1 //===--- Tooling.cpp - Running clang standalone tools ---------------------===//
2 //
3 //                     The LLVM Compiler Infrastructure
4 //
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
7 //
8 //===----------------------------------------------------------------------===//
9 //
10 //  This file implements functions to run clang tools standalone instead
11 //  of running them as a plugin.
12 //
13 //===----------------------------------------------------------------------===//
14 
15 #include "clang/Tooling/Tooling.h"
16 #include "clang/Driver/Compilation.h"
17 #include "clang/Driver/Driver.h"
18 #include "clang/Driver/Tool.h"
19 #include "clang/Frontend/CompilerInstance.h"
20 #include "clang/Frontend/FrontendDiagnostic.h"
21 #include "clang/Frontend/TextDiagnosticPrinter.h"
22 #include "clang/Tooling/ArgumentsAdjusters.h"
23 #include "clang/Tooling/CompilationDatabase.h"
24 #include "llvm/ADT/STLExtras.h"
25 #include "llvm/Option/Option.h"
26 #include "llvm/Support/Debug.h"
27 #include "llvm/Support/FileSystem.h"
28 #include "llvm/Support/Host.h"
29 #include "llvm/Support/raw_ostream.h"
30 
31 // For chdir, see the comment in ClangTool::run for more information.
32 #ifdef _WIN32
33 #  include <direct.h>
34 #else
35 #  include <unistd.h>
36 #endif
37 
38 namespace clang {
39 namespace tooling {
40 
41 FrontendActionFactory::~FrontendActionFactory() {}
42 
43 // FIXME: This file contains structural duplication with other parts of the
44 // code that sets up a compiler to run tools on it, and we should refactor
45 // it to be based on the same framework.
46 
47 /// \brief Builds a clang driver initialized for running clang tools.
48 static clang::driver::Driver *newDriver(clang::DiagnosticsEngine *Diagnostics,
49                                         const char *BinaryName) {
50   const std::string DefaultOutputName = "a.out";
51   clang::driver::Driver *CompilerDriver = new clang::driver::Driver(
52     BinaryName, llvm::sys::getDefaultTargetTriple(),
53     DefaultOutputName, *Diagnostics);
54   CompilerDriver->setTitle("clang_based_tool");
55   return CompilerDriver;
56 }
57 
58 /// \brief Retrieves the clang CC1 specific flags out of the compilation's jobs.
59 ///
60 /// Returns NULL on error.
61 static const llvm::opt::ArgStringList *getCC1Arguments(
62     clang::DiagnosticsEngine *Diagnostics,
63     clang::driver::Compilation *Compilation) {
64   // We expect to get back exactly one Command job, if we didn't something
65   // failed. Extract that job from the Compilation.
66   const clang::driver::JobList &Jobs = Compilation->getJobs();
67   if (Jobs.size() != 1 || !isa<clang::driver::Command>(*Jobs.begin())) {
68     SmallString<256> error_msg;
69     llvm::raw_svector_ostream error_stream(error_msg);
70     Compilation->PrintJob(error_stream, Compilation->getJobs(), "; ", true);
71     Diagnostics->Report(clang::diag::err_fe_expected_compiler_job)
72         << error_stream.str();
73     return NULL;
74   }
75 
76   // The one job we find should be to invoke clang again.
77   const clang::driver::Command *Cmd =
78       cast<clang::driver::Command>(*Jobs.begin());
79   if (StringRef(Cmd->getCreator().getName()) != "clang") {
80     Diagnostics->Report(clang::diag::err_fe_expected_clang_command);
81     return NULL;
82   }
83 
84   return &Cmd->getArguments();
85 }
86 
87 /// \brief Returns a clang build invocation initialized from the CC1 flags.
88 static clang::CompilerInvocation *newInvocation(
89     clang::DiagnosticsEngine *Diagnostics,
90     const llvm::opt::ArgStringList &CC1Args) {
91   assert(!CC1Args.empty() && "Must at least contain the program name!");
92   clang::CompilerInvocation *Invocation = new clang::CompilerInvocation;
93   clang::CompilerInvocation::CreateFromArgs(
94       *Invocation, CC1Args.data() + 1, CC1Args.data() + CC1Args.size(),
95       *Diagnostics);
96   Invocation->getFrontendOpts().DisableFree = false;
97   return Invocation;
98 }
99 
100 bool runToolOnCode(clang::FrontendAction *ToolAction, const Twine &Code,
101                    const Twine &FileName) {
102   return runToolOnCodeWithArgs(
103       ToolAction, Code, std::vector<std::string>(), FileName);
104 }
105 
106 bool runToolOnCodeWithArgs(clang::FrontendAction *ToolAction, const Twine &Code,
107                            const std::vector<std::string> &Args,
108                            const Twine &FileName) {
109   SmallString<16> FileNameStorage;
110   StringRef FileNameRef = FileName.toNullTerminatedStringRef(FileNameStorage);
111   std::vector<std::string> Commands;
112   Commands.push_back("clang-tool");
113   Commands.push_back("-fsyntax-only");
114   Commands.insert(Commands.end(), Args.begin(), Args.end());
115   Commands.push_back(FileNameRef.data());
116   FileManager Files((FileSystemOptions()));
117   ToolInvocation Invocation(Commands, ToolAction, &Files);
118 
119   SmallString<1024> CodeStorage;
120   Invocation.mapVirtualFile(FileNameRef,
121                             Code.toNullTerminatedStringRef(CodeStorage));
122   return Invocation.run();
123 }
124 
125 std::string getAbsolutePath(StringRef File) {
126   SmallString<1024> BaseDirectory;
127   if (const char *PWD = ::getenv("PWD"))
128     BaseDirectory = PWD;
129   else
130     llvm::sys::fs::current_path(BaseDirectory);
131   SmallString<1024> PathStorage;
132   if (llvm::sys::path::is_absolute(File)) {
133     llvm::sys::path::native(File, PathStorage);
134     return PathStorage.str();
135   }
136   StringRef RelativePath(File);
137   // FIXME: Should '.\\' be accepted on Win32?
138   if (RelativePath.startswith("./")) {
139     RelativePath = RelativePath.substr(strlen("./"));
140   }
141   SmallString<1024> AbsolutePath(BaseDirectory);
142   llvm::sys::path::append(AbsolutePath, RelativePath);
143   llvm::sys::path::native(Twine(AbsolutePath), PathStorage);
144   return PathStorage.str();
145 }
146 
147 ToolInvocation::ToolInvocation(
148     ArrayRef<std::string> CommandLine, FrontendAction *ToolAction,
149     FileManager *Files)
150     : CommandLine(CommandLine.vec()), ToolAction(ToolAction), Files(Files) {
151 }
152 
153 void ToolInvocation::mapVirtualFile(StringRef FilePath, StringRef Content) {
154   SmallString<1024> PathStorage;
155   llvm::sys::path::native(FilePath, PathStorage);
156   MappedFileContents[PathStorage] = Content;
157 }
158 
159 bool ToolInvocation::run() {
160   std::vector<const char*> Argv;
161   for (int I = 0, E = CommandLine.size(); I != E; ++I)
162     Argv.push_back(CommandLine[I].c_str());
163   const char *const BinaryName = Argv[0];
164   IntrusiveRefCntPtr<DiagnosticOptions> DiagOpts = new DiagnosticOptions();
165   TextDiagnosticPrinter DiagnosticPrinter(
166       llvm::errs(), &*DiagOpts);
167   DiagnosticsEngine Diagnostics(
168     IntrusiveRefCntPtr<clang::DiagnosticIDs>(new DiagnosticIDs()),
169     &*DiagOpts, &DiagnosticPrinter, false);
170 
171   const OwningPtr<clang::driver::Driver> Driver(
172       newDriver(&Diagnostics, BinaryName));
173   // Since the input might only be virtual, don't check whether it exists.
174   Driver->setCheckInputsExist(false);
175   const OwningPtr<clang::driver::Compilation> Compilation(
176       Driver->BuildCompilation(llvm::makeArrayRef(Argv)));
177   const llvm::opt::ArgStringList *const CC1Args = getCC1Arguments(
178       &Diagnostics, Compilation.get());
179   if (CC1Args == NULL) {
180     return false;
181   }
182   OwningPtr<clang::CompilerInvocation> Invocation(
183       newInvocation(&Diagnostics, *CC1Args));
184   return runInvocation(BinaryName, Compilation.get(), Invocation.take());
185 }
186 
187 bool ToolInvocation::runInvocation(
188     const char *BinaryName,
189     clang::driver::Compilation *Compilation,
190     clang::CompilerInvocation *Invocation) {
191   // Show the invocation, with -v.
192   if (Invocation->getHeaderSearchOpts().Verbose) {
193     llvm::errs() << "clang Invocation:\n";
194     Compilation->PrintJob(llvm::errs(), Compilation->getJobs(), "\n", true);
195     llvm::errs() << "\n";
196   }
197 
198   // Create a compiler instance to handle the actual work.
199   clang::CompilerInstance Compiler;
200   Compiler.setInvocation(Invocation);
201   Compiler.setFileManager(Files);
202   // FIXME: What about LangOpts?
203 
204   // ToolAction can have lifetime requirements for Compiler or its members, and
205   // we need to ensure it's deleted earlier than Compiler. So we pass it to an
206   // OwningPtr declared after the Compiler variable.
207   OwningPtr<FrontendAction> ScopedToolAction(ToolAction.take());
208 
209   // Create the compilers actual diagnostics engine.
210   Compiler.createDiagnostics();
211   if (!Compiler.hasDiagnostics())
212     return false;
213 
214   Compiler.createSourceManager(*Files);
215   addFileMappingsTo(Compiler.getSourceManager());
216 
217   const bool Success = Compiler.ExecuteAction(*ScopedToolAction);
218 
219   Compiler.resetAndLeakFileManager();
220   Files->clearStatCaches();
221   return Success;
222 }
223 
224 void ToolInvocation::addFileMappingsTo(SourceManager &Sources) {
225   for (llvm::StringMap<StringRef>::const_iterator
226            It = MappedFileContents.begin(), End = MappedFileContents.end();
227        It != End; ++It) {
228     // Inject the code as the given file name into the preprocessor options.
229     const llvm::MemoryBuffer *Input =
230         llvm::MemoryBuffer::getMemBuffer(It->getValue());
231     // FIXME: figure out what '0' stands for.
232     const FileEntry *FromFile = Files->getVirtualFile(
233         It->getKey(), Input->getBufferSize(), 0);
234     Sources.overrideFileContents(FromFile, Input);
235   }
236 }
237 
238 ClangTool::ClangTool(const CompilationDatabase &Compilations,
239                      ArrayRef<std::string> SourcePaths)
240     : Files((FileSystemOptions())) {
241   ArgsAdjusters.push_back(new ClangStripOutputAdjuster());
242   ArgsAdjusters.push_back(new ClangSyntaxOnlyAdjuster());
243   for (unsigned I = 0, E = SourcePaths.size(); I != E; ++I) {
244     SmallString<1024> File(getAbsolutePath(SourcePaths[I]));
245 
246     std::vector<CompileCommand> CompileCommandsForFile =
247       Compilations.getCompileCommands(File.str());
248     if (!CompileCommandsForFile.empty()) {
249       for (int I = 0, E = CompileCommandsForFile.size(); I != E; ++I) {
250         CompileCommands.push_back(std::make_pair(File.str(),
251                                   CompileCommandsForFile[I]));
252       }
253     } else {
254       // FIXME: There are two use cases here: doing a fuzzy
255       // "find . -name '*.cc' |xargs tool" match, where as a user I don't care
256       // about the .cc files that were not found, and the use case where I
257       // specify all files I want to run over explicitly, where this should
258       // be an error. We'll want to add an option for this.
259       llvm::outs() << "Skipping " << File << ". Command line not found.\n";
260     }
261   }
262 }
263 
264 void ClangTool::mapVirtualFile(StringRef FilePath, StringRef Content) {
265   MappedFileContents.push_back(std::make_pair(FilePath, Content));
266 }
267 
268 void ClangTool::setArgumentsAdjuster(ArgumentsAdjuster *Adjuster) {
269   clearArgumentsAdjusters();
270   appendArgumentsAdjuster(Adjuster);
271 }
272 
273 void ClangTool::appendArgumentsAdjuster(ArgumentsAdjuster *Adjuster) {
274   ArgsAdjusters.push_back(Adjuster);
275 }
276 
277 void ClangTool::clearArgumentsAdjusters() {
278   for (unsigned I = 0, E = ArgsAdjusters.size(); I != E; ++I)
279     delete ArgsAdjusters[I];
280   ArgsAdjusters.clear();
281 }
282 
283 int ClangTool::run(FrontendActionFactory *ActionFactory) {
284   // Exists solely for the purpose of lookup of the resource path.
285   // This just needs to be some symbol in the binary.
286   static int StaticSymbol;
287   // The driver detects the builtin header path based on the path of the
288   // executable.
289   // FIXME: On linux, GetMainExecutable is independent of the value of the
290   // first argument, thus allowing ClangTool and runToolOnCode to just
291   // pass in made-up names here. Make sure this works on other platforms.
292   std::string MainExecutable =
293     llvm::sys::Path::GetMainExecutable("clang_tool", &StaticSymbol).str();
294 
295   bool ProcessingFailed = false;
296   for (unsigned I = 0; I < CompileCommands.size(); ++I) {
297     std::string File = CompileCommands[I].first;
298     // FIXME: chdir is thread hostile; on the other hand, creating the same
299     // behavior as chdir is complex: chdir resolves the path once, thus
300     // guaranteeing that all subsequent relative path operations work
301     // on the same path the original chdir resulted in. This makes a difference
302     // for example on network filesystems, where symlinks might be switched
303     // during runtime of the tool. Fixing this depends on having a file system
304     // abstraction that allows openat() style interactions.
305     if (chdir(CompileCommands[I].second.Directory.c_str()))
306       llvm::report_fatal_error("Cannot chdir into \"" +
307                                CompileCommands[I].second.Directory + "\n!");
308     std::vector<std::string> CommandLine = CompileCommands[I].second.CommandLine;
309     for (unsigned I = 0, E = ArgsAdjusters.size(); I != E; ++I)
310       CommandLine = ArgsAdjusters[I]->Adjust(CommandLine);
311     assert(!CommandLine.empty());
312     CommandLine[0] = MainExecutable;
313     // FIXME: We need a callback mechanism for the tool writer to output a
314     // customized message for each file.
315     DEBUG({
316       llvm::dbgs() << "Processing: " << File << ".\n";
317     });
318     ToolInvocation Invocation(CommandLine, ActionFactory->create(), &Files);
319     for (int I = 0, E = MappedFileContents.size(); I != E; ++I) {
320       Invocation.mapVirtualFile(MappedFileContents[I].first,
321                                 MappedFileContents[I].second);
322     }
323     if (!Invocation.run()) {
324       // FIXME: Diagnostics should be used instead.
325       llvm::errs() << "Error while processing " << File << ".\n";
326       ProcessingFailed = true;
327     }
328   }
329   return ProcessingFailed ? 1 : 0;
330 }
331 
332 } // end namespace tooling
333 } // end namespace clang
334