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( 118 StringRef EscapedCommandLine) { 119 llvm::Triple Triple(llvm::sys::getProcessTriple()); 120 if (Triple.getOS() == llvm::Triple::OSType::Win32) { 121 // Assume Windows command line parsing on Win32 unless the triple explicitly 122 // tells us otherwise. 123 if (!Triple.hasEnvironment() || 124 Triple.getEnvironment() == llvm::Triple::EnvironmentType::MSVC) { 125 llvm::BumpPtrAllocator Alloc; 126 llvm::StringSaver Saver(Alloc); 127 llvm::SmallVector<const char *, 64> T; 128 llvm::cl::TokenizeWindowsCommandLine(EscapedCommandLine, Saver, T); 129 std::vector<std::string> Result(T.begin(), T.end()); 130 return Result; 131 } 132 } 133 134 CommandLineArgumentParser parser(EscapedCommandLine); 135 return parser.parse(); 136 } 137 138 class JSONCompilationDatabasePlugin : public CompilationDatabasePlugin { 139 std::unique_ptr<CompilationDatabase> 140 loadFromDirectory(StringRef Directory, std::string &ErrorMessage) override { 141 SmallString<1024> JSONDatabasePath(Directory); 142 llvm::sys::path::append(JSONDatabasePath, "compile_commands.json"); 143 std::unique_ptr<CompilationDatabase> Database( 144 JSONCompilationDatabase::loadFromFile(JSONDatabasePath, ErrorMessage)); 145 if (!Database) 146 return nullptr; 147 return Database; 148 } 149 }; 150 151 } // end namespace 152 153 // Register the JSONCompilationDatabasePlugin with the 154 // CompilationDatabasePluginRegistry using this statically initialized variable. 155 static CompilationDatabasePluginRegistry::Add<JSONCompilationDatabasePlugin> 156 X("json-compilation-database", "Reads JSON formatted compilation databases"); 157 158 // This anchor is used to force the linker to link in the generated object file 159 // and thus register the JSONCompilationDatabasePlugin. 160 volatile int JSONAnchorSource = 0; 161 162 std::unique_ptr<JSONCompilationDatabase> 163 JSONCompilationDatabase::loadFromFile(StringRef FilePath, 164 std::string &ErrorMessage) { 165 llvm::ErrorOr<std::unique_ptr<llvm::MemoryBuffer>> DatabaseBuffer = 166 llvm::MemoryBuffer::getFile(FilePath); 167 if (std::error_code Result = DatabaseBuffer.getError()) { 168 ErrorMessage = "Error while opening JSON database: " + Result.message(); 169 return nullptr; 170 } 171 std::unique_ptr<JSONCompilationDatabase> Database( 172 new JSONCompilationDatabase(std::move(*DatabaseBuffer))); 173 if (!Database->parse(ErrorMessage)) 174 return nullptr; 175 return Database; 176 } 177 178 std::unique_ptr<JSONCompilationDatabase> 179 JSONCompilationDatabase::loadFromBuffer(StringRef DatabaseString, 180 std::string &ErrorMessage) { 181 std::unique_ptr<llvm::MemoryBuffer> DatabaseBuffer( 182 llvm::MemoryBuffer::getMemBuffer(DatabaseString)); 183 std::unique_ptr<JSONCompilationDatabase> Database( 184 new JSONCompilationDatabase(std::move(DatabaseBuffer))); 185 if (!Database->parse(ErrorMessage)) 186 return nullptr; 187 return Database; 188 } 189 190 std::vector<CompileCommand> 191 JSONCompilationDatabase::getCompileCommands(StringRef FilePath) const { 192 SmallString<128> NativeFilePath; 193 llvm::sys::path::native(FilePath, NativeFilePath); 194 195 std::string Error; 196 llvm::raw_string_ostream ES(Error); 197 StringRef Match = MatchTrie.findEquivalent(NativeFilePath, ES); 198 if (Match.empty()) 199 return std::vector<CompileCommand>(); 200 llvm::StringMap< std::vector<CompileCommandRef> >::const_iterator 201 CommandsRefI = IndexByFile.find(Match); 202 if (CommandsRefI == IndexByFile.end()) 203 return std::vector<CompileCommand>(); 204 std::vector<CompileCommand> Commands; 205 getCommands(CommandsRefI->getValue(), Commands); 206 return Commands; 207 } 208 209 std::vector<std::string> 210 JSONCompilationDatabase::getAllFiles() const { 211 std::vector<std::string> Result; 212 213 llvm::StringMap< std::vector<CompileCommandRef> >::const_iterator 214 CommandsRefI = IndexByFile.begin(); 215 const llvm::StringMap< std::vector<CompileCommandRef> >::const_iterator 216 CommandsRefEnd = IndexByFile.end(); 217 for (; CommandsRefI != CommandsRefEnd; ++CommandsRefI) { 218 Result.push_back(CommandsRefI->first().str()); 219 } 220 221 return Result; 222 } 223 224 std::vector<CompileCommand> 225 JSONCompilationDatabase::getAllCompileCommands() const { 226 std::vector<CompileCommand> Commands; 227 getCommands(AllCommands, Commands); 228 return Commands; 229 } 230 231 static std::vector<std::string> 232 nodeToCommandLine(const std::vector<llvm::yaml::ScalarNode *> &Nodes) { 233 SmallString<1024> Storage; 234 if (Nodes.size() == 1) { 235 return unescapeCommandLine(Nodes[0]->getValue(Storage)); 236 } 237 std::vector<std::string> Arguments; 238 for (auto *Node : Nodes) { 239 Arguments.push_back(Node->getValue(Storage)); 240 } 241 return Arguments; 242 } 243 244 void JSONCompilationDatabase::getCommands( 245 ArrayRef<CompileCommandRef> CommandsRef, 246 std::vector<CompileCommand> &Commands) const { 247 for (int I = 0, E = CommandsRef.size(); I != E; ++I) { 248 SmallString<8> DirectoryStorage; 249 SmallString<32> FilenameStorage; 250 Commands.emplace_back( 251 std::get<0>(CommandsRef[I])->getValue(DirectoryStorage), 252 std::get<1>(CommandsRef[I])->getValue(FilenameStorage), 253 nodeToCommandLine(std::get<2>(CommandsRef[I]))); 254 } 255 } 256 257 bool JSONCompilationDatabase::parse(std::string &ErrorMessage) { 258 llvm::yaml::document_iterator I = YAMLStream.begin(); 259 if (I == YAMLStream.end()) { 260 ErrorMessage = "Error while parsing YAML."; 261 return false; 262 } 263 llvm::yaml::Node *Root = I->getRoot(); 264 if (!Root) { 265 ErrorMessage = "Error while parsing YAML."; 266 return false; 267 } 268 llvm::yaml::SequenceNode *Array = dyn_cast<llvm::yaml::SequenceNode>(Root); 269 if (!Array) { 270 ErrorMessage = "Expected array."; 271 return false; 272 } 273 for (auto& NextObject : *Array) { 274 llvm::yaml::MappingNode *Object = dyn_cast<llvm::yaml::MappingNode>(&NextObject); 275 if (!Object) { 276 ErrorMessage = "Expected object."; 277 return false; 278 } 279 llvm::yaml::ScalarNode *Directory = nullptr; 280 llvm::Optional<std::vector<llvm::yaml::ScalarNode *>> Command; 281 llvm::yaml::ScalarNode *File = nullptr; 282 for (auto& NextKeyValue : *Object) { 283 llvm::yaml::ScalarNode *KeyString = 284 dyn_cast<llvm::yaml::ScalarNode>(NextKeyValue.getKey()); 285 if (!KeyString) { 286 ErrorMessage = "Expected strings as key."; 287 return false; 288 } 289 SmallString<10> KeyStorage; 290 StringRef KeyValue = KeyString->getValue(KeyStorage); 291 llvm::yaml::Node *Value = NextKeyValue.getValue(); 292 if (!Value) { 293 ErrorMessage = "Expected value."; 294 return false; 295 } 296 llvm::yaml::ScalarNode *ValueString = 297 dyn_cast<llvm::yaml::ScalarNode>(Value); 298 llvm::yaml::SequenceNode *SequenceString = 299 dyn_cast<llvm::yaml::SequenceNode>(Value); 300 if (KeyValue == "arguments" && !SequenceString) { 301 ErrorMessage = "Expected sequence as value."; 302 return false; 303 } else if (KeyValue != "arguments" && !ValueString) { 304 ErrorMessage = "Expected string as value."; 305 return false; 306 } 307 if (KeyValue == "directory") { 308 Directory = ValueString; 309 } else if (KeyValue == "arguments") { 310 Command = std::vector<llvm::yaml::ScalarNode *>(); 311 for (auto &Argument : *SequenceString) { 312 auto Scalar = dyn_cast<llvm::yaml::ScalarNode>(&Argument); 313 if (!Scalar) { 314 ErrorMessage = "Only strings are allowed in 'arguments'."; 315 return false; 316 } 317 Command->push_back(Scalar); 318 } 319 } else if (KeyValue == "command") { 320 if (!Command) 321 Command = std::vector<llvm::yaml::ScalarNode *>(1, ValueString); 322 } else if (KeyValue == "file") { 323 File = ValueString; 324 } else { 325 ErrorMessage = ("Unknown key: \"" + 326 KeyString->getRawValue() + "\"").str(); 327 return false; 328 } 329 } 330 if (!File) { 331 ErrorMessage = "Missing key: \"file\"."; 332 return false; 333 } 334 if (!Command) { 335 ErrorMessage = "Missing key: \"command\" or \"arguments\"."; 336 return false; 337 } 338 if (!Directory) { 339 ErrorMessage = "Missing key: \"directory\"."; 340 return false; 341 } 342 SmallString<8> FileStorage; 343 StringRef FileName = File->getValue(FileStorage); 344 SmallString<128> NativeFilePath; 345 if (llvm::sys::path::is_relative(FileName)) { 346 SmallString<8> DirectoryStorage; 347 SmallString<128> AbsolutePath( 348 Directory->getValue(DirectoryStorage)); 349 llvm::sys::path::append(AbsolutePath, FileName); 350 llvm::sys::path::native(AbsolutePath, NativeFilePath); 351 } else { 352 llvm::sys::path::native(FileName, NativeFilePath); 353 } 354 auto Cmd = CompileCommandRef(Directory, File, *Command); 355 IndexByFile[NativeFilePath].push_back(Cmd); 356 AllCommands.push_back(Cmd); 357 MatchTrie.insert(NativeFilePath); 358 } 359 return true; 360 } 361 362 } // end namespace tooling 363 } // end namespace clang 364