FormatString.cpp revision 9dbe16eb808ed3b58be6be48bf4ae7317db63e89 
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') ?
187      ++I, LengthModifier::AsChar : LengthModifier::AsShort;
188      break;
189    case 'l':
190      ++I;
191      lmKind = (I != E && *I == 'l') ?
192      ++I, LengthModifier::AsLongLong : 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
209static bool hasSameSize(ASTContext &astContext, QualType typeA, QualType typeB) {
210  return astContext.getTypeSize(typeA) == astContext.getTypeSize(typeB);
211}
212
213bool ArgTypeResult::matchesType(ASTContext &C, QualType argTy) const {
214  switch (K) {
215    case InvalidTy:
216      assert(false && "ArgTypeResult must be valid");
217      return true;
218
219    case UnknownTy:
220      return true;
221
222    case SpecificTy: {
223      argTy = C.getCanonicalType(argTy).getUnqualifiedType();
224      if (T == argTy)
225        return true;
226      // Check for "compatible types".
227      if (const BuiltinType *BT = argTy->getAs<BuiltinType>()) {
228        if (!T->isIntegerType())
229          return false;
230        switch (BT->getKind()) {
231          default:
232            break;
233          case BuiltinType::Char_S:
234          case BuiltinType::SChar:
235          case BuiltinType::Char_U:
236          case BuiltinType::UChar:
237            return hasSameSize(C, T, C.UnsignedCharTy);
238          case BuiltinType::Short:
239          case BuiltinType::UShort:
240            return hasSameSize(C, T, C.ShortTy);
241          case BuiltinType::Int:
242          case BuiltinType::UInt:
243            return hasSameSize(C, T, C.IntTy);
244          case BuiltinType::Long:
245          case BuiltinType::ULong:
246            return hasSameSize(C, T, C.LongTy);
247          case BuiltinType::LongLong:
248          case BuiltinType::ULongLong:
249            return hasSameSize(C, T, C.LongLongTy);
250        }
251      }
252      return false;
253    }
254
255    case CStrTy: {
256      const PointerType *PT = argTy->getAs<PointerType>();
257      if (!PT)
258        return false;
259      QualType pointeeTy = PT->getPointeeType();
260      if (const BuiltinType *BT = pointeeTy->getAs<BuiltinType>())
261        switch (BT->getKind()) {
262          case BuiltinType::Void:
263          case BuiltinType::Char_U:
264          case BuiltinType::UChar:
265          case BuiltinType::Char_S:
266          case BuiltinType::SChar:
267            return true;
268          default:
269            break;
270        }
271
272      return false;
273    }
274
275    case WCStrTy: {
276      const PointerType *PT = argTy->getAs<PointerType>();
277      if (!PT)
278        return false;
279      QualType pointeeTy =
280        C.getCanonicalType(PT->getPointeeType()).getUnqualifiedType();
281      return pointeeTy == C.getWCharType();
282    }
283
284    case WIntTy: {
285      // Instead of doing a lookup for the definition of 'wint_t' (which
286      // is defined by the system headers) instead see if wchar_t and
287      // the argument type promote to the same type.
288      QualType PromoWChar =
289        C.getWCharType()->isPromotableIntegerType()
290          ? C.getPromotedIntegerType(C.getWCharType()) : C.getWCharType();
291      QualType PromoArg =
292        argTy->isPromotableIntegerType()
293          ? C.getPromotedIntegerType(argTy) : argTy;
294
295      PromoWChar = C.getCanonicalType(PromoWChar).getUnqualifiedType();
296      PromoArg = C.getCanonicalType(PromoArg).getUnqualifiedType();
297
298      return PromoWChar == PromoArg;
299    }
300
301    case CPointerTy:
302      return argTy->isPointerType() || argTy->isObjCObjectPointerType() ||
303        argTy->isNullPtrType();
304
305    case ObjCPointerTy:
306      return argTy->getAs<ObjCObjectPointerType>() != NULL;
307  }
308
309  // FIXME: Should be unreachable, but Clang is currently emitting
310  // a warning.
311  return false;
312}
313
314QualType ArgTypeResult::getRepresentativeType(ASTContext &C) const {
315  switch (K) {
316    case InvalidTy:
317      assert(false && "No representative type for Invalid ArgTypeResult");
318      // Fall-through.
319    case UnknownTy:
320      return QualType();
321    case SpecificTy:
322      return T;
323    case CStrTy:
324      return C.getPointerType(C.CharTy);
325    case WCStrTy:
326      return C.getPointerType(C.getWCharType());
327    case ObjCPointerTy:
328      return C.ObjCBuiltinIdTy;
329    case CPointerTy:
330      return C.VoidPtrTy;
331    case WIntTy: {
332      QualType WC = C.getWCharType();
333      return WC->isPromotableIntegerType() ? C.getPromotedIntegerType(WC) : WC;
334    }
335  }
336
337  // FIXME: Should be unreachable, but Clang is currently emitting
338  // a warning.
339  return QualType();
340}
341
342//===----------------------------------------------------------------------===//
343// Methods on OptionalAmount.
344//===----------------------------------------------------------------------===//
345
346ArgTypeResult
347analyze_format_string::OptionalAmount::getArgType(ASTContext &Ctx) const {
348  return Ctx.IntTy;
349}
350
351//===----------------------------------------------------------------------===//
352// Methods on LengthModifier.
353//===----------------------------------------------------------------------===//
354
355const char *
356analyze_format_string::LengthModifier::toString() const {
357  switch (kind) {
358  case AsChar:
359    return "hh";
360  case AsShort:
361    return "h";
362  case AsLong: // or AsWideChar
363    return "l";
364  case AsLongLong:
365    return "ll";
366  case AsIntMax:
367    return "j";
368  case AsSizeT:
369    return "z";
370  case AsPtrDiff:
371    return "t";
372  case AsLongDouble:
373    return "L";
374  case None:
375    return "";
376  }
377  return NULL;
378}
379
380//===----------------------------------------------------------------------===//
381// Methods on OptionalAmount.
382//===----------------------------------------------------------------------===//
383
384void OptionalAmount::toString(llvm::raw_ostream &os) const {
385  switch (hs) {
386  case Invalid:
387  case NotSpecified:
388    return;
389  case Arg:
390    if (UsesDotPrefix)
391        os << ".";
392    if (usesPositionalArg())
393      os << "*" << getPositionalArgIndex() << "$";
394    else
395      os << "*";
396    break;
397  case Constant:
398    if (UsesDotPrefix)
399        os << ".";
400    os << amt;
401    break;
402  }
403}
404
405//===----------------------------------------------------------------------===//
406// Methods on ConversionSpecifier.
407//===----------------------------------------------------------------------===//
408
409bool FormatSpecifier::hasValidLengthModifier() const {
410  switch (LM.getKind()) {
411    case LengthModifier::None:
412      return true;
413
414        // Handle most integer flags
415    case LengthModifier::AsChar:
416    case LengthModifier::AsShort:
417    case LengthModifier::AsLongLong:
418    case LengthModifier::AsIntMax:
419    case LengthModifier::AsSizeT:
420    case LengthModifier::AsPtrDiff:
421      switch (CS.getKind()) {
422        case ConversionSpecifier::dArg:
423        case ConversionSpecifier::iArg:
424        case ConversionSpecifier::oArg:
425        case ConversionSpecifier::uArg:
426        case ConversionSpecifier::xArg:
427        case ConversionSpecifier::XArg:
428        case ConversionSpecifier::nArg:
429          return true;
430        default:
431          return false;
432      }
433
434        // Handle 'l' flag
435    case LengthModifier::AsLong:
436      switch (CS.getKind()) {
437        case ConversionSpecifier::dArg:
438        case ConversionSpecifier::iArg:
439        case ConversionSpecifier::oArg:
440        case ConversionSpecifier::uArg:
441        case ConversionSpecifier::xArg:
442        case ConversionSpecifier::XArg:
443        case ConversionSpecifier::aArg:
444        case ConversionSpecifier::AArg:
445        case ConversionSpecifier::fArg:
446        case ConversionSpecifier::FArg:
447        case ConversionSpecifier::eArg:
448        case ConversionSpecifier::EArg:
449        case ConversionSpecifier::gArg:
450        case ConversionSpecifier::GArg:
451        case ConversionSpecifier::nArg:
452        case ConversionSpecifier::cArg:
453        case ConversionSpecifier::sArg:
454          return true;
455        default:
456          return false;
457      }
458
459    case LengthModifier::AsLongDouble:
460      switch (CS.getKind()) {
461        case ConversionSpecifier::aArg:
462        case ConversionSpecifier::AArg:
463        case ConversionSpecifier::fArg:
464        case ConversionSpecifier::FArg:
465        case ConversionSpecifier::eArg:
466        case ConversionSpecifier::EArg:
467        case ConversionSpecifier::gArg:
468        case ConversionSpecifier::GArg:
469          return true;
470        default:
471          return false;
472      }
473  }
474  return false;
475}
476
477
478