FormatString.cpp revision 101d4e0c8ffbcdaaa58cddf1c20f98aa1b4501c4 
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::ArgTypeResult;
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 ArgTypeResult.
233//===----------------------------------------------------------------------===//
234
235bool ArgTypeResult::matchesType(ASTContext &C, QualType argTy) const {
236  switch (K) {
237    case InvalidTy:
238      llvm_unreachable("ArgTypeResult must be valid");
239
240    case UnknownTy:
241      return true;
242
243    case AnyCharTy: {
244      if (const BuiltinType *BT = argTy->getAs<BuiltinType>())
245        switch (BT->getKind()) {
246          default:
247            break;
248          case BuiltinType::Char_S:
249          case BuiltinType::SChar:
250          case BuiltinType::UChar:
251          case BuiltinType::Char_U:
252            return true;
253        }
254      return false;
255    }
256
257    case SpecificTy: {
258      argTy = C.getCanonicalType(argTy).getUnqualifiedType();
259      if (T == argTy)
260        return true;
261      // Check for "compatible types".
262      if (const BuiltinType *BT = argTy->getAs<BuiltinType>())
263        switch (BT->getKind()) {
264          default:
265            break;
266          case BuiltinType::Char_S:
267          case BuiltinType::SChar:
268          case BuiltinType::Char_U:
269          case BuiltinType::UChar:
270            return T == C.UnsignedCharTy || T == C.SignedCharTy;
271          case BuiltinType::Short:
272            return T == C.UnsignedShortTy;
273          case BuiltinType::UShort:
274            return T == C.ShortTy;
275          case BuiltinType::Int:
276            return T == C.UnsignedIntTy;
277          case BuiltinType::UInt:
278            return T == C.IntTy;
279          case BuiltinType::Long:
280            return T == C.UnsignedLongTy;
281          case BuiltinType::ULong:
282            return T == C.LongTy;
283          case BuiltinType::LongLong:
284            return T == C.UnsignedLongLongTy;
285          case BuiltinType::ULongLong:
286            return T == C.LongLongTy;
287        }
288      return false;
289    }
290
291    case CStrTy: {
292      const PointerType *PT = argTy->getAs<PointerType>();
293      if (!PT)
294        return false;
295      QualType pointeeTy = PT->getPointeeType();
296      if (const BuiltinType *BT = pointeeTy->getAs<BuiltinType>())
297        switch (BT->getKind()) {
298          case BuiltinType::Void:
299          case BuiltinType::Char_U:
300          case BuiltinType::UChar:
301          case BuiltinType::Char_S:
302          case BuiltinType::SChar:
303            return true;
304          default:
305            break;
306        }
307
308      return false;
309    }
310
311    case WCStrTy: {
312      const PointerType *PT = argTy->getAs<PointerType>();
313      if (!PT)
314        return false;
315      QualType pointeeTy =
316        C.getCanonicalType(PT->getPointeeType()).getUnqualifiedType();
317      return pointeeTy == C.getWCharType();
318    }
319
320    case WIntTy: {
321
322      QualType PromoArg =
323        argTy->isPromotableIntegerType()
324          ? C.getPromotedIntegerType(argTy) : argTy;
325
326      QualType WInt = C.getCanonicalType(C.getWIntType()).getUnqualifiedType();
327      PromoArg = C.getCanonicalType(PromoArg).getUnqualifiedType();
328
329      // If the promoted argument is the corresponding signed type of the
330      // wint_t type, then it should match.
331      if (PromoArg->hasSignedIntegerRepresentation() &&
332          C.getCorrespondingUnsignedType(PromoArg) == WInt)
333        return true;
334
335      return WInt == PromoArg;
336    }
337
338    case CPointerTy:
339      return argTy->isPointerType() || argTy->isObjCObjectPointerType() ||
340             argTy->isBlockPointerType() || argTy->isNullPtrType();
341
342    case ObjCPointerTy: {
343      if (argTy->getAs<ObjCObjectPointerType>() ||
344          argTy->getAs<BlockPointerType>())
345        return true;
346
347      // Handle implicit toll-free bridging.
348      if (const PointerType *PT = argTy->getAs<PointerType>()) {
349        // Things such as CFTypeRef are really just opaque pointers
350        // to C structs representing CF types that can often be bridged
351        // to Objective-C objects.  Since the compiler doesn't know which
352        // structs can be toll-free bridged, we just accept them all.
353        QualType pointee = PT->getPointeeType();
354        if (pointee->getAsStructureType() || pointee->isVoidType())
355          return true;
356      }
357      return false;
358    }
359  }
360
361  llvm_unreachable("Invalid ArgTypeResult Kind!");
362}
363
364QualType ArgTypeResult::getRepresentativeType(ASTContext &C) const {
365  switch (K) {
366    case InvalidTy:
367      llvm_unreachable("No representative type for Invalid ArgTypeResult");
368    case UnknownTy:
369      return QualType();
370    case AnyCharTy:
371      return C.CharTy;
372    case SpecificTy:
373      return T;
374    case CStrTy:
375      return C.getPointerType(C.CharTy);
376    case WCStrTy:
377      return C.getPointerType(C.getWCharType());
378    case ObjCPointerTy:
379      return C.ObjCBuiltinIdTy;
380    case CPointerTy:
381      return C.VoidPtrTy;
382    case WIntTy: {
383      return C.getWIntType();
384    }
385  }
386
387  llvm_unreachable("Invalid ArgTypeResult Kind!");
388}
389
390std::string ArgTypeResult::getRepresentativeTypeName(ASTContext &C) const {
391  std::string S = getRepresentativeType(C).getAsString();
392  if (Name && S != Name)
393    return std::string("'") + Name + "' (aka '" + S + "')";
394  return std::string("'") + S + "'";
395}
396
397
398//===----------------------------------------------------------------------===//
399// Methods on OptionalAmount.
400//===----------------------------------------------------------------------===//
401
402ArgTypeResult
403analyze_format_string::OptionalAmount::getArgType(ASTContext &Ctx) const {
404  return Ctx.IntTy;
405}
406
407//===----------------------------------------------------------------------===//
408// Methods on LengthModifier.
409//===----------------------------------------------------------------------===//
410
411const char *
412analyze_format_string::LengthModifier::toString() const {
413  switch (kind) {
414  case AsChar:
415    return "hh";
416  case AsShort:
417    return "h";
418  case AsLong: // or AsWideChar
419    return "l";
420  case AsLongLong:
421    return "ll";
422  case AsQuad:
423    return "q";
424  case AsIntMax:
425    return "j";
426  case AsSizeT:
427    return "z";
428  case AsPtrDiff:
429    return "t";
430  case AsLongDouble:
431    return "L";
432  case AsAllocate:
433    return "a";
434  case AsMAllocate:
435    return "m";
436  case None:
437    return "";
438  }
439  return NULL;
440}
441
442//===----------------------------------------------------------------------===//
443// Methods on ConversionSpecifier.
444//===----------------------------------------------------------------------===//
445
446const char *ConversionSpecifier::toString() const {
447  switch (kind) {
448  case dArg: return "d";
449  case iArg: return "i";
450  case oArg: return "o";
451  case uArg: return "u";
452  case xArg: return "x";
453  case XArg: return "X";
454  case fArg: return "f";
455  case FArg: return "F";
456  case eArg: return "e";
457  case EArg: return "E";
458  case gArg: return "g";
459  case GArg: return "G";
460  case aArg: return "a";
461  case AArg: return "A";
462  case cArg: return "c";
463  case sArg: return "s";
464  case pArg: return "p";
465  case nArg: return "n";
466  case PercentArg:  return "%";
467  case ScanListArg: return "[";
468  case InvalidSpecifier: return NULL;
469
470  // MacOS X unicode extensions.
471  case CArg: return "C";
472  case SArg: return "S";
473
474  // Objective-C specific specifiers.
475  case ObjCObjArg: return "@";
476
477  // GlibC specific specifiers.
478  case PrintErrno: return "m";
479  }
480  return NULL;
481}
482
483//===----------------------------------------------------------------------===//
484// Methods on OptionalAmount.
485//===----------------------------------------------------------------------===//
486
487void OptionalAmount::toString(raw_ostream &os) const {
488  switch (hs) {
489  case Invalid:
490  case NotSpecified:
491    return;
492  case Arg:
493    if (UsesDotPrefix)
494        os << ".";
495    if (usesPositionalArg())
496      os << "*" << getPositionalArgIndex() << "$";
497    else
498      os << "*";
499    break;
500  case Constant:
501    if (UsesDotPrefix)
502        os << ".";
503    os << amt;
504    break;
505  }
506}
507
508bool FormatSpecifier::hasValidLengthModifier() const {
509  switch (LM.getKind()) {
510    case LengthModifier::None:
511      return true;
512
513    // Handle most integer flags
514    case LengthModifier::AsChar:
515    case LengthModifier::AsShort:
516    case LengthModifier::AsLongLong:
517    case LengthModifier::AsQuad:
518    case LengthModifier::AsIntMax:
519    case LengthModifier::AsSizeT:
520    case LengthModifier::AsPtrDiff:
521      switch (CS.getKind()) {
522        case ConversionSpecifier::dArg:
523        case ConversionSpecifier::iArg:
524        case ConversionSpecifier::oArg:
525        case ConversionSpecifier::uArg:
526        case ConversionSpecifier::xArg:
527        case ConversionSpecifier::XArg:
528        case ConversionSpecifier::nArg:
529          return true;
530        default:
531          return false;
532      }
533
534    // Handle 'l' flag
535    case LengthModifier::AsLong:
536      switch (CS.getKind()) {
537        case ConversionSpecifier::dArg:
538        case ConversionSpecifier::iArg:
539        case ConversionSpecifier::oArg:
540        case ConversionSpecifier::uArg:
541        case ConversionSpecifier::xArg:
542        case ConversionSpecifier::XArg:
543        case ConversionSpecifier::aArg:
544        case ConversionSpecifier::AArg:
545        case ConversionSpecifier::fArg:
546        case ConversionSpecifier::FArg:
547        case ConversionSpecifier::eArg:
548        case ConversionSpecifier::EArg:
549        case ConversionSpecifier::gArg:
550        case ConversionSpecifier::GArg:
551        case ConversionSpecifier::nArg:
552        case ConversionSpecifier::cArg:
553        case ConversionSpecifier::sArg:
554        case ConversionSpecifier::ScanListArg:
555          return true;
556        default:
557          return false;
558      }
559
560    case LengthModifier::AsLongDouble:
561      switch (CS.getKind()) {
562        case ConversionSpecifier::aArg:
563        case ConversionSpecifier::AArg:
564        case ConversionSpecifier::fArg:
565        case ConversionSpecifier::FArg:
566        case ConversionSpecifier::eArg:
567        case ConversionSpecifier::EArg:
568        case ConversionSpecifier::gArg:
569        case ConversionSpecifier::GArg:
570          return true;
571        // GNU extension.
572        case ConversionSpecifier::dArg:
573        case ConversionSpecifier::iArg:
574        case ConversionSpecifier::oArg:
575        case ConversionSpecifier::uArg:
576        case ConversionSpecifier::xArg:
577        case ConversionSpecifier::XArg:
578          return true;
579        default:
580          return false;
581      }
582
583    case LengthModifier::AsAllocate:
584      switch (CS.getKind()) {
585        case ConversionSpecifier::sArg:
586        case ConversionSpecifier::SArg:
587        case ConversionSpecifier::ScanListArg:
588          return true;
589        default:
590          return false;
591      }
592
593    case LengthModifier::AsMAllocate:
594      switch (CS.getKind()) {
595        case ConversionSpecifier::cArg:
596        case ConversionSpecifier::CArg:
597        case ConversionSpecifier::sArg:
598        case ConversionSpecifier::SArg:
599        case ConversionSpecifier::ScanListArg:
600          return true;
601        default:
602          return false;
603      }
604  }
605  llvm_unreachable("Invalid LengthModifier Kind!");
606}
607
608bool FormatSpecifier::hasStandardLengthModifier() const {
609  switch (LM.getKind()) {
610    case LengthModifier::None:
611    case LengthModifier::AsChar:
612    case LengthModifier::AsShort:
613    case LengthModifier::AsLong:
614    case LengthModifier::AsLongLong:
615    case LengthModifier::AsIntMax:
616    case LengthModifier::AsSizeT:
617    case LengthModifier::AsPtrDiff:
618    case LengthModifier::AsLongDouble:
619      return true;
620    case LengthModifier::AsAllocate:
621    case LengthModifier::AsMAllocate:
622    case LengthModifier::AsQuad:
623      return false;
624  }
625  llvm_unreachable("Invalid LengthModifier Kind!");
626}
627
628bool FormatSpecifier::hasStandardConversionSpecifier(const LangOptions &LangOpt) const {
629  switch (CS.getKind()) {
630    case ConversionSpecifier::cArg:
631    case ConversionSpecifier::dArg:
632    case ConversionSpecifier::iArg:
633    case ConversionSpecifier::oArg:
634    case ConversionSpecifier::uArg:
635    case ConversionSpecifier::xArg:
636    case ConversionSpecifier::XArg:
637    case ConversionSpecifier::fArg:
638    case ConversionSpecifier::FArg:
639    case ConversionSpecifier::eArg:
640    case ConversionSpecifier::EArg:
641    case ConversionSpecifier::gArg:
642    case ConversionSpecifier::GArg:
643    case ConversionSpecifier::aArg:
644    case ConversionSpecifier::AArg:
645    case ConversionSpecifier::sArg:
646    case ConversionSpecifier::pArg:
647    case ConversionSpecifier::nArg:
648    case ConversionSpecifier::ObjCObjArg:
649    case ConversionSpecifier::ScanListArg:
650    case ConversionSpecifier::PercentArg:
651      return true;
652    case ConversionSpecifier::CArg:
653    case ConversionSpecifier::SArg:
654      return LangOpt.ObjC1 || LangOpt.ObjC2;
655    case ConversionSpecifier::InvalidSpecifier:
656    case ConversionSpecifier::PrintErrno:
657      return false;
658  }
659  llvm_unreachable("Invalid ConversionSpecifier Kind!");
660}
661
662bool FormatSpecifier::hasStandardLengthConversionCombination() const {
663  if (LM.getKind() == LengthModifier::AsLongDouble) {
664    switch(CS.getKind()) {
665        case ConversionSpecifier::dArg:
666        case ConversionSpecifier::iArg:
667        case ConversionSpecifier::oArg:
668        case ConversionSpecifier::uArg:
669        case ConversionSpecifier::xArg:
670        case ConversionSpecifier::XArg:
671          return false;
672        default:
673          return true;
674    }
675  }
676  return true;
677}
678