xref: /llvm-project/clang/lib/Tooling/JSONCompilationDatabase.cpp (revision 3ecd8c0aabd30107d4fff8fdd9547f77b3b66d1a)
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/Path.h"
20 #include <system_error>
21 
22 namespace clang {
23 namespace tooling {
24 
25 namespace {
26 
27 /// \brief A parser for escaped strings of command line arguments.
28 ///
29 /// Assumes \-escaping for quoted arguments (see the documentation of
30 /// unescapeCommandLine(...)).
31 class CommandLineArgumentParser {
32  public:
33   CommandLineArgumentParser(StringRef CommandLine)
34       : Input(CommandLine), Position(Input.begin()-1) {}
35 
36   std::vector<std::string> parse() {
37     bool HasMoreInput = true;
38     while (HasMoreInput && nextNonWhitespace()) {
39       std::string Argument;
40       HasMoreInput = parseStringInto(Argument);
41       CommandLine.push_back(Argument);
42     }
43     return CommandLine;
44   }
45 
46  private:
47   // All private methods return true if there is more input available.
48 
49   bool parseStringInto(std::string &String) {
50     do {
51       if (*Position == '"') {
52         if (!parseDoubleQuotedStringInto(String)) return false;
53       } else if (*Position == '\'') {
54         if (!parseSingleQuotedStringInto(String)) return false;
55       } else {
56         if (!parseFreeStringInto(String)) return false;
57       }
58     } while (*Position != ' ');
59     return true;
60   }
61 
62   bool parseDoubleQuotedStringInto(std::string &String) {
63     if (!next()) return false;
64     while (*Position != '"') {
65       if (!skipEscapeCharacter()) return false;
66       String.push_back(*Position);
67       if (!next()) return false;
68     }
69     return next();
70   }
71 
72   bool parseSingleQuotedStringInto(std::string &String) {
73     if (!next()) return false;
74     while (*Position != '\'') {
75       String.push_back(*Position);
76       if (!next()) return false;
77     }
78     return next();
79   }
80 
81   bool parseFreeStringInto(std::string &String) {
82     do {
83       if (!skipEscapeCharacter()) return false;
84       String.push_back(*Position);
85       if (!next()) return false;
86     } while (*Position != ' ' && *Position != '"' && *Position != '\'');
87     return true;
88   }
89 
90   bool skipEscapeCharacter() {
91     if (*Position == '\\') {
92       return next();
93     }
94     return true;
95   }
96 
97   bool nextNonWhitespace() {
98     do {
99       if (!next()) return false;
100     } while (*Position == ' ');
101     return true;
102   }
103 
104   bool next() {
105     ++Position;
106     return Position != Input.end();
107   }
108 
109   const StringRef Input;
110   StringRef::iterator Position;
111   std::vector<std::string> CommandLine;
112 };
113 
114 std::vector<std::string> unescapeCommandLine(
115     StringRef EscapedCommandLine) {
116   CommandLineArgumentParser parser(EscapedCommandLine);
117   return parser.parse();
118 }
119 
120 class JSONCompilationDatabasePlugin : public CompilationDatabasePlugin {
121   std::unique_ptr<CompilationDatabase>
122   loadFromDirectory(StringRef Directory, std::string &ErrorMessage) override {
123     SmallString<1024> JSONDatabasePath(Directory);
124     llvm::sys::path::append(JSONDatabasePath, "compile_commands.json");
125     std::unique_ptr<CompilationDatabase> Database(
126         JSONCompilationDatabase::loadFromFile(JSONDatabasePath, ErrorMessage));
127     if (!Database)
128       return nullptr;
129     return Database;
130   }
131 };
132 
133 } // end namespace
134 
135 // Register the JSONCompilationDatabasePlugin with the
136 // CompilationDatabasePluginRegistry using this statically initialized variable.
137 static CompilationDatabasePluginRegistry::Add<JSONCompilationDatabasePlugin>
138 X("json-compilation-database", "Reads JSON formatted compilation databases");
139 
140 // This anchor is used to force the linker to link in the generated object file
141 // and thus register the JSONCompilationDatabasePlugin.
142 volatile int JSONAnchorSource = 0;
143 
144 std::unique_ptr<JSONCompilationDatabase>
145 JSONCompilationDatabase::loadFromFile(StringRef FilePath,
146                                       std::string &ErrorMessage) {
147   llvm::ErrorOr<std::unique_ptr<llvm::MemoryBuffer>> DatabaseBuffer =
148       llvm::MemoryBuffer::getFile(FilePath);
149   if (std::error_code Result = DatabaseBuffer.getError()) {
150     ErrorMessage = "Error while opening JSON database: " + Result.message();
151     return nullptr;
152   }
153   std::unique_ptr<JSONCompilationDatabase> Database(
154       new JSONCompilationDatabase(std::move(*DatabaseBuffer)));
155   if (!Database->parse(ErrorMessage))
156     return nullptr;
157   return Database;
158 }
159 
160 std::unique_ptr<JSONCompilationDatabase>
161 JSONCompilationDatabase::loadFromBuffer(StringRef DatabaseString,
162                                         std::string &ErrorMessage) {
163   std::unique_ptr<llvm::MemoryBuffer> DatabaseBuffer(
164       llvm::MemoryBuffer::getMemBuffer(DatabaseString));
165   std::unique_ptr<JSONCompilationDatabase> Database(
166       new JSONCompilationDatabase(std::move(DatabaseBuffer)));
167   if (!Database->parse(ErrorMessage))
168     return nullptr;
169   return Database;
170 }
171 
172 std::vector<CompileCommand>
173 JSONCompilationDatabase::getCompileCommands(StringRef FilePath) const {
174   SmallString<128> NativeFilePath;
175   llvm::sys::path::native(FilePath, NativeFilePath);
176 
177   std::string Error;
178   llvm::raw_string_ostream ES(Error);
179   StringRef Match = MatchTrie.findEquivalent(NativeFilePath, ES);
180   if (Match.empty())
181     return std::vector<CompileCommand>();
182   llvm::StringMap< std::vector<CompileCommandRef> >::const_iterator
183     CommandsRefI = IndexByFile.find(Match);
184   if (CommandsRefI == IndexByFile.end())
185     return std::vector<CompileCommand>();
186   std::vector<CompileCommand> Commands;
187   getCommands(CommandsRefI->getValue(), Commands);
188   return Commands;
189 }
190 
191 std::vector<std::string>
192 JSONCompilationDatabase::getAllFiles() const {
193   std::vector<std::string> Result;
194 
195   llvm::StringMap< std::vector<CompileCommandRef> >::const_iterator
196     CommandsRefI = IndexByFile.begin();
197   const llvm::StringMap< std::vector<CompileCommandRef> >::const_iterator
198     CommandsRefEnd = IndexByFile.end();
199   for (; CommandsRefI != CommandsRefEnd; ++CommandsRefI) {
200     Result.push_back(CommandsRefI->first().str());
201   }
202 
203   return Result;
204 }
205 
206 std::vector<CompileCommand>
207 JSONCompilationDatabase::getAllCompileCommands() const {
208   std::vector<CompileCommand> Commands;
209   for (llvm::StringMap< std::vector<CompileCommandRef> >::const_iterator
210         CommandsRefI = IndexByFile.begin(), CommandsRefEnd = IndexByFile.end();
211       CommandsRefI != CommandsRefEnd; ++CommandsRefI) {
212     getCommands(CommandsRefI->getValue(), Commands);
213   }
214   return Commands;
215 }
216 
217 static std::vector<std::string>
218 nodeToCommandLine(const std::vector<llvm::yaml::ScalarNode *> &Nodes) {
219   SmallString<1024> Storage;
220   if (Nodes.size() == 1) {
221     return unescapeCommandLine(Nodes[0]->getValue(Storage));
222   }
223   std::vector<std::string> Arguments;
224   for (auto *Node : Nodes) {
225     Arguments.push_back(Node->getValue(Storage));
226   }
227   return Arguments;
228 }
229 
230 void JSONCompilationDatabase::getCommands(
231     ArrayRef<CompileCommandRef> CommandsRef,
232     std::vector<CompileCommand> &Commands) const {
233   for (int I = 0, E = CommandsRef.size(); I != E; ++I) {
234     SmallString<8> DirectoryStorage;
235     Commands.emplace_back(CommandsRef[I].first->getValue(DirectoryStorage),
236                           nodeToCommandLine(CommandsRef[I].second));
237   }
238 }
239 
240 bool JSONCompilationDatabase::parse(std::string &ErrorMessage) {
241   llvm::yaml::document_iterator I = YAMLStream.begin();
242   if (I == YAMLStream.end()) {
243     ErrorMessage = "Error while parsing YAML.";
244     return false;
245   }
246   llvm::yaml::Node *Root = I->getRoot();
247   if (!Root) {
248     ErrorMessage = "Error while parsing YAML.";
249     return false;
250   }
251   llvm::yaml::SequenceNode *Array = dyn_cast<llvm::yaml::SequenceNode>(Root);
252   if (!Array) {
253     ErrorMessage = "Expected array.";
254     return false;
255   }
256   for (auto& NextObject : *Array) {
257     llvm::yaml::MappingNode *Object = dyn_cast<llvm::yaml::MappingNode>(&NextObject);
258     if (!Object) {
259       ErrorMessage = "Expected object.";
260       return false;
261     }
262     llvm::yaml::ScalarNode *Directory = nullptr;
263     llvm::Optional<std::vector<llvm::yaml::ScalarNode *>> Command;
264     llvm::yaml::ScalarNode *File = nullptr;
265     for (auto& NextKeyValue : *Object) {
266       llvm::yaml::ScalarNode *KeyString =
267           dyn_cast<llvm::yaml::ScalarNode>(NextKeyValue.getKey());
268       if (!KeyString) {
269         ErrorMessage = "Expected strings as key.";
270         return false;
271       }
272       SmallString<10> KeyStorage;
273       StringRef KeyValue = KeyString->getValue(KeyStorage);
274       llvm::yaml::Node *Value = NextKeyValue.getValue();
275       if (!Value) {
276         ErrorMessage = "Expected value.";
277         return false;
278       }
279       llvm::yaml::ScalarNode *ValueString =
280           dyn_cast<llvm::yaml::ScalarNode>(Value);
281       llvm::yaml::SequenceNode *SequenceString =
282           dyn_cast<llvm::yaml::SequenceNode>(Value);
283       if (KeyValue == "arguments" && !SequenceString) {
284         ErrorMessage = "Expected sequence as value.";
285         return false;
286       } else if (KeyValue != "arguments" && !ValueString) {
287         ErrorMessage = "Expected string as value.";
288         return false;
289       }
290       if (KeyValue == "directory") {
291         Directory = ValueString;
292       } else if (KeyValue == "arguments") {
293         Command = std::vector<llvm::yaml::ScalarNode *>();
294         for (auto &Argument : *SequenceString) {
295           auto Scalar = dyn_cast<llvm::yaml::ScalarNode>(&Argument);
296           if (!Scalar) {
297             ErrorMessage = "Only strings are allowed in 'arguments'.";
298             return false;
299           }
300           Command->push_back(Scalar);
301         }
302       } else if (KeyValue == "command") {
303         if (!Command)
304           Command = std::vector<llvm::yaml::ScalarNode *>(1, ValueString);
305       } else if (KeyValue == "file") {
306         File = ValueString;
307       } else {
308         ErrorMessage = ("Unknown key: \"" +
309                         KeyString->getRawValue() + "\"").str();
310         return false;
311       }
312     }
313     if (!File) {
314       ErrorMessage = "Missing key: \"file\".";
315       return false;
316     }
317     if (!Command) {
318       ErrorMessage = "Missing key: \"command\" or \"arguments\".";
319       return false;
320     }
321     if (!Directory) {
322       ErrorMessage = "Missing key: \"directory\".";
323       return false;
324     }
325     SmallString<8> FileStorage;
326     StringRef FileName = File->getValue(FileStorage);
327     SmallString<128> NativeFilePath;
328     if (llvm::sys::path::is_relative(FileName)) {
329       SmallString<8> DirectoryStorage;
330       SmallString<128> AbsolutePath(
331           Directory->getValue(DirectoryStorage));
332       llvm::sys::path::append(AbsolutePath, FileName);
333       llvm::sys::path::native(AbsolutePath, NativeFilePath);
334     } else {
335       llvm::sys::path::native(FileName, NativeFilePath);
336     }
337     IndexByFile[NativeFilePath].push_back(
338         CompileCommandRef(Directory, *Command));
339     MatchTrie.insert(NativeFilePath);
340   }
341   return true;
342 }
343 
344 } // end namespace tooling
345 } // end namespace clang
346