1 //===--- JSONCompilationDatabase.cpp - ------------------------------------===// 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 contains the implementation of the JSONCompilationDatabase. 11 // 12 //===----------------------------------------------------------------------===// 13 14 #include "clang/Tooling/JSONCompilationDatabase.h" 15 #include "clang/Tooling/CompilationDatabase.h" 16 #include "clang/Tooling/CompilationDatabasePluginRegistry.h" 17 #include "clang/Tooling/Tooling.h" 18 #include "llvm/ADT/SmallString.h" 19 #include "llvm/Support/Allocator.h" 20 #include "llvm/Support/CommandLine.h" 21 #include "llvm/Support/Path.h" 22 #include "llvm/Support/StringSaver.h" 23 #include <system_error> 24 25 namespace clang { 26 namespace tooling { 27 28 namespace { 29 30 /// \brief A parser for escaped strings of command line arguments. 31 /// 32 /// Assumes \-escaping for quoted arguments (see the documentation of 33 /// unescapeCommandLine(...)). 34 class CommandLineArgumentParser { 35 public: 36 CommandLineArgumentParser(StringRef CommandLine) 37 : Input(CommandLine), Position(Input.begin()-1) {} 38 39 std::vector<std::string> parse() { 40 bool HasMoreInput = true; 41 while (HasMoreInput && nextNonWhitespace()) { 42 std::string Argument; 43 HasMoreInput = parseStringInto(Argument); 44 CommandLine.push_back(Argument); 45 } 46 return CommandLine; 47 } 48 49 private: 50 // All private methods return true if there is more input available. 51 52 bool parseStringInto(std::string &String) { 53 do { 54 if (*Position == '"') { 55 if (!parseDoubleQuotedStringInto(String)) return false; 56 } else if (*Position == '\'') { 57 if (!parseSingleQuotedStringInto(String)) return false; 58 } else { 59 if (!parseFreeStringInto(String)) return false; 60 } 61 } while (*Position != ' '); 62 return true; 63 } 64 65 bool parseDoubleQuotedStringInto(std::string &String) { 66 if (!next()) return false; 67 while (*Position != '"') { 68 if (!skipEscapeCharacter()) return false; 69 String.push_back(*Position); 70 if (!next()) return false; 71 } 72 return next(); 73 } 74 75 bool parseSingleQuotedStringInto(std::string &String) { 76 if (!next()) return false; 77 while (*Position != '\'') { 78 String.push_back(*Position); 79 if (!next()) return false; 80 } 81 return next(); 82 } 83 84 bool parseFreeStringInto(std::string &String) { 85 do { 86 if (!skipEscapeCharacter()) return false; 87 String.push_back(*Position); 88 if (!next()) return false; 89 } while (*Position != ' ' && *Position != '"' && *Position != '\''); 90 return true; 91 } 92 93 bool skipEscapeCharacter() { 94 if (*Position == '\\') { 95 return next(); 96 } 97 return true; 98 } 99 100 bool nextNonWhitespace() { 101 do { 102 if (!next()) return false; 103 } while (*Position == ' '); 104 return true; 105 } 106 107 bool next() { 108 ++Position; 109 return Position != Input.end(); 110 } 111 112 const StringRef Input; 113 StringRef::iterator Position; 114 std::vector<std::string> CommandLine; 115 }; 116 117 std::vector<std::string> unescapeCommandLine(JSONCommandLineSyntax Syntax, 118 StringRef EscapedCommandLine) { 119 if (Syntax == JSONCommandLineSyntax::AutoDetect) { 120 llvm::Triple Triple(llvm::sys::getProcessTriple()); 121 if (Triple.getOS() == llvm::Triple::OSType::Win32) { 122 // Assume Windows command line parsing on Win32 unless the triple 123 // explicitly 124 // tells us otherwise. 125 if (!Triple.hasEnvironment() || 126 Triple.getEnvironment() == llvm::Triple::EnvironmentType::MSVC) 127 Syntax = JSONCommandLineSyntax::Windows; 128 else 129 Syntax = JSONCommandLineSyntax::Gnu; 130 } 131 } 132 133 if (Syntax == JSONCommandLineSyntax::Windows) { 134 llvm::BumpPtrAllocator Alloc; 135 llvm::StringSaver Saver(Alloc); 136 llvm::SmallVector<const char *, 64> T; 137 llvm::cl::TokenizeWindowsCommandLine(EscapedCommandLine, Saver, T); 138 std::vector<std::string> Result(T.begin(), T.end()); 139 return Result; 140 } 141 assert(Syntax == JSONCommandLineSyntax::Gnu); 142 CommandLineArgumentParser parser(EscapedCommandLine); 143 return parser.parse(); 144 } 145 146 class JSONCompilationDatabasePlugin : public CompilationDatabasePlugin { 147 std::unique_ptr<CompilationDatabase> 148 loadFromDirectory(StringRef Directory, std::string &ErrorMessage) override { 149 SmallString<1024> JSONDatabasePath(Directory); 150 llvm::sys::path::append(JSONDatabasePath, "compile_commands.json"); 151 std::unique_ptr<CompilationDatabase> Database( 152 JSONCompilationDatabase::loadFromFile( 153 JSONDatabasePath, ErrorMessage, JSONCommandLineSyntax::AutoDetect)); 154 if (!Database) 155 return nullptr; 156 return Database; 157 } 158 }; 159 160 } // end namespace 161 162 // Register the JSONCompilationDatabasePlugin with the 163 // CompilationDatabasePluginRegistry using this statically initialized variable. 164 static CompilationDatabasePluginRegistry::Add<JSONCompilationDatabasePlugin> 165 X("json-compilation-database", "Reads JSON formatted compilation databases"); 166 167 // This anchor is used to force the linker to link in the generated object file 168 // and thus register the JSONCompilationDatabasePlugin. 169 volatile int JSONAnchorSource = 0; 170 171 std::unique_ptr<JSONCompilationDatabase> 172 JSONCompilationDatabase::loadFromFile(StringRef FilePath, 173 std::string &ErrorMessage, 174 JSONCommandLineSyntax Syntax) { 175 llvm::ErrorOr<std::unique_ptr<llvm::MemoryBuffer>> DatabaseBuffer = 176 llvm::MemoryBuffer::getFile(FilePath); 177 if (std::error_code Result = DatabaseBuffer.getError()) { 178 ErrorMessage = "Error while opening JSON database: " + Result.message(); 179 return nullptr; 180 } 181 std::unique_ptr<JSONCompilationDatabase> Database( 182 new JSONCompilationDatabase(std::move(*DatabaseBuffer), Syntax)); 183 if (!Database->parse(ErrorMessage)) 184 return nullptr; 185 return Database; 186 } 187 188 std::unique_ptr<JSONCompilationDatabase> 189 JSONCompilationDatabase::loadFromBuffer(StringRef DatabaseString, 190 std::string &ErrorMessage, 191 JSONCommandLineSyntax Syntax) { 192 std::unique_ptr<llvm::MemoryBuffer> DatabaseBuffer( 193 llvm::MemoryBuffer::getMemBuffer(DatabaseString)); 194 std::unique_ptr<JSONCompilationDatabase> Database( 195 new JSONCompilationDatabase(std::move(DatabaseBuffer), Syntax)); 196 if (!Database->parse(ErrorMessage)) 197 return nullptr; 198 return Database; 199 } 200 201 std::vector<CompileCommand> 202 JSONCompilationDatabase::getCompileCommands(StringRef FilePath) const { 203 SmallString<128> NativeFilePath; 204 llvm::sys::path::native(FilePath, NativeFilePath); 205 206 std::string Error; 207 llvm::raw_string_ostream ES(Error); 208 StringRef Match = MatchTrie.findEquivalent(NativeFilePath, ES); 209 if (Match.empty()) 210 return std::vector<CompileCommand>(); 211 llvm::StringMap< std::vector<CompileCommandRef> >::const_iterator 212 CommandsRefI = IndexByFile.find(Match); 213 if (CommandsRefI == IndexByFile.end()) 214 return std::vector<CompileCommand>(); 215 std::vector<CompileCommand> Commands; 216 getCommands(CommandsRefI->getValue(), Commands); 217 return Commands; 218 } 219 220 std::vector<std::string> 221 JSONCompilationDatabase::getAllFiles() const { 222 std::vector<std::string> Result; 223 224 llvm::StringMap< std::vector<CompileCommandRef> >::const_iterator 225 CommandsRefI = IndexByFile.begin(); 226 const llvm::StringMap< std::vector<CompileCommandRef> >::const_iterator 227 CommandsRefEnd = IndexByFile.end(); 228 for (; CommandsRefI != CommandsRefEnd; ++CommandsRefI) { 229 Result.push_back(CommandsRefI->first().str()); 230 } 231 232 return Result; 233 } 234 235 std::vector<CompileCommand> 236 JSONCompilationDatabase::getAllCompileCommands() const { 237 std::vector<CompileCommand> Commands; 238 getCommands(AllCommands, Commands); 239 return Commands; 240 } 241 242 static std::vector<std::string> 243 nodeToCommandLine(JSONCommandLineSyntax Syntax, 244 const std::vector<llvm::yaml::ScalarNode *> &Nodes) { 245 SmallString<1024> Storage; 246 if (Nodes.size() == 1) { 247 return unescapeCommandLine(Syntax, Nodes[0]->getValue(Storage)); 248 } 249 std::vector<std::string> Arguments; 250 for (auto *Node : Nodes) { 251 Arguments.push_back(Node->getValue(Storage)); 252 } 253 return Arguments; 254 } 255 256 void JSONCompilationDatabase::getCommands( 257 ArrayRef<CompileCommandRef> CommandsRef, 258 std::vector<CompileCommand> &Commands) const { 259 for (int I = 0, E = CommandsRef.size(); I != E; ++I) { 260 SmallString<8> DirectoryStorage; 261 SmallString<32> FilenameStorage; 262 Commands.emplace_back( 263 std::get<0>(CommandsRef[I])->getValue(DirectoryStorage), 264 std::get<1>(CommandsRef[I])->getValue(FilenameStorage), 265 nodeToCommandLine(Syntax, std::get<2>(CommandsRef[I]))); 266 } 267 } 268 269 bool JSONCompilationDatabase::parse(std::string &ErrorMessage) { 270 llvm::yaml::document_iterator I = YAMLStream.begin(); 271 if (I == YAMLStream.end()) { 272 ErrorMessage = "Error while parsing YAML."; 273 return false; 274 } 275 llvm::yaml::Node *Root = I->getRoot(); 276 if (!Root) { 277 ErrorMessage = "Error while parsing YAML."; 278 return false; 279 } 280 llvm::yaml::SequenceNode *Array = dyn_cast<llvm::yaml::SequenceNode>(Root); 281 if (!Array) { 282 ErrorMessage = "Expected array."; 283 return false; 284 } 285 for (auto& NextObject : *Array) { 286 llvm::yaml::MappingNode *Object = dyn_cast<llvm::yaml::MappingNode>(&NextObject); 287 if (!Object) { 288 ErrorMessage = "Expected object."; 289 return false; 290 } 291 llvm::yaml::ScalarNode *Directory = nullptr; 292 llvm::Optional<std::vector<llvm::yaml::ScalarNode *>> Command; 293 llvm::yaml::ScalarNode *File = nullptr; 294 for (auto& NextKeyValue : *Object) { 295 llvm::yaml::ScalarNode *KeyString = 296 dyn_cast<llvm::yaml::ScalarNode>(NextKeyValue.getKey()); 297 if (!KeyString) { 298 ErrorMessage = "Expected strings as key."; 299 return false; 300 } 301 SmallString<10> KeyStorage; 302 StringRef KeyValue = KeyString->getValue(KeyStorage); 303 llvm::yaml::Node *Value = NextKeyValue.getValue(); 304 if (!Value) { 305 ErrorMessage = "Expected value."; 306 return false; 307 } 308 llvm::yaml::ScalarNode *ValueString = 309 dyn_cast<llvm::yaml::ScalarNode>(Value); 310 llvm::yaml::SequenceNode *SequenceString = 311 dyn_cast<llvm::yaml::SequenceNode>(Value); 312 if (KeyValue == "arguments" && !SequenceString) { 313 ErrorMessage = "Expected sequence as value."; 314 return false; 315 } else if (KeyValue != "arguments" && !ValueString) { 316 ErrorMessage = "Expected string as value."; 317 return false; 318 } 319 if (KeyValue == "directory") { 320 Directory = ValueString; 321 } else if (KeyValue == "arguments") { 322 Command = std::vector<llvm::yaml::ScalarNode *>(); 323 for (auto &Argument : *SequenceString) { 324 auto Scalar = dyn_cast<llvm::yaml::ScalarNode>(&Argument); 325 if (!Scalar) { 326 ErrorMessage = "Only strings are allowed in 'arguments'."; 327 return false; 328 } 329 Command->push_back(Scalar); 330 } 331 } else if (KeyValue == "command") { 332 if (!Command) 333 Command = std::vector<llvm::yaml::ScalarNode *>(1, ValueString); 334 } else if (KeyValue == "file") { 335 File = ValueString; 336 } else { 337 ErrorMessage = ("Unknown key: \"" + 338 KeyString->getRawValue() + "\"").str(); 339 return false; 340 } 341 } 342 if (!File) { 343 ErrorMessage = "Missing key: \"file\"."; 344 return false; 345 } 346 if (!Command) { 347 ErrorMessage = "Missing key: \"command\" or \"arguments\"."; 348 return false; 349 } 350 if (!Directory) { 351 ErrorMessage = "Missing key: \"directory\"."; 352 return false; 353 } 354 SmallString<8> FileStorage; 355 StringRef FileName = File->getValue(FileStorage); 356 SmallString<128> NativeFilePath; 357 if (llvm::sys::path::is_relative(FileName)) { 358 SmallString<8> DirectoryStorage; 359 SmallString<128> AbsolutePath( 360 Directory->getValue(DirectoryStorage)); 361 llvm::sys::path::append(AbsolutePath, FileName); 362 llvm::sys::path::native(AbsolutePath, NativeFilePath); 363 } else { 364 llvm::sys::path::native(FileName, NativeFilePath); 365 } 366 auto Cmd = CompileCommandRef(Directory, File, *Command); 367 IndexByFile[NativeFilePath].push_back(Cmd); 368 AllCommands.push_back(Cmd); 369 MatchTrie.insert(NativeFilePath); 370 } 371 return true; 372 } 373 374 } // end namespace tooling 375 } // end namespace clang 376