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 Syntax = JSONCommandLineSyntax::Gnu; 121 llvm::Triple Triple(llvm::sys::getProcessTriple()); 122 if (Triple.getOS() == llvm::Triple::OSType::Win32) { 123 // Assume Windows command line parsing on Win32 unless the triple 124 // explicitly tells us otherwise. 125 if (!Triple.hasEnvironment() || 126 Triple.getEnvironment() == llvm::Triple::EnvironmentType::MSVC) 127 Syntax = JSONCommandLineSyntax::Windows; 128 } 129 } 130 131 if (Syntax == JSONCommandLineSyntax::Windows) { 132 llvm::BumpPtrAllocator Alloc; 133 llvm::StringSaver Saver(Alloc); 134 llvm::SmallVector<const char *, 64> T; 135 llvm::cl::TokenizeWindowsCommandLine(EscapedCommandLine, Saver, T); 136 std::vector<std::string> Result(T.begin(), T.end()); 137 return Result; 138 } 139 assert(Syntax == JSONCommandLineSyntax::Gnu); 140 CommandLineArgumentParser parser(EscapedCommandLine); 141 return parser.parse(); 142 } 143 144 class JSONCompilationDatabasePlugin : public CompilationDatabasePlugin { 145 std::unique_ptr<CompilationDatabase> 146 loadFromDirectory(StringRef Directory, std::string &ErrorMessage) override { 147 SmallString<1024> JSONDatabasePath(Directory); 148 llvm::sys::path::append(JSONDatabasePath, "compile_commands.json"); 149 std::unique_ptr<CompilationDatabase> Database( 150 JSONCompilationDatabase::loadFromFile( 151 JSONDatabasePath, ErrorMessage, JSONCommandLineSyntax::AutoDetect)); 152 if (!Database) 153 return nullptr; 154 return Database; 155 } 156 }; 157 158 } // end namespace 159 160 // Register the JSONCompilationDatabasePlugin with the 161 // CompilationDatabasePluginRegistry using this statically initialized variable. 162 static CompilationDatabasePluginRegistry::Add<JSONCompilationDatabasePlugin> 163 X("json-compilation-database", "Reads JSON formatted compilation databases"); 164 165 // This anchor is used to force the linker to link in the generated object file 166 // and thus register the JSONCompilationDatabasePlugin. 167 volatile int JSONAnchorSource = 0; 168 169 std::unique_ptr<JSONCompilationDatabase> 170 JSONCompilationDatabase::loadFromFile(StringRef FilePath, 171 std::string &ErrorMessage, 172 JSONCommandLineSyntax Syntax) { 173 llvm::ErrorOr<std::unique_ptr<llvm::MemoryBuffer>> DatabaseBuffer = 174 llvm::MemoryBuffer::getFile(FilePath); 175 if (std::error_code Result = DatabaseBuffer.getError()) { 176 ErrorMessage = "Error while opening JSON database: " + Result.message(); 177 return nullptr; 178 } 179 std::unique_ptr<JSONCompilationDatabase> Database( 180 new JSONCompilationDatabase(std::move(*DatabaseBuffer), Syntax)); 181 if (!Database->parse(ErrorMessage)) 182 return nullptr; 183 return Database; 184 } 185 186 std::unique_ptr<JSONCompilationDatabase> 187 JSONCompilationDatabase::loadFromBuffer(StringRef DatabaseString, 188 std::string &ErrorMessage, 189 JSONCommandLineSyntax Syntax) { 190 std::unique_ptr<llvm::MemoryBuffer> DatabaseBuffer( 191 llvm::MemoryBuffer::getMemBuffer(DatabaseString)); 192 std::unique_ptr<JSONCompilationDatabase> Database( 193 new JSONCompilationDatabase(std::move(DatabaseBuffer), Syntax)); 194 if (!Database->parse(ErrorMessage)) 195 return nullptr; 196 return Database; 197 } 198 199 std::vector<CompileCommand> 200 JSONCompilationDatabase::getCompileCommands(StringRef FilePath) const { 201 SmallString<128> NativeFilePath; 202 llvm::sys::path::native(FilePath, NativeFilePath); 203 204 std::string Error; 205 llvm::raw_string_ostream ES(Error); 206 StringRef Match = MatchTrie.findEquivalent(NativeFilePath, ES); 207 if (Match.empty()) 208 return std::vector<CompileCommand>(); 209 llvm::StringMap< std::vector<CompileCommandRef> >::const_iterator 210 CommandsRefI = IndexByFile.find(Match); 211 if (CommandsRefI == IndexByFile.end()) 212 return std::vector<CompileCommand>(); 213 std::vector<CompileCommand> Commands; 214 getCommands(CommandsRefI->getValue(), Commands); 215 return Commands; 216 } 217 218 std::vector<std::string> 219 JSONCompilationDatabase::getAllFiles() const { 220 std::vector<std::string> Result; 221 222 llvm::StringMap< std::vector<CompileCommandRef> >::const_iterator 223 CommandsRefI = IndexByFile.begin(); 224 const llvm::StringMap< std::vector<CompileCommandRef> >::const_iterator 225 CommandsRefEnd = IndexByFile.end(); 226 for (; CommandsRefI != CommandsRefEnd; ++CommandsRefI) { 227 Result.push_back(CommandsRefI->first().str()); 228 } 229 230 return Result; 231 } 232 233 std::vector<CompileCommand> 234 JSONCompilationDatabase::getAllCompileCommands() const { 235 std::vector<CompileCommand> Commands; 236 getCommands(AllCommands, Commands); 237 return Commands; 238 } 239 240 static std::vector<std::string> 241 nodeToCommandLine(JSONCommandLineSyntax Syntax, 242 const std::vector<llvm::yaml::ScalarNode *> &Nodes) { 243 SmallString<1024> Storage; 244 if (Nodes.size() == 1) { 245 return unescapeCommandLine(Syntax, Nodes[0]->getValue(Storage)); 246 } 247 std::vector<std::string> Arguments; 248 for (auto *Node : Nodes) { 249 Arguments.push_back(Node->getValue(Storage)); 250 } 251 return Arguments; 252 } 253 254 void JSONCompilationDatabase::getCommands( 255 ArrayRef<CompileCommandRef> CommandsRef, 256 std::vector<CompileCommand> &Commands) const { 257 for (int I = 0, E = CommandsRef.size(); I != E; ++I) { 258 SmallString<8> DirectoryStorage; 259 SmallString<32> FilenameStorage; 260 Commands.emplace_back( 261 std::get<0>(CommandsRef[I])->getValue(DirectoryStorage), 262 std::get<1>(CommandsRef[I])->getValue(FilenameStorage), 263 nodeToCommandLine(Syntax, std::get<2>(CommandsRef[I]))); 264 } 265 } 266 267 bool JSONCompilationDatabase::parse(std::string &ErrorMessage) { 268 llvm::yaml::document_iterator I = YAMLStream.begin(); 269 if (I == YAMLStream.end()) { 270 ErrorMessage = "Error while parsing YAML."; 271 return false; 272 } 273 llvm::yaml::Node *Root = I->getRoot(); 274 if (!Root) { 275 ErrorMessage = "Error while parsing YAML."; 276 return false; 277 } 278 llvm::yaml::SequenceNode *Array = dyn_cast<llvm::yaml::SequenceNode>(Root); 279 if (!Array) { 280 ErrorMessage = "Expected array."; 281 return false; 282 } 283 for (auto& NextObject : *Array) { 284 llvm::yaml::MappingNode *Object = dyn_cast<llvm::yaml::MappingNode>(&NextObject); 285 if (!Object) { 286 ErrorMessage = "Expected object."; 287 return false; 288 } 289 llvm::yaml::ScalarNode *Directory = nullptr; 290 llvm::Optional<std::vector<llvm::yaml::ScalarNode *>> Command; 291 llvm::yaml::ScalarNode *File = nullptr; 292 for (auto& NextKeyValue : *Object) { 293 llvm::yaml::ScalarNode *KeyString = 294 dyn_cast<llvm::yaml::ScalarNode>(NextKeyValue.getKey()); 295 if (!KeyString) { 296 ErrorMessage = "Expected strings as key."; 297 return false; 298 } 299 SmallString<10> KeyStorage; 300 StringRef KeyValue = KeyString->getValue(KeyStorage); 301 llvm::yaml::Node *Value = NextKeyValue.getValue(); 302 if (!Value) { 303 ErrorMessage = "Expected value."; 304 return false; 305 } 306 llvm::yaml::ScalarNode *ValueString = 307 dyn_cast<llvm::yaml::ScalarNode>(Value); 308 llvm::yaml::SequenceNode *SequenceString = 309 dyn_cast<llvm::yaml::SequenceNode>(Value); 310 if (KeyValue == "arguments" && !SequenceString) { 311 ErrorMessage = "Expected sequence as value."; 312 return false; 313 } else if (KeyValue != "arguments" && !ValueString) { 314 ErrorMessage = "Expected string as value."; 315 return false; 316 } 317 if (KeyValue == "directory") { 318 Directory = ValueString; 319 } else if (KeyValue == "arguments") { 320 Command = std::vector<llvm::yaml::ScalarNode *>(); 321 for (auto &Argument : *SequenceString) { 322 auto Scalar = dyn_cast<llvm::yaml::ScalarNode>(&Argument); 323 if (!Scalar) { 324 ErrorMessage = "Only strings are allowed in 'arguments'."; 325 return false; 326 } 327 Command->push_back(Scalar); 328 } 329 } else if (KeyValue == "command") { 330 if (!Command) 331 Command = std::vector<llvm::yaml::ScalarNode *>(1, ValueString); 332 } else if (KeyValue == "file") { 333 File = ValueString; 334 } else { 335 ErrorMessage = ("Unknown key: \"" + 336 KeyString->getRawValue() + "\"").str(); 337 return false; 338 } 339 } 340 if (!File) { 341 ErrorMessage = "Missing key: \"file\"."; 342 return false; 343 } 344 if (!Command) { 345 ErrorMessage = "Missing key: \"command\" or \"arguments\"."; 346 return false; 347 } 348 if (!Directory) { 349 ErrorMessage = "Missing key: \"directory\"."; 350 return false; 351 } 352 SmallString<8> FileStorage; 353 StringRef FileName = File->getValue(FileStorage); 354 SmallString<128> NativeFilePath; 355 if (llvm::sys::path::is_relative(FileName)) { 356 SmallString<8> DirectoryStorage; 357 SmallString<128> AbsolutePath( 358 Directory->getValue(DirectoryStorage)); 359 llvm::sys::path::append(AbsolutePath, FileName); 360 llvm::sys::path::native(AbsolutePath, NativeFilePath); 361 } else { 362 llvm::sys::path::native(FileName, NativeFilePath); 363 } 364 auto Cmd = CompileCommandRef(Directory, File, *Command); 365 IndexByFile[NativeFilePath].push_back(Cmd); 366 AllCommands.push_back(Cmd); 367 MatchTrie.insert(NativeFilePath); 368 } 369 return true; 370 } 371 372 } // end namespace tooling 373 } // end namespace clang 374