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 "CompileCommands.h"
33 #include "GlobalCompilationDatabase.h"
34 #include "support/Logger.h"
35 #include "support/Path.h"
36 #include "support/Trace.h"
37 #include "clang/Basic/Diagnostic.h"
38 #include "clang/Basic/TargetInfo.h"
39 #include "clang/Basic/TargetOptions.h"
40 #include "clang/Driver/Types.h"
41 #include "clang/Tooling/CompilationDatabase.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/Support/FileSystem.h"
47 #include "llvm/Support/MemoryBuffer.h"
48 #include "llvm/Support/Path.h"
49 #include "llvm/Support/Program.h"
50 #include "llvm/Support/Regex.h"
51 #include "llvm/Support/ScopedPrinter.h"
52 #include <algorithm>
53 #include <iterator>
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> 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(llvm::SmallString<128> Driver,
140                                llvm::StringRef Lang,
141                                llvm::ArrayRef<std::string> CommandLine,
142                                const llvm::Regex &QueryDriverRegex) {
143   trace::Span Tracer("Extract system includes and target");
144 
145   if (!llvm::sys::path::is_absolute(Driver)) {
146     assert(llvm::none_of(
147         Driver, [](char C) { return llvm::sys::path::is_separator(C); }));
148     auto DriverProgram = llvm::sys::findProgramByName(Driver);
149     if (DriverProgram) {
150       vlog("System include extraction: driver {0} expanded to {1}", Driver,
151            *DriverProgram);
152       Driver = *DriverProgram;
153     } else {
154       elog("System include extraction: driver {0} not found in PATH", Driver);
155       return llvm::None;
156     }
157   }
158 
159   SPAN_ATTACH(Tracer, "driver", Driver);
160   SPAN_ATTACH(Tracer, "lang", Lang);
161 
162   if (!QueryDriverRegex.match(Driver)) {
163     vlog("System include extraction: not allowed driver {0}", Driver);
164     return llvm::None;
165   }
166 
167   llvm::SmallString<128> StdErrPath;
168   if (auto EC = llvm::sys::fs::createTemporaryFile("system-includes", "clangd",
169                                                    StdErrPath)) {
170     elog("System include extraction: failed to create temporary file with "
171          "error {0}",
172          EC.message());
173     return llvm::None;
174   }
175   auto CleanUp = llvm::make_scope_exit(
176       [&StdErrPath]() { llvm::sys::fs::remove(StdErrPath); });
177 
178   llvm::Optional<llvm::StringRef> Redirects[] = {{""}, {""}, StdErrPath.str()};
179 
180   llvm::SmallVector<llvm::StringRef> Args = {Driver, "-E", "-x",
181                                              Lang,   "-",  "-v"};
182 
183   // These flags will be preserved
184   const llvm::StringRef FlagsToPreserve[] = {
185       "-nostdinc", "--no-standard-includes", "-nostdinc++", "-nobuiltininc"};
186   // Preserves these flags and their values, either as separate args or with an
187   // equalsbetween them
188   const llvm::StringRef ArgsToPreserve[] = {"--sysroot", "-isysroot"};
189 
190   for (size_t I = 0, E = CommandLine.size(); I < E; ++I) {
191     llvm::StringRef Arg = CommandLine[I];
192     if (llvm::any_of(FlagsToPreserve,
193                      [&Arg](llvm::StringRef S) { return S == Arg; })) {
194       Args.push_back(Arg);
195     } else {
196       const auto *Found =
197           llvm::find_if(ArgsToPreserve, [&Arg](llvm::StringRef S) {
198             return Arg.startswith(S);
199           });
200       if (Found == std::end(ArgsToPreserve))
201         continue;
202       Arg = Arg.drop_front(Found->size());
203       if (Arg.empty() && I + 1 < E) {
204         Args.push_back(CommandLine[I]);
205         Args.push_back(CommandLine[++I]);
206       } else if (Arg.startswith("=")) {
207         Args.push_back(CommandLine[I]);
208       }
209     }
210   }
211 
212   std::string ErrMsg;
213   if (int RC = llvm::sys::ExecuteAndWait(Driver, Args, /*Env=*/llvm::None,
214                                          Redirects, /*SecondsToWait=*/0,
215                                          /*MemoryLimit=*/0, &ErrMsg)) {
216     elog("System include extraction: driver execution failed with return code: "
217          "{0} - '{1}'. Args: [{2}]",
218          llvm::to_string(RC), ErrMsg, printArgv(Args));
219     return llvm::None;
220   }
221 
222   auto BufOrError = llvm::MemoryBuffer::getFile(StdErrPath);
223   if (!BufOrError) {
224     elog("System include extraction: failed to read {0} with error {1}",
225          StdErrPath, BufOrError.getError().message());
226     return llvm::None;
227   }
228 
229   llvm::Optional<DriverInfo> Info =
230       parseDriverOutput(BufOrError->get()->getBuffer());
231   if (!Info)
232     return llvm::None;
233   log("System includes extractor: successfully executed {0}\n\tgot includes: "
234       "\"{1}\"\n\tgot target: \"{2}\"",
235       Driver, llvm::join(Info->SystemIncludes, ", "), Info->Target);
236   return Info;
237 }
238 
239 tooling::CompileCommand &
240 addSystemIncludes(tooling::CompileCommand &Cmd,
241                   llvm::ArrayRef<std::string> SystemIncludes) {
242   std::vector<std::string> ToAppend;
243   for (llvm::StringRef Include : SystemIncludes) {
244     // FIXME(kadircet): This doesn't work when we have "--driver-mode=cl"
245     ToAppend.push_back("-isystem");
246     ToAppend.push_back(Include.str());
247   }
248   if (!ToAppend.empty()) {
249     // Just append when `--` isn't present.
250     auto InsertAt = llvm::find(Cmd.CommandLine, "--");
251     Cmd.CommandLine.insert(InsertAt, std::make_move_iterator(ToAppend.begin()),
252                            std::make_move_iterator(ToAppend.end()));
253   }
254   return Cmd;
255 }
256 
257 tooling::CompileCommand &setTarget(tooling::CompileCommand &Cmd,
258                                    const std::string &Target) {
259   if (!Target.empty()) {
260     // We do not want to override existing target with extracted one.
261     for (llvm::StringRef Arg : Cmd.CommandLine) {
262       if (Arg == "-target" || Arg.startswith("--target="))
263         return Cmd;
264     }
265     // Just append when `--` isn't present.
266     auto InsertAt = llvm::find(Cmd.CommandLine, "--");
267     Cmd.CommandLine.insert(InsertAt, "--target=" + Target);
268   }
269   return Cmd;
270 }
271 
272 /// Converts a glob containing only ** or * into a regex.
273 std::string convertGlobToRegex(llvm::StringRef Glob) {
274   std::string RegText;
275   llvm::raw_string_ostream RegStream(RegText);
276   RegStream << '^';
277   for (size_t I = 0, E = Glob.size(); I < E; ++I) {
278     if (Glob[I] == '*') {
279       if (I + 1 < E && Glob[I + 1] == '*') {
280         // Double star, accept any sequence.
281         RegStream << ".*";
282         // Also skip the second star.
283         ++I;
284       } else {
285         // Single star, accept any sequence without a slash.
286         RegStream << "[^/]*";
287       }
288     } else if (llvm::sys::path::is_separator(Glob[I]) &&
289                llvm::sys::path::is_separator('/') &&
290                llvm::sys::path::is_separator('\\')) {
291       RegStream << R"([/\\])"; // Accept either slash on windows.
292     } else {
293       RegStream << llvm::Regex::escape(Glob.substr(I, 1));
294     }
295   }
296   RegStream << '$';
297   RegStream.flush();
298   return RegText;
299 }
300 
301 /// Converts a glob containing only ** or * into a regex.
302 llvm::Regex convertGlobsToRegex(llvm::ArrayRef<std::string> Globs) {
303   assert(!Globs.empty() && "Globs cannot be empty!");
304   std::vector<std::string> RegTexts;
305   RegTexts.reserve(Globs.size());
306   for (llvm::StringRef Glob : Globs)
307     RegTexts.push_back(convertGlobToRegex(Glob));
308 
309   // Tempting to pass IgnoreCase, but we don't know the FS sensitivity.
310   llvm::Regex Reg(llvm::join(RegTexts, "|"));
311   assert(Reg.isValid(RegTexts.front()) &&
312          "Created an invalid regex from globs");
313   return Reg;
314 }
315 
316 /// Extracts system includes from a trusted driver by parsing the output of
317 /// include search path and appends them to the commands coming from underlying
318 /// compilation database.
319 class QueryDriverDatabase : public DelegatingCDB {
320 public:
321   QueryDriverDatabase(llvm::ArrayRef<std::string> QueryDriverGlobs,
322                       std::unique_ptr<GlobalCompilationDatabase> Base)
323       : DelegatingCDB(std::move(Base)),
324         QueryDriverRegex(convertGlobsToRegex(QueryDriverGlobs)) {}
325 
326   llvm::Optional<tooling::CompileCommand>
327   getCompileCommand(PathRef File) const override {
328     auto Cmd = DelegatingCDB::getCompileCommand(File);
329     if (!Cmd || Cmd->CommandLine.empty())
330       return Cmd;
331 
332     llvm::StringRef Lang;
333     for (size_t I = 0, E = Cmd->CommandLine.size(); I < E; ++I) {
334       llvm::StringRef Arg = Cmd->CommandLine[I];
335       if (Arg == "-x" && I + 1 < E)
336         Lang = Cmd->CommandLine[I + 1];
337       else if (Arg.startswith("-x"))
338         Lang = Arg.drop_front(2).trim();
339     }
340     if (Lang.empty()) {
341       llvm::StringRef Ext = llvm::sys::path::extension(File).trim('.');
342       auto Type = driver::types::lookupTypeForExtension(Ext);
343       if (Type == driver::types::TY_INVALID) {
344         elog("System include extraction: invalid file type for {0}", Ext);
345         return {};
346       }
347       Lang = driver::types::getTypeName(Type);
348     }
349 
350     llvm::SmallString<128> Driver(Cmd->CommandLine.front());
351     if (llvm::any_of(Driver,
352                        [](char C) { return llvm::sys::path::is_separator(C); }))
353       // Driver is a not a single executable name but instead a path (either
354       // relative or absolute).
355       llvm::sys::fs::make_absolute(Cmd->Directory, Driver);
356 
357     if (auto Info =
358             QueriedDrivers.get(/*Key=*/(Driver + ":" + Lang).str(), [&] {
359               return extractSystemIncludesAndTarget(
360                   Driver, Lang, Cmd->CommandLine, QueryDriverRegex);
361             })) {
362       setTarget(addSystemIncludes(*Cmd, Info->SystemIncludes), Info->Target);
363     }
364     return Cmd;
365   }
366 
367 private:
368   // Caches includes extracted from a driver. Key is driver:lang.
369   Memoize<llvm::StringMap<llvm::Optional<DriverInfo>>> QueriedDrivers;
370   llvm::Regex QueryDriverRegex;
371 };
372 } // namespace
373 
374 std::unique_ptr<GlobalCompilationDatabase>
375 getQueryDriverDatabase(llvm::ArrayRef<std::string> QueryDriverGlobs,
376                        std::unique_ptr<GlobalCompilationDatabase> Base) {
377   assert(Base && "Null base to SystemIncludeExtractor");
378   if (QueryDriverGlobs.empty())
379     return Base;
380   return std::make_unique<QueryDriverDatabase>(QueryDriverGlobs,
381                                                std::move(Base));
382 }
383 
384 } // namespace clangd
385 } // namespace clang
386