blob: ca14da3bca3910c76bf69602ea63adce444813d3 [file] [log] [blame]
David Blaikied5321242012-06-06 18:52:13 +00001//===--- InclusionRewriter.cpp - Rewrite includes into their expansions ---===//
2//
3// The LLVM Compiler Infrastructure
4//
5// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
7//
8//===----------------------------------------------------------------------===//
9//
10// This code rewrites include invocations into their expansions. This gives you
11// a file with all included files merged into it.
12//
13//===----------------------------------------------------------------------===//
14
Ted Kremenekcdf81492012-09-01 05:09:24 +000015#include "clang/Rewrite/Frontend/Rewriters.h"
David Blaikied5321242012-06-06 18:52:13 +000016#include "clang/Basic/SourceManager.h"
17#include "clang/Frontend/PreprocessorOutputOptions.h"
Benjamin Kramerb10e6152013-04-16 19:08:41 +000018#include "clang/Lex/HeaderSearch.h"
Lubos Lunakba5ee4d2013-07-20 14:30:01 +000019#include "clang/Lex/Pragma.h"
Chandler Carruth3a022472012-12-04 09:13:33 +000020#include "clang/Lex/Preprocessor.h"
Benjamin Kramerb10e6152013-04-16 19:08:41 +000021#include "llvm/ADT/SmallString.h"
David Blaikied5321242012-06-06 18:52:13 +000022#include "llvm/Support/raw_ostream.h"
23
24using namespace clang;
25using namespace llvm;
26
27namespace {
28
29class InclusionRewriter : public PPCallbacks {
30 /// Information about which #includes were actually performed,
31 /// created by preprocessor callbacks.
Justin Bogner0707fd02015-07-01 04:40:10 +000032 struct IncludedFile {
David Blaikied5321242012-06-06 18:52:13 +000033 FileID Id;
34 SrcMgr::CharacteristicKind FileType;
Justin Bogner0707fd02015-07-01 04:40:10 +000035 IncludedFile(FileID Id, SrcMgr::CharacteristicKind FileType)
36 : Id(Id), FileType(FileType) {}
David Blaikied5321242012-06-06 18:52:13 +000037 };
Dmitri Gribenko4280e5c2012-06-08 23:13:42 +000038 Preprocessor &PP; ///< Used to find inclusion directives.
39 SourceManager &SM; ///< Used to read and manage source files.
40 raw_ostream &OS; ///< The destination stream for rewritten contents.
Reid Klecknere2793c02014-09-05 16:49:50 +000041 StringRef MainEOL; ///< The line ending marker to use.
Argyrios Kyrtzidis17ff2e52013-07-26 15:32:04 +000042 const llvm::MemoryBuffer *PredefinesBuffer; ///< The preprocessor predefines.
Dmitri Gribenko4280e5c2012-06-08 23:13:42 +000043 bool ShowLineMarkers; ///< Show #line markers.
Reid Kleckner1df0fea2015-02-26 00:17:25 +000044 bool UseLineDirectives; ///< Use of line directives or line markers.
Justin Bogner0707fd02015-07-01 04:40:10 +000045 /// Tracks where inclusions that change the file are found.
46 std::map<unsigned, IncludedFile> FileIncludes;
47 /// Tracks where inclusions that import modules are found.
48 std::map<unsigned, const Module *> ModuleIncludes;
49 /// Used transitively for building up the FileIncludes mapping over the
David Blaikied5321242012-06-06 18:52:13 +000050 /// various \c PPCallbacks callbacks.
Justin Bogner0707fd02015-07-01 04:40:10 +000051 SourceLocation LastInclusionLocation;
David Blaikied5321242012-06-06 18:52:13 +000052public:
Reid Kleckner1df0fea2015-02-26 00:17:25 +000053 InclusionRewriter(Preprocessor &PP, raw_ostream &OS, bool ShowLineMarkers,
54 bool UseLineDirectives);
David Blaikied5321242012-06-06 18:52:13 +000055 bool Process(FileID FileId, SrcMgr::CharacteristicKind FileType);
Argyrios Kyrtzidis17ff2e52013-07-26 15:32:04 +000056 void setPredefinesBuffer(const llvm::MemoryBuffer *Buf) {
57 PredefinesBuffer = Buf;
58 }
Reid Klecknere2793c02014-09-05 16:49:50 +000059 void detectMainFileEOL();
David Blaikied5321242012-06-06 18:52:13 +000060private:
Craig Topperfb6b25b2014-03-15 04:29:04 +000061 void FileChanged(SourceLocation Loc, FileChangeReason Reason,
62 SrcMgr::CharacteristicKind FileType,
63 FileID PrevFID) override;
Nikola Smiljanicfb891fc2015-05-12 11:48:05 +000064 void FileSkipped(const FileEntry &SkippedFile, const Token &FilenameTok,
Craig Topperfb6b25b2014-03-15 04:29:04 +000065 SrcMgr::CharacteristicKind FileType) override;
66 void InclusionDirective(SourceLocation HashLoc, const Token &IncludeTok,
67 StringRef FileName, bool IsAngled,
68 CharSourceRange FilenameRange, const FileEntry *File,
69 StringRef SearchPath, StringRef RelativePath,
70 const Module *Imported) override;
Mehdi Amini99d1b292016-10-01 16:38:28 +000071 void WriteLineInfo(StringRef Filename, int Line,
David Blaikied5321242012-06-06 18:52:13 +000072 SrcMgr::CharacteristicKind FileType,
Reid Klecknere2793c02014-09-05 16:49:50 +000073 StringRef Extra = StringRef());
74 void WriteImplicitModuleImport(const Module *Mod);
David Blaikied5321242012-06-06 18:52:13 +000075 void OutputContentUpTo(const MemoryBuffer &FromFile,
76 unsigned &WriteFrom, unsigned WriteTo,
77 StringRef EOL, int &lines,
Alp Toker08c25002013-12-13 17:04:55 +000078 bool EnsureNewline);
David Blaikied5321242012-06-06 18:52:13 +000079 void CommentOutDirective(Lexer &DirectivesLex, const Token &StartToken,
80 const MemoryBuffer &FromFile, StringRef EOL,
81 unsigned &NextToWrite, int &Lines);
Benjamin Kramerb10e6152013-04-16 19:08:41 +000082 bool HandleHasInclude(FileID FileId, Lexer &RawLex,
83 const DirectoryLookup *Lookup, Token &Tok,
84 bool &FileExists);
Justin Bogner0707fd02015-07-01 04:40:10 +000085 const IncludedFile *FindIncludeAtLocation(SourceLocation Loc) const;
86 const Module *FindModuleAtLocation(SourceLocation Loc) const;
David Blaikied5321242012-06-06 18:52:13 +000087 StringRef NextIdentifierName(Lexer &RawLex, Token &RawToken);
88};
89
90} // end anonymous namespace
91
92/// Initializes an InclusionRewriter with a \p PP source and \p OS destination.
93InclusionRewriter::InclusionRewriter(Preprocessor &PP, raw_ostream &OS,
Reid Kleckner1df0fea2015-02-26 00:17:25 +000094 bool ShowLineMarkers,
95 bool UseLineDirectives)
Reid Klecknere2793c02014-09-05 16:49:50 +000096 : PP(PP), SM(PP.getSourceManager()), OS(OS), MainEOL("\n"),
97 PredefinesBuffer(nullptr), ShowLineMarkers(ShowLineMarkers),
Eric Christopher8213f7f2015-02-26 00:29:54 +000098 UseLineDirectives(UseLineDirectives),
Justin Bogner0707fd02015-07-01 04:40:10 +000099 LastInclusionLocation(SourceLocation()) {}
David Blaikied5321242012-06-06 18:52:13 +0000100
101/// Write appropriate line information as either #line directives or GNU line
102/// markers depending on what mode we're in, including the \p Filename and
103/// \p Line we are located at, using the specified \p EOL line separator, and
104/// any \p Extra context specifiers in GNU line directives.
Mehdi Amini99d1b292016-10-01 16:38:28 +0000105void InclusionRewriter::WriteLineInfo(StringRef Filename, int Line,
David Blaikied5321242012-06-06 18:52:13 +0000106 SrcMgr::CharacteristicKind FileType,
Reid Klecknere2793c02014-09-05 16:49:50 +0000107 StringRef Extra) {
David Blaikied5321242012-06-06 18:52:13 +0000108 if (!ShowLineMarkers)
109 return;
Reid Kleckner1df0fea2015-02-26 00:17:25 +0000110 if (UseLineDirectives) {
Eli Friedman9fc443a2013-09-17 00:51:31 +0000111 OS << "#line" << ' ' << Line << ' ' << '"';
112 OS.write_escaped(Filename);
113 OS << '"';
David Blaikied5321242012-06-06 18:52:13 +0000114 } else {
115 // Use GNU linemarkers as described here:
116 // http://gcc.gnu.org/onlinedocs/cpp/Preprocessor-Output.html
Eli Friedman80e45b82013-08-29 01:42:42 +0000117 OS << '#' << ' ' << Line << ' ' << '"';
118 OS.write_escaped(Filename);
119 OS << '"';
David Blaikied5321242012-06-06 18:52:13 +0000120 if (!Extra.empty())
121 OS << Extra;
122 if (FileType == SrcMgr::C_System)
123 // "`3' This indicates that the following text comes from a system header
124 // file, so certain warnings should be suppressed."
125 OS << " 3";
126 else if (FileType == SrcMgr::C_ExternCSystem)
127 // as above for `3', plus "`4' This indicates that the following text
128 // should be treated as being wrapped in an implicit extern "C" block."
129 OS << " 3 4";
130 }
Reid Klecknere2793c02014-09-05 16:49:50 +0000131 OS << MainEOL;
David Blaikied5321242012-06-06 18:52:13 +0000132}
133
Reid Klecknere2793c02014-09-05 16:49:50 +0000134void InclusionRewriter::WriteImplicitModuleImport(const Module *Mod) {
Argyrios Kyrtzidiscf22d1f2013-04-10 01:53:50 +0000135 OS << "@import " << Mod->getFullModuleName() << ";"
Reid Klecknere2793c02014-09-05 16:49:50 +0000136 << " /* clang -frewrite-includes: implicit import */" << MainEOL;
Argyrios Kyrtzidiscf22d1f2013-04-10 01:53:50 +0000137}
138
David Blaikied5321242012-06-06 18:52:13 +0000139/// FileChanged - Whenever the preprocessor enters or exits a #include file
140/// it invokes this handler.
141void InclusionRewriter::FileChanged(SourceLocation Loc,
142 FileChangeReason Reason,
143 SrcMgr::CharacteristicKind NewFileType,
144 FileID) {
145 if (Reason != EnterFile)
146 return;
Justin Bogner0707fd02015-07-01 04:40:10 +0000147 if (LastInclusionLocation.isInvalid())
David Blaikied5321242012-06-06 18:52:13 +0000148 // we didn't reach this file (eg: the main file) via an inclusion directive
149 return;
Justin Bogner0707fd02015-07-01 04:40:10 +0000150 FileID Id = FullSourceLoc(Loc, SM).getFileID();
Justin Bogner879d4202015-07-01 04:53:19 +0000151 auto P = FileIncludes.insert(std::make_pair(
152 LastInclusionLocation.getRawEncoding(), IncludedFile(Id, NewFileType)));
Justin Bogner2510ba32015-07-01 05:41:50 +0000153 (void)P;
Justin Bogner0707fd02015-07-01 04:40:10 +0000154 assert(P.second && "Unexpected revisitation of the same include directive");
155 LastInclusionLocation = SourceLocation();
David Blaikied5321242012-06-06 18:52:13 +0000156}
157
158/// Called whenever an inclusion is skipped due to canonical header protection
159/// macros.
Nikola Smiljanicfb891fc2015-05-12 11:48:05 +0000160void InclusionRewriter::FileSkipped(const FileEntry &/*SkippedFile*/,
David Blaikied5321242012-06-06 18:52:13 +0000161 const Token &/*FilenameTok*/,
162 SrcMgr::CharacteristicKind /*FileType*/) {
Yaron Kerened1fe5d2015-10-03 05:15:57 +0000163 assert(LastInclusionLocation.isValid() &&
Justin Bogner0707fd02015-07-01 04:40:10 +0000164 "A file, that wasn't found via an inclusion directive, was skipped");
165 LastInclusionLocation = SourceLocation();
David Blaikied5321242012-06-06 18:52:13 +0000166}
167
168/// This should be called whenever the preprocessor encounters include
169/// directives. It does not say whether the file has been included, but it
170/// provides more information about the directive (hash location instead
171/// of location inside the included file). It is assumed that the matching
172/// FileChanged() or FileSkipped() is called after this.
173void InclusionRewriter::InclusionDirective(SourceLocation HashLoc,
174 const Token &/*IncludeTok*/,
175 StringRef /*FileName*/,
176 bool /*IsAngled*/,
Argyrios Kyrtzidis4fcd2882012-09-27 01:42:07 +0000177 CharSourceRange /*FilenameRange*/,
David Blaikied5321242012-06-06 18:52:13 +0000178 const FileEntry * /*File*/,
David Blaikied5321242012-06-06 18:52:13 +0000179 StringRef /*SearchPath*/,
Argyrios Kyrtzidis19d78b72012-09-29 01:06:10 +0000180 StringRef /*RelativePath*/,
Argyrios Kyrtzidiscf22d1f2013-04-10 01:53:50 +0000181 const Module *Imported) {
Justin Bogner0707fd02015-07-01 04:40:10 +0000182 assert(LastInclusionLocation.isInvalid() &&
183 "Another inclusion directive was found before the previous one "
184 "was processed");
185 if (Imported) {
Justin Bogner879d4202015-07-01 04:53:19 +0000186 auto P = ModuleIncludes.insert(
187 std::make_pair(HashLoc.getRawEncoding(), Imported));
Justin Bogner2510ba32015-07-01 05:41:50 +0000188 (void)P;
Justin Bogner0707fd02015-07-01 04:40:10 +0000189 assert(P.second && "Unexpected revisitation of the same include directive");
190 } else
191 LastInclusionLocation = HashLoc;
David Blaikied5321242012-06-06 18:52:13 +0000192}
193
194/// Simple lookup for a SourceLocation (specifically one denoting the hash in
195/// an inclusion directive) in the map of inclusion information, FileChanges.
Justin Bogner0707fd02015-07-01 04:40:10 +0000196const InclusionRewriter::IncludedFile *
197InclusionRewriter::FindIncludeAtLocation(SourceLocation Loc) const {
198 const auto I = FileIncludes.find(Loc.getRawEncoding());
199 if (I != FileIncludes.end())
David Blaikied5321242012-06-06 18:52:13 +0000200 return &I->second;
Craig Topper8ae12032014-05-07 06:21:57 +0000201 return nullptr;
David Blaikied5321242012-06-06 18:52:13 +0000202}
203
Justin Bogner0707fd02015-07-01 04:40:10 +0000204/// Simple lookup for a SourceLocation (specifically one denoting the hash in
205/// an inclusion directive) in the map of module inclusion information.
206const Module *
207InclusionRewriter::FindModuleAtLocation(SourceLocation Loc) const {
208 const auto I = ModuleIncludes.find(Loc.getRawEncoding());
209 if (I != ModuleIncludes.end())
210 return I->second;
211 return nullptr;
212}
213
David Blaikied5321242012-06-06 18:52:13 +0000214/// Detect the likely line ending style of \p FromFile by examining the first
215/// newline found within it.
216static StringRef DetectEOL(const MemoryBuffer &FromFile) {
Reid Klecknere2793c02014-09-05 16:49:50 +0000217 // Detect what line endings the file uses, so that added content does not mix
218 // the style. We need to check for "\r\n" first because "\n\r" will match
219 // "\r\n\r\n".
David Blaikied5321242012-06-06 18:52:13 +0000220 const char *Pos = strchr(FromFile.getBufferStart(), '\n');
Craig Topper8ae12032014-05-07 06:21:57 +0000221 if (!Pos)
David Blaikied5321242012-06-06 18:52:13 +0000222 return "\n";
David Blaikied5321242012-06-06 18:52:13 +0000223 if (Pos - 1 >= FromFile.getBufferStart() && Pos[-1] == '\r')
224 return "\r\n";
Reid Klecknere2793c02014-09-05 16:49:50 +0000225 if (Pos + 1 < FromFile.getBufferEnd() && Pos[1] == '\r')
226 return "\n\r";
David Blaikied5321242012-06-06 18:52:13 +0000227 return "\n";
228}
229
Reid Klecknere2793c02014-09-05 16:49:50 +0000230void InclusionRewriter::detectMainFileEOL() {
231 bool Invalid;
232 const MemoryBuffer &FromFile = *SM.getBuffer(SM.getMainFileID(), &Invalid);
233 assert(!Invalid);
234 if (Invalid)
235 return; // Should never happen, but whatever.
236 MainEOL = DetectEOL(FromFile);
237}
238
David Blaikied5321242012-06-06 18:52:13 +0000239/// Writes out bytes from \p FromFile, starting at \p NextToWrite and ending at
240/// \p WriteTo - 1.
241void InclusionRewriter::OutputContentUpTo(const MemoryBuffer &FromFile,
242 unsigned &WriteFrom, unsigned WriteTo,
Reid Klecknere2793c02014-09-05 16:49:50 +0000243 StringRef LocalEOL, int &Line,
David Blaikied5321242012-06-06 18:52:13 +0000244 bool EnsureNewline) {
245 if (WriteTo <= WriteFrom)
246 return;
Argyrios Kyrtzidis17ff2e52013-07-26 15:32:04 +0000247 if (&FromFile == PredefinesBuffer) {
248 // Ignore the #defines of the predefines buffer.
249 WriteFrom = WriteTo;
250 return;
251 }
Reid Klecknere2793c02014-09-05 16:49:50 +0000252
253 // If we would output half of a line ending, advance one character to output
254 // the whole line ending. All buffers are null terminated, so looking ahead
255 // one byte is safe.
256 if (LocalEOL.size() == 2 &&
257 LocalEOL[0] == (FromFile.getBufferStart() + WriteTo)[-1] &&
258 LocalEOL[1] == (FromFile.getBufferStart() + WriteTo)[0])
259 WriteTo++;
260
261 StringRef TextToWrite(FromFile.getBufferStart() + WriteFrom,
262 WriteTo - WriteFrom);
263
264 if (MainEOL == LocalEOL) {
265 OS << TextToWrite;
266 // count lines manually, it's faster than getPresumedLoc()
267 Line += TextToWrite.count(LocalEOL);
268 if (EnsureNewline && !TextToWrite.endswith(LocalEOL))
269 OS << MainEOL;
270 } else {
271 // Output the file one line at a time, rewriting the line endings as we go.
272 StringRef Rest = TextToWrite;
273 while (!Rest.empty()) {
274 StringRef LineText;
275 std::tie(LineText, Rest) = Rest.split(LocalEOL);
276 OS << LineText;
277 Line++;
278 if (!Rest.empty())
279 OS << MainEOL;
280 }
281 if (TextToWrite.endswith(LocalEOL) || EnsureNewline)
282 OS << MainEOL;
David Blaikied5321242012-06-06 18:52:13 +0000283 }
284 WriteFrom = WriteTo;
285}
286
287/// Print characters from \p FromFile starting at \p NextToWrite up until the
288/// inclusion directive at \p StartToken, then print out the inclusion
289/// inclusion directive disabled by a #if directive, updating \p NextToWrite
290/// and \p Line to track the number of source lines visited and the progress
291/// through the \p FromFile buffer.
292void InclusionRewriter::CommentOutDirective(Lexer &DirectiveLex,
293 const Token &StartToken,
294 const MemoryBuffer &FromFile,
Reid Klecknere2793c02014-09-05 16:49:50 +0000295 StringRef LocalEOL,
David Blaikied5321242012-06-06 18:52:13 +0000296 unsigned &NextToWrite, int &Line) {
297 OutputContentUpTo(FromFile, NextToWrite,
Reid Klecknere2793c02014-09-05 16:49:50 +0000298 SM.getFileOffset(StartToken.getLocation()), LocalEOL, Line,
299 false);
David Blaikied5321242012-06-06 18:52:13 +0000300 Token DirectiveToken;
301 do {
302 DirectiveLex.LexFromRawLexer(DirectiveToken);
303 } while (!DirectiveToken.is(tok::eod) && DirectiveToken.isNot(tok::eof));
Lubos Lunak72cad682014-05-01 21:10:08 +0000304 if (&FromFile == PredefinesBuffer) {
305 // OutputContentUpTo() would not output anything anyway.
306 return;
307 }
Reid Klecknere2793c02014-09-05 16:49:50 +0000308 OS << "#if 0 /* expanded by -frewrite-includes */" << MainEOL;
David Blaikied5321242012-06-06 18:52:13 +0000309 OutputContentUpTo(FromFile, NextToWrite,
Reid Klecknere2793c02014-09-05 16:49:50 +0000310 SM.getFileOffset(DirectiveToken.getLocation()) +
311 DirectiveToken.getLength(),
312 LocalEOL, Line, true);
313 OS << "#endif /* expanded by -frewrite-includes */" << MainEOL;
David Blaikied5321242012-06-06 18:52:13 +0000314}
315
316/// Find the next identifier in the pragma directive specified by \p RawToken.
317StringRef InclusionRewriter::NextIdentifierName(Lexer &RawLex,
318 Token &RawToken) {
319 RawLex.LexFromRawLexer(RawToken);
320 if (RawToken.is(tok::raw_identifier))
321 PP.LookUpIdentifierInfo(RawToken);
322 if (RawToken.is(tok::identifier))
323 return RawToken.getIdentifierInfo()->getName();
324 return StringRef();
325}
326
Benjamin Kramerb10e6152013-04-16 19:08:41 +0000327// Expand __has_include and __has_include_next if possible. If there's no
328// definitive answer return false.
329bool InclusionRewriter::HandleHasInclude(
330 FileID FileId, Lexer &RawLex, const DirectoryLookup *Lookup, Token &Tok,
331 bool &FileExists) {
332 // Lex the opening paren.
333 RawLex.LexFromRawLexer(Tok);
334 if (Tok.isNot(tok::l_paren))
335 return false;
336
337 RawLex.LexFromRawLexer(Tok);
338
339 SmallString<128> FilenameBuffer;
340 StringRef Filename;
341 // Since the raw lexer doesn't give us angle_literals we have to parse them
342 // ourselves.
343 // FIXME: What to do if the file name is a macro?
344 if (Tok.is(tok::less)) {
345 RawLex.LexFromRawLexer(Tok);
346
347 FilenameBuffer += '<';
348 do {
349 if (Tok.is(tok::eod)) // Sanity check.
350 return false;
351
352 if (Tok.is(tok::raw_identifier))
353 PP.LookUpIdentifierInfo(Tok);
354
355 // Get the string piece.
356 SmallVector<char, 128> TmpBuffer;
357 bool Invalid = false;
358 StringRef TmpName = PP.getSpelling(Tok, TmpBuffer, &Invalid);
359 if (Invalid)
360 return false;
361
362 FilenameBuffer += TmpName;
363
364 RawLex.LexFromRawLexer(Tok);
365 } while (Tok.isNot(tok::greater));
366
367 FilenameBuffer += '>';
368 Filename = FilenameBuffer;
369 } else {
370 if (Tok.isNot(tok::string_literal))
371 return false;
372
373 bool Invalid = false;
374 Filename = PP.getSpelling(Tok, FilenameBuffer, &Invalid);
375 if (Invalid)
376 return false;
377 }
378
379 // Lex the closing paren.
380 RawLex.LexFromRawLexer(Tok);
381 if (Tok.isNot(tok::r_paren))
382 return false;
383
384 // Now ask HeaderInfo if it knows about the header.
385 // FIXME: Subframeworks aren't handled here. Do we care?
386 bool isAngled = PP.GetIncludeFilenameSpelling(Tok.getLocation(), Filename);
387 const DirectoryLookup *CurDir;
Manuel Klimek9af34ae2014-08-12 08:25:57 +0000388 const FileEntry *FileEnt = PP.getSourceManager().getFileEntryForID(FileId);
389 SmallVector<std::pair<const FileEntry *, const DirectoryEntry *>, 1>
390 Includers;
391 Includers.push_back(std::make_pair(FileEnt, FileEnt->getDir()));
Richard Smith3d5b48c2015-10-16 21:42:56 +0000392 // FIXME: Why don't we call PP.LookupFile here?
Benjamin Kramerb10e6152013-04-16 19:08:41 +0000393 const FileEntry *File = PP.getHeaderSearchInfo().LookupFile(
Manuel Klimek9af34ae2014-08-12 08:25:57 +0000394 Filename, SourceLocation(), isAngled, nullptr, CurDir, Includers, nullptr,
Duncan P. N. Exon Smithcfc1f6a2017-04-27 21:41:51 +0000395 nullptr, nullptr, nullptr, nullptr);
Benjamin Kramerb10e6152013-04-16 19:08:41 +0000396
Craig Topper8ae12032014-05-07 06:21:57 +0000397 FileExists = File != nullptr;
Benjamin Kramerb10e6152013-04-16 19:08:41 +0000398 return true;
399}
400
Benjamin Kramere2881572013-10-13 12:02:16 +0000401/// Use a raw lexer to analyze \p FileId, incrementally copying parts of it
David Blaikied5321242012-06-06 18:52:13 +0000402/// and including content of included files recursively.
403bool InclusionRewriter::Process(FileID FileId,
404 SrcMgr::CharacteristicKind FileType)
405{
406 bool Invalid;
407 const MemoryBuffer &FromFile = *SM.getBuffer(FileId, &Invalid);
Justin Bogner0707fd02015-07-01 04:40:10 +0000408 assert(!Invalid && "Attempting to process invalid inclusion");
Mehdi Amini99d1b292016-10-01 16:38:28 +0000409 StringRef FileName = FromFile.getBufferIdentifier();
David Blaikied5321242012-06-06 18:52:13 +0000410 Lexer RawLex(FileId, &FromFile, PP.getSourceManager(), PP.getLangOpts());
411 RawLex.SetCommentRetentionState(false);
412
Reid Klecknere2793c02014-09-05 16:49:50 +0000413 StringRef LocalEOL = DetectEOL(FromFile);
David Blaikied5321242012-06-06 18:52:13 +0000414
Lubos Lunak10961c02014-05-01 13:50:44 +0000415 // Per the GNU docs: "1" indicates entering a new file.
Lubos Lunak72cad682014-05-01 21:10:08 +0000416 if (FileId == SM.getMainFileID() || FileId == PP.getPredefinesFileID())
Reid Klecknere2793c02014-09-05 16:49:50 +0000417 WriteLineInfo(FileName, 1, FileType, "");
Lubos Lunak10961c02014-05-01 13:50:44 +0000418 else
Reid Klecknere2793c02014-09-05 16:49:50 +0000419 WriteLineInfo(FileName, 1, FileType, " 1");
David Blaikied5321242012-06-06 18:52:13 +0000420
421 if (SM.getFileIDSize(FileId) == 0)
Argyrios Kyrtzidis953ef332013-04-10 01:53:37 +0000422 return false;
David Blaikied5321242012-06-06 18:52:13 +0000423
Alp Toker3dfeafd2013-11-28 07:21:44 +0000424 // The next byte to be copied from the source file, which may be non-zero if
425 // the lexer handled a BOM.
Alp Toker52937ab2013-12-05 17:28:42 +0000426 unsigned NextToWrite = SM.getFileOffset(RawLex.getSourceLocation());
427 assert(SM.getLineNumber(FileId, NextToWrite) == 1);
David Blaikied5321242012-06-06 18:52:13 +0000428 int Line = 1; // The current input file line number.
429
430 Token RawToken;
431 RawLex.LexFromRawLexer(RawToken);
432
433 // TODO: Consider adding a switch that strips possibly unimportant content,
434 // such as comments, to reduce the size of repro files.
435 while (RawToken.isNot(tok::eof)) {
436 if (RawToken.is(tok::hash) && RawToken.isAtStartOfLine()) {
437 RawLex.setParsingPreprocessorDirective(true);
438 Token HashToken = RawToken;
439 RawLex.LexFromRawLexer(RawToken);
440 if (RawToken.is(tok::raw_identifier))
441 PP.LookUpIdentifierInfo(RawToken);
Craig Topper8ae12032014-05-07 06:21:57 +0000442 if (RawToken.getIdentifierInfo() != nullptr) {
David Blaikied5321242012-06-06 18:52:13 +0000443 switch (RawToken.getIdentifierInfo()->getPPKeywordID()) {
444 case tok::pp_include:
445 case tok::pp_include_next:
446 case tok::pp_import: {
Reid Klecknere2793c02014-09-05 16:49:50 +0000447 CommentOutDirective(RawLex, HashToken, FromFile, LocalEOL, NextToWrite,
David Blaikied5321242012-06-06 18:52:13 +0000448 Line);
Lubos Lunak4526b462014-05-01 21:11:57 +0000449 if (FileId != PP.getPredefinesFileID())
Reid Klecknere2793c02014-09-05 16:49:50 +0000450 WriteLineInfo(FileName, Line - 1, FileType, "");
Argyrios Kyrtzidiscf22d1f2013-04-10 01:53:50 +0000451 StringRef LineInfoExtra;
Justin Bogner0707fd02015-07-01 04:40:10 +0000452 SourceLocation Loc = HashToken.getLocation();
Richard Smith466a15e2016-04-08 00:09:53 +0000453 if (const Module *Mod = PP.getLangOpts().ObjC2
454 ? FindModuleAtLocation(Loc)
455 : nullptr)
Justin Bogner0707fd02015-07-01 04:40:10 +0000456 WriteImplicitModuleImport(Mod);
457 else if (const IncludedFile *Inc = FindIncludeAtLocation(Loc)) {
458 // include and recursively process the file
459 if (Process(Inc->Id, Inc->FileType)) {
David Blaikied5321242012-06-06 18:52:13 +0000460 // and set lineinfo back to this file, if the nested one was
461 // actually included
462 // `2' indicates returning to a file (after having included
463 // another file.
Argyrios Kyrtzidiscf22d1f2013-04-10 01:53:50 +0000464 LineInfoExtra = " 2";
Argyrios Kyrtzidis953ef332013-04-10 01:53:37 +0000465 }
Argyrios Kyrtzidiscf22d1f2013-04-10 01:53:50 +0000466 }
467 // fix up lineinfo (since commented out directive changed line
468 // numbers) for inclusions that were skipped due to header guards
Reid Klecknere2793c02014-09-05 16:49:50 +0000469 WriteLineInfo(FileName, Line, FileType, LineInfoExtra);
David Blaikied5321242012-06-06 18:52:13 +0000470 break;
471 }
472 case tok::pp_pragma: {
473 StringRef Identifier = NextIdentifierName(RawLex, RawToken);
474 if (Identifier == "clang" || Identifier == "GCC") {
475 if (NextIdentifierName(RawLex, RawToken) == "system_header") {
476 // keep the directive in, commented out
Reid Klecknere2793c02014-09-05 16:49:50 +0000477 CommentOutDirective(RawLex, HashToken, FromFile, LocalEOL,
David Blaikied5321242012-06-06 18:52:13 +0000478 NextToWrite, Line);
479 // update our own type
480 FileType = SM.getFileCharacteristic(RawToken.getLocation());
Reid Klecknere2793c02014-09-05 16:49:50 +0000481 WriteLineInfo(FileName, Line, FileType);
David Blaikied5321242012-06-06 18:52:13 +0000482 }
483 } else if (Identifier == "once") {
484 // keep the directive in, commented out
Reid Klecknere2793c02014-09-05 16:49:50 +0000485 CommentOutDirective(RawLex, HashToken, FromFile, LocalEOL,
David Blaikied5321242012-06-06 18:52:13 +0000486 NextToWrite, Line);
Reid Klecknere2793c02014-09-05 16:49:50 +0000487 WriteLineInfo(FileName, Line, FileType);
David Blaikied5321242012-06-06 18:52:13 +0000488 }
489 break;
490 }
Benjamin Kramerb10e6152013-04-16 19:08:41 +0000491 case tok::pp_if:
Lubos Lunak4c22f6a2013-07-20 14:23:27 +0000492 case tok::pp_elif: {
493 bool elif = (RawToken.getIdentifierInfo()->getPPKeywordID() ==
494 tok::pp_elif);
Benjamin Kramerb10e6152013-04-16 19:08:41 +0000495 // Rewrite special builtin macros to avoid pulling in host details.
496 do {
497 // Walk over the directive.
498 RawLex.LexFromRawLexer(RawToken);
499 if (RawToken.is(tok::raw_identifier))
500 PP.LookUpIdentifierInfo(RawToken);
501
502 if (RawToken.is(tok::identifier)) {
503 bool HasFile;
504 SourceLocation Loc = RawToken.getLocation();
505
506 // Rewrite __has_include(x)
507 if (RawToken.getIdentifierInfo()->isStr("__has_include")) {
Craig Topper8ae12032014-05-07 06:21:57 +0000508 if (!HandleHasInclude(FileId, RawLex, nullptr, RawToken,
509 HasFile))
Benjamin Kramerb10e6152013-04-16 19:08:41 +0000510 continue;
511 // Rewrite __has_include_next(x)
512 } else if (RawToken.getIdentifierInfo()->isStr(
513 "__has_include_next")) {
514 const DirectoryLookup *Lookup = PP.GetCurDirLookup();
515 if (Lookup)
516 ++Lookup;
517
518 if (!HandleHasInclude(FileId, RawLex, Lookup, RawToken,
519 HasFile))
520 continue;
521 } else {
522 continue;
523 }
524 // Replace the macro with (0) or (1), followed by the commented
525 // out macro for reference.
526 OutputContentUpTo(FromFile, NextToWrite, SM.getFileOffset(Loc),
Reid Klecknere2793c02014-09-05 16:49:50 +0000527 LocalEOL, Line, false);
Benjamin Kramerb10e6152013-04-16 19:08:41 +0000528 OS << '(' << (int) HasFile << ")/*";
529 OutputContentUpTo(FromFile, NextToWrite,
530 SM.getFileOffset(RawToken.getLocation()) +
Reid Klecknere2793c02014-09-05 16:49:50 +0000531 RawToken.getLength(),
532 LocalEOL, Line, false);
Benjamin Kramerb10e6152013-04-16 19:08:41 +0000533 OS << "*/";
534 }
535 } while (RawToken.isNot(tok::eod));
Lubos Lunak4c22f6a2013-07-20 14:23:27 +0000536 if (elif) {
537 OutputContentUpTo(FromFile, NextToWrite,
538 SM.getFileOffset(RawToken.getLocation()) +
539 RawToken.getLength(),
Reid Klecknere2793c02014-09-05 16:49:50 +0000540 LocalEOL, Line, /*EnsureNewline=*/ true);
541 WriteLineInfo(FileName, Line, FileType);
Lubos Lunak4c22f6a2013-07-20 14:23:27 +0000542 }
Benjamin Kramerb10e6152013-04-16 19:08:41 +0000543 break;
Lubos Lunak4c22f6a2013-07-20 14:23:27 +0000544 }
545 case tok::pp_endif:
546 case tok::pp_else: {
547 // We surround every #include by #if 0 to comment it out, but that
548 // changes line numbers. These are fixed up right after that, but
549 // the whole #include could be inside a preprocessor conditional
550 // that is not processed. So it is necessary to fix the line
551 // numbers one the next line after each #else/#endif as well.
552 RawLex.SetKeepWhitespaceMode(true);
553 do {
554 RawLex.LexFromRawLexer(RawToken);
555 } while (RawToken.isNot(tok::eod) && RawToken.isNot(tok::eof));
Reid Klecknere2793c02014-09-05 16:49:50 +0000556 OutputContentUpTo(FromFile, NextToWrite,
557 SM.getFileOffset(RawToken.getLocation()) +
558 RawToken.getLength(),
559 LocalEOL, Line, /*EnsureNewline=*/ true);
560 WriteLineInfo(FileName, Line, FileType);
Lubos Lunak4c22f6a2013-07-20 14:23:27 +0000561 RawLex.SetKeepWhitespaceMode(false);
562 }
David Blaikied5321242012-06-06 18:52:13 +0000563 default:
564 break;
565 }
566 }
567 RawLex.setParsingPreprocessorDirective(false);
568 }
569 RawLex.LexFromRawLexer(RawToken);
570 }
571 OutputContentUpTo(FromFile, NextToWrite,
Reid Klecknere2793c02014-09-05 16:49:50 +0000572 SM.getFileOffset(SM.getLocForEndOfFile(FileId)), LocalEOL,
573 Line, /*EnsureNewline=*/true);
David Blaikied5321242012-06-06 18:52:13 +0000574 return true;
575}
576
David Blaikie619117a2012-06-14 17:36:01 +0000577/// InclusionRewriterInInput - Implement -frewrite-includes mode.
David Blaikied5321242012-06-06 18:52:13 +0000578void clang::RewriteIncludesInInput(Preprocessor &PP, raw_ostream *OS,
579 const PreprocessorOutputOptions &Opts) {
580 SourceManager &SM = PP.getSourceManager();
Reid Kleckner1df0fea2015-02-26 00:17:25 +0000581 InclusionRewriter *Rewrite = new InclusionRewriter(
582 PP, *OS, Opts.ShowLineMarkers, Opts.UseLineDirectives);
Reid Klecknere2793c02014-09-05 16:49:50 +0000583 Rewrite->detectMainFileEOL();
584
Craig Topperb8a70532014-09-10 04:53:53 +0000585 PP.addPPCallbacks(std::unique_ptr<PPCallbacks>(Rewrite));
Lubos Lunak576a0412014-05-01 12:54:03 +0000586 PP.IgnorePragmas();
David Blaikied5321242012-06-06 18:52:13 +0000587
588 // First let the preprocessor process the entire file and call callbacks.
589 // Callbacks will record which #include's were actually performed.
590 PP.EnterMainSourceFile();
591 Token Tok;
592 // Only preprocessor directives matter here, so disable macro expansion
593 // everywhere else as an optimization.
594 // TODO: It would be even faster if the preprocessor could be switched
595 // to a mode where it would parse only preprocessor directives and comments,
596 // nothing else matters for parsing or processing.
597 PP.SetMacroExpansionOnlyInDirectives();
598 do {
599 PP.Lex(Tok);
600 } while (Tok.isNot(tok::eof));
Argyrios Kyrtzidis17ff2e52013-07-26 15:32:04 +0000601 Rewrite->setPredefinesBuffer(SM.getBuffer(PP.getPredefinesFileID()));
602 Rewrite->Process(PP.getPredefinesFileID(), SrcMgr::C_User);
David Blaikied5321242012-06-06 18:52:13 +0000603 Rewrite->Process(SM.getMainFileID(), SrcMgr::C_User);
604 OS->flush();
605}