FormatString.cpp revision 28058d179ae40edc66135458849f1073c841bc74 
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++) == '$') {
159    // Special case: '%0$', since this is an easy mistake.
160    if (Amt.getConstantAmount() == 0) {
161      H.HandleZeroPosition(Start, I - Start);
162      return true;
163    }
164
165    FS.setArgIndex(Amt.getConstantAmount() - 1);
166    FS.setUsesPositionalArg();
167    // Update the caller's pointer if we decided to consume
168    // these characters.
169    Beg = I;
170    return false;
171  }
172
173  return false;
174}
175
176bool
177clang::analyze_format_string::ParseLengthModifier(FormatSpecifier &FS,
178                                                  const char *&I,
179                                                  const char *E,
180                                                  const LangOptions &LO,
181                                                  bool IsScanf) {
182  LengthModifier::Kind lmKind = LengthModifier::None;
183  const char *lmPosition = I;
184  switch (*I) {
185    default:
186      return false;
187    case 'h':
188      ++I;
189      lmKind = (I != E && *I == 'h') ? (++I, LengthModifier::AsChar)
190                                     : LengthModifier::AsShort;
191      break;
192    case 'l':
193      ++I;
194      lmKind = (I != E && *I == 'l') ? (++I, LengthModifier::AsLongLong)
195                                     : LengthModifier::AsLong;
196      break;
197    case 'j': lmKind = LengthModifier::AsIntMax;     ++I; break;
198    case 'z': lmKind = LengthModifier::AsSizeT;      ++I; break;
199    case 't': lmKind = LengthModifier::AsPtrDiff;    ++I; break;
200    case 'L': lmKind = LengthModifier::AsLongDouble; ++I; break;
201    case 'q': lmKind = LengthModifier::AsLongLong;   ++I; break;
202    case 'a':
203      if (IsScanf && !LO.C99 && !LO.CPlusPlus0x) {
204        // For scanf in C90, look at the next character to see if this should
205        // be parsed as the GNU extension 'a' length modifier. If not, this
206        // will be parsed as a conversion specifier.
207        ++I;
208        if (I != E && (*I == 's' || *I == 'S' || *I == '[')) {
209          lmKind = LengthModifier::AsAllocate;
210          break;
211        }
212        --I;
213      }
214      return false;
215  }
216  LengthModifier lm(lmPosition, lmKind);
217  FS.setLengthModifier(lm);
218  return true;
219}
220
221//===----------------------------------------------------------------------===//
222// Methods on ArgTypeResult.
223//===----------------------------------------------------------------------===//
224
225bool ArgTypeResult::matchesType(ASTContext &C, QualType argTy) const {
226  switch (K) {
227    case InvalidTy:
228      llvm_unreachable("ArgTypeResult must be valid");
229
230    case UnknownTy:
231      return true;
232
233    case AnyCharTy: {
234      if (const BuiltinType *BT = argTy->getAs<BuiltinType>())
235        switch (BT->getKind()) {
236          default:
237            break;
238          case BuiltinType::Char_S:
239          case BuiltinType::SChar:
240          case BuiltinType::UChar:
241          case BuiltinType::Char_U:
242            return true;
243        }
244      return false;
245    }
246
247    case SpecificTy: {
248      argTy = C.getCanonicalType(argTy).getUnqualifiedType();
249      if (T == argTy)
250        return true;
251      // Check for "compatible types".
252      if (const BuiltinType *BT = argTy->getAs<BuiltinType>())
253        switch (BT->getKind()) {
254          default:
255            break;
256          case BuiltinType::Char_S:
257          case BuiltinType::SChar:
258            return T == C.UnsignedCharTy;
259          case BuiltinType::Char_U:
260          case BuiltinType::UChar:
261            return T == C.SignedCharTy;
262          case BuiltinType::Short:
263            return T == C.UnsignedShortTy;
264          case BuiltinType::UShort:
265            return T == C.ShortTy;
266          case BuiltinType::Int:
267            return T == C.UnsignedIntTy;
268          case BuiltinType::UInt:
269            return T == C.IntTy;
270          case BuiltinType::Long:
271            return T == C.UnsignedLongTy;
272          case BuiltinType::ULong:
273            return T == C.LongTy;
274          case BuiltinType::LongLong:
275            return T == C.UnsignedLongLongTy;
276          case BuiltinType::ULongLong:
277            return T == C.LongLongTy;
278        }
279      return false;
280    }
281
282    case CStrTy: {
283      const PointerType *PT = argTy->getAs<PointerType>();
284      if (!PT)
285        return false;
286      QualType pointeeTy = PT->getPointeeType();
287      if (const BuiltinType *BT = pointeeTy->getAs<BuiltinType>())
288        switch (BT->getKind()) {
289          case BuiltinType::Void:
290          case BuiltinType::Char_U:
291          case BuiltinType::UChar:
292          case BuiltinType::Char_S:
293          case BuiltinType::SChar:
294            return true;
295          default:
296            break;
297        }
298
299      return false;
300    }
301
302    case WCStrTy: {
303      const PointerType *PT = argTy->getAs<PointerType>();
304      if (!PT)
305        return false;
306      QualType pointeeTy =
307        C.getCanonicalType(PT->getPointeeType()).getUnqualifiedType();
308      return pointeeTy == C.getWCharType();
309    }
310
311    case WIntTy: {
312      // Instead of doing a lookup for the definition of 'wint_t' (which
313      // is defined by the system headers) instead see if wchar_t and
314      // the argument type promote to the same type.
315      QualType PromoWChar =
316        C.getWCharType()->isPromotableIntegerType()
317          ? C.getPromotedIntegerType(C.getWCharType()) : C.getWCharType();
318      QualType PromoArg =
319        argTy->isPromotableIntegerType()
320          ? C.getPromotedIntegerType(argTy) : argTy;
321
322      PromoWChar = C.getCanonicalType(PromoWChar).getUnqualifiedType();
323      PromoArg = C.getCanonicalType(PromoArg).getUnqualifiedType();
324
325      return PromoWChar == PromoArg;
326    }
327
328    case CPointerTy:
329      return argTy->isPointerType() || argTy->isObjCObjectPointerType() ||
330        argTy->isNullPtrType();
331
332    case ObjCPointerTy:
333      return argTy->getAs<ObjCObjectPointerType>() != NULL;
334  }
335
336  // FIXME: Should be unreachable, but Clang is currently emitting
337  // a warning.
338  return false;
339}
340
341QualType ArgTypeResult::getRepresentativeType(ASTContext &C) const {
342  switch (K) {
343    case InvalidTy:
344      llvm_unreachable("No representative type for Invalid ArgTypeResult");
345    case UnknownTy:
346      return QualType();
347    case AnyCharTy:
348      return C.CharTy;
349    case SpecificTy:
350      return T;
351    case CStrTy:
352      return C.getPointerType(C.CharTy);
353    case WCStrTy:
354      return C.getPointerType(C.getWCharType());
355    case ObjCPointerTy:
356      return C.ObjCBuiltinIdTy;
357    case CPointerTy:
358      return C.VoidPtrTy;
359    case WIntTy: {
360      QualType WC = C.getWCharType();
361      return WC->isPromotableIntegerType() ? C.getPromotedIntegerType(WC) : WC;
362    }
363  }
364
365  // FIXME: Should be unreachable, but Clang is currently emitting
366  // a warning.
367  return QualType();
368}
369
370std::string ArgTypeResult::getRepresentativeTypeName(ASTContext &C) const {
371  std::string S = getRepresentativeType(C).getAsString();
372  if (Name)
373    return std::string("'") + Name + "' (aka '" + S + "')";
374  return std::string("'") + S + "'";
375}
376
377
378//===----------------------------------------------------------------------===//
379// Methods on OptionalAmount.
380//===----------------------------------------------------------------------===//
381
382ArgTypeResult
383analyze_format_string::OptionalAmount::getArgType(ASTContext &Ctx) const {
384  return Ctx.IntTy;
385}
386
387//===----------------------------------------------------------------------===//
388// Methods on LengthModifier.
389//===----------------------------------------------------------------------===//
390
391const char *
392analyze_format_string::LengthModifier::toString() const {
393  switch (kind) {
394  case AsChar:
395    return "hh";
396  case AsShort:
397    return "h";
398  case AsLong: // or AsWideChar
399    return "l";
400  case AsLongLong:
401    return "ll";
402  case AsIntMax:
403    return "j";
404  case AsSizeT:
405    return "z";
406  case AsPtrDiff:
407    return "t";
408  case AsLongDouble:
409    return "L";
410  case AsAllocate:
411    return "a";
412  case None:
413    return "";
414  }
415  return NULL;
416}
417
418//===----------------------------------------------------------------------===//
419// Methods on ConversionSpecifier.
420//===----------------------------------------------------------------------===//
421
422const char *ConversionSpecifier::toString() const {
423  switch (kind) {
424  case dArg: return "d";
425  case iArg: return "i";
426  case oArg: return "o";
427  case uArg: return "u";
428  case xArg: return "x";
429  case XArg: return "X";
430  case fArg: return "f";
431  case FArg: return "F";
432  case eArg: return "e";
433  case EArg: return "E";
434  case gArg: return "g";
435  case GArg: return "G";
436  case aArg: return "a";
437  case AArg: return "A";
438  case cArg: return "c";
439  case sArg: return "s";
440  case pArg: return "p";
441  case nArg: return "n";
442  case PercentArg:  return "%";
443  case ScanListArg: return "[";
444  case InvalidSpecifier: return NULL;
445
446  // MacOS X unicode extensions.
447  case CArg: return "C";
448  case SArg: return "S";
449
450  // Objective-C specific specifiers.
451  case ObjCObjArg: return "@";
452
453  // GlibC specific specifiers.
454  case PrintErrno: return "m";
455  }
456  return NULL;
457}
458
459//===----------------------------------------------------------------------===//
460// Methods on OptionalAmount.
461//===----------------------------------------------------------------------===//
462
463void OptionalAmount::toString(raw_ostream &os) const {
464  switch (hs) {
465  case Invalid:
466  case NotSpecified:
467    return;
468  case Arg:
469    if (UsesDotPrefix)
470        os << ".";
471    if (usesPositionalArg())
472      os << "*" << getPositionalArgIndex() << "$";
473    else
474      os << "*";
475    break;
476  case Constant:
477    if (UsesDotPrefix)
478        os << ".";
479    os << amt;
480    break;
481  }
482}
483
484bool FormatSpecifier::hasValidLengthModifier() const {
485  switch (LM.getKind()) {
486    case LengthModifier::None:
487      return true;
488
489        // Handle most integer flags
490    case LengthModifier::AsChar:
491    case LengthModifier::AsShort:
492    case LengthModifier::AsLongLong:
493    case LengthModifier::AsIntMax:
494    case LengthModifier::AsSizeT:
495    case LengthModifier::AsPtrDiff:
496      switch (CS.getKind()) {
497        case ConversionSpecifier::dArg:
498        case ConversionSpecifier::iArg:
499        case ConversionSpecifier::oArg:
500        case ConversionSpecifier::uArg:
501        case ConversionSpecifier::xArg:
502        case ConversionSpecifier::XArg:
503        case ConversionSpecifier::nArg:
504          return true;
505        default:
506          return false;
507      }
508
509        // Handle 'l' flag
510    case LengthModifier::AsLong:
511      switch (CS.getKind()) {
512        case ConversionSpecifier::dArg:
513        case ConversionSpecifier::iArg:
514        case ConversionSpecifier::oArg:
515        case ConversionSpecifier::uArg:
516        case ConversionSpecifier::xArg:
517        case ConversionSpecifier::XArg:
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        case ConversionSpecifier::nArg:
527        case ConversionSpecifier::cArg:
528        case ConversionSpecifier::sArg:
529          return true;
530        default:
531          return false;
532      }
533
534    case LengthModifier::AsLongDouble:
535      switch (CS.getKind()) {
536        case ConversionSpecifier::aArg:
537        case ConversionSpecifier::AArg:
538        case ConversionSpecifier::fArg:
539        case ConversionSpecifier::FArg:
540        case ConversionSpecifier::eArg:
541        case ConversionSpecifier::EArg:
542        case ConversionSpecifier::gArg:
543        case ConversionSpecifier::GArg:
544          return true;
545        default:
546          return false;
547      }
548
549    case LengthModifier::AsAllocate:
550      switch (CS.getKind()) {
551        case ConversionSpecifier::sArg:
552        case ConversionSpecifier::SArg:
553        case ConversionSpecifier::ScanListArg:
554          return true;
555        default:
556          return false;
557      }
558  }
559  return false;
560}
561