contrib/llvm/tools/clang/lib/StaticAnalyzer/Core/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/Core/AnalyzerOptions.h"
  15 #include "clang/Basic/FileManager.h"
  16 #include "clang/Basic/SourceManager.h"
  17 #include "clang/Basic/Version.h"
  18 #include "clang/Lex/Preprocessor.h"
  19 #include "clang/StaticAnalyzer/Core/BugReporter/PathDiagnostic.h"
  20 #include "clang/StaticAnalyzer/Core/PathDiagnosticConsumers.h"
  21 #include "llvm/ADT/DenseMap.h"
  22 #include "llvm/ADT/SmallVector.h"
  23 #include "llvm/Support/Casting.h"
  24 #include "llvm/Support/raw_ostream.h"
  25 using namespace clang;
  26 using namespace ento;
  27
  28 typedef llvm::DenseMap<FileID, unsigned> FIDMap;
  29
  30
  31 namespace {
  32   class PlistDiagnostics : public PathDiagnosticConsumer {
  33     const std::string OutputFile;
  34     const LangOptions &LangOpts;
  35     const bool SupportsCrossFileDiagnostics;
  36   public:
  37     PlistDiagnostics(AnalyzerOptions &AnalyzerOpts,
  38                      const std::string& prefix,
  39                      const LangOptions &LangOpts,
  40                      bool supportsMultipleFiles);
  41
  42     virtual ~PlistDiagnostics() {}
  43
  44     void FlushDiagnosticsImpl(std::vector<const PathDiagnostic *> &Diags,
  45                               FilesMade *filesMade);
  46
  47     virtual StringRef getName() const {
  48       return "PlistDiagnostics";
  49     }
  50
  51     PathGenerationScheme getGenerationScheme() const { return Extensive; }
  52     bool supportsLogicalOpControlFlow() const { return true; }
  53     bool supportsAllBlockEdges() const { return true; }
  54     virtual bool supportsCrossFileDiagnostics() const {
  55       return SupportsCrossFileDiagnostics;
  56     }
  57   };
  58 } // end anonymous namespace
  59
  60 PlistDiagnostics::PlistDiagnostics(AnalyzerOptions &AnalyzerOpts,
  61                                    const std::string& output,
  62                                    const LangOptions &LO,
  63                                    bool supportsMultipleFiles)
  64   : OutputFile(output),
  65     LangOpts(LO),
  66     SupportsCrossFileDiagnostics(supportsMultipleFiles) {}
  67
  68 void ento::createPlistDiagnosticConsumer(AnalyzerOptions &AnalyzerOpts,
  69                                          PathDiagnosticConsumers &C,
  70                                          const std::string& s,
  71                                          const Preprocessor &PP) {
  72   C.push_back(new PlistDiagnostics(AnalyzerOpts, s,
  73                                    PP.getLangOpts(), false));
  74 }
  75
  76 void ento::createPlistMultiFileDiagnosticConsumer(AnalyzerOptions &AnalyzerOpts,
  77                                                   PathDiagnosticConsumers &C,
  78                                                   const std::string &s,
  79                                                   const Preprocessor &PP) {
  80   C.push_back(new PlistDiagnostics(AnalyzerOpts, s,
  81                                    PP.getLangOpts(), true));
  82 }
  83
  84 static void AddFID(FIDMap &FIDs, SmallVectorImpl<FileID> &V,
  85                    const SourceManager* SM, SourceLocation L) {
  86
  87   FileID FID = SM->getFileID(SM->getExpansionLoc(L));
  88   FIDMap::iterator I = FIDs.find(FID);
  89   if (I != FIDs.end()) return;
  90   FIDs[FID] = V.size();
  91   V.push_back(FID);
  92 }
  93
  94 static unsigned GetFID(const FIDMap& FIDs, const SourceManager &SM,
  95                        SourceLocation L) {
  96   FileID FID = SM.getFileID(SM.getExpansionLoc(L));
  97   FIDMap::const_iterator I = FIDs.find(FID);
  98   assert(I != FIDs.end());
  99   return I->second;
 100 }
 101
 102 static raw_ostream &Indent(raw_ostream &o, const unsigned indent) {
 103   for (unsigned i = 0; i < indent; ++i) o << ' ';
 104   return o;
 105 }
 106
 107 static void EmitLocation(raw_ostream &o, const SourceManager &SM,
 108                          const LangOptions &LangOpts,
 109                          SourceLocation L, const FIDMap &FM,
 110                          unsigned indent, bool extend = false) {
 111
 112   FullSourceLoc Loc(SM.getExpansionLoc(L), const_cast<SourceManager&>(SM));
 113
 114   // Add in the length of the token, so that we cover multi-char tokens.
 115   unsigned offset =
 116     extend ? Lexer::MeasureTokenLength(Loc, SM, LangOpts) - 1 : 0;
 117
 118   Indent(o, indent) << "<dict>\n";
 119   Indent(o, indent) << " <key>line</key><integer>"
 120                     << Loc.getExpansionLineNumber() << "</integer>\n";
 121   Indent(o, indent) << " <key>col</key><integer>"
 122                     << Loc.getExpansionColumnNumber() + offset << "</integer>\n";
 123   Indent(o, indent) << " <key>file</key><integer>"
 124                     << GetFID(FM, SM, Loc) << "</integer>\n";
 125   Indent(o, indent) << "</dict>\n";
 126 }
 127
 128 static void EmitLocation(raw_ostream &o, const SourceManager &SM,
 129                          const LangOptions &LangOpts,
 130                          const PathDiagnosticLocation &L, const FIDMap& FM,
 131                          unsigned indent, bool extend = false) {
 132   EmitLocation(o, SM, LangOpts, L.asLocation(), FM, indent, extend);
 133 }
 134
 135 static void EmitRange(raw_ostream &o, const SourceManager &SM,
 136                       const LangOptions &LangOpts,
 137                       PathDiagnosticRange R, const FIDMap &FM,
 138                       unsigned indent) {
 139   Indent(o, indent) << "<array>\n";
 140   EmitLocation(o, SM, LangOpts, R.getBegin(), FM, indent+1);
 141   EmitLocation(o, SM, LangOpts, R.getEnd(), FM, indent+1, !R.isPoint);
 142   Indent(o, indent) << "</array>\n";
 143 }
 144
 145 static raw_ostream &EmitString(raw_ostream &o, StringRef s) {
 146   o << "<string>";
 147   for (StringRef::const_iterator I = s.begin(), E = s.end(); I != E; ++I) {
 148     char c = *I;
 149     switch (c) {
 150     default:   o << c; break;
 151     case '&':  o << "&amp;"; break;
 152     case '<':  o << "&lt;"; break;
 153     case '>':  o << "&gt;"; break;
 154     case '\'': o << "&apos;"; break;
 155     case '\"': o << "&quot;"; break;
 156     }
 157   }
 158   o << "</string>";
 159   return o;
 160 }
 161
 162 static void ReportControlFlow(raw_ostream &o,
 163                               const PathDiagnosticControlFlowPiece& P,
 164                               const FIDMap& FM,
 165                               const SourceManager &SM,
 166                               const LangOptions &LangOpts,
 167                               unsigned indent) {
 168
 169   Indent(o, indent) << "<dict>\n";
 170   ++indent;
 171
 172   Indent(o, indent) << "<key>kind</key><string>control</string>\n";
 173
 174   // Emit edges.
 175   Indent(o, indent) << "<key>edges</key>\n";
 176   ++indent;
 177   Indent(o, indent) << "<array>\n";
 178   ++indent;
 179   for (PathDiagnosticControlFlowPiece::const_iterator I=P.begin(), E=P.end();
 180        I!=E; ++I) {
 181     Indent(o, indent) << "<dict>\n";
 182     ++indent;
 183
 184     // Make the ranges of the start and end point self-consistent with adjacent edges
 185     // by forcing to use only the beginning of the range.  This simplifies the layout
 186     // logic for clients.
 187     Indent(o, indent) << "<key>start</key>\n";
 188     SourceLocation StartEdge = I->getStart().asRange().getBegin();
 189     EmitRange(o, SM, LangOpts, SourceRange(StartEdge, StartEdge), FM, indent+1);
 190
 191     Indent(o, indent) << "<key>end</key>\n";
 192     SourceLocation EndEdge = I->getEnd().asRange().getBegin();
 193     EmitRange(o, SM, LangOpts, SourceRange(EndEdge, EndEdge), FM, indent+1);
 194
 195     --indent;
 196     Indent(o, indent) << "</dict>\n";
 197   }
 198   --indent;
 199   Indent(o, indent) << "</array>\n";
 200   --indent;
 201
 202   // Output any helper text.
 203   const std::string& s = P.getString();
 204   if (!s.empty()) {
 205     Indent(o, indent) << "<key>alternate</key>";
 206     EmitString(o, s) << '\n';
 207   }
 208
 209   --indent;
 210   Indent(o, indent) << "</dict>\n";
 211 }
 212
 213 static void ReportEvent(raw_ostream &o, const PathDiagnosticPiece& P,
 214                         const FIDMap& FM,
 215                         const SourceManager &SM,
 216                         const LangOptions &LangOpts,
 217                         unsigned indent,
 218                         unsigned depth) {
 219
 220   Indent(o, indent) << "<dict>\n";
 221   ++indent;
 222
 223   Indent(o, indent) << "<key>kind</key><string>event</string>\n";
 224
 225   // Output the location.
 226   FullSourceLoc L = P.getLocation().asLocation();
 227
 228   Indent(o, indent) << "<key>location</key>\n";
 229   EmitLocation(o, SM, LangOpts, L, FM, indent);
 230
 231   // Output the ranges (if any).
 232   ArrayRef<SourceRange> Ranges = P.getRanges();
 233
 234   if (!Ranges.empty()) {
 235     Indent(o, indent) << "<key>ranges</key>\n";
 236     Indent(o, indent) << "<array>\n";
 237     ++indent;
 238     for (ArrayRef<SourceRange>::iterator I = Ranges.begin(), E = Ranges.end();
 239          I != E; ++I) {
 240       EmitRange(o, SM, LangOpts, *I, FM, indent+1);
 241     }
 242     --indent;
 243     Indent(o, indent) << "</array>\n";
 244   }
 245
 246   // Output the call depth.
 247   Indent(o, indent) << "<key>depth</key>"
 248                     << "<integer>" << depth << "</integer>\n";
 249
 250   // Output the text.
 251   assert(!P.getString().empty());
 252   Indent(o, indent) << "<key>extended_message</key>\n";
 253   Indent(o, indent);
 254   EmitString(o, P.getString()) << '\n';
 255
 256   // Output the short text.
 257   // FIXME: Really use a short string.
 258   Indent(o, indent) << "<key>message</key>\n";
 259   Indent(o, indent);
 260   EmitString(o, P.getString()) << '\n';
 261
 262   // Finish up.
 263   --indent;
 264   Indent(o, indent); o << "</dict>\n";
 265 }
 266
 267 static void ReportPiece(raw_ostream &o,
 268                         const PathDiagnosticPiece &P,
 269                         const FIDMap& FM, const SourceManager &SM,
 270                         const LangOptions &LangOpts,
 271                         unsigned indent,
 272                         unsigned depth,
 273                         bool includeControlFlow);
 274
 275 static void ReportCall(raw_ostream &o,
 276                        const PathDiagnosticCallPiece &P,
 277                        const FIDMap& FM, const SourceManager &SM,
 278                        const LangOptions &LangOpts,
 279                        unsigned indent,
 280                        unsigned depth) {
 281
 282   IntrusiveRefCntPtr<PathDiagnosticEventPiece> callEnter =
 283     P.getCallEnterEvent();
 284
 285   if (callEnter)
 286     ReportPiece(o, *callEnter, FM, SM, LangOpts, indent, depth, true);
 287
 288   IntrusiveRefCntPtr<PathDiagnosticEventPiece> callEnterWithinCaller =
 289     P.getCallEnterWithinCallerEvent();
 290
 291   ++depth;
 292
 293   if (callEnterWithinCaller)
 294     ReportPiece(o, *callEnterWithinCaller, FM, SM, LangOpts,
 295                 indent, depth, true);
 296
 297   for (PathPieces::const_iterator I = P.path.begin(), E = P.path.end();I!=E;++I)
 298     ReportPiece(o, **I, FM, SM, LangOpts, indent, depth, true);
 299
 300   --depth;
 301
 302   IntrusiveRefCntPtr<PathDiagnosticEventPiece> callExit =
 303     P.getCallExitEvent();
 304
 305   if (callExit)
 306     ReportPiece(o, *callExit, FM, SM, LangOpts, indent, depth, true);
 307 }
 308
 309 static void ReportMacro(raw_ostream &o,
 310                         const PathDiagnosticMacroPiece& P,
 311                         const FIDMap& FM, const SourceManager &SM,
 312                         const LangOptions &LangOpts,
 313                         unsigned indent,
 314                         unsigned depth) {
 315
 316   for (PathPieces::const_iterator I = P.subPieces.begin(), E=P.subPieces.end();
 317        I!=E; ++I) {
 318     ReportPiece(o, **I, FM, SM, LangOpts, indent, depth, false);
 319   }
 320 }
 321
 322 static void ReportDiag(raw_ostream &o, const PathDiagnosticPiece& P,
 323                        const FIDMap& FM, const SourceManager &SM,
 324                        const LangOptions &LangOpts) {
 325   ReportPiece(o, P, FM, SM, LangOpts, 4, 0, true);
 326 }
 327
 328 static void ReportPiece(raw_ostream &o,
 329                         const PathDiagnosticPiece &P,
 330                         const FIDMap& FM, const SourceManager &SM,
 331                         const LangOptions &LangOpts,
 332                         unsigned indent,
 333                         unsigned depth,
 334                         bool includeControlFlow) {
 335   switch (P.getKind()) {
 336     case PathDiagnosticPiece::ControlFlow:
 337       if (includeControlFlow)
 338         ReportControlFlow(o, cast<PathDiagnosticControlFlowPiece>(P), FM, SM,
 339                           LangOpts, indent);
 340       break;
 341     case PathDiagnosticPiece::Call:
 342       ReportCall(o, cast<PathDiagnosticCallPiece>(P), FM, SM, LangOpts,
 343                  indent, depth);
 344       break;
 345     case PathDiagnosticPiece::Event:
 346       ReportEvent(o, cast<PathDiagnosticSpotPiece>(P), FM, SM, LangOpts,
 347                   indent, depth);
 348       break;
 349     case PathDiagnosticPiece::Macro:
 350       ReportMacro(o, cast<PathDiagnosticMacroPiece>(P), FM, SM, LangOpts,
 351                   indent, depth);
 352       break;
 353   }
 354 }
 355
 356 void PlistDiagnostics::FlushDiagnosticsImpl(
 357                                     std::vector<const PathDiagnostic *> &Diags,
 358                                     FilesMade *filesMade) {
 359   // Build up a set of FIDs that we use by scanning the locations and
 360   // ranges of the diagnostics.
 361   FIDMap FM;
 362   SmallVector<FileID, 10> Fids;
 363   const SourceManager* SM = 0;
 364
 365   if (!Diags.empty())
 366     SM = &(*(*Diags.begin())->path.begin())->getLocation().getManager();
 367
 368
 369   for (std::vector<const PathDiagnostic*>::iterator DI = Diags.begin(),
 370        DE = Diags.end(); DI != DE; ++DI) {
 371
 372     const PathDiagnostic *D = *DI;
 373
 374     SmallVector<const PathPieces *, 5> WorkList;
 375     WorkList.push_back(&D->path);
 376
 377     while (!WorkList.empty()) {
 378       const PathPieces &path = *WorkList.back();
 379       WorkList.pop_back();
 380
 381       for (PathPieces::const_iterator I = path.begin(), E = path.end();
 382            I!=E; ++I) {
 383         const PathDiagnosticPiece *piece = I->getPtr();
 384         AddFID(FM, Fids, SM, piece->getLocation().asLocation());
 385         ArrayRef<SourceRange> Ranges = piece->getRanges();
 386         for (ArrayRef<SourceRange>::iterator I = Ranges.begin(),
 387                                              E = Ranges.end(); I != E; ++I) {
 388           AddFID(FM, Fids, SM, I->getBegin());
 389           AddFID(FM, Fids, SM, I->getEnd());
 390         }
 391
 392         if (const PathDiagnosticCallPiece *call =
 393             dyn_cast<PathDiagnosticCallPiece>(piece)) {
 394           IntrusiveRefCntPtr<PathDiagnosticEventPiece>
 395             callEnterWithin = call->getCallEnterWithinCallerEvent();
 396           if (callEnterWithin)
 397             AddFID(FM, Fids, SM, callEnterWithin->getLocation().asLocation());
 398
 399           WorkList.push_back(&call->path);
 400         }
 401         else if (const PathDiagnosticMacroPiece *macro =
 402                  dyn_cast<PathDiagnosticMacroPiece>(piece)) {
 403           WorkList.push_back(&macro->subPieces);
 404         }
 405       }
 406     }
 407   }
 408
 409   // Open the file.
 410   std::string ErrMsg;
 411   llvm::raw_fd_ostream o(OutputFile.c_str(), ErrMsg);
 412   if (!ErrMsg.empty()) {
 413     llvm::errs() << "warning: could not create file: " << OutputFile << '\n';
 414     return;
 415   }
 416
 417   // Write the plist header.
 418   o << "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
 419   "<!DOCTYPE plist PUBLIC \"-//Apple Computer//DTD PLIST 1.0//EN\" "
 420   "\"http://www.apple.com/DTDs/PropertyList-1.0.dtd\">\n"
 421   "<plist version=\"1.0\">\n";
 422
 423   // Write the root object: a <dict> containing...
 424   //  - "clang_version", the string representation of clang version
 425   //  - "files", an <array> mapping from FIDs to file names
 426   //  - "diagnostics", an <array> containing the path diagnostics
 427   o << "<dict>\n" <<
 428        " <key>clang_version</key>\n";
 429   EmitString(o, getClangFullVersion()) << '\n';
 430   o << " <key>files</key>\n"
 431        " <array>\n";
 432
 433   for (SmallVectorImpl<FileID>::iterator I=Fids.begin(), E=Fids.end();
 434        I!=E; ++I) {
 435     o << "  ";
 436     EmitString(o, SM->getFileEntryForID(*I)->getName()) << '\n';
 437   }
 438
 439   o << " </array>\n"
 440        " <key>diagnostics</key>\n"
 441        " <array>\n";
 442
 443   for (std::vector<const PathDiagnostic*>::iterator DI=Diags.begin(),
 444        DE = Diags.end(); DI!=DE; ++DI) {
 445
 446     o << "  <dict>\n"
 447          "   <key>path</key>\n";
 448
 449     const PathDiagnostic *D = *DI;
 450
 451     o << "   <array>\n";
 452
 453     for (PathPieces::const_iterator I = D->path.begin(), E = D->path.end();
 454          I != E; ++I)
 455       ReportDiag(o, **I, FM, *SM, LangOpts);
 456
 457     o << "   </array>\n";
 458
 459     // Output the bug type and bug category.
 460     o << "   <key>description</key>";
 461     EmitString(o, D->getShortDescription()) << '\n';
 462     o << "   <key>category</key>";
 463     EmitString(o, D->getCategory()) << '\n';
 464     o << "   <key>type</key>";
 465     EmitString(o, D->getBugType()) << '\n';
 466
 467     // Output information about the semantic context where
 468     // the issue occurred.
 469     if (const Decl *DeclWithIssue = D->getDeclWithIssue()) {
 470       // FIXME: handle blocks, which have no name.
 471       if (const NamedDecl *ND = dyn_cast<NamedDecl>(DeclWithIssue)) {
 472         StringRef declKind;
 473         switch (ND->getKind()) {
 474           case Decl::CXXRecord:
 475             declKind = "C++ class";
 476             break;
 477           case Decl::CXXMethod:
 478             declKind = "C++ method";
 479             break;
 480           case Decl::ObjCMethod:
 481             declKind = "Objective-C method";
 482             break;
 483           case Decl::Function:
 484             declKind = "function";
 485             break;
 486           default:
 487             break;
 488         }
 489         if (!declKind.empty()) {
 490           const std::string &declName = ND->getDeclName().getAsString();
 491           o << "  <key>issue_context_kind</key>";
 492           EmitString(o, declKind) << '\n';
 493           o << "  <key>issue_context</key>";
 494           EmitString(o, declName) << '\n';
 495         }
 496
 497         // Output the bug hash for issue unique-ing. Currently, it's just an
 498         // offset from the beginning of the function.
 499         if (const Stmt *Body = DeclWithIssue->getBody()) {
 500
 501           // If the bug uniqueing location exists, use it for the hash.
 502           // For example, this ensures that two leaks reported on the same line
 503           // will have different issue_hashes and that the hash will identify
 504           // the leak location even after code is added between the allocation
 505           // site and the end of scope (leak report location).
 506           PathDiagnosticLocation UPDLoc = D->getUniqueingLoc();
 507           if (UPDLoc.isValid()) {
 508             FullSourceLoc UL(SM->getExpansionLoc(UPDLoc.asLocation()),
 509                              *SM);
 510             FullSourceLoc UFunL(SM->getExpansionLoc(
 511               D->getUniqueingDecl()->getBody()->getLocStart()), *SM);
 512             o << "  <key>issue_hash</key><string>"
 513               << UL.getExpansionLineNumber() - UFunL.getExpansionLineNumber()
 514               << "</string>\n";
 515
 516           // Otherwise, use the location on which the bug is reported.
 517           } else {
 518             FullSourceLoc L(SM->getExpansionLoc(D->getLocation().asLocation()),
 519                             *SM);
 520             FullSourceLoc FunL(SM->getExpansionLoc(Body->getLocStart()), *SM);
 521             o << "  <key>issue_hash</key><string>"
 522               << L.getExpansionLineNumber() - FunL.getExpansionLineNumber()
 523               << "</string>\n";
 524           }
 525
 526         }
 527       }
 528     }
 529
 530     // Output the location of the bug.
 531     o << "  <key>location</key>\n";
 532     EmitLocation(o, *SM, LangOpts, D->getLocation(), FM, 2);
 533
 534     // Output the diagnostic to the sub-diagnostic client, if any.
 535     if (!filesMade->empty()) {
 536       StringRef lastName;
 537       PDFileEntry::ConsumerFiles *files = filesMade->getFiles(*D);
 538       if (files) {
 539         for (PDFileEntry::ConsumerFiles::const_iterator CI = files->begin(),
 540                 CE = files->end(); CI != CE; ++CI) {
 541           StringRef newName = CI->first;
 542           if (newName != lastName) {
 543             if (!lastName.empty()) {
 544               o << "  </array>\n";
 545             }
 546             lastName = newName;
 547             o <<  "  <key>" << lastName << "_files</key>\n";
 548             o << "  <array>\n";
 549           }
 550           o << "   <string>" << CI->second << "</string>\n";
 551         }
 552         o << "  </array>\n";
 553       }
 554     }
 555
 556     // Close up the entry.
 557     o << "  </dict>\n";
 558   }
 559
 560   o << " </array>\n";
 561
 562   // Finish.
 563   o << "</dict>\n</plist>";
 564 }