FormatString.cpp revision 58e1e54476d610d6c33ef483f216ed8a1282d35c 
1// FormatString.cpp - Common stuff for handling printf/scanf formats -*- 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// Shared details for processing format strings of printf and scanf
11// (and friends).
12//
13//===----------------------------------------------------------------------===//
14
15#include "FormatStringParsing.h"
16#include "clang/Basic/LangOptions.h"
17
18using clang::analyze_format_string::ArgType;
19using clang::analyze_format_string::FormatStringHandler;
20using clang::analyze_format_string::FormatSpecifier;
21using clang::analyze_format_string::LengthModifier;
22using clang::analyze_format_string::OptionalAmount;
23using clang::analyze_format_string::PositionContext;
24using clang::analyze_format_string::ConversionSpecifier;
25using namespace clang;
26
27// Key function to FormatStringHandler.
28FormatStringHandler::~FormatStringHandler() {}
29
30//===----------------------------------------------------------------------===//
31// Functions for parsing format strings components in both printf and
32// scanf format strings.
33//===----------------------------------------------------------------------===//
34
35OptionalAmount
36clang::analyze_format_string::ParseAmount(const char *&Beg, const char *E) {
37  const char *I = Beg;
38  UpdateOnReturn <const char*> UpdateBeg(Beg, I);
39
40  unsigned accumulator = 0;
41  bool hasDigits = false;
42
43  for ( ; I != E; ++I) {
44    char c = *I;
45    if (c >= '0' && c <= '9') {
46      hasDigits = true;
47      accumulator = (accumulator * 10) + (c - '0');
48      continue;
49    }
50
51    if (hasDigits)
52      return OptionalAmount(OptionalAmount::Constant, accumulator, Beg, I - Beg,
53          false);
54
55    break;
56  }
57
58  return OptionalAmount();
59}
60
61OptionalAmount
62clang::analyze_format_string::ParseNonPositionAmount(const char *&Beg,
63                                                     const char *E,
64                                                     unsigned &argIndex) {
65  if (*Beg == '*') {
66    ++Beg;
67    return OptionalAmount(OptionalAmount::Arg, argIndex++, Beg, 0, false);
68  }
69
70  return ParseAmount(Beg, E);
71}
72
73OptionalAmount
74clang::analyze_format_string::ParsePositionAmount(FormatStringHandler &H,
75                                                  const char *Start,
76                                                  const char *&Beg,
77                                                  const char *E,
78                                                  PositionContext p) {
79  if (*Beg == '*') {
80    const char *I = Beg + 1;
81    const OptionalAmount &Amt = ParseAmount(I, E);
82
83    if (Amt.getHowSpecified() == OptionalAmount::NotSpecified) {
84      H.HandleInvalidPosition(Beg, I - Beg, p);
85      return OptionalAmount(false);
86    }
87
88    if (I == E) {
89      // No more characters left?
90      H.HandleIncompleteSpecifier(Start, E - Start);
91      return OptionalAmount(false);
92    }
93
94    assert(Amt.getHowSpecified() == OptionalAmount::Constant);
95
96    if (*I == '$') {
97      // Handle positional arguments
98
99      // Special case: '*0$', since this is an easy mistake.
100      if (Amt.getConstantAmount() == 0) {
101        H.HandleZeroPosition(Beg, I - Beg + 1);
102        return OptionalAmount(false);
103      }
104
105      const char *Tmp = Beg;
106      Beg = ++I;
107
108      return OptionalAmount(OptionalAmount::Arg, Amt.getConstantAmount() - 1,
109                            Tmp, 0, true);
110    }
111
112    H.HandleInvalidPosition(Beg, I - Beg, p);
113    return OptionalAmount(false);
114  }
115
116  return ParseAmount(Beg, E);
117}
118
119
120bool
121clang::analyze_format_string::ParseFieldWidth(FormatStringHandler &H,
122                                              FormatSpecifier &CS,
123                                              const char *Start,
124                                              const char *&Beg, const char *E,
125                                              unsigned *argIndex) {
126  // FIXME: Support negative field widths.
127  if (argIndex) {
128    CS.setFieldWidth(ParseNonPositionAmount(Beg, E, *argIndex));
129  }
130  else {
131    const OptionalAmount Amt =
132      ParsePositionAmount(H, Start, Beg, E,
133                          analyze_format_string::FieldWidthPos);
134
135    if (Amt.isInvalid())
136      return true;
137    CS.setFieldWidth(Amt);
138  }
139  return false;
140}
141
142bool
143clang::analyze_format_string::ParseArgPosition(FormatStringHandler &H,
144                                               FormatSpecifier &FS,
145                                               const char *Start,
146                                               const char *&Beg,
147                                               const char *E) {
148  const char *I = Beg;
149
150  const OptionalAmount &Amt = ParseAmount(I, E);
151
152  if (I == E) {
153    // No more characters left?
154    H.HandleIncompleteSpecifier(Start, E - Start);
155    return true;
156  }
157
158  if (Amt.getHowSpecified() == OptionalAmount::Constant && *(I++) == '$') {
159    // Warn that positional arguments are non-standard.
160    H.HandlePosition(Start, I - Start);
161
162    // Special case: '%0$', since this is an easy mistake.
163    if (Amt.getConstantAmount() == 0) {
164      H.HandleZeroPosition(Start, I - Start);
165      return true;
166    }
167
168    FS.setArgIndex(Amt.getConstantAmount() - 1);
169    FS.setUsesPositionalArg();
170    // Update the caller's pointer if we decided to consume
171    // these characters.
172    Beg = I;
173    return false;
174  }
175
176  return false;
177}
178
179bool
180clang::analyze_format_string::ParseLengthModifier(FormatSpecifier &FS,
181                                                  const char *&I,
182                                                  const char *E,
183                                                  const LangOptions &LO,
184                                                  bool IsScanf) {
185  LengthModifier::Kind lmKind = LengthModifier::None;
186  const char *lmPosition = I;
187  switch (*I) {
188    default:
189      return false;
190    case 'h':
191      ++I;
192      lmKind = (I != E && *I == 'h') ? (++I, LengthModifier::AsChar)
193                                     : LengthModifier::AsShort;
194      break;
195    case 'l':
196      ++I;
197      lmKind = (I != E && *I == 'l') ? (++I, LengthModifier::AsLongLong)
198                                     : LengthModifier::AsLong;
199      break;
200    case 'j': lmKind = LengthModifier::AsIntMax;     ++I; break;
201    case 'z': lmKind = LengthModifier::AsSizeT;      ++I; break;
202    case 't': lmKind = LengthModifier::AsPtrDiff;    ++I; break;
203    case 'L': lmKind = LengthModifier::AsLongDouble; ++I; break;
204    case 'q': lmKind = LengthModifier::AsQuad;       ++I; break;
205    case 'a':
206      if (IsScanf && !LO.C99 && !LO.CPlusPlus0x) {
207        // For scanf in C90, look at the next character to see if this should
208        // be parsed as the GNU extension 'a' length modifier. If not, this
209        // will be parsed as a conversion specifier.
210        ++I;
211        if (I != E && (*I == 's' || *I == 'S' || *I == '[')) {
212          lmKind = LengthModifier::AsAllocate;
213          break;
214        }
215        --I;
216      }
217      return false;
218    case 'm':
219      if (IsScanf) {
220        lmKind = LengthModifier::AsMAllocate;
221        ++I;
222        break;
223      }
224      return false;
225  }
226  LengthModifier lm(lmPosition, lmKind);
227  FS.setLengthModifier(lm);
228  return true;
229}
230
231//===----------------------------------------------------------------------===//
232// Methods on ArgType.
233//===----------------------------------------------------------------------===//
234
235bool ArgType::matchesType(ASTContext &C, QualType argTy) const {
236  if (Ptr) {
237    // It has to be a pointer.
238    const PointerType *PT = argTy->getAs<PointerType>();
239    if (!PT)
240      return false;
241
242    // We cannot write through a const qualified pointer.
243    if (PT->getPointeeType().isConstQualified())
244      return false;
245
246    argTy = PT->getPointeeType();
247  }
248
249  switch (K) {
250    case InvalidTy:
251      llvm_unreachable("ArgType must be valid");
252
253    case UnknownTy:
254      return true;
255
256    case AnyCharTy: {
257      if (const EnumType *ETy = argTy->getAs<EnumType>())
258        argTy = ETy->getDecl()->getIntegerType();
259
260      if (const BuiltinType *BT = argTy->getAs<BuiltinType>())
261        switch (BT->getKind()) {
262          default:
263            break;
264          case BuiltinType::Char_S:
265          case BuiltinType::SChar:
266          case BuiltinType::UChar:
267          case BuiltinType::Char_U:
268            return true;
269        }
270      return false;
271    }
272
273    case SpecificTy: {
274      if (const EnumType *ETy = argTy->getAs<EnumType>())
275        argTy = ETy->getDecl()->getIntegerType();
276      argTy = C.getCanonicalType(argTy).getUnqualifiedType();
277
278      if (T == argTy)
279        return true;
280      // Check for "compatible types".
281      if (const BuiltinType *BT = argTy->getAs<BuiltinType>())
282        switch (BT->getKind()) {
283          default:
284            break;
285          case BuiltinType::Char_S:
286          case BuiltinType::SChar:
287          case BuiltinType::Char_U:
288          case BuiltinType::UChar:
289            return T == C.UnsignedCharTy || T == C.SignedCharTy;
290          case BuiltinType::Short:
291            return T == C.UnsignedShortTy;
292          case BuiltinType::UShort:
293            return T == C.ShortTy;
294          case BuiltinType::Int:
295            return T == C.UnsignedIntTy;
296          case BuiltinType::UInt:
297            return T == C.IntTy;
298          case BuiltinType::Long:
299            return T == C.UnsignedLongTy;
300          case BuiltinType::ULong:
301            return T == C.LongTy;
302          case BuiltinType::LongLong:
303            return T == C.UnsignedLongLongTy;
304          case BuiltinType::ULongLong:
305            return T == C.LongLongTy;
306        }
307      return false;
308    }
309
310    case CStrTy: {
311      const PointerType *PT = argTy->getAs<PointerType>();
312      if (!PT)
313        return false;
314      QualType pointeeTy = PT->getPointeeType();
315      if (const BuiltinType *BT = pointeeTy->getAs<BuiltinType>())
316        switch (BT->getKind()) {
317          case BuiltinType::Void:
318          case BuiltinType::Char_U:
319          case BuiltinType::UChar:
320          case BuiltinType::Char_S:
321          case BuiltinType::SChar:
322            return true;
323          default:
324            break;
325        }
326
327      return false;
328    }
329
330    case WCStrTy: {
331      const PointerType *PT = argTy->getAs<PointerType>();
332      if (!PT)
333        return false;
334      QualType pointeeTy =
335        C.getCanonicalType(PT->getPointeeType()).getUnqualifiedType();
336      return pointeeTy == C.getWCharType();
337    }
338
339    case WIntTy: {
340
341      QualType PromoArg =
342        argTy->isPromotableIntegerType()
343          ? C.getPromotedIntegerType(argTy) : argTy;
344
345      QualType WInt = C.getCanonicalType(C.getWIntType()).getUnqualifiedType();
346      PromoArg = C.getCanonicalType(PromoArg).getUnqualifiedType();
347
348      // If the promoted argument is the corresponding signed type of the
349      // wint_t type, then it should match.
350      if (PromoArg->hasSignedIntegerRepresentation() &&
351          C.getCorrespondingUnsignedType(PromoArg) == WInt)
352        return true;
353
354      return WInt == PromoArg;
355    }
356
357    case CPointerTy:
358      return argTy->isPointerType() || argTy->isObjCObjectPointerType() ||
359             argTy->isBlockPointerType() || argTy->isNullPtrType();
360
361    case ObjCPointerTy: {
362      if (argTy->getAs<ObjCObjectPointerType>() ||
363          argTy->getAs<BlockPointerType>())
364        return true;
365
366      // Handle implicit toll-free bridging.
367      if (const PointerType *PT = argTy->getAs<PointerType>()) {
368        // Things such as CFTypeRef are really just opaque pointers
369        // to C structs representing CF types that can often be bridged
370        // to Objective-C objects.  Since the compiler doesn't know which
371        // structs can be toll-free bridged, we just accept them all.
372        QualType pointee = PT->getPointeeType();
373        if (pointee->getAsStructureType() || pointee->isVoidType())
374          return true;
375      }
376      return false;
377    }
378  }
379
380  llvm_unreachable("Invalid ArgType Kind!");
381}
382
383QualType ArgType::getRepresentativeType(ASTContext &C) const {
384  QualType Res;
385  switch (K) {
386    case InvalidTy:
387      llvm_unreachable("No representative type for Invalid ArgType");
388    case UnknownTy:
389      llvm_unreachable("No representative type for Unknown ArgType");
390    case AnyCharTy:
391      Res = C.CharTy;
392      break;
393    case SpecificTy:
394      Res = T;
395      break;
396    case CStrTy:
397      Res = C.getPointerType(C.CharTy);
398      break;
399    case WCStrTy:
400      Res = C.getPointerType(C.getWCharType());
401      break;
402    case ObjCPointerTy:
403      Res = C.ObjCBuiltinIdTy;
404      break;
405    case CPointerTy:
406      Res = C.VoidPtrTy;
407      break;
408    case WIntTy: {
409      Res = C.getWIntType();
410      break;
411    }
412  }
413
414  if (Ptr)
415    Res = C.getPointerType(Res);
416  return Res;
417}
418
419std::string ArgType::getRepresentativeTypeName(ASTContext &C) const {
420  std::string S = getRepresentativeType(C).getAsString();
421
422  std::string Alias;
423  if (Name) {
424    // Use a specific name for this type, e.g. "size_t".
425    Alias = Name;
426    if (Ptr) {
427      // If ArgType is actually a pointer to T, append an asterisk.
428      Alias += (Alias[Alias.size()-1] == '*') ? "*" : " *";
429    }
430    // If Alias is the same as the underlying type, e.g. wchar_t, then drop it.
431    if (S == Alias)
432      Alias.clear();
433  }
434
435  if (!Alias.empty())
436    return std::string("'") + Alias + "' (aka '" + S + "')";
437  return std::string("'") + S + "'";
438}
439
440
441//===----------------------------------------------------------------------===//
442// Methods on OptionalAmount.
443//===----------------------------------------------------------------------===//
444
445ArgType
446analyze_format_string::OptionalAmount::getArgType(ASTContext &Ctx) const {
447  return Ctx.IntTy;
448}
449
450//===----------------------------------------------------------------------===//
451// Methods on LengthModifier.
452//===----------------------------------------------------------------------===//
453
454const char *
455analyze_format_string::LengthModifier::toString() const {
456  switch (kind) {
457  case AsChar:
458    return "hh";
459  case AsShort:
460    return "h";
461  case AsLong: // or AsWideChar
462    return "l";
463  case AsLongLong:
464    return "ll";
465  case AsQuad:
466    return "q";
467  case AsIntMax:
468    return "j";
469  case AsSizeT:
470    return "z";
471  case AsPtrDiff:
472    return "t";
473  case AsLongDouble:
474    return "L";
475  case AsAllocate:
476    return "a";
477  case AsMAllocate:
478    return "m";
479  case None:
480    return "";
481  }
482  return NULL;
483}
484
485//===----------------------------------------------------------------------===//
486// Methods on ConversionSpecifier.
487//===----------------------------------------------------------------------===//
488
489const char *ConversionSpecifier::toString() const {
490  switch (kind) {
491  case dArg: return "d";
492  case iArg: return "i";
493  case oArg: return "o";
494  case uArg: return "u";
495  case xArg: return "x";
496  case XArg: return "X";
497  case fArg: return "f";
498  case FArg: return "F";
499  case eArg: return "e";
500  case EArg: return "E";
501  case gArg: return "g";
502  case GArg: return "G";
503  case aArg: return "a";
504  case AArg: return "A";
505  case cArg: return "c";
506  case sArg: return "s";
507  case pArg: return "p";
508  case nArg: return "n";
509  case PercentArg:  return "%";
510  case ScanListArg: return "[";
511  case InvalidSpecifier: return NULL;
512
513  // MacOS X unicode extensions.
514  case CArg: return "C";
515  case SArg: return "S";
516
517  // Objective-C specific specifiers.
518  case ObjCObjArg: return "@";
519
520  // GlibC specific specifiers.
521  case PrintErrno: return "m";
522  }
523  return NULL;
524}
525
526//===----------------------------------------------------------------------===//
527// Methods on OptionalAmount.
528//===----------------------------------------------------------------------===//
529
530void OptionalAmount::toString(raw_ostream &os) const {
531  switch (hs) {
532  case Invalid:
533  case NotSpecified:
534    return;
535  case Arg:
536    if (UsesDotPrefix)
537        os << ".";
538    if (usesPositionalArg())
539      os << "*" << getPositionalArgIndex() << "$";
540    else
541      os << "*";
542    break;
543  case Constant:
544    if (UsesDotPrefix)
545        os << ".";
546    os << amt;
547    break;
548  }
549}
550
551bool FormatSpecifier::hasValidLengthModifier() const {
552  switch (LM.getKind()) {
553    case LengthModifier::None:
554      return true;
555
556    // Handle most integer flags
557    case LengthModifier::AsChar:
558    case LengthModifier::AsShort:
559    case LengthModifier::AsLongLong:
560    case LengthModifier::AsQuad:
561    case LengthModifier::AsIntMax:
562    case LengthModifier::AsSizeT:
563    case LengthModifier::AsPtrDiff:
564      switch (CS.getKind()) {
565        case ConversionSpecifier::dArg:
566        case ConversionSpecifier::iArg:
567        case ConversionSpecifier::oArg:
568        case ConversionSpecifier::uArg:
569        case ConversionSpecifier::xArg:
570        case ConversionSpecifier::XArg:
571        case ConversionSpecifier::nArg:
572          return true;
573        default:
574          return false;
575      }
576
577    // Handle 'l' flag
578    case LengthModifier::AsLong:
579      switch (CS.getKind()) {
580        case ConversionSpecifier::dArg:
581        case ConversionSpecifier::iArg:
582        case ConversionSpecifier::oArg:
583        case ConversionSpecifier::uArg:
584        case ConversionSpecifier::xArg:
585        case ConversionSpecifier::XArg:
586        case ConversionSpecifier::aArg:
587        case ConversionSpecifier::AArg:
588        case ConversionSpecifier::fArg:
589        case ConversionSpecifier::FArg:
590        case ConversionSpecifier::eArg:
591        case ConversionSpecifier::EArg:
592        case ConversionSpecifier::gArg:
593        case ConversionSpecifier::GArg:
594        case ConversionSpecifier::nArg:
595        case ConversionSpecifier::cArg:
596        case ConversionSpecifier::sArg:
597        case ConversionSpecifier::ScanListArg:
598          return true;
599        default:
600          return false;
601      }
602
603    case LengthModifier::AsLongDouble:
604      switch (CS.getKind()) {
605        case ConversionSpecifier::aArg:
606        case ConversionSpecifier::AArg:
607        case ConversionSpecifier::fArg:
608        case ConversionSpecifier::FArg:
609        case ConversionSpecifier::eArg:
610        case ConversionSpecifier::EArg:
611        case ConversionSpecifier::gArg:
612        case ConversionSpecifier::GArg:
613          return true;
614        // GNU extension.
615        case ConversionSpecifier::dArg:
616        case ConversionSpecifier::iArg:
617        case ConversionSpecifier::oArg:
618        case ConversionSpecifier::uArg:
619        case ConversionSpecifier::xArg:
620        case ConversionSpecifier::XArg:
621          return true;
622        default:
623          return false;
624      }
625
626    case LengthModifier::AsAllocate:
627      switch (CS.getKind()) {
628        case ConversionSpecifier::sArg:
629        case ConversionSpecifier::SArg:
630        case ConversionSpecifier::ScanListArg:
631          return true;
632        default:
633          return false;
634      }
635
636    case LengthModifier::AsMAllocate:
637      switch (CS.getKind()) {
638        case ConversionSpecifier::cArg:
639        case ConversionSpecifier::CArg:
640        case ConversionSpecifier::sArg:
641        case ConversionSpecifier::SArg:
642        case ConversionSpecifier::ScanListArg:
643          return true;
644        default:
645          return false;
646      }
647  }
648  llvm_unreachable("Invalid LengthModifier Kind!");
649}
650
651bool FormatSpecifier::hasStandardLengthModifier() const {
652  switch (LM.getKind()) {
653    case LengthModifier::None:
654    case LengthModifier::AsChar:
655    case LengthModifier::AsShort:
656    case LengthModifier::AsLong:
657    case LengthModifier::AsLongLong:
658    case LengthModifier::AsIntMax:
659    case LengthModifier::AsSizeT:
660    case LengthModifier::AsPtrDiff:
661    case LengthModifier::AsLongDouble:
662      return true;
663    case LengthModifier::AsAllocate:
664    case LengthModifier::AsMAllocate:
665    case LengthModifier::AsQuad:
666      return false;
667  }
668  llvm_unreachable("Invalid LengthModifier Kind!");
669}
670
671bool FormatSpecifier::hasStandardConversionSpecifier(const LangOptions &LangOpt) const {
672  switch (CS.getKind()) {
673    case ConversionSpecifier::cArg:
674    case ConversionSpecifier::dArg:
675    case ConversionSpecifier::iArg:
676    case ConversionSpecifier::oArg:
677    case ConversionSpecifier::uArg:
678    case ConversionSpecifier::xArg:
679    case ConversionSpecifier::XArg:
680    case ConversionSpecifier::fArg:
681    case ConversionSpecifier::FArg:
682    case ConversionSpecifier::eArg:
683    case ConversionSpecifier::EArg:
684    case ConversionSpecifier::gArg:
685    case ConversionSpecifier::GArg:
686    case ConversionSpecifier::aArg:
687    case ConversionSpecifier::AArg:
688    case ConversionSpecifier::sArg:
689    case ConversionSpecifier::pArg:
690    case ConversionSpecifier::nArg:
691    case ConversionSpecifier::ObjCObjArg:
692    case ConversionSpecifier::ScanListArg:
693    case ConversionSpecifier::PercentArg:
694      return true;
695    case ConversionSpecifier::CArg:
696    case ConversionSpecifier::SArg:
697      return LangOpt.ObjC1 || LangOpt.ObjC2;
698    case ConversionSpecifier::InvalidSpecifier:
699    case ConversionSpecifier::PrintErrno:
700      return false;
701  }
702  llvm_unreachable("Invalid ConversionSpecifier Kind!");
703}
704
705bool FormatSpecifier::hasStandardLengthConversionCombination() const {
706  if (LM.getKind() == LengthModifier::AsLongDouble) {
707    switch(CS.getKind()) {
708        case ConversionSpecifier::dArg:
709        case ConversionSpecifier::iArg:
710        case ConversionSpecifier::oArg:
711        case ConversionSpecifier::uArg:
712        case ConversionSpecifier::xArg:
713        case ConversionSpecifier::XArg:
714          return false;
715        default:
716          return true;
717    }
718  }
719  return true;
720}
721
722bool FormatSpecifier::namedTypeToLengthModifier(QualType QT,
723                                                LengthModifier &LM) {
724  assert(isa<TypedefType>(QT) && "Expected a TypedefType");
725  const TypedefNameDecl *Typedef = cast<TypedefType>(QT)->getDecl();
726
727  for (;;) {
728    const IdentifierInfo *Identifier = Typedef->getIdentifier();
729    if (Identifier->getName() == "size_t") {
730      LM.setKind(LengthModifier::AsSizeT);
731      return true;
732    } else if (Identifier->getName() == "ssize_t") {
733      // Not C99, but common in Unix.
734      LM.setKind(LengthModifier::AsSizeT);
735      return true;
736    } else if (Identifier->getName() == "intmax_t") {
737      LM.setKind(LengthModifier::AsIntMax);
738      return true;
739    } else if (Identifier->getName() == "uintmax_t") {
740      LM.setKind(LengthModifier::AsIntMax);
741      return true;
742    } else if (Identifier->getName() == "ptrdiff_t") {
743      LM.setKind(LengthModifier::AsPtrDiff);
744      return true;
745    }
746
747    QualType T = Typedef->getUnderlyingType();
748    if (!isa<TypedefType>(T))
749      break;
750
751    Typedef = cast<TypedefType>(T)->getDecl();
752  }
753  return false;
754}
755