16ed1f85cSDaniel Jasper //===--- JSONCompilationDatabase.cpp - ------------------------------------===//
26ed1f85cSDaniel Jasper //
36ed1f85cSDaniel Jasper //                     The LLVM Compiler Infrastructure
46ed1f85cSDaniel Jasper //
56ed1f85cSDaniel Jasper // This file is distributed under the University of Illinois Open Source
66ed1f85cSDaniel Jasper // License. See LICENSE.TXT for details.
76ed1f85cSDaniel Jasper //
86ed1f85cSDaniel Jasper //===----------------------------------------------------------------------===//
96ed1f85cSDaniel Jasper //
106ed1f85cSDaniel Jasper //  This file contains the implementation of the JSONCompilationDatabase.
116ed1f85cSDaniel Jasper //
126ed1f85cSDaniel Jasper //===----------------------------------------------------------------------===//
136ed1f85cSDaniel Jasper 
146ed1f85cSDaniel Jasper #include "clang/Tooling/JSONCompilationDatabase.h"
156ed1f85cSDaniel Jasper #include "clang/Tooling/CompilationDatabase.h"
166ed1f85cSDaniel Jasper #include "clang/Tooling/CompilationDatabasePluginRegistry.h"
176ed1f85cSDaniel Jasper #include "clang/Tooling/Tooling.h"
186ed1f85cSDaniel Jasper #include "llvm/ADT/SmallString.h"
196ed1f85cSDaniel Jasper #include "llvm/Support/Path.h"
208a8e554aSRafael Espindola #include <system_error>
216ed1f85cSDaniel Jasper 
226ed1f85cSDaniel Jasper namespace clang {
236ed1f85cSDaniel Jasper namespace tooling {
246ed1f85cSDaniel Jasper 
256ed1f85cSDaniel Jasper namespace {
266ed1f85cSDaniel Jasper 
276ed1f85cSDaniel Jasper /// \brief A parser for escaped strings of command line arguments.
286ed1f85cSDaniel Jasper ///
296ed1f85cSDaniel Jasper /// Assumes \-escaping for quoted arguments (see the documentation of
306ed1f85cSDaniel Jasper /// unescapeCommandLine(...)).
316ed1f85cSDaniel Jasper class CommandLineArgumentParser {
326ed1f85cSDaniel Jasper  public:
336ed1f85cSDaniel Jasper   CommandLineArgumentParser(StringRef CommandLine)
346ed1f85cSDaniel Jasper       : Input(CommandLine), Position(Input.begin()-1) {}
356ed1f85cSDaniel Jasper 
366ed1f85cSDaniel Jasper   std::vector<std::string> parse() {
376ed1f85cSDaniel Jasper     bool HasMoreInput = true;
386ed1f85cSDaniel Jasper     while (HasMoreInput && nextNonWhitespace()) {
396ed1f85cSDaniel Jasper       std::string Argument;
406ed1f85cSDaniel Jasper       HasMoreInput = parseStringInto(Argument);
416ed1f85cSDaniel Jasper       CommandLine.push_back(Argument);
426ed1f85cSDaniel Jasper     }
436ed1f85cSDaniel Jasper     return CommandLine;
446ed1f85cSDaniel Jasper   }
456ed1f85cSDaniel Jasper 
466ed1f85cSDaniel Jasper  private:
476ed1f85cSDaniel Jasper   // All private methods return true if there is more input available.
486ed1f85cSDaniel Jasper 
496ed1f85cSDaniel Jasper   bool parseStringInto(std::string &String) {
506ed1f85cSDaniel Jasper     do {
516ed1f85cSDaniel Jasper       if (*Position == '"') {
52fe7a3486SPeter Collingbourne         if (!parseDoubleQuotedStringInto(String)) return false;
53fe7a3486SPeter Collingbourne       } else if (*Position == '\'') {
54fe7a3486SPeter Collingbourne         if (!parseSingleQuotedStringInto(String)) return false;
556ed1f85cSDaniel Jasper       } else {
566ed1f85cSDaniel Jasper         if (!parseFreeStringInto(String)) return false;
576ed1f85cSDaniel Jasper       }
586ed1f85cSDaniel Jasper     } while (*Position != ' ');
596ed1f85cSDaniel Jasper     return true;
606ed1f85cSDaniel Jasper   }
616ed1f85cSDaniel Jasper 
62fe7a3486SPeter Collingbourne   bool parseDoubleQuotedStringInto(std::string &String) {
636ed1f85cSDaniel Jasper     if (!next()) return false;
646ed1f85cSDaniel Jasper     while (*Position != '"') {
656ed1f85cSDaniel Jasper       if (!skipEscapeCharacter()) return false;
666ed1f85cSDaniel Jasper       String.push_back(*Position);
676ed1f85cSDaniel Jasper       if (!next()) return false;
686ed1f85cSDaniel Jasper     }
696ed1f85cSDaniel Jasper     return next();
706ed1f85cSDaniel Jasper   }
716ed1f85cSDaniel Jasper 
72fe7a3486SPeter Collingbourne   bool parseSingleQuotedStringInto(std::string &String) {
73fe7a3486SPeter Collingbourne     if (!next()) return false;
74fe7a3486SPeter Collingbourne     while (*Position != '\'') {
75fe7a3486SPeter Collingbourne       String.push_back(*Position);
76fe7a3486SPeter Collingbourne       if (!next()) return false;
77fe7a3486SPeter Collingbourne     }
78fe7a3486SPeter Collingbourne     return next();
79fe7a3486SPeter Collingbourne   }
80fe7a3486SPeter Collingbourne 
816ed1f85cSDaniel Jasper   bool parseFreeStringInto(std::string &String) {
826ed1f85cSDaniel Jasper     do {
836ed1f85cSDaniel Jasper       if (!skipEscapeCharacter()) return false;
846ed1f85cSDaniel Jasper       String.push_back(*Position);
856ed1f85cSDaniel Jasper       if (!next()) return false;
86fe7a3486SPeter Collingbourne     } while (*Position != ' ' && *Position != '"' && *Position != '\'');
876ed1f85cSDaniel Jasper     return true;
886ed1f85cSDaniel Jasper   }
896ed1f85cSDaniel Jasper 
906ed1f85cSDaniel Jasper   bool skipEscapeCharacter() {
916ed1f85cSDaniel Jasper     if (*Position == '\\') {
926ed1f85cSDaniel Jasper       return next();
936ed1f85cSDaniel Jasper     }
946ed1f85cSDaniel Jasper     return true;
956ed1f85cSDaniel Jasper   }
966ed1f85cSDaniel Jasper 
976ed1f85cSDaniel Jasper   bool nextNonWhitespace() {
986ed1f85cSDaniel Jasper     do {
996ed1f85cSDaniel Jasper       if (!next()) return false;
1006ed1f85cSDaniel Jasper     } while (*Position == ' ');
1016ed1f85cSDaniel Jasper     return true;
1026ed1f85cSDaniel Jasper   }
1036ed1f85cSDaniel Jasper 
1046ed1f85cSDaniel Jasper   bool next() {
1056ed1f85cSDaniel Jasper     ++Position;
1066ed1f85cSDaniel Jasper     return Position != Input.end();
1076ed1f85cSDaniel Jasper   }
1086ed1f85cSDaniel Jasper 
1096ed1f85cSDaniel Jasper   const StringRef Input;
1106ed1f85cSDaniel Jasper   StringRef::iterator Position;
1116ed1f85cSDaniel Jasper   std::vector<std::string> CommandLine;
1126ed1f85cSDaniel Jasper };
1136ed1f85cSDaniel Jasper 
1146ed1f85cSDaniel Jasper std::vector<std::string> unescapeCommandLine(
1156ed1f85cSDaniel Jasper     StringRef EscapedCommandLine) {
1166ed1f85cSDaniel Jasper   CommandLineArgumentParser parser(EscapedCommandLine);
1176ed1f85cSDaniel Jasper   return parser.parse();
1186ed1f85cSDaniel Jasper }
1196ed1f85cSDaniel Jasper 
1206ed1f85cSDaniel Jasper class JSONCompilationDatabasePlugin : public CompilationDatabasePlugin {
121*cdba84c0SDavid Blaikie   std::unique_ptr<CompilationDatabase>
122*cdba84c0SDavid Blaikie   loadFromDirectory(StringRef Directory, std::string &ErrorMessage) override {
123f857950dSDmitri Gribenko     SmallString<1024> JSONDatabasePath(Directory);
1246ed1f85cSDaniel Jasper     llvm::sys::path::append(JSONDatabasePath, "compile_commands.json");
125b8984329SAhmed Charles     std::unique_ptr<CompilationDatabase> Database(
1266ed1f85cSDaniel Jasper         JSONCompilationDatabase::loadFromFile(JSONDatabasePath, ErrorMessage));
1276ed1f85cSDaniel Jasper     if (!Database)
128ccbc35edSCraig Topper       return nullptr;
129*cdba84c0SDavid Blaikie     return Database;
1306ed1f85cSDaniel Jasper   }
1316ed1f85cSDaniel Jasper };
1326ed1f85cSDaniel Jasper 
13369b6277aSCraig Topper } // end namespace
13469b6277aSCraig Topper 
1356ed1f85cSDaniel Jasper // Register the JSONCompilationDatabasePlugin with the
1366ed1f85cSDaniel Jasper // CompilationDatabasePluginRegistry using this statically initialized variable.
1376ed1f85cSDaniel Jasper static CompilationDatabasePluginRegistry::Add<JSONCompilationDatabasePlugin>
1386ed1f85cSDaniel Jasper X("json-compilation-database", "Reads JSON formatted compilation databases");
1396ed1f85cSDaniel Jasper 
1406ed1f85cSDaniel Jasper // This anchor is used to force the linker to link in the generated object file
1416ed1f85cSDaniel Jasper // and thus register the JSONCompilationDatabasePlugin.
142d574ac2fSNAKAMURA Takumi volatile int JSONAnchorSource = 0;
1436ed1f85cSDaniel Jasper 
144*cdba84c0SDavid Blaikie std::unique_ptr<JSONCompilationDatabase>
1456ed1f85cSDaniel Jasper JSONCompilationDatabase::loadFromFile(StringRef FilePath,
1466ed1f85cSDaniel Jasper                                       std::string &ErrorMessage) {
1472d2b420aSRafael Espindola   llvm::ErrorOr<std::unique_ptr<llvm::MemoryBuffer>> DatabaseBuffer =
1482d2b420aSRafael Espindola       llvm::MemoryBuffer::getFile(FilePath);
1492d2b420aSRafael Espindola   if (std::error_code Result = DatabaseBuffer.getError()) {
1506ed1f85cSDaniel Jasper     ErrorMessage = "Error while opening JSON database: " + Result.message();
151ccbc35edSCraig Topper     return nullptr;
1526ed1f85cSDaniel Jasper   }
153b8984329SAhmed Charles   std::unique_ptr<JSONCompilationDatabase> Database(
1542d2b420aSRafael Espindola       new JSONCompilationDatabase(DatabaseBuffer->release()));
1556ed1f85cSDaniel Jasper   if (!Database->parse(ErrorMessage))
156ccbc35edSCraig Topper     return nullptr;
157*cdba84c0SDavid Blaikie   return Database;
1586ed1f85cSDaniel Jasper }
1596ed1f85cSDaniel Jasper 
160*cdba84c0SDavid Blaikie std::unique_ptr<JSONCompilationDatabase>
1616ed1f85cSDaniel Jasper JSONCompilationDatabase::loadFromBuffer(StringRef DatabaseString,
1626ed1f85cSDaniel Jasper                                         std::string &ErrorMessage) {
163b8984329SAhmed Charles   std::unique_ptr<llvm::MemoryBuffer> DatabaseBuffer(
1646ed1f85cSDaniel Jasper       llvm::MemoryBuffer::getMemBuffer(DatabaseString));
165b8984329SAhmed Charles   std::unique_ptr<JSONCompilationDatabase> Database(
1669a16beb8SAhmed Charles       new JSONCompilationDatabase(DatabaseBuffer.release()));
1676ed1f85cSDaniel Jasper   if (!Database->parse(ErrorMessage))
168ccbc35edSCraig Topper     return nullptr;
169*cdba84c0SDavid Blaikie   return Database;
1706ed1f85cSDaniel Jasper }
1716ed1f85cSDaniel Jasper 
1726ed1f85cSDaniel Jasper std::vector<CompileCommand>
1736ed1f85cSDaniel Jasper JSONCompilationDatabase::getCompileCommands(StringRef FilePath) const {
174f857950dSDmitri Gribenko   SmallString<128> NativeFilePath;
1756ed1f85cSDaniel Jasper   llvm::sys::path::native(FilePath, NativeFilePath);
176965f8825SAlp Toker 
17726cf9c43SDaniel Jasper   std::string Error;
17826cf9c43SDaniel Jasper   llvm::raw_string_ostream ES(Error);
17926cf9c43SDaniel Jasper   StringRef Match = MatchTrie.findEquivalent(NativeFilePath.str(), ES);
1803128a11eSArnaud A. de Grandmaison   if (Match.empty())
18126cf9c43SDaniel Jasper     return std::vector<CompileCommand>();
1826ed1f85cSDaniel Jasper   llvm::StringMap< std::vector<CompileCommandRef> >::const_iterator
18326cf9c43SDaniel Jasper     CommandsRefI = IndexByFile.find(Match);
1846ed1f85cSDaniel Jasper   if (CommandsRefI == IndexByFile.end())
1856ed1f85cSDaniel Jasper     return std::vector<CompileCommand>();
1866ed1f85cSDaniel Jasper   std::vector<CompileCommand> Commands;
187251ad5e0SArgyrios Kyrtzidis   getCommands(CommandsRefI->getValue(), Commands);
1886ed1f85cSDaniel Jasper   return Commands;
1896ed1f85cSDaniel Jasper }
1906ed1f85cSDaniel Jasper 
1916ed1f85cSDaniel Jasper std::vector<std::string>
1926ed1f85cSDaniel Jasper JSONCompilationDatabase::getAllFiles() const {
1936ed1f85cSDaniel Jasper   std::vector<std::string> Result;
1946ed1f85cSDaniel Jasper 
1956ed1f85cSDaniel Jasper   llvm::StringMap< std::vector<CompileCommandRef> >::const_iterator
1966ed1f85cSDaniel Jasper     CommandsRefI = IndexByFile.begin();
1976ed1f85cSDaniel Jasper   const llvm::StringMap< std::vector<CompileCommandRef> >::const_iterator
1986ed1f85cSDaniel Jasper     CommandsRefEnd = IndexByFile.end();
1996ed1f85cSDaniel Jasper   for (; CommandsRefI != CommandsRefEnd; ++CommandsRefI) {
2006ed1f85cSDaniel Jasper     Result.push_back(CommandsRefI->first().str());
2016ed1f85cSDaniel Jasper   }
2026ed1f85cSDaniel Jasper 
2036ed1f85cSDaniel Jasper   return Result;
2046ed1f85cSDaniel Jasper }
2056ed1f85cSDaniel Jasper 
206251ad5e0SArgyrios Kyrtzidis std::vector<CompileCommand>
207251ad5e0SArgyrios Kyrtzidis JSONCompilationDatabase::getAllCompileCommands() const {
208251ad5e0SArgyrios Kyrtzidis   std::vector<CompileCommand> Commands;
209251ad5e0SArgyrios Kyrtzidis   for (llvm::StringMap< std::vector<CompileCommandRef> >::const_iterator
210251ad5e0SArgyrios Kyrtzidis         CommandsRefI = IndexByFile.begin(), CommandsRefEnd = IndexByFile.end();
211251ad5e0SArgyrios Kyrtzidis       CommandsRefI != CommandsRefEnd; ++CommandsRefI) {
212251ad5e0SArgyrios Kyrtzidis     getCommands(CommandsRefI->getValue(), Commands);
213251ad5e0SArgyrios Kyrtzidis   }
214251ad5e0SArgyrios Kyrtzidis   return Commands;
215251ad5e0SArgyrios Kyrtzidis }
216251ad5e0SArgyrios Kyrtzidis 
217251ad5e0SArgyrios Kyrtzidis void JSONCompilationDatabase::getCommands(
218251ad5e0SArgyrios Kyrtzidis                                   ArrayRef<CompileCommandRef> CommandsRef,
219251ad5e0SArgyrios Kyrtzidis                                   std::vector<CompileCommand> &Commands) const {
220251ad5e0SArgyrios Kyrtzidis   for (int I = 0, E = CommandsRef.size(); I != E; ++I) {
221f857950dSDmitri Gribenko     SmallString<8> DirectoryStorage;
222f857950dSDmitri Gribenko     SmallString<1024> CommandStorage;
223251ad5e0SArgyrios Kyrtzidis     Commands.push_back(CompileCommand(
224251ad5e0SArgyrios Kyrtzidis       // FIXME: Escape correctly:
225251ad5e0SArgyrios Kyrtzidis       CommandsRef[I].first->getValue(DirectoryStorage),
226251ad5e0SArgyrios Kyrtzidis       unescapeCommandLine(CommandsRef[I].second->getValue(CommandStorage))));
227251ad5e0SArgyrios Kyrtzidis   }
228251ad5e0SArgyrios Kyrtzidis }
229251ad5e0SArgyrios Kyrtzidis 
2306ed1f85cSDaniel Jasper bool JSONCompilationDatabase::parse(std::string &ErrorMessage) {
2316ed1f85cSDaniel Jasper   llvm::yaml::document_iterator I = YAMLStream.begin();
2326ed1f85cSDaniel Jasper   if (I == YAMLStream.end()) {
2336ed1f85cSDaniel Jasper     ErrorMessage = "Error while parsing YAML.";
2346ed1f85cSDaniel Jasper     return false;
2356ed1f85cSDaniel Jasper   }
2366ed1f85cSDaniel Jasper   llvm::yaml::Node *Root = I->getRoot();
237ccbc35edSCraig Topper   if (!Root) {
2386ed1f85cSDaniel Jasper     ErrorMessage = "Error while parsing YAML.";
2396ed1f85cSDaniel Jasper     return false;
2406ed1f85cSDaniel Jasper   }
241f857950dSDmitri Gribenko   llvm::yaml::SequenceNode *Array = dyn_cast<llvm::yaml::SequenceNode>(Root);
242ccbc35edSCraig Topper   if (!Array) {
2436ed1f85cSDaniel Jasper     ErrorMessage = "Expected array.";
2446ed1f85cSDaniel Jasper     return false;
2456ed1f85cSDaniel Jasper   }
2466ed1f85cSDaniel Jasper   for (llvm::yaml::SequenceNode::iterator AI = Array->begin(),
2476ed1f85cSDaniel Jasper                                           AE = Array->end();
2486ed1f85cSDaniel Jasper        AI != AE; ++AI) {
249f857950dSDmitri Gribenko     llvm::yaml::MappingNode *Object = dyn_cast<llvm::yaml::MappingNode>(&*AI);
250ccbc35edSCraig Topper     if (!Object) {
2516ed1f85cSDaniel Jasper       ErrorMessage = "Expected object.";
2526ed1f85cSDaniel Jasper       return false;
2536ed1f85cSDaniel Jasper     }
254ccbc35edSCraig Topper     llvm::yaml::ScalarNode *Directory = nullptr;
255ccbc35edSCraig Topper     llvm::yaml::ScalarNode *Command = nullptr;
256ccbc35edSCraig Topper     llvm::yaml::ScalarNode *File = nullptr;
2576ed1f85cSDaniel Jasper     for (llvm::yaml::MappingNode::iterator KVI = Object->begin(),
2586ed1f85cSDaniel Jasper                                            KVE = Object->end();
2596ed1f85cSDaniel Jasper          KVI != KVE; ++KVI) {
2606ed1f85cSDaniel Jasper       llvm::yaml::Node *Value = (*KVI).getValue();
261ccbc35edSCraig Topper       if (!Value) {
2626ed1f85cSDaniel Jasper         ErrorMessage = "Expected value.";
2636ed1f85cSDaniel Jasper         return false;
2646ed1f85cSDaniel Jasper       }
2656ed1f85cSDaniel Jasper       llvm::yaml::ScalarNode *ValueString =
266f857950dSDmitri Gribenko           dyn_cast<llvm::yaml::ScalarNode>(Value);
267ccbc35edSCraig Topper       if (!ValueString) {
2686ed1f85cSDaniel Jasper         ErrorMessage = "Expected string as value.";
2696ed1f85cSDaniel Jasper         return false;
2706ed1f85cSDaniel Jasper       }
2716ed1f85cSDaniel Jasper       llvm::yaml::ScalarNode *KeyString =
272f857950dSDmitri Gribenko           dyn_cast<llvm::yaml::ScalarNode>((*KVI).getKey());
273ccbc35edSCraig Topper       if (!KeyString) {
2746ed1f85cSDaniel Jasper         ErrorMessage = "Expected strings as key.";
2756ed1f85cSDaniel Jasper         return false;
2766ed1f85cSDaniel Jasper       }
277f857950dSDmitri Gribenko       SmallString<8> KeyStorage;
2786ed1f85cSDaniel Jasper       if (KeyString->getValue(KeyStorage) == "directory") {
2796ed1f85cSDaniel Jasper         Directory = ValueString;
2806ed1f85cSDaniel Jasper       } else if (KeyString->getValue(KeyStorage) == "command") {
2816ed1f85cSDaniel Jasper         Command = ValueString;
2826ed1f85cSDaniel Jasper       } else if (KeyString->getValue(KeyStorage) == "file") {
2836ed1f85cSDaniel Jasper         File = ValueString;
2846ed1f85cSDaniel Jasper       } else {
2856ed1f85cSDaniel Jasper         ErrorMessage = ("Unknown key: \"" +
2866ed1f85cSDaniel Jasper                         KeyString->getRawValue() + "\"").str();
2876ed1f85cSDaniel Jasper         return false;
2886ed1f85cSDaniel Jasper       }
2896ed1f85cSDaniel Jasper     }
2906ed1f85cSDaniel Jasper     if (!File) {
2916ed1f85cSDaniel Jasper       ErrorMessage = "Missing key: \"file\".";
2926ed1f85cSDaniel Jasper       return false;
2936ed1f85cSDaniel Jasper     }
2946ed1f85cSDaniel Jasper     if (!Command) {
2956ed1f85cSDaniel Jasper       ErrorMessage = "Missing key: \"command\".";
2966ed1f85cSDaniel Jasper       return false;
2976ed1f85cSDaniel Jasper     }
2986ed1f85cSDaniel Jasper     if (!Directory) {
2996ed1f85cSDaniel Jasper       ErrorMessage = "Missing key: \"directory\".";
3006ed1f85cSDaniel Jasper       return false;
3016ed1f85cSDaniel Jasper     }
302f857950dSDmitri Gribenko     SmallString<8> FileStorage;
30326cf9c43SDaniel Jasper     StringRef FileName = File->getValue(FileStorage);
304f857950dSDmitri Gribenko     SmallString<128> NativeFilePath;
30526cf9c43SDaniel Jasper     if (llvm::sys::path::is_relative(FileName)) {
306f857950dSDmitri Gribenko       SmallString<8> DirectoryStorage;
307f857950dSDmitri Gribenko       SmallString<128> AbsolutePath(
30826cf9c43SDaniel Jasper           Directory->getValue(DirectoryStorage));
30926cf9c43SDaniel Jasper       llvm::sys::path::append(AbsolutePath, FileName);
31026cf9c43SDaniel Jasper       llvm::sys::path::native(AbsolutePath.str(), NativeFilePath);
31126cf9c43SDaniel Jasper     } else {
31226cf9c43SDaniel Jasper       llvm::sys::path::native(FileName, NativeFilePath);
31326cf9c43SDaniel Jasper     }
3146ed1f85cSDaniel Jasper     IndexByFile[NativeFilePath].push_back(
3156ed1f85cSDaniel Jasper         CompileCommandRef(Directory, Command));
31626cf9c43SDaniel Jasper     MatchTrie.insert(NativeFilePath.str());
3176ed1f85cSDaniel Jasper   }
3186ed1f85cSDaniel Jasper   return true;
3196ed1f85cSDaniel Jasper }
3206ed1f85cSDaniel Jasper 
3216ed1f85cSDaniel Jasper } // end namespace tooling
3226ed1f85cSDaniel Jasper } // end namespace clang
323