FormatString.cpp revision f4f0c6095d1f481b94c6821c65e3bf1c9df42af7 
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
17using clang::analyze_format_string::ArgTypeResult;
18using clang::analyze_format_string::FormatStringHandler;
19using clang::analyze_format_string::FormatSpecifier;
20using clang::analyze_format_string::LengthModifier;
21using clang::analyze_format_string::OptionalAmount;
22using clang::analyze_format_string::PositionContext;
23using clang::analyze_format_string::ConversionSpecifier;
24using namespace clang;
25
26// Key function to FormatStringHandler.
27FormatStringHandler::~FormatStringHandler() {}
28
29//===----------------------------------------------------------------------===//
30// Functions for parsing format strings components in both printf and
31// scanf format strings.
32//===----------------------------------------------------------------------===//
33
34OptionalAmount
35clang::analyze_format_string::ParseAmount(const char *&Beg, const char *E) {
36  const char *I = Beg;
37  UpdateOnReturn <const char*> UpdateBeg(Beg, I);
38
39  unsigned accumulator = 0;
40  bool hasDigits = false;
41
42  for ( ; I != E; ++I) {
43    char c = *I;
44    if (c >= '0' && c <= '9') {
45      hasDigits = true;
46      accumulator = (accumulator * 10) + (c - '0');
47      continue;
48    }
49
50    if (hasDigits)
51      return OptionalAmount(OptionalAmount::Constant, accumulator, Beg, I - Beg,
52          false);
53
54    break;
55  }
56
57  return OptionalAmount();
58}
59
60OptionalAmount
61clang::analyze_format_string::ParseNonPositionAmount(const char *&Beg,
62                                                     const char *E,
63                                                     unsigned &argIndex) {
64  if (*Beg == '*') {
65    ++Beg;
66    return OptionalAmount(OptionalAmount::Arg, argIndex++, Beg, 0, false);
67  }
68
69  return ParseAmount(Beg, E);
70}
71
72OptionalAmount
73clang::analyze_format_string::ParsePositionAmount(FormatStringHandler &H,
74                                                  const char *Start,
75                                                  const char *&Beg,
76                                                  const char *E,
77                                                  PositionContext p) {
78  if (*Beg == '*') {
79    const char *I = Beg + 1;
80    const OptionalAmount &Amt = ParseAmount(I, E);
81
82    if (Amt.getHowSpecified() == OptionalAmount::NotSpecified) {
83      H.HandleInvalidPosition(Beg, I - Beg, p);
84      return OptionalAmount(false);
85    }
86
87    if (I == E) {
88      // No more characters left?
89      H.HandleIncompleteSpecifier(Start, E - Start);
90      return OptionalAmount(false);
91    }
92
93    assert(Amt.getHowSpecified() == OptionalAmount::Constant);
94
95    if (*I == '$') {
96      // Handle positional arguments
97
98      // Special case: '*0$', since this is an easy mistake.
99      if (Amt.getConstantAmount() == 0) {
100        H.HandleZeroPosition(Beg, I - Beg + 1);
101        return OptionalAmount(false);
102      }
103
104      const char *Tmp = Beg;
105      Beg = ++I;
106
107      return OptionalAmount(OptionalAmount::Arg, Amt.getConstantAmount() - 1,
108                            Tmp, 0, true);
109    }
110
111    H.HandleInvalidPosition(Beg, I - Beg, p);
112    return OptionalAmount(false);
113  }
114
115  return ParseAmount(Beg, E);
116}
117
118
119bool
120clang::analyze_format_string::ParseFieldWidth(FormatStringHandler &H,
121                                              FormatSpecifier &CS,
122                                              const char *Start,
123                                              const char *&Beg, const char *E,
124                                              unsigned *argIndex) {
125  // FIXME: Support negative field widths.
126  if (argIndex) {
127    CS.setFieldWidth(ParseNonPositionAmount(Beg, E, *argIndex));
128  }
129  else {
130    const OptionalAmount Amt =
131      ParsePositionAmount(H, Start, Beg, E,
132                          analyze_format_string::FieldWidthPos);
133
134    if (Amt.isInvalid())
135      return true;
136    CS.setFieldWidth(Amt);
137  }
138  return false;
139}
140
141bool
142clang::analyze_format_string::ParseArgPosition(FormatStringHandler &H,
143                                               FormatSpecifier &FS,
144                                               const char *Start,
145                                               const char *&Beg,
146                                               const char *E) {
147  const char *I = Beg;
148
149  const OptionalAmount &Amt = ParseAmount(I, E);
150
151  if (I == E) {
152    // No more characters left?
153    H.HandleIncompleteSpecifier(Start, E - Start);
154    return true;
155  }
156
157  if (Amt.getHowSpecified() == OptionalAmount::Constant && *(I++) == '$') {
158    // Special case: '%0$', since this is an easy mistake.
159    if (Amt.getConstantAmount() == 0) {
160      H.HandleZeroPosition(Start, I - Start);
161      return true;
162    }
163
164    FS.setArgIndex(Amt.getConstantAmount() - 1);
165    FS.setUsesPositionalArg();
166    // Update the caller's pointer if we decided to consume
167    // these characters.
168    Beg = I;
169    return false;
170  }
171
172  return false;
173}
174
175bool
176clang::analyze_format_string::ParseLengthModifier(FormatSpecifier &FS,
177                                                  const char *&I,
178                                                  const char *E) {
179  LengthModifier::Kind lmKind = LengthModifier::None;
180  const char *lmPosition = I;
181  switch (*I) {
182    default:
183      return false;
184    case 'h':
185      ++I;
186      lmKind = (I != E && *I == 'h') ? (++I, LengthModifier::AsChar)
187                                     : LengthModifier::AsShort;
188      break;
189    case 'l':
190      ++I;
191      lmKind = (I != E && *I == 'l') ? (++I, LengthModifier::AsLongLong)
192                                     : LengthModifier::AsLong;
193      break;
194    case 'j': lmKind = LengthModifier::AsIntMax;     ++I; break;
195    case 'z': lmKind = LengthModifier::AsSizeT;      ++I; break;
196    case 't': lmKind = LengthModifier::AsPtrDiff;    ++I; break;
197    case 'L': lmKind = LengthModifier::AsLongDouble; ++I; break;
198    case 'q': lmKind = LengthModifier::AsLongLong;   ++I; break;
199  }
200  LengthModifier lm(lmPosition, lmKind);
201  FS.setLengthModifier(lm);
202  return true;
203}
204
205//===----------------------------------------------------------------------===//
206// Methods on ArgTypeResult.
207//===----------------------------------------------------------------------===//
208
209bool ArgTypeResult::matchesType(ASTContext &C, QualType argTy) const {
210  switch (K) {
211    case InvalidTy:
212      llvm_unreachable("ArgTypeResult must be valid");
213
214    case UnknownTy:
215      return true;
216
217    case AnyCharTy: {
218      if (const BuiltinType *BT = argTy->getAs<BuiltinType>())
219        switch (BT->getKind()) {
220          default:
221            break;
222          case BuiltinType::Char_S:
223          case BuiltinType::SChar:
224          case BuiltinType::UChar:
225          case BuiltinType::Char_U:
226            return true;
227        }
228      return false;
229    }
230
231    case SpecificTy: {
232      argTy = C.getCanonicalType(argTy).getUnqualifiedType();
233      if (T == argTy)
234        return true;
235      // Check for "compatible types".
236      if (const BuiltinType *BT = argTy->getAs<BuiltinType>())
237        switch (BT->getKind()) {
238          default:
239            break;
240          case BuiltinType::Char_S:
241          case BuiltinType::SChar:
242            return T == C.UnsignedCharTy;
243          case BuiltinType::Char_U:
244          case BuiltinType::UChar:
245            return T == C.SignedCharTy;
246          case BuiltinType::Short:
247            return T == C.UnsignedShortTy;
248          case BuiltinType::UShort:
249            return T == C.ShortTy;
250          case BuiltinType::Int:
251            return T == C.UnsignedIntTy;
252          case BuiltinType::UInt:
253            return T == C.IntTy;
254          case BuiltinType::Long:
255            return T == C.UnsignedLongTy;
256          case BuiltinType::ULong:
257            return T == C.LongTy;
258          case BuiltinType::LongLong:
259            return T == C.UnsignedLongLongTy;
260          case BuiltinType::ULongLong:
261            return T == C.LongLongTy;
262        }
263      return false;
264    }
265
266    case CStrTy: {
267      const PointerType *PT = argTy->getAs<PointerType>();
268      if (!PT)
269        return false;
270      QualType pointeeTy = PT->getPointeeType();
271      if (const BuiltinType *BT = pointeeTy->getAs<BuiltinType>())
272        switch (BT->getKind()) {
273          case BuiltinType::Void:
274          case BuiltinType::Char_U:
275          case BuiltinType::UChar:
276          case BuiltinType::Char_S:
277          case BuiltinType::SChar:
278            return true;
279          default:
280            break;
281        }
282
283      return false;
284    }
285
286    case WCStrTy: {
287      const PointerType *PT = argTy->getAs<PointerType>();
288      if (!PT)
289        return false;
290      QualType pointeeTy =
291        C.getCanonicalType(PT->getPointeeType()).getUnqualifiedType();
292      return pointeeTy == C.getWCharType();
293    }
294
295    case WIntTy: {
296      // Instead of doing a lookup for the definition of 'wint_t' (which
297      // is defined by the system headers) instead see if wchar_t and
298      // the argument type promote to the same type.
299      QualType PromoWChar =
300        C.getWCharType()->isPromotableIntegerType()
301          ? C.getPromotedIntegerType(C.getWCharType()) : C.getWCharType();
302      QualType PromoArg =
303        argTy->isPromotableIntegerType()
304          ? C.getPromotedIntegerType(argTy) : argTy;
305
306      PromoWChar = C.getCanonicalType(PromoWChar).getUnqualifiedType();
307      PromoArg = C.getCanonicalType(PromoArg).getUnqualifiedType();
308
309      return PromoWChar == PromoArg;
310    }
311
312    case CPointerTy:
313      return argTy->isPointerType() || argTy->isObjCObjectPointerType() ||
314        argTy->isNullPtrType();
315
316    case ObjCPointerTy:
317      return argTy->getAs<ObjCObjectPointerType>() != NULL;
318  }
319
320  // FIXME: Should be unreachable, but Clang is currently emitting
321  // a warning.
322  return false;
323}
324
325QualType ArgTypeResult::getRepresentativeType(ASTContext &C) const {
326  switch (K) {
327    case InvalidTy:
328      llvm_unreachable("No representative type for Invalid ArgTypeResult");
329    case UnknownTy:
330      return QualType();
331    case AnyCharTy:
332      return C.CharTy;
333    case SpecificTy:
334      return T;
335    case CStrTy:
336      return C.getPointerType(C.CharTy);
337    case WCStrTy:
338      return C.getPointerType(C.getWCharType());
339    case ObjCPointerTy:
340      return C.ObjCBuiltinIdTy;
341    case CPointerTy:
342      return C.VoidPtrTy;
343    case WIntTy: {
344      QualType WC = C.getWCharType();
345      return WC->isPromotableIntegerType() ? C.getPromotedIntegerType(WC) : WC;
346    }
347  }
348
349  // FIXME: Should be unreachable, but Clang is currently emitting
350  // a warning.
351  return QualType();
352}
353
354std::string ArgTypeResult::getRepresentativeTypeName(ASTContext &C) const {
355  std::string S = getRepresentativeType(C).getAsString();
356  if (Name)
357    return std::string("'") + Name + "' (aka '" + S + "')";
358  return std::string("'") + S + "'";
359}
360
361
362//===----------------------------------------------------------------------===//
363// Methods on OptionalAmount.
364//===----------------------------------------------------------------------===//
365
366ArgTypeResult
367analyze_format_string::OptionalAmount::getArgType(ASTContext &Ctx) const {
368  return Ctx.IntTy;
369}
370
371//===----------------------------------------------------------------------===//
372// Methods on LengthModifier.
373//===----------------------------------------------------------------------===//
374
375const char *
376analyze_format_string::LengthModifier::toString() const {
377  switch (kind) {
378  case AsChar:
379    return "hh";
380  case AsShort:
381    return "h";
382  case AsLong: // or AsWideChar
383    return "l";
384  case AsLongLong:
385    return "ll";
386  case AsIntMax:
387    return "j";
388  case AsSizeT:
389    return "z";
390  case AsPtrDiff:
391    return "t";
392  case AsLongDouble:
393    return "L";
394  case None:
395    return "";
396  }
397  return NULL;
398}
399
400//===----------------------------------------------------------------------===//
401// Methods on ConversionSpecifier.
402//===----------------------------------------------------------------------===//
403
404const char *ConversionSpecifier::toString() const {
405  switch (kind) {
406  case dArg: return "d";
407  case iArg: return "i";
408  case oArg: return "o";
409  case uArg: return "u";
410  case xArg: return "x";
411  case XArg: return "X";
412  case fArg: return "f";
413  case FArg: return "F";
414  case eArg: return "e";
415  case EArg: return "E";
416  case gArg: return "g";
417  case GArg: return "G";
418  case aArg: return "a";
419  case AArg: return "A";
420  case cArg: return "c";
421  case sArg: return "s";
422  case pArg: return "p";
423  case nArg: return "n";
424  case PercentArg:  return "%";
425  case ScanListArg: return "[";
426  case InvalidSpecifier: return NULL;
427
428  // MacOS X unicode extensions.
429  case CArg: return "C";
430  case SArg: return "S";
431
432  // Objective-C specific specifiers.
433  case ObjCObjArg: return "@";
434
435  // GlibC specific specifiers.
436  case PrintErrno: return "m";
437  }
438  return NULL;
439}
440
441//===----------------------------------------------------------------------===//
442// Methods on OptionalAmount.
443//===----------------------------------------------------------------------===//
444
445void OptionalAmount::toString(raw_ostream &os) const {
446  switch (hs) {
447  case Invalid:
448  case NotSpecified:
449    return;
450  case Arg:
451    if (UsesDotPrefix)
452        os << ".";
453    if (usesPositionalArg())
454      os << "*" << getPositionalArgIndex() << "$";
455    else
456      os << "*";
457    break;
458  case Constant:
459    if (UsesDotPrefix)
460        os << ".";
461    os << amt;
462    break;
463  }
464}
465
466bool FormatSpecifier::hasValidLengthModifier() const {
467  switch (LM.getKind()) {
468    case LengthModifier::None:
469      return true;
470
471        // Handle most integer flags
472    case LengthModifier::AsChar:
473    case LengthModifier::AsShort:
474    case LengthModifier::AsLongLong:
475    case LengthModifier::AsIntMax:
476    case LengthModifier::AsSizeT:
477    case LengthModifier::AsPtrDiff:
478      switch (CS.getKind()) {
479        case ConversionSpecifier::dArg:
480        case ConversionSpecifier::iArg:
481        case ConversionSpecifier::oArg:
482        case ConversionSpecifier::uArg:
483        case ConversionSpecifier::xArg:
484        case ConversionSpecifier::XArg:
485        case ConversionSpecifier::nArg:
486          return true;
487        default:
488          return false;
489      }
490
491        // Handle 'l' flag
492    case LengthModifier::AsLong:
493      switch (CS.getKind()) {
494        case ConversionSpecifier::dArg:
495        case ConversionSpecifier::iArg:
496        case ConversionSpecifier::oArg:
497        case ConversionSpecifier::uArg:
498        case ConversionSpecifier::xArg:
499        case ConversionSpecifier::XArg:
500        case ConversionSpecifier::aArg:
501        case ConversionSpecifier::AArg:
502        case ConversionSpecifier::fArg:
503        case ConversionSpecifier::FArg:
504        case ConversionSpecifier::eArg:
505        case ConversionSpecifier::EArg:
506        case ConversionSpecifier::gArg:
507        case ConversionSpecifier::GArg:
508        case ConversionSpecifier::nArg:
509        case ConversionSpecifier::cArg:
510        case ConversionSpecifier::sArg:
511          return true;
512        default:
513          return false;
514      }
515
516    case LengthModifier::AsLongDouble:
517      switch (CS.getKind()) {
518        case ConversionSpecifier::aArg:
519        case ConversionSpecifier::AArg:
520        case ConversionSpecifier::fArg:
521        case ConversionSpecifier::FArg:
522        case ConversionSpecifier::eArg:
523        case ConversionSpecifier::EArg:
524        case ConversionSpecifier::gArg:
525        case ConversionSpecifier::GArg:
526          return true;
527        default:
528          return false;
529      }
530  }
531  return false;
532}
533