PlistDiagnostics.cpp revision 5a8e1ad062420ef74707bf093889403d07664b17 
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    bool supportsAllBlockEdges() const { return true; }
54    virtual bool supportsCrossFileDiagnostics() const {
55      return SupportsCrossFileDiagnostics;
56    }
57  };
58} // end anonymous namespace
59
60PlistDiagnostics::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
68void 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
76void 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
84static 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
94static 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
102static raw_ostream &Indent(raw_ostream &o, const unsigned indent) {
103  for (unsigned i = 0; i < indent; ++i) o << ' ';
104  return o;
105}
106
107static 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
128static 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
135static 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
145static 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
162static 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
213static 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                        bool isKeyEvent = false) {
220
221  Indent(o, indent) << "<dict>\n";
222  ++indent;
223
224  Indent(o, indent) << "<key>kind</key><string>event</string>\n";
225
226  if (isKeyEvent) {
227    Indent(o, indent) << "<key>key_event</key><true/>\n";
228  }
229
230  // Output the location.
231  FullSourceLoc L = P.getLocation().asLocation();
232
233  Indent(o, indent) << "<key>location</key>\n";
234  EmitLocation(o, SM, LangOpts, L, FM, indent);
235
236  // Output the ranges (if any).
237  ArrayRef<SourceRange> Ranges = P.getRanges();
238
239  if (!Ranges.empty()) {
240    Indent(o, indent) << "<key>ranges</key>\n";
241    Indent(o, indent) << "<array>\n";
242    ++indent;
243    for (ArrayRef<SourceRange>::iterator I = Ranges.begin(), E = Ranges.end();
244         I != E; ++I) {
245      EmitRange(o, SM, LangOpts, *I, FM, indent+1);
246    }
247    --indent;
248    Indent(o, indent) << "</array>\n";
249  }
250
251  // Output the call depth.
252  Indent(o, indent) << "<key>depth</key>"
253                    << "<integer>" << depth << "</integer>\n";
254
255  // Output the text.
256  assert(!P.getString().empty());
257  Indent(o, indent) << "<key>extended_message</key>\n";
258  Indent(o, indent);
259  EmitString(o, P.getString()) << '\n';
260
261  // Output the short text.
262  // FIXME: Really use a short string.
263  Indent(o, indent) << "<key>message</key>\n";
264  Indent(o, indent);
265  EmitString(o, P.getString()) << '\n';
266
267  // Finish up.
268  --indent;
269  Indent(o, indent); o << "</dict>\n";
270}
271
272static void ReportPiece(raw_ostream &o,
273                        const PathDiagnosticPiece &P,
274                        const FIDMap& FM, const SourceManager &SM,
275                        const LangOptions &LangOpts,
276                        unsigned indent,
277                        unsigned depth,
278                        bool includeControlFlow,
279                        bool isKeyEvent = false);
280
281static void ReportCall(raw_ostream &o,
282                       const PathDiagnosticCallPiece &P,
283                       const FIDMap& FM, const SourceManager &SM,
284                       const LangOptions &LangOpts,
285                       unsigned indent,
286                       unsigned depth) {
287
288  IntrusiveRefCntPtr<PathDiagnosticEventPiece> callEnter =
289    P.getCallEnterEvent();
290
291  if (callEnter)
292    ReportPiece(o, *callEnter, FM, SM, LangOpts, indent, depth, true,
293                P.isLastInMainSourceFile());
294
295  IntrusiveRefCntPtr<PathDiagnosticEventPiece> callEnterWithinCaller =
296    P.getCallEnterWithinCallerEvent();
297
298  ++depth;
299
300  if (callEnterWithinCaller)
301    ReportPiece(o, *callEnterWithinCaller, FM, SM, LangOpts,
302                indent, depth, true);
303
304  for (PathPieces::const_iterator I = P.path.begin(), E = P.path.end();I!=E;++I)
305    ReportPiece(o, **I, FM, SM, LangOpts, indent, depth, true);
306
307  --depth;
308
309  IntrusiveRefCntPtr<PathDiagnosticEventPiece> callExit =
310    P.getCallExitEvent();
311
312  if (callExit)
313    ReportPiece(o, *callExit, FM, SM, LangOpts, indent, depth, true);
314}
315
316static void ReportMacro(raw_ostream &o,
317                        const PathDiagnosticMacroPiece& P,
318                        const FIDMap& FM, const SourceManager &SM,
319                        const LangOptions &LangOpts,
320                        unsigned indent,
321                        unsigned depth) {
322
323  for (PathPieces::const_iterator I = P.subPieces.begin(), E=P.subPieces.end();
324       I!=E; ++I) {
325    ReportPiece(o, **I, FM, SM, LangOpts, indent, depth, false);
326  }
327}
328
329static void ReportDiag(raw_ostream &o, const PathDiagnosticPiece& P,
330                       const FIDMap& FM, const SourceManager &SM,
331                       const LangOptions &LangOpts) {
332  ReportPiece(o, P, FM, SM, LangOpts, 4, 0, true);
333}
334
335static void ReportPiece(raw_ostream &o,
336                        const PathDiagnosticPiece &P,
337                        const FIDMap& FM, const SourceManager &SM,
338                        const LangOptions &LangOpts,
339                        unsigned indent,
340                        unsigned depth,
341                        bool includeControlFlow,
342                        bool isKeyEvent) {
343  switch (P.getKind()) {
344    case PathDiagnosticPiece::ControlFlow:
345      if (includeControlFlow)
346        ReportControlFlow(o, cast<PathDiagnosticControlFlowPiece>(P), FM, SM,
347                          LangOpts, indent);
348      break;
349    case PathDiagnosticPiece::Call:
350      ReportCall(o, cast<PathDiagnosticCallPiece>(P), FM, SM, LangOpts,
351                 indent, depth);
352      break;
353    case PathDiagnosticPiece::Event:
354      ReportEvent(o, cast<PathDiagnosticSpotPiece>(P), FM, SM, LangOpts,
355                  indent, depth, isKeyEvent);
356      break;
357    case PathDiagnosticPiece::Macro:
358      ReportMacro(o, cast<PathDiagnosticMacroPiece>(P), FM, SM, LangOpts,
359                  indent, depth);
360      break;
361  }
362}
363
364void PlistDiagnostics::FlushDiagnosticsImpl(
365                                    std::vector<const PathDiagnostic *> &Diags,
366                                    FilesMade *filesMade) {
367  // Build up a set of FIDs that we use by scanning the locations and
368  // ranges of the diagnostics.
369  FIDMap FM;
370  SmallVector<FileID, 10> Fids;
371  const SourceManager* SM = 0;
372
373  if (!Diags.empty())
374    SM = &(*(*Diags.begin())->path.begin())->getLocation().getManager();
375
376
377  for (std::vector<const PathDiagnostic*>::iterator DI = Diags.begin(),
378       DE = Diags.end(); DI != DE; ++DI) {
379
380    const PathDiagnostic *D = *DI;
381
382    SmallVector<const PathPieces *, 5> WorkList;
383    WorkList.push_back(&D->path);
384
385    while (!WorkList.empty()) {
386      const PathPieces &path = *WorkList.back();
387      WorkList.pop_back();
388
389      for (PathPieces::const_iterator I = path.begin(), E = path.end();
390           I!=E; ++I) {
391        const PathDiagnosticPiece *piece = I->getPtr();
392        AddFID(FM, Fids, SM, piece->getLocation().asLocation());
393        ArrayRef<SourceRange> Ranges = piece->getRanges();
394        for (ArrayRef<SourceRange>::iterator I = Ranges.begin(),
395                                             E = Ranges.end(); I != E; ++I) {
396          AddFID(FM, Fids, SM, I->getBegin());
397          AddFID(FM, Fids, SM, I->getEnd());
398        }
399
400        if (const PathDiagnosticCallPiece *call =
401            dyn_cast<PathDiagnosticCallPiece>(piece)) {
402          IntrusiveRefCntPtr<PathDiagnosticEventPiece>
403            callEnterWithin = call->getCallEnterWithinCallerEvent();
404          if (callEnterWithin)
405            AddFID(FM, Fids, SM, callEnterWithin->getLocation().asLocation());
406
407          WorkList.push_back(&call->path);
408        }
409        else if (const PathDiagnosticMacroPiece *macro =
410                 dyn_cast<PathDiagnosticMacroPiece>(piece)) {
411          WorkList.push_back(&macro->subPieces);
412        }
413      }
414    }
415  }
416
417  // Open the file.
418  std::string ErrMsg;
419  llvm::raw_fd_ostream o(OutputFile.c_str(), ErrMsg);
420  if (!ErrMsg.empty()) {
421    llvm::errs() << "warning: could not create file: " << OutputFile << '\n';
422    return;
423  }
424
425  // Write the plist header.
426  o << "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
427  "<!DOCTYPE plist PUBLIC \"-//Apple Computer//DTD PLIST 1.0//EN\" "
428  "\"http://www.apple.com/DTDs/PropertyList-1.0.dtd\">\n"
429  "<plist version=\"1.0\">\n";
430
431  // Write the root object: a <dict> containing...
432  //  - "clang_version", the string representation of clang version
433  //  - "files", an <array> mapping from FIDs to file names
434  //  - "diagnostics", an <array> containing the path diagnostics
435  o << "<dict>\n" <<
436       " <key>clang_version</key>\n";
437  EmitString(o, getClangFullVersion()) << '\n';
438  o << " <key>files</key>\n"
439       " <array>\n";
440
441  for (SmallVectorImpl<FileID>::iterator I=Fids.begin(), E=Fids.end();
442       I!=E; ++I) {
443    o << "  ";
444    EmitString(o, SM->getFileEntryForID(*I)->getName()) << '\n';
445  }
446
447  o << " </array>\n"
448       " <key>diagnostics</key>\n"
449       " <array>\n";
450
451  for (std::vector<const PathDiagnostic*>::iterator DI=Diags.begin(),
452       DE = Diags.end(); DI!=DE; ++DI) {
453
454    o << "  <dict>\n"
455         "   <key>path</key>\n";
456
457    const PathDiagnostic *D = *DI;
458
459    o << "   <array>\n";
460
461    for (PathPieces::const_iterator I = D->path.begin(), E = D->path.end();
462         I != E; ++I)
463      ReportDiag(o, **I, FM, *SM, LangOpts);
464
465    o << "   </array>\n";
466
467    // Output the bug type and bug category.
468    o << "   <key>description</key>";
469    EmitString(o, D->getShortDescription()) << '\n';
470    o << "   <key>category</key>";
471    EmitString(o, D->getCategory()) << '\n';
472    o << "   <key>type</key>";
473    EmitString(o, D->getBugType()) << '\n';
474
475    // Output information about the semantic context where
476    // the issue occurred.
477    if (const Decl *DeclWithIssue = D->getDeclWithIssue()) {
478      // FIXME: handle blocks, which have no name.
479      if (const NamedDecl *ND = dyn_cast<NamedDecl>(DeclWithIssue)) {
480        StringRef declKind;
481        switch (ND->getKind()) {
482          case Decl::CXXRecord:
483            declKind = "C++ class";
484            break;
485          case Decl::CXXMethod:
486            declKind = "C++ method";
487            break;
488          case Decl::ObjCMethod:
489            declKind = "Objective-C method";
490            break;
491          case Decl::Function:
492            declKind = "function";
493            break;
494          default:
495            break;
496        }
497        if (!declKind.empty()) {
498          const std::string &declName = ND->getDeclName().getAsString();
499          o << "  <key>issue_context_kind</key>";
500          EmitString(o, declKind) << '\n';
501          o << "  <key>issue_context</key>";
502          EmitString(o, declName) << '\n';
503        }
504
505        // Output the bug hash for issue unique-ing. Currently, it's just an
506        // offset from the beginning of the function.
507        if (const Stmt *Body = DeclWithIssue->getBody()) {
508
509          // If the bug uniqueing location exists, use it for the hash.
510          // For example, this ensures that two leaks reported on the same line
511          // will have different issue_hashes and that the hash will identify
512          // the leak location even after code is added between the allocation
513          // site and the end of scope (leak report location).
514          PathDiagnosticLocation UPDLoc = D->getUniqueingLoc();
515          if (UPDLoc.isValid()) {
516            FullSourceLoc UL(SM->getExpansionLoc(UPDLoc.asLocation()),
517                             *SM);
518            FullSourceLoc UFunL(SM->getExpansionLoc(
519              D->getUniqueingDecl()->getBody()->getLocStart()), *SM);
520            o << "  <key>issue_hash</key><string>"
521              << UL.getExpansionLineNumber() - UFunL.getExpansionLineNumber()
522              << "</string>\n";
523
524          // Otherwise, use the location on which the bug is reported.
525          } else {
526            FullSourceLoc L(SM->getExpansionLoc(D->getLocation().asLocation()),
527                            *SM);
528            FullSourceLoc FunL(SM->getExpansionLoc(Body->getLocStart()), *SM);
529            o << "  <key>issue_hash</key><string>"
530              << L.getExpansionLineNumber() - FunL.getExpansionLineNumber()
531              << "</string>\n";
532          }
533
534        }
535      }
536    }
537
538    // Output the location of the bug.
539    o << "  <key>location</key>\n";
540    EmitLocation(o, *SM, LangOpts, D->getLocation(), FM, 2);
541
542    // Output the diagnostic to the sub-diagnostic client, if any.
543    if (!filesMade->empty()) {
544      StringRef lastName;
545      PDFileEntry::ConsumerFiles *files = filesMade->getFiles(*D);
546      if (files) {
547        for (PDFileEntry::ConsumerFiles::const_iterator CI = files->begin(),
548                CE = files->end(); CI != CE; ++CI) {
549          StringRef newName = CI->first;
550          if (newName != lastName) {
551            if (!lastName.empty()) {
552              o << "  </array>\n";
553            }
554            lastName = newName;
555            o <<  "  <key>" << lastName << "_files</key>\n";
556            o << "  <array>\n";
557          }
558          o << "   <string>" << CI->second << "</string>\n";
559        }
560        o << "  </array>\n";
561      }
562    }
563
564    // Close up the entry.
565    o << "  </dict>\n";
566  }
567
568  o << " </array>\n";
569
570  // Finish.
571  o << "</dict>\n</plist>";
572}
573