PrintfFormatString.cpp revision 6ecb950c65329f8d6ce9ad0514632df35a5ab61f 
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  }
42  else {
43    const OptionalAmount Amt = ParsePositionAmount(H, Start, Beg, E,
44                                           analyze_format_string::PrecisionPos);
45    if (Amt.isInvalid())
46      return true;
47    FS.setPrecision(Amt);
48  }
49  return false;
50}
51
52static PrintfSpecifierResult ParsePrintfSpecifier(FormatStringHandler &H,
53                                                  const char *&Beg,
54                                                  const char *E,
55                                                  unsigned &argIndex) {
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 '-': FS.setIsLeftJustified(I); break;
104      case '+': FS.setHasPlusPrefix(I); break;
105      case ' ': FS.setHasSpacePrefix(I); break;
106      case '#': FS.setHasAlternativeForm(I); break;
107      case '0': FS.setHasLeadingZeros(I); break;
108    }
109    if (!hasMore)
110      break;
111  }
112
113  if (I == E) {
114    // No more characters left?
115    H.HandleIncompleteSpecifier(Start, E - Start);
116    return true;
117  }
118
119  // Look for the field width (if any).
120  if (ParseFieldWidth(H, FS, Start, I, E,
121                      FS.usesPositionalArg() ? 0 : &argIndex))
122    return true;
123
124  if (I == E) {
125    // No more characters left?
126    H.HandleIncompleteSpecifier(Start, E - Start);
127    return true;
128  }
129
130  // Look for the precision (if any).
131  if (*I == '.') {
132    ++I;
133    if (I == E) {
134      H.HandleIncompleteSpecifier(Start, E - Start);
135      return true;
136    }
137
138    if (ParsePrecision(H, FS, Start, I, E,
139                       FS.usesPositionalArg() ? 0 : &argIndex))
140      return true;
141
142    if (I == E) {
143      // No more characters left?
144      H.HandleIncompleteSpecifier(Start, E - Start);
145      return true;
146    }
147  }
148
149  // Look for the length modifier.
150  if (ParseLengthModifier(FS, I, E) && I == E) {
151    // No more characters left?
152    H.HandleIncompleteSpecifier(Start, E - Start);
153    return true;
154  }
155
156  if (*I == '\0') {
157    // Detect spurious null characters, which are likely errors.
158    H.HandleNullChar(I);
159    return true;
160  }
161
162  // Finally, look for the conversion specifier.
163  const char *conversionPosition = I++;
164  ConversionSpecifier::Kind k = ConversionSpecifier::InvalidSpecifier;
165  switch (*conversionPosition) {
166    default:
167      break;
168    // C99: 7.19.6.1 (section 8).
169    case '%': k = ConversionSpecifier::PercentArg;   break;
170    case 'A': k = ConversionSpecifier::AArg; break;
171    case 'E': k = ConversionSpecifier::EArg; break;
172    case 'F': k = ConversionSpecifier::FArg; break;
173    case 'G': k = ConversionSpecifier::GArg; break;
174    case 'X': k = ConversionSpecifier::XArg; break;
175    case 'a': k = ConversionSpecifier::aArg; break;
176    case 'c': k = ConversionSpecifier::cArg; break;
177    case 'd': k = ConversionSpecifier::dArg; break;
178    case 'e': k = ConversionSpecifier::eArg; break;
179    case 'f': k = ConversionSpecifier::fArg; break;
180    case 'g': k = ConversionSpecifier::gArg; break;
181    case 'i': k = ConversionSpecifier::iArg; break;
182    case 'n': k = ConversionSpecifier::nArg; break;
183    case 'o': k = ConversionSpecifier::oArg; break;
184    case 'p': k = ConversionSpecifier::pArg;   break;
185    case 's': k = ConversionSpecifier::sArg;      break;
186    case 'u': k = ConversionSpecifier::uArg; break;
187    case 'x': k = ConversionSpecifier::xArg; break;
188    // Mac OS X (unicode) specific
189    case 'C': k = ConversionSpecifier::CArg; break;
190    case 'S': k = ConversionSpecifier::SArg; break;
191    // Objective-C.
192    case '@': k = ConversionSpecifier::ObjCObjArg; break;
193    // Glibc specific.
194    case 'm': k = ConversionSpecifier::PrintErrno; break;
195  }
196  PrintfConversionSpecifier CS(conversionPosition, k);
197  FS.setConversionSpecifier(CS);
198  if (CS.consumesDataArgument() && !FS.usesPositionalArg())
199    FS.setArgIndex(argIndex++);
200
201  if (k == ConversionSpecifier::InvalidSpecifier) {
202    // Assume the conversion takes one argument.
203    return !H.HandleInvalidPrintfConversionSpecifier(FS, Beg, I - Beg);
204  }
205  return PrintfSpecifierResult(Start, FS);
206}
207
208bool clang::analyze_format_string::ParsePrintfString(FormatStringHandler &H,
209                                                     const char *I,
210                                                     const char *E) {
211
212  unsigned argIndex = 0;
213
214  // Keep looking for a format specifier until we have exhausted the string.
215  while (I != E) {
216    const PrintfSpecifierResult &FSR = ParsePrintfSpecifier(H, I, E, argIndex);
217    // Did a fail-stop error of any kind occur when parsing the specifier?
218    // If so, don't do any more processing.
219    if (FSR.shouldStop())
220      return true;;
221    // Did we exhaust the string or encounter an error that
222    // we can recover from?
223    if (!FSR.hasValue())
224      continue;
225    // We have a format specifier.  Pass it to the callback.
226    if (!H.HandlePrintfSpecifier(FSR.getValue(), FSR.getStart(),
227                                 I - FSR.getStart()))
228      return true;
229  }
230  assert(I == E && "Format string not exhausted");
231  return false;
232}
233
234//===----------------------------------------------------------------------===//
235// Methods on ConversionSpecifier.
236//===----------------------------------------------------------------------===//
237const char *ConversionSpecifier::toString() const {
238  switch (kind) {
239  case dArg: return "d";
240  case iArg: return "i";
241  case oArg: return "o";
242  case uArg: return "u";
243  case xArg: return "x";
244  case XArg: return "X";
245  case fArg: return "f";
246  case FArg: return "F";
247  case eArg: return "e";
248  case EArg: return "E";
249  case gArg: return "g";
250  case GArg: return "G";
251  case aArg: return "a";
252  case AArg: return "A";
253  case cArg:     return "c";
254  case sArg:          return "s";
255  case pArg:       return "p";
256  case nArg:     return "n";
257  case PercentArg:       return "%";
258  case InvalidSpecifier: return NULL;
259
260  // MacOS X unicode extensions.
261  case CArg:          return "C";
262  case SArg: return "S";
263
264  // Objective-C specific specifiers.
265  case ObjCObjArg: return "@";
266
267  // GlibC specific specifiers.
268  case PrintErrno: return "m";
269  }
270  return NULL;
271}
272
273//===----------------------------------------------------------------------===//
274// Methods on PrintfSpecifier.
275//===----------------------------------------------------------------------===//
276
277ArgTypeResult PrintfSpecifier::getArgType(ASTContext &Ctx) const {
278  const PrintfConversionSpecifier &CS = getConversionSpecifier();
279
280  if (!CS.consumesDataArgument())
281    return ArgTypeResult::Invalid();
282
283  if (CS.isIntArg())
284    switch (LM.getKind()) {
285      case LengthModifier::AsLongDouble:
286        return ArgTypeResult::Invalid();
287      case LengthModifier::None: return Ctx.IntTy;
288      case LengthModifier::AsChar: return Ctx.SignedCharTy;
289      case LengthModifier::AsShort: return Ctx.ShortTy;
290      case LengthModifier::AsLong: return Ctx.LongTy;
291      case LengthModifier::AsLongLong: return Ctx.LongLongTy;
292      case LengthModifier::AsIntMax:
293        // FIXME: Return unknown for now.
294        return ArgTypeResult();
295      case LengthModifier::AsSizeT: return Ctx.getSizeType();
296      case LengthModifier::AsPtrDiff: return Ctx.getPointerDiffType();
297    }
298
299  if (CS.isUIntArg())
300    switch (LM.getKind()) {
301      case LengthModifier::AsLongDouble:
302        return ArgTypeResult::Invalid();
303      case LengthModifier::None: return Ctx.UnsignedIntTy;
304      case LengthModifier::AsChar: return Ctx.UnsignedCharTy;
305      case LengthModifier::AsShort: return Ctx.UnsignedShortTy;
306      case LengthModifier::AsLong: return Ctx.UnsignedLongTy;
307      case LengthModifier::AsLongLong: return Ctx.UnsignedLongLongTy;
308      case LengthModifier::AsIntMax:
309        // FIXME: Return unknown for now.
310        return ArgTypeResult();
311      case LengthModifier::AsSizeT:
312        // FIXME: How to get the corresponding unsigned
313        // version of size_t?
314        return ArgTypeResult();
315      case LengthModifier::AsPtrDiff:
316        // FIXME: How to get the corresponding unsigned
317        // version of ptrdiff_t?
318        return ArgTypeResult();
319    }
320
321  if (CS.isDoubleArg()) {
322    if (LM.getKind() == LengthModifier::AsLongDouble)
323      return Ctx.LongDoubleTy;
324    return Ctx.DoubleTy;
325  }
326
327  switch (CS.getKind()) {
328    case ConversionSpecifier::sArg:
329      return ArgTypeResult(LM.getKind() == LengthModifier::AsWideChar ?
330          ArgTypeResult::WCStrTy : ArgTypeResult::CStrTy);
331    case ConversionSpecifier::SArg:
332      // FIXME: This appears to be Mac OS X specific.
333      return ArgTypeResult::WCStrTy;
334    case ConversionSpecifier::CArg:
335      return Ctx.WCharTy;
336    case ConversionSpecifier::pArg:
337      return ArgTypeResult::CPointerTy;
338    default:
339      break;
340  }
341
342  // FIXME: Handle other cases.
343  return ArgTypeResult();
344}
345
346bool PrintfSpecifier::fixType(QualType QT) {
347  // Handle strings first (char *, wchar_t *)
348  if (QT->isPointerType() && (QT->getPointeeType()->isAnyCharacterType())) {
349    CS.setKind(ConversionSpecifier::sArg);
350
351    // Disable irrelevant flags
352    HasAlternativeForm = 0;
353    HasLeadingZeroes = 0;
354
355    // Set the long length modifier for wide characters
356    if (QT->getPointeeType()->isWideCharType())
357      LM.setKind(LengthModifier::AsWideChar);
358
359    return true;
360  }
361
362  // We can only work with builtin types.
363  if (!QT->isBuiltinType())
364    return false;
365
366  // Everything else should be a base type
367  const BuiltinType *BT = QT->getAs<BuiltinType>();
368
369  // Set length modifier
370  switch (BT->getKind()) {
371  default:
372    // The rest of the conversions are either optional or for non-builtin types
373    LM.setKind(LengthModifier::None);
374    break;
375
376  case BuiltinType::WChar:
377  case BuiltinType::Long:
378  case BuiltinType::ULong:
379    LM.setKind(LengthModifier::AsLong);
380    break;
381
382  case BuiltinType::LongLong:
383  case BuiltinType::ULongLong:
384    LM.setKind(LengthModifier::AsLongLong);
385    break;
386
387  case BuiltinType::LongDouble:
388    LM.setKind(LengthModifier::AsLongDouble);
389    break;
390  }
391
392  // Set conversion specifier and disable any flags which do not apply to it.
393  if (QT->isAnyCharacterType()) {
394    CS.setKind(ConversionSpecifier::cArg);
395    Precision.setHowSpecified(OptionalAmount::NotSpecified);
396    HasAlternativeForm = 0;
397    HasLeadingZeroes = 0;
398    HasPlusPrefix = 0;
399  }
400  // Test for Floating type first as LongDouble can pass isUnsignedIntegerType
401  else if (QT->isRealFloatingType()) {
402    CS.setKind(ConversionSpecifier::fArg);
403  }
404  else if (QT->isPointerType()) {
405    CS.setKind(ConversionSpecifier::pArg);
406    Precision.setHowSpecified(OptionalAmount::NotSpecified);
407    HasAlternativeForm = 0;
408    HasLeadingZeroes = 0;
409    HasPlusPrefix = 0;
410  }
411  else if (QT->isSignedIntegerType()) {
412    CS.setKind(ConversionSpecifier::dArg);
413    HasAlternativeForm = 0;
414  }
415  else if (QT->isUnsignedIntegerType()) {
416    CS.setKind(ConversionSpecifier::uArg);
417    HasAlternativeForm = 0;
418    HasPlusPrefix = 0;
419  }
420  else {
421    return false;
422  }
423
424  return true;
425}
426
427void PrintfSpecifier::toString(llvm::raw_ostream &os) const {
428  // Whilst some features have no defined order, we are using the order
429  // appearing in the C99 standard (ISO/IEC 9899:1999 (E) ¤7.19.6.1)
430  os << "%";
431
432  // Positional args
433  if (usesPositionalArg()) {
434    os << getPositionalArgIndex() << "$";
435  }
436
437  // Conversion flags
438  if (IsLeftJustified)    os << "-";
439  if (HasPlusPrefix)      os << "+";
440  if (HasSpacePrefix)     os << " ";
441  if (HasAlternativeForm) os << "#";
442  if (HasLeadingZeroes)   os << "0";
443
444  // Minimum field width
445  FieldWidth.toString(os);
446  // Precision
447  Precision.toString(os);
448  // Length modifier
449  os << LM.toString();
450  // Conversion specifier
451  os << CS.toString();
452}
453
454bool PrintfSpecifier::hasValidPlusPrefix() const {
455  if (!HasPlusPrefix)
456    return true;
457
458  // The plus prefix only makes sense for signed conversions
459  switch (CS.getKind()) {
460  case ConversionSpecifier::dArg:
461  case ConversionSpecifier::iArg:
462  case ConversionSpecifier::fArg:
463  case ConversionSpecifier::FArg:
464  case ConversionSpecifier::eArg:
465  case ConversionSpecifier::EArg:
466  case ConversionSpecifier::gArg:
467  case ConversionSpecifier::GArg:
468  case ConversionSpecifier::aArg:
469  case ConversionSpecifier::AArg:
470    return true;
471
472  default:
473    return false;
474  }
475}
476
477bool PrintfSpecifier::hasValidAlternativeForm() const {
478  if (!HasAlternativeForm)
479    return true;
480
481  // Alternate form flag only valid with the oxaAeEfFgG conversions
482  switch (CS.getKind()) {
483  case ConversionSpecifier::oArg:
484  case ConversionSpecifier::xArg:
485  case ConversionSpecifier::aArg:
486  case ConversionSpecifier::AArg:
487  case ConversionSpecifier::eArg:
488  case ConversionSpecifier::EArg:
489  case ConversionSpecifier::fArg:
490  case ConversionSpecifier::FArg:
491  case ConversionSpecifier::gArg:
492  case ConversionSpecifier::GArg:
493    return true;
494
495  default:
496    return false;
497  }
498}
499
500bool PrintfSpecifier::hasValidLeadingZeros() const {
501  if (!HasLeadingZeroes)
502    return true;
503
504  // Leading zeroes flag only valid with the diouxXaAeEfFgG conversions
505  switch (CS.getKind()) {
506  case ConversionSpecifier::dArg:
507  case ConversionSpecifier::iArg:
508  case ConversionSpecifier::oArg:
509  case ConversionSpecifier::uArg:
510  case ConversionSpecifier::xArg:
511  case ConversionSpecifier::XArg:
512  case ConversionSpecifier::aArg:
513  case ConversionSpecifier::AArg:
514  case ConversionSpecifier::eArg:
515  case ConversionSpecifier::EArg:
516  case ConversionSpecifier::fArg:
517  case ConversionSpecifier::FArg:
518  case ConversionSpecifier::gArg:
519  case ConversionSpecifier::GArg:
520    return true;
521
522  default:
523    return false;
524  }
525}
526
527bool PrintfSpecifier::hasValidSpacePrefix() const {
528  if (!HasSpacePrefix)
529    return true;
530
531  // The space prefix only makes sense for signed conversions
532  switch (CS.getKind()) {
533  case ConversionSpecifier::dArg:
534  case ConversionSpecifier::iArg:
535  case ConversionSpecifier::fArg:
536  case ConversionSpecifier::FArg:
537  case ConversionSpecifier::eArg:
538  case ConversionSpecifier::EArg:
539  case ConversionSpecifier::gArg:
540  case ConversionSpecifier::GArg:
541  case ConversionSpecifier::aArg:
542  case ConversionSpecifier::AArg:
543    return true;
544
545  default:
546    return false;
547  }
548}
549
550bool PrintfSpecifier::hasValidLeftJustified() const {
551  if (!IsLeftJustified)
552    return true;
553
554  // The left justified flag is valid for all conversions except n
555  switch (CS.getKind()) {
556  case ConversionSpecifier::nArg:
557    return false;
558
559  default:
560    return true;
561  }
562}
563
564bool PrintfSpecifier::hasValidLengthModifier() const {
565  switch (LM.getKind()) {
566  case LengthModifier::None:
567    return true;
568
569  // Handle most integer flags
570  case LengthModifier::AsChar:
571  case LengthModifier::AsShort:
572  case LengthModifier::AsLongLong:
573  case LengthModifier::AsIntMax:
574  case LengthModifier::AsSizeT:
575  case LengthModifier::AsPtrDiff:
576    switch (CS.getKind()) {
577    case ConversionSpecifier::dArg:
578    case ConversionSpecifier::iArg:
579    case ConversionSpecifier::oArg:
580    case ConversionSpecifier::uArg:
581    case ConversionSpecifier::xArg:
582    case ConversionSpecifier::XArg:
583    case ConversionSpecifier::nArg:
584      return true;
585    default:
586      return false;
587    }
588
589  // Handle 'l' flag
590  case LengthModifier::AsLong:
591    switch (CS.getKind()) {
592    case ConversionSpecifier::dArg:
593    case ConversionSpecifier::iArg:
594    case ConversionSpecifier::oArg:
595    case ConversionSpecifier::uArg:
596    case ConversionSpecifier::xArg:
597    case ConversionSpecifier::XArg:
598    case ConversionSpecifier::aArg:
599    case ConversionSpecifier::AArg:
600    case ConversionSpecifier::fArg:
601    case ConversionSpecifier::FArg:
602    case ConversionSpecifier::eArg:
603    case ConversionSpecifier::EArg:
604    case ConversionSpecifier::gArg:
605    case ConversionSpecifier::GArg:
606    case ConversionSpecifier::nArg:
607    case ConversionSpecifier::cArg:
608    case ConversionSpecifier::sArg:
609      return true;
610    default:
611      return false;
612    }
613
614  case LengthModifier::AsLongDouble:
615    switch (CS.getKind()) {
616    case ConversionSpecifier::aArg:
617    case ConversionSpecifier::AArg:
618    case ConversionSpecifier::fArg:
619    case ConversionSpecifier::FArg:
620    case ConversionSpecifier::eArg:
621    case ConversionSpecifier::EArg:
622    case ConversionSpecifier::gArg:
623    case ConversionSpecifier::GArg:
624      return true;
625    default:
626      return false;
627    }
628  }
629  return false;
630}
631
632bool PrintfSpecifier::hasValidPrecision() const {
633  if (Precision.getHowSpecified() == OptionalAmount::NotSpecified)
634    return true;
635
636  // Precision is only valid with the diouxXaAeEfFgGs conversions
637  switch (CS.getKind()) {
638  case ConversionSpecifier::dArg:
639  case ConversionSpecifier::iArg:
640  case ConversionSpecifier::oArg:
641  case ConversionSpecifier::uArg:
642  case ConversionSpecifier::xArg:
643  case ConversionSpecifier::XArg:
644  case ConversionSpecifier::aArg:
645  case ConversionSpecifier::AArg:
646  case ConversionSpecifier::eArg:
647  case ConversionSpecifier::EArg:
648  case ConversionSpecifier::fArg:
649  case ConversionSpecifier::FArg:
650  case ConversionSpecifier::gArg:
651  case ConversionSpecifier::GArg:
652  case ConversionSpecifier::sArg:
653    return true;
654
655  default:
656    return false;
657  }
658}
659bool PrintfSpecifier::hasValidFieldWidth() const {
660  if (FieldWidth.getHowSpecified() == OptionalAmount::NotSpecified)
661      return true;
662
663  // The field width is valid for all conversions except n
664  switch (CS.getKind()) {
665  case ConversionSpecifier::nArg:
666    return false;
667
668  default:
669    return true;
670  }
671}
672