FormatString.cpp revision eb2d1f1c88836bd5382e5d7aa8f6b85148a88b27 
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      llvm_unreachable("ArgTypeResult must be valid");
213
214    case UnknownTy:
215      return true;
216
217    case SpecificTy: {
218      argTy = C.getCanonicalType(argTy).getUnqualifiedType();
219      if (T == argTy)
220        return true;
221      // Check for "compatible types".
222      if (const BuiltinType *BT = argTy->getAs<BuiltinType>())
223        switch (BT->getKind()) {
224          default:
225            break;
226          case BuiltinType::Char_S:
227          case BuiltinType::SChar:
228            return T == C.UnsignedCharTy;
229          case BuiltinType::Char_U:
230          case BuiltinType::UChar:
231            return T == C.SignedCharTy;
232          case BuiltinType::Short:
233            return T == C.UnsignedShortTy;
234          case BuiltinType::UShort:
235            return T == C.ShortTy;
236          case BuiltinType::Int:
237            return T == C.UnsignedIntTy;
238          case BuiltinType::UInt:
239            return T == C.IntTy;
240          case BuiltinType::Long:
241            return T == C.UnsignedLongTy;
242          case BuiltinType::ULong:
243            return T == C.LongTy;
244          case BuiltinType::LongLong:
245            return T == C.UnsignedLongLongTy;
246          case BuiltinType::ULongLong:
247            return T == C.LongLongTy;
248        }
249      return false;
250    }
251
252    case CStrTy: {
253      const PointerType *PT = argTy->getAs<PointerType>();
254      if (!PT)
255        return false;
256      QualType pointeeTy = PT->getPointeeType();
257      if (const BuiltinType *BT = pointeeTy->getAs<BuiltinType>())
258        switch (BT->getKind()) {
259          case BuiltinType::Void:
260          case BuiltinType::Char_U:
261          case BuiltinType::UChar:
262          case BuiltinType::Char_S:
263          case BuiltinType::SChar:
264            return true;
265          default:
266            break;
267        }
268
269      return false;
270    }
271
272    case WCStrTy: {
273      const PointerType *PT = argTy->getAs<PointerType>();
274      if (!PT)
275        return false;
276      QualType pointeeTy =
277        C.getCanonicalType(PT->getPointeeType()).getUnqualifiedType();
278      return pointeeTy == C.getWCharType();
279    }
280
281    case WIntTy: {
282      // Instead of doing a lookup for the definition of 'wint_t' (which
283      // is defined by the system headers) instead see if wchar_t and
284      // the argument type promote to the same type.
285      QualType PromoWChar =
286        C.getWCharType()->isPromotableIntegerType()
287          ? C.getPromotedIntegerType(C.getWCharType()) : C.getWCharType();
288      QualType PromoArg =
289        argTy->isPromotableIntegerType()
290          ? C.getPromotedIntegerType(argTy) : argTy;
291
292      PromoWChar = C.getCanonicalType(PromoWChar).getUnqualifiedType();
293      PromoArg = C.getCanonicalType(PromoArg).getUnqualifiedType();
294
295      return PromoWChar == PromoArg;
296    }
297
298    case CPointerTy:
299      return argTy->isPointerType() || argTy->isObjCObjectPointerType() ||
300        argTy->isNullPtrType();
301
302    case ObjCPointerTy:
303      return argTy->getAs<ObjCObjectPointerType>() != NULL;
304  }
305
306  // FIXME: Should be unreachable, but Clang is currently emitting
307  // a warning.
308  return false;
309}
310
311QualType ArgTypeResult::getRepresentativeType(ASTContext &C) const {
312  switch (K) {
313    case InvalidTy:
314      llvm_unreachable("No representative type for Invalid ArgTypeResult");
315    case UnknownTy:
316      return QualType();
317    case SpecificTy:
318      return T;
319    case CStrTy:
320      return C.getPointerType(C.CharTy);
321    case WCStrTy:
322      return C.getPointerType(C.getWCharType());
323    case ObjCPointerTy:
324      return C.ObjCBuiltinIdTy;
325    case CPointerTy:
326      return C.VoidPtrTy;
327    case WIntTy: {
328      QualType WC = C.getWCharType();
329      return WC->isPromotableIntegerType() ? C.getPromotedIntegerType(WC) : WC;
330    }
331  }
332
333  // FIXME: Should be unreachable, but Clang is currently emitting
334  // a warning.
335  return QualType();
336}
337
338//===----------------------------------------------------------------------===//
339// Methods on OptionalAmount.
340//===----------------------------------------------------------------------===//
341
342ArgTypeResult
343analyze_format_string::OptionalAmount::getArgType(ASTContext &Ctx) const {
344  return Ctx.IntTy;
345}
346
347//===----------------------------------------------------------------------===//
348// Methods on LengthModifier.
349//===----------------------------------------------------------------------===//
350
351const char *
352analyze_format_string::LengthModifier::toString() const {
353  switch (kind) {
354  case AsChar:
355    return "hh";
356  case AsShort:
357    return "h";
358  case AsLong: // or AsWideChar
359    return "l";
360  case AsLongLong:
361    return "ll";
362  case AsIntMax:
363    return "j";
364  case AsSizeT:
365    return "z";
366  case AsPtrDiff:
367    return "t";
368  case AsLongDouble:
369    return "L";
370  case None:
371    return "";
372  }
373  return NULL;
374}
375
376//===----------------------------------------------------------------------===//
377// Methods on OptionalAmount.
378//===----------------------------------------------------------------------===//
379
380void OptionalAmount::toString(raw_ostream &os) const {
381  switch (hs) {
382  case Invalid:
383  case NotSpecified:
384    return;
385  case Arg:
386    if (UsesDotPrefix)
387        os << ".";
388    if (usesPositionalArg())
389      os << "*" << getPositionalArgIndex() << "$";
390    else
391      os << "*";
392    break;
393  case Constant:
394    if (UsesDotPrefix)
395        os << ".";
396    os << amt;
397    break;
398  }
399}
400
401//===----------------------------------------------------------------------===//
402// Methods on ConversionSpecifier.
403//===----------------------------------------------------------------------===//
404
405bool FormatSpecifier::hasValidLengthModifier() const {
406  switch (LM.getKind()) {
407    case LengthModifier::None:
408      return true;
409
410        // Handle most integer flags
411    case LengthModifier::AsChar:
412    case LengthModifier::AsShort:
413    case LengthModifier::AsLongLong:
414    case LengthModifier::AsIntMax:
415    case LengthModifier::AsSizeT:
416    case LengthModifier::AsPtrDiff:
417      switch (CS.getKind()) {
418        case ConversionSpecifier::dArg:
419        case ConversionSpecifier::iArg:
420        case ConversionSpecifier::oArg:
421        case ConversionSpecifier::uArg:
422        case ConversionSpecifier::xArg:
423        case ConversionSpecifier::XArg:
424        case ConversionSpecifier::nArg:
425          return true;
426        default:
427          return false;
428      }
429
430        // Handle 'l' flag
431    case LengthModifier::AsLong:
432      switch (CS.getKind()) {
433        case ConversionSpecifier::dArg:
434        case ConversionSpecifier::iArg:
435        case ConversionSpecifier::oArg:
436        case ConversionSpecifier::uArg:
437        case ConversionSpecifier::xArg:
438        case ConversionSpecifier::XArg:
439        case ConversionSpecifier::aArg:
440        case ConversionSpecifier::AArg:
441        case ConversionSpecifier::fArg:
442        case ConversionSpecifier::FArg:
443        case ConversionSpecifier::eArg:
444        case ConversionSpecifier::EArg:
445        case ConversionSpecifier::gArg:
446        case ConversionSpecifier::GArg:
447        case ConversionSpecifier::nArg:
448        case ConversionSpecifier::cArg:
449        case ConversionSpecifier::sArg:
450          return true;
451        default:
452          return false;
453      }
454
455    case LengthModifier::AsLongDouble:
456      switch (CS.getKind()) {
457        case ConversionSpecifier::aArg:
458        case ConversionSpecifier::AArg:
459        case ConversionSpecifier::fArg:
460        case ConversionSpecifier::FArg:
461        case ConversionSpecifier::eArg:
462        case ConversionSpecifier::EArg:
463        case ConversionSpecifier::gArg:
464        case ConversionSpecifier::GArg:
465          return true;
466        default:
467          return false;
468      }
469  }
470  return false;
471}
472
473
474