1 //===-- clang-import-test.cpp - ASTImporter/ExternalASTSource testbed -----===//
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 #include "clang/AST/ASTContext.h"
10 #include "clang/AST/ASTImporter.h"
11 #include "clang/AST/DeclObjC.h"
12 #include "clang/AST/ExternalASTMerger.h"
13 #include "clang/Basic/Builtins.h"
14 #include "clang/Basic/FileManager.h"
15 #include "clang/Basic/IdentifierTable.h"
16 #include "clang/Basic/SourceLocation.h"
17 #include "clang/Basic/TargetInfo.h"
18 #include "clang/Basic/TargetOptions.h"
19 #include "clang/CodeGen/ModuleBuilder.h"
20 #include "clang/Driver/Types.h"
21 #include "clang/Frontend/ASTConsumers.h"
22 #include "clang/Frontend/CompilerInstance.h"
23 #include "clang/Frontend/MultiplexConsumer.h"
24 #include "clang/Frontend/TextDiagnosticBuffer.h"
25 #include "clang/Lex/Lexer.h"
26 #include "clang/Lex/Preprocessor.h"
27 #include "clang/Parse/ParseAST.h"
28 
29 #include "llvm/IR/LLVMContext.h"
30 #include "llvm/IR/Module.h"
31 #include "llvm/Support/CommandLine.h"
32 #include "llvm/Support/Error.h"
33 #include "llvm/Support/Host.h"
34 #include "llvm/Support/Signals.h"
35 
36 #include <memory>
37 #include <string>
38 
39 using namespace clang;
40 
41 static llvm::cl::opt<std::string> Expression(
42     "expression", llvm::cl::Required,
43     llvm::cl::desc("Path to a file containing the expression to parse"));
44 
45 static llvm::cl::list<std::string>
46     Imports("import", llvm::cl::ZeroOrMore,
47             llvm::cl::desc("Path to a file containing declarations to import"));
48 
49 static llvm::cl::opt<bool>
50     Direct("direct", llvm::cl::Optional,
51            llvm::cl::desc("Use the parsed declarations without indirection"));
52 
53 static llvm::cl::opt<bool> UseOrigins(
54     "use-origins", llvm::cl::Optional,
55     llvm::cl::desc(
56         "Use DeclContext origin information for more accurate lookups"));
57 
58 static llvm::cl::list<std::string>
59     ClangArgs("Xcc", llvm::cl::ZeroOrMore,
60               llvm::cl::desc("Argument to pass to the CompilerInvocation"),
61               llvm::cl::CommaSeparated);
62 
63 static llvm::cl::opt<std::string>
64     Input("x", llvm::cl::Optional,
65           llvm::cl::desc("The language to parse (default: c++)"),
66           llvm::cl::init("c++"));
67 
68 static llvm::cl::opt<bool> ObjCARC("objc-arc", llvm::cl::init(false),
69                                    llvm::cl::desc("Emable ObjC ARC"));
70 
71 static llvm::cl::opt<bool> DumpAST("dump-ast", llvm::cl::init(false),
72                                    llvm::cl::desc("Dump combined AST"));
73 
74 static llvm::cl::opt<bool> DumpIR("dump-ir", llvm::cl::init(false),
75                                   llvm::cl::desc("Dump IR from final parse"));
76 
77 namespace init_convenience {
78 class TestDiagnosticConsumer : public DiagnosticConsumer {
79 private:
80   std::unique_ptr<TextDiagnosticBuffer> Passthrough;
81   const LangOptions *LangOpts = nullptr;
82 
83 public:
TestDiagnosticConsumer()84   TestDiagnosticConsumer()
85       : Passthrough(std::make_unique<TextDiagnosticBuffer>()) {}
86 
BeginSourceFile(const LangOptions & LangOpts,const Preprocessor * PP=nullptr)87   virtual void BeginSourceFile(const LangOptions &LangOpts,
88                                const Preprocessor *PP = nullptr) override {
89     this->LangOpts = &LangOpts;
90     return Passthrough->BeginSourceFile(LangOpts, PP);
91   }
92 
EndSourceFile()93   virtual void EndSourceFile() override {
94     this->LangOpts = nullptr;
95     Passthrough->EndSourceFile();
96   }
97 
IncludeInDiagnosticCounts() const98   virtual bool IncludeInDiagnosticCounts() const override {
99     return Passthrough->IncludeInDiagnosticCounts();
100   }
101 
102 private:
PrintSourceForLocation(const SourceLocation & Loc,SourceManager & SM)103   static void PrintSourceForLocation(const SourceLocation &Loc,
104                                      SourceManager &SM) {
105     const char *LocData = SM.getCharacterData(Loc, /*Invalid=*/nullptr);
106     unsigned LocColumn =
107         SM.getSpellingColumnNumber(Loc, /*Invalid=*/nullptr) - 1;
108     FileID FID = SM.getFileID(Loc);
109     llvm::MemoryBufferRef Buffer = SM.getBufferOrFake(FID, Loc);
110 
111     assert(LocData >= Buffer.getBufferStart() &&
112            LocData < Buffer.getBufferEnd());
113 
114     const char *LineBegin = LocData - LocColumn;
115 
116     assert(LineBegin >= Buffer.getBufferStart());
117 
118     const char *LineEnd = nullptr;
119 
120     for (LineEnd = LineBegin; *LineEnd != '\n' && *LineEnd != '\r' &&
121                               LineEnd < Buffer.getBufferEnd();
122          ++LineEnd)
123       ;
124 
125     llvm::StringRef LineString(LineBegin, LineEnd - LineBegin);
126 
127     llvm::errs() << LineString << '\n';
128     llvm::errs().indent(LocColumn);
129     llvm::errs() << '^';
130     llvm::errs() << '\n';
131   }
132 
HandleDiagnostic(DiagnosticsEngine::Level DiagLevel,const Diagnostic & Info)133   virtual void HandleDiagnostic(DiagnosticsEngine::Level DiagLevel,
134                                 const Diagnostic &Info) override {
135     if (Info.hasSourceManager() && LangOpts) {
136       SourceManager &SM = Info.getSourceManager();
137 
138       if (Info.getLocation().isValid()) {
139         Info.getLocation().print(llvm::errs(), SM);
140         llvm::errs() << ": ";
141       }
142 
143       SmallString<16> DiagText;
144       Info.FormatDiagnostic(DiagText);
145       llvm::errs() << DiagText << '\n';
146 
147       if (Info.getLocation().isValid()) {
148         PrintSourceForLocation(Info.getLocation(), SM);
149       }
150 
151       for (const CharSourceRange &Range : Info.getRanges()) {
152         bool Invalid = true;
153         StringRef Ref = Lexer::getSourceText(Range, SM, *LangOpts, &Invalid);
154         if (!Invalid) {
155           llvm::errs() << Ref << '\n';
156         }
157       }
158     }
159     DiagnosticConsumer::HandleDiagnostic(DiagLevel, Info);
160   }
161 };
162 
BuildCompilerInstance()163 std::unique_ptr<CompilerInstance> BuildCompilerInstance() {
164   auto Ins = std::make_unique<CompilerInstance>();
165   auto DC = std::make_unique<TestDiagnosticConsumer>();
166   const bool ShouldOwnClient = true;
167   Ins->createDiagnostics(DC.release(), ShouldOwnClient);
168 
169   auto Inv = std::make_unique<CompilerInvocation>();
170 
171   std::vector<const char *> ClangArgv(ClangArgs.size());
172   std::transform(ClangArgs.begin(), ClangArgs.end(), ClangArgv.begin(),
173                  [](const std::string &s) -> const char * { return s.data(); });
174   CompilerInvocation::CreateFromArgs(*Inv, ClangArgv, Ins->getDiagnostics());
175 
176   {
177     using namespace driver::types;
178     ID Id = lookupTypeForTypeSpecifier(Input.c_str());
179     assert(Id != TY_INVALID);
180     if (isCXX(Id)) {
181       Inv->getLangOpts()->CPlusPlus = true;
182       Inv->getLangOpts()->CPlusPlus11 = true;
183       Inv->getHeaderSearchOpts().UseLibcxx = true;
184     }
185     if (isObjC(Id)) {
186       Inv->getLangOpts()->ObjC = 1;
187     }
188   }
189   Inv->getLangOpts()->ObjCAutoRefCount = ObjCARC;
190 
191   Inv->getLangOpts()->Bool = true;
192   Inv->getLangOpts()->WChar = true;
193   Inv->getLangOpts()->Blocks = true;
194   Inv->getLangOpts()->DebuggerSupport = true;
195   Inv->getLangOpts()->SpellChecking = false;
196   Inv->getLangOpts()->ThreadsafeStatics = false;
197   Inv->getLangOpts()->AccessControl = false;
198   Inv->getLangOpts()->DollarIdents = true;
199   Inv->getLangOpts()->Exceptions = true;
200   Inv->getLangOpts()->CXXExceptions = true;
201   // Needed for testing dynamic_cast.
202   Inv->getLangOpts()->RTTI = true;
203   Inv->getCodeGenOpts().setDebugInfo(codegenoptions::FullDebugInfo);
204   Inv->getTargetOpts().Triple = llvm::sys::getDefaultTargetTriple();
205 
206   Ins->setInvocation(std::move(Inv));
207 
208   TargetInfo *TI = TargetInfo::CreateTargetInfo(
209       Ins->getDiagnostics(), Ins->getInvocation().TargetOpts);
210   Ins->setTarget(TI);
211   Ins->getTarget().adjust(Ins->getLangOpts());
212   Ins->createFileManager();
213   Ins->createSourceManager(Ins->getFileManager());
214   Ins->createPreprocessor(TU_Complete);
215 
216   return Ins;
217 }
218 
219 std::unique_ptr<ASTContext>
BuildASTContext(CompilerInstance & CI,SelectorTable & ST,Builtin::Context & BC)220 BuildASTContext(CompilerInstance &CI, SelectorTable &ST, Builtin::Context &BC) {
221   auto AST = std::make_unique<ASTContext>(
222       CI.getLangOpts(), CI.getSourceManager(),
223       CI.getPreprocessor().getIdentifierTable(), ST, BC);
224   AST->InitBuiltinTypes(CI.getTarget());
225   return AST;
226 }
227 
BuildCodeGen(CompilerInstance & CI,llvm::LLVMContext & LLVMCtx)228 std::unique_ptr<CodeGenerator> BuildCodeGen(CompilerInstance &CI,
229                                             llvm::LLVMContext &LLVMCtx) {
230   StringRef ModuleName("$__module");
231   return std::unique_ptr<CodeGenerator>(CreateLLVMCodeGen(
232       CI.getDiagnostics(), ModuleName, CI.getHeaderSearchOpts(),
233       CI.getPreprocessorOpts(), CI.getCodeGenOpts(), LLVMCtx));
234 }
235 } // namespace init_convenience
236 
237 namespace {
238 
239 /// A container for a CompilerInstance (possibly with an ExternalASTMerger
240 /// attached to its ASTContext).
241 ///
242 /// Provides an accessor for the DeclContext origins associated with the
243 /// ExternalASTMerger (or an empty list of origins if no ExternalASTMerger is
244 /// attached).
245 ///
246 /// This is the main unit of parsed source code maintained by clang-import-test.
247 struct CIAndOrigins {
248   using OriginMap = clang::ExternalASTMerger::OriginMap;
249   std::unique_ptr<CompilerInstance> CI;
250 
getASTContext__anon5a8deb180211::CIAndOrigins251   ASTContext &getASTContext() { return CI->getASTContext(); }
getFileManager__anon5a8deb180211::CIAndOrigins252   FileManager &getFileManager() { return CI->getFileManager(); }
getOriginMap__anon5a8deb180211::CIAndOrigins253   const OriginMap &getOriginMap() {
254     static const OriginMap EmptyOriginMap{};
255     if (ExternalASTSource *Source = CI->getASTContext().getExternalSource())
256       return static_cast<ExternalASTMerger *>(Source)->GetOrigins();
257     return EmptyOriginMap;
258   }
getDiagnosticClient__anon5a8deb180211::CIAndOrigins259   DiagnosticConsumer &getDiagnosticClient() {
260     return CI->getDiagnosticClient();
261   }
getCompilerInstance__anon5a8deb180211::CIAndOrigins262   CompilerInstance &getCompilerInstance() { return *CI; }
263 };
264 
AddExternalSource(CIAndOrigins & CI,llvm::MutableArrayRef<CIAndOrigins> Imports)265 void AddExternalSource(CIAndOrigins &CI,
266                        llvm::MutableArrayRef<CIAndOrigins> Imports) {
267   ExternalASTMerger::ImporterTarget Target(
268       {CI.getASTContext(), CI.getFileManager()});
269   llvm::SmallVector<ExternalASTMerger::ImporterSource, 3> Sources;
270   for (CIAndOrigins &Import : Imports)
271     Sources.emplace_back(Import.getASTContext(), Import.getFileManager(),
272                          Import.getOriginMap());
273   auto ES = std::make_unique<ExternalASTMerger>(Target, Sources);
274   CI.getASTContext().setExternalSource(ES.release());
275   CI.getASTContext().getTranslationUnitDecl()->setHasExternalVisibleStorage();
276 }
277 
BuildIndirect(CIAndOrigins & CI)278 CIAndOrigins BuildIndirect(CIAndOrigins &CI) {
279   CIAndOrigins IndirectCI{init_convenience::BuildCompilerInstance()};
280   auto ST = std::make_unique<SelectorTable>();
281   auto BC = std::make_unique<Builtin::Context>();
282   std::unique_ptr<ASTContext> AST = init_convenience::BuildASTContext(
283       IndirectCI.getCompilerInstance(), *ST, *BC);
284   IndirectCI.getCompilerInstance().setASTContext(AST.release());
285   AddExternalSource(IndirectCI, CI);
286   return IndirectCI;
287 }
288 
ParseSource(const std::string & Path,CompilerInstance & CI,ASTConsumer & Consumer)289 llvm::Error ParseSource(const std::string &Path, CompilerInstance &CI,
290                         ASTConsumer &Consumer) {
291   SourceManager &SM = CI.getSourceManager();
292   auto FE = CI.getFileManager().getFileRef(Path);
293   if (!FE) {
294     llvm::consumeError(FE.takeError());
295     return llvm::make_error<llvm::StringError>(
296         llvm::Twine("No such file or directory: ", Path), std::error_code());
297   }
298   SM.setMainFileID(SM.createFileID(*FE, SourceLocation(), SrcMgr::C_User));
299   ParseAST(CI.getPreprocessor(), &Consumer, CI.getASTContext());
300   return llvm::Error::success();
301 }
302 
Parse(const std::string & Path,llvm::MutableArrayRef<CIAndOrigins> Imports,bool ShouldDumpAST,bool ShouldDumpIR)303 llvm::Expected<CIAndOrigins> Parse(const std::string &Path,
304                                    llvm::MutableArrayRef<CIAndOrigins> Imports,
305                                    bool ShouldDumpAST, bool ShouldDumpIR) {
306   CIAndOrigins CI{init_convenience::BuildCompilerInstance()};
307   auto ST = std::make_unique<SelectorTable>();
308   auto BC = std::make_unique<Builtin::Context>();
309   std::unique_ptr<ASTContext> AST =
310       init_convenience::BuildASTContext(CI.getCompilerInstance(), *ST, *BC);
311   CI.getCompilerInstance().setASTContext(AST.release());
312   if (Imports.size())
313     AddExternalSource(CI, Imports);
314 
315   std::vector<std::unique_ptr<ASTConsumer>> ASTConsumers;
316 
317   auto LLVMCtx = std::make_unique<llvm::LLVMContext>();
318   ASTConsumers.push_back(
319       init_convenience::BuildCodeGen(CI.getCompilerInstance(), *LLVMCtx));
320   auto &CG = *static_cast<CodeGenerator *>(ASTConsumers.back().get());
321 
322   if (ShouldDumpAST)
323     ASTConsumers.push_back(CreateASTDumper(nullptr /*Dump to stdout.*/, "",
324                                            true, false, false, false,
325                                            clang::ADOF_Default));
326 
327   CI.getDiagnosticClient().BeginSourceFile(
328       CI.getCompilerInstance().getLangOpts(),
329       &CI.getCompilerInstance().getPreprocessor());
330   MultiplexConsumer Consumers(std::move(ASTConsumers));
331   Consumers.Initialize(CI.getASTContext());
332 
333   if (llvm::Error PE = ParseSource(Path, CI.getCompilerInstance(), Consumers))
334     return std::move(PE);
335   CI.getDiagnosticClient().EndSourceFile();
336   if (ShouldDumpIR)
337     CG.GetModule()->print(llvm::outs(), nullptr);
338   if (CI.getDiagnosticClient().getNumErrors())
339     return llvm::make_error<llvm::StringError>(
340         "Errors occurred while parsing the expression.", std::error_code());
341   return std::move(CI);
342 }
343 
Forget(CIAndOrigins & CI,llvm::MutableArrayRef<CIAndOrigins> Imports)344 void Forget(CIAndOrigins &CI, llvm::MutableArrayRef<CIAndOrigins> Imports) {
345   llvm::SmallVector<ExternalASTMerger::ImporterSource, 3> Sources;
346   for (CIAndOrigins &Import : Imports)
347     Sources.push_back({Import.getASTContext(), Import.getFileManager(),
348                        Import.getOriginMap()});
349   ExternalASTSource *Source = CI.CI->getASTContext().getExternalSource();
350   auto *Merger = static_cast<ExternalASTMerger *>(Source);
351   Merger->RemoveSources(Sources);
352 }
353 
354 } // end namespace
355 
main(int argc,const char ** argv)356 int main(int argc, const char **argv) {
357   const bool DisableCrashReporting = true;
358   llvm::sys::PrintStackTraceOnErrorSignal(argv[0], DisableCrashReporting);
359   llvm::cl::ParseCommandLineOptions(argc, argv);
360   std::vector<CIAndOrigins> ImportCIs;
361   for (auto I : Imports) {
362     llvm::Expected<CIAndOrigins> ImportCI = Parse(I, {}, false, false);
363     if (auto E = ImportCI.takeError()) {
364       llvm::errs() << "error: " << llvm::toString(std::move(E)) << "\n";
365       exit(-1);
366     }
367     ImportCIs.push_back(std::move(*ImportCI));
368   }
369   std::vector<CIAndOrigins> IndirectCIs;
370   if (!Direct || UseOrigins) {
371     for (auto &ImportCI : ImportCIs) {
372       CIAndOrigins IndirectCI = BuildIndirect(ImportCI);
373       IndirectCIs.push_back(std::move(IndirectCI));
374     }
375   }
376   if (UseOrigins)
377     for (auto &ImportCI : ImportCIs)
378       IndirectCIs.push_back(std::move(ImportCI));
379   llvm::Expected<CIAndOrigins> ExpressionCI =
380       Parse(Expression, (Direct && !UseOrigins) ? ImportCIs : IndirectCIs,
381             DumpAST, DumpIR);
382   if (auto E = ExpressionCI.takeError()) {
383     llvm::errs() << "error: " << llvm::toString(std::move(E)) << "\n";
384     exit(-1);
385   }
386   Forget(*ExpressionCI, (Direct && !UseOrigins) ? ImportCIs : IndirectCIs);
387   return 0;
388 }
389