lib/StaticAnalyzer/EntoSA/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/StaticAnalyzer/PathDiagnosticClients.h"
  15 #include "clang/StaticAnalyzer/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 namespace ento;
  25 using llvm::cast;
  26
  27 typedef llvm::DenseMap<FileID, unsigned> FIDMap;
  28
  29 namespace clang {
  30   class Preprocessor;
  31 }
  32
  33 namespace {
  34 struct CompareDiagnostics {
  35   // Compare if 'X' is "<" than 'Y'.
  36   bool operator()(const PathDiagnostic *X, const PathDiagnostic *Y) const {
  37     // First compare by location
  38     const FullSourceLoc &XLoc = X->getLocation().asLocation();
  39     const FullSourceLoc &YLoc = Y->getLocation().asLocation();
  40     if (XLoc < YLoc)
  41       return true;
  42     if (XLoc != YLoc)
  43       return false;
  44
  45     // Next, compare by bug type.
  46     llvm::StringRef XBugType = X->getBugType();
  47     llvm::StringRef YBugType = Y->getBugType();
  48     if (XBugType < YBugType)
  49       return true;
  50     if (XBugType != YBugType)
  51       return false;
  52
  53     // Next, compare by bug description.
  54     llvm::StringRef XDesc = X->getDescription();
  55     llvm::StringRef YDesc = Y->getDescription();
  56     if (XDesc < YDesc)
  57       return true;
  58     if (XDesc != YDesc)
  59       return false;
  60
  61     // FIXME: Further refine by comparing PathDiagnosticPieces?
  62     return false;
  63   }
  64 };
  65 }
  66
  67 namespace {
  68   class PlistDiagnostics : public PathDiagnosticClient {
  69     std::vector<const PathDiagnostic*> BatchedDiags;
  70     const std::string OutputFile;
  71     const LangOptions &LangOpts;
  72     llvm::OwningPtr<PathDiagnosticClient> SubPD;
  73     bool flushed;
  74   public:
  75     PlistDiagnostics(const std::string& prefix, const LangOptions &LangOpts,
  76                      PathDiagnosticClient *subPD);
  77
  78     ~PlistDiagnostics() { FlushDiagnostics(NULL); }
  79
  80     void FlushDiagnostics(llvm::SmallVectorImpl<std::string> *FilesMade);
  81
  82     void HandlePathDiagnostic(const PathDiagnostic* D);
  83
  84     virtual llvm::StringRef getName() const {
  85       return "PlistDiagnostics";
  86     }
  87
  88     PathGenerationScheme getGenerationScheme() const;
  89     bool supportsLogicalOpControlFlow() const { return true; }
  90     bool supportsAllBlockEdges() const { return true; }
  91     virtual bool useVerboseDescription() const { return false; }
  92   };
  93 } // end anonymous namespace
  94
  95 PlistDiagnostics::PlistDiagnostics(const std::string& output,
  96                                    const LangOptions &LO,
  97                                    PathDiagnosticClient *subPD)
  98   : OutputFile(output), LangOpts(LO), SubPD(subPD), flushed(false) {}
  99
 100 PathDiagnosticClient*
 101 ento::createPlistDiagnosticClient(const std::string& s, const Preprocessor &PP,
 102                                   PathDiagnosticClient *subPD) {
 103   return new PlistDiagnostics(s, PP.getLangOptions(), subPD);
 104 }
 105
 106 PathDiagnosticClient::PathGenerationScheme
 107 PlistDiagnostics::getGenerationScheme() const {
 108   if (const PathDiagnosticClient *PD = SubPD.get())
 109     return PD->getGenerationScheme();
 110
 111   return Extensive;
 112 }
 113
 114 static void AddFID(FIDMap &FIDs, llvm::SmallVectorImpl<FileID> &V,
 115                    const SourceManager* SM, SourceLocation L) {
 116
 117   FileID FID = SM->getFileID(SM->getInstantiationLoc(L));
 118   FIDMap::iterator I = FIDs.find(FID);
 119   if (I != FIDs.end()) return;
 120   FIDs[FID] = V.size();
 121   V.push_back(FID);
 122 }
 123
 124 static unsigned GetFID(const FIDMap& FIDs, const SourceManager &SM,
 125                        SourceLocation L) {
 126   FileID FID = SM.getFileID(SM.getInstantiationLoc(L));
 127   FIDMap::const_iterator I = FIDs.find(FID);
 128   assert(I != FIDs.end());
 129   return I->second;
 130 }
 131
 132 static llvm::raw_ostream& Indent(llvm::raw_ostream& o, const unsigned indent) {
 133   for (unsigned i = 0; i < indent; ++i) o << ' ';
 134   return o;
 135 }
 136
 137 static void EmitLocation(llvm::raw_ostream& o, const SourceManager &SM,
 138                          const LangOptions &LangOpts,
 139                          SourceLocation L, const FIDMap &FM,
 140                          unsigned indent, bool extend = false) {
 141
 142   FullSourceLoc Loc(SM.getInstantiationLoc(L), const_cast<SourceManager&>(SM));
 143
 144   // Add in the length of the token, so that we cover multi-char tokens.
 145   unsigned offset =
 146     extend ? Lexer::MeasureTokenLength(Loc, SM, LangOpts) - 1 : 0;
 147
 148   Indent(o, indent) << "<dict>\n";
 149   Indent(o, indent) << " <key>line</key><integer>"
 150                     << Loc.getInstantiationLineNumber() << "</integer>\n";
 151   Indent(o, indent) << " <key>col</key><integer>"
 152                     << Loc.getInstantiationColumnNumber() + offset << "</integer>\n";
 153   Indent(o, indent) << " <key>file</key><integer>"
 154                     << GetFID(FM, SM, Loc) << "</integer>\n";
 155   Indent(o, indent) << "</dict>\n";
 156 }
 157
 158 static void EmitLocation(llvm::raw_ostream& o, const SourceManager &SM,
 159                          const LangOptions &LangOpts,
 160                          const PathDiagnosticLocation &L, const FIDMap& FM,
 161                          unsigned indent, bool extend = false) {
 162   EmitLocation(o, SM, LangOpts, L.asLocation(), FM, indent, extend);
 163 }
 164
 165 static void EmitRange(llvm::raw_ostream& o, const SourceManager &SM,
 166                       const LangOptions &LangOpts,
 167                       PathDiagnosticRange R, const FIDMap &FM,
 168                       unsigned indent) {
 169   Indent(o, indent) << "<array>\n";
 170   EmitLocation(o, SM, LangOpts, R.getBegin(), FM, indent+1);
 171   EmitLocation(o, SM, LangOpts, R.getEnd(), FM, indent+1, !R.isPoint);
 172   Indent(o, indent) << "</array>\n";
 173 }
 174
 175 static llvm::raw_ostream& EmitString(llvm::raw_ostream& o,
 176                                      const std::string& s) {
 177   o << "<string>";
 178   for (std::string::const_iterator I=s.begin(), E=s.end(); I!=E; ++I) {
 179     char c = *I;
 180     switch (c) {
 181     default:   o << c; break;
 182     case '&':  o << "&amp;"; break;
 183     case '<':  o << "&lt;"; break;
 184     case '>':  o << "&gt;"; break;
 185     case '\'': o << "&apos;"; break;
 186     case '\"': o << "&quot;"; break;
 187     }
 188   }
 189   o << "</string>";
 190   return o;
 191 }
 192
 193 static void ReportControlFlow(llvm::raw_ostream& o,
 194                               const PathDiagnosticControlFlowPiece& P,
 195                               const FIDMap& FM,
 196                               const SourceManager &SM,
 197                               const LangOptions &LangOpts,
 198                               unsigned indent) {
 199
 200   Indent(o, indent) << "<dict>\n";
 201   ++indent;
 202
 203   Indent(o, indent) << "<key>kind</key><string>control</string>\n";
 204
 205   // Emit edges.
 206   Indent(o, indent) << "<key>edges</key>\n";
 207   ++indent;
 208   Indent(o, indent) << "<array>\n";
 209   ++indent;
 210   for (PathDiagnosticControlFlowPiece::const_iterator I=P.begin(), E=P.end();
 211        I!=E; ++I) {
 212     Indent(o, indent) << "<dict>\n";
 213     ++indent;
 214     Indent(o, indent) << "<key>start</key>\n";
 215     EmitRange(o, SM, LangOpts, I->getStart().asRange(), FM, indent+1);
 216     Indent(o, indent) << "<key>end</key>\n";
 217     EmitRange(o, SM, LangOpts, I->getEnd().asRange(), FM, indent+1);
 218     --indent;
 219     Indent(o, indent) << "</dict>\n";
 220   }
 221   --indent;
 222   Indent(o, indent) << "</array>\n";
 223   --indent;
 224
 225   // Output any helper text.
 226   const std::string& s = P.getString();
 227   if (!s.empty()) {
 228     Indent(o, indent) << "<key>alternate</key>";
 229     EmitString(o, s) << '\n';
 230   }
 231
 232   --indent;
 233   Indent(o, indent) << "</dict>\n";
 234 }
 235
 236 static void ReportEvent(llvm::raw_ostream& o, const PathDiagnosticPiece& P,
 237                         const FIDMap& FM,
 238                         const SourceManager &SM,
 239                         const LangOptions &LangOpts,
 240                         unsigned indent) {
 241
 242   Indent(o, indent) << "<dict>\n";
 243   ++indent;
 244
 245   Indent(o, indent) << "<key>kind</key><string>event</string>\n";
 246
 247   // Output the location.
 248   FullSourceLoc L = P.getLocation().asLocation();
 249
 250   Indent(o, indent) << "<key>location</key>\n";
 251   EmitLocation(o, SM, LangOpts, L, FM, indent);
 252
 253   // Output the ranges (if any).
 254   PathDiagnosticPiece::range_iterator RI = P.ranges_begin(),
 255   RE = P.ranges_end();
 256
 257   if (RI != RE) {
 258     Indent(o, indent) << "<key>ranges</key>\n";
 259     Indent(o, indent) << "<array>\n";
 260     ++indent;
 261     for (; RI != RE; ++RI)
 262       EmitRange(o, SM, LangOpts, *RI, FM, indent+1);
 263     --indent;
 264     Indent(o, indent) << "</array>\n";
 265   }
 266
 267   // Output the text.
 268   assert(!P.getString().empty());
 269   Indent(o, indent) << "<key>extended_message</key>\n";
 270   Indent(o, indent);
 271   EmitString(o, P.getString()) << '\n';
 272
 273   // Output the short text.
 274   // FIXME: Really use a short string.
 275   Indent(o, indent) << "<key>message</key>\n";
 276   EmitString(o, P.getString()) << '\n';
 277
 278   // Finish up.
 279   --indent;
 280   Indent(o, indent); o << "</dict>\n";
 281 }
 282
 283 static void ReportMacro(llvm::raw_ostream& o,
 284                         const PathDiagnosticMacroPiece& P,
 285                         const FIDMap& FM, const SourceManager &SM,
 286                         const LangOptions &LangOpts,
 287                         unsigned indent) {
 288
 289   for (PathDiagnosticMacroPiece::const_iterator I=P.begin(), E=P.end();
 290        I!=E; ++I) {
 291
 292     switch ((*I)->getKind()) {
 293     default:
 294       break;
 295     case PathDiagnosticPiece::Event:
 296       ReportEvent(o, cast<PathDiagnosticEventPiece>(**I), FM, SM, LangOpts,
 297                   indent);
 298       break;
 299     case PathDiagnosticPiece::Macro:
 300       ReportMacro(o, cast<PathDiagnosticMacroPiece>(**I), FM, SM, LangOpts,
 301                   indent);
 302       break;
 303     }
 304   }
 305 }
 306
 307 static void ReportDiag(llvm::raw_ostream& o, const PathDiagnosticPiece& P,
 308                        const FIDMap& FM, const SourceManager &SM,
 309                        const LangOptions &LangOpts) {
 310
 311   unsigned indent = 4;
 312
 313   switch (P.getKind()) {
 314   case PathDiagnosticPiece::ControlFlow:
 315     ReportControlFlow(o, cast<PathDiagnosticControlFlowPiece>(P), FM, SM,
 316                       LangOpts, indent);
 317     break;
 318   case PathDiagnosticPiece::Event:
 319     ReportEvent(o, cast<PathDiagnosticEventPiece>(P), FM, SM, LangOpts,
 320                 indent);
 321     break;
 322   case PathDiagnosticPiece::Macro:
 323     ReportMacro(o, cast<PathDiagnosticMacroPiece>(P), FM, SM, LangOpts,
 324                 indent);
 325     break;
 326   }
 327 }
 328
 329 void PlistDiagnostics::HandlePathDiagnostic(const PathDiagnostic* D) {
 330   if (!D)
 331     return;
 332
 333   if (D->empty()) {
 334     delete D;
 335     return;
 336   }
 337
 338   // We need to flatten the locations (convert Stmt* to locations) because
 339   // the referenced statements may be freed by the time the diagnostics
 340   // are emitted.
 341   const_cast<PathDiagnostic*>(D)->flattenLocations();
 342   BatchedDiags.push_back(D);
 343 }
 344
 345 void PlistDiagnostics::FlushDiagnostics(llvm::SmallVectorImpl<std::string>
 346                                         *FilesMade) {
 347
 348   if (flushed)
 349     return;
 350
 351   flushed = true;
 352
 353   // Sort the diagnostics so that they are always emitted in a deterministic
 354   // order.
 355   if (!BatchedDiags.empty())
 356     std::sort(BatchedDiags.begin(), BatchedDiags.end(), CompareDiagnostics());
 357
 358   // Build up a set of FIDs that we use by scanning the locations and
 359   // ranges of the diagnostics.
 360   FIDMap FM;
 361   llvm::SmallVector<FileID, 10> Fids;
 362   const SourceManager* SM = 0;
 363
 364   if (!BatchedDiags.empty())
 365     SM = &(*BatchedDiags.begin())->begin()->getLocation().getManager();
 366
 367   for (std::vector<const PathDiagnostic*>::iterator DI = BatchedDiags.begin(),
 368        DE = BatchedDiags.end(); DI != DE; ++DI) {
 369
 370     const PathDiagnostic *D = *DI;
 371
 372     for (PathDiagnostic::const_iterator I=D->begin(), E=D->end(); I!=E; ++I) {
 373       AddFID(FM, Fids, SM, I->getLocation().asLocation());
 374
 375       for (PathDiagnosticPiece::range_iterator RI=I->ranges_begin(),
 376            RE=I->ranges_end(); RI!=RE; ++RI) {
 377         AddFID(FM, Fids, SM, RI->getBegin());
 378         AddFID(FM, Fids, SM, RI->getEnd());
 379       }
 380     }
 381   }
 382
 383   // Open the file.
 384   std::string ErrMsg;
 385   llvm::raw_fd_ostream o(OutputFile.c_str(), ErrMsg);
 386   if (!ErrMsg.empty()) {
 387     llvm::errs() << "warning: could not creat file: " << OutputFile << '\n';
 388     return;
 389   }
 390
 391   // Write the plist header.
 392   o << "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
 393   "<!DOCTYPE plist PUBLIC \"-//Apple Computer//DTD PLIST 1.0//EN\" "
 394   "\"http://www.apple.com/DTDs/PropertyList-1.0.dtd\">\n"
 395   "<plist version=\"1.0\">\n";
 396
 397   // Write the root object: a <dict> containing...
 398   //  - "files", an <array> mapping from FIDs to file names
 399   //  - "diagnostics", an <array> containing the path diagnostics
 400   o << "<dict>\n"
 401        " <key>files</key>\n"
 402        " <array>\n";
 403
 404   for (llvm::SmallVectorImpl<FileID>::iterator I=Fids.begin(), E=Fids.end();
 405        I!=E; ++I) {
 406     o << "  ";
 407     EmitString(o, SM->getFileEntryForID(*I)->getName()) << '\n';
 408   }
 409
 410   o << " </array>\n"
 411        " <key>diagnostics</key>\n"
 412        " <array>\n";
 413
 414   for (std::vector<const PathDiagnostic*>::iterator DI=BatchedDiags.begin(),
 415        DE = BatchedDiags.end(); DI!=DE; ++DI) {
 416
 417     o << "  <dict>\n"
 418          "   <key>path</key>\n";
 419
 420     const PathDiagnostic *D = *DI;
 421     // Create an owning smart pointer for 'D' just so that we auto-free it
 422     // when we exit this method.
 423     llvm::OwningPtr<PathDiagnostic> OwnedD(const_cast<PathDiagnostic*>(D));
 424
 425     o << "   <array>\n";
 426
 427     for (PathDiagnostic::const_iterator I=D->begin(), E=D->end(); I != E; ++I)
 428       ReportDiag(o, *I, FM, *SM, LangOpts);
 429
 430     o << "   </array>\n";
 431
 432     // Output the bug type and bug category.
 433     o << "   <key>description</key>";
 434     EmitString(o, D->getDescription()) << '\n';
 435     o << "   <key>category</key>";
 436     EmitString(o, D->getCategory()) << '\n';
 437     o << "   <key>type</key>";
 438     EmitString(o, D->getBugType()) << '\n';
 439
 440     // Output the location of the bug.
 441     o << "  <key>location</key>\n";
 442     EmitLocation(o, *SM, LangOpts, D->getLocation(), FM, 2);
 443
 444     // Output the diagnostic to the sub-diagnostic client, if any.
 445     if (SubPD) {
 446       SubPD->HandlePathDiagnostic(OwnedD.take());
 447       llvm::SmallVector<std::string, 1> SubFilesMade;
 448       SubPD->FlushDiagnostics(SubFilesMade);
 449
 450       if (!SubFilesMade.empty()) {
 451         o << "  <key>" << SubPD->getName() << "_files</key>\n";
 452         o << "  <array>\n";
 453         for (size_t i = 0, n = SubFilesMade.size(); i < n ; ++i)
 454           o << "   <string>" << SubFilesMade[i] << "</string>\n";
 455         o << "  </array>\n";
 456       }
 457     }
 458
 459     // Close up the entry.
 460     o << "  </dict>\n";
 461   }
 462
 463   o << " </array>\n";
 464
 465   // Finish.
 466   o << "</dict>\n</plist>";
 467
 468   if (FilesMade)
 469     FilesMade->push_back(OutputFile);
 470
 471   BatchedDiags.clear();
 472 }