PlistDiagnostics.cpp revision 5fba5a789a238c29ef811a39a39be722443ec1b1 
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"
25using namespace clang;
26using namespace ento;
27
28typedef llvm::DenseMap<FileID, unsigned> FIDMap;
29
30
31namespace {
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
59PlistDiagnostics::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
67void 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
75void 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
83static 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
93static 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
101static raw_ostream &Indent(raw_ostream &o, const unsigned indent) {
102  for (unsigned i = 0; i < indent; ++i) o << ' ';
103  return o;
104}
105
106static 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
127static 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
134static 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
144static 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
161static 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
212static 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
271static 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
280static 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
315static 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
328static 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
334static 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
363void 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.back();
386      WorkList.pop_back();
387
388      for (PathPieces::const_iterator I = path.begin(), E = path.end();
389           I!=E; ++I) {
390        const PathDiagnosticPiece *piece = I->getPtr();
391        AddFID(FM, Fids, SM, piece->getLocation().asLocation());
392        ArrayRef<SourceRange> Ranges = piece->getRanges();
393        for (ArrayRef<SourceRange>::iterator I = Ranges.begin(),
394                                             E = Ranges.end(); I != E; ++I) {
395          AddFID(FM, Fids, SM, I->getBegin());
396          AddFID(FM, Fids, SM, I->getEnd());
397        }
398
399        if (const PathDiagnosticCallPiece *call =
400            dyn_cast<PathDiagnosticCallPiece>(piece)) {
401          IntrusiveRefCntPtr<PathDiagnosticEventPiece>
402            callEnterWithin = call->getCallEnterWithinCallerEvent();
403          if (callEnterWithin)
404            AddFID(FM, Fids, SM, callEnterWithin->getLocation().asLocation());
405
406          WorkList.push_back(&call->path);
407        }
408        else if (const PathDiagnosticMacroPiece *macro =
409                 dyn_cast<PathDiagnosticMacroPiece>(piece)) {
410          WorkList.push_back(&macro->subPieces);
411        }
412      }
413    }
414  }
415
416  // Open the file.
417  std::string ErrMsg;
418  llvm::raw_fd_ostream o(OutputFile.c_str(), ErrMsg);
419  if (!ErrMsg.empty()) {
420    llvm::errs() << "warning: could not create file: " << OutputFile << '\n';
421    return;
422  }
423
424  // Write the plist header.
425  o << "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
426  "<!DOCTYPE plist PUBLIC \"-//Apple Computer//DTD PLIST 1.0//EN\" "
427  "\"http://www.apple.com/DTDs/PropertyList-1.0.dtd\">\n"
428  "<plist version=\"1.0\">\n";
429
430  // Write the root object: a <dict> containing...
431  //  - "clang_version", the string representation of clang version
432  //  - "files", an <array> mapping from FIDs to file names
433  //  - "diagnostics", an <array> containing the path diagnostics
434  o << "<dict>\n" <<
435       " <key>clang_version</key>\n";
436  EmitString(o, getClangFullVersion()) << '\n';
437  o << " <key>files</key>\n"
438       " <array>\n";
439
440  for (SmallVectorImpl<FileID>::iterator I=Fids.begin(), E=Fids.end();
441       I!=E; ++I) {
442    o << "  ";
443    EmitString(o, SM->getFileEntryForID(*I)->getName()) << '\n';
444  }
445
446  o << " </array>\n"
447       " <key>diagnostics</key>\n"
448       " <array>\n";
449
450  for (std::vector<const PathDiagnostic*>::iterator DI=Diags.begin(),
451       DE = Diags.end(); DI!=DE; ++DI) {
452
453    o << "  <dict>\n"
454         "   <key>path</key>\n";
455
456    const PathDiagnostic *D = *DI;
457
458    o << "   <array>\n";
459
460    for (PathPieces::const_iterator I = D->path.begin(), E = D->path.end();
461         I != E; ++I)
462      ReportDiag(o, **I, FM, *SM, LangOpts);
463
464    o << "   </array>\n";
465
466    // Output the bug type and bug category.
467    o << "   <key>description</key>";
468    EmitString(o, D->getShortDescription()) << '\n';
469    o << "   <key>category</key>";
470    EmitString(o, D->getCategory()) << '\n';
471    o << "   <key>type</key>";
472    EmitString(o, D->getBugType()) << '\n';
473
474    // Output information about the semantic context where
475    // the issue occurred.
476    if (const Decl *DeclWithIssue = D->getDeclWithIssue()) {
477      // FIXME: handle blocks, which have no name.
478      if (const NamedDecl *ND = dyn_cast<NamedDecl>(DeclWithIssue)) {
479        StringRef declKind;
480        switch (ND->getKind()) {
481          case Decl::CXXRecord:
482            declKind = "C++ class";
483            break;
484          case Decl::CXXMethod:
485            declKind = "C++ method";
486            break;
487          case Decl::ObjCMethod:
488            declKind = "Objective-C method";
489            break;
490          case Decl::Function:
491            declKind = "function";
492            break;
493          default:
494            break;
495        }
496        if (!declKind.empty()) {
497          const std::string &declName = ND->getDeclName().getAsString();
498          o << "  <key>issue_context_kind</key>";
499          EmitString(o, declKind) << '\n';
500          o << "  <key>issue_context</key>";
501          EmitString(o, declName) << '\n';
502        }
503
504        // Output the bug hash for issue unique-ing. Currently, it's just an
505        // offset from the beginning of the function.
506        if (const Stmt *Body = DeclWithIssue->getBody()) {
507
508          // If the bug uniqueing location exists, use it for the hash.
509          // For example, this ensures that two leaks reported on the same line
510          // will have different issue_hashes and that the hash will identify
511          // the leak location even after code is added between the allocation
512          // site and the end of scope (leak report location).
513          PathDiagnosticLocation UPDLoc = D->getUniqueingLoc();
514          if (UPDLoc.isValid()) {
515            FullSourceLoc UL(SM->getExpansionLoc(UPDLoc.asLocation()),
516                             *SM);
517            FullSourceLoc UFunL(SM->getExpansionLoc(
518              D->getUniqueingDecl()->getBody()->getLocStart()), *SM);
519            o << "  <key>issue_hash</key><string>"
520              << UL.getExpansionLineNumber() - UFunL.getExpansionLineNumber()
521              << "</string>\n";
522
523          // Otherwise, use the location on which the bug is reported.
524          } else {
525            FullSourceLoc L(SM->getExpansionLoc(D->getLocation().asLocation()),
526                            *SM);
527            FullSourceLoc FunL(SM->getExpansionLoc(Body->getLocStart()), *SM);
528            o << "  <key>issue_hash</key><string>"
529              << L.getExpansionLineNumber() - FunL.getExpansionLineNumber()
530              << "</string>\n";
531          }
532
533        }
534      }
535    }
536
537    // Output the location of the bug.
538    o << "  <key>location</key>\n";
539    EmitLocation(o, *SM, LangOpts, D->getLocation(), FM, 2);
540
541    // Output the diagnostic to the sub-diagnostic client, if any.
542    if (!filesMade->empty()) {
543      StringRef lastName;
544      PDFileEntry::ConsumerFiles *files = filesMade->getFiles(*D);
545      if (files) {
546        for (PDFileEntry::ConsumerFiles::const_iterator CI = files->begin(),
547                CE = files->end(); CI != CE; ++CI) {
548          StringRef newName = CI->first;
549          if (newName != lastName) {
550            if (!lastName.empty()) {
551              o << "  </array>\n";
552            }
553            lastName = newName;
554            o <<  "  <key>" << lastName << "_files</key>\n";
555            o << "  <array>\n";
556          }
557          o << "   <string>" << CI->second << "</string>\n";
558        }
559        o << "  </array>\n";
560      }
561    }
562
563    // Close up the entry.
564    o << "  </dict>\n";
565  }
566
567  o << " </array>\n";
568
569  // Finish.
570  o << "</dict>\n</plist>";
571}
572