FormatString.cpp revision afcd1954be90965d832ae56f1413beae836c3346 
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            return T == C.UnsignedCharTy;
269          case BuiltinType::Char_U:
270          case BuiltinType::UChar:
271            return T == C.SignedCharTy;
272          case BuiltinType::Short:
273            return T == C.UnsignedShortTy;
274          case BuiltinType::UShort:
275            return T == C.ShortTy;
276          case BuiltinType::Int:
277            return T == C.UnsignedIntTy;
278          case BuiltinType::UInt:
279            return T == C.IntTy;
280          case BuiltinType::Long:
281            return T == C.UnsignedLongTy;
282          case BuiltinType::ULong:
283            return T == C.LongTy;
284          case BuiltinType::LongLong:
285            return T == C.UnsignedLongLongTy;
286          case BuiltinType::ULongLong:
287            return T == C.LongLongTy;
288        }
289      return false;
290    }
291
292    case CStrTy: {
293      const PointerType *PT = argTy->getAs<PointerType>();
294      if (!PT)
295        return false;
296      QualType pointeeTy = PT->getPointeeType();
297      if (const BuiltinType *BT = pointeeTy->getAs<BuiltinType>())
298        switch (BT->getKind()) {
299          case BuiltinType::Void:
300          case BuiltinType::Char_U:
301          case BuiltinType::UChar:
302          case BuiltinType::Char_S:
303          case BuiltinType::SChar:
304            return true;
305          default:
306            break;
307        }
308
309      return false;
310    }
311
312    case WCStrTy: {
313      const PointerType *PT = argTy->getAs<PointerType>();
314      if (!PT)
315        return false;
316      QualType pointeeTy =
317        C.getCanonicalType(PT->getPointeeType()).getUnqualifiedType();
318      return pointeeTy == C.getWCharType();
319    }
320
321    case WIntTy: {
322      // Instead of doing a lookup for the definition of 'wint_t' (which
323      // is defined by the system headers) instead see if wchar_t and
324      // the argument type promote to the same type.
325      QualType PromoWChar =
326        C.getWCharType()->isPromotableIntegerType()
327          ? C.getPromotedIntegerType(C.getWCharType()) : C.getWCharType();
328      QualType PromoArg =
329        argTy->isPromotableIntegerType()
330          ? C.getPromotedIntegerType(argTy) : argTy;
331
332      PromoWChar = C.getCanonicalType(PromoWChar).getUnqualifiedType();
333      PromoArg = C.getCanonicalType(PromoArg).getUnqualifiedType();
334
335      return PromoWChar == 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      QualType WC = C.getWCharType();
384      return WC->isPromotableIntegerType() ? C.getPromotedIntegerType(WC) : WC;
385    }
386  }
387
388  llvm_unreachable("Invalid ArgTypeResult Kind!");
389}
390
391std::string ArgTypeResult::getRepresentativeTypeName(ASTContext &C) const {
392  std::string S = getRepresentativeType(C).getAsString();
393  if (Name && S != Name)
394    return std::string("'") + Name + "' (aka '" + S + "')";
395  return std::string("'") + S + "'";
396}
397
398
399//===----------------------------------------------------------------------===//
400// Methods on OptionalAmount.
401//===----------------------------------------------------------------------===//
402
403ArgTypeResult
404analyze_format_string::OptionalAmount::getArgType(ASTContext &Ctx) const {
405  return Ctx.IntTy;
406}
407
408//===----------------------------------------------------------------------===//
409// Methods on LengthModifier.
410//===----------------------------------------------------------------------===//
411
412const char *
413analyze_format_string::LengthModifier::toString() const {
414  switch (kind) {
415  case AsChar:
416    return "hh";
417  case AsShort:
418    return "h";
419  case AsLong: // or AsWideChar
420    return "l";
421  case AsLongLong:
422    return "ll";
423  case AsQuad:
424    return "q";
425  case AsIntMax:
426    return "j";
427  case AsSizeT:
428    return "z";
429  case AsPtrDiff:
430    return "t";
431  case AsLongDouble:
432    return "L";
433  case AsAllocate:
434    return "a";
435  case AsMAllocate:
436    return "m";
437  case None:
438    return "";
439  }
440  return NULL;
441}
442
443//===----------------------------------------------------------------------===//
444// Methods on ConversionSpecifier.
445//===----------------------------------------------------------------------===//
446
447const char *ConversionSpecifier::toString() const {
448  switch (kind) {
449  case dArg: return "d";
450  case iArg: return "i";
451  case oArg: return "o";
452  case uArg: return "u";
453  case xArg: return "x";
454  case XArg: return "X";
455  case fArg: return "f";
456  case FArg: return "F";
457  case eArg: return "e";
458  case EArg: return "E";
459  case gArg: return "g";
460  case GArg: return "G";
461  case aArg: return "a";
462  case AArg: return "A";
463  case cArg: return "c";
464  case sArg: return "s";
465  case pArg: return "p";
466  case nArg: return "n";
467  case PercentArg:  return "%";
468  case ScanListArg: return "[";
469  case InvalidSpecifier: return NULL;
470
471  // MacOS X unicode extensions.
472  case CArg: return "C";
473  case SArg: return "S";
474
475  // Objective-C specific specifiers.
476  case ObjCObjArg: return "@";
477
478  // GlibC specific specifiers.
479  case PrintErrno: return "m";
480  }
481  return NULL;
482}
483
484//===----------------------------------------------------------------------===//
485// Methods on OptionalAmount.
486//===----------------------------------------------------------------------===//
487
488void OptionalAmount::toString(raw_ostream &os) const {
489  switch (hs) {
490  case Invalid:
491  case NotSpecified:
492    return;
493  case Arg:
494    if (UsesDotPrefix)
495        os << ".";
496    if (usesPositionalArg())
497      os << "*" << getPositionalArgIndex() << "$";
498    else
499      os << "*";
500    break;
501  case Constant:
502    if (UsesDotPrefix)
503        os << ".";
504    os << amt;
505    break;
506  }
507}
508
509bool FormatSpecifier::hasValidLengthModifier() const {
510  switch (LM.getKind()) {
511    case LengthModifier::None:
512      return true;
513
514    // Handle most integer flags
515    case LengthModifier::AsChar:
516    case LengthModifier::AsShort:
517    case LengthModifier::AsLongLong:
518    case LengthModifier::AsQuad:
519    case LengthModifier::AsIntMax:
520    case LengthModifier::AsSizeT:
521    case LengthModifier::AsPtrDiff:
522      switch (CS.getKind()) {
523        case ConversionSpecifier::dArg:
524        case ConversionSpecifier::iArg:
525        case ConversionSpecifier::oArg:
526        case ConversionSpecifier::uArg:
527        case ConversionSpecifier::xArg:
528        case ConversionSpecifier::XArg:
529        case ConversionSpecifier::nArg:
530          return true;
531        default:
532          return false;
533      }
534
535    // Handle 'l' flag
536    case LengthModifier::AsLong:
537      switch (CS.getKind()) {
538        case ConversionSpecifier::dArg:
539        case ConversionSpecifier::iArg:
540        case ConversionSpecifier::oArg:
541        case ConversionSpecifier::uArg:
542        case ConversionSpecifier::xArg:
543        case ConversionSpecifier::XArg:
544        case ConversionSpecifier::aArg:
545        case ConversionSpecifier::AArg:
546        case ConversionSpecifier::fArg:
547        case ConversionSpecifier::FArg:
548        case ConversionSpecifier::eArg:
549        case ConversionSpecifier::EArg:
550        case ConversionSpecifier::gArg:
551        case ConversionSpecifier::GArg:
552        case ConversionSpecifier::nArg:
553        case ConversionSpecifier::cArg:
554        case ConversionSpecifier::sArg:
555        case ConversionSpecifier::ScanListArg:
556          return true;
557        default:
558          return false;
559      }
560
561    case LengthModifier::AsLongDouble:
562      switch (CS.getKind()) {
563        case ConversionSpecifier::aArg:
564        case ConversionSpecifier::AArg:
565        case ConversionSpecifier::fArg:
566        case ConversionSpecifier::FArg:
567        case ConversionSpecifier::eArg:
568        case ConversionSpecifier::EArg:
569        case ConversionSpecifier::gArg:
570        case ConversionSpecifier::GArg:
571          return true;
572        // GNU extension.
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          return true;
580        default:
581          return false;
582      }
583
584    case LengthModifier::AsAllocate:
585      switch (CS.getKind()) {
586        case ConversionSpecifier::sArg:
587        case ConversionSpecifier::SArg:
588        case ConversionSpecifier::ScanListArg:
589          return true;
590        default:
591          return false;
592      }
593
594    case LengthModifier::AsMAllocate:
595      switch (CS.getKind()) {
596        case ConversionSpecifier::cArg:
597        case ConversionSpecifier::CArg:
598        case ConversionSpecifier::sArg:
599        case ConversionSpecifier::SArg:
600        case ConversionSpecifier::ScanListArg:
601          return true;
602        default:
603          return false;
604      }
605  }
606  llvm_unreachable("Invalid LengthModifier Kind!");
607}
608
609bool FormatSpecifier::hasStandardLengthModifier() const {
610  switch (LM.getKind()) {
611    case LengthModifier::None:
612    case LengthModifier::AsChar:
613    case LengthModifier::AsShort:
614    case LengthModifier::AsLong:
615    case LengthModifier::AsLongLong:
616    case LengthModifier::AsIntMax:
617    case LengthModifier::AsSizeT:
618    case LengthModifier::AsPtrDiff:
619    case LengthModifier::AsLongDouble:
620      return true;
621    case LengthModifier::AsAllocate:
622    case LengthModifier::AsMAllocate:
623    case LengthModifier::AsQuad:
624      return false;
625  }
626  llvm_unreachable("Invalid LengthModifier Kind!");
627}
628
629bool FormatSpecifier::hasStandardConversionSpecifier(const LangOptions &LangOpt) const {
630  switch (CS.getKind()) {
631    case ConversionSpecifier::cArg:
632    case ConversionSpecifier::dArg:
633    case ConversionSpecifier::iArg:
634    case ConversionSpecifier::oArg:
635    case ConversionSpecifier::uArg:
636    case ConversionSpecifier::xArg:
637    case ConversionSpecifier::XArg:
638    case ConversionSpecifier::fArg:
639    case ConversionSpecifier::FArg:
640    case ConversionSpecifier::eArg:
641    case ConversionSpecifier::EArg:
642    case ConversionSpecifier::gArg:
643    case ConversionSpecifier::GArg:
644    case ConversionSpecifier::aArg:
645    case ConversionSpecifier::AArg:
646    case ConversionSpecifier::sArg:
647    case ConversionSpecifier::pArg:
648    case ConversionSpecifier::nArg:
649    case ConversionSpecifier::ObjCObjArg:
650    case ConversionSpecifier::ScanListArg:
651    case ConversionSpecifier::PercentArg:
652      return true;
653    case ConversionSpecifier::CArg:
654    case ConversionSpecifier::SArg:
655      return LangOpt.ObjC1 || LangOpt.ObjC2;
656    case ConversionSpecifier::InvalidSpecifier:
657    case ConversionSpecifier::PrintErrno:
658      return false;
659  }
660  llvm_unreachable("Invalid ConversionSpecifier Kind!");
661}
662
663bool FormatSpecifier::hasStandardLengthConversionCombination() const {
664  if (LM.getKind() == LengthModifier::AsLongDouble) {
665    switch(CS.getKind()) {
666        case ConversionSpecifier::dArg:
667        case ConversionSpecifier::iArg:
668        case ConversionSpecifier::oArg:
669        case ConversionSpecifier::uArg:
670        case ConversionSpecifier::xArg:
671        case ConversionSpecifier::XArg:
672          return false;
673        default:
674          return true;
675    }
676  }
677  return true;
678}
679