blob: ff18c3f32456f4195faec4d67cdfb1d7705d8c38 [file] [log] [blame]
Chris Lattner22eb9722006-06-18 05:43:12 +00001//===--- Preprocess.cpp - C Language Family Preprocessor Implementation ---===//
2//
3// The LLVM Compiler Infrastructure
4//
Chris Lattner5b12ab82007-12-29 19:59:25 +00005// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
Chris Lattner22eb9722006-06-18 05:43:12 +00007//
8//===----------------------------------------------------------------------===//
9//
10// This file implements the Preprocessor interface.
11//
12//===----------------------------------------------------------------------===//
13//
Chris Lattner22eb9722006-06-18 05:43:12 +000014// Options to support:
15// -H - Print the name of each header file used.
Chris Lattner1630c3c2009-02-06 06:45:26 +000016// -d[DNI] - Dump various things.
Chris Lattner22eb9722006-06-18 05:43:12 +000017// -fworking-directory - #line's with preprocessor's working dir.
18// -fpreprocessed
19// -dependency-file,-M,-MM,-MF,-MG,-MP,-MT,-MQ,-MD,-MMD
20// -W*
21// -w
22//
23// Messages to emit:
24// "Multiple include guards may be useful for:\n"
25//
Chris Lattner22eb9722006-06-18 05:43:12 +000026//===----------------------------------------------------------------------===//
27
28#include "clang/Lex/Preprocessor.h"
Chris Lattnerd19564b2009-12-15 01:51:03 +000029#include "MacroArgs.h"
Douglas Gregor9882a5a2010-01-04 19:18:44 +000030#include "clang/Lex/ExternalPreprocessorSource.h"
Chris Lattner07b019a2006-10-22 07:28:56 +000031#include "clang/Lex/HeaderSearch.h"
Chris Lattner22eb9722006-06-18 05:43:12 +000032#include "clang/Lex/MacroInfo.h"
Chris Lattnerb8761832006-06-24 21:31:03 +000033#include "clang/Lex/Pragma.h"
Douglas Gregor7f6d60d2010-03-19 16:15:56 +000034#include "clang/Lex/PreprocessingRecord.h"
Chris Lattner0b8cfc22006-06-28 06:49:17 +000035#include "clang/Lex/ScratchBuffer.h"
Chris Lattner60f36222009-01-29 05:15:15 +000036#include "clang/Lex/LexDiagnostic.h"
Douglas Gregor3a7ad252010-08-24 19:08:16 +000037#include "clang/Lex/CodeCompletionHandler.h"
Chris Lattner22eb9722006-06-18 05:43:12 +000038#include "clang/Basic/SourceManager.h"
Ted Kremeneka5c2c272009-02-12 03:26:59 +000039#include "clang/Basic/FileManager.h"
Chris Lattner81278c62006-10-14 19:03:49 +000040#include "clang/Basic/TargetInfo.h"
Chris Lattner5cd83512008-10-05 20:40:30 +000041#include "llvm/ADT/APFloat.h"
Chris Lattner7a4af3b2006-07-26 06:26:52 +000042#include "llvm/ADT/SmallVector.h"
Chris Lattner8a7003c2007-07-16 06:48:38 +000043#include "llvm/Support/MemoryBuffer.h"
Benjamin Kramer89b422c2009-08-23 12:08:50 +000044#include "llvm/Support/raw_ostream.h"
Chris Lattner22eb9722006-06-18 05:43:12 +000045using namespace clang;
46
47//===----------------------------------------------------------------------===//
Douglas Gregor9882a5a2010-01-04 19:18:44 +000048ExternalPreprocessorSource::~ExternalPreprocessorSource() { }
Chris Lattner22eb9722006-06-18 05:43:12 +000049
Chris Lattner02dffbd2006-10-14 07:50:21 +000050Preprocessor::Preprocessor(Diagnostic &diags, const LangOptions &opts,
Daniel Dunbar1b444192009-11-13 05:51:54 +000051 const TargetInfo &target, SourceManager &SM,
Ted Kremeneka705b042009-01-15 18:47:46 +000052 HeaderSearch &Headers,
Daniel Dunbar0c6c9302009-11-11 21:44:21 +000053 IdentifierInfoLookup* IILookup,
54 bool OwnsHeaders)
Chris Lattner83aba002009-03-13 21:17:43 +000055 : Diags(&diags), Features(opts), Target(target),FileMgr(Headers.getFileMgr()),
Argyrios Kyrtzidis71731d62010-11-03 22:45:23 +000056 FileSystemOpts(Headers.getFileSystemOpts()), SourceMgr(SM),
57 HeaderInfo(Headers), ExternalSource(0),
Douglas Gregor3a7ad252010-08-24 19:08:16 +000058 Identifiers(opts, IILookup), BuiltinInfo(Target), CodeComplete(0),
59 CodeCompletionFile(0), SkipMainFilePreamble(0, true), CurPPLexer(0),
Ted Kremenekc8456f82010-10-19 22:15:20 +000060 CurDirLookup(0), Callbacks(0), MacroArgCache(0), Record(0), MIChainHead(0),
61 MICache(0) {
Chris Lattner0b8cfc22006-06-28 06:49:17 +000062 ScratchBuf = new ScratchBuffer(SourceMgr);
Chris Lattner0af3ba12009-04-13 01:29:17 +000063 CounterValue = 0; // __COUNTER__ starts at 0.
Daniel Dunbar0c6c9302009-11-11 21:44:21 +000064 OwnsHeaderSearch = OwnsHeaders;
Mike Stump11289f42009-09-09 15:08:12 +000065
Chris Lattner22eb9722006-06-18 05:43:12 +000066 // Clear stats.
Chris Lattner59a9ebd2006-10-18 05:34:33 +000067 NumDirectives = NumDefined = NumUndefined = NumPragma = 0;
Chris Lattner22eb9722006-06-18 05:43:12 +000068 NumIf = NumElse = NumEndif = 0;
Chris Lattner78186052006-07-09 00:45:31 +000069 NumEnteredSourceFiles = 0;
70 NumMacroExpanded = NumFnMacroExpanded = NumBuiltinMacroExpanded = 0;
Chris Lattner510ab612006-07-20 04:47:30 +000071 NumFastMacroExpanded = NumTokenPaste = NumFastTokenPaste = 0;
Mike Stump11289f42009-09-09 15:08:12 +000072 MaxIncludeStackDepth = 0;
Chris Lattner22eb9722006-06-18 05:43:12 +000073 NumSkipped = 0;
Chris Lattnerb352e3e2006-11-21 06:17:10 +000074
75 // Default to discarding comments.
76 KeepComments = false;
77 KeepMacroComments = false;
Mike Stump11289f42009-09-09 15:08:12 +000078
Chris Lattner22eb9722006-06-18 05:43:12 +000079 // Macro expansion is enabled.
80 DisableMacroExpansion = false;
Chris Lattneree8760b2006-07-15 07:42:55 +000081 InMacroArgs = false;
Chris Lattner285c0c12008-03-09 02:26:03 +000082 NumCachedTokenLexers = 0;
Chris Lattner0c885f52006-06-21 06:50:18 +000083
Argyrios Kyrtzidisb3dd1e02008-08-10 13:15:22 +000084 CachedLexPos = 0;
85
Douglas Gregor9882a5a2010-01-04 19:18:44 +000086 // We haven't read anything from the external source.
87 ReadMacrosFromExternalSource = false;
Kovarththanan Rajaratnamba2c6522010-03-13 10:17:05 +000088
Chris Lattner8ff71992006-07-06 05:17:39 +000089 // "Poison" __VA_ARGS__, which can only appear in the expansion of a macro.
90 // This gets unpoisoned where it is allowed.
91 (Ident__VA_ARGS__ = getIdentifierInfo("__VA_ARGS__"))->setIsPoisoned();
Mike Stump11289f42009-09-09 15:08:12 +000092
Chris Lattnerb8761832006-06-24 21:31:03 +000093 // Initialize the pragma handlers.
Argyrios Kyrtzidis36745fd2010-07-13 09:07:17 +000094 PragmaHandlers = new PragmaNamespace(llvm::StringRef());
Chris Lattnerb8761832006-06-24 21:31:03 +000095 RegisterBuiltinPragmas();
Mike Stump11289f42009-09-09 15:08:12 +000096
Chris Lattner677757a2006-06-28 05:26:32 +000097 // Initialize builtin macros like __LINE__ and friends.
98 RegisterBuiltinMacros();
Chris Lattner22eb9722006-06-18 05:43:12 +000099}
100
101Preprocessor::~Preprocessor() {
Argyrios Kyrtzidis5d240d02008-08-23 12:12:06 +0000102 assert(BacktrackPositions.empty() && "EnableBacktrack/Backtrack imbalance!");
103
Chris Lattner69772b02006-07-02 20:34:39 +0000104 while (!IncludeMacroStack.empty()) {
105 delete IncludeMacroStack.back().TheLexer;
Chris Lattner285c0c12008-03-09 02:26:03 +0000106 delete IncludeMacroStack.back().TheTokenLexer;
Chris Lattner69772b02006-07-02 20:34:39 +0000107 IncludeMacroStack.pop_back();
Chris Lattner22eb9722006-06-18 05:43:12 +0000108 }
Chris Lattnerc43ddc82007-10-07 08:44:20 +0000109
110 // Free any macro definitions.
Ted Kremenekb865f7e2010-10-19 21:30:11 +0000111 for (MacroInfoChain *I = MIChainHead ; I ; I = I->Next)
Ted Kremenek1f1e4bd2010-10-19 18:16:54 +0000112 I->MI.Destroy();
Mike Stump11289f42009-09-09 15:08:12 +0000113
Chris Lattnerc02c4ab2007-07-15 00:25:26 +0000114 // Free any cached macro expanders.
Chris Lattner285c0c12008-03-09 02:26:03 +0000115 for (unsigned i = 0, e = NumCachedTokenLexers; i != e; ++i)
116 delete TokenLexerCache[i];
Kovarththanan Rajaratnamba2c6522010-03-13 10:17:05 +0000117
Chris Lattnerd19564b2009-12-15 01:51:03 +0000118 // Free any cached MacroArgs.
119 for (MacroArgs *ArgList = MacroArgCache; ArgList; )
120 ArgList = ArgList->deallocate();
Mike Stump11289f42009-09-09 15:08:12 +0000121
Chris Lattnerb8761832006-06-24 21:31:03 +0000122 // Release pragma information.
123 delete PragmaHandlers;
Chris Lattner0b8cfc22006-06-28 06:49:17 +0000124
125 // Delete the scratch buffer info.
126 delete ScratchBuf;
Chris Lattner22192932008-03-14 06:07:05 +0000127
Daniel Dunbar0c6c9302009-11-11 21:44:21 +0000128 // Delete the header search info, if we own it.
129 if (OwnsHeaderSearch)
130 delete &HeaderInfo;
131
Chris Lattner22192932008-03-14 06:07:05 +0000132 delete Callbacks;
Chris Lattner22eb9722006-06-18 05:43:12 +0000133}
134
Ted Kremeneka5c2c272009-02-12 03:26:59 +0000135void Preprocessor::setPTHManager(PTHManager* pm) {
136 PTH.reset(pm);
Douglas Gregord2eb58a2009-10-16 18:18:30 +0000137 FileMgr.addStatCache(PTH->createStatCache());
Ted Kremeneka5c2c272009-02-12 03:26:59 +0000138}
139
Chris Lattner146762e2007-07-20 16:59:19 +0000140void Preprocessor::DumpToken(const Token &Tok, bool DumpFlags) const {
Benjamin Kramer89b422c2009-08-23 12:08:50 +0000141 llvm::errs() << tok::getTokenName(Tok.getKind()) << " '"
142 << getSpelling(Tok) << "'";
Mike Stump11289f42009-09-09 15:08:12 +0000143
Chris Lattnerd01e2912006-06-18 16:22:51 +0000144 if (!DumpFlags) return;
Mike Stump11289f42009-09-09 15:08:12 +0000145
Benjamin Kramer89b422c2009-08-23 12:08:50 +0000146 llvm::errs() << "\t";
Chris Lattnerd01e2912006-06-18 16:22:51 +0000147 if (Tok.isAtStartOfLine())
Benjamin Kramer89b422c2009-08-23 12:08:50 +0000148 llvm::errs() << " [StartOfLine]";
Chris Lattnerd01e2912006-06-18 16:22:51 +0000149 if (Tok.hasLeadingSpace())
Benjamin Kramer89b422c2009-08-23 12:08:50 +0000150 llvm::errs() << " [LeadingSpace]";
Chris Lattner6e4bf522006-07-27 06:59:25 +0000151 if (Tok.isExpandDisabled())
Benjamin Kramer89b422c2009-08-23 12:08:50 +0000152 llvm::errs() << " [ExpandDisabled]";
Chris Lattnerd01e2912006-06-18 16:22:51 +0000153 if (Tok.needsCleaning()) {
Chris Lattner50b497e2006-06-18 16:32:35 +0000154 const char *Start = SourceMgr.getCharacterData(Tok.getLocation());
Benjamin Kramere8394df2010-08-11 14:47:12 +0000155 llvm::errs() << " [UnClean='" << llvm::StringRef(Start, Tok.getLength())
Benjamin Kramer89b422c2009-08-23 12:08:50 +0000156 << "']";
Chris Lattnerd01e2912006-06-18 16:22:51 +0000157 }
Mike Stump11289f42009-09-09 15:08:12 +0000158
Benjamin Kramer89b422c2009-08-23 12:08:50 +0000159 llvm::errs() << "\tLoc=<";
Chris Lattner615315f2007-12-09 20:31:55 +0000160 DumpLocation(Tok.getLocation());
Benjamin Kramer89b422c2009-08-23 12:08:50 +0000161 llvm::errs() << ">";
Chris Lattner615315f2007-12-09 20:31:55 +0000162}
163
164void Preprocessor::DumpLocation(SourceLocation Loc) const {
Chris Lattnerf1ca7d32009-01-27 07:57:44 +0000165 Loc.dump(SourceMgr);
Chris Lattnerd01e2912006-06-18 16:22:51 +0000166}
167
168void Preprocessor::DumpMacro(const MacroInfo &MI) const {
Benjamin Kramer89b422c2009-08-23 12:08:50 +0000169 llvm::errs() << "MACRO: ";
Chris Lattnerd01e2912006-06-18 16:22:51 +0000170 for (unsigned i = 0, e = MI.getNumTokens(); i != e; ++i) {
171 DumpToken(MI.getReplacementToken(i));
Benjamin Kramer89b422c2009-08-23 12:08:50 +0000172 llvm::errs() << " ";
Chris Lattnerd01e2912006-06-18 16:22:51 +0000173 }
Benjamin Kramer89b422c2009-08-23 12:08:50 +0000174 llvm::errs() << "\n";
Chris Lattnerd01e2912006-06-18 16:22:51 +0000175}
176
Chris Lattner22eb9722006-06-18 05:43:12 +0000177void Preprocessor::PrintStats() {
Benjamin Kramer89b422c2009-08-23 12:08:50 +0000178 llvm::errs() << "\n*** Preprocessor Stats:\n";
179 llvm::errs() << NumDirectives << " directives found:\n";
180 llvm::errs() << " " << NumDefined << " #define.\n";
181 llvm::errs() << " " << NumUndefined << " #undef.\n";
182 llvm::errs() << " #include/#include_next/#import:\n";
183 llvm::errs() << " " << NumEnteredSourceFiles << " source files entered.\n";
184 llvm::errs() << " " << MaxIncludeStackDepth << " max include stack depth\n";
185 llvm::errs() << " " << NumIf << " #if/#ifndef/#ifdef.\n";
186 llvm::errs() << " " << NumElse << " #else/#elif.\n";
187 llvm::errs() << " " << NumEndif << " #endif.\n";
188 llvm::errs() << " " << NumPragma << " #pragma.\n";
189 llvm::errs() << NumSkipped << " #if/#ifndef#ifdef regions skipped\n";
Chris Lattner22eb9722006-06-18 05:43:12 +0000190
Benjamin Kramer89b422c2009-08-23 12:08:50 +0000191 llvm::errs() << NumMacroExpanded << "/" << NumFnMacroExpanded << "/"
Ted Kremeneka0a3e9b2008-01-14 16:44:48 +0000192 << NumBuiltinMacroExpanded << " obj/fn/builtin macros expanded, "
193 << NumFastMacroExpanded << " on the fast path.\n";
Benjamin Kramer89b422c2009-08-23 12:08:50 +0000194 llvm::errs() << (NumFastTokenPaste+NumTokenPaste)
Ted Kremeneka0a3e9b2008-01-14 16:44:48 +0000195 << " token paste (##) operations performed, "
196 << NumFastTokenPaste << " on the fast path.\n";
Chris Lattner22eb9722006-06-18 05:43:12 +0000197}
198
Kovarththanan Rajaratnamba2c6522010-03-13 10:17:05 +0000199Preprocessor::macro_iterator
200Preprocessor::macro_begin(bool IncludeExternalMacros) const {
201 if (IncludeExternalMacros && ExternalSource &&
Douglas Gregor9882a5a2010-01-04 19:18:44 +0000202 !ReadMacrosFromExternalSource) {
203 ReadMacrosFromExternalSource = true;
204 ExternalSource->ReadDefinedMacros();
205 }
Kovarththanan Rajaratnamba2c6522010-03-13 10:17:05 +0000206
207 return Macros.begin();
Douglas Gregor9882a5a2010-01-04 19:18:44 +0000208}
209
Kovarththanan Rajaratnamba2c6522010-03-13 10:17:05 +0000210Preprocessor::macro_iterator
211Preprocessor::macro_end(bool IncludeExternalMacros) const {
212 if (IncludeExternalMacros && ExternalSource &&
Douglas Gregor9882a5a2010-01-04 19:18:44 +0000213 !ReadMacrosFromExternalSource) {
214 ReadMacrosFromExternalSource = true;
215 ExternalSource->ReadDefinedMacros();
216 }
Kovarththanan Rajaratnamba2c6522010-03-13 10:17:05 +0000217
218 return Macros.end();
Douglas Gregor9882a5a2010-01-04 19:18:44 +0000219}
220
Kovarththanan Rajaratnamba2c6522010-03-13 10:17:05 +0000221bool Preprocessor::SetCodeCompletionPoint(const FileEntry *File,
222 unsigned TruncateAtLine,
Douglas Gregor53ad6b92009-12-02 06:49:09 +0000223 unsigned TruncateAtColumn) {
224 using llvm::MemoryBuffer;
225
226 CodeCompletionFile = File;
227
228 // Okay to clear out the code-completion point by passing NULL.
229 if (!CodeCompletionFile)
230 return false;
231
232 // Load the actual file's contents.
Douglas Gregor26266da2010-03-16 19:49:24 +0000233 bool Invalid = false;
234 const MemoryBuffer *Buffer = SourceMgr.getMemoryBufferForFile(File, &Invalid);
235 if (Invalid)
Douglas Gregor53ad6b92009-12-02 06:49:09 +0000236 return true;
237
238 // Find the byte position of the truncation point.
239 const char *Position = Buffer->getBufferStart();
240 for (unsigned Line = 1; Line < TruncateAtLine; ++Line) {
241 for (; *Position; ++Position) {
242 if (*Position != '\r' && *Position != '\n')
243 continue;
Kovarththanan Rajaratnamba2c6522010-03-13 10:17:05 +0000244
Douglas Gregor53ad6b92009-12-02 06:49:09 +0000245 // Eat \r\n or \n\r as a single line.
246 if ((Position[1] == '\r' || Position[1] == '\n') &&
247 Position[0] != Position[1])
248 ++Position;
249 ++Position;
250 break;
251 }
252 }
Kovarththanan Rajaratnamba2c6522010-03-13 10:17:05 +0000253
Douglas Gregor9291ab62009-12-08 21:45:46 +0000254 Position += TruncateAtColumn - 1;
Kovarththanan Rajaratnamba2c6522010-03-13 10:17:05 +0000255
Douglas Gregor53ad6b92009-12-02 06:49:09 +0000256 // Truncate the buffer.
Douglas Gregor9291ab62009-12-08 21:45:46 +0000257 if (Position < Buffer->getBufferEnd()) {
Chris Lattner58c79342010-04-05 22:42:27 +0000258 llvm::StringRef Data(Buffer->getBufferStart(),
259 Position-Buffer->getBufferStart());
Kovarththanan Rajaratnamba2c6522010-03-13 10:17:05 +0000260 MemoryBuffer *TruncatedBuffer
Chris Lattner58c79342010-04-05 22:42:27 +0000261 = MemoryBuffer::getMemBufferCopy(Data, Buffer->getBufferIdentifier());
Douglas Gregor53ad6b92009-12-02 06:49:09 +0000262 SourceMgr.overrideFileContents(File, TruncatedBuffer);
263 }
264
265 return false;
266}
267
Douglas Gregor5f498832009-12-03 17:05:59 +0000268bool Preprocessor::isCodeCompletionFile(SourceLocation FileLoc) const {
Douglas Gregor53ad6b92009-12-02 06:49:09 +0000269 return CodeCompletionFile && FileLoc.isFileID() &&
270 SourceMgr.getFileEntryForID(SourceMgr.getFileID(FileLoc))
271 == CodeCompletionFile;
272}
273
Douglas Gregor11583702010-08-25 17:04:25 +0000274void Preprocessor::CodeCompleteNaturalLanguage() {
275 SetCodeCompletionPoint(0, 0, 0);
276 getDiagnostics().setSuppressAllDiagnostics(true);
277 if (CodeComplete)
278 CodeComplete->CodeCompleteNaturalLanguage();
279}
280
Chris Lattner22eb9722006-06-18 05:43:12 +0000281//===----------------------------------------------------------------------===//
Chris Lattnerd01e2912006-06-18 16:22:51 +0000282// Token Spelling
283//===----------------------------------------------------------------------===//
284
Chris Lattnerd01e2912006-06-18 16:22:51 +0000285/// getSpelling() - Return the 'spelling' of this token. The spelling of a
286/// token are the characters used to represent the token in the source file
287/// after trigraph expansion and escaped-newline folding. In particular, this
288/// wants to get the true, uncanonicalized, spelling of things like digraphs
289/// UCNs, etc.
Daniel Dunbarbf410c62009-11-14 01:20:48 +0000290std::string Preprocessor::getSpelling(const Token &Tok,
291 const SourceManager &SourceMgr,
Douglas Gregor7bda4b82010-03-16 05:20:39 +0000292 const LangOptions &Features,
293 bool *Invalid) {
Chris Lattnerd01e2912006-06-18 16:22:51 +0000294 assert((int)Tok.getLength() >= 0 && "Token character range is bogus!");
Ted Kremenek8d178f42009-01-27 00:01:05 +0000295
Chris Lattnerd01e2912006-06-18 16:22:51 +0000296 // If this token contains nothing interesting, return it directly.
Douglas Gregor7bda4b82010-03-16 05:20:39 +0000297 bool CharDataInvalid = false;
298 const char* TokStart = SourceMgr.getCharacterData(Tok.getLocation(),
299 &CharDataInvalid);
300 if (Invalid)
301 *Invalid = CharDataInvalid;
302 if (CharDataInvalid)
303 return std::string();
304
Chris Lattnerd01e2912006-06-18 16:22:51 +0000305 if (!Tok.needsCleaning())
306 return std::string(TokStart, TokStart+Tok.getLength());
Mike Stump11289f42009-09-09 15:08:12 +0000307
Chris Lattnerd01e2912006-06-18 16:22:51 +0000308 std::string Result;
309 Result.reserve(Tok.getLength());
Mike Stump11289f42009-09-09 15:08:12 +0000310
Chris Lattneref9eae12006-07-04 22:33:12 +0000311 // Otherwise, hard case, relex the characters into the string.
Chris Lattnerd01e2912006-06-18 16:22:51 +0000312 for (const char *Ptr = TokStart, *End = TokStart+Tok.getLength();
313 Ptr != End; ) {
314 unsigned CharSize;
315 Result.push_back(Lexer::getCharAndSizeNoWarn(Ptr, CharSize, Features));
316 Ptr += CharSize;
317 }
318 assert(Result.size() != unsigned(Tok.getLength()) &&
319 "NeedsCleaning flag set on something that didn't need cleaning!");
320 return Result;
321}
322
Daniel Dunbarbf410c62009-11-14 01:20:48 +0000323/// getSpelling() - Return the 'spelling' of this token. The spelling of a
324/// token are the characters used to represent the token in the source file
325/// after trigraph expansion and escaped-newline folding. In particular, this
326/// wants to get the true, uncanonicalized, spelling of things like digraphs
327/// UCNs, etc.
Douglas Gregor7bda4b82010-03-16 05:20:39 +0000328std::string Preprocessor::getSpelling(const Token &Tok, bool *Invalid) const {
329 return getSpelling(Tok, SourceMgr, Features, Invalid);
Daniel Dunbarbf410c62009-11-14 01:20:48 +0000330}
331
Chris Lattnerd01e2912006-06-18 16:22:51 +0000332/// getSpelling - This method is used to get the spelling of a token into a
333/// preallocated buffer, instead of as an std::string. The caller is required
334/// to allocate enough space for the token, which is guaranteed to be at least
335/// Tok.getLength() bytes long. The actual length of the token is returned.
Chris Lattneref9eae12006-07-04 22:33:12 +0000336///
337/// Note that this method may do two possible things: it may either fill in
338/// the buffer specified with characters, or it may *change the input pointer*
339/// to point to a constant buffer with the data already in it (avoiding a
340/// copy). The caller is not allowed to modify the returned buffer pointer
341/// if an internal buffer is returned.
Alexis Hunt3b791862010-08-30 17:47:05 +0000342unsigned Preprocessor::getSpelling(const Token &Tok,
343 const char *&Buffer, bool *Invalid) const {
Chris Lattnerd01e2912006-06-18 16:22:51 +0000344 assert((int)Tok.getLength() >= 0 && "Token character range is bogus!");
Mike Stump11289f42009-09-09 15:08:12 +0000345
Chris Lattnerd3a15f72006-07-04 23:01:03 +0000346 // If this token is an identifier, just return the string from the identifier
347 // table, which is very quick.
348 if (const IdentifierInfo *II = Tok.getIdentifierInfo()) {
Alexis Hunt3b791862010-08-30 17:47:05 +0000349 Buffer = II->getNameStart();
350 return II->getLength();
Chris Lattnerd3a15f72006-07-04 23:01:03 +0000351 }
Ted Kremenek884a5582009-01-08 02:47:16 +0000352
Chris Lattnerd3a15f72006-07-04 23:01:03 +0000353 // Otherwise, compute the start of the token in the input lexer buffer.
Chris Lattner5a7971e2009-01-26 19:29:26 +0000354 const char *TokStart = 0;
Mike Stump11289f42009-09-09 15:08:12 +0000355
Chris Lattner5a7971e2009-01-26 19:29:26 +0000356 if (Tok.isLiteral())
357 TokStart = Tok.getLiteralData();
Mike Stump11289f42009-09-09 15:08:12 +0000358
Douglas Gregor7bda4b82010-03-16 05:20:39 +0000359 if (TokStart == 0) {
360 bool CharDataInvalid = false;
361 TokStart = SourceMgr.getCharacterData(Tok.getLocation(), &CharDataInvalid);
362 if (Invalid)
363 *Invalid = CharDataInvalid;
364 if (CharDataInvalid) {
365 Buffer = "";
366 return 0;
367 }
368 }
Chris Lattnerd01e2912006-06-18 16:22:51 +0000369
370 // If this token contains nothing interesting, return it directly.
Alexis Hunt3b791862010-08-30 17:47:05 +0000371 if (!Tok.needsCleaning()) {
Chris Lattneref9eae12006-07-04 22:33:12 +0000372 Buffer = TokStart;
Alexis Hunt3b791862010-08-30 17:47:05 +0000373 return Tok.getLength();
Chris Lattnerd01e2912006-06-18 16:22:51 +0000374 }
Mike Stump11289f42009-09-09 15:08:12 +0000375
Chris Lattnerd01e2912006-06-18 16:22:51 +0000376 // Otherwise, hard case, relex the characters into the string.
Chris Lattneref9eae12006-07-04 22:33:12 +0000377 char *OutBuf = const_cast<char*>(Buffer);
Alexis Hunt3b791862010-08-30 17:47:05 +0000378 for (const char *Ptr = TokStart, *End = TokStart+Tok.getLength();
Chris Lattnerd01e2912006-06-18 16:22:51 +0000379 Ptr != End; ) {
380 unsigned CharSize;
381 *OutBuf++ = Lexer::getCharAndSizeNoWarn(Ptr, CharSize, Features);
382 Ptr += CharSize;
383 }
Alexis Hunt3b791862010-08-30 17:47:05 +0000384 assert(unsigned(OutBuf-Buffer) != Tok.getLength() &&
Chris Lattnerd01e2912006-06-18 16:22:51 +0000385 "NeedsCleaning flag set on something that didn't need cleaning!");
Mike Stump11289f42009-09-09 15:08:12 +0000386
Chris Lattnerd01e2912006-06-18 16:22:51 +0000387 return OutBuf-Buffer;
388}
389
Benjamin Kramera197fb62010-02-27 17:05:45 +0000390/// getSpelling - This method is used to get the spelling of a token into a
391/// SmallVector. Note that the returned StringRef may not point to the
392/// supplied buffer if a copy can be avoided.
393llvm::StringRef Preprocessor::getSpelling(const Token &Tok,
Douglas Gregor7bda4b82010-03-16 05:20:39 +0000394 llvm::SmallVectorImpl<char> &Buffer,
395 bool *Invalid) const {
Benjamin Kramera197fb62010-02-27 17:05:45 +0000396 // Try the fast path.
397 if (const IdentifierInfo *II = Tok.getIdentifierInfo())
398 return II->getName();
399
400 // Resize the buffer if we need to copy into it.
401 if (Tok.needsCleaning())
402 Buffer.resize(Tok.getLength());
403
404 const char *Ptr = Buffer.data();
Douglas Gregor7bda4b82010-03-16 05:20:39 +0000405 unsigned Len = getSpelling(Tok, Ptr, Invalid);
Benjamin Kramera197fb62010-02-27 17:05:45 +0000406 return llvm::StringRef(Ptr, Len);
407}
408
Chris Lattnerb94ec7b2006-07-14 06:54:10 +0000409/// CreateString - Plop the specified string into a scratch buffer and return a
410/// location for it. If specified, the source location provides a source
411/// location for the token.
Chris Lattner5a7971e2009-01-26 19:29:26 +0000412void Preprocessor::CreateString(const char *Buf, unsigned Len, Token &Tok,
413 SourceLocation InstantiationLoc) {
414 Tok.setLength(Len);
Mike Stump11289f42009-09-09 15:08:12 +0000415
Chris Lattner5a7971e2009-01-26 19:29:26 +0000416 const char *DestPtr;
417 SourceLocation Loc = ScratchBuf->getToken(Buf, Len, DestPtr);
Mike Stump11289f42009-09-09 15:08:12 +0000418
Chris Lattner5a7971e2009-01-26 19:29:26 +0000419 if (InstantiationLoc.isValid())
Chris Lattner9dc9c202009-02-15 20:52:18 +0000420 Loc = SourceMgr.createInstantiationLoc(Loc, InstantiationLoc,
421 InstantiationLoc, Len);
Chris Lattner5a7971e2009-01-26 19:29:26 +0000422 Tok.setLocation(Loc);
Mike Stump11289f42009-09-09 15:08:12 +0000423
Chris Lattner5a7971e2009-01-26 19:29:26 +0000424 // If this is a literal token, set the pointer data.
425 if (Tok.isLiteral())
426 Tok.setLiteralData(DestPtr);
Chris Lattnerb94ec7b2006-07-14 06:54:10 +0000427}
428
429
Chris Lattner8a7003c2007-07-16 06:48:38 +0000430/// AdvanceToTokenCharacter - Given a location that specifies the start of a
431/// token, return a new location that specifies a character within the token.
Mike Stump11289f42009-09-09 15:08:12 +0000432SourceLocation Preprocessor::AdvanceToTokenCharacter(SourceLocation TokStart,
Chris Lattner8a7003c2007-07-16 06:48:38 +0000433 unsigned CharNo) {
Chris Lattner8a425862009-01-16 07:36:28 +0000434 // Figure out how many physical characters away the specified instantiation
Chris Lattner8a7003c2007-07-16 06:48:38 +0000435 // character is. This needs to take into consideration newlines and
436 // trigraphs.
Douglas Gregor42fe8582010-03-16 20:46:42 +0000437 bool Invalid = false;
438 const char *TokPtr = SourceMgr.getCharacterData(TokStart, &Invalid);
Mike Stump11289f42009-09-09 15:08:12 +0000439
Chris Lattner93017cc2009-04-18 22:28:58 +0000440 // If they request the first char of the token, we're trivially done.
Douglas Gregor42fe8582010-03-16 20:46:42 +0000441 if (Invalid || (CharNo == 0 && Lexer::isObviouslySimpleCharacter(*TokPtr)))
Chris Lattner93017cc2009-04-18 22:28:58 +0000442 return TokStart;
Mike Stump11289f42009-09-09 15:08:12 +0000443
Chris Lattnerdc5c0552007-07-20 16:37:10 +0000444 unsigned PhysOffset = 0;
Mike Stump11289f42009-09-09 15:08:12 +0000445
Chris Lattner8a7003c2007-07-16 06:48:38 +0000446 // The usual case is that tokens don't contain anything interesting. Skip
447 // over the uninteresting characters. If a token only consists of simple
448 // chars, this method is extremely fast.
Chris Lattner93017cc2009-04-18 22:28:58 +0000449 while (Lexer::isObviouslySimpleCharacter(*TokPtr)) {
450 if (CharNo == 0)
451 return TokStart.getFileLocWithOffset(PhysOffset);
Chris Lattnerdc5c0552007-07-20 16:37:10 +0000452 ++TokPtr, --CharNo, ++PhysOffset;
Chris Lattner93017cc2009-04-18 22:28:58 +0000453 }
Mike Stump11289f42009-09-09 15:08:12 +0000454
Chris Lattner8ddb5cf2009-01-17 07:57:25 +0000455 // If we have a character that may be a trigraph or escaped newline, use a
Chris Lattner8a7003c2007-07-16 06:48:38 +0000456 // lexer to parse it correctly.
Chris Lattner93017cc2009-04-18 22:28:58 +0000457 for (; CharNo; --CharNo) {
458 unsigned Size;
459 Lexer::getCharAndSizeNoWarn(TokPtr, Size, Features);
460 TokPtr += Size;
461 PhysOffset += Size;
Chris Lattner8a7003c2007-07-16 06:48:38 +0000462 }
Mike Stump11289f42009-09-09 15:08:12 +0000463
Chris Lattner93017cc2009-04-18 22:28:58 +0000464 // Final detail: if we end up on an escaped newline, we want to return the
465 // location of the actual byte of the token. For example foo\<newline>bar
466 // advanced by 3 should return the location of b, not of \\. One compounding
467 // detail of this is that the escape may be made by a trigraph.
468 if (!Lexer::isObviouslySimpleCharacter(*TokPtr))
Ted Kremenekdb4b7712010-01-29 19:38:24 +0000469 PhysOffset += Lexer::SkipEscapedNewLines(TokPtr)-TokPtr;
Mike Stump11289f42009-09-09 15:08:12 +0000470
Chris Lattnerdc5c0552007-07-20 16:37:10 +0000471 return TokStart.getFileLocWithOffset(PhysOffset);
Chris Lattner8a7003c2007-07-16 06:48:38 +0000472}
473
Kovarththanan Rajaratnamba2c6522010-03-13 10:17:05 +0000474SourceLocation Preprocessor::getLocForEndOfToken(SourceLocation Loc,
Douglas Gregor562c1f92010-01-22 19:49:59 +0000475 unsigned Offset) {
Douglas Gregor96977da2009-02-27 17:53:17 +0000476 if (Loc.isInvalid() || !Loc.isFileID())
477 return SourceLocation();
478
Chris Lattner184e65d2009-04-14 23:22:57 +0000479 unsigned Len = Lexer::MeasureTokenLength(Loc, getSourceManager(), Features);
Douglas Gregor562c1f92010-01-22 19:49:59 +0000480 if (Len > Offset)
481 Len = Len - Offset;
482 else
483 return Loc;
Kovarththanan Rajaratnamba2c6522010-03-13 10:17:05 +0000484
Douglas Gregor96977da2009-02-27 17:53:17 +0000485 return AdvanceToTokenCharacter(Loc, Len);
486}
487
488
Chris Lattner8a7003c2007-07-16 06:48:38 +0000489
Chris Lattner1f1b0db2007-10-09 22:10:18 +0000490//===----------------------------------------------------------------------===//
491// Preprocessor Initialization Methods
492//===----------------------------------------------------------------------===//
493
Chris Lattner1f1b0db2007-10-09 22:10:18 +0000494
495/// EnterMainSourceFile - Enter the specified FileID as the main source file,
Nate Begemanf7c3ff62008-01-07 04:01:26 +0000496/// which implicitly adds the builtin defines etc.
Chris Lattnerfb24a3a2010-04-20 20:35:58 +0000497void Preprocessor::EnterMainSourceFile() {
Chris Lattner9ef847b2009-02-13 19:33:24 +0000498 // We do not allow the preprocessor to reenter the main file. Doing so will
499 // cause FileID's to accumulate information from both runs (e.g. #line
500 // information) and predefined macros aren't guaranteed to be set properly.
501 assert(NumEnteredSourceFiles == 0 && "Cannot reenter the main file!");
Chris Lattnerd32480d2009-01-17 06:22:33 +0000502 FileID MainFileID = SourceMgr.getMainFileID();
Mike Stump11289f42009-09-09 15:08:12 +0000503
Chris Lattner1f1b0db2007-10-09 22:10:18 +0000504 // Enter the main file source buffer.
Chris Lattnerfb24a3a2010-04-20 20:35:58 +0000505 EnterSourceFile(MainFileID, 0, SourceLocation());
Mike Stump11289f42009-09-09 15:08:12 +0000506
Douglas Gregor3f4bea02010-07-26 21:36:20 +0000507 // If we've been asked to skip bytes in the main file (e.g., as part of a
508 // precompiled preamble), do so now.
509 if (SkipMainFilePreamble.first > 0)
510 CurLexer->SkipBytes(SkipMainFilePreamble.first,
511 SkipMainFilePreamble.second);
512
Chris Lattner609d4132007-11-15 19:07:47 +0000513 // Tell the header info that the main file was entered. If the file is later
514 // #imported, it won't be re-entered.
Chris Lattnerd32480d2009-01-17 06:22:33 +0000515 if (const FileEntry *FE = SourceMgr.getFileEntryForID(MainFileID))
Chris Lattner609d4132007-11-15 19:07:47 +0000516 HeaderInfo.IncrementIncludeCount(FE);
Mike Stump11289f42009-09-09 15:08:12 +0000517
Benjamin Kramerd77adb52009-12-31 15:33:09 +0000518 // Preprocess Predefines to populate the initial preprocessor state.
Mike Stump11289f42009-09-09 15:08:12 +0000519 llvm::MemoryBuffer *SB =
Chris Lattner58c79342010-04-05 22:42:27 +0000520 llvm::MemoryBuffer::getMemBufferCopy(Predefines, "<built-in>");
Douglas Gregor33551892010-08-26 14:07:34 +0000521 assert(SB && "Cannot create predefined source buffer");
Chris Lattnerd32480d2009-01-17 06:22:33 +0000522 FileID FID = SourceMgr.createFileIDForMemBuffer(SB);
523 assert(!FID.isInvalid() && "Could not create FileID for predefines?");
Mike Stump11289f42009-09-09 15:08:12 +0000524
Chris Lattner1f1b0db2007-10-09 22:10:18 +0000525 // Start parsing the predefines.
Chris Lattnerfb24a3a2010-04-20 20:35:58 +0000526 EnterSourceFile(FID, 0, SourceLocation());
Chris Lattner1f1b0db2007-10-09 22:10:18 +0000527}
Chris Lattner8a7003c2007-07-16 06:48:38 +0000528
Daniel Dunbarcb9eaf52010-03-23 05:09:10 +0000529void Preprocessor::EndSourceFile() {
530 // Notify the client that we reached the end of the source file.
531 if (Callbacks)
532 Callbacks->EndOfMainFile();
533}
Chris Lattner677757a2006-06-28 05:26:32 +0000534
535//===----------------------------------------------------------------------===//
536// Lexer Event Handling.
537//===----------------------------------------------------------------------===//
538
Chris Lattnercefc7682006-07-08 08:28:12 +0000539/// LookUpIdentifierInfo - Given a tok::identifier token, look up the
540/// identifier information for the token and install it into the token.
Chris Lattner146762e2007-07-20 16:59:19 +0000541IdentifierInfo *Preprocessor::LookUpIdentifierInfo(Token &Identifier,
Daniel Dunbar07dcd8b2009-11-05 01:53:52 +0000542 const char *BufPtr) const {
Chris Lattner98c1f7c2007-10-09 18:02:16 +0000543 assert(Identifier.is(tok::identifier) && "Not an identifier!");
Chris Lattnercefc7682006-07-08 08:28:12 +0000544 assert(Identifier.getIdentifierInfo() == 0 && "Identinfo already exists!");
Mike Stump11289f42009-09-09 15:08:12 +0000545
Chris Lattnercefc7682006-07-08 08:28:12 +0000546 // Look up this token, see if it is a macro, or if it is a language keyword.
547 IdentifierInfo *II;
548 if (BufPtr && !Identifier.needsCleaning()) {
549 // No cleaning needed, just use the characters from the lexed buffer.
Daniel Dunbarf539bfe2009-11-05 01:53:39 +0000550 II = getIdentifierInfo(llvm::StringRef(BufPtr, Identifier.getLength()));
Chris Lattnercefc7682006-07-08 08:28:12 +0000551 } else {
552 // Cleaning needed, alloca a buffer, clean into it, then use the buffer.
Kovarththanan Rajaratnamba2c6522010-03-13 10:17:05 +0000553 llvm::SmallString<64> IdentifierBuffer;
Benjamin Kramer0a1abd42010-02-27 13:44:12 +0000554 llvm::StringRef CleanedStr = getSpelling(Identifier, IdentifierBuffer);
555 II = getIdentifierInfo(CleanedStr);
Chris Lattnercefc7682006-07-08 08:28:12 +0000556 }
Chris Lattner8c204872006-10-14 05:19:21 +0000557 Identifier.setIdentifierInfo(II);
Chris Lattnercefc7682006-07-08 08:28:12 +0000558 return II;
559}
560
561
Chris Lattner677757a2006-06-28 05:26:32 +0000562/// HandleIdentifier - This callback is invoked when the lexer reads an
563/// identifier. This callback looks up the identifier in the map and/or
564/// potentially macro expands it or turns it into a named token (like 'for').
Chris Lattnerad89ec02009-01-21 07:43:11 +0000565///
566/// Note that callers of this method are guarded by checking the
567/// IdentifierInfo's 'isHandleIdentifierCase' bit. If this method changes, the
568/// IdentifierInfo methods that compute these properties will need to change to
569/// match.
Chris Lattner146762e2007-07-20 16:59:19 +0000570void Preprocessor::HandleIdentifier(Token &Identifier) {
Chris Lattner0f1f5052006-07-20 04:16:23 +0000571 assert(Identifier.getIdentifierInfo() &&
572 "Can't handle identifiers without identifier info!");
Mike Stump11289f42009-09-09 15:08:12 +0000573
Chris Lattnerc79f6fb2006-07-04 17:53:21 +0000574 IdentifierInfo &II = *Identifier.getIdentifierInfo();
Chris Lattner677757a2006-06-28 05:26:32 +0000575
576 // If this identifier was poisoned, and if it was not produced from a macro
577 // expansion, emit an error.
Ted Kremeneka2c3c8d2008-11-19 22:43:49 +0000578 if (II.isPoisoned() && CurPPLexer) {
Chris Lattner8ff71992006-07-06 05:17:39 +0000579 if (&II != Ident__VA_ARGS__) // We warn about __VA_ARGS__ with poisoning.
580 Diag(Identifier, diag::err_pp_used_poisoned_id);
581 else
582 Diag(Identifier, diag::ext_pp_bad_vaargs_use);
583 }
Mike Stump11289f42009-09-09 15:08:12 +0000584
Chris Lattner78186052006-07-09 00:45:31 +0000585 // If this is a macro to be expanded, do it.
Chris Lattnerc43ddc82007-10-07 08:44:20 +0000586 if (MacroInfo *MI = getMacroInfo(&II)) {
Chris Lattner6e4bf522006-07-27 06:59:25 +0000587 if (!DisableMacroExpansion && !Identifier.isExpandDisabled()) {
588 if (MI->isEnabled()) {
589 if (!HandleMacroExpandedIdentifier(Identifier, MI))
590 return;
591 } else {
592 // C99 6.10.3.4p2 says that a disabled macro may never again be
593 // expanded, even if it's in a context where it could be expanded in the
594 // future.
Chris Lattner146762e2007-07-20 16:59:19 +0000595 Identifier.setFlag(Token::DisableExpand);
Chris Lattner6e4bf522006-07-27 06:59:25 +0000596 }
597 }
Chris Lattner063400e2006-10-14 19:54:15 +0000598 }
Chris Lattner677757a2006-06-28 05:26:32 +0000599
Chris Lattner5b9f4892006-11-21 17:23:33 +0000600 // C++ 2.11p2: If this is an alternative representation of a C++ operator,
601 // then we act as if it is the actual operator and not the textual
602 // representation of it.
Fariborz Jahanian9e42a952010-09-03 17:33:04 +0000603 if (II.isCPlusPlusOperatorKeyword())
Chris Lattner5b9f4892006-11-21 17:23:33 +0000604 Identifier.setIdentifierInfo(0);
605
Chris Lattner677757a2006-06-28 05:26:32 +0000606 // If this is an extension token, diagnose its use.
Steve Naroffc84e8b72008-09-02 18:50:17 +0000607 // We avoid diagnosing tokens that originate from macro definitions.
Eli Friedman6bba2ad2009-04-28 03:59:15 +0000608 // FIXME: This warning is disabled in cases where it shouldn't be,
609 // like "#define TY typeof", "TY(1) x".
610 if (II.isExtensionToken() && !DisableMacroExpansion)
Chris Lattner53621a52007-06-13 20:44:40 +0000611 Diag(Identifier, diag::ext_token_used);
Chris Lattner677757a2006-06-28 05:26:32 +0000612}
Douglas Gregorc6d5edd2009-07-02 17:08:52 +0000613
614void Preprocessor::AddCommentHandler(CommentHandler *Handler) {
615 assert(Handler && "NULL comment handler");
616 assert(std::find(CommentHandlers.begin(), CommentHandlers.end(), Handler) ==
617 CommentHandlers.end() && "Comment handler already registered");
618 CommentHandlers.push_back(Handler);
619}
620
621void Preprocessor::RemoveCommentHandler(CommentHandler *Handler) {
622 std::vector<CommentHandler *>::iterator Pos
623 = std::find(CommentHandlers.begin(), CommentHandlers.end(), Handler);
624 assert(Pos != CommentHandlers.end() && "Comment handler not registered");
625 CommentHandlers.erase(Pos);
626}
627
Chris Lattner87d02082010-01-18 22:35:47 +0000628bool Preprocessor::HandleComment(Token &result, SourceRange Comment) {
629 bool AnyPendingTokens = false;
Douglas Gregorc6d5edd2009-07-02 17:08:52 +0000630 for (std::vector<CommentHandler *>::iterator H = CommentHandlers.begin(),
631 HEnd = CommentHandlers.end();
Chris Lattner87d02082010-01-18 22:35:47 +0000632 H != HEnd; ++H) {
633 if ((*H)->HandleComment(*this, Comment))
634 AnyPendingTokens = true;
635 }
636 if (!AnyPendingTokens || getCommentRetentionState())
637 return false;
638 Lex(result);
639 return true;
Douglas Gregorc6d5edd2009-07-02 17:08:52 +0000640}
641
642CommentHandler::~CommentHandler() { }
Douglas Gregor7f6d60d2010-03-19 16:15:56 +0000643
Douglas Gregor3a7ad252010-08-24 19:08:16 +0000644CodeCompletionHandler::~CodeCompletionHandler() { }
645
Douglas Gregor7f6d60d2010-03-19 16:15:56 +0000646void Preprocessor::createPreprocessingRecord() {
647 if (Record)
648 return;
649
Douglas Gregor7dc87222010-03-19 17:12:43 +0000650 Record = new PreprocessingRecord;
651 addPPCallbacks(Record);
Douglas Gregor7f6d60d2010-03-19 16:15:56 +0000652}