(); Decl::CollectingStats(false); Stmt::CollectingStats(false); } }
/// ParseAST - Parse the entire file specified, notifying the ASTConsumer as /// the file is parsed. This inserts the parsed decls into the translation unit /// held by Ctx. /// void clang::ParseAST(Preprocessor &PP, ASTConsumer *Consumer, ASTContext &Ctx, bool PrintStats, bool CompleteTranslationUnit, CodeCompleteConsumer *CompletionConsumer) { // Collect global stats on Decls/Stmts (until we have a module streamer). if (PrintStats) { Decl::CollectingStats(true); Stmt::CollectingStats(true); } Sema S(PP, Ctx, *Consumer, CompleteTranslationUnit, CompletionConsumer); Parser P(PP, S); PP.EnterMainSourceFile(); // Initialize the parser. P.Initialize(); Consumer->Initialize(Ctx); if (SemaConsumer *SC = dyn_cast<SemaConsumer>(Consumer)) SC->InitializeSema(S); if (ExternalASTSource *External = Ctx.getExternalSource()) { if (ExternalSemaSource *ExternalSema = dyn_cast<ExternalSemaSource>(External)) ExternalSema->InitializeSema(S); External->StartTranslationUnit(Consumer); } Parser::DeclGroupPtrTy ADecl; while (!P.ParseTopLevelDecl(ADecl)) { // Not end of file. // If we got a null return and something *was* parsed, ignore it. This // is due to a top-level semicolon, an action override, or a parse error // skipping something. if (ADecl) Consumer->HandleTopLevelDecl(ADecl.getAsVal<DeclGroupRef>()); }; // Check for any pending objective-c implementation decl. while ((ADecl = P.FinishPendingObjCActions())) Consumer->HandleTopLevelDecl(ADecl.getAsVal<DeclGroupRef>()); // Process any TopLevelDecls generated by #pragma weak. for (llvm::SmallVector<Decl*,2>::iterator I = S.WeakTopLevelDecls().begin(), E = S.WeakTopLevelDecls().end(); I != E; ++I) Consumer->HandleTopLevelDecl(DeclGroupRef(*I)); // Dump record layouts, if requested. if (PP.getLangOptions().DumpRecordLayouts) DumpRecordLayouts(Ctx); Consumer->HandleTranslationUnit(Ctx); if (ExternalSemaSource *ESS = dyn_cast_or_null<ExternalSemaSource>(Ctx.getExternalSource())) ESS->ForgetSema(); if (SemaConsumer *SC = dyn_cast<SemaConsumer>(Consumer)) SC->ForgetSema(); if (PrintStats) { fprintf(stderr, "\nSTATISTICS:\n"); P.getActions().PrintStats(); Ctx.PrintStats(); Decl::PrintStats(); Stmt::PrintStats(); Consumer->PrintStats(); } }
/// RewriteMacrosInInput - Implement -rewrite-macros mode. void clang::RewriteMacrosInInput(Preprocessor &PP, raw_ostream *OS) { SourceManager &SM = PP.getSourceManager(); Rewriter Rewrite; Rewrite.setSourceMgr(SM, PP.getLangOpts()); RewriteBuffer &RB = Rewrite.getEditBuffer(SM.getMainFileID()); std::vector<Token> RawTokens; LexRawTokensFromMainFile(PP, RawTokens); unsigned CurRawTok = 0; Token RawTok = GetNextRawTok(RawTokens, CurRawTok, false); // Get the first preprocessing token. PP.EnterMainSourceFile(); Token PPTok; PP.Lex(PPTok); // Preprocess the input file in parallel with raw lexing the main file. Ignore // all tokens that are preprocessed from a file other than the main file (e.g. // a header). If we see tokens that are in the preprocessed file but not the // lexed file, we have a macro expansion. If we see tokens in the lexed file // that aren't in the preprocessed view, we have macros that expand to no // tokens, or macro arguments etc. while (RawTok.isNot(tok::eof) || PPTok.isNot(tok::eof)) { SourceLocation PPLoc = SM.getExpansionLoc(PPTok.getLocation()); // If PPTok is from a different source file, ignore it. if (!SM.isFromMainFile(PPLoc)) { PP.Lex(PPTok); continue; } // If the raw file hits a preprocessor directive, they will be extra tokens // in the raw file that don't exist in the preprocsesed file. However, we // choose to preserve them in the output file and otherwise handle them // specially. if (RawTok.is(tok::hash) && RawTok.isAtStartOfLine()) { // If this is a #warning directive or #pragma mark (GNU extensions), // comment the line out. if (RawTokens[CurRawTok].is(tok::identifier)) { const IdentifierInfo *II = RawTokens[CurRawTok].getIdentifierInfo(); if (II->getName() == "warning") { // Comment out #warning. RB.InsertTextAfter(SM.getFileOffset(RawTok.getLocation()), "//"); } else if (II->getName() == "pragma" && RawTokens[CurRawTok+1].is(tok::identifier) && (RawTokens[CurRawTok+1].getIdentifierInfo()->getName() == "mark")) { // Comment out #pragma mark. RB.InsertTextAfter(SM.getFileOffset(RawTok.getLocation()), "//"); } } // Otherwise, if this is a #include or some other directive, just leave it // in the file by skipping over the line. RawTok = GetNextRawTok(RawTokens, CurRawTok, false); while (!RawTok.isAtStartOfLine() && RawTok.isNot(tok::eof)) RawTok = GetNextRawTok(RawTokens, CurRawTok, false); continue; } // Okay, both tokens are from the same file. Get their offsets from the // start of the file. unsigned PPOffs = SM.getFileOffset(PPLoc); unsigned RawOffs = SM.getFileOffset(RawTok.getLocation()); // If the offsets are the same and the token kind is the same, ignore them. if (PPOffs == RawOffs && isSameToken(RawTok, PPTok)) { RawTok = GetNextRawTok(RawTokens, CurRawTok, false); PP.Lex(PPTok); continue; } // If the PP token is farther along than the raw token, something was // deleted. Comment out the raw token. if (RawOffs <= PPOffs) { // Comment out a whole run of tokens instead of bracketing each one with // comments. Add a leading space if RawTok didn't have one. bool HasSpace = RawTok.hasLeadingSpace(); RB.InsertTextAfter(RawOffs, &" /*"[HasSpace]); unsigned EndPos; do { EndPos = RawOffs+RawTok.getLength(); RawTok = GetNextRawTok(RawTokens, CurRawTok, true); RawOffs = SM.getFileOffset(RawTok.getLocation()); if (RawTok.is(tok::comment)) { // Skip past the comment. RawTok = GetNextRawTok(RawTokens, CurRawTok, false); break; } } while (RawOffs <= PPOffs && !RawTok.isAtStartOfLine() && (PPOffs != RawOffs || !isSameToken(RawTok, PPTok))); RB.InsertTextBefore(EndPos, "*/"); continue; } // Otherwise, there was a replacement an expansion. Insert the new token // in the output buffer. Insert the whole run of new tokens at once to get // them in the right order. unsigned InsertPos = PPOffs; std::string Expansion; while (PPOffs < RawOffs) { Expansion += ' ' + PP.getSpelling(PPTok); PP.Lex(PPTok); PPLoc = SM.getExpansionLoc(PPTok.getLocation()); PPOffs = SM.getFileOffset(PPLoc); } Expansion += ' '; RB.InsertTextBefore(InsertPos, Expansion); } // Get the buffer corresponding to MainFileID. If we haven't changed it, then // we are done. if (const RewriteBuffer *RewriteBuf = Rewrite.getRewriteBufferFor(SM.getMainFileID())) { //printf("Changed:\n"); *OS << std::string(RewriteBuf->begin(), RewriteBuf->end()); } else { fprintf(stderr, "No changes\n"); } OS->flush(); }