|
150
|
1 //===- Preprocessor.cpp - C Language Family Preprocessor Implementation ---===//
|
|
|
2 //
|
|
|
3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
|
|
|
4 // See https://llvm.org/LICENSE.txt for license information.
|
|
|
5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
|
|
|
6 //
|
|
|
7 //===----------------------------------------------------------------------===//
|
|
|
8 //
|
|
|
9 // This file implements the Preprocessor interface.
|
|
|
10 //
|
|
|
11 //===----------------------------------------------------------------------===//
|
|
|
12 //
|
|
|
13 // Options to support:
|
|
|
14 // -H - Print the name of each header file used.
|
|
|
15 // -d[DNI] - Dump various things.
|
|
|
16 // -fworking-directory - #line's with preprocessor's working dir.
|
|
|
17 // -fpreprocessed
|
|
|
18 // -dependency-file,-M,-MM,-MF,-MG,-MP,-MT,-MQ,-MD,-MMD
|
|
|
19 // -W*
|
|
|
20 // -w
|
|
|
21 //
|
|
|
22 // Messages to emit:
|
|
|
23 // "Multiple include guards may be useful for:\n"
|
|
|
24 //
|
|
|
25 //===----------------------------------------------------------------------===//
|
|
|
26
|
|
|
27 #include "clang/Lex/Preprocessor.h"
|
|
|
28 #include "clang/Basic/Builtins.h"
|
|
|
29 #include "clang/Basic/FileManager.h"
|
|
|
30 #include "clang/Basic/FileSystemStatCache.h"
|
|
|
31 #include "clang/Basic/IdentifierTable.h"
|
|
|
32 #include "clang/Basic/LLVM.h"
|
|
|
33 #include "clang/Basic/LangOptions.h"
|
|
|
34 #include "clang/Basic/Module.h"
|
|
|
35 #include "clang/Basic/SourceLocation.h"
|
|
|
36 #include "clang/Basic/SourceManager.h"
|
|
|
37 #include "clang/Basic/TargetInfo.h"
|
|
|
38 #include "clang/Lex/CodeCompletionHandler.h"
|
|
|
39 #include "clang/Lex/ExternalPreprocessorSource.h"
|
|
|
40 #include "clang/Lex/HeaderSearch.h"
|
|
|
41 #include "clang/Lex/LexDiagnostic.h"
|
|
|
42 #include "clang/Lex/Lexer.h"
|
|
|
43 #include "clang/Lex/LiteralSupport.h"
|
|
|
44 #include "clang/Lex/MacroArgs.h"
|
|
|
45 #include "clang/Lex/MacroInfo.h"
|
|
|
46 #include "clang/Lex/ModuleLoader.h"
|
|
|
47 #include "clang/Lex/Pragma.h"
|
|
|
48 #include "clang/Lex/PreprocessingRecord.h"
|
|
|
49 #include "clang/Lex/PreprocessorLexer.h"
|
|
|
50 #include "clang/Lex/PreprocessorOptions.h"
|
|
|
51 #include "clang/Lex/ScratchBuffer.h"
|
|
|
52 #include "clang/Lex/Token.h"
|
|
|
53 #include "clang/Lex/TokenLexer.h"
|
|
|
54 #include "llvm/ADT/APInt.h"
|
|
|
55 #include "llvm/ADT/ArrayRef.h"
|
|
|
56 #include "llvm/ADT/DenseMap.h"
|
|
|
57 #include "llvm/ADT/STLExtras.h"
|
|
|
58 #include "llvm/ADT/SmallString.h"
|
|
|
59 #include "llvm/ADT/SmallVector.h"
|
|
|
60 #include "llvm/ADT/StringRef.h"
|
|
|
61 #include "llvm/Support/Capacity.h"
|
|
|
62 #include "llvm/Support/ErrorHandling.h"
|
|
|
63 #include "llvm/Support/MemoryBuffer.h"
|
|
|
64 #include "llvm/Support/raw_ostream.h"
|
|
|
65 #include <algorithm>
|
|
|
66 #include <cassert>
|
|
|
67 #include <memory>
|
|
252
|
68 #include <optional>
|
|
150
|
69 #include <string>
|
|
|
70 #include <utility>
|
|
|
71 #include <vector>
|
|
|
72
|
|
|
73 using namespace clang;
|
|
|
74
|
|
|
75 LLVM_INSTANTIATE_REGISTRY(PragmaHandlerRegistry)
|
|
|
76
|
|
|
77 ExternalPreprocessorSource::~ExternalPreprocessorSource() = default;
|
|
|
78
|
|
|
79 Preprocessor::Preprocessor(std::shared_ptr<PreprocessorOptions> PPOpts,
|
|
|
80 DiagnosticsEngine &diags, LangOptions &opts,
|
|
|
81 SourceManager &SM, HeaderSearch &Headers,
|
|
|
82 ModuleLoader &TheModuleLoader,
|
|
|
83 IdentifierInfoLookup *IILookup, bool OwnsHeaders,
|
|
|
84 TranslationUnitKind TUKind)
|
|
|
85 : PPOpts(std::move(PPOpts)), Diags(&diags), LangOpts(opts),
|
|
|
86 FileMgr(Headers.getFileMgr()), SourceMgr(SM),
|
|
|
87 ScratchBuf(new ScratchBuffer(SourceMgr)), HeaderInfo(Headers),
|
|
|
88 TheModuleLoader(TheModuleLoader), ExternalSource(nullptr),
|
|
|
89 // As the language options may have not been loaded yet (when
|
|
|
90 // deserializing an ASTUnit), adding keywords to the identifier table is
|
|
|
91 // deferred to Preprocessor::Initialize().
|
|
|
92 Identifiers(IILookup), PragmaHandlers(new PragmaNamespace(StringRef())),
|
|
|
93 TUKind(TUKind), SkipMainFilePreamble(0, true),
|
|
|
94 CurSubmoduleState(&NullSubmoduleState) {
|
|
|
95 OwnsHeaderSearch = OwnsHeaders;
|
|
|
96
|
|
|
97 // Default to discarding comments.
|
|
|
98 KeepComments = false;
|
|
|
99 KeepMacroComments = false;
|
|
|
100 SuppressIncludeNotFoundError = false;
|
|
|
101
|
|
|
102 // Macro expansion is enabled.
|
|
|
103 DisableMacroExpansion = false;
|
|
|
104 MacroExpansionInDirectivesOverride = false;
|
|
|
105 InMacroArgs = false;
|
|
|
106 ArgMacro = nullptr;
|
|
|
107 InMacroArgPreExpansion = false;
|
|
|
108 NumCachedTokenLexers = 0;
|
|
|
109 PragmasEnabled = true;
|
|
|
110 ParsingIfOrElifDirective = false;
|
|
|
111 PreprocessedOutput = false;
|
|
|
112
|
|
|
113 // We haven't read anything from the external source.
|
|
|
114 ReadMacrosFromExternalSource = false;
|
|
|
115
|
|
|
116 BuiltinInfo = std::make_unique<Builtin::Context>();
|
|
|
117
|
|
|
118 // "Poison" __VA_ARGS__, __VA_OPT__ which can only appear in the expansion of
|
|
|
119 // a macro. They get unpoisoned where it is allowed.
|
|
|
120 (Ident__VA_ARGS__ = getIdentifierInfo("__VA_ARGS__"))->setIsPoisoned();
|
|
|
121 SetPoisonReason(Ident__VA_ARGS__,diag::ext_pp_bad_vaargs_use);
|
|
207
|
122 (Ident__VA_OPT__ = getIdentifierInfo("__VA_OPT__"))->setIsPoisoned();
|
|
|
123 SetPoisonReason(Ident__VA_OPT__,diag::ext_pp_bad_vaopt_use);
|
|
150
|
124
|
|
|
125 // Initialize the pragma handlers.
|
|
|
126 RegisterBuiltinPragmas();
|
|
|
127
|
|
|
128 // Initialize builtin macros like __LINE__ and friends.
|
|
|
129 RegisterBuiltinMacros();
|
|
|
130
|
|
|
131 if(LangOpts.Borland) {
|
|
|
132 Ident__exception_info = getIdentifierInfo("_exception_info");
|
|
|
133 Ident___exception_info = getIdentifierInfo("__exception_info");
|
|
|
134 Ident_GetExceptionInfo = getIdentifierInfo("GetExceptionInformation");
|
|
|
135 Ident__exception_code = getIdentifierInfo("_exception_code");
|
|
|
136 Ident___exception_code = getIdentifierInfo("__exception_code");
|
|
|
137 Ident_GetExceptionCode = getIdentifierInfo("GetExceptionCode");
|
|
|
138 Ident__abnormal_termination = getIdentifierInfo("_abnormal_termination");
|
|
|
139 Ident___abnormal_termination = getIdentifierInfo("__abnormal_termination");
|
|
|
140 Ident_AbnormalTermination = getIdentifierInfo("AbnormalTermination");
|
|
|
141 } else {
|
|
|
142 Ident__exception_info = Ident__exception_code = nullptr;
|
|
|
143 Ident__abnormal_termination = Ident___exception_info = nullptr;
|
|
|
144 Ident___exception_code = Ident___abnormal_termination = nullptr;
|
|
|
145 Ident_GetExceptionInfo = Ident_GetExceptionCode = nullptr;
|
|
|
146 Ident_AbnormalTermination = nullptr;
|
|
|
147 }
|
|
|
148
|
|
|
149 // If using a PCH where a #pragma hdrstop is expected, start skipping tokens.
|
|
|
150 if (usingPCHWithPragmaHdrStop())
|
|
|
151 SkippingUntilPragmaHdrStop = true;
|
|
|
152
|
|
|
153 // If using a PCH with a through header, start skipping tokens.
|
|
|
154 if (!this->PPOpts->PCHThroughHeader.empty() &&
|
|
|
155 !this->PPOpts->ImplicitPCHInclude.empty())
|
|
|
156 SkippingUntilPCHThroughHeader = true;
|
|
|
157
|
|
240
|
158 #ifndef noCbC
|
|
|
159 SavedDepth = 0;
|
|
|
160 SavedTokenFlag = false;
|
|
|
161 #endif
|
|
|
162
|
|
150
|
163 if (this->PPOpts->GeneratePreamble)
|
|
|
164 PreambleConditionalStack.startRecording();
|
|
|
165
|
|
|
166 MaxTokens = LangOpts.MaxTokens;
|
|
|
167 }
|
|
|
168
|
|
|
169 Preprocessor::~Preprocessor() {
|
|
|
170 assert(BacktrackPositions.empty() && "EnableBacktrack/Backtrack imbalance!");
|
|
|
171
|
|
|
172 IncludeMacroStack.clear();
|
|
|
173
|
|
|
174 // Free any cached macro expanders.
|
|
|
175 // This populates MacroArgCache, so all TokenLexers need to be destroyed
|
|
|
176 // before the code below that frees up the MacroArgCache list.
|
|
|
177 std::fill(TokenLexerCache, TokenLexerCache + NumCachedTokenLexers, nullptr);
|
|
|
178 CurTokenLexer.reset();
|
|
|
179
|
|
|
180 // Free any cached MacroArgs.
|
|
|
181 for (MacroArgs *ArgList = MacroArgCache; ArgList;)
|
|
|
182 ArgList = ArgList->deallocate();
|
|
|
183
|
|
|
184 // Delete the header search info, if we own it.
|
|
|
185 if (OwnsHeaderSearch)
|
|
|
186 delete &HeaderInfo;
|
|
|
187 }
|
|
|
188
|
|
|
189 void Preprocessor::Initialize(const TargetInfo &Target,
|
|
|
190 const TargetInfo *AuxTarget) {
|
|
|
191 assert((!this->Target || this->Target == &Target) &&
|
|
|
192 "Invalid override of target information");
|
|
|
193 this->Target = &Target;
|
|
|
194
|
|
|
195 assert((!this->AuxTarget || this->AuxTarget == AuxTarget) &&
|
|
|
196 "Invalid override of aux target information.");
|
|
|
197 this->AuxTarget = AuxTarget;
|
|
|
198
|
|
|
199 // Initialize information about built-ins.
|
|
|
200 BuiltinInfo->InitializeTarget(Target, AuxTarget);
|
|
|
201 HeaderInfo.setTarget(Target);
|
|
|
202
|
|
|
203 // Populate the identifier table with info about keywords for the current language.
|
|
|
204 Identifiers.AddKeywords(LangOpts);
|
|
236
|
205
|
|
|
206 // Initialize the __FTL_EVAL_METHOD__ macro to the TargetInfo.
|
|
|
207 setTUFPEvalMethod(getTargetInfo().getFPEvalMethod());
|
|
|
208
|
|
|
209 if (getLangOpts().getFPEvalMethod() == LangOptions::FEM_UnsetOnCommandLine)
|
|
|
210 // Use setting from TargetInfo.
|
|
|
211 setCurrentFPEvalMethod(SourceLocation(), Target.getFPEvalMethod());
|
|
|
212 else
|
|
|
213 // Set initial value of __FLT_EVAL_METHOD__ from the command line.
|
|
|
214 setCurrentFPEvalMethod(SourceLocation(), getLangOpts().getFPEvalMethod());
|
|
150
|
215 }
|
|
|
216
|
|
|
217 void Preprocessor::InitializeForModelFile() {
|
|
|
218 NumEnteredSourceFiles = 0;
|
|
|
219
|
|
|
220 // Reset pragmas
|
|
|
221 PragmaHandlersBackup = std::move(PragmaHandlers);
|
|
|
222 PragmaHandlers = std::make_unique<PragmaNamespace>(StringRef());
|
|
|
223 RegisterBuiltinPragmas();
|
|
|
224
|
|
|
225 // Reset PredefinesFileID
|
|
|
226 PredefinesFileID = FileID();
|
|
|
227 }
|
|
|
228
|
|
|
229 void Preprocessor::FinalizeForModelFile() {
|
|
|
230 NumEnteredSourceFiles = 1;
|
|
|
231
|
|
|
232 PragmaHandlers = std::move(PragmaHandlersBackup);
|
|
|
233 }
|
|
|
234
|
|
|
235 void Preprocessor::DumpToken(const Token &Tok, bool DumpFlags) const {
|
|
236
|
236 llvm::errs() << tok::getTokenName(Tok.getKind());
|
|
|
237
|
|
|
238 if (!Tok.isAnnotation())
|
|
|
239 llvm::errs() << " '" << getSpelling(Tok) << "'";
|
|
150
|
240
|
|
|
241 if (!DumpFlags) return;
|
|
|
242
|
|
|
243 llvm::errs() << "\t";
|
|
|
244 if (Tok.isAtStartOfLine())
|
|
|
245 llvm::errs() << " [StartOfLine]";
|
|
|
246 if (Tok.hasLeadingSpace())
|
|
|
247 llvm::errs() << " [LeadingSpace]";
|
|
|
248 if (Tok.isExpandDisabled())
|
|
|
249 llvm::errs() << " [ExpandDisabled]";
|
|
|
250 if (Tok.needsCleaning()) {
|
|
|
251 const char *Start = SourceMgr.getCharacterData(Tok.getLocation());
|
|
|
252 llvm::errs() << " [UnClean='" << StringRef(Start, Tok.getLength())
|
|
|
253 << "']";
|
|
|
254 }
|
|
|
255
|
|
|
256 llvm::errs() << "\tLoc=<";
|
|
|
257 DumpLocation(Tok.getLocation());
|
|
|
258 llvm::errs() << ">";
|
|
|
259 }
|
|
|
260
|
|
|
261 void Preprocessor::DumpLocation(SourceLocation Loc) const {
|
|
|
262 Loc.print(llvm::errs(), SourceMgr);
|
|
|
263 }
|
|
|
264
|
|
|
265 void Preprocessor::DumpMacro(const MacroInfo &MI) const {
|
|
|
266 llvm::errs() << "MACRO: ";
|
|
|
267 for (unsigned i = 0, e = MI.getNumTokens(); i != e; ++i) {
|
|
|
268 DumpToken(MI.getReplacementToken(i));
|
|
|
269 llvm::errs() << " ";
|
|
|
270 }
|
|
|
271 llvm::errs() << "\n";
|
|
|
272 }
|
|
|
273
|
|
|
274 void Preprocessor::PrintStats() {
|
|
|
275 llvm::errs() << "\n*** Preprocessor Stats:\n";
|
|
|
276 llvm::errs() << NumDirectives << " directives found:\n";
|
|
|
277 llvm::errs() << " " << NumDefined << " #define.\n";
|
|
|
278 llvm::errs() << " " << NumUndefined << " #undef.\n";
|
|
|
279 llvm::errs() << " #include/#include_next/#import:\n";
|
|
|
280 llvm::errs() << " " << NumEnteredSourceFiles << " source files entered.\n";
|
|
|
281 llvm::errs() << " " << MaxIncludeStackDepth << " max include stack depth\n";
|
|
|
282 llvm::errs() << " " << NumIf << " #if/#ifndef/#ifdef.\n";
|
|
207
|
283 llvm::errs() << " " << NumElse << " #else/#elif/#elifdef/#elifndef.\n";
|
|
150
|
284 llvm::errs() << " " << NumEndif << " #endif.\n";
|
|
|
285 llvm::errs() << " " << NumPragma << " #pragma.\n";
|
|
|
286 llvm::errs() << NumSkipped << " #if/#ifndef#ifdef regions skipped\n";
|
|
|
287
|
|
|
288 llvm::errs() << NumMacroExpanded << "/" << NumFnMacroExpanded << "/"
|
|
|
289 << NumBuiltinMacroExpanded << " obj/fn/builtin macros expanded, "
|
|
|
290 << NumFastMacroExpanded << " on the fast path.\n";
|
|
|
291 llvm::errs() << (NumFastTokenPaste+NumTokenPaste)
|
|
|
292 << " token paste (##) operations performed, "
|
|
|
293 << NumFastTokenPaste << " on the fast path.\n";
|
|
|
294
|
|
|
295 llvm::errs() << "\nPreprocessor Memory: " << getTotalMemory() << "B total";
|
|
|
296
|
|
|
297 llvm::errs() << "\n BumpPtr: " << BP.getTotalMemory();
|
|
|
298 llvm::errs() << "\n Macro Expanded Tokens: "
|
|
|
299 << llvm::capacity_in_bytes(MacroExpandedTokens);
|
|
|
300 llvm::errs() << "\n Predefines Buffer: " << Predefines.capacity();
|
|
|
301 // FIXME: List information for all submodules.
|
|
|
302 llvm::errs() << "\n Macros: "
|
|
|
303 << llvm::capacity_in_bytes(CurSubmoduleState->Macros);
|
|
|
304 llvm::errs() << "\n #pragma push_macro Info: "
|
|
|
305 << llvm::capacity_in_bytes(PragmaPushMacroInfo);
|
|
|
306 llvm::errs() << "\n Poison Reasons: "
|
|
|
307 << llvm::capacity_in_bytes(PoisonReasons);
|
|
|
308 llvm::errs() << "\n Comment Handlers: "
|
|
|
309 << llvm::capacity_in_bytes(CommentHandlers) << "\n";
|
|
|
310 }
|
|
|
311
|
|
|
312 Preprocessor::macro_iterator
|
|
|
313 Preprocessor::macro_begin(bool IncludeExternalMacros) const {
|
|
|
314 if (IncludeExternalMacros && ExternalSource &&
|
|
|
315 !ReadMacrosFromExternalSource) {
|
|
|
316 ReadMacrosFromExternalSource = true;
|
|
|
317 ExternalSource->ReadDefinedMacros();
|
|
|
318 }
|
|
|
319
|
|
|
320 // Make sure we cover all macros in visible modules.
|
|
|
321 for (const ModuleMacro &Macro : ModuleMacros)
|
|
|
322 CurSubmoduleState->Macros.insert(std::make_pair(Macro.II, MacroState()));
|
|
|
323
|
|
|
324 return CurSubmoduleState->Macros.begin();
|
|
|
325 }
|
|
|
326
|
|
|
327 size_t Preprocessor::getTotalMemory() const {
|
|
|
328 return BP.getTotalMemory()
|
|
|
329 + llvm::capacity_in_bytes(MacroExpandedTokens)
|
|
|
330 + Predefines.capacity() /* Predefines buffer. */
|
|
|
331 // FIXME: Include sizes from all submodules, and include MacroInfo sizes,
|
|
|
332 // and ModuleMacros.
|
|
|
333 + llvm::capacity_in_bytes(CurSubmoduleState->Macros)
|
|
|
334 + llvm::capacity_in_bytes(PragmaPushMacroInfo)
|
|
|
335 + llvm::capacity_in_bytes(PoisonReasons)
|
|
|
336 + llvm::capacity_in_bytes(CommentHandlers);
|
|
|
337 }
|
|
|
338
|
|
|
339 Preprocessor::macro_iterator
|
|
|
340 Preprocessor::macro_end(bool IncludeExternalMacros) const {
|
|
|
341 if (IncludeExternalMacros && ExternalSource &&
|
|
|
342 !ReadMacrosFromExternalSource) {
|
|
|
343 ReadMacrosFromExternalSource = true;
|
|
|
344 ExternalSource->ReadDefinedMacros();
|
|
|
345 }
|
|
|
346
|
|
|
347 return CurSubmoduleState->Macros.end();
|
|
|
348 }
|
|
|
349
|
|
|
350 /// Compares macro tokens with a specified token value sequence.
|
|
|
351 static bool MacroDefinitionEquals(const MacroInfo *MI,
|
|
|
352 ArrayRef<TokenValue> Tokens) {
|
|
|
353 return Tokens.size() == MI->getNumTokens() &&
|
|
|
354 std::equal(Tokens.begin(), Tokens.end(), MI->tokens_begin());
|
|
|
355 }
|
|
|
356
|
|
|
357 StringRef Preprocessor::getLastMacroWithSpelling(
|
|
|
358 SourceLocation Loc,
|
|
|
359 ArrayRef<TokenValue> Tokens) const {
|
|
|
360 SourceLocation BestLocation;
|
|
|
361 StringRef BestSpelling;
|
|
|
362 for (Preprocessor::macro_iterator I = macro_begin(), E = macro_end();
|
|
|
363 I != E; ++I) {
|
|
|
364 const MacroDirective::DefInfo
|
|
|
365 Def = I->second.findDirectiveAtLoc(Loc, SourceMgr);
|
|
|
366 if (!Def || !Def.getMacroInfo())
|
|
|
367 continue;
|
|
|
368 if (!Def.getMacroInfo()->isObjectLike())
|
|
|
369 continue;
|
|
|
370 if (!MacroDefinitionEquals(Def.getMacroInfo(), Tokens))
|
|
|
371 continue;
|
|
|
372 SourceLocation Location = Def.getLocation();
|
|
|
373 // Choose the macro defined latest.
|
|
|
374 if (BestLocation.isInvalid() ||
|
|
|
375 (Location.isValid() &&
|
|
|
376 SourceMgr.isBeforeInTranslationUnit(BestLocation, Location))) {
|
|
|
377 BestLocation = Location;
|
|
|
378 BestSpelling = I->first->getName();
|
|
|
379 }
|
|
|
380 }
|
|
|
381 return BestSpelling;
|
|
|
382 }
|
|
|
383
|
|
|
384 void Preprocessor::recomputeCurLexerKind() {
|
|
|
385 if (CurLexer)
|
|
236
|
386 CurLexerKind = CurLexer->isDependencyDirectivesLexer()
|
|
|
387 ? CLK_DependencyDirectivesLexer
|
|
|
388 : CLK_Lexer;
|
|
150
|
389 else if (CurTokenLexer)
|
|
|
390 CurLexerKind = CLK_TokenLexer;
|
|
|
391 else
|
|
|
392 CurLexerKind = CLK_CachingLexer;
|
|
|
393 }
|
|
|
394
|
|
|
395 bool Preprocessor::SetCodeCompletionPoint(const FileEntry *File,
|
|
|
396 unsigned CompleteLine,
|
|
|
397 unsigned CompleteColumn) {
|
|
|
398 assert(File);
|
|
|
399 assert(CompleteLine && CompleteColumn && "Starts from 1:1");
|
|
|
400 assert(!CodeCompletionFile && "Already set");
|
|
|
401
|
|
|
402 // Load the actual file's contents.
|
|
252
|
403 std::optional<llvm::MemoryBufferRef> Buffer =
|
|
207
|
404 SourceMgr.getMemoryBufferForFileOrNone(File);
|
|
|
405 if (!Buffer)
|
|
150
|
406 return true;
|
|
|
407
|
|
|
408 // Find the byte position of the truncation point.
|
|
|
409 const char *Position = Buffer->getBufferStart();
|
|
|
410 for (unsigned Line = 1; Line < CompleteLine; ++Line) {
|
|
|
411 for (; *Position; ++Position) {
|
|
|
412 if (*Position != '\r' && *Position != '\n')
|
|
|
413 continue;
|
|
|
414
|
|
|
415 // Eat \r\n or \n\r as a single line.
|
|
|
416 if ((Position[1] == '\r' || Position[1] == '\n') &&
|
|
|
417 Position[0] != Position[1])
|
|
|
418 ++Position;
|
|
|
419 ++Position;
|
|
|
420 break;
|
|
|
421 }
|
|
|
422 }
|
|
|
423
|
|
|
424 Position += CompleteColumn - 1;
|
|
|
425
|
|
|
426 // If pointing inside the preamble, adjust the position at the beginning of
|
|
|
427 // the file after the preamble.
|
|
|
428 if (SkipMainFilePreamble.first &&
|
|
|
429 SourceMgr.getFileEntryForID(SourceMgr.getMainFileID()) == File) {
|
|
|
430 if (Position - Buffer->getBufferStart() < SkipMainFilePreamble.first)
|
|
|
431 Position = Buffer->getBufferStart() + SkipMainFilePreamble.first;
|
|
|
432 }
|
|
|
433
|
|
|
434 if (Position > Buffer->getBufferEnd())
|
|
|
435 Position = Buffer->getBufferEnd();
|
|
|
436
|
|
|
437 CodeCompletionFile = File;
|
|
|
438 CodeCompletionOffset = Position - Buffer->getBufferStart();
|
|
|
439
|
|
|
440 auto NewBuffer = llvm::WritableMemoryBuffer::getNewUninitMemBuffer(
|
|
|
441 Buffer->getBufferSize() + 1, Buffer->getBufferIdentifier());
|
|
|
442 char *NewBuf = NewBuffer->getBufferStart();
|
|
|
443 char *NewPos = std::copy(Buffer->getBufferStart(), Position, NewBuf);
|
|
|
444 *NewPos = '\0';
|
|
|
445 std::copy(Position, Buffer->getBufferEnd(), NewPos+1);
|
|
|
446 SourceMgr.overrideFileContents(File, std::move(NewBuffer));
|
|
|
447
|
|
|
448 return false;
|
|
|
449 }
|
|
|
450
|
|
|
451 void Preprocessor::CodeCompleteIncludedFile(llvm::StringRef Dir,
|
|
|
452 bool IsAngled) {
|
|
207
|
453 setCodeCompletionReached();
|
|
150
|
454 if (CodeComplete)
|
|
|
455 CodeComplete->CodeCompleteIncludedFile(Dir, IsAngled);
|
|
|
456 }
|
|
|
457
|
|
|
458 void Preprocessor::CodeCompleteNaturalLanguage() {
|
|
207
|
459 setCodeCompletionReached();
|
|
150
|
460 if (CodeComplete)
|
|
|
461 CodeComplete->CodeCompleteNaturalLanguage();
|
|
|
462 }
|
|
|
463
|
|
|
464 /// getSpelling - This method is used to get the spelling of a token into a
|
|
|
465 /// SmallVector. Note that the returned StringRef may not point to the
|
|
|
466 /// supplied buffer if a copy can be avoided.
|
|
|
467 StringRef Preprocessor::getSpelling(const Token &Tok,
|
|
|
468 SmallVectorImpl<char> &Buffer,
|
|
|
469 bool *Invalid) const {
|
|
|
470 // NOTE: this has to be checked *before* testing for an IdentifierInfo.
|
|
|
471 if (Tok.isNot(tok::raw_identifier) && !Tok.hasUCN()) {
|
|
|
472 // Try the fast path.
|
|
|
473 if (const IdentifierInfo *II = Tok.getIdentifierInfo())
|
|
|
474 return II->getName();
|
|
|
475 }
|
|
|
476
|
|
|
477 // Resize the buffer if we need to copy into it.
|
|
|
478 if (Tok.needsCleaning())
|
|
|
479 Buffer.resize(Tok.getLength());
|
|
|
480
|
|
|
481 const char *Ptr = Buffer.data();
|
|
|
482 unsigned Len = getSpelling(Tok, Ptr, Invalid);
|
|
|
483 return StringRef(Ptr, Len);
|
|
|
484 }
|
|
|
485
|
|
|
486 /// CreateString - Plop the specified string into a scratch buffer and return a
|
|
|
487 /// location for it. If specified, the source location provides a source
|
|
|
488 /// location for the token.
|
|
|
489 void Preprocessor::CreateString(StringRef Str, Token &Tok,
|
|
|
490 SourceLocation ExpansionLocStart,
|
|
|
491 SourceLocation ExpansionLocEnd) {
|
|
|
492 Tok.setLength(Str.size());
|
|
|
493
|
|
|
494 const char *DestPtr;
|
|
|
495 SourceLocation Loc = ScratchBuf->getToken(Str.data(), Str.size(), DestPtr);
|
|
|
496
|
|
|
497 if (ExpansionLocStart.isValid())
|
|
|
498 Loc = SourceMgr.createExpansionLoc(Loc, ExpansionLocStart,
|
|
|
499 ExpansionLocEnd, Str.size());
|
|
|
500 Tok.setLocation(Loc);
|
|
|
501
|
|
|
502 // If this is a raw identifier or a literal token, set the pointer data.
|
|
|
503 if (Tok.is(tok::raw_identifier))
|
|
|
504 Tok.setRawIdentifierData(DestPtr);
|
|
|
505 else if (Tok.isLiteral())
|
|
|
506 Tok.setLiteralData(DestPtr);
|
|
|
507 }
|
|
|
508
|
|
|
509 SourceLocation Preprocessor::SplitToken(SourceLocation Loc, unsigned Length) {
|
|
|
510 auto &SM = getSourceManager();
|
|
|
511 SourceLocation SpellingLoc = SM.getSpellingLoc(Loc);
|
|
|
512 std::pair<FileID, unsigned> LocInfo = SM.getDecomposedLoc(SpellingLoc);
|
|
|
513 bool Invalid = false;
|
|
|
514 StringRef Buffer = SM.getBufferData(LocInfo.first, &Invalid);
|
|
|
515 if (Invalid)
|
|
|
516 return SourceLocation();
|
|
|
517
|
|
|
518 // FIXME: We could consider re-using spelling for tokens we see repeatedly.
|
|
|
519 const char *DestPtr;
|
|
|
520 SourceLocation Spelling =
|
|
|
521 ScratchBuf->getToken(Buffer.data() + LocInfo.second, Length, DestPtr);
|
|
|
522 return SM.createTokenSplitLoc(Spelling, Loc, Loc.getLocWithOffset(Length));
|
|
|
523 }
|
|
|
524
|
|
|
525 Module *Preprocessor::getCurrentModule() {
|
|
|
526 if (!getLangOpts().isCompilingModule())
|
|
|
527 return nullptr;
|
|
|
528
|
|
|
529 return getHeaderSearchInfo().lookupModule(getLangOpts().CurrentModule);
|
|
|
530 }
|
|
|
531
|
|
236
|
532 Module *Preprocessor::getCurrentModuleImplementation() {
|
|
|
533 if (!getLangOpts().isCompilingModuleImplementation())
|
|
|
534 return nullptr;
|
|
|
535
|
|
|
536 return getHeaderSearchInfo().lookupModule(getLangOpts().ModuleName);
|
|
|
537 }
|
|
|
538
|
|
150
|
539 //===----------------------------------------------------------------------===//
|
|
|
540 // Preprocessor Initialization Methods
|
|
|
541 //===----------------------------------------------------------------------===//
|
|
|
542
|
|
|
543 /// EnterMainSourceFile - Enter the specified FileID as the main source file,
|
|
|
544 /// which implicitly adds the builtin defines etc.
|
|
|
545 void Preprocessor::EnterMainSourceFile() {
|
|
|
546 // We do not allow the preprocessor to reenter the main file. Doing so will
|
|
|
547 // cause FileID's to accumulate information from both runs (e.g. #line
|
|
|
548 // information) and predefined macros aren't guaranteed to be set properly.
|
|
|
549 assert(NumEnteredSourceFiles == 0 && "Cannot reenter the main file!");
|
|
|
550 FileID MainFileID = SourceMgr.getMainFileID();
|
|
|
551
|
|
|
552 // If MainFileID is loaded it means we loaded an AST file, no need to enter
|
|
|
553 // a main file.
|
|
|
554 if (!SourceMgr.isLoadedFileID(MainFileID)) {
|
|
|
555 // Enter the main file source buffer.
|
|
|
556 EnterSourceFile(MainFileID, nullptr, SourceLocation());
|
|
|
557
|
|
|
558 // If we've been asked to skip bytes in the main file (e.g., as part of a
|
|
|
559 // precompiled preamble), do so now.
|
|
|
560 if (SkipMainFilePreamble.first > 0)
|
|
|
561 CurLexer->SetByteOffset(SkipMainFilePreamble.first,
|
|
|
562 SkipMainFilePreamble.second);
|
|
|
563
|
|
|
564 // Tell the header info that the main file was entered. If the file is later
|
|
|
565 // #imported, it won't be re-entered.
|
|
|
566 if (const FileEntry *FE = SourceMgr.getFileEntryForID(MainFileID))
|
|
236
|
567 markIncluded(FE);
|
|
150
|
568 }
|
|
|
569
|
|
|
570 // Preprocess Predefines to populate the initial preprocessor state.
|
|
|
571 std::unique_ptr<llvm::MemoryBuffer> SB =
|
|
|
572 llvm::MemoryBuffer::getMemBufferCopy(Predefines, "<built-in>");
|
|
|
573 assert(SB && "Cannot create predefined source buffer");
|
|
|
574 FileID FID = SourceMgr.createFileID(std::move(SB));
|
|
|
575 assert(FID.isValid() && "Could not create FileID for predefines?");
|
|
|
576 setPredefinesFileID(FID);
|
|
|
577
|
|
|
578 // Start parsing the predefines.
|
|
|
579 EnterSourceFile(FID, nullptr, SourceLocation());
|
|
|
580
|
|
|
581 if (!PPOpts->PCHThroughHeader.empty()) {
|
|
|
582 // Lookup and save the FileID for the through header. If it isn't found
|
|
|
583 // in the search path, it's a fatal error.
|
|
252
|
584 OptionalFileEntryRef File = LookupFile(
|
|
150
|
585 SourceLocation(), PPOpts->PCHThroughHeader,
|
|
236
|
586 /*isAngled=*/false, /*FromDir=*/nullptr, /*FromFile=*/nullptr,
|
|
|
587 /*CurDir=*/nullptr, /*SearchPath=*/nullptr, /*RelativePath=*/nullptr,
|
|
150
|
588 /*SuggestedModule=*/nullptr, /*IsMapped=*/nullptr,
|
|
|
589 /*IsFrameworkFound=*/nullptr);
|
|
|
590 if (!File) {
|
|
|
591 Diag(SourceLocation(), diag::err_pp_through_header_not_found)
|
|
|
592 << PPOpts->PCHThroughHeader;
|
|
|
593 return;
|
|
|
594 }
|
|
|
595 setPCHThroughHeaderFileID(
|
|
|
596 SourceMgr.createFileID(*File, SourceLocation(), SrcMgr::C_User));
|
|
|
597 }
|
|
|
598
|
|
|
599 // Skip tokens from the Predefines and if needed the main file.
|
|
|
600 if ((usingPCHWithThroughHeader() && SkippingUntilPCHThroughHeader) ||
|
|
|
601 (usingPCHWithPragmaHdrStop() && SkippingUntilPragmaHdrStop))
|
|
|
602 SkipTokensWhileUsingPCH();
|
|
|
603 }
|
|
|
604
|
|
|
605 void Preprocessor::setPCHThroughHeaderFileID(FileID FID) {
|
|
|
606 assert(PCHThroughHeaderFileID.isInvalid() &&
|
|
|
607 "PCHThroughHeaderFileID already set!");
|
|
|
608 PCHThroughHeaderFileID = FID;
|
|
|
609 }
|
|
|
610
|
|
|
611 bool Preprocessor::isPCHThroughHeader(const FileEntry *FE) {
|
|
|
612 assert(PCHThroughHeaderFileID.isValid() &&
|
|
|
613 "Invalid PCH through header FileID");
|
|
|
614 return FE == SourceMgr.getFileEntryForID(PCHThroughHeaderFileID);
|
|
|
615 }
|
|
|
616
|
|
|
617 bool Preprocessor::creatingPCHWithThroughHeader() {
|
|
|
618 return TUKind == TU_Prefix && !PPOpts->PCHThroughHeader.empty() &&
|
|
|
619 PCHThroughHeaderFileID.isValid();
|
|
|
620 }
|
|
|
621
|
|
|
622 bool Preprocessor::usingPCHWithThroughHeader() {
|
|
|
623 return TUKind != TU_Prefix && !PPOpts->PCHThroughHeader.empty() &&
|
|
|
624 PCHThroughHeaderFileID.isValid();
|
|
|
625 }
|
|
|
626
|
|
|
627 bool Preprocessor::creatingPCHWithPragmaHdrStop() {
|
|
|
628 return TUKind == TU_Prefix && PPOpts->PCHWithHdrStop;
|
|
|
629 }
|
|
|
630
|
|
|
631 bool Preprocessor::usingPCHWithPragmaHdrStop() {
|
|
|
632 return TUKind != TU_Prefix && PPOpts->PCHWithHdrStop;
|
|
|
633 }
|
|
|
634
|
|
|
635 /// Skip tokens until after the #include of the through header or
|
|
|
636 /// until after a #pragma hdrstop is seen. Tokens in the predefines file
|
|
|
637 /// and the main file may be skipped. If the end of the predefines file
|
|
|
638 /// is reached, skipping continues into the main file. If the end of the
|
|
|
639 /// main file is reached, it's a fatal error.
|
|
|
640 void Preprocessor::SkipTokensWhileUsingPCH() {
|
|
|
641 bool ReachedMainFileEOF = false;
|
|
|
642 bool UsingPCHThroughHeader = SkippingUntilPCHThroughHeader;
|
|
|
643 bool UsingPragmaHdrStop = SkippingUntilPragmaHdrStop;
|
|
|
644 Token Tok;
|
|
|
645 while (true) {
|
|
|
646 bool InPredefines =
|
|
|
647 (CurLexer && CurLexer->getFileID() == getPredefinesFileID());
|
|
|
648 switch (CurLexerKind) {
|
|
|
649 case CLK_Lexer:
|
|
|
650 CurLexer->Lex(Tok);
|
|
|
651 break;
|
|
|
652 case CLK_TokenLexer:
|
|
|
653 CurTokenLexer->Lex(Tok);
|
|
|
654 break;
|
|
|
655 case CLK_CachingLexer:
|
|
|
656 CachingLex(Tok);
|
|
|
657 break;
|
|
236
|
658 case CLK_DependencyDirectivesLexer:
|
|
|
659 CurLexer->LexDependencyDirectiveToken(Tok);
|
|
|
660 break;
|
|
150
|
661 case CLK_LexAfterModuleImport:
|
|
|
662 LexAfterModuleImport(Tok);
|
|
|
663 break;
|
|
|
664 }
|
|
|
665 if (Tok.is(tok::eof) && !InPredefines) {
|
|
|
666 ReachedMainFileEOF = true;
|
|
|
667 break;
|
|
|
668 }
|
|
|
669 if (UsingPCHThroughHeader && !SkippingUntilPCHThroughHeader)
|
|
|
670 break;
|
|
|
671 if (UsingPragmaHdrStop && !SkippingUntilPragmaHdrStop)
|
|
|
672 break;
|
|
|
673 }
|
|
|
674 if (ReachedMainFileEOF) {
|
|
|
675 if (UsingPCHThroughHeader)
|
|
|
676 Diag(SourceLocation(), diag::err_pp_through_header_not_seen)
|
|
|
677 << PPOpts->PCHThroughHeader << 1;
|
|
|
678 else if (!PPOpts->PCHWithHdrStopCreate)
|
|
|
679 Diag(SourceLocation(), diag::err_pp_pragma_hdrstop_not_seen);
|
|
|
680 }
|
|
|
681 }
|
|
|
682
|
|
|
683 void Preprocessor::replayPreambleConditionalStack() {
|
|
|
684 // Restore the conditional stack from the preamble, if there is one.
|
|
|
685 if (PreambleConditionalStack.isReplaying()) {
|
|
|
686 assert(CurPPLexer &&
|
|
|
687 "CurPPLexer is null when calling replayPreambleConditionalStack.");
|
|
|
688 CurPPLexer->setConditionalLevels(PreambleConditionalStack.getStack());
|
|
|
689 PreambleConditionalStack.doneReplaying();
|
|
|
690 if (PreambleConditionalStack.reachedEOFWhileSkipping())
|
|
|
691 SkipExcludedConditionalBlock(
|
|
|
692 PreambleConditionalStack.SkipInfo->HashTokenLoc,
|
|
|
693 PreambleConditionalStack.SkipInfo->IfTokenLoc,
|
|
|
694 PreambleConditionalStack.SkipInfo->FoundNonSkipPortion,
|
|
|
695 PreambleConditionalStack.SkipInfo->FoundElse,
|
|
|
696 PreambleConditionalStack.SkipInfo->ElseLoc);
|
|
|
697 }
|
|
|
698 }
|
|
|
699
|
|
|
700 void Preprocessor::EndSourceFile() {
|
|
|
701 // Notify the client that we reached the end of the source file.
|
|
|
702 if (Callbacks)
|
|
|
703 Callbacks->EndOfMainFile();
|
|
|
704 }
|
|
|
705
|
|
|
706 //===----------------------------------------------------------------------===//
|
|
|
707 // Lexer Event Handling.
|
|
|
708 //===----------------------------------------------------------------------===//
|
|
|
709
|
|
|
710 /// LookUpIdentifierInfo - Given a tok::raw_identifier token, look up the
|
|
|
711 /// identifier information for the token and install it into the token,
|
|
|
712 /// updating the token kind accordingly.
|
|
|
713 IdentifierInfo *Preprocessor::LookUpIdentifierInfo(Token &Identifier) const {
|
|
|
714 assert(!Identifier.getRawIdentifier().empty() && "No raw identifier data!");
|
|
|
715
|
|
|
716 // Look up this token, see if it is a macro, or if it is a language keyword.
|
|
|
717 IdentifierInfo *II;
|
|
|
718 if (!Identifier.needsCleaning() && !Identifier.hasUCN()) {
|
|
|
719 // No cleaning needed, just use the characters from the lexed buffer.
|
|
|
720 II = getIdentifierInfo(Identifier.getRawIdentifier());
|
|
|
721 } else {
|
|
|
722 // Cleaning needed, alloca a buffer, clean into it, then use the buffer.
|
|
|
723 SmallString<64> IdentifierBuffer;
|
|
|
724 StringRef CleanedStr = getSpelling(Identifier, IdentifierBuffer);
|
|
|
725
|
|
|
726 if (Identifier.hasUCN()) {
|
|
|
727 SmallString<64> UCNIdentifierBuffer;
|
|
|
728 expandUCNs(UCNIdentifierBuffer, CleanedStr);
|
|
|
729 II = getIdentifierInfo(UCNIdentifierBuffer);
|
|
|
730 } else {
|
|
|
731 II = getIdentifierInfo(CleanedStr);
|
|
|
732 }
|
|
|
733 }
|
|
|
734
|
|
|
735 // Update the token info (identifier info and appropriate token kind).
|
|
236
|
736 // FIXME: the raw_identifier may contain leading whitespace which is removed
|
|
|
737 // from the cleaned identifier token. The SourceLocation should be updated to
|
|
|
738 // refer to the non-whitespace character. For instance, the text "\\\nB" (a
|
|
|
739 // line continuation before 'B') is parsed as a single tok::raw_identifier and
|
|
|
740 // is cleaned to tok::identifier "B". After cleaning the token's length is
|
|
|
741 // still 3 and the SourceLocation refers to the location of the backslash.
|
|
150
|
742 Identifier.setIdentifierInfo(II);
|
|
236
|
743 Identifier.setKind(II->getTokenID());
|
|
150
|
744
|
|
|
745 return II;
|
|
|
746 }
|
|
|
747
|
|
|
748 void Preprocessor::SetPoisonReason(IdentifierInfo *II, unsigned DiagID) {
|
|
|
749 PoisonReasons[II] = DiagID;
|
|
|
750 }
|
|
|
751
|
|
|
752 void Preprocessor::PoisonSEHIdentifiers(bool Poison) {
|
|
|
753 assert(Ident__exception_code && Ident__exception_info);
|
|
|
754 assert(Ident___exception_code && Ident___exception_info);
|
|
|
755 Ident__exception_code->setIsPoisoned(Poison);
|
|
|
756 Ident___exception_code->setIsPoisoned(Poison);
|
|
|
757 Ident_GetExceptionCode->setIsPoisoned(Poison);
|
|
|
758 Ident__exception_info->setIsPoisoned(Poison);
|
|
|
759 Ident___exception_info->setIsPoisoned(Poison);
|
|
|
760 Ident_GetExceptionInfo->setIsPoisoned(Poison);
|
|
|
761 Ident__abnormal_termination->setIsPoisoned(Poison);
|
|
|
762 Ident___abnormal_termination->setIsPoisoned(Poison);
|
|
|
763 Ident_AbnormalTermination->setIsPoisoned(Poison);
|
|
|
764 }
|
|
|
765
|
|
|
766 void Preprocessor::HandlePoisonedIdentifier(Token & Identifier) {
|
|
|
767 assert(Identifier.getIdentifierInfo() &&
|
|
|
768 "Can't handle identifiers without identifier info!");
|
|
|
769 llvm::DenseMap<IdentifierInfo*,unsigned>::const_iterator it =
|
|
|
770 PoisonReasons.find(Identifier.getIdentifierInfo());
|
|
|
771 if(it == PoisonReasons.end())
|
|
|
772 Diag(Identifier, diag::err_pp_used_poisoned_id);
|
|
|
773 else
|
|
|
774 Diag(Identifier,it->second) << Identifier.getIdentifierInfo();
|
|
|
775 }
|
|
|
776
|
|
|
777 void Preprocessor::updateOutOfDateIdentifier(IdentifierInfo &II) const {
|
|
|
778 assert(II.isOutOfDate() && "not out of date");
|
|
|
779 getExternalSource()->updateOutOfDateIdentifier(II);
|
|
|
780 }
|
|
|
781
|
|
|
782 /// HandleIdentifier - This callback is invoked when the lexer reads an
|
|
|
783 /// identifier. This callback looks up the identifier in the map and/or
|
|
|
784 /// potentially macro expands it or turns it into a named token (like 'for').
|
|
|
785 ///
|
|
|
786 /// Note that callers of this method are guarded by checking the
|
|
|
787 /// IdentifierInfo's 'isHandleIdentifierCase' bit. If this method changes, the
|
|
|
788 /// IdentifierInfo methods that compute these properties will need to change to
|
|
|
789 /// match.
|
|
|
790 bool Preprocessor::HandleIdentifier(Token &Identifier) {
|
|
|
791 assert(Identifier.getIdentifierInfo() &&
|
|
|
792 "Can't handle identifiers without identifier info!");
|
|
|
793
|
|
|
794 IdentifierInfo &II = *Identifier.getIdentifierInfo();
|
|
|
795
|
|
|
796 // If the information about this identifier is out of date, update it from
|
|
|
797 // the external source.
|
|
|
798 // We have to treat __VA_ARGS__ in a special way, since it gets
|
|
|
799 // serialized with isPoisoned = true, but our preprocessor may have
|
|
|
800 // unpoisoned it if we're defining a C99 macro.
|
|
|
801 if (II.isOutOfDate()) {
|
|
|
802 bool CurrentIsPoisoned = false;
|
|
|
803 const bool IsSpecialVariadicMacro =
|
|
|
804 &II == Ident__VA_ARGS__ || &II == Ident__VA_OPT__;
|
|
|
805 if (IsSpecialVariadicMacro)
|
|
|
806 CurrentIsPoisoned = II.isPoisoned();
|
|
|
807
|
|
|
808 updateOutOfDateIdentifier(II);
|
|
|
809 Identifier.setKind(II.getTokenID());
|
|
|
810
|
|
|
811 if (IsSpecialVariadicMacro)
|
|
|
812 II.setIsPoisoned(CurrentIsPoisoned);
|
|
|
813 }
|
|
|
814
|
|
|
815 // If this identifier was poisoned, and if it was not produced from a macro
|
|
|
816 // expansion, emit an error.
|
|
|
817 if (II.isPoisoned() && CurPPLexer) {
|
|
|
818 HandlePoisonedIdentifier(Identifier);
|
|
|
819 }
|
|
|
820
|
|
|
821 // If this is a macro to be expanded, do it.
|
|
|
822 if (MacroDefinition MD = getMacroDefinition(&II)) {
|
|
|
823 auto *MI = MD.getMacroInfo();
|
|
|
824 assert(MI && "macro definition with no macro info?");
|
|
|
825 if (!DisableMacroExpansion) {
|
|
|
826 if (!Identifier.isExpandDisabled() && MI->isEnabled()) {
|
|
|
827 // C99 6.10.3p10: If the preprocessing token immediately after the
|
|
|
828 // macro name isn't a '(', this macro should not be expanded.
|
|
|
829 if (!MI->isFunctionLike() || isNextPPTokenLParen())
|
|
|
830 return HandleMacroExpandedIdentifier(Identifier, MD);
|
|
|
831 } else {
|
|
|
832 // C99 6.10.3.4p2 says that a disabled macro may never again be
|
|
|
833 // expanded, even if it's in a context where it could be expanded in the
|
|
|
834 // future.
|
|
|
835 Identifier.setFlag(Token::DisableExpand);
|
|
|
836 if (MI->isObjectLike() || isNextPPTokenLParen())
|
|
|
837 Diag(Identifier, diag::pp_disabled_macro_expansion);
|
|
|
838 }
|
|
|
839 }
|
|
|
840 }
|
|
|
841
|
|
|
842 // If this identifier is a keyword in a newer Standard or proposed Standard,
|
|
|
843 // produce a warning. Don't warn if we're not considering macro expansion,
|
|
|
844 // since this identifier might be the name of a macro.
|
|
|
845 // FIXME: This warning is disabled in cases where it shouldn't be, like
|
|
|
846 // "#define constexpr constexpr", "int constexpr;"
|
|
|
847 if (II.isFutureCompatKeyword() && !DisableMacroExpansion) {
|
|
236
|
848 Diag(Identifier, getIdentifierTable().getFutureCompatDiagKind(II, getLangOpts()))
|
|
150
|
849 << II.getName();
|
|
|
850 // Don't diagnose this keyword again in this translation unit.
|
|
|
851 II.setIsFutureCompatKeyword(false);
|
|
|
852 }
|
|
|
853
|
|
|
854 // If this is an extension token, diagnose its use.
|
|
|
855 // We avoid diagnosing tokens that originate from macro definitions.
|
|
|
856 // FIXME: This warning is disabled in cases where it shouldn't be,
|
|
|
857 // like "#define TY typeof", "TY(1) x".
|
|
|
858 if (II.isExtensionToken() && !DisableMacroExpansion)
|
|
|
859 Diag(Identifier, diag::ext_token_used);
|
|
|
860
|
|
|
861 // If this is the 'import' contextual keyword following an '@', note
|
|
|
862 // that the next token indicates a module name.
|
|
|
863 //
|
|
|
864 // Note that we do not treat 'import' as a contextual
|
|
|
865 // keyword when we're in a caching lexer, because caching lexers only get
|
|
|
866 // used in contexts where import declarations are disallowed.
|
|
|
867 //
|
|
252
|
868 // Likewise if this is the standard C++ import keyword.
|
|
150
|
869 if (((LastTokenWasAt && II.isModulesImport()) ||
|
|
|
870 Identifier.is(tok::kw_import)) &&
|
|
|
871 !InMacroArgs && !DisableMacroExpansion &&
|
|
|
872 (getLangOpts().Modules || getLangOpts().DebuggerSupport) &&
|
|
|
873 CurLexerKind != CLK_CachingLexer) {
|
|
|
874 ModuleImportLoc = Identifier.getLocation();
|
|
236
|
875 NamedModuleImportPath.clear();
|
|
252
|
876 IsAtImport = true;
|
|
150
|
877 ModuleImportExpectsIdentifier = true;
|
|
|
878 CurLexerKind = CLK_LexAfterModuleImport;
|
|
|
879 }
|
|
|
880 return true;
|
|
|
881 }
|
|
|
882
|
|
|
883 void Preprocessor::Lex(Token &Result) {
|
|
|
884 ++LexLevel;
|
|
|
885
|
|
|
886 // We loop here until a lex function returns a token; this avoids recursion.
|
|
|
887 bool ReturnedToken;
|
|
|
888 do {
|
|
|
889 switch (CurLexerKind) {
|
|
|
890 case CLK_Lexer:
|
|
|
891 ReturnedToken = CurLexer->Lex(Result);
|
|
|
892 break;
|
|
|
893 case CLK_TokenLexer:
|
|
|
894 ReturnedToken = CurTokenLexer->Lex(Result);
|
|
|
895 break;
|
|
|
896 case CLK_CachingLexer:
|
|
|
897 CachingLex(Result);
|
|
|
898 ReturnedToken = true;
|
|
|
899 break;
|
|
236
|
900 case CLK_DependencyDirectivesLexer:
|
|
|
901 ReturnedToken = CurLexer->LexDependencyDirectiveToken(Result);
|
|
|
902 break;
|
|
150
|
903 case CLK_LexAfterModuleImport:
|
|
|
904 ReturnedToken = LexAfterModuleImport(Result);
|
|
|
905 break;
|
|
|
906 }
|
|
|
907 } while (!ReturnedToken);
|
|
|
908
|
|
|
909 if (Result.is(tok::unknown) && TheModuleLoader.HadFatalFailure)
|
|
|
910 return;
|
|
|
911
|
|
|
912 if (Result.is(tok::code_completion) && Result.getIdentifierInfo()) {
|
|
|
913 // Remember the identifier before code completion token.
|
|
|
914 setCodeCompletionIdentifierInfo(Result.getIdentifierInfo());
|
|
|
915 setCodeCompletionTokenRange(Result.getLocation(), Result.getEndLoc());
|
|
|
916 // Set IdenfitierInfo to null to avoid confusing code that handles both
|
|
|
917 // identifiers and completion tokens.
|
|
|
918 Result.setIdentifierInfo(nullptr);
|
|
|
919 }
|
|
|
920
|
|
236
|
921 // Update StdCXXImportSeqState to track our position within a C++20 import-seq
|
|
150
|
922 // if this token is being produced as a result of phase 4 of translation.
|
|
236
|
923 // Update TrackGMFState to decide if we are currently in a Global Module
|
|
|
924 // Fragment. GMF state updates should precede StdCXXImportSeq ones, since GMF state
|
|
|
925 // depends on the prevailing StdCXXImportSeq state in two cases.
|
|
150
|
926 if (getLangOpts().CPlusPlusModules && LexLevel == 1 &&
|
|
|
927 !Result.getFlag(Token::IsReinjected)) {
|
|
|
928 switch (Result.getKind()) {
|
|
|
929 case tok::l_paren: case tok::l_square: case tok::l_brace:
|
|
236
|
930 StdCXXImportSeqState.handleOpenBracket();
|
|
150
|
931 break;
|
|
|
932 case tok::r_paren: case tok::r_square:
|
|
236
|
933 StdCXXImportSeqState.handleCloseBracket();
|
|
150
|
934 break;
|
|
|
935 case tok::r_brace:
|
|
236
|
936 StdCXXImportSeqState.handleCloseBrace();
|
|
150
|
937 break;
|
|
236
|
938 // This token is injected to represent the translation of '#include "a.h"'
|
|
|
939 // into "import a.h;". Mimic the notional ';'.
|
|
|
940 case tok::annot_module_include:
|
|
150
|
941 case tok::semi:
|
|
236
|
942 TrackGMFState.handleSemi();
|
|
|
943 StdCXXImportSeqState.handleSemi();
|
|
252
|
944 ModuleDeclState.handleSemi();
|
|
150
|
945 break;
|
|
|
946 case tok::header_name:
|
|
|
947 case tok::annot_header_unit:
|
|
236
|
948 StdCXXImportSeqState.handleHeaderName();
|
|
150
|
949 break;
|
|
|
950 case tok::kw_export:
|
|
236
|
951 TrackGMFState.handleExport();
|
|
|
952 StdCXXImportSeqState.handleExport();
|
|
252
|
953 ModuleDeclState.handleExport();
|
|
|
954 break;
|
|
|
955 case tok::colon:
|
|
|
956 ModuleDeclState.handleColon();
|
|
|
957 break;
|
|
|
958 case tok::period:
|
|
|
959 ModuleDeclState.handlePeriod();
|
|
150
|
960 break;
|
|
|
961 case tok::identifier:
|
|
|
962 if (Result.getIdentifierInfo()->isModulesImport()) {
|
|
236
|
963 TrackGMFState.handleImport(StdCXXImportSeqState.afterTopLevelSeq());
|
|
|
964 StdCXXImportSeqState.handleImport();
|
|
|
965 if (StdCXXImportSeqState.afterImportSeq()) {
|
|
150
|
966 ModuleImportLoc = Result.getLocation();
|
|
236
|
967 NamedModuleImportPath.clear();
|
|
252
|
968 IsAtImport = false;
|
|
150
|
969 ModuleImportExpectsIdentifier = true;
|
|
|
970 CurLexerKind = CLK_LexAfterModuleImport;
|
|
|
971 }
|
|
|
972 break;
|
|
236
|
973 } else if (Result.getIdentifierInfo() == getIdentifierInfo("module")) {
|
|
|
974 TrackGMFState.handleModule(StdCXXImportSeqState.afterTopLevelSeq());
|
|
252
|
975 ModuleDeclState.handleModule();
|
|
236
|
976 break;
|
|
252
|
977 } else {
|
|
|
978 ModuleDeclState.handleIdentifier(Result.getIdentifierInfo());
|
|
|
979 if (ModuleDeclState.isModuleCandidate())
|
|
|
980 break;
|
|
150
|
981 }
|
|
236
|
982 [[fallthrough]];
|
|
150
|
983 default:
|
|
236
|
984 TrackGMFState.handleMisc();
|
|
|
985 StdCXXImportSeqState.handleMisc();
|
|
252
|
986 ModuleDeclState.handleMisc();
|
|
150
|
987 break;
|
|
|
988 }
|
|
|
989 }
|
|
|
990
|
|
|
991 LastTokenWasAt = Result.is(tok::at);
|
|
|
992 --LexLevel;
|
|
|
993
|
|
207
|
994 if ((LexLevel == 0 || PreprocessToken) &&
|
|
|
995 !Result.getFlag(Token::IsReinjected)) {
|
|
|
996 if (LexLevel == 0)
|
|
|
997 ++TokenCount;
|
|
150
|
998 if (OnToken)
|
|
|
999 OnToken(Result);
|
|
|
1000 }
|
|
|
1001 }
|
|
|
1002
|
|
|
1003 /// Lex a header-name token (including one formed from header-name-tokens if
|
|
|
1004 /// \p AllowConcatenation is \c true).
|
|
|
1005 ///
|
|
|
1006 /// \param FilenameTok Filled in with the next token. On success, this will
|
|
|
1007 /// be either a header_name token. On failure, it will be whatever other
|
|
|
1008 /// token was found instead.
|
|
|
1009 /// \param AllowMacroExpansion If \c true, allow the header name to be formed
|
|
|
1010 /// by macro expansion (concatenating tokens as necessary if the first
|
|
|
1011 /// token is a '<').
|
|
|
1012 /// \return \c true if we reached EOD or EOF while looking for a > token in
|
|
|
1013 /// a concatenated header name and diagnosed it. \c false otherwise.
|
|
|
1014 bool Preprocessor::LexHeaderName(Token &FilenameTok, bool AllowMacroExpansion) {
|
|
|
1015 // Lex using header-name tokenization rules if tokens are being lexed from
|
|
|
1016 // a file. Just grab a token normally if we're in a macro expansion.
|
|
|
1017 if (CurPPLexer)
|
|
|
1018 CurPPLexer->LexIncludeFilename(FilenameTok);
|
|
|
1019 else
|
|
|
1020 Lex(FilenameTok);
|
|
|
1021
|
|
|
1022 // This could be a <foo/bar.h> file coming from a macro expansion. In this
|
|
|
1023 // case, glue the tokens together into an angle_string_literal token.
|
|
|
1024 SmallString<128> FilenameBuffer;
|
|
|
1025 if (FilenameTok.is(tok::less) && AllowMacroExpansion) {
|
|
|
1026 bool StartOfLine = FilenameTok.isAtStartOfLine();
|
|
|
1027 bool LeadingSpace = FilenameTok.hasLeadingSpace();
|
|
|
1028 bool LeadingEmptyMacro = FilenameTok.hasLeadingEmptyMacro();
|
|
|
1029
|
|
|
1030 SourceLocation Start = FilenameTok.getLocation();
|
|
|
1031 SourceLocation End;
|
|
|
1032 FilenameBuffer.push_back('<');
|
|
|
1033
|
|
|
1034 // Consume tokens until we find a '>'.
|
|
|
1035 // FIXME: A header-name could be formed starting or ending with an
|
|
|
1036 // alternative token. It's not clear whether that's ill-formed in all
|
|
|
1037 // cases.
|
|
|
1038 while (FilenameTok.isNot(tok::greater)) {
|
|
|
1039 Lex(FilenameTok);
|
|
|
1040 if (FilenameTok.isOneOf(tok::eod, tok::eof)) {
|
|
|
1041 Diag(FilenameTok.getLocation(), diag::err_expected) << tok::greater;
|
|
|
1042 Diag(Start, diag::note_matching) << tok::less;
|
|
|
1043 return true;
|
|
|
1044 }
|
|
|
1045
|
|
|
1046 End = FilenameTok.getLocation();
|
|
|
1047
|
|
|
1048 // FIXME: Provide code completion for #includes.
|
|
|
1049 if (FilenameTok.is(tok::code_completion)) {
|
|
|
1050 setCodeCompletionReached();
|
|
|
1051 Lex(FilenameTok);
|
|
|
1052 continue;
|
|
|
1053 }
|
|
|
1054
|
|
|
1055 // Append the spelling of this token to the buffer. If there was a space
|
|
|
1056 // before it, add it now.
|
|
|
1057 if (FilenameTok.hasLeadingSpace())
|
|
|
1058 FilenameBuffer.push_back(' ');
|
|
|
1059
|
|
|
1060 // Get the spelling of the token, directly into FilenameBuffer if
|
|
|
1061 // possible.
|
|
|
1062 size_t PreAppendSize = FilenameBuffer.size();
|
|
|
1063 FilenameBuffer.resize(PreAppendSize + FilenameTok.getLength());
|
|
|
1064
|
|
|
1065 const char *BufPtr = &FilenameBuffer[PreAppendSize];
|
|
|
1066 unsigned ActualLen = getSpelling(FilenameTok, BufPtr);
|
|
|
1067
|
|
|
1068 // If the token was spelled somewhere else, copy it into FilenameBuffer.
|
|
|
1069 if (BufPtr != &FilenameBuffer[PreAppendSize])
|
|
|
1070 memcpy(&FilenameBuffer[PreAppendSize], BufPtr, ActualLen);
|
|
|
1071
|
|
|
1072 // Resize FilenameBuffer to the correct size.
|
|
|
1073 if (FilenameTok.getLength() != ActualLen)
|
|
|
1074 FilenameBuffer.resize(PreAppendSize + ActualLen);
|
|
|
1075 }
|
|
|
1076
|
|
|
1077 FilenameTok.startToken();
|
|
|
1078 FilenameTok.setKind(tok::header_name);
|
|
|
1079 FilenameTok.setFlagValue(Token::StartOfLine, StartOfLine);
|
|
|
1080 FilenameTok.setFlagValue(Token::LeadingSpace, LeadingSpace);
|
|
|
1081 FilenameTok.setFlagValue(Token::LeadingEmptyMacro, LeadingEmptyMacro);
|
|
|
1082 CreateString(FilenameBuffer, FilenameTok, Start, End);
|
|
|
1083 } else if (FilenameTok.is(tok::string_literal) && AllowMacroExpansion) {
|
|
|
1084 // Convert a string-literal token of the form " h-char-sequence "
|
|
|
1085 // (produced by macro expansion) into a header-name token.
|
|
|
1086 //
|
|
|
1087 // The rules for header-names don't quite match the rules for
|
|
|
1088 // string-literals, but all the places where they differ result in
|
|
|
1089 // undefined behavior, so we can and do treat them the same.
|
|
|
1090 //
|
|
|
1091 // A string-literal with a prefix or suffix is not translated into a
|
|
|
1092 // header-name. This could theoretically be observable via the C++20
|
|
|
1093 // context-sensitive header-name formation rules.
|
|
|
1094 StringRef Str = getSpelling(FilenameTok, FilenameBuffer);
|
|
|
1095 if (Str.size() >= 2 && Str.front() == '"' && Str.back() == '"')
|
|
|
1096 FilenameTok.setKind(tok::header_name);
|
|
|
1097 }
|
|
|
1098
|
|
|
1099 return false;
|
|
|
1100 }
|
|
|
1101
|
|
|
1102 /// Collect the tokens of a C++20 pp-import-suffix.
|
|
|
1103 void Preprocessor::CollectPpImportSuffix(SmallVectorImpl<Token> &Toks) {
|
|
|
1104 // FIXME: For error recovery, consider recognizing attribute syntax here
|
|
|
1105 // and terminating / diagnosing a missing semicolon if we find anything
|
|
|
1106 // else? (Can we leave that to the parser?)
|
|
|
1107 unsigned BracketDepth = 0;
|
|
|
1108 while (true) {
|
|
|
1109 Toks.emplace_back();
|
|
|
1110 Lex(Toks.back());
|
|
|
1111
|
|
|
1112 switch (Toks.back().getKind()) {
|
|
|
1113 case tok::l_paren: case tok::l_square: case tok::l_brace:
|
|
|
1114 ++BracketDepth;
|
|
|
1115 break;
|
|
|
1116
|
|
|
1117 case tok::r_paren: case tok::r_square: case tok::r_brace:
|
|
|
1118 if (BracketDepth == 0)
|
|
|
1119 return;
|
|
|
1120 --BracketDepth;
|
|
|
1121 break;
|
|
|
1122
|
|
|
1123 case tok::semi:
|
|
|
1124 if (BracketDepth == 0)
|
|
|
1125 return;
|
|
|
1126 break;
|
|
|
1127
|
|
|
1128 case tok::eof:
|
|
|
1129 return;
|
|
|
1130
|
|
|
1131 default:
|
|
|
1132 break;
|
|
|
1133 }
|
|
|
1134 }
|
|
|
1135 }
|
|
|
1136
|
|
|
1137
|
|
|
1138 /// Lex a token following the 'import' contextual keyword.
|
|
|
1139 ///
|
|
|
1140 /// pp-import: [C++20]
|
|
|
1141 /// import header-name pp-import-suffix[opt] ;
|
|
|
1142 /// import header-name-tokens pp-import-suffix[opt] ;
|
|
|
1143 /// [ObjC] @ import module-name ;
|
|
|
1144 /// [Clang] import module-name ;
|
|
|
1145 ///
|
|
|
1146 /// header-name-tokens:
|
|
|
1147 /// string-literal
|
|
|
1148 /// < [any sequence of preprocessing-tokens other than >] >
|
|
|
1149 ///
|
|
|
1150 /// module-name:
|
|
|
1151 /// module-name-qualifier[opt] identifier
|
|
|
1152 ///
|
|
|
1153 /// module-name-qualifier
|
|
|
1154 /// module-name-qualifier[opt] identifier .
|
|
|
1155 ///
|
|
|
1156 /// We respond to a pp-import by importing macros from the named module.
|
|
|
1157 bool Preprocessor::LexAfterModuleImport(Token &Result) {
|
|
|
1158 // Figure out what kind of lexer we actually have.
|
|
|
1159 recomputeCurLexerKind();
|
|
|
1160
|
|
|
1161 // Lex the next token. The header-name lexing rules are used at the start of
|
|
|
1162 // a pp-import.
|
|
|
1163 //
|
|
|
1164 // For now, we only support header-name imports in C++20 mode.
|
|
|
1165 // FIXME: Should we allow this in all language modes that support an import
|
|
|
1166 // declaration as an extension?
|
|
236
|
1167 if (NamedModuleImportPath.empty() && getLangOpts().CPlusPlusModules) {
|
|
150
|
1168 if (LexHeaderName(Result))
|
|
|
1169 return true;
|
|
252
|
1170
|
|
|
1171 if (Result.is(tok::colon) && ModuleDeclState.isNamedModule()) {
|
|
|
1172 std::string Name = ModuleDeclState.getPrimaryName().str();
|
|
|
1173 Name += ":";
|
|
|
1174 NamedModuleImportPath.push_back(
|
|
|
1175 {getIdentifierInfo(Name), Result.getLocation()});
|
|
|
1176 CurLexerKind = CLK_LexAfterModuleImport;
|
|
|
1177 return true;
|
|
|
1178 }
|
|
150
|
1179 } else {
|
|
|
1180 Lex(Result);
|
|
|
1181 }
|
|
|
1182
|
|
|
1183 // Allocate a holding buffer for a sequence of tokens and introduce it into
|
|
|
1184 // the token stream.
|
|
|
1185 auto EnterTokens = [this](ArrayRef<Token> Toks) {
|
|
|
1186 auto ToksCopy = std::make_unique<Token[]>(Toks.size());
|
|
|
1187 std::copy(Toks.begin(), Toks.end(), ToksCopy.get());
|
|
|
1188 EnterTokenStream(std::move(ToksCopy), Toks.size(),
|
|
|
1189 /*DisableMacroExpansion*/ true, /*IsReinject*/ false);
|
|
|
1190 };
|
|
|
1191
|
|
252
|
1192 bool ImportingHeader = Result.is(tok::header_name);
|
|
150
|
1193 // Check for a header-name.
|
|
|
1194 SmallVector<Token, 32> Suffix;
|
|
252
|
1195 if (ImportingHeader) {
|
|
150
|
1196 // Enter the header-name token into the token stream; a Lex action cannot
|
|
|
1197 // both return a token and cache tokens (doing so would corrupt the token
|
|
|
1198 // cache if the call to Lex comes from CachingLex / PeekAhead).
|
|
|
1199 Suffix.push_back(Result);
|
|
|
1200
|
|
|
1201 // Consume the pp-import-suffix and expand any macros in it now. We'll add
|
|
|
1202 // it back into the token stream later.
|
|
|
1203 CollectPpImportSuffix(Suffix);
|
|
|
1204 if (Suffix.back().isNot(tok::semi)) {
|
|
|
1205 // This is not a pp-import after all.
|
|
|
1206 EnterTokens(Suffix);
|
|
|
1207 return false;
|
|
|
1208 }
|
|
|
1209
|
|
|
1210 // C++2a [cpp.module]p1:
|
|
|
1211 // The ';' preprocessing-token terminating a pp-import shall not have
|
|
|
1212 // been produced by macro replacement.
|
|
|
1213 SourceLocation SemiLoc = Suffix.back().getLocation();
|
|
|
1214 if (SemiLoc.isMacroID())
|
|
|
1215 Diag(SemiLoc, diag::err_header_import_semi_in_macro);
|
|
|
1216
|
|
|
1217 // Reconstitute the import token.
|
|
|
1218 Token ImportTok;
|
|
|
1219 ImportTok.startToken();
|
|
|
1220 ImportTok.setKind(tok::kw_import);
|
|
|
1221 ImportTok.setLocation(ModuleImportLoc);
|
|
|
1222 ImportTok.setIdentifierInfo(getIdentifierInfo("import"));
|
|
|
1223 ImportTok.setLength(6);
|
|
|
1224
|
|
|
1225 auto Action = HandleHeaderIncludeOrImport(
|
|
|
1226 /*HashLoc*/ SourceLocation(), ImportTok, Suffix.front(), SemiLoc);
|
|
|
1227 switch (Action.Kind) {
|
|
|
1228 case ImportAction::None:
|
|
|
1229 break;
|
|
|
1230
|
|
|
1231 case ImportAction::ModuleBegin:
|
|
|
1232 // Let the parser know we're textually entering the module.
|
|
|
1233 Suffix.emplace_back();
|
|
|
1234 Suffix.back().startToken();
|
|
|
1235 Suffix.back().setKind(tok::annot_module_begin);
|
|
|
1236 Suffix.back().setLocation(SemiLoc);
|
|
|
1237 Suffix.back().setAnnotationEndLoc(SemiLoc);
|
|
|
1238 Suffix.back().setAnnotationValue(Action.ModuleForHeader);
|
|
236
|
1239 [[fallthrough]];
|
|
150
|
1240
|
|
|
1241 case ImportAction::ModuleImport:
|
|
236
|
1242 case ImportAction::HeaderUnitImport:
|
|
150
|
1243 case ImportAction::SkippedModuleImport:
|
|
|
1244 // We chose to import (or textually enter) the file. Convert the
|
|
|
1245 // header-name token into a header unit annotation token.
|
|
|
1246 Suffix[0].setKind(tok::annot_header_unit);
|
|
|
1247 Suffix[0].setAnnotationEndLoc(Suffix[0].getLocation());
|
|
|
1248 Suffix[0].setAnnotationValue(Action.ModuleForHeader);
|
|
|
1249 // FIXME: Call the moduleImport callback?
|
|
|
1250 break;
|
|
|
1251 case ImportAction::Failure:
|
|
|
1252 assert(TheModuleLoader.HadFatalFailure &&
|
|
|
1253 "This should be an early exit only to a fatal error");
|
|
|
1254 Result.setKind(tok::eof);
|
|
|
1255 CurLexer->cutOffLexing();
|
|
|
1256 EnterTokens(Suffix);
|
|
|
1257 return true;
|
|
|
1258 }
|
|
|
1259
|
|
|
1260 EnterTokens(Suffix);
|
|
|
1261 return false;
|
|
|
1262 }
|
|
|
1263
|
|
|
1264 // The token sequence
|
|
|
1265 //
|
|
|
1266 // import identifier (. identifier)*
|
|
|
1267 //
|
|
|
1268 // indicates a module import directive. We already saw the 'import'
|
|
|
1269 // contextual keyword, so now we're looking for the identifiers.
|
|
|
1270 if (ModuleImportExpectsIdentifier && Result.getKind() == tok::identifier) {
|
|
|
1271 // We expected to see an identifier here, and we did; continue handling
|
|
|
1272 // identifiers.
|
|
252
|
1273 NamedModuleImportPath.push_back(
|
|
|
1274 std::make_pair(Result.getIdentifierInfo(), Result.getLocation()));
|
|
150
|
1275 ModuleImportExpectsIdentifier = false;
|
|
|
1276 CurLexerKind = CLK_LexAfterModuleImport;
|
|
|
1277 return true;
|
|
|
1278 }
|
|
|
1279
|
|
|
1280 // If we're expecting a '.' or a ';', and we got a '.', then wait until we
|
|
|
1281 // see the next identifier. (We can also see a '[[' that begins an
|
|
252
|
1282 // attribute-specifier-seq here under the Standard C++ Modules.)
|
|
150
|
1283 if (!ModuleImportExpectsIdentifier && Result.getKind() == tok::period) {
|
|
|
1284 ModuleImportExpectsIdentifier = true;
|
|
|
1285 CurLexerKind = CLK_LexAfterModuleImport;
|
|
|
1286 return true;
|
|
|
1287 }
|
|
|
1288
|
|
|
1289 // If we didn't recognize a module name at all, this is not a (valid) import.
|
|
236
|
1290 if (NamedModuleImportPath.empty() || Result.is(tok::eof))
|
|
150
|
1291 return true;
|
|
|
1292
|
|
|
1293 // Consume the pp-import-suffix and expand any macros in it now, if we're not
|
|
|
1294 // at the semicolon already.
|
|
|
1295 SourceLocation SemiLoc = Result.getLocation();
|
|
|
1296 if (Result.isNot(tok::semi)) {
|
|
|
1297 Suffix.push_back(Result);
|
|
|
1298 CollectPpImportSuffix(Suffix);
|
|
|
1299 if (Suffix.back().isNot(tok::semi)) {
|
|
|
1300 // This is not an import after all.
|
|
|
1301 EnterTokens(Suffix);
|
|
|
1302 return false;
|
|
|
1303 }
|
|
|
1304 SemiLoc = Suffix.back().getLocation();
|
|
|
1305 }
|
|
|
1306
|
|
252
|
1307 // Under the standard C++ Modules, the dot is just part of the module name,
|
|
|
1308 // and not a real hierarchy separator. Flatten such module names now.
|
|
150
|
1309 //
|
|
|
1310 // FIXME: Is this the right level to be performing this transformation?
|
|
|
1311 std::string FlatModuleName;
|
|
252
|
1312 if (getLangOpts().CPlusPlusModules) {
|
|
236
|
1313 for (auto &Piece : NamedModuleImportPath) {
|
|
252
|
1314 // If the FlatModuleName ends with colon, it implies it is a partition.
|
|
|
1315 if (!FlatModuleName.empty() && FlatModuleName.back() != ':')
|
|
150
|
1316 FlatModuleName += ".";
|
|
|
1317 FlatModuleName += Piece.first->getName();
|
|
|
1318 }
|
|
236
|
1319 SourceLocation FirstPathLoc = NamedModuleImportPath[0].second;
|
|
|
1320 NamedModuleImportPath.clear();
|
|
|
1321 NamedModuleImportPath.push_back(
|
|
150
|
1322 std::make_pair(getIdentifierInfo(FlatModuleName), FirstPathLoc));
|
|
|
1323 }
|
|
|
1324
|
|
|
1325 Module *Imported = nullptr;
|
|
252
|
1326 // We don't/shouldn't load the standard c++20 modules when preprocessing.
|
|
|
1327 if (getLangOpts().Modules && !isInImportingCXXNamedModules()) {
|
|
150
|
1328 Imported = TheModuleLoader.loadModule(ModuleImportLoc,
|
|
236
|
1329 NamedModuleImportPath,
|
|
150
|
1330 Module::Hidden,
|
|
|
1331 /*IsInclusionDirective=*/false);
|
|
|
1332 if (Imported)
|
|
|
1333 makeModuleVisible(Imported, SemiLoc);
|
|
|
1334 }
|
|
252
|
1335
|
|
150
|
1336 if (Callbacks)
|
|
236
|
1337 Callbacks->moduleImport(ModuleImportLoc, NamedModuleImportPath, Imported);
|
|
150
|
1338
|
|
|
1339 if (!Suffix.empty()) {
|
|
|
1340 EnterTokens(Suffix);
|
|
|
1341 return false;
|
|
|
1342 }
|
|
|
1343 return true;
|
|
|
1344 }
|
|
|
1345
|
|
|
1346 void Preprocessor::makeModuleVisible(Module *M, SourceLocation Loc) {
|
|
|
1347 CurSubmoduleState->VisibleModules.setVisible(
|
|
|
1348 M, Loc, [](Module *) {},
|
|
|
1349 [&](ArrayRef<Module *> Path, Module *Conflict, StringRef Message) {
|
|
|
1350 // FIXME: Include the path in the diagnostic.
|
|
|
1351 // FIXME: Include the import location for the conflicting module.
|
|
|
1352 Diag(ModuleImportLoc, diag::warn_module_conflict)
|
|
|
1353 << Path[0]->getFullModuleName()
|
|
|
1354 << Conflict->getFullModuleName()
|
|
|
1355 << Message;
|
|
|
1356 });
|
|
|
1357
|
|
|
1358 // Add this module to the imports list of the currently-built submodule.
|
|
|
1359 if (!BuildingSubmoduleStack.empty() && M != BuildingSubmoduleStack.back().M)
|
|
|
1360 BuildingSubmoduleStack.back().M->Imports.insert(M);
|
|
|
1361 }
|
|
|
1362
|
|
|
1363 bool Preprocessor::FinishLexStringLiteral(Token &Result, std::string &String,
|
|
|
1364 const char *DiagnosticTag,
|
|
|
1365 bool AllowMacroExpansion) {
|
|
|
1366 // We need at least one string literal.
|
|
|
1367 if (Result.isNot(tok::string_literal)) {
|
|
|
1368 Diag(Result, diag::err_expected_string_literal)
|
|
|
1369 << /*Source='in...'*/0 << DiagnosticTag;
|
|
|
1370 return false;
|
|
|
1371 }
|
|
|
1372
|
|
|
1373 // Lex string literal tokens, optionally with macro expansion.
|
|
|
1374 SmallVector<Token, 4> StrToks;
|
|
|
1375 do {
|
|
|
1376 StrToks.push_back(Result);
|
|
|
1377
|
|
|
1378 if (Result.hasUDSuffix())
|
|
|
1379 Diag(Result, diag::err_invalid_string_udl);
|
|
|
1380
|
|
|
1381 if (AllowMacroExpansion)
|
|
|
1382 Lex(Result);
|
|
|
1383 else
|
|
|
1384 LexUnexpandedToken(Result);
|
|
|
1385 } while (Result.is(tok::string_literal));
|
|
|
1386
|
|
|
1387 // Concatenate and parse the strings.
|
|
|
1388 StringLiteralParser Literal(StrToks, *this);
|
|
236
|
1389 assert(Literal.isOrdinary() && "Didn't allow wide strings in");
|
|
150
|
1390
|
|
|
1391 if (Literal.hadError)
|
|
|
1392 return false;
|
|
|
1393
|
|
|
1394 if (Literal.Pascal) {
|
|
|
1395 Diag(StrToks[0].getLocation(), diag::err_expected_string_literal)
|
|
|
1396 << /*Source='in...'*/0 << DiagnosticTag;
|
|
|
1397 return false;
|
|
|
1398 }
|
|
|
1399
|
|
|
1400 String = std::string(Literal.GetString());
|
|
|
1401 return true;
|
|
|
1402 }
|
|
|
1403
|
|
|
1404 bool Preprocessor::parseSimpleIntegerLiteral(Token &Tok, uint64_t &Value) {
|
|
|
1405 assert(Tok.is(tok::numeric_constant));
|
|
|
1406 SmallString<8> IntegerBuffer;
|
|
|
1407 bool NumberInvalid = false;
|
|
|
1408 StringRef Spelling = getSpelling(Tok, IntegerBuffer, &NumberInvalid);
|
|
|
1409 if (NumberInvalid)
|
|
|
1410 return false;
|
|
207
|
1411 NumericLiteralParser Literal(Spelling, Tok.getLocation(), getSourceManager(),
|
|
|
1412 getLangOpts(), getTargetInfo(),
|
|
|
1413 getDiagnostics());
|
|
150
|
1414 if (Literal.hadError || !Literal.isIntegerLiteral() || Literal.hasUDSuffix())
|
|
|
1415 return false;
|
|
|
1416 llvm::APInt APVal(64, 0);
|
|
|
1417 if (Literal.GetIntegerValue(APVal))
|
|
|
1418 return false;
|
|
|
1419 Lex(Tok);
|
|
|
1420 Value = APVal.getLimitedValue();
|
|
|
1421 return true;
|
|
|
1422 }
|
|
|
1423
|
|
|
1424 void Preprocessor::addCommentHandler(CommentHandler *Handler) {
|
|
|
1425 assert(Handler && "NULL comment handler");
|
|
236
|
1426 assert(!llvm::is_contained(CommentHandlers, Handler) &&
|
|
150
|
1427 "Comment handler already registered");
|
|
|
1428 CommentHandlers.push_back(Handler);
|
|
|
1429 }
|
|
|
1430
|
|
|
1431 void Preprocessor::removeCommentHandler(CommentHandler *Handler) {
|
|
|
1432 std::vector<CommentHandler *>::iterator Pos =
|
|
|
1433 llvm::find(CommentHandlers, Handler);
|
|
|
1434 assert(Pos != CommentHandlers.end() && "Comment handler not registered");
|
|
|
1435 CommentHandlers.erase(Pos);
|
|
|
1436 }
|
|
|
1437
|
|
|
1438 bool Preprocessor::HandleComment(Token &result, SourceRange Comment) {
|
|
|
1439 bool AnyPendingTokens = false;
|
|
|
1440 for (std::vector<CommentHandler *>::iterator H = CommentHandlers.begin(),
|
|
|
1441 HEnd = CommentHandlers.end();
|
|
|
1442 H != HEnd; ++H) {
|
|
|
1443 if ((*H)->HandleComment(*this, Comment))
|
|
|
1444 AnyPendingTokens = true;
|
|
|
1445 }
|
|
|
1446 if (!AnyPendingTokens || getCommentRetentionState())
|
|
|
1447 return false;
|
|
|
1448 Lex(result);
|
|
|
1449 return true;
|
|
|
1450 }
|
|
|
1451
|
|
236
|
1452 void Preprocessor::emitMacroDeprecationWarning(const Token &Identifier) const {
|
|
|
1453 const MacroAnnotations &A =
|
|
|
1454 getMacroAnnotations(Identifier.getIdentifierInfo());
|
|
|
1455 assert(A.DeprecationInfo &&
|
|
|
1456 "Macro deprecation warning without recorded annotation!");
|
|
|
1457 const MacroAnnotationInfo &Info = *A.DeprecationInfo;
|
|
|
1458 if (Info.Message.empty())
|
|
|
1459 Diag(Identifier, diag::warn_pragma_deprecated_macro_use)
|
|
|
1460 << Identifier.getIdentifierInfo() << 0;
|
|
|
1461 else
|
|
|
1462 Diag(Identifier, diag::warn_pragma_deprecated_macro_use)
|
|
|
1463 << Identifier.getIdentifierInfo() << 1 << Info.Message;
|
|
|
1464 Diag(Info.Location, diag::note_pp_macro_annotation) << 0;
|
|
|
1465 }
|
|
|
1466
|
|
|
1467 void Preprocessor::emitRestrictExpansionWarning(const Token &Identifier) const {
|
|
|
1468 const MacroAnnotations &A =
|
|
|
1469 getMacroAnnotations(Identifier.getIdentifierInfo());
|
|
|
1470 assert(A.RestrictExpansionInfo &&
|
|
|
1471 "Macro restricted expansion warning without recorded annotation!");
|
|
|
1472 const MacroAnnotationInfo &Info = *A.RestrictExpansionInfo;
|
|
|
1473 if (Info.Message.empty())
|
|
|
1474 Diag(Identifier, diag::warn_pragma_restrict_expansion_macro_use)
|
|
|
1475 << Identifier.getIdentifierInfo() << 0;
|
|
|
1476 else
|
|
|
1477 Diag(Identifier, diag::warn_pragma_restrict_expansion_macro_use)
|
|
|
1478 << Identifier.getIdentifierInfo() << 1 << Info.Message;
|
|
|
1479 Diag(Info.Location, diag::note_pp_macro_annotation) << 1;
|
|
|
1480 }
|
|
|
1481
|
|
|
1482 void Preprocessor::emitFinalMacroWarning(const Token &Identifier,
|
|
|
1483 bool IsUndef) const {
|
|
|
1484 const MacroAnnotations &A =
|
|
|
1485 getMacroAnnotations(Identifier.getIdentifierInfo());
|
|
|
1486 assert(A.FinalAnnotationLoc &&
|
|
|
1487 "Final macro warning without recorded annotation!");
|
|
|
1488
|
|
|
1489 Diag(Identifier, diag::warn_pragma_final_macro)
|
|
|
1490 << Identifier.getIdentifierInfo() << (IsUndef ? 0 : 1);
|
|
|
1491 Diag(*A.FinalAnnotationLoc, diag::note_pp_macro_annotation) << 2;
|
|
|
1492 }
|
|
|
1493
|
|
252
|
1494 bool Preprocessor::isSafeBufferOptOut(const SourceManager &SourceMgr,
|
|
|
1495 const SourceLocation &Loc) const {
|
|
|
1496 // Try to find a region in `SafeBufferOptOutMap` where `Loc` is in:
|
|
|
1497 auto FirstRegionEndingAfterLoc = llvm::partition_point(
|
|
|
1498 SafeBufferOptOutMap,
|
|
|
1499 [&SourceMgr,
|
|
|
1500 &Loc](const std::pair<SourceLocation, SourceLocation> &Region) {
|
|
|
1501 return SourceMgr.isBeforeInTranslationUnit(Region.second, Loc);
|
|
|
1502 });
|
|
|
1503
|
|
|
1504 if (FirstRegionEndingAfterLoc != SafeBufferOptOutMap.end()) {
|
|
|
1505 // To test if the start location of the found region precedes `Loc`:
|
|
|
1506 return SourceMgr.isBeforeInTranslationUnit(FirstRegionEndingAfterLoc->first,
|
|
|
1507 Loc);
|
|
|
1508 }
|
|
|
1509 // If we do not find a region whose end location passes `Loc`, we want to
|
|
|
1510 // check if the current region is still open:
|
|
|
1511 if (!SafeBufferOptOutMap.empty() &&
|
|
|
1512 SafeBufferOptOutMap.back().first == SafeBufferOptOutMap.back().second)
|
|
|
1513 return SourceMgr.isBeforeInTranslationUnit(SafeBufferOptOutMap.back().first,
|
|
|
1514 Loc);
|
|
|
1515 return false;
|
|
|
1516 }
|
|
|
1517
|
|
|
1518 bool Preprocessor::enterOrExitSafeBufferOptOutRegion(
|
|
|
1519 bool isEnter, const SourceLocation &Loc) {
|
|
|
1520 if (isEnter) {
|
|
|
1521 if (isPPInSafeBufferOptOutRegion())
|
|
|
1522 return true; // invalid enter action
|
|
|
1523 InSafeBufferOptOutRegion = true;
|
|
|
1524 CurrentSafeBufferOptOutStart = Loc;
|
|
|
1525
|
|
|
1526 // To set the start location of a new region:
|
|
|
1527
|
|
|
1528 if (!SafeBufferOptOutMap.empty()) {
|
|
|
1529 [[maybe_unused]] auto *PrevRegion = &SafeBufferOptOutMap.back();
|
|
|
1530 assert(PrevRegion->first != PrevRegion->second &&
|
|
|
1531 "Shall not begin a safe buffer opt-out region before closing the "
|
|
|
1532 "previous one.");
|
|
|
1533 }
|
|
|
1534 // If the start location equals to the end location, we call the region a
|
|
|
1535 // open region or a unclosed region (i.e., end location has not been set
|
|
|
1536 // yet).
|
|
|
1537 SafeBufferOptOutMap.emplace_back(Loc, Loc);
|
|
|
1538 } else {
|
|
|
1539 if (!isPPInSafeBufferOptOutRegion())
|
|
|
1540 return true; // invalid enter action
|
|
|
1541 InSafeBufferOptOutRegion = false;
|
|
|
1542
|
|
|
1543 // To set the end location of the current open region:
|
|
|
1544
|
|
|
1545 assert(!SafeBufferOptOutMap.empty() &&
|
|
|
1546 "Misordered safe buffer opt-out regions");
|
|
|
1547 auto *CurrRegion = &SafeBufferOptOutMap.back();
|
|
|
1548 assert(CurrRegion->first == CurrRegion->second &&
|
|
|
1549 "Set end location to a closed safe buffer opt-out region");
|
|
|
1550 CurrRegion->second = Loc;
|
|
|
1551 }
|
|
|
1552 return false;
|
|
|
1553 }
|
|
|
1554
|
|
|
1555 bool Preprocessor::isPPInSafeBufferOptOutRegion() {
|
|
|
1556 return InSafeBufferOptOutRegion;
|
|
|
1557 }
|
|
|
1558 bool Preprocessor::isPPInSafeBufferOptOutRegion(SourceLocation &StartLoc) {
|
|
|
1559 StartLoc = CurrentSafeBufferOptOutStart;
|
|
|
1560 return InSafeBufferOptOutRegion;
|
|
|
1561 }
|
|
|
1562
|
|
150
|
1563 ModuleLoader::~ModuleLoader() = default;
|
|
|
1564
|
|
|
1565 CommentHandler::~CommentHandler() = default;
|
|
|
1566
|
|
207
|
1567 EmptylineHandler::~EmptylineHandler() = default;
|
|
|
1568
|
|
150
|
1569 CodeCompletionHandler::~CodeCompletionHandler() = default;
|
|
|
1570
|
|
|
1571 void Preprocessor::createPreprocessingRecord() {
|
|
|
1572 if (Record)
|
|
|
1573 return;
|
|
|
1574
|
|
|
1575 Record = new PreprocessingRecord(getSourceManager());
|
|
|
1576 addPPCallbacks(std::unique_ptr<PPCallbacks>(Record));
|
|
|
1577 }
|
|
240
|
1578
|
|
|
1579 #ifndef noCbC
|
|
|
1580
|
|
|
1581 Token Preprocessor::ReadFromString(const char *src , SourceLocation Loc) {
|
|
|
1582 // Push the ( "string" ) tokens into the token stream.
|
|
|
1583 MacroInfo *MI = AllocateMacroInfo(Loc);
|
|
242
|
1584 SmallVector<Token, 16> Tokens;
|
|
240
|
1585 Token Tok;
|
|
|
1586 std::unique_ptr<Lexer> lx(new Lexer(CurLexer->getFileID(),getSourceManager().getBufferOrFake(CurLexer->getFileID(), Loc),*this));
|
|
|
1587 lx->InitLexer(src,src,src + strlen(src));
|
|
|
1588 lx->Lex(Tok);
|
|
|
1589 CurLexer.swap(lx);
|
|
|
1590 int i = 0;
|
|
|
1591 while (Tok.getKind() != tok::TokenKind::eof) {
|
|
|
1592 Tok.setLocation(Loc);
|
|
242
|
1593 Tokens.push_back(Tok);
|
|
240
|
1594 Lex(Tok); i++;
|
|
|
1595 }
|
|
|
1596 Tok.setLocation(Loc);
|
|
242
|
1597 Tokens.push_back(Tok);
|
|
|
1598 i++;
|
|
240
|
1599 MI->DefinitionLength = i;
|
|
|
1600 CurLexer = std::move(lx);
|
|
|
1601 CurPPLexer = CurLexer.get();
|
|
242
|
1602 MI->setTokens(Tokens, BP);
|
|
240
|
1603 EnterMacro(Tok, Loc, MI , 0 );
|
|
|
1604 CurTokenLexer->MacroDefLength = i;
|
|
|
1605 return Tok;
|
|
|
1606 }
|
|
|
1607
|
|
|
1608 #endif
|
