PrintfFormatString.cpp revision 58e1e54476d610d6c33ef483f216ed8a1282d35c 
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::ArgType;
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                                                  const LangOptions &LO) {
56
57  using namespace clang::analyze_format_string;
58  using namespace clang::analyze_printf;
59
60  const char *I = Beg;
61  const char *Start = 0;
62  UpdateOnReturn <const char*> UpdateBeg(Beg, I);
63
64  // Look for a '%' character that indicates the start of a format specifier.
65  for ( ; I != E ; ++I) {
66    char c = *I;
67    if (c == '\0') {
68      // Detect spurious null characters, which are likely errors.
69      H.HandleNullChar(I);
70      return true;
71    }
72    if (c == '%') {
73      Start = I++;  // Record the start of the format specifier.
74      break;
75    }
76  }
77
78  // No format specifier found?
79  if (!Start)
80    return false;
81
82  if (I == E) {
83    // No more characters left?
84    H.HandleIncompleteSpecifier(Start, E - Start);
85    return true;
86  }
87
88  PrintfSpecifier FS;
89  if (ParseArgPosition(H, FS, Start, I, E))
90    return true;
91
92  if (I == E) {
93    // No more characters left?
94    H.HandleIncompleteSpecifier(Start, E - Start);
95    return true;
96  }
97
98  // Look for flags (if any).
99  bool hasMore = true;
100  for ( ; I != E; ++I) {
101    switch (*I) {
102      default: hasMore = false; break;
103      case '\'':
104        // FIXME: POSIX specific.  Always accept?
105        FS.setHasThousandsGrouping(I);
106        break;
107      case '-': FS.setIsLeftJustified(I); break;
108      case '+': FS.setHasPlusPrefix(I); break;
109      case ' ': FS.setHasSpacePrefix(I); break;
110      case '#': FS.setHasAlternativeForm(I); break;
111      case '0': FS.setHasLeadingZeros(I); break;
112    }
113    if (!hasMore)
114      break;
115  }
116
117  if (I == E) {
118    // No more characters left?
119    H.HandleIncompleteSpecifier(Start, E - Start);
120    return true;
121  }
122
123  // Look for the field width (if any).
124  if (ParseFieldWidth(H, FS, Start, I, E,
125                      FS.usesPositionalArg() ? 0 : &argIndex))
126    return true;
127
128  if (I == E) {
129    // No more characters left?
130    H.HandleIncompleteSpecifier(Start, E - Start);
131    return true;
132  }
133
134  // Look for the precision (if any).
135  if (*I == '.') {
136    ++I;
137    if (I == E) {
138      H.HandleIncompleteSpecifier(Start, E - Start);
139      return true;
140    }
141
142    if (ParsePrecision(H, FS, Start, I, E,
143                       FS.usesPositionalArg() ? 0 : &argIndex))
144      return true;
145
146    if (I == E) {
147      // No more characters left?
148      H.HandleIncompleteSpecifier(Start, E - Start);
149      return true;
150    }
151  }
152
153  // Look for the length modifier.
154  if (ParseLengthModifier(FS, I, E, LO) && I == E) {
155    // No more characters left?
156    H.HandleIncompleteSpecifier(Start, E - Start);
157    return true;
158  }
159
160  if (*I == '\0') {
161    // Detect spurious null characters, which are likely errors.
162    H.HandleNullChar(I);
163    return true;
164  }
165
166  // Finally, look for the conversion specifier.
167  const char *conversionPosition = I++;
168  ConversionSpecifier::Kind k = ConversionSpecifier::InvalidSpecifier;
169  switch (*conversionPosition) {
170    default:
171      break;
172    // C99: 7.19.6.1 (section 8).
173    case '%': k = ConversionSpecifier::PercentArg;   break;
174    case 'A': k = ConversionSpecifier::AArg; break;
175    case 'E': k = ConversionSpecifier::EArg; break;
176    case 'F': k = ConversionSpecifier::FArg; break;
177    case 'G': k = ConversionSpecifier::GArg; break;
178    case 'X': k = ConversionSpecifier::XArg; break;
179    case 'a': k = ConversionSpecifier::aArg; break;
180    case 'c': k = ConversionSpecifier::cArg; break;
181    case 'd': k = ConversionSpecifier::dArg; break;
182    case 'e': k = ConversionSpecifier::eArg; break;
183    case 'f': k = ConversionSpecifier::fArg; break;
184    case 'g': k = ConversionSpecifier::gArg; break;
185    case 'i': k = ConversionSpecifier::iArg; break;
186    case 'n': k = ConversionSpecifier::nArg; break;
187    case 'o': k = ConversionSpecifier::oArg; break;
188    case 'p': k = ConversionSpecifier::pArg;   break;
189    case 's': k = ConversionSpecifier::sArg;      break;
190    case 'u': k = ConversionSpecifier::uArg; break;
191    case 'x': k = ConversionSpecifier::xArg; break;
192    // POSIX specific.
193    case 'C': k = ConversionSpecifier::CArg; break;
194    case 'S': k = ConversionSpecifier::SArg; break;
195    // Objective-C.
196    case '@': k = ConversionSpecifier::ObjCObjArg; break;
197    // Glibc specific.
198    case 'm': k = ConversionSpecifier::PrintErrno; break;
199  }
200  PrintfConversionSpecifier CS(conversionPosition, k);
201  FS.setConversionSpecifier(CS);
202  if (CS.consumesDataArgument() && !FS.usesPositionalArg())
203    FS.setArgIndex(argIndex++);
204
205  if (k == ConversionSpecifier::InvalidSpecifier) {
206    // Assume the conversion takes one argument.
207    return !H.HandleInvalidPrintfConversionSpecifier(FS, Start, I - Start);
208  }
209  return PrintfSpecifierResult(Start, FS);
210}
211
212bool clang::analyze_format_string::ParsePrintfString(FormatStringHandler &H,
213                                                     const char *I,
214                                                     const char *E,
215                                                     const LangOptions &LO) {
216
217  unsigned argIndex = 0;
218
219  // Keep looking for a format specifier until we have exhausted the string.
220  while (I != E) {
221    const PrintfSpecifierResult &FSR = ParsePrintfSpecifier(H, I, E, argIndex,
222                                                            LO);
223    // Did a fail-stop error of any kind occur when parsing the specifier?
224    // If so, don't do any more processing.
225    if (FSR.shouldStop())
226      return true;;
227    // Did we exhaust the string or encounter an error that
228    // we can recover from?
229    if (!FSR.hasValue())
230      continue;
231    // We have a format specifier.  Pass it to the callback.
232    if (!H.HandlePrintfSpecifier(FSR.getValue(), FSR.getStart(),
233                                 I - FSR.getStart()))
234      return true;
235  }
236  assert(I == E && "Format string not exhausted");
237  return false;
238}
239
240//===----------------------------------------------------------------------===//
241// Methods on PrintfSpecifier.
242//===----------------------------------------------------------------------===//
243
244ArgType PrintfSpecifier::getArgType(ASTContext &Ctx,
245                                    bool IsObjCLiteral) const {
246  const PrintfConversionSpecifier &CS = getConversionSpecifier();
247
248  if (!CS.consumesDataArgument())
249    return ArgType::Invalid();
250
251  if (CS.getKind() == ConversionSpecifier::cArg)
252    switch (LM.getKind()) {
253      case LengthModifier::None: return Ctx.IntTy;
254      case LengthModifier::AsLong:
255        return ArgType(ArgType::WIntTy, "wint_t");
256      default:
257        return ArgType::Invalid();
258    }
259
260  if (CS.isIntArg())
261    switch (LM.getKind()) {
262      case LengthModifier::AsLongDouble:
263        // GNU extension.
264        return Ctx.LongLongTy;
265      case LengthModifier::None: return Ctx.IntTy;
266      case LengthModifier::AsChar: return ArgType::AnyCharTy;
267      case LengthModifier::AsShort: return Ctx.ShortTy;
268      case LengthModifier::AsLong: return Ctx.LongTy;
269      case LengthModifier::AsLongLong:
270      case LengthModifier::AsQuad:
271        return Ctx.LongLongTy;
272      case LengthModifier::AsIntMax:
273        return ArgType(Ctx.getIntMaxType(), "intmax_t");
274      case LengthModifier::AsSizeT:
275        // FIXME: How to get the corresponding signed version of size_t?
276        return ArgType();
277      case LengthModifier::AsPtrDiff:
278        return ArgType(Ctx.getPointerDiffType(), "ptrdiff_t");
279      case LengthModifier::AsAllocate:
280      case LengthModifier::AsMAllocate:
281        return ArgType::Invalid();
282    }
283
284  if (CS.isUIntArg())
285    switch (LM.getKind()) {
286      case LengthModifier::AsLongDouble:
287        // GNU extension.
288        return Ctx.UnsignedLongLongTy;
289      case LengthModifier::None: return Ctx.UnsignedIntTy;
290      case LengthModifier::AsChar: return Ctx.UnsignedCharTy;
291      case LengthModifier::AsShort: return Ctx.UnsignedShortTy;
292      case LengthModifier::AsLong: return Ctx.UnsignedLongTy;
293      case LengthModifier::AsLongLong:
294      case LengthModifier::AsQuad:
295        return Ctx.UnsignedLongLongTy;
296      case LengthModifier::AsIntMax:
297        return ArgType(Ctx.getUIntMaxType(), "uintmax_t");
298      case LengthModifier::AsSizeT:
299        return ArgType(Ctx.getSizeType(), "size_t");
300      case LengthModifier::AsPtrDiff:
301        // FIXME: How to get the corresponding unsigned
302        // version of ptrdiff_t?
303        return ArgType();
304      case LengthModifier::AsAllocate:
305      case LengthModifier::AsMAllocate:
306        return ArgType::Invalid();
307    }
308
309  if (CS.isDoubleArg()) {
310    if (LM.getKind() == LengthModifier::AsLongDouble)
311      return Ctx.LongDoubleTy;
312    return Ctx.DoubleTy;
313  }
314
315  switch (CS.getKind()) {
316    case ConversionSpecifier::sArg:
317      if (LM.getKind() == LengthModifier::AsWideChar) {
318        if (IsObjCLiteral)
319          return Ctx.getPointerType(Ctx.UnsignedShortTy.withConst());
320        return ArgType(ArgType::WCStrTy, "wchar_t *");
321      }
322      return ArgType::CStrTy;
323    case ConversionSpecifier::SArg:
324      if (IsObjCLiteral)
325        return Ctx.getPointerType(Ctx.UnsignedShortTy.withConst());
326      return ArgType(ArgType::WCStrTy, "wchar_t *");
327    case ConversionSpecifier::CArg:
328      if (IsObjCLiteral)
329        return Ctx.UnsignedShortTy;
330      return ArgType(Ctx.WCharTy, "wchar_t");
331    case ConversionSpecifier::pArg:
332      return ArgType::CPointerTy;
333    case ConversionSpecifier::nArg:
334      return ArgType::PtrTo(Ctx.IntTy);
335    case ConversionSpecifier::ObjCObjArg:
336      return ArgType::ObjCPointerTy;
337    default:
338      break;
339  }
340
341  // FIXME: Handle other cases.
342  return ArgType();
343}
344
345bool PrintfSpecifier::fixType(QualType QT, const LangOptions &LangOpt,
346                              ASTContext &Ctx, bool IsObjCLiteral) {
347  // %n is different from other conversion specifiers; don't try to fix it.
348  if (CS.getKind() == ConversionSpecifier::nArg)
349    return false;
350
351  // Handle Objective-C objects first. Note that while the '%@' specifier will
352  // not warn for structure pointer or void pointer arguments (because that's
353  // how CoreFoundation objects are implemented), we only show a fixit for '%@'
354  // if we know it's an object (block, id, class, or __attribute__((NSObject))).
355  if (QT->isObjCRetainableType()) {
356    if (!IsObjCLiteral)
357      return false;
358
359    CS.setKind(ConversionSpecifier::ObjCObjArg);
360
361    // Disable irrelevant flags
362    HasThousandsGrouping = false;
363    HasPlusPrefix = false;
364    HasSpacePrefix = false;
365    HasAlternativeForm = false;
366    HasLeadingZeroes = false;
367    Precision.setHowSpecified(OptionalAmount::NotSpecified);
368    LM.setKind(LengthModifier::None);
369
370    return true;
371  }
372
373  // Handle strings next (char *, wchar_t *)
374  if (QT->isPointerType() && (QT->getPointeeType()->isAnyCharacterType())) {
375    CS.setKind(ConversionSpecifier::sArg);
376
377    // Disable irrelevant flags
378    HasAlternativeForm = 0;
379    HasLeadingZeroes = 0;
380
381    // Set the long length modifier for wide characters
382    if (QT->getPointeeType()->isWideCharType())
383      LM.setKind(LengthModifier::AsWideChar);
384    else
385      LM.setKind(LengthModifier::None);
386
387    return true;
388  }
389
390  // If it's an enum, get its underlying type.
391  if (const EnumType *ETy = QT->getAs<EnumType>())
392    QT = ETy->getDecl()->getIntegerType();
393
394  // We can only work with builtin types.
395  const BuiltinType *BT = QT->getAs<BuiltinType>();
396  if (!BT)
397    return false;
398
399  // Set length modifier
400  switch (BT->getKind()) {
401  case BuiltinType::Bool:
402  case BuiltinType::WChar_U:
403  case BuiltinType::WChar_S:
404  case BuiltinType::Char16:
405  case BuiltinType::Char32:
406  case BuiltinType::UInt128:
407  case BuiltinType::Int128:
408  case BuiltinType::Half:
409    // Various types which are non-trivial to correct.
410    return false;
411
412#define SIGNED_TYPE(Id, SingletonId)
413#define UNSIGNED_TYPE(Id, SingletonId)
414#define FLOATING_TYPE(Id, SingletonId)
415#define BUILTIN_TYPE(Id, SingletonId) \
416  case BuiltinType::Id:
417#include "clang/AST/BuiltinTypes.def"
418    // Misc other stuff which doesn't make sense here.
419    return false;
420
421  case BuiltinType::UInt:
422  case BuiltinType::Int:
423  case BuiltinType::Float:
424  case BuiltinType::Double:
425    LM.setKind(LengthModifier::None);
426    break;
427
428  case BuiltinType::Char_U:
429  case BuiltinType::UChar:
430  case BuiltinType::Char_S:
431  case BuiltinType::SChar:
432    LM.setKind(LengthModifier::AsChar);
433    break;
434
435  case BuiltinType::Short:
436  case BuiltinType::UShort:
437    LM.setKind(LengthModifier::AsShort);
438    break;
439
440  case BuiltinType::Long:
441  case BuiltinType::ULong:
442    LM.setKind(LengthModifier::AsLong);
443    break;
444
445  case BuiltinType::LongLong:
446  case BuiltinType::ULongLong:
447    LM.setKind(LengthModifier::AsLongLong);
448    break;
449
450  case BuiltinType::LongDouble:
451    LM.setKind(LengthModifier::AsLongDouble);
452    break;
453  }
454
455  // Handle size_t, ptrdiff_t, etc. that have dedicated length modifiers in C99.
456  if (isa<TypedefType>(QT) && (LangOpt.C99 || LangOpt.CPlusPlus0x))
457    namedTypeToLengthModifier(QT, LM);
458
459  // If fixing the length modifier was enough, we are done.
460  const analyze_printf::ArgType &ATR = getArgType(Ctx, IsObjCLiteral);
461  if (hasValidLengthModifier() && ATR.isValid() && ATR.matchesType(Ctx, QT))
462    return true;
463
464  // Set conversion specifier and disable any flags which do not apply to it.
465  // Let typedefs to char fall through to int, as %c is silly for uint8_t.
466  if (isa<TypedefType>(QT) && QT->isAnyCharacterType()) {
467    CS.setKind(ConversionSpecifier::cArg);
468    LM.setKind(LengthModifier::None);
469    Precision.setHowSpecified(OptionalAmount::NotSpecified);
470    HasAlternativeForm = 0;
471    HasLeadingZeroes = 0;
472    HasPlusPrefix = 0;
473  }
474  // Test for Floating type first as LongDouble can pass isUnsignedIntegerType
475  else if (QT->isRealFloatingType()) {
476    CS.setKind(ConversionSpecifier::fArg);
477  }
478  else if (QT->isSignedIntegerType()) {
479    CS.setKind(ConversionSpecifier::dArg);
480    HasAlternativeForm = 0;
481  }
482  else if (QT->isUnsignedIntegerType()) {
483    CS.setKind(ConversionSpecifier::uArg);
484    HasAlternativeForm = 0;
485    HasPlusPrefix = 0;
486  } else {
487    llvm_unreachable("Unexpected type");
488  }
489
490  return true;
491}
492
493void PrintfSpecifier::toString(raw_ostream &os) const {
494  // Whilst some features have no defined order, we are using the order
495  // appearing in the C99 standard (ISO/IEC 9899:1999 (E) 7.19.6.1)
496  os << "%";
497
498  // Positional args
499  if (usesPositionalArg()) {
500    os << getPositionalArgIndex() << "$";
501  }
502
503  // Conversion flags
504  if (IsLeftJustified)    os << "-";
505  if (HasPlusPrefix)      os << "+";
506  if (HasSpacePrefix)     os << " ";
507  if (HasAlternativeForm) os << "#";
508  if (HasLeadingZeroes)   os << "0";
509
510  // Minimum field width
511  FieldWidth.toString(os);
512  // Precision
513  Precision.toString(os);
514  // Length modifier
515  os << LM.toString();
516  // Conversion specifier
517  os << CS.toString();
518}
519
520bool PrintfSpecifier::hasValidPlusPrefix() const {
521  if (!HasPlusPrefix)
522    return true;
523
524  // The plus prefix only makes sense for signed conversions
525  switch (CS.getKind()) {
526  case ConversionSpecifier::dArg:
527  case ConversionSpecifier::iArg:
528  case ConversionSpecifier::fArg:
529  case ConversionSpecifier::FArg:
530  case ConversionSpecifier::eArg:
531  case ConversionSpecifier::EArg:
532  case ConversionSpecifier::gArg:
533  case ConversionSpecifier::GArg:
534  case ConversionSpecifier::aArg:
535  case ConversionSpecifier::AArg:
536    return true;
537
538  default:
539    return false;
540  }
541}
542
543bool PrintfSpecifier::hasValidAlternativeForm() const {
544  if (!HasAlternativeForm)
545    return true;
546
547  // Alternate form flag only valid with the oxXaAeEfFgG conversions
548  switch (CS.getKind()) {
549  case ConversionSpecifier::oArg:
550  case ConversionSpecifier::xArg:
551  case ConversionSpecifier::XArg:
552  case ConversionSpecifier::aArg:
553  case ConversionSpecifier::AArg:
554  case ConversionSpecifier::eArg:
555  case ConversionSpecifier::EArg:
556  case ConversionSpecifier::fArg:
557  case ConversionSpecifier::FArg:
558  case ConversionSpecifier::gArg:
559  case ConversionSpecifier::GArg:
560    return true;
561
562  default:
563    return false;
564  }
565}
566
567bool PrintfSpecifier::hasValidLeadingZeros() const {
568  if (!HasLeadingZeroes)
569    return true;
570
571  // Leading zeroes flag only valid with the diouxXaAeEfFgG conversions
572  switch (CS.getKind()) {
573  case ConversionSpecifier::dArg:
574  case ConversionSpecifier::iArg:
575  case ConversionSpecifier::oArg:
576  case ConversionSpecifier::uArg:
577  case ConversionSpecifier::xArg:
578  case ConversionSpecifier::XArg:
579  case ConversionSpecifier::aArg:
580  case ConversionSpecifier::AArg:
581  case ConversionSpecifier::eArg:
582  case ConversionSpecifier::EArg:
583  case ConversionSpecifier::fArg:
584  case ConversionSpecifier::FArg:
585  case ConversionSpecifier::gArg:
586  case ConversionSpecifier::GArg:
587    return true;
588
589  default:
590    return false;
591  }
592}
593
594bool PrintfSpecifier::hasValidSpacePrefix() const {
595  if (!HasSpacePrefix)
596    return true;
597
598  // The space prefix only makes sense for signed conversions
599  switch (CS.getKind()) {
600  case ConversionSpecifier::dArg:
601  case ConversionSpecifier::iArg:
602  case ConversionSpecifier::fArg:
603  case ConversionSpecifier::FArg:
604  case ConversionSpecifier::eArg:
605  case ConversionSpecifier::EArg:
606  case ConversionSpecifier::gArg:
607  case ConversionSpecifier::GArg:
608  case ConversionSpecifier::aArg:
609  case ConversionSpecifier::AArg:
610    return true;
611
612  default:
613    return false;
614  }
615}
616
617bool PrintfSpecifier::hasValidLeftJustified() const {
618  if (!IsLeftJustified)
619    return true;
620
621  // The left justified flag is valid for all conversions except n
622  switch (CS.getKind()) {
623  case ConversionSpecifier::nArg:
624    return false;
625
626  default:
627    return true;
628  }
629}
630
631bool PrintfSpecifier::hasValidThousandsGroupingPrefix() const {
632  if (!HasThousandsGrouping)
633    return true;
634
635  switch (CS.getKind()) {
636    case ConversionSpecifier::dArg:
637    case ConversionSpecifier::iArg:
638    case ConversionSpecifier::uArg:
639    case ConversionSpecifier::fArg:
640    case ConversionSpecifier::FArg:
641    case ConversionSpecifier::gArg:
642    case ConversionSpecifier::GArg:
643      return true;
644    default:
645      return false;
646  }
647}
648
649bool PrintfSpecifier::hasValidPrecision() const {
650  if (Precision.getHowSpecified() == OptionalAmount::NotSpecified)
651    return true;
652
653  // Precision is only valid with the diouxXaAeEfFgGs conversions
654  switch (CS.getKind()) {
655  case ConversionSpecifier::dArg:
656  case ConversionSpecifier::iArg:
657  case ConversionSpecifier::oArg:
658  case ConversionSpecifier::uArg:
659  case ConversionSpecifier::xArg:
660  case ConversionSpecifier::XArg:
661  case ConversionSpecifier::aArg:
662  case ConversionSpecifier::AArg:
663  case ConversionSpecifier::eArg:
664  case ConversionSpecifier::EArg:
665  case ConversionSpecifier::fArg:
666  case ConversionSpecifier::FArg:
667  case ConversionSpecifier::gArg:
668  case ConversionSpecifier::GArg:
669  case ConversionSpecifier::sArg:
670    return true;
671
672  default:
673    return false;
674  }
675}
676bool PrintfSpecifier::hasValidFieldWidth() const {
677  if (FieldWidth.getHowSpecified() == OptionalAmount::NotSpecified)
678      return true;
679
680  // The field width is valid for all conversions except n
681  switch (CS.getKind()) {
682  case ConversionSpecifier::nArg:
683    return false;
684
685  default:
686    return true;
687  }
688}
689