PrintfFormatString.cpp revision 630821869c4ec4604ab479d66e5ff81147a858e1 
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 PrintfSpecifier.
239//===----------------------------------------------------------------------===//
240
241ArgTypeResult PrintfSpecifier::getArgType(ASTContext &Ctx) const {
242  const PrintfConversionSpecifier &CS = getConversionSpecifier();
243
244  if (!CS.consumesDataArgument())
245    return ArgTypeResult::Invalid();
246
247  if (CS.getKind() == ConversionSpecifier::cArg)
248    switch (LM.getKind()) {
249      case LengthModifier::None: return Ctx.IntTy;
250      case LengthModifier::AsLong:
251        return ArgTypeResult(ArgTypeResult::WIntTy, "wint_t");
252      default:
253        return ArgTypeResult::Invalid();
254    }
255
256  if (CS.isIntArg())
257    switch (LM.getKind()) {
258      case LengthModifier::AsLongDouble:
259        return ArgTypeResult::Invalid();
260      case LengthModifier::None: return Ctx.IntTy;
261      case LengthModifier::AsChar: return ArgTypeResult::AnyCharTy;
262      case LengthModifier::AsShort: return Ctx.ShortTy;
263      case LengthModifier::AsLong: return Ctx.LongTy;
264      case LengthModifier::AsLongLong: return Ctx.LongLongTy;
265      case LengthModifier::AsIntMax:
266        return ArgTypeResult(Ctx.getIntMaxType(), "intmax_t");
267      case LengthModifier::AsSizeT:
268        // FIXME: How to get the corresponding signed version of size_t?
269        return ArgTypeResult();
270      case LengthModifier::AsPtrDiff:
271        return ArgTypeResult(Ctx.getPointerDiffType(), "ptrdiff_t");
272    }
273
274  if (CS.isUIntArg())
275    switch (LM.getKind()) {
276      case LengthModifier::AsLongDouble:
277        return ArgTypeResult::Invalid();
278      case LengthModifier::None: return Ctx.UnsignedIntTy;
279      case LengthModifier::AsChar: return Ctx.UnsignedCharTy;
280      case LengthModifier::AsShort: return Ctx.UnsignedShortTy;
281      case LengthModifier::AsLong: return Ctx.UnsignedLongTy;
282      case LengthModifier::AsLongLong: return Ctx.UnsignedLongLongTy;
283      case LengthModifier::AsIntMax:
284        return ArgTypeResult(Ctx.getUIntMaxType(), "uintmax_t");
285      case LengthModifier::AsSizeT:
286        return ArgTypeResult(Ctx.getSizeType(), "size_t");
287      case LengthModifier::AsPtrDiff:
288        // FIXME: How to get the corresponding unsigned
289        // version of ptrdiff_t?
290        return ArgTypeResult();
291    }
292
293  if (CS.isDoubleArg()) {
294    if (LM.getKind() == LengthModifier::AsLongDouble)
295      return Ctx.LongDoubleTy;
296    return Ctx.DoubleTy;
297  }
298
299  switch (CS.getKind()) {
300    case ConversionSpecifier::sArg:
301      if (LM.getKind() == LengthModifier::AsWideChar)
302        return ArgTypeResult(ArgTypeResult::WCStrTy, "wchar_t *");
303      return ArgTypeResult::CStrTy;
304    case ConversionSpecifier::SArg:
305      // FIXME: This appears to be Mac OS X specific.
306      return ArgTypeResult(ArgTypeResult::WCStrTy, "wchar_t *");
307    case ConversionSpecifier::CArg:
308      return ArgTypeResult(Ctx.WCharTy, "wchar_t");
309    case ConversionSpecifier::pArg:
310      return ArgTypeResult::CPointerTy;
311    default:
312      break;
313  }
314
315  // FIXME: Handle other cases.
316  return ArgTypeResult();
317}
318
319bool PrintfSpecifier::fixType(QualType QT, const LangOptions &LangOpt) {
320  // Handle strings first (char *, wchar_t *)
321  if (QT->isPointerType() && (QT->getPointeeType()->isAnyCharacterType())) {
322    CS.setKind(ConversionSpecifier::sArg);
323
324    // Disable irrelevant flags
325    HasAlternativeForm = 0;
326    HasLeadingZeroes = 0;
327
328    // Set the long length modifier for wide characters
329    if (QT->getPointeeType()->isWideCharType())
330      LM.setKind(LengthModifier::AsWideChar);
331    else
332      LM.setKind(LengthModifier::None);
333
334    return true;
335  }
336
337  // We can only work with builtin types.
338  const BuiltinType *BT = QT->getAs<BuiltinType>();
339  if (!BT)
340    return false;
341
342  // Set length modifier
343  switch (BT->getKind()) {
344  case BuiltinType::Bool:
345  case BuiltinType::WChar_U:
346  case BuiltinType::WChar_S:
347  case BuiltinType::Char16:
348  case BuiltinType::Char32:
349  case BuiltinType::UInt128:
350  case BuiltinType::Int128:
351  case BuiltinType::Half:
352    // Various types which are non-trivial to correct.
353    return false;
354
355#define SIGNED_TYPE(Id, SingletonId)
356#define UNSIGNED_TYPE(Id, SingletonId)
357#define FLOATING_TYPE(Id, SingletonId)
358#define BUILTIN_TYPE(Id, SingletonId) \
359  case BuiltinType::Id:
360#include "clang/AST/BuiltinTypes.def"
361    // Misc other stuff which doesn't make sense here.
362    return false;
363
364  case BuiltinType::UInt:
365  case BuiltinType::Int:
366  case BuiltinType::Float:
367  case BuiltinType::Double:
368    LM.setKind(LengthModifier::None);
369    break;
370
371  case BuiltinType::Char_U:
372  case BuiltinType::UChar:
373  case BuiltinType::Char_S:
374  case BuiltinType::SChar:
375    LM.setKind(LengthModifier::AsChar);
376    break;
377
378  case BuiltinType::Short:
379  case BuiltinType::UShort:
380    LM.setKind(LengthModifier::AsShort);
381    break;
382
383  case BuiltinType::Long:
384  case BuiltinType::ULong:
385    LM.setKind(LengthModifier::AsLong);
386    break;
387
388  case BuiltinType::LongLong:
389  case BuiltinType::ULongLong:
390    LM.setKind(LengthModifier::AsLongLong);
391    break;
392
393  case BuiltinType::LongDouble:
394    LM.setKind(LengthModifier::AsLongDouble);
395    break;
396  }
397
398  // Handle size_t, ptrdiff_t, etc. that have dedicated length modifiers in C99.
399  if (isa<TypedefType>(QT) && (LangOpt.C99 || LangOpt.CPlusPlus0x)) {
400    const IdentifierInfo *Identifier = QT.getBaseTypeIdentifier();
401    if (Identifier->getName() == "size_t") {
402      LM.setKind(LengthModifier::AsSizeT);
403    } else if (Identifier->getName() == "ssize_t") {
404      // Not C99, but common in Unix.
405      LM.setKind(LengthModifier::AsSizeT);
406    } else if (Identifier->getName() == "intmax_t") {
407      LM.setKind(LengthModifier::AsIntMax);
408    } else if (Identifier->getName() == "uintmax_t") {
409      LM.setKind(LengthModifier::AsIntMax);
410    } else if (Identifier->getName() == "ptrdiff_t") {
411      LM.setKind(LengthModifier::AsPtrDiff);
412    }
413  }
414
415  // Set conversion specifier and disable any flags which do not apply to it.
416  // Let typedefs to char fall through to int, as %c is silly for uint8_t.
417  if (isa<TypedefType>(QT) && QT->isAnyCharacterType()) {
418    CS.setKind(ConversionSpecifier::cArg);
419    LM.setKind(LengthModifier::None);
420    Precision.setHowSpecified(OptionalAmount::NotSpecified);
421    HasAlternativeForm = 0;
422    HasLeadingZeroes = 0;
423    HasPlusPrefix = 0;
424  }
425  // Test for Floating type first as LongDouble can pass isUnsignedIntegerType
426  else if (QT->isRealFloatingType()) {
427    CS.setKind(ConversionSpecifier::fArg);
428  }
429  else if (QT->isSignedIntegerType()) {
430    CS.setKind(ConversionSpecifier::dArg);
431    HasAlternativeForm = 0;
432  }
433  else if (QT->isUnsignedIntegerType()) {
434    // Preserve the original formatting, e.g. 'X', 'o'.
435    if (!cast<PrintfConversionSpecifier>(CS).isUIntArg())
436      CS.setKind(ConversionSpecifier::uArg);
437    HasAlternativeForm = 0;
438    HasPlusPrefix = 0;
439  } else {
440    llvm_unreachable("Unexpected type");
441  }
442
443  return true;
444}
445
446void PrintfSpecifier::toString(raw_ostream &os) const {
447  // Whilst some features have no defined order, we are using the order
448  // appearing in the C99 standard (ISO/IEC 9899:1999 (E) 7.19.6.1)
449  os << "%";
450
451  // Positional args
452  if (usesPositionalArg()) {
453    os << getPositionalArgIndex() << "$";
454  }
455
456  // Conversion flags
457  if (IsLeftJustified)    os << "-";
458  if (HasPlusPrefix)      os << "+";
459  if (HasSpacePrefix)     os << " ";
460  if (HasAlternativeForm) os << "#";
461  if (HasLeadingZeroes)   os << "0";
462
463  // Minimum field width
464  FieldWidth.toString(os);
465  // Precision
466  Precision.toString(os);
467  // Length modifier
468  os << LM.toString();
469  // Conversion specifier
470  os << CS.toString();
471}
472
473bool PrintfSpecifier::hasValidPlusPrefix() const {
474  if (!HasPlusPrefix)
475    return true;
476
477  // The plus prefix only makes sense for signed conversions
478  switch (CS.getKind()) {
479  case ConversionSpecifier::dArg:
480  case ConversionSpecifier::iArg:
481  case ConversionSpecifier::fArg:
482  case ConversionSpecifier::FArg:
483  case ConversionSpecifier::eArg:
484  case ConversionSpecifier::EArg:
485  case ConversionSpecifier::gArg:
486  case ConversionSpecifier::GArg:
487  case ConversionSpecifier::aArg:
488  case ConversionSpecifier::AArg:
489    return true;
490
491  default:
492    return false;
493  }
494}
495
496bool PrintfSpecifier::hasValidAlternativeForm() const {
497  if (!HasAlternativeForm)
498    return true;
499
500  // Alternate form flag only valid with the oxXaAeEfFgG conversions
501  switch (CS.getKind()) {
502  case ConversionSpecifier::oArg:
503  case ConversionSpecifier::xArg:
504  case ConversionSpecifier::XArg:
505  case ConversionSpecifier::aArg:
506  case ConversionSpecifier::AArg:
507  case ConversionSpecifier::eArg:
508  case ConversionSpecifier::EArg:
509  case ConversionSpecifier::fArg:
510  case ConversionSpecifier::FArg:
511  case ConversionSpecifier::gArg:
512  case ConversionSpecifier::GArg:
513    return true;
514
515  default:
516    return false;
517  }
518}
519
520bool PrintfSpecifier::hasValidLeadingZeros() const {
521  if (!HasLeadingZeroes)
522    return true;
523
524  // Leading zeroes flag only valid with the diouxXaAeEfFgG conversions
525  switch (CS.getKind()) {
526  case ConversionSpecifier::dArg:
527  case ConversionSpecifier::iArg:
528  case ConversionSpecifier::oArg:
529  case ConversionSpecifier::uArg:
530  case ConversionSpecifier::xArg:
531  case ConversionSpecifier::XArg:
532  case ConversionSpecifier::aArg:
533  case ConversionSpecifier::AArg:
534  case ConversionSpecifier::eArg:
535  case ConversionSpecifier::EArg:
536  case ConversionSpecifier::fArg:
537  case ConversionSpecifier::FArg:
538  case ConversionSpecifier::gArg:
539  case ConversionSpecifier::GArg:
540    return true;
541
542  default:
543    return false;
544  }
545}
546
547bool PrintfSpecifier::hasValidSpacePrefix() const {
548  if (!HasSpacePrefix)
549    return true;
550
551  // The space prefix only makes sense for signed conversions
552  switch (CS.getKind()) {
553  case ConversionSpecifier::dArg:
554  case ConversionSpecifier::iArg:
555  case ConversionSpecifier::fArg:
556  case ConversionSpecifier::FArg:
557  case ConversionSpecifier::eArg:
558  case ConversionSpecifier::EArg:
559  case ConversionSpecifier::gArg:
560  case ConversionSpecifier::GArg:
561  case ConversionSpecifier::aArg:
562  case ConversionSpecifier::AArg:
563    return true;
564
565  default:
566    return false;
567  }
568}
569
570bool PrintfSpecifier::hasValidLeftJustified() const {
571  if (!IsLeftJustified)
572    return true;
573
574  // The left justified flag is valid for all conversions except n
575  switch (CS.getKind()) {
576  case ConversionSpecifier::nArg:
577    return false;
578
579  default:
580    return true;
581  }
582}
583
584bool PrintfSpecifier::hasValidThousandsGroupingPrefix() const {
585  if (!HasThousandsGrouping)
586    return true;
587
588  switch (CS.getKind()) {
589    case ConversionSpecifier::dArg:
590    case ConversionSpecifier::iArg:
591    case ConversionSpecifier::uArg:
592    case ConversionSpecifier::fArg:
593    case ConversionSpecifier::FArg:
594    case ConversionSpecifier::gArg:
595    case ConversionSpecifier::GArg:
596      return true;
597    default:
598      return false;
599  }
600}
601
602bool PrintfSpecifier::hasValidPrecision() const {
603  if (Precision.getHowSpecified() == OptionalAmount::NotSpecified)
604    return true;
605
606  // Precision is only valid with the diouxXaAeEfFgGs conversions
607  switch (CS.getKind()) {
608  case ConversionSpecifier::dArg:
609  case ConversionSpecifier::iArg:
610  case ConversionSpecifier::oArg:
611  case ConversionSpecifier::uArg:
612  case ConversionSpecifier::xArg:
613  case ConversionSpecifier::XArg:
614  case ConversionSpecifier::aArg:
615  case ConversionSpecifier::AArg:
616  case ConversionSpecifier::eArg:
617  case ConversionSpecifier::EArg:
618  case ConversionSpecifier::fArg:
619  case ConversionSpecifier::FArg:
620  case ConversionSpecifier::gArg:
621  case ConversionSpecifier::GArg:
622  case ConversionSpecifier::sArg:
623    return true;
624
625  default:
626    return false;
627  }
628}
629bool PrintfSpecifier::hasValidFieldWidth() const {
630  if (FieldWidth.getHowSpecified() == OptionalAmount::NotSpecified)
631      return true;
632
633  // The field width is valid for all conversions except n
634  switch (CS.getKind()) {
635  case ConversionSpecifier::nArg:
636    return false;
637
638  default:
639    return true;
640  }
641}
642