PrintfFormatString.cpp revision 0ddaeb9b031070ec64afe92d9892875ac44df427 
1//== PrintfFormatString.cpp - Analysis of printf format strings --*- 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// Handling of format string in printf and friends.  The structure of format
11// strings for fprintf() are described in C99 7.19.6.1.
12//
13//===----------------------------------------------------------------------===//
14
15#include "clang/Analysis/Analyses/FormatString.h"
16#include "FormatStringParsing.h"
17
18using clang::analyze_format_string::ArgTypeResult;
19using clang::analyze_format_string::FormatStringHandler;
20using clang::analyze_format_string::LengthModifier;
21using clang::analyze_format_string::OptionalAmount;
22using clang::analyze_format_string::ConversionSpecifier;
23using clang::analyze_printf::PrintfSpecifier;
24
25using namespace clang;
26
27typedef clang::analyze_format_string::SpecifierResult<PrintfSpecifier>
28        PrintfSpecifierResult;
29
30//===----------------------------------------------------------------------===//
31// Methods for parsing format strings.
32//===----------------------------------------------------------------------===//
33
34using analyze_format_string::ParseNonPositionAmount;
35
36static bool ParsePrecision(FormatStringHandler &H, PrintfSpecifier &FS,
37                           const char *Start, const char *&Beg, const char *E,
38                           unsigned *argIndex) {
39  if (argIndex) {
40    FS.setPrecision(ParseNonPositionAmount(Beg, E, *argIndex));
41  } else {
42    const OptionalAmount Amt = ParsePositionAmount(H, Start, Beg, E,
43                                           analyze_format_string::PrecisionPos);
44    if (Amt.isInvalid())
45      return true;
46    FS.setPrecision(Amt);
47  }
48  return false;
49}
50
51static PrintfSpecifierResult ParsePrintfSpecifier(FormatStringHandler &H,
52                                                  const char *&Beg,
53                                                  const char *E,
54                                                  unsigned &argIndex) {
55
56  using namespace clang::analyze_format_string;
57  using namespace clang::analyze_printf;
58
59  const char *I = Beg;
60  const char *Start = 0;
61  UpdateOnReturn <const char*> UpdateBeg(Beg, I);
62
63  // Look for a '%' character that indicates the start of a format specifier.
64  for ( ; I != E ; ++I) {
65    char c = *I;
66    if (c == '\0') {
67      // Detect spurious null characters, which are likely errors.
68      H.HandleNullChar(I);
69      return true;
70    }
71    if (c == '%') {
72      Start = I++;  // Record the start of the format specifier.
73      break;
74    }
75  }
76
77  // No format specifier found?
78  if (!Start)
79    return false;
80
81  if (I == E) {
82    // No more characters left?
83    H.HandleIncompleteSpecifier(Start, E - Start);
84    return true;
85  }
86
87  PrintfSpecifier FS;
88  if (ParseArgPosition(H, FS, Start, I, E))
89    return true;
90
91  if (I == E) {
92    // No more characters left?
93    H.HandleIncompleteSpecifier(Start, E - Start);
94    return true;
95  }
96
97  // Look for flags (if any).
98  bool hasMore = true;
99  for ( ; I != E; ++I) {
100    switch (*I) {
101      default: hasMore = false; break;
102      case '\'':
103        // FIXME: POSIX specific.  Always accept?
104        FS.setHasThousandsGrouping(I);
105        break;
106      case '-': FS.setIsLeftJustified(I); break;
107      case '+': FS.setHasPlusPrefix(I); break;
108      case ' ': FS.setHasSpacePrefix(I); break;
109      case '#': FS.setHasAlternativeForm(I); break;
110      case '0': FS.setHasLeadingZeros(I); break;
111    }
112    if (!hasMore)
113      break;
114  }
115
116  if (I == E) {
117    // No more characters left?
118    H.HandleIncompleteSpecifier(Start, E - Start);
119    return true;
120  }
121
122  // Look for the field width (if any).
123  if (ParseFieldWidth(H, FS, Start, I, E,
124                      FS.usesPositionalArg() ? 0 : &argIndex))
125    return true;
126
127  if (I == E) {
128    // No more characters left?
129    H.HandleIncompleteSpecifier(Start, E - Start);
130    return true;
131  }
132
133  // Look for the precision (if any).
134  if (*I == '.') {
135    ++I;
136    if (I == E) {
137      H.HandleIncompleteSpecifier(Start, E - Start);
138      return true;
139    }
140
141    if (ParsePrecision(H, FS, Start, I, E,
142                       FS.usesPositionalArg() ? 0 : &argIndex))
143      return true;
144
145    if (I == E) {
146      // No more characters left?
147      H.HandleIncompleteSpecifier(Start, E - Start);
148      return true;
149    }
150  }
151
152  // Look for the length modifier.
153  if (ParseLengthModifier(FS, I, E) && I == E) {
154    // No more characters left?
155    H.HandleIncompleteSpecifier(Start, E - Start);
156    return true;
157  }
158
159  if (*I == '\0') {
160    // Detect spurious null characters, which are likely errors.
161    H.HandleNullChar(I);
162    return true;
163  }
164
165  // Finally, look for the conversion specifier.
166  const char *conversionPosition = I++;
167  ConversionSpecifier::Kind k = ConversionSpecifier::InvalidSpecifier;
168  switch (*conversionPosition) {
169    default:
170      break;
171    // C99: 7.19.6.1 (section 8).
172    case '%': k = ConversionSpecifier::PercentArg;   break;
173    case 'A': k = ConversionSpecifier::AArg; break;
174    case 'E': k = ConversionSpecifier::EArg; break;
175    case 'F': k = ConversionSpecifier::FArg; break;
176    case 'G': k = ConversionSpecifier::GArg; break;
177    case 'X': k = ConversionSpecifier::XArg; break;
178    case 'a': k = ConversionSpecifier::aArg; break;
179    case 'c': k = ConversionSpecifier::cArg; break;
180    case 'd': k = ConversionSpecifier::dArg; break;
181    case 'e': k = ConversionSpecifier::eArg; break;
182    case 'f': k = ConversionSpecifier::fArg; break;
183    case 'g': k = ConversionSpecifier::gArg; break;
184    case 'i': k = ConversionSpecifier::iArg; break;
185    case 'n': k = ConversionSpecifier::nArg; break;
186    case 'o': k = ConversionSpecifier::oArg; break;
187    case 'p': k = ConversionSpecifier::pArg;   break;
188    case 's': k = ConversionSpecifier::sArg;      break;
189    case 'u': k = ConversionSpecifier::uArg; break;
190    case 'x': k = ConversionSpecifier::xArg; break;
191    // POSIX specific.
192    case 'C': k = ConversionSpecifier::CArg; break;
193    case 'S': k = ConversionSpecifier::SArg; break;
194    // Objective-C.
195    case '@': k = ConversionSpecifier::ObjCObjArg; break;
196    // Glibc specific.
197    case 'm': k = ConversionSpecifier::PrintErrno; break;
198  }
199  PrintfConversionSpecifier CS(conversionPosition, k);
200  FS.setConversionSpecifier(CS);
201  if (CS.consumesDataArgument() && !FS.usesPositionalArg())
202    FS.setArgIndex(argIndex++);
203
204  if (k == ConversionSpecifier::InvalidSpecifier) {
205    // Assume the conversion takes one argument.
206    return !H.HandleInvalidPrintfConversionSpecifier(FS, Start, I - Start);
207  }
208  return PrintfSpecifierResult(Start, FS);
209}
210
211bool clang::analyze_format_string::ParsePrintfString(FormatStringHandler &H,
212                                                     const char *I,
213                                                     const char *E) {
214
215  unsigned argIndex = 0;
216
217  // Keep looking for a format specifier until we have exhausted the string.
218  while (I != E) {
219    const PrintfSpecifierResult &FSR = ParsePrintfSpecifier(H, I, E, argIndex);
220    // Did a fail-stop error of any kind occur when parsing the specifier?
221    // If so, don't do any more processing.
222    if (FSR.shouldStop())
223      return true;;
224    // Did we exhaust the string or encounter an error that
225    // we can recover from?
226    if (!FSR.hasValue())
227      continue;
228    // We have a format specifier.  Pass it to the callback.
229    if (!H.HandlePrintfSpecifier(FSR.getValue(), FSR.getStart(),
230                                 I - FSR.getStart()))
231      return true;
232  }
233  assert(I == E && "Format string not exhausted");
234  return false;
235}
236
237//===----------------------------------------------------------------------===//
238// Methods on ConversionSpecifier.
239//===----------------------------------------------------------------------===//
240const char *ConversionSpecifier::toString() const {
241  switch (kind) {
242  case dArg: return "d";
243  case iArg: return "i";
244  case oArg: return "o";
245  case uArg: return "u";
246  case xArg: return "x";
247  case XArg: return "X";
248  case fArg: return "f";
249  case FArg: return "F";
250  case eArg: return "e";
251  case EArg: return "E";
252  case gArg: return "g";
253  case GArg: return "G";
254  case aArg: return "a";
255  case AArg: return "A";
256  case cArg: return "c";
257  case sArg: return "s";
258  case pArg: return "p";
259  case nArg: return "n";
260  case PercentArg:  return "%";
261  case ScanListArg: return "[";
262  case InvalidSpecifier: return NULL;
263
264  // MacOS X unicode extensions.
265  case CArg: return "C";
266  case SArg: return "S";
267
268  // Objective-C specific specifiers.
269  case ObjCObjArg: return "@";
270
271  // GlibC specific specifiers.
272  case PrintErrno: return "m";
273  }
274  return NULL;
275}
276
277//===----------------------------------------------------------------------===//
278// Methods on PrintfSpecifier.
279//===----------------------------------------------------------------------===//
280
281ArgTypeResult PrintfSpecifier::getArgType(ASTContext &Ctx) const {
282  const PrintfConversionSpecifier &CS = getConversionSpecifier();
283
284  if (!CS.consumesDataArgument())
285    return ArgTypeResult::Invalid();
286
287  if (CS.getKind() == ConversionSpecifier::cArg)
288    switch (LM.getKind()) {
289      case LengthModifier::None: return Ctx.IntTy;
290      case LengthModifier::AsLong: return ArgTypeResult::WIntTy;
291      default:
292        return ArgTypeResult::Invalid();
293    }
294
295  if (CS.isIntArg())
296    switch (LM.getKind()) {
297      case LengthModifier::AsLongDouble:
298        return ArgTypeResult::Invalid();
299      case LengthModifier::None: return Ctx.IntTy;
300      case LengthModifier::AsChar: return Ctx.SignedCharTy;
301      case LengthModifier::AsShort: return Ctx.ShortTy;
302      case LengthModifier::AsLong: return Ctx.LongTy;
303      case LengthModifier::AsLongLong: return Ctx.LongLongTy;
304      case LengthModifier::AsIntMax:
305        // FIXME: Return unknown for now.
306        return ArgTypeResult();
307      case LengthModifier::AsSizeT: return Ctx.getSizeType();
308      case LengthModifier::AsPtrDiff: return Ctx.getPointerDiffType();
309    }
310
311  if (CS.isUIntArg())
312    switch (LM.getKind()) {
313      case LengthModifier::AsLongDouble:
314        return ArgTypeResult::Invalid();
315      case LengthModifier::None: return Ctx.UnsignedIntTy;
316      case LengthModifier::AsChar: return Ctx.UnsignedCharTy;
317      case LengthModifier::AsShort: return Ctx.UnsignedShortTy;
318      case LengthModifier::AsLong: return Ctx.UnsignedLongTy;
319      case LengthModifier::AsLongLong: return Ctx.UnsignedLongLongTy;
320      case LengthModifier::AsIntMax:
321        // FIXME: Return unknown for now.
322        return ArgTypeResult();
323      case LengthModifier::AsSizeT:
324        // FIXME: How to get the corresponding unsigned
325        // version of size_t?
326        return ArgTypeResult();
327      case LengthModifier::AsPtrDiff:
328        // FIXME: How to get the corresponding unsigned
329        // version of ptrdiff_t?
330        return ArgTypeResult();
331    }
332
333  if (CS.isDoubleArg()) {
334    if (LM.getKind() == LengthModifier::AsLongDouble)
335      return Ctx.LongDoubleTy;
336    return Ctx.DoubleTy;
337  }
338
339  switch (CS.getKind()) {
340    case ConversionSpecifier::sArg:
341      return ArgTypeResult(LM.getKind() == LengthModifier::AsWideChar ?
342          ArgTypeResult::WCStrTy : ArgTypeResult::CStrTy);
343    case ConversionSpecifier::SArg:
344      // FIXME: This appears to be Mac OS X specific.
345      return ArgTypeResult::WCStrTy;
346    case ConversionSpecifier::CArg:
347      return Ctx.WCharTy;
348    case ConversionSpecifier::pArg:
349      return ArgTypeResult::CPointerTy;
350    default:
351      break;
352  }
353
354  // FIXME: Handle other cases.
355  return ArgTypeResult();
356}
357
358bool PrintfSpecifier::fixType(QualType QT) {
359  // Handle strings first (char *, wchar_t *)
360  if (QT->isPointerType() && (QT->getPointeeType()->isAnyCharacterType())) {
361    CS.setKind(ConversionSpecifier::sArg);
362
363    // Disable irrelevant flags
364    HasAlternativeForm = 0;
365    HasLeadingZeroes = 0;
366
367    // Set the long length modifier for wide characters
368    if (QT->getPointeeType()->isWideCharType())
369      LM.setKind(LengthModifier::AsWideChar);
370
371    return true;
372  }
373
374  // We can only work with builtin types.
375  if (!QT->isBuiltinType())
376    return false;
377
378  // Everything else should be a base type
379  const BuiltinType *BT = QT->getAs<BuiltinType>();
380
381  // Set length modifier
382  switch (BT->getKind()) {
383  case BuiltinType::Bool:
384  case BuiltinType::WChar_U:
385  case BuiltinType::WChar_S:
386  case BuiltinType::Char16:
387  case BuiltinType::Char32:
388  case BuiltinType::UInt128:
389  case BuiltinType::Int128:
390  case BuiltinType::Half:
391    // Integral types which are non-trivial to correct.
392    return false;
393
394  case BuiltinType::Void:
395  case BuiltinType::NullPtr:
396  case BuiltinType::ObjCId:
397  case BuiltinType::ObjCClass:
398  case BuiltinType::ObjCSel:
399  case BuiltinType::Dependent:
400  case BuiltinType::Overload:
401  case BuiltinType::BoundMember:
402  case BuiltinType::UnknownAny:
403  case BuiltinType::ARCUnbridgedCast:
404    // Misc other stuff which doesn't make sense here.
405    return false;
406
407  case BuiltinType::UInt:
408  case BuiltinType::Int:
409  case BuiltinType::Float:
410  case BuiltinType::Double:
411    LM.setKind(LengthModifier::None);
412    break;
413
414  case BuiltinType::Char_U:
415  case BuiltinType::UChar:
416  case BuiltinType::Char_S:
417  case BuiltinType::SChar:
418    LM.setKind(LengthModifier::AsChar);
419    break;
420
421  case BuiltinType::Short:
422  case BuiltinType::UShort:
423    LM.setKind(LengthModifier::AsShort);
424    break;
425
426  case BuiltinType::Long:
427  case BuiltinType::ULong:
428    LM.setKind(LengthModifier::AsLong);
429    break;
430
431  case BuiltinType::LongLong:
432  case BuiltinType::ULongLong:
433    LM.setKind(LengthModifier::AsLongLong);
434    break;
435
436  case BuiltinType::LongDouble:
437    LM.setKind(LengthModifier::AsLongDouble);
438    break;
439  }
440
441  // Set conversion specifier and disable any flags which do not apply to it.
442  // Let typedefs to char fall through to int, as %c is silly for uint8_t.
443  if (isa<TypedefType>(QT) && QT->isAnyCharacterType()) {
444    CS.setKind(ConversionSpecifier::cArg);
445    LM.setKind(LengthModifier::None);
446    Precision.setHowSpecified(OptionalAmount::NotSpecified);
447    HasAlternativeForm = 0;
448    HasLeadingZeroes = 0;
449    HasPlusPrefix = 0;
450  }
451  // Test for Floating type first as LongDouble can pass isUnsignedIntegerType
452  else if (QT->isRealFloatingType()) {
453    CS.setKind(ConversionSpecifier::fArg);
454  }
455  else if (QT->isSignedIntegerType()) {
456    CS.setKind(ConversionSpecifier::dArg);
457    HasAlternativeForm = 0;
458  }
459  else if (QT->isUnsignedIntegerType()) {
460    // Preserve the original formatting, e.g. 'X', 'o'.
461    if (!cast<PrintfConversionSpecifier>(CS).isUIntArg())
462      CS.setKind(ConversionSpecifier::uArg);
463    HasAlternativeForm = 0;
464    HasPlusPrefix = 0;
465  } else {
466    llvm_unreachable("Unexpected type");
467  }
468
469  return true;
470}
471
472void PrintfSpecifier::toString(raw_ostream &os) const {
473  // Whilst some features have no defined order, we are using the order
474  // appearing in the C99 standard (ISO/IEC 9899:1999 (E) 7.19.6.1)
475  os << "%";
476
477  // Positional args
478  if (usesPositionalArg()) {
479    os << getPositionalArgIndex() << "$";
480  }
481
482  // Conversion flags
483  if (IsLeftJustified)    os << "-";
484  if (HasPlusPrefix)      os << "+";
485  if (HasSpacePrefix)     os << " ";
486  if (HasAlternativeForm) os << "#";
487  if (HasLeadingZeroes)   os << "0";
488
489  // Minimum field width
490  FieldWidth.toString(os);
491  // Precision
492  Precision.toString(os);
493  // Length modifier
494  os << LM.toString();
495  // Conversion specifier
496  os << CS.toString();
497}
498
499bool PrintfSpecifier::hasValidPlusPrefix() const {
500  if (!HasPlusPrefix)
501    return true;
502
503  // The plus prefix only makes sense for signed conversions
504  switch (CS.getKind()) {
505  case ConversionSpecifier::dArg:
506  case ConversionSpecifier::iArg:
507  case ConversionSpecifier::fArg:
508  case ConversionSpecifier::FArg:
509  case ConversionSpecifier::eArg:
510  case ConversionSpecifier::EArg:
511  case ConversionSpecifier::gArg:
512  case ConversionSpecifier::GArg:
513  case ConversionSpecifier::aArg:
514  case ConversionSpecifier::AArg:
515    return true;
516
517  default:
518    return false;
519  }
520}
521
522bool PrintfSpecifier::hasValidAlternativeForm() const {
523  if (!HasAlternativeForm)
524    return true;
525
526  // Alternate form flag only valid with the oxXaAeEfFgG conversions
527  switch (CS.getKind()) {
528  case ConversionSpecifier::oArg:
529  case ConversionSpecifier::xArg:
530  case ConversionSpecifier::XArg:
531  case ConversionSpecifier::aArg:
532  case ConversionSpecifier::AArg:
533  case ConversionSpecifier::eArg:
534  case ConversionSpecifier::EArg:
535  case ConversionSpecifier::fArg:
536  case ConversionSpecifier::FArg:
537  case ConversionSpecifier::gArg:
538  case ConversionSpecifier::GArg:
539    return true;
540
541  default:
542    return false;
543  }
544}
545
546bool PrintfSpecifier::hasValidLeadingZeros() const {
547  if (!HasLeadingZeroes)
548    return true;
549
550  // Leading zeroes flag only valid with the diouxXaAeEfFgG conversions
551  switch (CS.getKind()) {
552  case ConversionSpecifier::dArg:
553  case ConversionSpecifier::iArg:
554  case ConversionSpecifier::oArg:
555  case ConversionSpecifier::uArg:
556  case ConversionSpecifier::xArg:
557  case ConversionSpecifier::XArg:
558  case ConversionSpecifier::aArg:
559  case ConversionSpecifier::AArg:
560  case ConversionSpecifier::eArg:
561  case ConversionSpecifier::EArg:
562  case ConversionSpecifier::fArg:
563  case ConversionSpecifier::FArg:
564  case ConversionSpecifier::gArg:
565  case ConversionSpecifier::GArg:
566    return true;
567
568  default:
569    return false;
570  }
571}
572
573bool PrintfSpecifier::hasValidSpacePrefix() const {
574  if (!HasSpacePrefix)
575    return true;
576
577  // The space prefix only makes sense for signed conversions
578  switch (CS.getKind()) {
579  case ConversionSpecifier::dArg:
580  case ConversionSpecifier::iArg:
581  case ConversionSpecifier::fArg:
582  case ConversionSpecifier::FArg:
583  case ConversionSpecifier::eArg:
584  case ConversionSpecifier::EArg:
585  case ConversionSpecifier::gArg:
586  case ConversionSpecifier::GArg:
587  case ConversionSpecifier::aArg:
588  case ConversionSpecifier::AArg:
589    return true;
590
591  default:
592    return false;
593  }
594}
595
596bool PrintfSpecifier::hasValidLeftJustified() const {
597  if (!IsLeftJustified)
598    return true;
599
600  // The left justified flag is valid for all conversions except n
601  switch (CS.getKind()) {
602  case ConversionSpecifier::nArg:
603    return false;
604
605  default:
606    return true;
607  }
608}
609
610bool PrintfSpecifier::hasValidThousandsGroupingPrefix() const {
611  if (!HasThousandsGrouping)
612    return true;
613
614  switch (CS.getKind()) {
615    case ConversionSpecifier::dArg:
616    case ConversionSpecifier::iArg:
617    case ConversionSpecifier::uArg:
618    case ConversionSpecifier::fArg:
619    case ConversionSpecifier::FArg:
620    case ConversionSpecifier::gArg:
621    case ConversionSpecifier::GArg:
622      return true;
623    default:
624      return false;
625  }
626}
627
628bool PrintfSpecifier::hasValidPrecision() const {
629  if (Precision.getHowSpecified() == OptionalAmount::NotSpecified)
630    return true;
631
632  // Precision is only valid with the diouxXaAeEfFgGs conversions
633  switch (CS.getKind()) {
634  case ConversionSpecifier::dArg:
635  case ConversionSpecifier::iArg:
636  case ConversionSpecifier::oArg:
637  case ConversionSpecifier::uArg:
638  case ConversionSpecifier::xArg:
639  case ConversionSpecifier::XArg:
640  case ConversionSpecifier::aArg:
641  case ConversionSpecifier::AArg:
642  case ConversionSpecifier::eArg:
643  case ConversionSpecifier::EArg:
644  case ConversionSpecifier::fArg:
645  case ConversionSpecifier::FArg:
646  case ConversionSpecifier::gArg:
647  case ConversionSpecifier::GArg:
648  case ConversionSpecifier::sArg:
649    return true;
650
651  default:
652    return false;
653  }
654}
655bool PrintfSpecifier::hasValidFieldWidth() const {
656  if (FieldWidth.getHowSpecified() == OptionalAmount::NotSpecified)
657      return true;
658
659  // The field width is valid for all conversions except n
660  switch (CS.getKind()) {
661  case ConversionSpecifier::nArg:
662    return false;
663
664  default:
665    return true;
666  }
667}
668