HTMLDiagnostics.cpp revision 4970ef8e3527ac356c3e9fde0710561fcb63e424 
1//===--- HTMLDiagnostics.cpp - HTML 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 HTMLDiagnostics object.
11//
12//===----------------------------------------------------------------------===//
13
14#include "clang/StaticAnalyzer/Core/PathDiagnosticConsumers.h"
15#include "clang/StaticAnalyzer/Core/BugReporter/PathDiagnostic.h"
16#include "clang/AST/ASTContext.h"
17#include "clang/AST/Decl.h"
18#include "clang/Basic/SourceManager.h"
19#include "clang/Basic/FileManager.h"
20#include "clang/Rewrite/Rewriter.h"
21#include "clang/Rewrite/HTMLRewrite.h"
22#include "clang/Lex/Lexer.h"
23#include "clang/Lex/Preprocessor.h"
24#include "llvm/Support/FileSystem.h"
25#include "llvm/Support/MemoryBuffer.h"
26#include "llvm/Support/raw_ostream.h"
27#include "llvm/Support/Path.h"
28
29using namespace clang;
30using namespace ento;
31
32//===----------------------------------------------------------------------===//
33// Boilerplate.
34//===----------------------------------------------------------------------===//
35
36namespace {
37
38class HTMLDiagnostics : public PathDiagnosticConsumer {
39  llvm::sys::Path Directory, FilePrefix;
40  bool createdDir, noDir;
41  const Preprocessor &PP;
42public:
43  HTMLDiagnostics(const std::string& prefix, const Preprocessor &pp);
44
45  virtual ~HTMLDiagnostics() { FlushDiagnostics(NULL); }
46
47  virtual void FlushDiagnosticsImpl(std::vector<const PathDiagnostic *> &Diags,
48                                    SmallVectorImpl<std::string> *FilesMade);
49
50  virtual StringRef getName() const {
51    return "HTMLDiagnostics";
52  }
53
54  unsigned ProcessMacroPiece(raw_ostream &os,
55                             const PathDiagnosticMacroPiece& P,
56                             unsigned num);
57
58  void HandlePiece(Rewriter& R, FileID BugFileID,
59                   const PathDiagnosticPiece& P, unsigned num, unsigned max);
60
61  void HighlightRange(Rewriter& R, FileID BugFileID, SourceRange Range,
62                      const char *HighlightStart = "<span class=\"mrange\">",
63                      const char *HighlightEnd = "</span>");
64
65  void ReportDiag(const PathDiagnostic& D,
66                  SmallVectorImpl<std::string> *FilesMade);
67};
68
69} // end anonymous namespace
70
71HTMLDiagnostics::HTMLDiagnostics(const std::string& prefix,
72                                 const Preprocessor &pp)
73  : Directory(prefix), FilePrefix(prefix), createdDir(false), noDir(false),
74    PP(pp) {
75  // All html files begin with "report"
76  FilePrefix.appendComponent("report");
77}
78
79PathDiagnosticConsumer*
80ento::createHTMLDiagnosticConsumer(const std::string& prefix,
81                                 const Preprocessor &PP) {
82  return new HTMLDiagnostics(prefix, PP);
83}
84
85//===----------------------------------------------------------------------===//
86// Report processing.
87//===----------------------------------------------------------------------===//
88
89void HTMLDiagnostics::FlushDiagnosticsImpl(
90  std::vector<const PathDiagnostic *> &Diags,
91  SmallVectorImpl<std::string> *FilesMade) {
92  for (std::vector<const PathDiagnostic *>::iterator it = Diags.begin(),
93       et = Diags.end(); it != et; ++it) {
94    ReportDiag(**it, FilesMade);
95  }
96}
97
98static void flattenPath(PathPieces &path, const PathPieces &oldPath) {
99  for (PathPieces::const_iterator it = oldPath.begin(), et = oldPath.end();
100       it != et; ++it ) {
101    PathDiagnosticPiece *piece = it->getPtr();
102    if (const PathDiagnosticCallPiece *call =
103        dyn_cast<PathDiagnosticCallPiece>(piece)) {
104      IntrusiveRefCntPtr<PathDiagnosticEventPiece> callEnter =
105        call->getCallEnterEvent();
106      if (callEnter)
107        path.push_back(callEnter);
108      flattenPath(path, call->path);
109      IntrusiveRefCntPtr<PathDiagnosticEventPiece> callExit =
110        call->getCallExitEvent();
111      if (callExit)
112        path.push_back(callExit);
113      continue;
114    }
115
116    path.push_back(piece);
117  }
118}
119
120void HTMLDiagnostics::ReportDiag(const PathDiagnostic& D,
121                                 SmallVectorImpl<std::string> *FilesMade) {
122
123  // Create the HTML directory if it is missing.
124  if (!createdDir) {
125    createdDir = true;
126    std::string ErrorMsg;
127    Directory.createDirectoryOnDisk(true, &ErrorMsg);
128
129    bool IsDirectory;
130    if (llvm::sys::fs::is_directory(Directory.str(), IsDirectory) ||
131        !IsDirectory) {
132      llvm::errs() << "warning: could not create directory '"
133                   << Directory.str() << "'\n"
134                   << "reason: " << ErrorMsg << '\n';
135
136      noDir = true;
137
138      return;
139    }
140  }
141
142  if (noDir)
143    return;
144
145  // First flatten out the entire path to make it easier to use.
146  PathPieces path;
147  flattenPath(path, D.path);
148
149  const SourceManager &SMgr = (*path.begin())->getLocation().getManager();
150  FileID FID;
151
152  // Verify that the entire path is from the same FileID.
153  for (PathPieces::const_iterator I = path.begin(), E = path.end();
154       I != E; ++I) {
155    FullSourceLoc L = (*I)->getLocation().asLocation().getExpansionLoc();
156
157    if (FID.isInvalid()) {
158      FID = SMgr.getFileID(L);
159    } else if (SMgr.getFileID(L) != FID)
160      return; // FIXME: Emit a warning?
161
162    // Check the source ranges.
163    for (PathDiagnosticPiece::range_iterator RI = (*I)->ranges_begin(),
164                                             RE = (*I)->ranges_end();
165                                             RI != RE; ++RI) {
166      SourceLocation L = SMgr.getExpansionLoc(RI->getBegin());
167      if (!L.isFileID() || SMgr.getFileID(L) != FID)
168        return; // FIXME: Emit a warning?
169      L = SMgr.getExpansionLoc(RI->getEnd());
170      if (!L.isFileID() || SMgr.getFileID(L) != FID)
171        return; // FIXME: Emit a warning?
172    }
173  }
174
175  if (FID.isInvalid())
176    return; // FIXME: Emit a warning?
177
178  // Create a new rewriter to generate HTML.
179  Rewriter R(const_cast<SourceManager&>(SMgr), PP.getLangOptions());
180
181  // Process the path.
182  unsigned n = path.size();
183  unsigned max = n;
184
185  for (PathPieces::const_reverse_iterator I = path.rbegin(),
186       E = path.rend();
187        I != E; ++I, --n)
188    HandlePiece(R, FID, **I, n, max);
189
190  // Add line numbers, header, footer, etc.
191
192  // unsigned FID = R.getSourceMgr().getMainFileID();
193  html::EscapeText(R, FID);
194  html::AddLineNumbers(R, FID);
195
196  // If we have a preprocessor, relex the file and syntax highlight.
197  // We might not have a preprocessor if we come from a deserialized AST file,
198  // for example.
199
200  html::SyntaxHighlight(R, FID, PP);
201  html::HighlightMacros(R, FID, PP);
202
203  // Get the full directory name of the analyzed file.
204
205  const FileEntry* Entry = SMgr.getFileEntryForID(FID);
206
207  // This is a cludge; basically we want to append either the full
208  // working directory if we have no directory information.  This is
209  // a work in progress.
210
211  std::string DirName = "";
212
213  if (llvm::sys::path::is_relative(Entry->getName())) {
214    llvm::sys::Path P = llvm::sys::Path::GetCurrentDirectory();
215    DirName = P.str() + "/";
216  }
217
218  // Add the name of the file as an <h1> tag.
219
220  {
221    std::string s;
222    llvm::raw_string_ostream os(s);
223
224    os << "<!-- REPORTHEADER -->\n"
225      << "<h3>Bug Summary</h3>\n<table class=\"simpletable\">\n"
226          "<tr><td class=\"rowname\">File:</td><td>"
227      << html::EscapeText(DirName)
228      << html::EscapeText(Entry->getName())
229      << "</td></tr>\n<tr><td class=\"rowname\">Location:</td><td>"
230         "<a href=\"#EndPath\">line "
231      << (*path.rbegin())->getLocation().asLocation().getExpansionLineNumber()
232      << ", column "
233      << (*path.rbegin())->getLocation().asLocation().getExpansionColumnNumber()
234      << "</a></td></tr>\n"
235         "<tr><td class=\"rowname\">Description:</td><td>"
236      << D.getDescription() << "</td></tr>\n";
237
238    // Output any other meta data.
239
240    for (PathDiagnostic::meta_iterator I=D.meta_begin(), E=D.meta_end();
241         I!=E; ++I) {
242      os << "<tr><td></td><td>" << html::EscapeText(*I) << "</td></tr>\n";
243    }
244
245    os << "</table>\n<!-- REPORTSUMMARYEXTRA -->\n"
246          "<h3>Annotated Source Code</h3>\n";
247
248    R.InsertTextBefore(SMgr.getLocForStartOfFile(FID), os.str());
249  }
250
251  // Embed meta-data tags.
252  {
253    std::string s;
254    llvm::raw_string_ostream os(s);
255
256    const std::string& BugDesc = D.getDescription();
257    if (!BugDesc.empty())
258      os << "\n<!-- BUGDESC " << BugDesc << " -->\n";
259
260    const std::string& BugType = D.getBugType();
261    if (!BugType.empty())
262      os << "\n<!-- BUGTYPE " << BugType << " -->\n";
263
264    const std::string& BugCategory = D.getCategory();
265    if (!BugCategory.empty())
266      os << "\n<!-- BUGCATEGORY " << BugCategory << " -->\n";
267
268    os << "\n<!-- BUGFILE " << DirName << Entry->getName() << " -->\n";
269
270    os << "\n<!-- BUGLINE "
271       << path.back()->getLocation().asLocation().getExpansionLineNumber()
272       << " -->\n";
273
274    os << "\n<!-- BUGPATHLENGTH " << path.size() << " -->\n";
275
276    // Mark the end of the tags.
277    os << "\n<!-- BUGMETAEND -->\n";
278
279    // Insert the text.
280    R.InsertTextBefore(SMgr.getLocForStartOfFile(FID), os.str());
281  }
282
283  // Add CSS, header, and footer.
284
285  html::AddHeaderFooterInternalBuiltinCSS(R, FID, Entry->getName());
286
287  // Get the rewrite buffer.
288  const RewriteBuffer *Buf = R.getRewriteBufferFor(FID);
289
290  if (!Buf) {
291    llvm::errs() << "warning: no diagnostics generated for main file.\n";
292    return;
293  }
294
295  // Create a path for the target HTML file.
296  llvm::sys::Path F(FilePrefix);
297  F.makeUnique(false, NULL);
298
299  // Rename the file with an HTML extension.
300  llvm::sys::Path H(F);
301  H.appendSuffix("html");
302  F.renamePathOnDisk(H, NULL);
303
304  std::string ErrorMsg;
305  llvm::raw_fd_ostream os(H.c_str(), ErrorMsg);
306
307  if (!ErrorMsg.empty()) {
308    llvm::errs() << "warning: could not create file '" << F.str()
309                 << "'\n";
310    return;
311  }
312
313  if (FilesMade)
314    FilesMade->push_back(llvm::sys::path::filename(H.str()));
315
316  // Emit the HTML to disk.
317  for (RewriteBuffer::iterator I = Buf->begin(), E = Buf->end(); I!=E; ++I)
318      os << *I;
319}
320
321void HTMLDiagnostics::HandlePiece(Rewriter& R, FileID BugFileID,
322                                  const PathDiagnosticPiece& P,
323                                  unsigned num, unsigned max) {
324
325  // For now, just draw a box above the line in question, and emit the
326  // warning.
327  FullSourceLoc Pos = P.getLocation().asLocation();
328
329  if (!Pos.isValid())
330    return;
331
332  SourceManager &SM = R.getSourceMgr();
333  assert(&Pos.getManager() == &SM && "SourceManagers are different!");
334  std::pair<FileID, unsigned> LPosInfo = SM.getDecomposedExpansionLoc(Pos);
335
336  if (LPosInfo.first != BugFileID)
337    return;
338
339  const llvm::MemoryBuffer *Buf = SM.getBuffer(LPosInfo.first);
340  const char* FileStart = Buf->getBufferStart();
341
342  // Compute the column number.  Rewind from the current position to the start
343  // of the line.
344  unsigned ColNo = SM.getColumnNumber(LPosInfo.first, LPosInfo.second);
345  const char *TokInstantiationPtr =Pos.getExpansionLoc().getCharacterData();
346  const char *LineStart = TokInstantiationPtr-ColNo;
347
348  // Compute LineEnd.
349  const char *LineEnd = TokInstantiationPtr;
350  const char* FileEnd = Buf->getBufferEnd();
351  while (*LineEnd != '\n' && LineEnd != FileEnd)
352    ++LineEnd;
353
354  // Compute the margin offset by counting tabs and non-tabs.
355  unsigned PosNo = 0;
356  for (const char* c = LineStart; c != TokInstantiationPtr; ++c)
357    PosNo += *c == '\t' ? 8 : 1;
358
359  // Create the html for the message.
360
361  const char *Kind = 0;
362  switch (P.getKind()) {
363  case PathDiagnosticPiece::Call:
364      llvm_unreachable("Calls should already be handled");
365  case PathDiagnosticPiece::Event:  Kind = "Event"; break;
366  case PathDiagnosticPiece::ControlFlow: Kind = "Control"; break;
367    // Setting Kind to "Control" is intentional.
368  case PathDiagnosticPiece::Macro: Kind = "Control"; break;
369  }
370
371  std::string sbuf;
372  llvm::raw_string_ostream os(sbuf);
373
374  os << "\n<tr><td class=\"num\"></td><td class=\"line\"><div id=\"";
375
376  if (num == max)
377    os << "EndPath";
378  else
379    os << "Path" << num;
380
381  os << "\" class=\"msg";
382  if (Kind)
383    os << " msg" << Kind;
384  os << "\" style=\"margin-left:" << PosNo << "ex";
385
386  // Output a maximum size.
387  if (!isa<PathDiagnosticMacroPiece>(P)) {
388    // Get the string and determining its maximum substring.
389    const std::string& Msg = P.getString();
390    unsigned max_token = 0;
391    unsigned cnt = 0;
392    unsigned len = Msg.size();
393
394    for (std::string::const_iterator I=Msg.begin(), E=Msg.end(); I!=E; ++I)
395      switch (*I) {
396      default:
397        ++cnt;
398        continue;
399      case ' ':
400      case '\t':
401      case '\n':
402        if (cnt > max_token) max_token = cnt;
403        cnt = 0;
404      }
405
406    if (cnt > max_token)
407      max_token = cnt;
408
409    // Determine the approximate size of the message bubble in em.
410    unsigned em;
411    const unsigned max_line = 120;
412
413    if (max_token >= max_line)
414      em = max_token / 2;
415    else {
416      unsigned characters = max_line;
417      unsigned lines = len / max_line;
418
419      if (lines > 0) {
420        for (; characters > max_token; --characters)
421          if (len / characters > lines) {
422            ++characters;
423            break;
424          }
425      }
426
427      em = characters / 2;
428    }
429
430    if (em < max_line/2)
431      os << "; max-width:" << em << "em";
432  }
433  else
434    os << "; max-width:100em";
435
436  os << "\">";
437
438  if (max > 1) {
439    os << "<table class=\"msgT\"><tr><td valign=\"top\">";
440    os << "<div class=\"PathIndex";
441    if (Kind) os << " PathIndex" << Kind;
442    os << "\">" << num << "</div>";
443    os << "</td><td>";
444  }
445
446  if (const PathDiagnosticMacroPiece *MP =
447        dyn_cast<PathDiagnosticMacroPiece>(&P)) {
448
449    os << "Within the expansion of the macro '";
450
451    // Get the name of the macro by relexing it.
452    {
453      FullSourceLoc L = MP->getLocation().asLocation().getExpansionLoc();
454      assert(L.isFileID());
455      StringRef BufferInfo = L.getBufferData();
456      const char* MacroName = L.getDecomposedLoc().second + BufferInfo.data();
457      Lexer rawLexer(L, PP.getLangOptions(), BufferInfo.begin(),
458                     MacroName, BufferInfo.end());
459
460      Token TheTok;
461      rawLexer.LexFromRawLexer(TheTok);
462      for (unsigned i = 0, n = TheTok.getLength(); i < n; ++i)
463        os << MacroName[i];
464    }
465
466    os << "':\n";
467
468    if (max > 1)
469      os << "</td></tr></table>";
470
471    // Within a macro piece.  Write out each event.
472    ProcessMacroPiece(os, *MP, 0);
473  }
474  else {
475    os << html::EscapeText(P.getString());
476
477    if (max > 1)
478      os << "</td></tr></table>";
479  }
480
481  os << "</div></td></tr>";
482
483  // Insert the new html.
484  unsigned DisplayPos = LineEnd - FileStart;
485  SourceLocation Loc =
486    SM.getLocForStartOfFile(LPosInfo.first).getLocWithOffset(DisplayPos);
487
488  R.InsertTextBefore(Loc, os.str());
489
490  // Now highlight the ranges.
491  for (const SourceRange *I = P.ranges_begin(), *E = P.ranges_end();
492        I != E; ++I)
493    HighlightRange(R, LPosInfo.first, *I);
494
495#if 0
496  // If there is a code insertion hint, insert that code.
497  // FIXME: This code is disabled because it seems to mangle the HTML
498  // output. I'm leaving it here because it's generally the right idea,
499  // but needs some help from someone more familiar with the rewriter.
500  for (const FixItHint *Hint = P.fixit_begin(), *HintEnd = P.fixit_end();
501       Hint != HintEnd; ++Hint) {
502    if (Hint->RemoveRange.isValid()) {
503      HighlightRange(R, LPosInfo.first, Hint->RemoveRange,
504                     "<span class=\"CodeRemovalHint\">", "</span>");
505    }
506    if (Hint->InsertionLoc.isValid()) {
507      std::string EscapedCode = html::EscapeText(Hint->CodeToInsert, true);
508      EscapedCode = "<span class=\"CodeInsertionHint\">" + EscapedCode
509        + "</span>";
510      R.InsertTextBefore(Hint->InsertionLoc, EscapedCode);
511    }
512  }
513#endif
514}
515
516static void EmitAlphaCounter(raw_ostream &os, unsigned n) {
517  unsigned x = n % ('z' - 'a');
518  n /= 'z' - 'a';
519
520  if (n > 0)
521    EmitAlphaCounter(os, n);
522
523  os << char('a' + x);
524}
525
526unsigned HTMLDiagnostics::ProcessMacroPiece(raw_ostream &os,
527                                            const PathDiagnosticMacroPiece& P,
528                                            unsigned num) {
529
530  for (PathPieces::const_iterator I = P.subPieces.begin(), E=P.subPieces.end();
531        I!=E; ++I) {
532
533    if (const PathDiagnosticMacroPiece *MP =
534          dyn_cast<PathDiagnosticMacroPiece>(*I)) {
535      num = ProcessMacroPiece(os, *MP, num);
536      continue;
537    }
538
539    if (PathDiagnosticEventPiece *EP = dyn_cast<PathDiagnosticEventPiece>(*I)) {
540      os << "<div class=\"msg msgEvent\" style=\"width:94%; "
541            "margin-left:5px\">"
542            "<table class=\"msgT\"><tr>"
543            "<td valign=\"top\"><div class=\"PathIndex PathIndexEvent\">";
544      EmitAlphaCounter(os, num++);
545      os << "</div></td><td valign=\"top\">"
546         << html::EscapeText(EP->getString())
547         << "</td></tr></table></div>\n";
548    }
549  }
550
551  return num;
552}
553
554void HTMLDiagnostics::HighlightRange(Rewriter& R, FileID BugFileID,
555                                     SourceRange Range,
556                                     const char *HighlightStart,
557                                     const char *HighlightEnd) {
558  SourceManager &SM = R.getSourceMgr();
559  const LangOptions &LangOpts = R.getLangOpts();
560
561  SourceLocation InstantiationStart = SM.getExpansionLoc(Range.getBegin());
562  unsigned StartLineNo = SM.getExpansionLineNumber(InstantiationStart);
563
564  SourceLocation InstantiationEnd = SM.getExpansionLoc(Range.getEnd());
565  unsigned EndLineNo = SM.getExpansionLineNumber(InstantiationEnd);
566
567  if (EndLineNo < StartLineNo)
568    return;
569
570  if (SM.getFileID(InstantiationStart) != BugFileID ||
571      SM.getFileID(InstantiationEnd) != BugFileID)
572    return;
573
574  // Compute the column number of the end.
575  unsigned EndColNo = SM.getExpansionColumnNumber(InstantiationEnd);
576  unsigned OldEndColNo = EndColNo;
577
578  if (EndColNo) {
579    // Add in the length of the token, so that we cover multi-char tokens.
580    EndColNo += Lexer::MeasureTokenLength(Range.getEnd(), SM, LangOpts)-1;
581  }
582
583  // Highlight the range.  Make the span tag the outermost tag for the
584  // selected range.
585
586  SourceLocation E =
587    InstantiationEnd.getLocWithOffset(EndColNo - OldEndColNo);
588
589  html::HighlightRange(R, InstantiationStart, E, HighlightStart, HighlightEnd);
590}
591