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     virtual bool supportsCrossFileDiagnostics() const {
  54       return SupportsCrossFileDiagnostics;
  55     }
  56   };
  57 } // end anonymous namespace
  58
  59 PlistDiagnostics::PlistDiagnostics(AnalyzerOptions &AnalyzerOpts,
  60                                    const std::string& output,
  61                                    const LangOptions &LO,
  62                                    bool supportsMultipleFiles)
  63   : OutputFile(output),
  64     LangOpts(LO),
  65     SupportsCrossFileDiagnostics(supportsMultipleFiles) {}
  66
  67 void ento::createPlistDiagnosticConsumer(AnalyzerOptions &AnalyzerOpts,
  68                                          PathDiagnosticConsumers &C,
  69                                          const std::string& s,
  70                                          const Preprocessor &PP) {
  71   C.push_back(new PlistDiagnostics(AnalyzerOpts, s,
  72                                    PP.getLangOpts(), false));
  73 }
  74
  75 void ento::createPlistMultiFileDiagnosticConsumer(AnalyzerOptions &AnalyzerOpts,
  76                                                   PathDiagnosticConsumers &C,
  77                                                   const std::string &s,
  78                                                   const Preprocessor &PP) {
  79   C.push_back(new PlistDiagnostics(AnalyzerOpts, s,
  80                                    PP.getLangOpts(), true));
  81 }
  82
  83 static void AddFID(FIDMap &FIDs, SmallVectorImpl<FileID> &V,
  84                    const SourceManager* SM, SourceLocation L) {
  85
  86   FileID FID = SM->getFileID(SM->getExpansionLoc(L));
  87   FIDMap::iterator I = FIDs.find(FID);
  88   if (I != FIDs.end()) return;
  89   FIDs[FID] = V.size();
  90   V.push_back(FID);
  91 }
  92
  93 static unsigned GetFID(const FIDMap& FIDs, const SourceManager &SM,
  94                        SourceLocation L) {
  95   FileID FID = SM.getFileID(SM.getExpansionLoc(L));
  96   FIDMap::const_iterator I = FIDs.find(FID);
  97   assert(I != FIDs.end());
  98   return I->second;
  99 }
 100
 101 static raw_ostream &Indent(raw_ostream &o, const unsigned indent) {
 102   for (unsigned i = 0; i < indent; ++i) o << ' ';
 103   return o;
 104 }
 105
 106 static void EmitLocation(raw_ostream &o, const SourceManager &SM,
 107                          const LangOptions &LangOpts,
 108                          SourceLocation L, const FIDMap &FM,
 109                          unsigned indent, bool extend = false) {
 110
 111   FullSourceLoc Loc(SM.getExpansionLoc(L), const_cast<SourceManager&>(SM));
 112
 113   // Add in the length of the token, so that we cover multi-char tokens.
 114   unsigned offset =
 115     extend ? Lexer::MeasureTokenLength(Loc, SM, LangOpts) - 1 : 0;
 116
 117   Indent(o, indent) << "<dict>\n";
 118   Indent(o, indent) << " <key>line</key><integer>"
 119                     << Loc.getExpansionLineNumber() << "</integer>\n";
 120   Indent(o, indent) << " <key>col</key><integer>"
 121                     << Loc.getExpansionColumnNumber() + offset << "</integer>\n";
 122   Indent(o, indent) << " <key>file</key><integer>"
 123                     << GetFID(FM, SM, Loc) << "</integer>\n";
 124   Indent(o, indent) << "</dict>\n";
 125 }
 126
 127 static void EmitLocation(raw_ostream &o, const SourceManager &SM,
 128                          const LangOptions &LangOpts,
 129                          const PathDiagnosticLocation &L, const FIDMap& FM,
 130                          unsigned indent, bool extend = false) {
 131   EmitLocation(o, SM, LangOpts, L.asLocation(), FM, indent, extend);
 132 }
 133
 134 static void EmitRange(raw_ostream &o, const SourceManager &SM,
 135                       const LangOptions &LangOpts,
 136                       PathDiagnosticRange R, const FIDMap &FM,
 137                       unsigned indent) {
 138   Indent(o, indent) << "<array>\n";
 139   EmitLocation(o, SM, LangOpts, R.getBegin(), FM, indent+1);
 140   EmitLocation(o, SM, LangOpts, R.getEnd(), FM, indent+1, !R.isPoint);
 141   Indent(o, indent) << "</array>\n";
 142 }
 143
 144 static raw_ostream &EmitString(raw_ostream &o, StringRef s) {
 145   o << "<string>";
 146   for (StringRef::const_iterator I = s.begin(), E = s.end(); I != E; ++I) {
 147     char c = *I;
 148     switch (c) {
 149     default:   o << c; break;
 150     case '&':  o << "&amp;"; break;
 151     case '<':  o << "&lt;"; break;
 152     case '>':  o << "&gt;"; break;
 153     case '\'': o << "&apos;"; break;
 154     case '\"': o << "&quot;"; break;
 155     }
 156   }
 157   o << "</string>";
 158   return o;
 159 }
 160
 161 static void ReportControlFlow(raw_ostream &o,
 162                               const PathDiagnosticControlFlowPiece& P,
 163                               const FIDMap& FM,
 164                               const SourceManager &SM,
 165                               const LangOptions &LangOpts,
 166                               unsigned indent) {
 167
 168   Indent(o, indent) << "<dict>\n";
 169   ++indent;
 170
 171   Indent(o, indent) << "<key>kind</key><string>control</string>\n";
 172
 173   // Emit edges.
 174   Indent(o, indent) << "<key>edges</key>\n";
 175   ++indent;
 176   Indent(o, indent) << "<array>\n";
 177   ++indent;
 178   for (PathDiagnosticControlFlowPiece::const_iterator I=P.begin(), E=P.end();
 179        I!=E; ++I) {
 180     Indent(o, indent) << "<dict>\n";
 181     ++indent;
 182
 183     // Make the ranges of the start and end point self-consistent with adjacent edges
 184     // by forcing to use only the beginning of the range.  This simplifies the layout
 185     // logic for clients.
 186     Indent(o, indent) << "<key>start</key>\n";
 187     SourceLocation StartEdge = I->getStart().asRange().getBegin();
 188     EmitRange(o, SM, LangOpts, SourceRange(StartEdge, StartEdge), FM, indent+1);
 189
 190     Indent(o, indent) << "<key>end</key>\n";
 191     SourceLocation EndEdge = I->getEnd().asRange().getBegin();
 192     EmitRange(o, SM, LangOpts, SourceRange(EndEdge, EndEdge), FM, indent+1);
 193
 194     --indent;
 195     Indent(o, indent) << "</dict>\n";
 196   }
 197   --indent;
 198   Indent(o, indent) << "</array>\n";
 199   --indent;
 200
 201   // Output any helper text.
 202   const std::string& s = P.getString();
 203   if (!s.empty()) {
 204     Indent(o, indent) << "<key>alternate</key>";
 205     EmitString(o, s) << '\n';
 206   }
 207
 208   --indent;
 209   Indent(o, indent) << "</dict>\n";
 210 }
 211
 212 static void ReportEvent(raw_ostream &o, const PathDiagnosticPiece& P,
 213                         const FIDMap& FM,
 214                         const SourceManager &SM,
 215                         const LangOptions &LangOpts,
 216                         unsigned indent,
 217                         unsigned depth,
 218                         bool isKeyEvent = false) {
 219
 220   Indent(o, indent) << "<dict>\n";
 221   ++indent;
 222
 223   Indent(o, indent) << "<key>kind</key><string>event</string>\n";
 224
 225   if (isKeyEvent) {
 226     Indent(o, indent) << "<key>key_event</key><true/>\n";
 227   }
 228
 229   // Output the location.
 230   FullSourceLoc L = P.getLocation().asLocation();
 231
 232   Indent(o, indent) << "<key>location</key>\n";
 233   EmitLocation(o, SM, LangOpts, L, FM, indent);
 234
 235   // Output the ranges (if any).
 236   ArrayRef<SourceRange> Ranges = P.getRanges();
 237
 238   if (!Ranges.empty()) {
 239     Indent(o, indent) << "<key>ranges</key>\n";
 240     Indent(o, indent) << "<array>\n";
 241     ++indent;
 242     for (ArrayRef<SourceRange>::iterator I = Ranges.begin(), E = Ranges.end();
 243          I != E; ++I) {
 244       EmitRange(o, SM, LangOpts, *I, FM, indent+1);
 245     }
 246     --indent;
 247     Indent(o, indent) << "</array>\n";
 248   }
 249
 250   // Output the call depth.
 251   Indent(o, indent) << "<key>depth</key>"
 252                     << "<integer>" << depth << "</integer>\n";
 253
 254   // Output the text.
 255   assert(!P.getString().empty());
 256   Indent(o, indent) << "<key>extended_message</key>\n";
 257   Indent(o, indent);
 258   EmitString(o, P.getString()) << '\n';
 259
 260   // Output the short text.
 261   // FIXME: Really use a short string.
 262   Indent(o, indent) << "<key>message</key>\n";
 263   Indent(o, indent);
 264   EmitString(o, P.getString()) << '\n';
 265
 266   // Finish up.
 267   --indent;
 268   Indent(o, indent); o << "</dict>\n";
 269 }
 270
 271 static void ReportPiece(raw_ostream &o,
 272                         const PathDiagnosticPiece &P,
 273                         const FIDMap& FM, const SourceManager &SM,
 274                         const LangOptions &LangOpts,
 275                         unsigned indent,
 276                         unsigned depth,
 277                         bool includeControlFlow,
 278                         bool isKeyEvent = false);
 279
 280 static void ReportCall(raw_ostream &o,
 281                        const PathDiagnosticCallPiece &P,
 282                        const FIDMap& FM, const SourceManager &SM,
 283                        const LangOptions &LangOpts,
 284                        unsigned indent,
 285                        unsigned depth) {
 286
 287   IntrusiveRefCntPtr<PathDiagnosticEventPiece> callEnter =
 288     P.getCallEnterEvent();
 289
 290   if (callEnter)
 291     ReportPiece(o, *callEnter, FM, SM, LangOpts, indent, depth, true,
 292                 P.isLastInMainSourceFile());
 293
 294   IntrusiveRefCntPtr<PathDiagnosticEventPiece> callEnterWithinCaller =
 295     P.getCallEnterWithinCallerEvent();
 296
 297   ++depth;
 298
 299   if (callEnterWithinCaller)
 300     ReportPiece(o, *callEnterWithinCaller, FM, SM, LangOpts,
 301                 indent, depth, true);
 302
 303   for (PathPieces::const_iterator I = P.path.begin(), E = P.path.end();I!=E;++I)
 304     ReportPiece(o, **I, FM, SM, LangOpts, indent, depth, true);
 305
 306   --depth;
 307
 308   IntrusiveRefCntPtr<PathDiagnosticEventPiece> callExit =
 309     P.getCallExitEvent();
 310
 311   if (callExit)
 312     ReportPiece(o, *callExit, FM, SM, LangOpts, indent, depth, true);
 313 }
 314
 315 static void ReportMacro(raw_ostream &o,
 316                         const PathDiagnosticMacroPiece& P,
 317                         const FIDMap& FM, const SourceManager &SM,
 318                         const LangOptions &LangOpts,
 319                         unsigned indent,
 320                         unsigned depth) {
 321
 322   for (PathPieces::const_iterator I = P.subPieces.begin(), E=P.subPieces.end();
 323        I!=E; ++I) {
 324     ReportPiece(o, **I, FM, SM, LangOpts, indent, depth, false);
 325   }
 326 }
 327
 328 static void ReportDiag(raw_ostream &o, const PathDiagnosticPiece& P,
 329                        const FIDMap& FM, const SourceManager &SM,
 330                        const LangOptions &LangOpts) {
 331   ReportPiece(o, P, FM, SM, LangOpts, 4, 0, true);
 332 }
 333
 334 static void ReportPiece(raw_ostream &o,
 335                         const PathDiagnosticPiece &P,
 336                         const FIDMap& FM, const SourceManager &SM,
 337                         const LangOptions &LangOpts,
 338                         unsigned indent,
 339                         unsigned depth,
 340                         bool includeControlFlow,
 341                         bool isKeyEvent) {
 342   switch (P.getKind()) {
 343     case PathDiagnosticPiece::ControlFlow:
 344       if (includeControlFlow)
 345         ReportControlFlow(o, cast<PathDiagnosticControlFlowPiece>(P), FM, SM,
 346                           LangOpts, indent);
 347       break;
 348     case PathDiagnosticPiece::Call:
 349       ReportCall(o, cast<PathDiagnosticCallPiece>(P), FM, SM, LangOpts,
 350                  indent, depth);
 351       break;
 352     case PathDiagnosticPiece::Event:
 353       ReportEvent(o, cast<PathDiagnosticSpotPiece>(P), FM, SM, LangOpts,
 354                   indent, depth, isKeyEvent);
 355       break;
 356     case PathDiagnosticPiece::Macro:
 357       ReportMacro(o, cast<PathDiagnosticMacroPiece>(P), FM, SM, LangOpts,
 358                   indent, depth);
 359       break;
 360   }
 361 }
 362
 363 void PlistDiagnostics::FlushDiagnosticsImpl(
 364                                     std::vector<const PathDiagnostic *> &Diags,
 365                                     FilesMade *filesMade) {
 366   // Build up a set of FIDs that we use by scanning the locations and
 367   // ranges of the diagnostics.
 368   FIDMap FM;
 369   SmallVector<FileID, 10> Fids;
 370   const SourceManager* SM = 0;
 371
 372   if (!Diags.empty())
 373     SM = &(*(*Diags.begin())->path.begin())->getLocation().getManager();
 374
 375
 376   for (std::vector<const PathDiagnostic*>::iterator DI = Diags.begin(),
 377        DE = Diags.end(); DI != DE; ++DI) {
 378
 379     const PathDiagnostic *D = *DI;
 380
 381     SmallVector<const PathPieces *, 5> WorkList;
 382     WorkList.push_back(&D->path);
 383
 384     while (!WorkList.empty()) {
 385       const PathPieces &path = *WorkList.pop_back_val();
 386
 387       for (PathPieces::const_iterator I = path.begin(), E = path.end(); I != E;
 388            ++I) {
 389         const PathDiagnosticPiece *piece = I->getPtr();
 390         AddFID(FM, Fids, SM, piece->getLocation().asLocation());
 391         ArrayRef<SourceRange> Ranges = piece->getRanges();
 392         for (ArrayRef<SourceRange>::iterator I = Ranges.begin(),
 393                                              E = Ranges.end(); I != E; ++I) {
 394           AddFID(FM, Fids, SM, I->getBegin());
 395           AddFID(FM, Fids, SM, I->getEnd());
 396         }
 397
 398         if (const PathDiagnosticCallPiece *call =
 399             dyn_cast<PathDiagnosticCallPiece>(piece)) {
 400           IntrusiveRefCntPtr<PathDiagnosticEventPiece>
 401             callEnterWithin = call->getCallEnterWithinCallerEvent();
 402           if (callEnterWithin)
 403             AddFID(FM, Fids, SM, callEnterWithin->getLocation().asLocation());
 404
 405           WorkList.push_back(&call->path);
 406         }
 407         else if (const PathDiagnosticMacroPiece *macro =
 408                  dyn_cast<PathDiagnosticMacroPiece>(piece)) {
 409           WorkList.push_back(&macro->subPieces);
 410         }
 411       }
 412     }
 413   }
 414
 415   // Open the file.
 416   std::string ErrMsg;
 417   llvm::raw_fd_ostream o(OutputFile.c_str(), ErrMsg);
 418   if (!ErrMsg.empty()) {
 419     llvm::errs() << "warning: could not create file: " << OutputFile << '\n';
 420     return;
 421   }
 422
 423   // Write the plist header.
 424   o << "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
 425   "<!DOCTYPE plist PUBLIC \"-//Apple Computer//DTD PLIST 1.0//EN\" "
 426   "\"http://www.apple.com/DTDs/PropertyList-1.0.dtd\">\n"
 427   "<plist version=\"1.0\">\n";
 428
 429   // Write the root object: a <dict> containing...
 430   //  - "clang_version", the string representation of clang version
 431   //  - "files", an <array> mapping from FIDs to file names
 432   //  - "diagnostics", an <array> containing the path diagnostics
 433   o << "<dict>\n" <<
 434        " <key>clang_version</key>\n";
 435   EmitString(o, getClangFullVersion()) << '\n';
 436   o << " <key>files</key>\n"
 437        " <array>\n";
 438
 439   for (SmallVectorImpl<FileID>::iterator I=Fids.begin(), E=Fids.end();
 440        I!=E; ++I) {
 441     o << "  ";
 442     EmitString(o, SM->getFileEntryForID(*I)->getName()) << '\n';
 443   }
 444
 445   o << " </array>\n"
 446        " <key>diagnostics</key>\n"
 447        " <array>\n";
 448
 449   for (std::vector<const PathDiagnostic*>::iterator DI=Diags.begin(),
 450        DE = Diags.end(); DI!=DE; ++DI) {
 451
 452     o << "  <dict>\n"
 453          "   <key>path</key>\n";
 454
 455     const PathDiagnostic *D = *DI;
 456
 457     o << "   <array>\n";
 458
 459     for (PathPieces::const_iterator I = D->path.begin(), E = D->path.end();
 460          I != E; ++I)
 461       ReportDiag(o, **I, FM, *SM, LangOpts);
 462
 463     o << "   </array>\n";
 464
 465     // Output the bug type and bug category.
 466     o << "   <key>description</key>";
 467     EmitString(o, D->getShortDescription()) << '\n';
 468     o << "   <key>category</key>";
 469     EmitString(o, D->getCategory()) << '\n';
 470     o << "   <key>type</key>";
 471     EmitString(o, D->getBugType()) << '\n';
 472
 473     // Output information about the semantic context where
 474     // the issue occurred.
 475     if (const Decl *DeclWithIssue = D->getDeclWithIssue()) {
 476       // FIXME: handle blocks, which have no name.
 477       if (const NamedDecl *ND = dyn_cast<NamedDecl>(DeclWithIssue)) {
 478         StringRef declKind;
 479         switch (ND->getKind()) {
 480           case Decl::CXXRecord:
 481             declKind = "C++ class";
 482             break;
 483           case Decl::CXXMethod:
 484             declKind = "C++ method";
 485             break;
 486           case Decl::ObjCMethod:
 487             declKind = "Objective-C method";
 488             break;
 489           case Decl::Function:
 490             declKind = "function";
 491             break;
 492           default:
 493             break;
 494         }
 495         if (!declKind.empty()) {
 496           const std::string &declName = ND->getDeclName().getAsString();
 497           o << "  <key>issue_context_kind</key>";
 498           EmitString(o, declKind) << '\n';
 499           o << "  <key>issue_context</key>";
 500           EmitString(o, declName) << '\n';
 501         }
 502
 503         // Output the bug hash for issue unique-ing. Currently, it's just an
 504         // offset from the beginning of the function.
 505         if (const Stmt *Body = DeclWithIssue->getBody()) {
 506
 507           // If the bug uniqueing location exists, use it for the hash.
 508           // For example, this ensures that two leaks reported on the same line
 509           // will have different issue_hashes and that the hash will identify
 510           // the leak location even after code is added between the allocation
 511           // site and the end of scope (leak report location).
 512           PathDiagnosticLocation UPDLoc = D->getUniqueingLoc();
 513           if (UPDLoc.isValid()) {
 514             FullSourceLoc UL(SM->getExpansionLoc(UPDLoc.asLocation()),
 515                              *SM);
 516             FullSourceLoc UFunL(SM->getExpansionLoc(
 517               D->getUniqueingDecl()->getBody()->getLocStart()), *SM);
 518             o << "  <key>issue_hash</key><string>"
 519               << UL.getExpansionLineNumber() - UFunL.getExpansionLineNumber()
 520               << "</string>\n";
 521
 522           // Otherwise, use the location on which the bug is reported.
 523           } else {
 524             FullSourceLoc L(SM->getExpansionLoc(D->getLocation().asLocation()),
 525                             *SM);
 526             FullSourceLoc FunL(SM->getExpansionLoc(Body->getLocStart()), *SM);
 527             o << "  <key>issue_hash</key><string>"
 528               << L.getExpansionLineNumber() - FunL.getExpansionLineNumber()
 529               << "</string>\n";
 530           }
 531
 532         }
 533       }
 534     }
 535
 536     // Output the location of the bug.
 537     o << "  <key>location</key>\n";
 538     EmitLocation(o, *SM, LangOpts, D->getLocation(), FM, 2);
 539
 540     // Output the diagnostic to the sub-diagnostic client, if any.
 541     if (!filesMade->empty()) {
 542       StringRef lastName;
 543       PDFileEntry::ConsumerFiles *files = filesMade->getFiles(*D);
 544       if (files) {
 545         for (PDFileEntry::ConsumerFiles::const_iterator CI = files->begin(),
 546                 CE = files->end(); CI != CE; ++CI) {
 547           StringRef newName = CI->first;
 548           if (newName != lastName) {
 549             if (!lastName.empty()) {
 550               o << "  </array>\n";
 551             }
 552             lastName = newName;
 553             o <<  "  <key>" << lastName << "_files</key>\n";
 554             o << "  <array>\n";
 555           }
 556           o << "   <string>" << CI->second << "</string>\n";
 557         }
 558         o << "  </array>\n";
 559       }
 560     }
 561
 562     // Close up the entry.
 563     o << "  </dict>\n";
 564   }
 565
 566   o << " </array>\n";
 567
 568   // Finish.
 569   o << "</dict>\n</plist>";
 570 }