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/Basic/LLVM.h" 16 #include "clang/Tooling/CompilationDatabase.h" 17 #include "clang/Tooling/CompilationDatabasePluginRegistry.h" 18 #include "llvm/ADT/Optional.h" 19 #include "llvm/ADT/SmallString.h" 20 #include "llvm/ADT/SmallVector.h" 21 #include "llvm/ADT/StringRef.h" 22 #include "llvm/ADT/Triple.h" 23 #include "llvm/Support/Allocator.h" 24 #include "llvm/Support/Casting.h" 25 #include "llvm/Support/CommandLine.h" 26 #include "llvm/Support/ErrorOr.h" 27 #include "llvm/Support/Host.h" 28 #include "llvm/Support/MemoryBuffer.h" 29 #include "llvm/Support/Path.h" 30 #include "llvm/Support/StringSaver.h" 31 #include "llvm/Support/YAMLParser.h" 32 #include "llvm/Support/raw_ostream.h" 33 #include <cassert> 34 #include <memory> 35 #include <string> 36 #include <system_error> 37 #include <tuple> 38 #include <utility> 39 #include <vector> 40 41 using namespace clang; 42 using namespace tooling; 43 44 namespace { 45 46 /// \brief A parser for escaped strings of command line arguments. 47 /// 48 /// Assumes \-escaping for quoted arguments (see the documentation of 49 /// unescapeCommandLine(...)). 50 class CommandLineArgumentParser { 51 public: 52 CommandLineArgumentParser(StringRef CommandLine) 53 : Input(CommandLine), Position(Input.begin()-1) {} 54 55 std::vector<std::string> parse() { 56 bool HasMoreInput = true; 57 while (HasMoreInput && nextNonWhitespace()) { 58 std::string Argument; 59 HasMoreInput = parseStringInto(Argument); 60 CommandLine.push_back(Argument); 61 } 62 return CommandLine; 63 } 64 65 private: 66 // All private methods return true if there is more input available. 67 68 bool parseStringInto(std::string &String) { 69 do { 70 if (*Position == '"') { 71 if (!parseDoubleQuotedStringInto(String)) return false; 72 } else if (*Position == '\'') { 73 if (!parseSingleQuotedStringInto(String)) return false; 74 } else { 75 if (!parseFreeStringInto(String)) return false; 76 } 77 } while (*Position != ' '); 78 return true; 79 } 80 81 bool parseDoubleQuotedStringInto(std::string &String) { 82 if (!next()) return false; 83 while (*Position != '"') { 84 if (!skipEscapeCharacter()) return false; 85 String.push_back(*Position); 86 if (!next()) return false; 87 } 88 return next(); 89 } 90 91 bool parseSingleQuotedStringInto(std::string &String) { 92 if (!next()) return false; 93 while (*Position != '\'') { 94 String.push_back(*Position); 95 if (!next()) return false; 96 } 97 return next(); 98 } 99 100 bool parseFreeStringInto(std::string &String) { 101 do { 102 if (!skipEscapeCharacter()) return false; 103 String.push_back(*Position); 104 if (!next()) return false; 105 } while (*Position != ' ' && *Position != '"' && *Position != '\''); 106 return true; 107 } 108 109 bool skipEscapeCharacter() { 110 if (*Position == '\\') { 111 return next(); 112 } 113 return true; 114 } 115 116 bool nextNonWhitespace() { 117 do { 118 if (!next()) return false; 119 } while (*Position == ' '); 120 return true; 121 } 122 123 bool next() { 124 ++Position; 125 return Position != Input.end(); 126 } 127 128 const StringRef Input; 129 StringRef::iterator Position; 130 std::vector<std::string> CommandLine; 131 }; 132 133 std::vector<std::string> unescapeCommandLine(JSONCommandLineSyntax Syntax, 134 StringRef EscapedCommandLine) { 135 if (Syntax == JSONCommandLineSyntax::AutoDetect) { 136 Syntax = JSONCommandLineSyntax::Gnu; 137 llvm::Triple Triple(llvm::sys::getProcessTriple()); 138 if (Triple.getOS() == llvm::Triple::OSType::Win32) { 139 // Assume Windows command line parsing on Win32 unless the triple 140 // explicitly tells us otherwise. 141 if (!Triple.hasEnvironment() || 142 Triple.getEnvironment() == llvm::Triple::EnvironmentType::MSVC) 143 Syntax = JSONCommandLineSyntax::Windows; 144 } 145 } 146 147 if (Syntax == JSONCommandLineSyntax::Windows) { 148 llvm::BumpPtrAllocator Alloc; 149 llvm::StringSaver Saver(Alloc); 150 llvm::SmallVector<const char *, 64> T; 151 llvm::cl::TokenizeWindowsCommandLine(EscapedCommandLine, Saver, T); 152 std::vector<std::string> Result(T.begin(), T.end()); 153 return Result; 154 } 155 assert(Syntax == JSONCommandLineSyntax::Gnu); 156 CommandLineArgumentParser parser(EscapedCommandLine); 157 return parser.parse(); 158 } 159 160 class JSONCompilationDatabasePlugin : public CompilationDatabasePlugin { 161 std::unique_ptr<CompilationDatabase> 162 loadFromDirectory(StringRef Directory, std::string &ErrorMessage) override { 163 SmallString<1024> JSONDatabasePath(Directory); 164 llvm::sys::path::append(JSONDatabasePath, "compile_commands.json"); 165 return JSONCompilationDatabase::loadFromFile( 166 JSONDatabasePath, ErrorMessage, JSONCommandLineSyntax::AutoDetect); 167 } 168 }; 169 170 } // namespace 171 172 // Register the JSONCompilationDatabasePlugin with the 173 // CompilationDatabasePluginRegistry using this statically initialized variable. 174 static CompilationDatabasePluginRegistry::Add<JSONCompilationDatabasePlugin> 175 X("json-compilation-database", "Reads JSON formatted compilation databases"); 176 177 namespace clang { 178 namespace tooling { 179 180 // This anchor is used to force the linker to link in the generated object file 181 // and thus register the JSONCompilationDatabasePlugin. 182 volatile int JSONAnchorSource = 0; 183 184 } // namespace tooling 185 } // namespace clang 186 187 std::unique_ptr<JSONCompilationDatabase> 188 JSONCompilationDatabase::loadFromFile(StringRef FilePath, 189 std::string &ErrorMessage, 190 JSONCommandLineSyntax Syntax) { 191 llvm::ErrorOr<std::unique_ptr<llvm::MemoryBuffer>> DatabaseBuffer = 192 llvm::MemoryBuffer::getFile(FilePath); 193 if (std::error_code Result = DatabaseBuffer.getError()) { 194 ErrorMessage = "Error while opening JSON database: " + Result.message(); 195 return nullptr; 196 } 197 std::unique_ptr<JSONCompilationDatabase> Database( 198 new JSONCompilationDatabase(std::move(*DatabaseBuffer), Syntax)); 199 if (!Database->parse(ErrorMessage)) 200 return nullptr; 201 return Database; 202 } 203 204 std::unique_ptr<JSONCompilationDatabase> 205 JSONCompilationDatabase::loadFromBuffer(StringRef DatabaseString, 206 std::string &ErrorMessage, 207 JSONCommandLineSyntax Syntax) { 208 std::unique_ptr<llvm::MemoryBuffer> DatabaseBuffer( 209 llvm::MemoryBuffer::getMemBuffer(DatabaseString)); 210 std::unique_ptr<JSONCompilationDatabase> Database( 211 new JSONCompilationDatabase(std::move(DatabaseBuffer), Syntax)); 212 if (!Database->parse(ErrorMessage)) 213 return nullptr; 214 return Database; 215 } 216 217 std::vector<CompileCommand> 218 JSONCompilationDatabase::getCompileCommands(StringRef FilePath) const { 219 SmallString<128> NativeFilePath; 220 llvm::sys::path::native(FilePath, NativeFilePath); 221 222 std::string Error; 223 llvm::raw_string_ostream ES(Error); 224 StringRef Match = MatchTrie.findEquivalent(NativeFilePath, ES); 225 if (Match.empty()) 226 return {}; 227 const auto CommandsRefI = IndexByFile.find(Match); 228 if (CommandsRefI == IndexByFile.end()) 229 return {}; 230 std::vector<CompileCommand> Commands; 231 getCommands(CommandsRefI->getValue(), Commands); 232 return Commands; 233 } 234 235 std::vector<std::string> 236 JSONCompilationDatabase::getAllFiles() const { 237 std::vector<std::string> Result; 238 for (const auto &CommandRef : IndexByFile) 239 Result.push_back(CommandRef.first().str()); 240 return Result; 241 } 242 243 std::vector<CompileCommand> 244 JSONCompilationDatabase::getAllCompileCommands() const { 245 std::vector<CompileCommand> Commands; 246 getCommands(AllCommands, Commands); 247 return Commands; 248 } 249 250 static std::vector<std::string> 251 nodeToCommandLine(JSONCommandLineSyntax Syntax, 252 const std::vector<llvm::yaml::ScalarNode *> &Nodes) { 253 SmallString<1024> Storage; 254 if (Nodes.size() == 1) 255 return unescapeCommandLine(Syntax, Nodes[0]->getValue(Storage)); 256 std::vector<std::string> Arguments; 257 for (const auto *Node : Nodes) 258 Arguments.push_back(Node->getValue(Storage)); 259 return Arguments; 260 } 261 262 void JSONCompilationDatabase::getCommands( 263 ArrayRef<CompileCommandRef> CommandsRef, 264 std::vector<CompileCommand> &Commands) const { 265 for (const auto &CommandRef : CommandsRef) { 266 SmallString<8> DirectoryStorage; 267 SmallString<32> FilenameStorage; 268 SmallString<32> OutputStorage; 269 auto Output = std::get<3>(CommandRef); 270 Commands.emplace_back( 271 std::get<0>(CommandRef)->getValue(DirectoryStorage), 272 std::get<1>(CommandRef)->getValue(FilenameStorage), 273 nodeToCommandLine(Syntax, std::get<2>(CommandRef)), 274 Output ? Output->getValue(OutputStorage) : ""); 275 } 276 } 277 278 bool JSONCompilationDatabase::parse(std::string &ErrorMessage) { 279 llvm::yaml::document_iterator I = YAMLStream.begin(); 280 if (I == YAMLStream.end()) { 281 ErrorMessage = "Error while parsing YAML."; 282 return false; 283 } 284 llvm::yaml::Node *Root = I->getRoot(); 285 if (!Root) { 286 ErrorMessage = "Error while parsing YAML."; 287 return false; 288 } 289 auto *Array = dyn_cast<llvm::yaml::SequenceNode>(Root); 290 if (!Array) { 291 ErrorMessage = "Expected array."; 292 return false; 293 } 294 for (auto &NextObject : *Array) { 295 auto *Object = dyn_cast<llvm::yaml::MappingNode>(&NextObject); 296 if (!Object) { 297 ErrorMessage = "Expected object."; 298 return false; 299 } 300 llvm::yaml::ScalarNode *Directory = nullptr; 301 llvm::Optional<std::vector<llvm::yaml::ScalarNode *>> Command; 302 llvm::yaml::ScalarNode *File = nullptr; 303 llvm::yaml::ScalarNode *Output = nullptr; 304 for (auto& NextKeyValue : *Object) { 305 auto *KeyString = dyn_cast<llvm::yaml::ScalarNode>(NextKeyValue.getKey()); 306 if (!KeyString) { 307 ErrorMessage = "Expected strings as key."; 308 return false; 309 } 310 SmallString<10> KeyStorage; 311 StringRef KeyValue = KeyString->getValue(KeyStorage); 312 llvm::yaml::Node *Value = NextKeyValue.getValue(); 313 if (!Value) { 314 ErrorMessage = "Expected value."; 315 return false; 316 } 317 auto *ValueString = dyn_cast<llvm::yaml::ScalarNode>(Value); 318 auto *SequenceString = dyn_cast<llvm::yaml::SequenceNode>(Value); 319 if (KeyValue == "arguments" && !SequenceString) { 320 ErrorMessage = "Expected sequence as value."; 321 return false; 322 } else if (KeyValue != "arguments" && !ValueString) { 323 ErrorMessage = "Expected string as value."; 324 return false; 325 } 326 if (KeyValue == "directory") { 327 Directory = ValueString; 328 } else if (KeyValue == "arguments") { 329 Command = std::vector<llvm::yaml::ScalarNode *>(); 330 for (auto &Argument : *SequenceString) { 331 auto *Scalar = dyn_cast<llvm::yaml::ScalarNode>(&Argument); 332 if (!Scalar) { 333 ErrorMessage = "Only strings are allowed in 'arguments'."; 334 return false; 335 } 336 Command->push_back(Scalar); 337 } 338 } else if (KeyValue == "command") { 339 if (!Command) 340 Command = std::vector<llvm::yaml::ScalarNode *>(1, ValueString); 341 } else if (KeyValue == "file") { 342 File = ValueString; 343 } else if (KeyValue == "output") { 344 Output = ValueString; 345 } else { 346 ErrorMessage = ("Unknown key: \"" + 347 KeyString->getRawValue() + "\"").str(); 348 return false; 349 } 350 } 351 if (!File) { 352 ErrorMessage = "Missing key: \"file\"."; 353 return false; 354 } 355 if (!Command) { 356 ErrorMessage = "Missing key: \"command\" or \"arguments\"."; 357 return false; 358 } 359 if (!Directory) { 360 ErrorMessage = "Missing key: \"directory\"."; 361 return false; 362 } 363 SmallString<8> FileStorage; 364 StringRef FileName = File->getValue(FileStorage); 365 SmallString<128> NativeFilePath; 366 if (llvm::sys::path::is_relative(FileName)) { 367 SmallString<8> DirectoryStorage; 368 SmallString<128> AbsolutePath( 369 Directory->getValue(DirectoryStorage)); 370 llvm::sys::path::append(AbsolutePath, FileName); 371 llvm::sys::path::native(AbsolutePath, NativeFilePath); 372 } else { 373 llvm::sys::path::native(FileName, NativeFilePath); 374 } 375 auto Cmd = CompileCommandRef(Directory, File, *Command, Output); 376 IndexByFile[NativeFilePath].push_back(Cmd); 377 AllCommands.push_back(Cmd); 378 MatchTrie.insert(NativeFilePath); 379 } 380 return true; 381 } 382