10b57cec5SDimitry Andric //===--- InclusionRewriter.cpp - Rewrite includes into their expansions ---===// 20b57cec5SDimitry Andric // 30b57cec5SDimitry Andric // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. 40b57cec5SDimitry Andric // See https://llvm.org/LICENSE.txt for license information. 50b57cec5SDimitry Andric // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception 60b57cec5SDimitry Andric // 70b57cec5SDimitry Andric //===----------------------------------------------------------------------===// 80b57cec5SDimitry Andric // 90b57cec5SDimitry Andric // This code rewrites include invocations into their expansions. This gives you 100b57cec5SDimitry Andric // a file with all included files merged into it. 110b57cec5SDimitry Andric // 120b57cec5SDimitry Andric //===----------------------------------------------------------------------===// 130b57cec5SDimitry Andric 140b57cec5SDimitry Andric #include "clang/Rewrite/Frontend/Rewriters.h" 150b57cec5SDimitry Andric #include "clang/Basic/SourceManager.h" 160b57cec5SDimitry Andric #include "clang/Frontend/PreprocessorOutputOptions.h" 170b57cec5SDimitry Andric #include "clang/Lex/Pragma.h" 180b57cec5SDimitry Andric #include "clang/Lex/Preprocessor.h" 190b57cec5SDimitry Andric #include "llvm/ADT/SmallString.h" 200b57cec5SDimitry Andric #include "llvm/Support/raw_ostream.h" 21bdd1243dSDimitry Andric #include <optional> 220b57cec5SDimitry Andric 230b57cec5SDimitry Andric using namespace clang; 240b57cec5SDimitry Andric using namespace llvm; 250b57cec5SDimitry Andric 260b57cec5SDimitry Andric namespace { 270b57cec5SDimitry Andric 280b57cec5SDimitry Andric class InclusionRewriter : public PPCallbacks { 290b57cec5SDimitry Andric /// Information about which #includes were actually performed, 300b57cec5SDimitry Andric /// created by preprocessor callbacks. 310b57cec5SDimitry Andric struct IncludedFile { 320b57cec5SDimitry Andric FileID Id; 330b57cec5SDimitry Andric SrcMgr::CharacteristicKind FileType; 3404eeddc0SDimitry Andric IncludedFile(FileID Id, SrcMgr::CharacteristicKind FileType) 3504eeddc0SDimitry Andric : Id(Id), FileType(FileType) {} 360b57cec5SDimitry Andric }; 370b57cec5SDimitry Andric Preprocessor &PP; ///< Used to find inclusion directives. 380b57cec5SDimitry Andric SourceManager &SM; ///< Used to read and manage source files. 390b57cec5SDimitry Andric raw_ostream &OS; ///< The destination stream for rewritten contents. 400b57cec5SDimitry Andric StringRef MainEOL; ///< The line ending marker to use. 41e8d8bef9SDimitry Andric llvm::MemoryBufferRef PredefinesBuffer; ///< The preprocessor predefines. 420b57cec5SDimitry Andric bool ShowLineMarkers; ///< Show #line markers. 430b57cec5SDimitry Andric bool UseLineDirectives; ///< Use of line directives or line markers. 440b57cec5SDimitry Andric /// Tracks where inclusions that change the file are found. 45e8d8bef9SDimitry Andric std::map<SourceLocation, IncludedFile> FileIncludes; 460b57cec5SDimitry Andric /// Tracks where inclusions that import modules are found. 47e8d8bef9SDimitry Andric std::map<SourceLocation, const Module *> ModuleIncludes; 480b57cec5SDimitry Andric /// Tracks where inclusions that enter modules (in a module build) are found. 49e8d8bef9SDimitry Andric std::map<SourceLocation, const Module *> ModuleEntryIncludes; 50a7dea167SDimitry Andric /// Tracks where #if and #elif directives get evaluated and whether to true. 51e8d8bef9SDimitry Andric std::map<SourceLocation, bool> IfConditions; 520b57cec5SDimitry Andric /// Used transitively for building up the FileIncludes mapping over the 530b57cec5SDimitry Andric /// various \c PPCallbacks callbacks. 540b57cec5SDimitry Andric SourceLocation LastInclusionLocation; 550b57cec5SDimitry Andric public: 560b57cec5SDimitry Andric InclusionRewriter(Preprocessor &PP, raw_ostream &OS, bool ShowLineMarkers, 570b57cec5SDimitry Andric bool UseLineDirectives); 5804eeddc0SDimitry Andric void Process(FileID FileId, SrcMgr::CharacteristicKind FileType); 59e8d8bef9SDimitry Andric void setPredefinesBuffer(const llvm::MemoryBufferRef &Buf) { 600b57cec5SDimitry Andric PredefinesBuffer = Buf; 610b57cec5SDimitry Andric } 620b57cec5SDimitry Andric void detectMainFileEOL(); 630b57cec5SDimitry Andric void handleModuleBegin(Token &Tok) { 640b57cec5SDimitry Andric assert(Tok.getKind() == tok::annot_module_begin); 65e8d8bef9SDimitry Andric ModuleEntryIncludes.insert( 66e8d8bef9SDimitry Andric {Tok.getLocation(), (Module *)Tok.getAnnotationValue()}); 670b57cec5SDimitry Andric } 680b57cec5SDimitry Andric private: 690b57cec5SDimitry Andric void FileChanged(SourceLocation Loc, FileChangeReason Reason, 700b57cec5SDimitry Andric SrcMgr::CharacteristicKind FileType, 710b57cec5SDimitry Andric FileID PrevFID) override; 72a7dea167SDimitry Andric void FileSkipped(const FileEntryRef &SkippedFile, const Token &FilenameTok, 730b57cec5SDimitry Andric SrcMgr::CharacteristicKind FileType) override; 740b57cec5SDimitry Andric void InclusionDirective(SourceLocation HashLoc, const Token &IncludeTok, 750b57cec5SDimitry Andric StringRef FileName, bool IsAngled, 7681ad6265SDimitry Andric CharSourceRange FilenameRange, 77bdd1243dSDimitry Andric OptionalFileEntryRef File, StringRef SearchPath, 7881ad6265SDimitry Andric StringRef RelativePath, const Module *Imported, 790b57cec5SDimitry Andric SrcMgr::CharacteristicKind FileType) override; 80a7dea167SDimitry Andric void If(SourceLocation Loc, SourceRange ConditionRange, 81a7dea167SDimitry Andric ConditionValueKind ConditionValue) override; 82a7dea167SDimitry Andric void Elif(SourceLocation Loc, SourceRange ConditionRange, 83a7dea167SDimitry Andric ConditionValueKind ConditionValue, SourceLocation IfLoc) override; 840b57cec5SDimitry Andric void WriteLineInfo(StringRef Filename, int Line, 850b57cec5SDimitry Andric SrcMgr::CharacteristicKind FileType, 860b57cec5SDimitry Andric StringRef Extra = StringRef()); 870b57cec5SDimitry Andric void WriteImplicitModuleImport(const Module *Mod); 88e8d8bef9SDimitry Andric void OutputContentUpTo(const MemoryBufferRef &FromFile, unsigned &WriteFrom, 89e8d8bef9SDimitry Andric unsigned WriteTo, StringRef EOL, int &lines, 900b57cec5SDimitry Andric bool EnsureNewline); 910b57cec5SDimitry Andric void CommentOutDirective(Lexer &DirectivesLex, const Token &StartToken, 92e8d8bef9SDimitry Andric const MemoryBufferRef &FromFile, StringRef EOL, 93*5f757f3fSDimitry Andric unsigned &NextToWrite, int &Lines, 94*5f757f3fSDimitry Andric const IncludedFile *Inc = nullptr); 950b57cec5SDimitry Andric const IncludedFile *FindIncludeAtLocation(SourceLocation Loc) const; 96*5f757f3fSDimitry Andric StringRef getIncludedFileName(const IncludedFile *Inc) const; 970b57cec5SDimitry Andric const Module *FindModuleAtLocation(SourceLocation Loc) const; 980b57cec5SDimitry Andric const Module *FindEnteredModule(SourceLocation Loc) const; 99a7dea167SDimitry Andric bool IsIfAtLocationTrue(SourceLocation Loc) const; 1000b57cec5SDimitry Andric StringRef NextIdentifierName(Lexer &RawLex, Token &RawToken); 1010b57cec5SDimitry Andric }; 1020b57cec5SDimitry Andric 1030b57cec5SDimitry Andric } // end anonymous namespace 1040b57cec5SDimitry Andric 1050b57cec5SDimitry Andric /// Initializes an InclusionRewriter with a \p PP source and \p OS destination. 1060b57cec5SDimitry Andric InclusionRewriter::InclusionRewriter(Preprocessor &PP, raw_ostream &OS, 1070b57cec5SDimitry Andric bool ShowLineMarkers, 1080b57cec5SDimitry Andric bool UseLineDirectives) 1090b57cec5SDimitry Andric : PP(PP), SM(PP.getSourceManager()), OS(OS), MainEOL("\n"), 110e8d8bef9SDimitry Andric ShowLineMarkers(ShowLineMarkers), UseLineDirectives(UseLineDirectives), 1110b57cec5SDimitry Andric LastInclusionLocation(SourceLocation()) {} 1120b57cec5SDimitry Andric 1130b57cec5SDimitry Andric /// Write appropriate line information as either #line directives or GNU line 1140b57cec5SDimitry Andric /// markers depending on what mode we're in, including the \p Filename and 1150b57cec5SDimitry Andric /// \p Line we are located at, using the specified \p EOL line separator, and 1160b57cec5SDimitry Andric /// any \p Extra context specifiers in GNU line directives. 1170b57cec5SDimitry Andric void InclusionRewriter::WriteLineInfo(StringRef Filename, int Line, 1180b57cec5SDimitry Andric SrcMgr::CharacteristicKind FileType, 1190b57cec5SDimitry Andric StringRef Extra) { 1200b57cec5SDimitry Andric if (!ShowLineMarkers) 1210b57cec5SDimitry Andric return; 1220b57cec5SDimitry Andric if (UseLineDirectives) { 1230b57cec5SDimitry Andric OS << "#line" << ' ' << Line << ' ' << '"'; 1240b57cec5SDimitry Andric OS.write_escaped(Filename); 1250b57cec5SDimitry Andric OS << '"'; 1260b57cec5SDimitry Andric } else { 1270b57cec5SDimitry Andric // Use GNU linemarkers as described here: 1280b57cec5SDimitry Andric // http://gcc.gnu.org/onlinedocs/cpp/Preprocessor-Output.html 1290b57cec5SDimitry Andric OS << '#' << ' ' << Line << ' ' << '"'; 1300b57cec5SDimitry Andric OS.write_escaped(Filename); 1310b57cec5SDimitry Andric OS << '"'; 1320b57cec5SDimitry Andric if (!Extra.empty()) 1330b57cec5SDimitry Andric OS << Extra; 1340b57cec5SDimitry Andric if (FileType == SrcMgr::C_System) 1350b57cec5SDimitry Andric // "`3' This indicates that the following text comes from a system header 1360b57cec5SDimitry Andric // file, so certain warnings should be suppressed." 1370b57cec5SDimitry Andric OS << " 3"; 1380b57cec5SDimitry Andric else if (FileType == SrcMgr::C_ExternCSystem) 1390b57cec5SDimitry Andric // as above for `3', plus "`4' This indicates that the following text 1400b57cec5SDimitry Andric // should be treated as being wrapped in an implicit extern "C" block." 1410b57cec5SDimitry Andric OS << " 3 4"; 1420b57cec5SDimitry Andric } 1430b57cec5SDimitry Andric OS << MainEOL; 1440b57cec5SDimitry Andric } 1450b57cec5SDimitry Andric 1460b57cec5SDimitry Andric void InclusionRewriter::WriteImplicitModuleImport(const Module *Mod) { 1470b57cec5SDimitry Andric OS << "#pragma clang module import " << Mod->getFullModuleName(true) 1480b57cec5SDimitry Andric << " /* clang -frewrite-includes: implicit import */" << MainEOL; 1490b57cec5SDimitry Andric } 1500b57cec5SDimitry Andric 1510b57cec5SDimitry Andric /// FileChanged - Whenever the preprocessor enters or exits a #include file 1520b57cec5SDimitry Andric /// it invokes this handler. 1530b57cec5SDimitry Andric void InclusionRewriter::FileChanged(SourceLocation Loc, 1540b57cec5SDimitry Andric FileChangeReason Reason, 1550b57cec5SDimitry Andric SrcMgr::CharacteristicKind NewFileType, 1560b57cec5SDimitry Andric FileID) { 1570b57cec5SDimitry Andric if (Reason != EnterFile) 1580b57cec5SDimitry Andric return; 1590b57cec5SDimitry Andric if (LastInclusionLocation.isInvalid()) 1600b57cec5SDimitry Andric // we didn't reach this file (eg: the main file) via an inclusion directive 1610b57cec5SDimitry Andric return; 1620b57cec5SDimitry Andric FileID Id = FullSourceLoc(Loc, SM).getFileID(); 1630b57cec5SDimitry Andric auto P = FileIncludes.insert( 16404eeddc0SDimitry Andric std::make_pair(LastInclusionLocation, IncludedFile(Id, NewFileType))); 1650b57cec5SDimitry Andric (void)P; 1660b57cec5SDimitry Andric assert(P.second && "Unexpected revisitation of the same include directive"); 1670b57cec5SDimitry Andric LastInclusionLocation = SourceLocation(); 1680b57cec5SDimitry Andric } 1690b57cec5SDimitry Andric 1700b57cec5SDimitry Andric /// Called whenever an inclusion is skipped due to canonical header protection 1710b57cec5SDimitry Andric /// macros. 172a7dea167SDimitry Andric void InclusionRewriter::FileSkipped(const FileEntryRef & /*SkippedFile*/, 1730b57cec5SDimitry Andric const Token & /*FilenameTok*/, 1740b57cec5SDimitry Andric SrcMgr::CharacteristicKind /*FileType*/) { 1750b57cec5SDimitry Andric assert(LastInclusionLocation.isValid() && 1760b57cec5SDimitry Andric "A file, that wasn't found via an inclusion directive, was skipped"); 1770b57cec5SDimitry Andric LastInclusionLocation = SourceLocation(); 1780b57cec5SDimitry Andric } 1790b57cec5SDimitry Andric 1800b57cec5SDimitry Andric /// This should be called whenever the preprocessor encounters include 1810b57cec5SDimitry Andric /// directives. It does not say whether the file has been included, but it 1820b57cec5SDimitry Andric /// provides more information about the directive (hash location instead 1830b57cec5SDimitry Andric /// of location inside the included file). It is assumed that the matching 1840b57cec5SDimitry Andric /// FileChanged() or FileSkipped() is called after this (or neither is 1850b57cec5SDimitry Andric /// called if this #include results in an error or does not textually include 1860b57cec5SDimitry Andric /// anything). 187bdd1243dSDimitry Andric void InclusionRewriter::InclusionDirective( 188bdd1243dSDimitry Andric SourceLocation HashLoc, const Token & /*IncludeTok*/, 189bdd1243dSDimitry Andric StringRef /*FileName*/, bool /*IsAngled*/, 190bdd1243dSDimitry Andric CharSourceRange /*FilenameRange*/, OptionalFileEntryRef /*File*/, 191bdd1243dSDimitry Andric StringRef /*SearchPath*/, StringRef /*RelativePath*/, 192bdd1243dSDimitry Andric const Module *Imported, SrcMgr::CharacteristicKind FileType) { 1930b57cec5SDimitry Andric if (Imported) { 194e8d8bef9SDimitry Andric auto P = ModuleIncludes.insert(std::make_pair(HashLoc, Imported)); 1950b57cec5SDimitry Andric (void)P; 1960b57cec5SDimitry Andric assert(P.second && "Unexpected revisitation of the same include directive"); 1970b57cec5SDimitry Andric } else 1980b57cec5SDimitry Andric LastInclusionLocation = HashLoc; 1990b57cec5SDimitry Andric } 2000b57cec5SDimitry Andric 201a7dea167SDimitry Andric void InclusionRewriter::If(SourceLocation Loc, SourceRange ConditionRange, 202a7dea167SDimitry Andric ConditionValueKind ConditionValue) { 203e8d8bef9SDimitry Andric auto P = IfConditions.insert(std::make_pair(Loc, ConditionValue == CVK_True)); 204a7dea167SDimitry Andric (void)P; 205a7dea167SDimitry Andric assert(P.second && "Unexpected revisitation of the same if directive"); 206a7dea167SDimitry Andric } 207a7dea167SDimitry Andric 208a7dea167SDimitry Andric void InclusionRewriter::Elif(SourceLocation Loc, SourceRange ConditionRange, 209a7dea167SDimitry Andric ConditionValueKind ConditionValue, 210a7dea167SDimitry Andric SourceLocation IfLoc) { 211e8d8bef9SDimitry Andric auto P = IfConditions.insert(std::make_pair(Loc, ConditionValue == CVK_True)); 212a7dea167SDimitry Andric (void)P; 213a7dea167SDimitry Andric assert(P.second && "Unexpected revisitation of the same elif directive"); 214a7dea167SDimitry Andric } 215a7dea167SDimitry Andric 2160b57cec5SDimitry Andric /// Simple lookup for a SourceLocation (specifically one denoting the hash in 2170b57cec5SDimitry Andric /// an inclusion directive) in the map of inclusion information, FileChanges. 2180b57cec5SDimitry Andric const InclusionRewriter::IncludedFile * 2190b57cec5SDimitry Andric InclusionRewriter::FindIncludeAtLocation(SourceLocation Loc) const { 220e8d8bef9SDimitry Andric const auto I = FileIncludes.find(Loc); 2210b57cec5SDimitry Andric if (I != FileIncludes.end()) 2220b57cec5SDimitry Andric return &I->second; 2230b57cec5SDimitry Andric return nullptr; 2240b57cec5SDimitry Andric } 2250b57cec5SDimitry Andric 2260b57cec5SDimitry Andric /// Simple lookup for a SourceLocation (specifically one denoting the hash in 2270b57cec5SDimitry Andric /// an inclusion directive) in the map of module inclusion information. 2280b57cec5SDimitry Andric const Module * 2290b57cec5SDimitry Andric InclusionRewriter::FindModuleAtLocation(SourceLocation Loc) const { 230e8d8bef9SDimitry Andric const auto I = ModuleIncludes.find(Loc); 2310b57cec5SDimitry Andric if (I != ModuleIncludes.end()) 2320b57cec5SDimitry Andric return I->second; 2330b57cec5SDimitry Andric return nullptr; 2340b57cec5SDimitry Andric } 2350b57cec5SDimitry Andric 2360b57cec5SDimitry Andric /// Simple lookup for a SourceLocation (specifically one denoting the hash in 2370b57cec5SDimitry Andric /// an inclusion directive) in the map of module entry information. 2380b57cec5SDimitry Andric const Module * 2390b57cec5SDimitry Andric InclusionRewriter::FindEnteredModule(SourceLocation Loc) const { 240e8d8bef9SDimitry Andric const auto I = ModuleEntryIncludes.find(Loc); 2410b57cec5SDimitry Andric if (I != ModuleEntryIncludes.end()) 2420b57cec5SDimitry Andric return I->second; 2430b57cec5SDimitry Andric return nullptr; 2440b57cec5SDimitry Andric } 2450b57cec5SDimitry Andric 246a7dea167SDimitry Andric bool InclusionRewriter::IsIfAtLocationTrue(SourceLocation Loc) const { 247e8d8bef9SDimitry Andric const auto I = IfConditions.find(Loc); 248a7dea167SDimitry Andric if (I != IfConditions.end()) 249a7dea167SDimitry Andric return I->second; 250a7dea167SDimitry Andric return false; 251a7dea167SDimitry Andric } 252a7dea167SDimitry Andric 2530b57cec5SDimitry Andric void InclusionRewriter::detectMainFileEOL() { 254bdd1243dSDimitry Andric std::optional<MemoryBufferRef> FromFile = 255bdd1243dSDimitry Andric *SM.getBufferOrNone(SM.getMainFileID()); 256e8d8bef9SDimitry Andric assert(FromFile); 257e8d8bef9SDimitry Andric if (!FromFile) 2580b57cec5SDimitry Andric return; // Should never happen, but whatever. 25904eeddc0SDimitry Andric MainEOL = FromFile->getBuffer().detectEOL(); 2600b57cec5SDimitry Andric } 2610b57cec5SDimitry Andric 2620b57cec5SDimitry Andric /// Writes out bytes from \p FromFile, starting at \p NextToWrite and ending at 2630b57cec5SDimitry Andric /// \p WriteTo - 1. 264e8d8bef9SDimitry Andric void InclusionRewriter::OutputContentUpTo(const MemoryBufferRef &FromFile, 2650b57cec5SDimitry Andric unsigned &WriteFrom, unsigned WriteTo, 2660b57cec5SDimitry Andric StringRef LocalEOL, int &Line, 2670b57cec5SDimitry Andric bool EnsureNewline) { 2680b57cec5SDimitry Andric if (WriteTo <= WriteFrom) 2690b57cec5SDimitry Andric return; 270e8d8bef9SDimitry Andric if (FromFile == PredefinesBuffer) { 2710b57cec5SDimitry Andric // Ignore the #defines of the predefines buffer. 2720b57cec5SDimitry Andric WriteFrom = WriteTo; 2730b57cec5SDimitry Andric return; 2740b57cec5SDimitry Andric } 2750b57cec5SDimitry Andric 2760b57cec5SDimitry Andric // If we would output half of a line ending, advance one character to output 2770b57cec5SDimitry Andric // the whole line ending. All buffers are null terminated, so looking ahead 2780b57cec5SDimitry Andric // one byte is safe. 2790b57cec5SDimitry Andric if (LocalEOL.size() == 2 && 2800b57cec5SDimitry Andric LocalEOL[0] == (FromFile.getBufferStart() + WriteTo)[-1] && 2810b57cec5SDimitry Andric LocalEOL[1] == (FromFile.getBufferStart() + WriteTo)[0]) 2820b57cec5SDimitry Andric WriteTo++; 2830b57cec5SDimitry Andric 2840b57cec5SDimitry Andric StringRef TextToWrite(FromFile.getBufferStart() + WriteFrom, 2850b57cec5SDimitry Andric WriteTo - WriteFrom); 286bdd1243dSDimitry Andric // count lines manually, it's faster than getPresumedLoc() 287bdd1243dSDimitry Andric Line += TextToWrite.count(LocalEOL); 2880b57cec5SDimitry Andric 2890b57cec5SDimitry Andric if (MainEOL == LocalEOL) { 2900b57cec5SDimitry Andric OS << TextToWrite; 2910b57cec5SDimitry Andric } else { 2920b57cec5SDimitry Andric // Output the file one line at a time, rewriting the line endings as we go. 2930b57cec5SDimitry Andric StringRef Rest = TextToWrite; 2940b57cec5SDimitry Andric while (!Rest.empty()) { 295bdd1243dSDimitry Andric // Identify and output the next line excluding an EOL sequence if present. 296bdd1243dSDimitry Andric size_t Idx = Rest.find(LocalEOL); 297bdd1243dSDimitry Andric StringRef LineText = Rest.substr(0, Idx); 2980b57cec5SDimitry Andric OS << LineText; 299bdd1243dSDimitry Andric if (Idx != StringRef::npos) { 300bdd1243dSDimitry Andric // An EOL sequence was present, output the EOL sequence for the 301bdd1243dSDimitry Andric // main source file and skip past the local EOL sequence. 3020b57cec5SDimitry Andric OS << MainEOL; 303bdd1243dSDimitry Andric Idx += LocalEOL.size(); 3040b57cec5SDimitry Andric } 305bdd1243dSDimitry Andric // Strip the line just handled. If Idx is npos or matches the end of the 306bdd1243dSDimitry Andric // text, Rest will be set to an empty string and the loop will terminate. 307bdd1243dSDimitry Andric Rest = Rest.substr(Idx); 308bdd1243dSDimitry Andric } 309bdd1243dSDimitry Andric } 310*5f757f3fSDimitry Andric if (EnsureNewline && !TextToWrite.ends_with(LocalEOL)) 3110b57cec5SDimitry Andric OS << MainEOL; 312bdd1243dSDimitry Andric 3130b57cec5SDimitry Andric WriteFrom = WriteTo; 3140b57cec5SDimitry Andric } 3150b57cec5SDimitry Andric 316*5f757f3fSDimitry Andric StringRef 317*5f757f3fSDimitry Andric InclusionRewriter::getIncludedFileName(const IncludedFile *Inc) const { 318*5f757f3fSDimitry Andric if (Inc) { 319*5f757f3fSDimitry Andric auto B = SM.getBufferOrNone(Inc->Id); 320*5f757f3fSDimitry Andric assert(B && "Attempting to process invalid inclusion"); 321*5f757f3fSDimitry Andric if (B) 322*5f757f3fSDimitry Andric return llvm::sys::path::filename(B->getBufferIdentifier()); 323*5f757f3fSDimitry Andric } 324*5f757f3fSDimitry Andric return StringRef(); 325*5f757f3fSDimitry Andric } 326*5f757f3fSDimitry Andric 3270b57cec5SDimitry Andric /// Print characters from \p FromFile starting at \p NextToWrite up until the 3280b57cec5SDimitry Andric /// inclusion directive at \p StartToken, then print out the inclusion 3290b57cec5SDimitry Andric /// inclusion directive disabled by a #if directive, updating \p NextToWrite 3300b57cec5SDimitry Andric /// and \p Line to track the number of source lines visited and the progress 3310b57cec5SDimitry Andric /// through the \p FromFile buffer. 3320b57cec5SDimitry Andric void InclusionRewriter::CommentOutDirective(Lexer &DirectiveLex, 3330b57cec5SDimitry Andric const Token &StartToken, 334e8d8bef9SDimitry Andric const MemoryBufferRef &FromFile, 3350b57cec5SDimitry Andric StringRef LocalEOL, 336*5f757f3fSDimitry Andric unsigned &NextToWrite, int &Line, 337*5f757f3fSDimitry Andric const IncludedFile *Inc) { 3380b57cec5SDimitry Andric OutputContentUpTo(FromFile, NextToWrite, 3390b57cec5SDimitry Andric SM.getFileOffset(StartToken.getLocation()), LocalEOL, Line, 3400b57cec5SDimitry Andric false); 3410b57cec5SDimitry Andric Token DirectiveToken; 3420b57cec5SDimitry Andric do { 3430b57cec5SDimitry Andric DirectiveLex.LexFromRawLexer(DirectiveToken); 3440b57cec5SDimitry Andric } while (!DirectiveToken.is(tok::eod) && DirectiveToken.isNot(tok::eof)); 345e8d8bef9SDimitry Andric if (FromFile == PredefinesBuffer) { 3460b57cec5SDimitry Andric // OutputContentUpTo() would not output anything anyway. 3470b57cec5SDimitry Andric return; 3480b57cec5SDimitry Andric } 349*5f757f3fSDimitry Andric if (Inc) { 350*5f757f3fSDimitry Andric OS << "#if defined(__CLANG_REWRITTEN_INCLUDES) "; 351*5f757f3fSDimitry Andric if (isSystem(Inc->FileType)) 352*5f757f3fSDimitry Andric OS << "|| defined(__CLANG_REWRITTEN_SYSTEM_INCLUDES) "; 353*5f757f3fSDimitry Andric OS << "/* " << getIncludedFileName(Inc); 354*5f757f3fSDimitry Andric } else { 355*5f757f3fSDimitry Andric OS << "#if 0 /*"; 356*5f757f3fSDimitry Andric } 357*5f757f3fSDimitry Andric OS << " expanded by -frewrite-includes */" << MainEOL; 3580b57cec5SDimitry Andric OutputContentUpTo(FromFile, NextToWrite, 3590b57cec5SDimitry Andric SM.getFileOffset(DirectiveToken.getLocation()) + 3600b57cec5SDimitry Andric DirectiveToken.getLength(), 3610b57cec5SDimitry Andric LocalEOL, Line, true); 362*5f757f3fSDimitry Andric OS << (Inc ? "#else /* " : "#endif /*") << getIncludedFileName(Inc) 363*5f757f3fSDimitry Andric << " expanded by -frewrite-includes */" << MainEOL; 3640b57cec5SDimitry Andric } 3650b57cec5SDimitry Andric 3660b57cec5SDimitry Andric /// Find the next identifier in the pragma directive specified by \p RawToken. 3670b57cec5SDimitry Andric StringRef InclusionRewriter::NextIdentifierName(Lexer &RawLex, 3680b57cec5SDimitry Andric Token &RawToken) { 3690b57cec5SDimitry Andric RawLex.LexFromRawLexer(RawToken); 3700b57cec5SDimitry Andric if (RawToken.is(tok::raw_identifier)) 3710b57cec5SDimitry Andric PP.LookUpIdentifierInfo(RawToken); 3720b57cec5SDimitry Andric if (RawToken.is(tok::identifier)) 3730b57cec5SDimitry Andric return RawToken.getIdentifierInfo()->getName(); 3740b57cec5SDimitry Andric return StringRef(); 3750b57cec5SDimitry Andric } 3760b57cec5SDimitry Andric 3770b57cec5SDimitry Andric /// Use a raw lexer to analyze \p FileId, incrementally copying parts of it 3780b57cec5SDimitry Andric /// and including content of included files recursively. 3790b57cec5SDimitry Andric void InclusionRewriter::Process(FileID FileId, 38004eeddc0SDimitry Andric SrcMgr::CharacteristicKind FileType) { 381e8d8bef9SDimitry Andric MemoryBufferRef FromFile; 382e8d8bef9SDimitry Andric { 383e8d8bef9SDimitry Andric auto B = SM.getBufferOrNone(FileId); 384e8d8bef9SDimitry Andric assert(B && "Attempting to process invalid inclusion"); 385e8d8bef9SDimitry Andric if (B) 386e8d8bef9SDimitry Andric FromFile = *B; 387e8d8bef9SDimitry Andric } 3880b57cec5SDimitry Andric StringRef FileName = FromFile.getBufferIdentifier(); 389e8d8bef9SDimitry Andric Lexer RawLex(FileId, FromFile, PP.getSourceManager(), PP.getLangOpts()); 3900b57cec5SDimitry Andric RawLex.SetCommentRetentionState(false); 3910b57cec5SDimitry Andric 39204eeddc0SDimitry Andric StringRef LocalEOL = FromFile.getBuffer().detectEOL(); 3930b57cec5SDimitry Andric 3940b57cec5SDimitry Andric // Per the GNU docs: "1" indicates entering a new file. 3950b57cec5SDimitry Andric if (FileId == SM.getMainFileID() || FileId == PP.getPredefinesFileID()) 3960b57cec5SDimitry Andric WriteLineInfo(FileName, 1, FileType, ""); 3970b57cec5SDimitry Andric else 3980b57cec5SDimitry Andric WriteLineInfo(FileName, 1, FileType, " 1"); 3990b57cec5SDimitry Andric 4000b57cec5SDimitry Andric if (SM.getFileIDSize(FileId) == 0) 4010b57cec5SDimitry Andric return; 4020b57cec5SDimitry Andric 4030b57cec5SDimitry Andric // The next byte to be copied from the source file, which may be non-zero if 4040b57cec5SDimitry Andric // the lexer handled a BOM. 4050b57cec5SDimitry Andric unsigned NextToWrite = SM.getFileOffset(RawLex.getSourceLocation()); 4060b57cec5SDimitry Andric assert(SM.getLineNumber(FileId, NextToWrite) == 1); 4070b57cec5SDimitry Andric int Line = 1; // The current input file line number. 4080b57cec5SDimitry Andric 4090b57cec5SDimitry Andric Token RawToken; 4100b57cec5SDimitry Andric RawLex.LexFromRawLexer(RawToken); 4110b57cec5SDimitry Andric 4120b57cec5SDimitry Andric // TODO: Consider adding a switch that strips possibly unimportant content, 4130b57cec5SDimitry Andric // such as comments, to reduce the size of repro files. 4140b57cec5SDimitry Andric while (RawToken.isNot(tok::eof)) { 4150b57cec5SDimitry Andric if (RawToken.is(tok::hash) && RawToken.isAtStartOfLine()) { 4160b57cec5SDimitry Andric RawLex.setParsingPreprocessorDirective(true); 4170b57cec5SDimitry Andric Token HashToken = RawToken; 4180b57cec5SDimitry Andric RawLex.LexFromRawLexer(RawToken); 4190b57cec5SDimitry Andric if (RawToken.is(tok::raw_identifier)) 4200b57cec5SDimitry Andric PP.LookUpIdentifierInfo(RawToken); 4210b57cec5SDimitry Andric if (RawToken.getIdentifierInfo() != nullptr) { 4220b57cec5SDimitry Andric switch (RawToken.getIdentifierInfo()->getPPKeywordID()) { 4230b57cec5SDimitry Andric case tok::pp_include: 4240b57cec5SDimitry Andric case tok::pp_include_next: 4250b57cec5SDimitry Andric case tok::pp_import: { 426*5f757f3fSDimitry Andric SourceLocation Loc = HashToken.getLocation(); 427*5f757f3fSDimitry Andric const IncludedFile *Inc = FindIncludeAtLocation(Loc); 428*5f757f3fSDimitry Andric CommentOutDirective(RawLex, HashToken, FromFile, LocalEOL, 429*5f757f3fSDimitry Andric NextToWrite, Line, Inc); 4300b57cec5SDimitry Andric if (FileId != PP.getPredefinesFileID()) 4310b57cec5SDimitry Andric WriteLineInfo(FileName, Line - 1, FileType, ""); 4320b57cec5SDimitry Andric StringRef LineInfoExtra; 4330b57cec5SDimitry Andric if (const Module *Mod = FindModuleAtLocation(Loc)) 4340b57cec5SDimitry Andric WriteImplicitModuleImport(Mod); 435*5f757f3fSDimitry Andric else if (Inc) { 4360b57cec5SDimitry Andric const Module *Mod = FindEnteredModule(Loc); 4370b57cec5SDimitry Andric if (Mod) 4380b57cec5SDimitry Andric OS << "#pragma clang module begin " 4390b57cec5SDimitry Andric << Mod->getFullModuleName(true) << "\n"; 4400b57cec5SDimitry Andric 4410b57cec5SDimitry Andric // Include and recursively process the file. 44204eeddc0SDimitry Andric Process(Inc->Id, Inc->FileType); 4430b57cec5SDimitry Andric 4440b57cec5SDimitry Andric if (Mod) 4450b57cec5SDimitry Andric OS << "#pragma clang module end /*" 4460b57cec5SDimitry Andric << Mod->getFullModuleName(true) << "*/\n"; 447*5f757f3fSDimitry Andric // There's no #include, therefore no #if, for -include files. 448*5f757f3fSDimitry Andric if (FromFile != PredefinesBuffer) { 449*5f757f3fSDimitry Andric OS << "#endif /* " << getIncludedFileName(Inc) 450*5f757f3fSDimitry Andric << " expanded by -frewrite-includes */" << LocalEOL; 451*5f757f3fSDimitry Andric } 4520b57cec5SDimitry Andric 4530b57cec5SDimitry Andric // Add line marker to indicate we're returning from an included 4540b57cec5SDimitry Andric // file. 4550b57cec5SDimitry Andric LineInfoExtra = " 2"; 4560b57cec5SDimitry Andric } 4570b57cec5SDimitry Andric // fix up lineinfo (since commented out directive changed line 4580b57cec5SDimitry Andric // numbers) for inclusions that were skipped due to header guards 4590b57cec5SDimitry Andric WriteLineInfo(FileName, Line, FileType, LineInfoExtra); 4600b57cec5SDimitry Andric break; 4610b57cec5SDimitry Andric } 4620b57cec5SDimitry Andric case tok::pp_pragma: { 4630b57cec5SDimitry Andric StringRef Identifier = NextIdentifierName(RawLex, RawToken); 4640b57cec5SDimitry Andric if (Identifier == "clang" || Identifier == "GCC") { 4650b57cec5SDimitry Andric if (NextIdentifierName(RawLex, RawToken) == "system_header") { 4660b57cec5SDimitry Andric // keep the directive in, commented out 4670b57cec5SDimitry Andric CommentOutDirective(RawLex, HashToken, FromFile, LocalEOL, 4680b57cec5SDimitry Andric NextToWrite, Line); 4690b57cec5SDimitry Andric // update our own type 4700b57cec5SDimitry Andric FileType = SM.getFileCharacteristic(RawToken.getLocation()); 4710b57cec5SDimitry Andric WriteLineInfo(FileName, Line, FileType); 4720b57cec5SDimitry Andric } 4730b57cec5SDimitry Andric } else if (Identifier == "once") { 4740b57cec5SDimitry Andric // keep the directive in, commented out 4750b57cec5SDimitry Andric CommentOutDirective(RawLex, HashToken, FromFile, LocalEOL, 4760b57cec5SDimitry Andric NextToWrite, Line); 4770b57cec5SDimitry Andric WriteLineInfo(FileName, Line, FileType); 4780b57cec5SDimitry Andric } 4790b57cec5SDimitry Andric break; 4800b57cec5SDimitry Andric } 4810b57cec5SDimitry Andric case tok::pp_if: 4820b57cec5SDimitry Andric case tok::pp_elif: { 4830b57cec5SDimitry Andric bool elif = (RawToken.getIdentifierInfo()->getPPKeywordID() == 4840b57cec5SDimitry Andric tok::pp_elif); 485a7dea167SDimitry Andric bool isTrue = IsIfAtLocationTrue(RawToken.getLocation()); 4860b57cec5SDimitry Andric OutputContentUpTo(FromFile, NextToWrite, 487a7dea167SDimitry Andric SM.getFileOffset(HashToken.getLocation()), 488a7dea167SDimitry Andric LocalEOL, Line, /*EnsureNewline=*/true); 489a7dea167SDimitry Andric do { 490a7dea167SDimitry Andric RawLex.LexFromRawLexer(RawToken); 491a7dea167SDimitry Andric } while (!RawToken.is(tok::eod) && RawToken.isNot(tok::eof)); 492a7dea167SDimitry Andric // We need to disable the old condition, but that is tricky. 493a7dea167SDimitry Andric // Trying to comment it out can easily lead to comment nesting. 494a7dea167SDimitry Andric // So instead make the condition harmless by making it enclose 495a7dea167SDimitry Andric // and empty block. Moreover, put it itself inside an #if 0 block 496a7dea167SDimitry Andric // to disable it from getting evaluated (e.g. __has_include_next 497a7dea167SDimitry Andric // warns if used from the primary source file). 498a7dea167SDimitry Andric OS << "#if 0 /* disabled by -frewrite-includes */" << MainEOL; 4990b57cec5SDimitry Andric if (elif) { 500a7dea167SDimitry Andric OS << "#if 0" << MainEOL; 501a7dea167SDimitry Andric } 5020b57cec5SDimitry Andric OutputContentUpTo(FromFile, NextToWrite, 5030b57cec5SDimitry Andric SM.getFileOffset(RawToken.getLocation()) + 5040b57cec5SDimitry Andric RawToken.getLength(), 5050b57cec5SDimitry Andric LocalEOL, Line, /*EnsureNewline=*/true); 506a7dea167SDimitry Andric // Close the empty block and the disabling block. 507a7dea167SDimitry Andric OS << "#endif" << MainEOL; 508a7dea167SDimitry Andric OS << "#endif /* disabled by -frewrite-includes */" << MainEOL; 509a7dea167SDimitry Andric OS << (elif ? "#elif " : "#if ") << (isTrue ? "1" : "0") 510a7dea167SDimitry Andric << " /* evaluated by -frewrite-includes */" << MainEOL; 5110b57cec5SDimitry Andric WriteLineInfo(FileName, Line, FileType); 5120b57cec5SDimitry Andric break; 5130b57cec5SDimitry Andric } 5140b57cec5SDimitry Andric case tok::pp_endif: 5150b57cec5SDimitry Andric case tok::pp_else: { 5160b57cec5SDimitry Andric // We surround every #include by #if 0 to comment it out, but that 5170b57cec5SDimitry Andric // changes line numbers. These are fixed up right after that, but 5180b57cec5SDimitry Andric // the whole #include could be inside a preprocessor conditional 5190b57cec5SDimitry Andric // that is not processed. So it is necessary to fix the line 5200b57cec5SDimitry Andric // numbers one the next line after each #else/#endif as well. 5210b57cec5SDimitry Andric RawLex.SetKeepWhitespaceMode(true); 5220b57cec5SDimitry Andric do { 5230b57cec5SDimitry Andric RawLex.LexFromRawLexer(RawToken); 5240b57cec5SDimitry Andric } while (RawToken.isNot(tok::eod) && RawToken.isNot(tok::eof)); 5250b57cec5SDimitry Andric OutputContentUpTo(FromFile, NextToWrite, 5260b57cec5SDimitry Andric SM.getFileOffset(RawToken.getLocation()) + 5270b57cec5SDimitry Andric RawToken.getLength(), 5280b57cec5SDimitry Andric LocalEOL, Line, /*EnsureNewline=*/ true); 5290b57cec5SDimitry Andric WriteLineInfo(FileName, Line, FileType); 5300b57cec5SDimitry Andric RawLex.SetKeepWhitespaceMode(false); 5310b57cec5SDimitry Andric break; 5320b57cec5SDimitry Andric } 5330b57cec5SDimitry Andric default: 5340b57cec5SDimitry Andric break; 5350b57cec5SDimitry Andric } 5360b57cec5SDimitry Andric } 5370b57cec5SDimitry Andric RawLex.setParsingPreprocessorDirective(false); 5380b57cec5SDimitry Andric } 5390b57cec5SDimitry Andric RawLex.LexFromRawLexer(RawToken); 5400b57cec5SDimitry Andric } 5410b57cec5SDimitry Andric OutputContentUpTo(FromFile, NextToWrite, 5420b57cec5SDimitry Andric SM.getFileOffset(SM.getLocForEndOfFile(FileId)), LocalEOL, 5430b57cec5SDimitry Andric Line, /*EnsureNewline=*/true); 5440b57cec5SDimitry Andric } 5450b57cec5SDimitry Andric 5460b57cec5SDimitry Andric /// InclusionRewriterInInput - Implement -frewrite-includes mode. 5470b57cec5SDimitry Andric void clang::RewriteIncludesInInput(Preprocessor &PP, raw_ostream *OS, 5480b57cec5SDimitry Andric const PreprocessorOutputOptions &Opts) { 5490b57cec5SDimitry Andric SourceManager &SM = PP.getSourceManager(); 5500b57cec5SDimitry Andric InclusionRewriter *Rewrite = new InclusionRewriter( 5510b57cec5SDimitry Andric PP, *OS, Opts.ShowLineMarkers, Opts.UseLineDirectives); 5520b57cec5SDimitry Andric Rewrite->detectMainFileEOL(); 5530b57cec5SDimitry Andric 5540b57cec5SDimitry Andric PP.addPPCallbacks(std::unique_ptr<PPCallbacks>(Rewrite)); 5550b57cec5SDimitry Andric PP.IgnorePragmas(); 5560b57cec5SDimitry Andric 5570b57cec5SDimitry Andric // First let the preprocessor process the entire file and call callbacks. 5580b57cec5SDimitry Andric // Callbacks will record which #include's were actually performed. 5590b57cec5SDimitry Andric PP.EnterMainSourceFile(); 5600b57cec5SDimitry Andric Token Tok; 5610b57cec5SDimitry Andric // Only preprocessor directives matter here, so disable macro expansion 5620b57cec5SDimitry Andric // everywhere else as an optimization. 5630b57cec5SDimitry Andric // TODO: It would be even faster if the preprocessor could be switched 5640b57cec5SDimitry Andric // to a mode where it would parse only preprocessor directives and comments, 5650b57cec5SDimitry Andric // nothing else matters for parsing or processing. 5660b57cec5SDimitry Andric PP.SetMacroExpansionOnlyInDirectives(); 5670b57cec5SDimitry Andric do { 5680b57cec5SDimitry Andric PP.Lex(Tok); 5690b57cec5SDimitry Andric if (Tok.is(tok::annot_module_begin)) 5700b57cec5SDimitry Andric Rewrite->handleModuleBegin(Tok); 5710b57cec5SDimitry Andric } while (Tok.isNot(tok::eof)); 572e8d8bef9SDimitry Andric Rewrite->setPredefinesBuffer(SM.getBufferOrFake(PP.getPredefinesFileID())); 57304eeddc0SDimitry Andric Rewrite->Process(PP.getPredefinesFileID(), SrcMgr::C_User); 57404eeddc0SDimitry Andric Rewrite->Process(SM.getMainFileID(), SrcMgr::C_User); 5750b57cec5SDimitry Andric OS->flush(); 5760b57cec5SDimitry Andric } 577