FormatString.cpp revision dc00d8158db573f4a1f91cfaa2a89171c2e5f637 
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
209bool ArgTypeResult::matchesType(ASTContext &C, QualType argTy) const {
210  switch (K) {
211    case InvalidTy:
212      assert(false && "ArgTypeResult must be valid");
213      return true;
214
215    case UnknownTy:
216      return true;
217
218    case SpecificTy: {
219      argTy = C.getCanonicalType(argTy).getUnqualifiedType();
220      if (T == argTy)
221        return true;
222      // Check for "compatible types".
223      if (const BuiltinType *BT = argTy->getAs<BuiltinType>())
224        switch (BT->getKind()) {
225          default:
226            break;
227          case BuiltinType::Char_S:
228          case BuiltinType::SChar:
229            return T == C.UnsignedCharTy;
230          case BuiltinType::Char_U:
231          case BuiltinType::UChar:
232            return T == C.SignedCharTy;
233          case BuiltinType::Short:
234            return T == C.UnsignedShortTy;
235          case BuiltinType::UShort:
236            return T == C.ShortTy;
237          case BuiltinType::Int:
238            return T == C.UnsignedIntTy;
239          case BuiltinType::UInt:
240            return T == C.IntTy;
241          case BuiltinType::Long:
242            return T == C.UnsignedLongTy;
243          case BuiltinType::ULong:
244            return T == C.LongTy;
245          case BuiltinType::LongLong:
246            return T == C.UnsignedLongLongTy;
247          case BuiltinType::ULongLong:
248            return T == C.LongLongTy;
249        }
250      return false;
251    }
252
253    case CStrTy: {
254      const PointerType *PT = argTy->getAs<PointerType>();
255      if (!PT)
256        return false;
257      QualType pointeeTy = PT->getPointeeType();
258      if (const BuiltinType *BT = pointeeTy->getAs<BuiltinType>())
259        switch (BT->getKind()) {
260          case BuiltinType::Void:
261          case BuiltinType::Char_U:
262          case BuiltinType::UChar:
263          case BuiltinType::Char_S:
264          case BuiltinType::SChar:
265            return true;
266          default:
267            break;
268        }
269
270      return false;
271    }
272
273    case WCStrTy: {
274      const PointerType *PT = argTy->getAs<PointerType>();
275      if (!PT)
276        return false;
277      QualType pointeeTy =
278        C.getCanonicalType(PT->getPointeeType()).getUnqualifiedType();
279      return pointeeTy == C.getWCharType();
280    }
281
282    case WIntTy: {
283      // Instead of doing a lookup for the definition of 'wint_t' (which
284      // is defined by the system headers) instead see if wchar_t and
285      // the argument type promote to the same type.
286      QualType PromoWChar =
287        C.getWCharType()->isPromotableIntegerType()
288          ? C.getPromotedIntegerType(C.getWCharType()) : C.getWCharType();
289      QualType PromoArg =
290        argTy->isPromotableIntegerType()
291          ? C.getPromotedIntegerType(argTy) : argTy;
292
293      PromoWChar = C.getCanonicalType(PromoWChar).getUnqualifiedType();
294      PromoArg = C.getCanonicalType(PromoArg).getUnqualifiedType();
295
296      return PromoWChar == PromoArg;
297    }
298
299    case CPointerTy:
300      return argTy->isPointerType() || argTy->isObjCObjectPointerType() ||
301        argTy->isNullPtrType();
302
303    case ObjCPointerTy:
304      return argTy->getAs<ObjCObjectPointerType>() != NULL;
305  }
306
307  // FIXME: Should be unreachable, but Clang is currently emitting
308  // a warning.
309  return false;
310}
311
312QualType ArgTypeResult::getRepresentativeType(ASTContext &C) const {
313  switch (K) {
314    case InvalidTy:
315      assert(false && "No representative type for Invalid ArgTypeResult");
316      // Fall-through.
317    case UnknownTy:
318      return QualType();
319    case SpecificTy:
320      return T;
321    case CStrTy:
322      return C.getPointerType(C.CharTy);
323    case WCStrTy:
324      return C.getPointerType(C.getWCharType());
325    case ObjCPointerTy:
326      return C.ObjCBuiltinIdTy;
327    case CPointerTy:
328      return C.VoidPtrTy;
329    case WIntTy: {
330      QualType WC = C.getWCharType();
331      return WC->isPromotableIntegerType() ? C.getPromotedIntegerType(WC) : WC;
332    }
333  }
334
335  // FIXME: Should be unreachable, but Clang is currently emitting
336  // a warning.
337  return QualType();
338}
339
340//===----------------------------------------------------------------------===//
341// Methods on OptionalAmount.
342//===----------------------------------------------------------------------===//
343
344ArgTypeResult
345analyze_format_string::OptionalAmount::getArgType(ASTContext &Ctx) const {
346  return Ctx.IntTy;
347}
348
349//===----------------------------------------------------------------------===//
350// Methods on LengthModifier.
351//===----------------------------------------------------------------------===//
352
353const char *
354analyze_format_string::LengthModifier::toString() const {
355  switch (kind) {
356  case AsChar:
357    return "hh";
358  case AsShort:
359    return "h";
360  case AsLong: // or AsWideChar
361    return "l";
362  case AsLongLong:
363    return "ll";
364  case AsIntMax:
365    return "j";
366  case AsSizeT:
367    return "z";
368  case AsPtrDiff:
369    return "t";
370  case AsLongDouble:
371    return "L";
372  case None:
373    return "";
374  }
375  return NULL;
376}
377
378//===----------------------------------------------------------------------===//
379// Methods on OptionalAmount.
380//===----------------------------------------------------------------------===//
381
382void OptionalAmount::toString(llvm::raw_ostream &os) const {
383  switch (hs) {
384  case Invalid:
385  case NotSpecified:
386    return;
387  case Arg:
388    if (UsesDotPrefix)
389        os << ".";
390    if (usesPositionalArg())
391      os << "*" << getPositionalArgIndex() << "$";
392    else
393      os << "*";
394    break;
395  case Constant:
396    if (UsesDotPrefix)
397        os << ".";
398    os << amt;
399    break;
400  }
401}
402
403//===----------------------------------------------------------------------===//
404// Methods on ConversionSpecifier.
405//===----------------------------------------------------------------------===//
406
407bool FormatSpecifier::hasValidLengthModifier() const {
408  switch (LM.getKind()) {
409    case LengthModifier::None:
410      return true;
411
412        // Handle most integer flags
413    case LengthModifier::AsChar:
414    case LengthModifier::AsShort:
415    case LengthModifier::AsLongLong:
416    case LengthModifier::AsIntMax:
417    case LengthModifier::AsSizeT:
418    case LengthModifier::AsPtrDiff:
419      switch (CS.getKind()) {
420        case ConversionSpecifier::dArg:
421        case ConversionSpecifier::iArg:
422        case ConversionSpecifier::oArg:
423        case ConversionSpecifier::uArg:
424        case ConversionSpecifier::xArg:
425        case ConversionSpecifier::XArg:
426        case ConversionSpecifier::nArg:
427          return true;
428        default:
429          return false;
430      }
431
432        // Handle 'l' flag
433    case LengthModifier::AsLong:
434      switch (CS.getKind()) {
435        case ConversionSpecifier::dArg:
436        case ConversionSpecifier::iArg:
437        case ConversionSpecifier::oArg:
438        case ConversionSpecifier::uArg:
439        case ConversionSpecifier::xArg:
440        case ConversionSpecifier::XArg:
441        case ConversionSpecifier::aArg:
442        case ConversionSpecifier::AArg:
443        case ConversionSpecifier::fArg:
444        case ConversionSpecifier::FArg:
445        case ConversionSpecifier::eArg:
446        case ConversionSpecifier::EArg:
447        case ConversionSpecifier::gArg:
448        case ConversionSpecifier::GArg:
449        case ConversionSpecifier::nArg:
450        case ConversionSpecifier::cArg:
451        case ConversionSpecifier::sArg:
452          return true;
453        default:
454          return false;
455      }
456
457    case LengthModifier::AsLongDouble:
458      switch (CS.getKind()) {
459        case ConversionSpecifier::aArg:
460        case ConversionSpecifier::AArg:
461        case ConversionSpecifier::fArg:
462        case ConversionSpecifier::FArg:
463        case ConversionSpecifier::eArg:
464        case ConversionSpecifier::EArg:
465        case ConversionSpecifier::gArg:
466        case ConversionSpecifier::GArg:
467          return true;
468        default:
469          return false;
470      }
471  }
472  return false;
473}
474
475
476