1 //===-- ClangIncludeFixer.cpp - Standalone include fixer ------------------===//
2 //
3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4 // See https://llvm.org/LICENSE.txt for license information.
5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6 //
7 //===----------------------------------------------------------------------===//
8 
9 #include "FuzzySymbolIndex.h"
10 #include "InMemorySymbolIndex.h"
11 #include "IncludeFixer.h"
12 #include "IncludeFixerContext.h"
13 #include "SymbolIndexManager.h"
14 #include "YamlSymbolIndex.h"
15 #include "clang/Format/Format.h"
16 #include "clang/Frontend/TextDiagnosticPrinter.h"
17 #include "clang/Rewrite/Core/Rewriter.h"
18 #include "clang/Tooling/CommonOptionsParser.h"
19 #include "clang/Tooling/Core/Replacement.h"
20 #include "clang/Tooling/Tooling.h"
21 #include "llvm/Support/CommandLine.h"
22 #include "llvm/Support/Path.h"
23 #include "llvm/Support/YAMLTraits.h"
24 
25 using namespace clang;
26 using namespace llvm;
27 using clang::include_fixer::IncludeFixerContext;
28 
29 LLVM_YAML_IS_DOCUMENT_LIST_VECTOR(IncludeFixerContext)
30 LLVM_YAML_IS_FLOW_SEQUENCE_VECTOR(IncludeFixerContext::HeaderInfo)
31 LLVM_YAML_IS_FLOW_SEQUENCE_VECTOR(IncludeFixerContext::QuerySymbolInfo)
32 
33 namespace llvm {
34 namespace yaml {
35 
36 template <> struct MappingTraits<tooling::Range> {
37   struct NormalizedRange {
38     NormalizedRange(const IO &) : Offset(0), Length(0) {}
39 
40     NormalizedRange(const IO &, const tooling::Range &R)
41         : Offset(R.getOffset()), Length(R.getLength()) {}
42 
43     tooling::Range denormalize(const IO &) {
44       return tooling::Range(Offset, Length);
45     }
46 
47     unsigned Offset;
48     unsigned Length;
49   };
50   static void mapping(IO &IO, tooling::Range &Info) {
51     MappingNormalization<NormalizedRange, tooling::Range> Keys(IO, Info);
52     IO.mapRequired("Offset", Keys->Offset);
53     IO.mapRequired("Length", Keys->Length);
54   }
55 };
56 
57 template <> struct MappingTraits<IncludeFixerContext::HeaderInfo> {
58   static void mapping(IO &io, IncludeFixerContext::HeaderInfo &Info) {
59     io.mapRequired("Header", Info.Header);
60     io.mapRequired("QualifiedName", Info.QualifiedName);
61   }
62 };
63 
64 template <> struct MappingTraits<IncludeFixerContext::QuerySymbolInfo> {
65   static void mapping(IO &io, IncludeFixerContext::QuerySymbolInfo &Info) {
66     io.mapRequired("RawIdentifier", Info.RawIdentifier);
67     io.mapRequired("Range", Info.Range);
68   }
69 };
70 
71 template <> struct MappingTraits<IncludeFixerContext> {
72   static void mapping(IO &IO, IncludeFixerContext &Context) {
73     IO.mapRequired("QuerySymbolInfos", Context.QuerySymbolInfos);
74     IO.mapRequired("HeaderInfos", Context.HeaderInfos);
75     IO.mapRequired("FilePath", Context.FilePath);
76   }
77 };
78 } // namespace yaml
79 } // namespace llvm
80 
81 namespace {
82 cl::OptionCategory IncludeFixerCategory("Tool options");
83 
84 enum DatabaseFormatTy {
85   fixed,     ///< Hard-coded mapping.
86   yaml,      ///< Yaml database created by find-all-symbols.
87   fuzzyYaml, ///< Yaml database with fuzzy-matched identifiers.
88 };
89 
90 cl::opt<DatabaseFormatTy> DatabaseFormat(
91     "db", cl::desc("Specify input format"),
92     cl::values(clEnumVal(fixed, "Hard-coded mapping"),
93                clEnumVal(yaml, "Yaml database created by find-all-symbols"),
94                clEnumVal(fuzzyYaml, "Yaml database, with fuzzy-matched names")),
95     cl::init(yaml), cl::cat(IncludeFixerCategory));
96 
97 cl::opt<std::string> Input("input",
98                            cl::desc("String to initialize the database"),
99                            cl::cat(IncludeFixerCategory));
100 
101 cl::opt<std::string>
102     QuerySymbol("query-symbol",
103                  cl::desc("Query a given symbol (e.g. \"a::b::foo\") in\n"
104                           "database directly without parsing the file."),
105                  cl::cat(IncludeFixerCategory));
106 
107 cl::opt<bool>
108     MinimizeIncludePaths("minimize-paths",
109                          cl::desc("Whether to minimize added include paths"),
110                          cl::init(true), cl::cat(IncludeFixerCategory));
111 
112 cl::opt<bool> Quiet("q", cl::desc("Reduce terminal output"), cl::init(false),
113                     cl::cat(IncludeFixerCategory));
114 
115 cl::opt<bool>
116     STDINMode("stdin",
117               cl::desc("Override source file's content (in the overlaying\n"
118                        "virtual file system) with input from <stdin> and run\n"
119                        "the tool on the new content with the compilation\n"
120                        "options of the source file. This mode is currently\n"
121                        "used for editor integration."),
122               cl::init(false), cl::cat(IncludeFixerCategory));
123 
124 cl::opt<bool> OutputHeaders(
125     "output-headers",
126     cl::desc("Print the symbol being queried and all its relevant headers in\n"
127              "JSON format to stdout:\n"
128              "  {\n"
129              "    \"FilePath\": \"/path/to/foo.cc\",\n"
130              "    \"QuerySymbolInfos\": [\n"
131              "       {\"RawIdentifier\": \"foo\",\n"
132              "        \"Range\": {\"Offset\": 0, \"Length\": 3}}\n"
133              "    ],\n"
134              "    \"HeaderInfos\": [ {\"Header\": \"\\\"foo_a.h\\\"\",\n"
135              "                      \"QualifiedName\": \"a::foo\"} ]\n"
136              "  }"),
137     cl::init(false), cl::cat(IncludeFixerCategory));
138 
139 cl::opt<std::string> InsertHeader(
140     "insert-header",
141     cl::desc("Insert a specific header. This should run with STDIN mode.\n"
142              "The result is written to stdout. It is currently used for\n"
143              "editor integration. Support YAML/JSON format:\n"
144              "  -insert-header=\"{\n"
145              "     FilePath: \"/path/to/foo.cc\",\n"
146              "     QuerySymbolInfos: [\n"
147              "       {RawIdentifier: foo,\n"
148              "        Range: {Offset: 0, Length: 3}}\n"
149              "     ],\n"
150              "     HeaderInfos: [ {Headers: \"\\\"foo_a.h\\\"\",\n"
151              "                     QualifiedName: \"a::foo\"} ]}\""),
152     cl::init(""), cl::cat(IncludeFixerCategory));
153 
154 cl::opt<std::string>
155     Style("style",
156           cl::desc("Fallback style for reformatting after inserting new\n"
157                    "headers if there is no clang-format config file found."),
158           cl::init("llvm"), cl::cat(IncludeFixerCategory));
159 
160 std::unique_ptr<include_fixer::SymbolIndexManager>
161 createSymbolIndexManager(StringRef FilePath) {
162   using find_all_symbols::SymbolInfo;
163 
164   auto SymbolIndexMgr = llvm::make_unique<include_fixer::SymbolIndexManager>();
165   switch (DatabaseFormat) {
166   case fixed: {
167     // Parse input and fill the database with it.
168     // <symbol>=<header><, header...>
169     // Multiple symbols can be given, separated by semicolons.
170     std::map<std::string, std::vector<std::string>> SymbolsMap;
171     SmallVector<StringRef, 4> SemicolonSplits;
172     StringRef(Input).split(SemicolonSplits, ";");
173     std::vector<find_all_symbols::SymbolAndSignals> Symbols;
174     for (StringRef Pair : SemicolonSplits) {
175       auto Split = Pair.split('=');
176       std::vector<std::string> Headers;
177       SmallVector<StringRef, 4> CommaSplits;
178       Split.second.split(CommaSplits, ",");
179       for (size_t I = 0, E = CommaSplits.size(); I != E; ++I)
180         Symbols.push_back(
181             {SymbolInfo(Split.first.trim(), SymbolInfo::SymbolKind::Unknown,
182                         CommaSplits[I].trim(), {}),
183              // Use fake "seen" signal for tests, so first header wins.
184              SymbolInfo::Signals(/*Seen=*/static_cast<unsigned>(E - I),
185                                  /*Used=*/0)});
186     }
187     SymbolIndexMgr->addSymbolIndex([=]() {
188       return llvm::make_unique<include_fixer::InMemorySymbolIndex>(Symbols);
189     });
190     break;
191   }
192   case yaml: {
193     auto CreateYamlIdx = [=]() -> std::unique_ptr<include_fixer::SymbolIndex> {
194       llvm::ErrorOr<std::unique_ptr<include_fixer::YamlSymbolIndex>> DB(
195           nullptr);
196       if (!Input.empty()) {
197         DB = include_fixer::YamlSymbolIndex::createFromFile(Input);
198       } else {
199         // If we don't have any input file, look in the directory of the
200         // first
201         // file and its parents.
202         SmallString<128> AbsolutePath(tooling::getAbsolutePath(FilePath));
203         StringRef Directory = llvm::sys::path::parent_path(AbsolutePath);
204         DB = include_fixer::YamlSymbolIndex::createFromDirectory(
205             Directory, "find_all_symbols_db.yaml");
206       }
207 
208       if (!DB) {
209         llvm::errs() << "Couldn't find YAML db: " << DB.getError().message()
210                      << '\n';
211         return nullptr;
212       }
213       return std::move(*DB);
214     };
215 
216     SymbolIndexMgr->addSymbolIndex(std::move(CreateYamlIdx));
217     break;
218   }
219   case fuzzyYaml: {
220     // This mode is not very useful, because we don't correct the identifier.
221     // It's main purpose is to expose FuzzySymbolIndex to tests.
222     SymbolIndexMgr->addSymbolIndex(
223         []() -> std::unique_ptr<include_fixer::SymbolIndex> {
224           auto DB = include_fixer::FuzzySymbolIndex::createFromYAML(Input);
225           if (!DB) {
226             llvm::errs() << "Couldn't load fuzzy YAML db: "
227                          << llvm::toString(DB.takeError()) << '\n';
228             return nullptr;
229           }
230           return std::move(*DB);
231         });
232     break;
233   }
234   }
235   return SymbolIndexMgr;
236 }
237 
238 void writeToJson(llvm::raw_ostream &OS, const IncludeFixerContext& Context) {
239   OS << "{\n"
240      << "  \"FilePath\": \""
241      << llvm::yaml::escape(Context.getFilePath()) << "\",\n"
242      << "  \"QuerySymbolInfos\": [\n";
243   for (const auto &Info : Context.getQuerySymbolInfos()) {
244     OS << "     {\"RawIdentifier\": \"" << Info.RawIdentifier << "\",\n";
245     OS << "      \"Range\":{";
246     OS << "\"Offset\":" << Info.Range.getOffset() << ",";
247     OS << "\"Length\":" << Info.Range.getLength() << "}}";
248     if (&Info != &Context.getQuerySymbolInfos().back())
249       OS << ",\n";
250   }
251   OS << "\n  ],\n";
252   OS << "  \"HeaderInfos\": [\n";
253   const auto &HeaderInfos = Context.getHeaderInfos();
254   for (const auto &Info : HeaderInfos) {
255     OS << "     {\"Header\": \"" << llvm::yaml::escape(Info.Header) << "\",\n"
256        << "      \"QualifiedName\": \"" << Info.QualifiedName << "\"}";
257     if (&Info != &HeaderInfos.back())
258       OS << ",\n";
259   }
260   OS << "\n";
261   OS << "  ]\n";
262   OS << "}\n";
263 }
264 
265 int includeFixerMain(int argc, const char **argv) {
266   tooling::CommonOptionsParser options(argc, argv, IncludeFixerCategory);
267   tooling::ClangTool tool(options.getCompilations(),
268                           options.getSourcePathList());
269 
270   llvm::StringRef SourceFilePath = options.getSourcePathList().front();
271   // In STDINMode, we override the file content with the <stdin> input.
272   // Since `tool.mapVirtualFile` takes `StringRef`, we define `Code` outside of
273   // the if-block so that `Code` is not released after the if-block.
274   std::unique_ptr<llvm::MemoryBuffer> Code;
275   if (STDINMode) {
276     assert(options.getSourcePathList().size() == 1 &&
277            "Expect exactly one file path in STDINMode.");
278     llvm::ErrorOr<std::unique_ptr<llvm::MemoryBuffer>> CodeOrErr =
279         MemoryBuffer::getSTDIN();
280     if (std::error_code EC = CodeOrErr.getError()) {
281       errs() << EC.message() << "\n";
282       return 1;
283     }
284     Code = std::move(CodeOrErr.get());
285     if (Code->getBufferSize() == 0)
286       return 0;  // Skip empty files.
287 
288     tool.mapVirtualFile(SourceFilePath, Code->getBuffer());
289   }
290 
291   if (!InsertHeader.empty()) {
292     if (!STDINMode) {
293       errs() << "Should be running in STDIN mode\n";
294       return 1;
295     }
296 
297     llvm::yaml::Input yin(InsertHeader);
298     IncludeFixerContext Context;
299     yin >> Context;
300 
301     const auto &HeaderInfos = Context.getHeaderInfos();
302     assert(!HeaderInfos.empty());
303     // We only accept one unique header.
304     // Check all elements in HeaderInfos have the same header.
305     bool IsUniqueHeader = std::equal(
306         HeaderInfos.begin()+1, HeaderInfos.end(), HeaderInfos.begin(),
307         [](const IncludeFixerContext::HeaderInfo &LHS,
308            const IncludeFixerContext::HeaderInfo &RHS) {
309           return LHS.Header == RHS.Header;
310         });
311     if (!IsUniqueHeader) {
312       errs() << "Expect exactly one unique header.\n";
313       return 1;
314     }
315 
316     // If a header has multiple symbols, we won't add the missing namespace
317     // qualifiers because we don't know which one is exactly used.
318     //
319     // Check whether all elements in HeaderInfos have the same qualified name.
320     bool IsUniqueQualifiedName = std::equal(
321         HeaderInfos.begin() + 1, HeaderInfos.end(), HeaderInfos.begin(),
322         [](const IncludeFixerContext::HeaderInfo &LHS,
323            const IncludeFixerContext::HeaderInfo &RHS) {
324           return LHS.QualifiedName == RHS.QualifiedName;
325         });
326     auto InsertStyle = format::getStyle(format::DefaultFormatStyle,
327                                         Context.getFilePath(), Style);
328     if (!InsertStyle) {
329       llvm::errs() << llvm::toString(InsertStyle.takeError()) << "\n";
330       return 1;
331     }
332     auto Replacements = clang::include_fixer::createIncludeFixerReplacements(
333         Code->getBuffer(), Context, *InsertStyle,
334         /*AddQualifiers=*/IsUniqueQualifiedName);
335     if (!Replacements) {
336       errs() << "Failed to create replacements: "
337              << llvm::toString(Replacements.takeError()) << "\n";
338       return 1;
339     }
340 
341     auto ChangedCode =
342         tooling::applyAllReplacements(Code->getBuffer(), *Replacements);
343     if (!ChangedCode) {
344       llvm::errs() << llvm::toString(ChangedCode.takeError()) << "\n";
345       return 1;
346     }
347     llvm::outs() << *ChangedCode;
348     return 0;
349   }
350 
351   // Set up data source.
352   std::unique_ptr<include_fixer::SymbolIndexManager> SymbolIndexMgr =
353       createSymbolIndexManager(SourceFilePath);
354   if (!SymbolIndexMgr)
355     return 1;
356 
357   // Query symbol mode.
358   if (!QuerySymbol.empty()) {
359     auto MatchedSymbols = SymbolIndexMgr->search(
360         QuerySymbol, /*IsNestedSearch=*/true, SourceFilePath);
361     for (auto &Symbol : MatchedSymbols) {
362       std::string HeaderPath = Symbol.getFilePath().str();
363       Symbol.SetFilePath(((HeaderPath[0] == '"' || HeaderPath[0] == '<')
364                               ? HeaderPath
365                               : "\"" + HeaderPath + "\""));
366     }
367 
368     // We leave an empty symbol range as we don't know the range of the symbol
369     // being queried in this mode. clang-include-fixer won't add namespace
370     // qualifiers if the symbol range is empty, which also fits this case.
371     IncludeFixerContext::QuerySymbolInfo Symbol;
372     Symbol.RawIdentifier = QuerySymbol;
373     auto Context =
374         IncludeFixerContext(SourceFilePath, {Symbol}, MatchedSymbols);
375     writeToJson(llvm::outs(), Context);
376     return 0;
377   }
378 
379   // Now run our tool.
380   std::vector<include_fixer::IncludeFixerContext> Contexts;
381   include_fixer::IncludeFixerActionFactory Factory(*SymbolIndexMgr, Contexts,
382                                                    Style, MinimizeIncludePaths);
383 
384   if (tool.run(&Factory) != 0) {
385     // We suppress all Clang diagnostics (because they would be wrong,
386     // clang-include-fixer does custom recovery) but still want to give some
387     // feedback in case there was a compiler error we couldn't recover from.
388     // The most common case for this is a #include in the file that couldn't be
389     // found.
390     llvm::errs() << "Fatal compiler error occurred while parsing file!"
391                     " (incorrect include paths?)\n";
392     return 1;
393   }
394 
395   assert(!Contexts.empty());
396 
397   if (OutputHeaders) {
398     // FIXME: Print contexts of all processing files instead of the first one.
399     writeToJson(llvm::outs(), Contexts.front());
400     return 0;
401   }
402 
403   std::vector<tooling::Replacements> FixerReplacements;
404   for (const auto &Context : Contexts) {
405     StringRef FilePath = Context.getFilePath();
406     auto InsertStyle =
407         format::getStyle(format::DefaultFormatStyle, FilePath, Style);
408     if (!InsertStyle) {
409       llvm::errs() << llvm::toString(InsertStyle.takeError()) << "\n";
410       return 1;
411     }
412     auto Buffer = llvm::MemoryBuffer::getFile(FilePath);
413     if (!Buffer) {
414       errs() << "Couldn't open file: " + FilePath.str() + ": "
415              << Buffer.getError().message() + "\n";
416       return 1;
417     }
418 
419     auto Replacements = clang::include_fixer::createIncludeFixerReplacements(
420         Buffer.get()->getBuffer(), Context, *InsertStyle);
421     if (!Replacements) {
422       errs() << "Failed to create replacement: "
423              << llvm::toString(Replacements.takeError()) << "\n";
424       return 1;
425     }
426     FixerReplacements.push_back(*Replacements);
427   }
428 
429   if (!Quiet) {
430     for (const auto &Context : Contexts) {
431       if (!Context.getHeaderInfos().empty()) {
432         llvm::errs() << "Added #include "
433                      << Context.getHeaderInfos().front().Header << " for "
434                      << Context.getFilePath() << "\n";
435       }
436     }
437   }
438 
439   if (STDINMode) {
440     assert(FixerReplacements.size() == 1);
441     auto ChangedCode = tooling::applyAllReplacements(Code->getBuffer(),
442                                                      FixerReplacements.front());
443     if (!ChangedCode) {
444       llvm::errs() << llvm::toString(ChangedCode.takeError()) << "\n";
445       return 1;
446     }
447     llvm::outs() << *ChangedCode;
448     return 0;
449   }
450 
451   // Set up a new source manager for applying the resulting replacements.
452   IntrusiveRefCntPtr<DiagnosticOptions> DiagOpts(new DiagnosticOptions);
453   DiagnosticsEngine Diagnostics(new DiagnosticIDs, &*DiagOpts);
454   TextDiagnosticPrinter DiagnosticPrinter(outs(), &*DiagOpts);
455   SourceManager SM(Diagnostics, tool.getFiles());
456   Diagnostics.setClient(&DiagnosticPrinter, false);
457 
458   // Write replacements to disk.
459   Rewriter Rewrites(SM, LangOptions());
460   for (const auto &Replacement : FixerReplacements) {
461     if (!tooling::applyAllReplacements(Replacement, Rewrites)) {
462       llvm::errs() << "Failed to apply replacements.\n";
463       return 1;
464     }
465   }
466   return Rewrites.overwriteChangedFiles();
467 }
468 
469 } // namespace
470 
471 int main(int argc, const char **argv) {
472   return includeFixerMain(argc, argv);
473 }
474