1 //===--- QueryDriverDatabase.cpp ---------------------------------*- C++-*-===//
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 // Some compiler drivers have implicit search mechanism for system headers.
9 // This compilation database implementation tries to extract that information by
10 // executing the driver in verbose mode. gcc-compatible drivers print something
11 // like:
12 // ....
13 // ....
14 // #include <...> search starts here:
15 //  /usr/lib/gcc/x86_64-linux-gnu/7/include
16 //  /usr/local/include
17 //  /usr/lib/gcc/x86_64-linux-gnu/7/include-fixed
18 //  /usr/include/x86_64-linux-gnu
19 //  /usr/include
20 // End of search list.
21 // ....
22 // ....
23 // This component parses that output and adds each path to command line args
24 // provided by Base, after prepending them with -isystem. Therefore current
25 // implementation would not work with a driver that is not gcc-compatible.
26 //
27 // First argument of the command line received from underlying compilation
28 // database is used as compiler driver path. Due to this arbitrary binary
29 // execution, this mechanism is not used by default and only executes binaries
30 // in the paths that are explicitly included by the user.
31 
32 #include "GlobalCompilationDatabase.h"
33 #include "support/Logger.h"
34 #include "support/Path.h"
35 #include "support/Trace.h"
36 #include "clang/Basic/Diagnostic.h"
37 #include "clang/Basic/TargetInfo.h"
38 #include "clang/Basic/TargetOptions.h"
39 #include "clang/Driver/Types.h"
40 #include "clang/Tooling/CompilationDatabase.h"
41 #include "llvm/ADT/DenseMap.h"
42 #include "llvm/ADT/ScopeExit.h"
43 #include "llvm/ADT/SmallString.h"
44 #include "llvm/ADT/StringExtras.h"
45 #include "llvm/ADT/StringRef.h"
46 #include "llvm/ADT/iterator_range.h"
47 #include "llvm/Support/FileSystem.h"
48 #include "llvm/Support/MemoryBuffer.h"
49 #include "llvm/Support/Path.h"
50 #include "llvm/Support/Program.h"
51 #include "llvm/Support/Regex.h"
52 #include "llvm/Support/ScopedPrinter.h"
53 #include <algorithm>
54 #include <map>
55 #include <string>
56 #include <vector>
57 
58 namespace clang {
59 namespace clangd {
60 namespace {
61 
62 struct DriverInfo {
63   std::vector<std::string> SystemIncludes;
64   std::string Target;
65 };
66 
67 bool isValidTarget(llvm::StringRef Triple) {
68   std::shared_ptr<TargetOptions> TargetOpts(new TargetOptions);
69   TargetOpts->Triple = Triple.str();
70   DiagnosticsEngine Diags(new DiagnosticIDs, new DiagnosticOptions,
71                           new IgnoringDiagConsumer);
72   IntrusiveRefCntPtr<TargetInfo> Target =
73       TargetInfo::CreateTargetInfo(Diags, TargetOpts);
74   return bool(Target);
75 }
76 
77 llvm::Optional<DriverInfo> parseDriverOutput(llvm::StringRef Output) {
78   DriverInfo Info;
79   const char SIS[] = "#include <...> search starts here:";
80   const char SIE[] = "End of search list.";
81   const char TS[] = "Target: ";
82   llvm::SmallVector<llvm::StringRef, 8> Lines;
83   Output.split(Lines, '\n', /*MaxSplit=*/-1, /*KeepEmpty=*/false);
84 
85   enum {
86     Initial,            // Initial state: searching for target or includes list.
87     IncludesExtracting, // Includes extracting.
88     Done                // Includes and target extraction done.
89   } State = Initial;
90   bool SeenIncludes = false;
91   bool SeenTarget = false;
92   for (auto *It = Lines.begin(); State != Done && It != Lines.end(); ++It) {
93     auto Line = *It;
94     switch (State) {
95     case Initial:
96       if (!SeenIncludes && Line.trim() == SIS) {
97         SeenIncludes = true;
98         State = IncludesExtracting;
99       } else if (!SeenTarget && Line.trim().startswith(TS)) {
100         SeenTarget = true;
101         llvm::StringRef TargetLine = Line.trim();
102         TargetLine.consume_front(TS);
103         // Only detect targets that clang understands
104         if (!isValidTarget(TargetLine)) {
105           elog("System include extraction: invalid target \"{0}\", ignoring",
106                TargetLine);
107         } else {
108           Info.Target = TargetLine.str();
109           vlog("System include extraction: target extracted: \"{0}\"",
110                TargetLine);
111         }
112       }
113       break;
114     case IncludesExtracting:
115       if (Line.trim() == SIE) {
116         State = SeenTarget ? Done : Initial;
117       } else {
118         Info.SystemIncludes.push_back(Line.trim().str());
119         vlog("System include extraction: adding {0}", Line);
120       }
121       break;
122     default:
123       llvm_unreachable("Impossible state of the driver output parser");
124       break;
125     }
126   }
127   if (!SeenIncludes) {
128     elog("System include extraction: start marker not found: {0}", Output);
129     return llvm::None;
130   }
131   if (State == IncludesExtracting) {
132     elog("System include extraction: end marker missing: {0}", Output);
133     return llvm::None;
134   }
135   return std::move(Info);
136 }
137 
138 llvm::Optional<DriverInfo>
139 extractSystemIncludesAndTarget(PathRef Driver, llvm::StringRef Lang,
140                                llvm::ArrayRef<std::string> CommandLine,
141                                const llvm::Regex &QueryDriverRegex) {
142   trace::Span Tracer("Extract system includes and target");
143   SPAN_ATTACH(Tracer, "driver", Driver);
144   SPAN_ATTACH(Tracer, "lang", Lang);
145 
146   if (!QueryDriverRegex.match(Driver)) {
147     vlog("System include extraction: not allowed driver {0}", Driver);
148     return llvm::None;
149   }
150 
151   if (!llvm::sys::fs::exists(Driver)) {
152     elog("System include extraction: {0} does not exist.", Driver);
153     return llvm::None;
154   }
155   if (!llvm::sys::fs::can_execute(Driver)) {
156     elog("System include extraction: {0} is not executable.", Driver);
157     return llvm::None;
158   }
159 
160   llvm::SmallString<128> StdErrPath;
161   if (auto EC = llvm::sys::fs::createTemporaryFile("system-includes", "clangd",
162                                                    StdErrPath)) {
163     elog("System include extraction: failed to create temporary file with "
164          "error {0}",
165          EC.message());
166     return llvm::None;
167   }
168   auto CleanUp = llvm::make_scope_exit(
169       [&StdErrPath]() { llvm::sys::fs::remove(StdErrPath); });
170 
171   llvm::Optional<llvm::StringRef> Redirects[] = {
172       {""}, {""}, llvm::StringRef(StdErrPath)};
173 
174   llvm::SmallVector<llvm::StringRef, 12> Args = {Driver, "-E", "-x",
175                                                  Lang,   "-",  "-v"};
176 
177   // These flags will be preserved
178   const llvm::StringRef FlagsToPreserve[] = {
179       "-nostdinc", "--no-standard-includes", "-nostdinc++", "-nobuiltininc"};
180   // Preserves these flags and their values, either as separate args or with an
181   // equalsbetween them
182   const llvm::StringRef ArgsToPreserve[] = {"--sysroot", "-isysroot"};
183 
184   for (size_t I = 0, E = CommandLine.size(); I < E; ++I) {
185     llvm::StringRef Arg = CommandLine[I];
186     if (llvm::any_of(FlagsToPreserve,
187                      [&Arg](llvm::StringRef S) { return S == Arg; })) {
188       Args.push_back(Arg);
189     } else {
190       const auto *Found =
191           llvm::find_if(ArgsToPreserve, [&Arg](llvm::StringRef S) {
192             return Arg.startswith(S);
193           });
194       if (Found == std::end(ArgsToPreserve))
195         continue;
196       Arg = Arg.drop_front(Found->size());
197       if (Arg.empty() && I + 1 < E) {
198         Args.push_back(CommandLine[I]);
199         Args.push_back(CommandLine[++I]);
200       } else if (Arg.startswith("=")) {
201         Args.push_back(CommandLine[I]);
202       }
203     }
204   }
205 
206   if (int RC = llvm::sys::ExecuteAndWait(Driver, Args, /*Env=*/llvm::None,
207                                          Redirects)) {
208     elog("System include extraction: driver execution failed with return code: "
209          "{0}. Args: ['{1}']",
210          llvm::to_string(RC), llvm::join(Args, "', '"));
211     return llvm::None;
212   }
213 
214   auto BufOrError = llvm::MemoryBuffer::getFile(StdErrPath);
215   if (!BufOrError) {
216     elog("System include extraction: failed to read {0} with error {1}",
217          StdErrPath, BufOrError.getError().message());
218     return llvm::None;
219   }
220 
221   llvm::Optional<DriverInfo> Info =
222       parseDriverOutput(BufOrError->get()->getBuffer());
223   if (!Info)
224     return llvm::None;
225   log("System includes extractor: successfully executed {0}\n\tgot includes: "
226       "\"{1}\"\n\tgot target: \"{2}\"",
227       Driver, llvm::join(Info->SystemIncludes, ", "), Info->Target);
228   return Info;
229 }
230 
231 tooling::CompileCommand &
232 addSystemIncludes(tooling::CompileCommand &Cmd,
233                   llvm::ArrayRef<std::string> SystemIncludes) {
234   for (llvm::StringRef Include : SystemIncludes) {
235     // FIXME(kadircet): This doesn't work when we have "--driver-mode=cl"
236     Cmd.CommandLine.push_back("-isystem");
237     Cmd.CommandLine.push_back(Include.str());
238   }
239   return Cmd;
240 }
241 
242 tooling::CompileCommand &setTarget(tooling::CompileCommand &Cmd,
243                                    const std::string &Target) {
244   if (!Target.empty()) {
245     // We do not want to override existing target with extracted one.
246     for (llvm::StringRef Arg : Cmd.CommandLine) {
247       if (Arg == "-target" || Arg.startswith("--target="))
248         return Cmd;
249     }
250     Cmd.CommandLine.push_back("--target=" + Target);
251   }
252   return Cmd;
253 }
254 
255 /// Converts a glob containing only ** or * into a regex.
256 std::string convertGlobToRegex(llvm::StringRef Glob) {
257   std::string RegText;
258   llvm::raw_string_ostream RegStream(RegText);
259   RegStream << '^';
260   for (size_t I = 0, E = Glob.size(); I < E; ++I) {
261     if (Glob[I] == '*') {
262       if (I + 1 < E && Glob[I + 1] == '*') {
263         // Double star, accept any sequence.
264         RegStream << ".*";
265         // Also skip the second star.
266         ++I;
267       } else {
268         // Single star, accept any sequence without a slash.
269         RegStream << "[^/]*";
270       }
271     } else {
272       RegStream << llvm::Regex::escape(Glob.substr(I, 1));
273     }
274   }
275   RegStream << '$';
276   RegStream.flush();
277   return RegText;
278 }
279 
280 /// Converts a glob containing only ** or * into a regex.
281 llvm::Regex convertGlobsToRegex(llvm::ArrayRef<std::string> Globs) {
282   assert(!Globs.empty() && "Globs cannot be empty!");
283   std::vector<std::string> RegTexts;
284   RegTexts.reserve(Globs.size());
285   for (llvm::StringRef Glob : Globs)
286     RegTexts.push_back(convertGlobToRegex(Glob));
287 
288   llvm::Regex Reg(llvm::join(RegTexts, "|"));
289   assert(Reg.isValid(RegTexts.front()) &&
290          "Created an invalid regex from globs");
291   return Reg;
292 }
293 
294 /// Extracts system includes from a trusted driver by parsing the output of
295 /// include search path and appends them to the commands coming from underlying
296 /// compilation database.
297 class QueryDriverDatabase : public GlobalCompilationDatabase {
298 public:
299   QueryDriverDatabase(llvm::ArrayRef<std::string> QueryDriverGlobs,
300                       std::unique_ptr<GlobalCompilationDatabase> Base)
301       : QueryDriverRegex(convertGlobsToRegex(QueryDriverGlobs)),
302         Base(std::move(Base)) {
303     assert(this->Base);
304     BaseChanged =
305         this->Base->watch([this](const std::vector<std::string> &Changes) {
306           OnCommandChanged.broadcast(Changes);
307         });
308   }
309 
310   llvm::Optional<tooling::CompileCommand>
311   getCompileCommand(PathRef File) const override {
312     auto Cmd = Base->getCompileCommand(File);
313     if (!Cmd || Cmd->CommandLine.empty())
314       return Cmd;
315 
316     llvm::StringRef Lang;
317     for (size_t I = 0, E = Cmd->CommandLine.size(); I < E; ++I) {
318       llvm::StringRef Arg = Cmd->CommandLine[I];
319       if (Arg == "-x" && I + 1 < E)
320         Lang = Cmd->CommandLine[I + 1];
321       else if (Arg.startswith("-x"))
322         Lang = Arg.drop_front(2).trim();
323     }
324     if (Lang.empty()) {
325       llvm::StringRef Ext = llvm::sys::path::extension(File).trim('.');
326       auto Type = driver::types::lookupTypeForExtension(Ext);
327       if (Type == driver::types::TY_INVALID) {
328         elog("System include extraction: invalid file type for {0}", Ext);
329         return {};
330       }
331       Lang = driver::types::getTypeName(Type);
332     }
333 
334     llvm::SmallString<128> Driver(Cmd->CommandLine.front());
335     llvm::sys::fs::make_absolute(Cmd->Directory, Driver);
336 
337     if (auto Info =
338             QueriedDrivers.get(/*Key=*/(Driver + ":" + Lang).str(), [&] {
339               return extractSystemIncludesAndTarget(
340                   Driver, Lang, Cmd->CommandLine, QueryDriverRegex);
341             })) {
342       setTarget(addSystemIncludes(*Cmd, Info->SystemIncludes), Info->Target);
343     }
344     return Cmd;
345   }
346 
347   llvm::Optional<ProjectInfo> getProjectInfo(PathRef File) const override {
348     return Base->getProjectInfo(File);
349   }
350 
351 private:
352   // Caches includes extracted from a driver. Key is driver:lang.
353   Memoize<llvm::StringMap<llvm::Optional<DriverInfo>>> QueriedDrivers;
354   llvm::Regex QueryDriverRegex;
355 
356   std::unique_ptr<GlobalCompilationDatabase> Base;
357   CommandChanged::Subscription BaseChanged;
358 };
359 } // namespace
360 
361 std::unique_ptr<GlobalCompilationDatabase>
362 getQueryDriverDatabase(llvm::ArrayRef<std::string> QueryDriverGlobs,
363                        std::unique_ptr<GlobalCompilationDatabase> Base) {
364   assert(Base && "Null base to SystemIncludeExtractor");
365   if (QueryDriverGlobs.empty())
366     return Base;
367   return std::make_unique<QueryDriverDatabase>(QueryDriverGlobs,
368                                                std::move(Base));
369 }
370 
371 } // namespace clangd
372 } // namespace clang
373