lib/Checker/PlistDiagnostics.cpp

   1 //===--- PlistDiagnostics.cpp - Plist Diagnostics for Paths -----*- C++ -*-===//
   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 defines the PlistDiagnostics object.
  11 //
  12 //===----------------------------------------------------------------------===//
  13
  14 #include "clang/Checker/PathDiagnosticClients.h"
  15 #include "clang/Checker/BugReporter/PathDiagnostic.h"
  16 #include "clang/Basic/SourceManager.h"
  17 #include "clang/Basic/FileManager.h"
  18 #include "clang/Lex/Preprocessor.h"
  19 #include "llvm/Support/raw_ostream.h"
  20 #include "llvm/Support/Casting.h"
  21 #include "llvm/ADT/DenseMap.h"
  22 #include "llvm/ADT/SmallVector.h"
  23 using namespace clang;
  24 using llvm::cast;
  25
  26 typedef llvm::DenseMap<FileID, unsigned> FIDMap;
  27
  28 namespace clang {
  29   class Preprocessor;
  30 }
  31
  32 namespace {
  33 struct CompareDiagnostics {
  34   // Compare if 'X' is "<" than 'Y'.
  35   bool operator()(const PathDiagnostic *X, const PathDiagnostic *Y) const {
  36     // First compare by location
  37     const FullSourceLoc &XLoc = X->getLocation().asLocation();
  38     const FullSourceLoc &YLoc = Y->getLocation().asLocation();
  39     if (XLoc < YLoc)
  40       return true;
  41     if (XLoc != YLoc)
  42       return false;
  43
  44     // Next, compare by bug type.
  45     llvm::StringRef XBugType = X->getBugType();
  46     llvm::StringRef YBugType = Y->getBugType();
  47     if (XBugType < YBugType)
  48       return true;
  49     if (XBugType != YBugType)
  50       return false;
  51
  52     // Next, compare by bug description.
  53     llvm::StringRef XDesc = X->getDescription();
  54     llvm::StringRef YDesc = Y->getDescription();
  55     if (XDesc < YDesc)
  56       return true;
  57     if (XDesc != YDesc)
  58       return false;
  59
  60     // FIXME: Further refine by comparing PathDiagnosticPieces?
  61     return false;
  62   }
  63 };
  64 }
  65
  66 namespace {
  67   class PlistDiagnostics : public PathDiagnosticClient {
  68     std::vector<const PathDiagnostic*> BatchedDiags;
  69     const std::string OutputFile;
  70     const LangOptions &LangOpts;
  71     llvm::OwningPtr<PathDiagnosticClient> SubPD;
  72     bool flushed;
  73   public:
  74     PlistDiagnostics(const std::string& prefix, const LangOptions &LangOpts,
  75                      PathDiagnosticClient *subPD);
  76
  77     ~PlistDiagnostics() { FlushDiagnostics(NULL); }
  78
  79     void FlushDiagnostics(llvm::SmallVectorImpl<std::string> *FilesMade);
  80
  81     void HandlePathDiagnostic(const PathDiagnostic* D);
  82
  83     virtual llvm::StringRef getName() const {
  84       return "PlistDiagnostics";
  85     }
  86
  87     PathGenerationScheme getGenerationScheme() const;
  88     bool supportsLogicalOpControlFlow() const { return true; }
  89     bool supportsAllBlockEdges() const { return true; }
  90     virtual bool useVerboseDescription() const { return false; }
  91   };
  92 } // end anonymous namespace
  93
  94 PlistDiagnostics::PlistDiagnostics(const std::string& output,
  95                                    const LangOptions &LO,
  96                                    PathDiagnosticClient *subPD)
  97   : OutputFile(output), LangOpts(LO), SubPD(subPD), flushed(false) {}
  98
  99 PathDiagnosticClient*
 100 clang::CreatePlistDiagnosticClient(const std::string& s, const Preprocessor &PP,
 101                                    PathDiagnosticClient *subPD) {
 102   return new PlistDiagnostics(s, PP.getLangOptions(), subPD);
 103 }
 104
 105 PathDiagnosticClient::PathGenerationScheme
 106 PlistDiagnostics::getGenerationScheme() const {
 107   if (const PathDiagnosticClient *PD = SubPD.get())
 108     return PD->getGenerationScheme();
 109
 110   return Extensive;
 111 }
 112
 113 static void AddFID(FIDMap &FIDs, llvm::SmallVectorImpl<FileID> &V,
 114                    const SourceManager* SM, SourceLocation L) {
 115
 116   FileID FID = SM->getFileID(SM->getInstantiationLoc(L));
 117   FIDMap::iterator I = FIDs.find(FID);
 118   if (I != FIDs.end()) return;
 119   FIDs[FID] = V.size();
 120   V.push_back(FID);
 121 }
 122
 123 static unsigned GetFID(const FIDMap& FIDs, const SourceManager &SM,
 124                        SourceLocation L) {
 125   FileID FID = SM.getFileID(SM.getInstantiationLoc(L));
 126   FIDMap::const_iterator I = FIDs.find(FID);
 127   assert(I != FIDs.end());
 128   return I->second;
 129 }
 130
 131 static llvm::raw_ostream& Indent(llvm::raw_ostream& o, const unsigned indent) {
 132   for (unsigned i = 0; i < indent; ++i) o << ' ';
 133   return o;
 134 }
 135
 136 static void EmitLocation(llvm::raw_ostream& o, const SourceManager &SM,
 137                          const LangOptions &LangOpts,
 138                          SourceLocation L, const FIDMap &FM,
 139                          unsigned indent, bool extend = false) {
 140
 141   FullSourceLoc Loc(SM.getInstantiationLoc(L), const_cast<SourceManager&>(SM));
 142
 143   // Add in the length of the token, so that we cover multi-char tokens.
 144   unsigned offset =
 145     extend ? Lexer::MeasureTokenLength(Loc, SM, LangOpts) - 1 : 0;
 146
 147   Indent(o, indent) << "<dict>\n";
 148   Indent(o, indent) << " <key>line</key><integer>"
 149                     << Loc.getInstantiationLineNumber() << "</integer>\n";
 150   Indent(o, indent) << " <key>col</key><integer>"
 151                     << Loc.getInstantiationColumnNumber() + offset << "</integer>\n";
 152   Indent(o, indent) << " <key>file</key><integer>"
 153                     << GetFID(FM, SM, Loc) << "</integer>\n";
 154   Indent(o, indent) << "</dict>\n";
 155 }
 156
 157 static void EmitLocation(llvm::raw_ostream& o, const SourceManager &SM,
 158                          const LangOptions &LangOpts,
 159                          const PathDiagnosticLocation &L, const FIDMap& FM,
 160                          unsigned indent, bool extend = false) {
 161   EmitLocation(o, SM, LangOpts, L.asLocation(), FM, indent, extend);
 162 }
 163
 164 static void EmitRange(llvm::raw_ostream& o, const SourceManager &SM,
 165                       const LangOptions &LangOpts,
 166                       PathDiagnosticRange R, const FIDMap &FM,
 167                       unsigned indent) {
 168   Indent(o, indent) << "<array>\n";
 169   EmitLocation(o, SM, LangOpts, R.getBegin(), FM, indent+1);
 170   EmitLocation(o, SM, LangOpts, R.getEnd(), FM, indent+1, !R.isPoint);
 171   Indent(o, indent) << "</array>\n";
 172 }
 173
 174 static llvm::raw_ostream& EmitString(llvm::raw_ostream& o,
 175                                      const std::string& s) {
 176   o << "<string>";
 177   for (std::string::const_iterator I=s.begin(), E=s.end(); I!=E; ++I) {
 178     char c = *I;
 179     switch (c) {
 180     default:   o << c; break;
 181     case '&':  o << "&amp;"; break;
 182     case '<':  o << "&lt;"; break;
 183     case '>':  o << "&gt;"; break;
 184     case '\'': o << "&apos;"; break;
 185     case '\"': o << "&quot;"; break;
 186     }
 187   }
 188   o << "</string>";
 189   return o;
 190 }
 191
 192 static void ReportControlFlow(llvm::raw_ostream& o,
 193                               const PathDiagnosticControlFlowPiece& P,
 194                               const FIDMap& FM,
 195                               const SourceManager &SM,
 196                               const LangOptions &LangOpts,
 197                               unsigned indent) {
 198
 199   Indent(o, indent) << "<dict>\n";
 200   ++indent;
 201
 202   Indent(o, indent) << "<key>kind</key><string>control</string>\n";
 203
 204   // Emit edges.
 205   Indent(o, indent) << "<key>edges</key>\n";
 206   ++indent;
 207   Indent(o, indent) << "<array>\n";
 208   ++indent;
 209   for (PathDiagnosticControlFlowPiece::const_iterator I=P.begin(), E=P.end();
 210        I!=E; ++I) {
 211     Indent(o, indent) << "<dict>\n";
 212     ++indent;
 213     Indent(o, indent) << "<key>start</key>\n";
 214     EmitRange(o, SM, LangOpts, I->getStart().asRange(), FM, indent+1);
 215     Indent(o, indent) << "<key>end</key>\n";
 216     EmitRange(o, SM, LangOpts, I->getEnd().asRange(), FM, indent+1);
 217     --indent;
 218     Indent(o, indent) << "</dict>\n";
 219   }
 220   --indent;
 221   Indent(o, indent) << "</array>\n";
 222   --indent;
 223
 224   // Output any helper text.
 225   const std::string& s = P.getString();
 226   if (!s.empty()) {
 227     Indent(o, indent) << "<key>alternate</key>";
 228     EmitString(o, s) << '\n';
 229   }
 230
 231   --indent;
 232   Indent(o, indent) << "</dict>\n";
 233 }
 234
 235 static void ReportEvent(llvm::raw_ostream& o, const PathDiagnosticPiece& P,
 236                         const FIDMap& FM,
 237                         const SourceManager &SM,
 238                         const LangOptions &LangOpts,
 239                         unsigned indent) {
 240
 241   Indent(o, indent) << "<dict>\n";
 242   ++indent;
 243
 244   Indent(o, indent) << "<key>kind</key><string>event</string>\n";
 245
 246   // Output the location.
 247   FullSourceLoc L = P.getLocation().asLocation();
 248
 249   Indent(o, indent) << "<key>location</key>\n";
 250   EmitLocation(o, SM, LangOpts, L, FM, indent);
 251
 252   // Output the ranges (if any).
 253   PathDiagnosticPiece::range_iterator RI = P.ranges_begin(),
 254   RE = P.ranges_end();
 255
 256   if (RI != RE) {
 257     Indent(o, indent) << "<key>ranges</key>\n";
 258     Indent(o, indent) << "<array>\n";
 259     ++indent;
 260     for (; RI != RE; ++RI)
 261       EmitRange(o, SM, LangOpts, *RI, FM, indent+1);
 262     --indent;
 263     Indent(o, indent) << "</array>\n";
 264   }
 265
 266   // Output the text.
 267   assert(!P.getString().empty());
 268   Indent(o, indent) << "<key>extended_message</key>\n";
 269   Indent(o, indent);
 270   EmitString(o, P.getString()) << '\n';
 271
 272   // Output the short text.
 273   // FIXME: Really use a short string.
 274   Indent(o, indent) << "<key>message</key>\n";
 275   EmitString(o, P.getString()) << '\n';
 276
 277   // Finish up.
 278   --indent;
 279   Indent(o, indent); o << "</dict>\n";
 280 }
 281
 282 static void ReportMacro(llvm::raw_ostream& o,
 283                         const PathDiagnosticMacroPiece& P,
 284                         const FIDMap& FM, const SourceManager &SM,
 285                         const LangOptions &LangOpts,
 286                         unsigned indent) {
 287
 288   for (PathDiagnosticMacroPiece::const_iterator I=P.begin(), E=P.end();
 289        I!=E; ++I) {
 290
 291     switch ((*I)->getKind()) {
 292     default:
 293       break;
 294     case PathDiagnosticPiece::Event:
 295       ReportEvent(o, cast<PathDiagnosticEventPiece>(**I), FM, SM, LangOpts,
 296                   indent);
 297       break;
 298     case PathDiagnosticPiece::Macro:
 299       ReportMacro(o, cast<PathDiagnosticMacroPiece>(**I), FM, SM, LangOpts,
 300                   indent);
 301       break;
 302     }
 303   }
 304 }
 305
 306 static void ReportDiag(llvm::raw_ostream& o, const PathDiagnosticPiece& P,
 307                        const FIDMap& FM, const SourceManager &SM,
 308                        const LangOptions &LangOpts) {
 309
 310   unsigned indent = 4;
 311
 312   switch (P.getKind()) {
 313   case PathDiagnosticPiece::ControlFlow:
 314     ReportControlFlow(o, cast<PathDiagnosticControlFlowPiece>(P), FM, SM,
 315                       LangOpts, indent);
 316     break;
 317   case PathDiagnosticPiece::Event:
 318     ReportEvent(o, cast<PathDiagnosticEventPiece>(P), FM, SM, LangOpts,
 319                 indent);
 320     break;
 321   case PathDiagnosticPiece::Macro:
 322     ReportMacro(o, cast<PathDiagnosticMacroPiece>(P), FM, SM, LangOpts,
 323                 indent);
 324     break;
 325   }
 326 }
 327
 328 void PlistDiagnostics::HandlePathDiagnostic(const PathDiagnostic* D) {
 329   if (!D)
 330     return;
 331
 332   if (D->empty()) {
 333     delete D;
 334     return;
 335   }
 336
 337   // We need to flatten the locations (convert Stmt* to locations) because
 338   // the referenced statements may be freed by the time the diagnostics
 339   // are emitted.
 340   const_cast<PathDiagnostic*>(D)->flattenLocations();
 341   BatchedDiags.push_back(D);
 342 }
 343
 344 void PlistDiagnostics::FlushDiagnostics(llvm::SmallVectorImpl<std::string>
 345                                         *FilesMade) {
 346
 347   if (flushed)
 348     return;
 349
 350   flushed = true;
 351
 352   // Sort the diagnostics so that they are always emitted in a deterministic
 353   // order.
 354   if (!BatchedDiags.empty())
 355     std::sort(BatchedDiags.begin(), BatchedDiags.end(), CompareDiagnostics());
 356
 357   // Build up a set of FIDs that we use by scanning the locations and
 358   // ranges of the diagnostics.
 359   FIDMap FM;
 360   llvm::SmallVector<FileID, 10> Fids;
 361   const SourceManager* SM = 0;
 362
 363   if (!BatchedDiags.empty())
 364     SM = &(*BatchedDiags.begin())->begin()->getLocation().getManager();
 365
 366   for (std::vector<const PathDiagnostic*>::iterator DI = BatchedDiags.begin(),
 367        DE = BatchedDiags.end(); DI != DE; ++DI) {
 368
 369     const PathDiagnostic *D = *DI;
 370
 371     for (PathDiagnostic::const_iterator I=D->begin(), E=D->end(); I!=E; ++I) {
 372       AddFID(FM, Fids, SM, I->getLocation().asLocation());
 373
 374       for (PathDiagnosticPiece::range_iterator RI=I->ranges_begin(),
 375            RE=I->ranges_end(); RI!=RE; ++RI) {
 376         AddFID(FM, Fids, SM, RI->getBegin());
 377         AddFID(FM, Fids, SM, RI->getEnd());
 378       }
 379     }
 380   }
 381
 382   // Open the file.
 383   std::string ErrMsg;
 384   llvm::raw_fd_ostream o(OutputFile.c_str(), ErrMsg);
 385   if (!ErrMsg.empty()) {
 386     llvm::errs() << "warning: could not creat file: " << OutputFile << '\n';
 387     return;
 388   }
 389
 390   // Write the plist header.
 391   o << "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
 392   "<!DOCTYPE plist PUBLIC \"-//Apple Computer//DTD PLIST 1.0//EN\" "
 393   "\"http://www.apple.com/DTDs/PropertyList-1.0.dtd\">\n"
 394   "<plist version=\"1.0\">\n";
 395
 396   // Write the root object: a <dict> containing...
 397   //  - "files", an <array> mapping from FIDs to file names
 398   //  - "diagnostics", an <array> containing the path diagnostics
 399   o << "<dict>\n"
 400        " <key>files</key>\n"
 401        " <array>\n";
 402
 403   for (llvm::SmallVectorImpl<FileID>::iterator I=Fids.begin(), E=Fids.end();
 404        I!=E; ++I) {
 405     o << "  ";
 406     EmitString(o, SM->getFileEntryForID(*I)->getName()) << '\n';
 407   }
 408
 409   o << " </array>\n"
 410        " <key>diagnostics</key>\n"
 411        " <array>\n";
 412
 413   for (std::vector<const PathDiagnostic*>::iterator DI=BatchedDiags.begin(),
 414        DE = BatchedDiags.end(); DI!=DE; ++DI) {
 415
 416     o << "  <dict>\n"
 417          "   <key>path</key>\n";
 418
 419     const PathDiagnostic *D = *DI;
 420     // Create an owning smart pointer for 'D' just so that we auto-free it
 421     // when we exit this method.
 422     llvm::OwningPtr<PathDiagnostic> OwnedD(const_cast<PathDiagnostic*>(D));
 423
 424     o << "   <array>\n";
 425
 426     for (PathDiagnostic::const_iterator I=D->begin(), E=D->end(); I != E; ++I)
 427       ReportDiag(o, *I, FM, *SM, LangOpts);
 428
 429     o << "   </array>\n";
 430
 431     // Output the bug type and bug category.
 432     o << "   <key>description</key>";
 433     EmitString(o, D->getDescription()) << '\n';
 434     o << "   <key>category</key>";
 435     EmitString(o, D->getCategory()) << '\n';
 436     o << "   <key>type</key>";
 437     EmitString(o, D->getBugType()) << '\n';
 438
 439     // Output the location of the bug.
 440     o << "  <key>location</key>\n";
 441     EmitLocation(o, *SM, LangOpts, D->getLocation(), FM, 2);
 442
 443     // Output the diagnostic to the sub-diagnostic client, if any.
 444     if (SubPD) {
 445       SubPD->HandlePathDiagnostic(OwnedD.take());
 446       llvm::SmallVector<std::string, 1> SubFilesMade;
 447       SubPD->FlushDiagnostics(SubFilesMade);
 448
 449       if (!SubFilesMade.empty()) {
 450         o << "  <key>" << SubPD->getName() << "_files</key>\n";
 451         o << "  <array>\n";
 452         for (size_t i = 0, n = SubFilesMade.size(); i < n ; ++i)
 453           o << "   <string>" << SubFilesMade[i] << "</string>\n";
 454         o << "  </array>\n";
 455       }
 456     }
 457
 458     // Close up the entry.
 459     o << "  </dict>\n";
 460   }
 461
 462   o << " </array>\n";
 463
 464   // Finish.
 465   o << "</dict>\n</plist>";
 466
 467   if (FilesMade)
 468     FilesMade->push_back(OutputFile);
 469
 470   BatchedDiags.clear();
 471 }