xref: /freebsd/contrib/llvm-project/clang/lib/AST/FormatString.cpp (revision 0fca6ea1d4eea4c934cfff25ac9ee8ad6fe95583) 
1 // FormatString.cpp - Common stuff for handling printf/scanf formats -*- C++ -*-
2 //
3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4 // See https://llvm.org/LICENSE.txt for license information.
5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6 //
7 //===----------------------------------------------------------------------===//
8 //
9 // Shared details for processing format strings of printf and scanf
10 // (and friends).
11 //
12 //===----------------------------------------------------------------------===//
13 
14 #include "FormatStringParsing.h"
15 #include "clang/Basic/LangOptions.h"
16 #include "clang/Basic/TargetInfo.h"
17 #include "llvm/Support/ConvertUTF.h"
18 #include <optional>
19 
20 using clang::analyze_format_string::ArgType;
21 using clang::analyze_format_string::FormatStringHandler;
22 using clang::analyze_format_string::FormatSpecifier;
23 using clang::analyze_format_string::LengthModifier;
24 using clang::analyze_format_string::OptionalAmount;
25 using clang::analyze_format_string::ConversionSpecifier;
26 using namespace clang;
27 
28 // Key function to FormatStringHandler.
~FormatStringHandler()29 FormatStringHandler::~FormatStringHandler() {}
30 
31 //===----------------------------------------------------------------------===//
32 // Functions for parsing format strings components in both printf and
33 // scanf format strings.
34 //===----------------------------------------------------------------------===//
35 
36 OptionalAmount
ParseAmount(const char * & Beg,const char * E)37 clang::analyze_format_string::ParseAmount(const char *&Beg, const char *E) {
38   const char *I = Beg;
39   UpdateOnReturn <const char*> UpdateBeg(Beg, I);
40 
41   unsigned accumulator = 0;
42   bool hasDigits = false;
43 
44   for ( ; I != E; ++I) {
45     char c = *I;
46     if (c >= '0' && c <= '9') {
47       hasDigits = true;
48       accumulator = (accumulator * 10) + (c - '0');
49       continue;
50     }
51 
52     if (hasDigits)
53       return OptionalAmount(OptionalAmount::Constant, accumulator, Beg, I - Beg,
54           false);
55 
56     break;
57   }
58 
59   return OptionalAmount();
60 }
61 
62 OptionalAmount
ParseNonPositionAmount(const char * & Beg,const char * E,unsigned & argIndex)63 clang::analyze_format_string::ParseNonPositionAmount(const char *&Beg,
64                                                      const char *E,
65                                                      unsigned &argIndex) {
66   if (*Beg == '*') {
67     ++Beg;
68     return OptionalAmount(OptionalAmount::Arg, argIndex++, Beg, 0, false);
69   }
70 
71   return ParseAmount(Beg, E);
72 }
73 
74 OptionalAmount
ParsePositionAmount(FormatStringHandler & H,const char * Start,const char * & Beg,const char * E,PositionContext p)75 clang::analyze_format_string::ParsePositionAmount(FormatStringHandler &H,
76                                                   const char *Start,
77                                                   const char *&Beg,
78                                                   const char *E,
79                                                   PositionContext p) {
80   if (*Beg == '*') {
81     const char *I = Beg + 1;
82     const OptionalAmount &Amt = ParseAmount(I, E);
83 
84     if (Amt.getHowSpecified() == OptionalAmount::NotSpecified) {
85       H.HandleInvalidPosition(Beg, I - Beg, p);
86       return OptionalAmount(false);
87     }
88 
89     if (I == E) {
90       // No more characters left?
91       H.HandleIncompleteSpecifier(Start, E - Start);
92       return OptionalAmount(false);
93     }
94 
95     assert(Amt.getHowSpecified() == OptionalAmount::Constant);
96 
97     if (*I == '$') {
98       // Handle positional arguments
99 
100       // Special case: '*0$', since this is an easy mistake.
101       if (Amt.getConstantAmount() == 0) {
102         H.HandleZeroPosition(Beg, I - Beg + 1);
103         return OptionalAmount(false);
104       }
105 
106       const char *Tmp = Beg;
107       Beg = ++I;
108 
109       return OptionalAmount(OptionalAmount::Arg, Amt.getConstantAmount() - 1,
110                             Tmp, 0, true);
111     }
112 
113     H.HandleInvalidPosition(Beg, I - Beg, p);
114     return OptionalAmount(false);
115   }
116 
117   return ParseAmount(Beg, E);
118 }
119 
120 
121 bool
ParseFieldWidth(FormatStringHandler & H,FormatSpecifier & CS,const char * Start,const char * & Beg,const char * E,unsigned * argIndex)122 clang::analyze_format_string::ParseFieldWidth(FormatStringHandler &H,
123                                               FormatSpecifier &CS,
124                                               const char *Start,
125                                               const char *&Beg, const char *E,
126                                               unsigned *argIndex) {
127   // FIXME: Support negative field widths.
128   if (argIndex) {
129     CS.setFieldWidth(ParseNonPositionAmount(Beg, E, *argIndex));
130   }
131   else {
132     const OptionalAmount Amt =
133       ParsePositionAmount(H, Start, Beg, E,
134                           analyze_format_string::FieldWidthPos);
135 
136     if (Amt.isInvalid())
137       return true;
138     CS.setFieldWidth(Amt);
139   }
140   return false;
141 }
142 
143 bool
ParseArgPosition(FormatStringHandler & H,FormatSpecifier & FS,const char * Start,const char * & Beg,const char * E)144 clang::analyze_format_string::ParseArgPosition(FormatStringHandler &H,
145                                                FormatSpecifier &FS,
146                                                const char *Start,
147                                                const char *&Beg,
148                                                const char *E) {
149   const char *I = Beg;
150 
151   const OptionalAmount &Amt = ParseAmount(I, E);
152 
153   if (I == E) {
154     // No more characters left?
155     H.HandleIncompleteSpecifier(Start, E - Start);
156     return true;
157   }
158 
159   if (Amt.getHowSpecified() == OptionalAmount::Constant && *(I++) == '$') {
160     // Warn that positional arguments are non-standard.
161     H.HandlePosition(Start, I - Start);
162 
163     // Special case: '%0$', since this is an easy mistake.
164     if (Amt.getConstantAmount() == 0) {
165       H.HandleZeroPosition(Start, I - Start);
166       return true;
167     }
168 
169     FS.setArgIndex(Amt.getConstantAmount() - 1);
170     FS.setUsesPositionalArg();
171     // Update the caller's pointer if we decided to consume
172     // these characters.
173     Beg = I;
174     return false;
175   }
176 
177   return false;
178 }
179 
180 bool
ParseVectorModifier(FormatStringHandler & H,FormatSpecifier & FS,const char * & I,const char * E,const LangOptions & LO)181 clang::analyze_format_string::ParseVectorModifier(FormatStringHandler &H,
182                                                   FormatSpecifier &FS,
183                                                   const char *&I,
184                                                   const char *E,
185                                                   const LangOptions &LO) {
186   if (!LO.OpenCL)
187     return false;
188 
189   const char *Start = I;
190   if (*I == 'v') {
191     ++I;
192 
193     if (I == E) {
194       H.HandleIncompleteSpecifier(Start, E - Start);
195       return true;
196     }
197 
198     OptionalAmount NumElts = ParseAmount(I, E);
199     if (NumElts.getHowSpecified() != OptionalAmount::Constant) {
200       H.HandleIncompleteSpecifier(Start, E - Start);
201       return true;
202     }
203 
204     FS.setVectorNumElts(NumElts);
205   }
206 
207   return false;
208 }
209 
210 bool
ParseLengthModifier(FormatSpecifier & FS,const char * & I,const char * E,const LangOptions & LO,bool IsScanf)211 clang::analyze_format_string::ParseLengthModifier(FormatSpecifier &FS,
212                                                   const char *&I,
213                                                   const char *E,
214                                                   const LangOptions &LO,
215                                                   bool IsScanf) {
216   LengthModifier::Kind lmKind = LengthModifier::None;
217   const char *lmPosition = I;
218   switch (*I) {
219     default:
220       return false;
221     case 'h':
222       ++I;
223       if (I != E && *I == 'h') {
224         ++I;
225         lmKind = LengthModifier::AsChar;
226       } else if (I != E && *I == 'l' && LO.OpenCL) {
227         ++I;
228         lmKind = LengthModifier::AsShortLong;
229       } else {
230         lmKind = LengthModifier::AsShort;
231       }
232       break;
233     case 'l':
234       ++I;
235       if (I != E && *I == 'l') {
236         ++I;
237         lmKind = LengthModifier::AsLongLong;
238       } else {
239         lmKind = LengthModifier::AsLong;
240       }
241       break;
242     case 'j': lmKind = LengthModifier::AsIntMax;     ++I; break;
243     case 'z': lmKind = LengthModifier::AsSizeT;      ++I; break;
244     case 't': lmKind = LengthModifier::AsPtrDiff;    ++I; break;
245     case 'L': lmKind = LengthModifier::AsLongDouble; ++I; break;
246     case 'q': lmKind = LengthModifier::AsQuad;       ++I; break;
247     case 'a':
248       if (IsScanf && !LO.C99 && !LO.CPlusPlus11) {
249         // For scanf in C90, look at the next character to see if this should
250         // be parsed as the GNU extension 'a' length modifier. If not, this
251         // will be parsed as a conversion specifier.
252         ++I;
253         if (I != E && (*I == 's' || *I == 'S' || *I == '[')) {
254           lmKind = LengthModifier::AsAllocate;
255           break;
256         }
257         --I;
258       }
259       return false;
260     case 'm':
261       if (IsScanf) {
262         lmKind = LengthModifier::AsMAllocate;
263         ++I;
264         break;
265       }
266       return false;
267     // printf: AsInt64, AsInt32, AsInt3264
268     // scanf:  AsInt64
269     case 'I':
270       if (I + 1 != E && I + 2 != E) {
271         if (I[1] == '6' && I[2] == '4') {
272           I += 3;
273           lmKind = LengthModifier::AsInt64;
274           break;
275         }
276         if (IsScanf)
277           return false;
278 
279         if (I[1] == '3' && I[2] == '2') {
280           I += 3;
281           lmKind = LengthModifier::AsInt32;
282           break;
283         }
284       }
285       ++I;
286       lmKind = LengthModifier::AsInt3264;
287       break;
288     case 'w':
289       lmKind = LengthModifier::AsWide; ++I; break;
290   }
291   LengthModifier lm(lmPosition, lmKind);
292   FS.setLengthModifier(lm);
293   return true;
294 }
295 
ParseUTF8InvalidSpecifier(const char * SpecifierBegin,const char * FmtStrEnd,unsigned & Len)296 bool clang::analyze_format_string::ParseUTF8InvalidSpecifier(
297     const char *SpecifierBegin, const char *FmtStrEnd, unsigned &Len) {
298   if (SpecifierBegin + 1 >= FmtStrEnd)
299     return false;
300 
301   const llvm::UTF8 *SB =
302       reinterpret_cast<const llvm::UTF8 *>(SpecifierBegin + 1);
303   const llvm::UTF8 *SE = reinterpret_cast<const llvm::UTF8 *>(FmtStrEnd);
304   const char FirstByte = *SB;
305 
306   // If the invalid specifier is a multibyte UTF-8 string, return the
307   // total length accordingly so that the conversion specifier can be
308   // properly updated to reflect a complete UTF-8 specifier.
309   unsigned NumBytes = llvm::getNumBytesForUTF8(FirstByte);
310   if (NumBytes == 1)
311     return false;
312   if (SB + NumBytes > SE)
313     return false;
314 
315   Len = NumBytes + 1;
316   return true;
317 }
318 
319 //===----------------------------------------------------------------------===//
320 // Methods on ArgType.
321 //===----------------------------------------------------------------------===//
322 
323 clang::analyze_format_string::ArgType::MatchKind
matchesType(ASTContext & C,QualType argTy) const324 ArgType::matchesType(ASTContext &C, QualType argTy) const {
325   // When using the format attribute in C++, you can receive a function or an
326   // array that will necessarily decay to a pointer when passed to the final
327   // format consumer. Apply decay before type comparison.
328   if (argTy->canDecayToPointerType())
329     argTy = C.getDecayedType(argTy);
330 
331   if (Ptr) {
332     // It has to be a pointer.
333     const PointerType *PT = argTy->getAs<PointerType>();
334     if (!PT)
335       return NoMatch;
336 
337     // We cannot write through a const qualified pointer.
338     if (PT->getPointeeType().isConstQualified())
339       return NoMatch;
340 
341     argTy = PT->getPointeeType();
342   }
343 
344   switch (K) {
345     case InvalidTy:
346       llvm_unreachable("ArgType must be valid");
347 
348     case UnknownTy:
349       return Match;
350 
351     case AnyCharTy: {
352       if (const auto *ETy = argTy->getAs<EnumType>()) {
353         // If the enum is incomplete we know nothing about the underlying type.
354         // Assume that it's 'int'. Do not use the underlying type for a scoped
355         // enumeration.
356         if (!ETy->getDecl()->isComplete())
357           return NoMatch;
358         if (ETy->isUnscopedEnumerationType())
359           argTy = ETy->getDecl()->getIntegerType();
360       }
361 
362       if (const auto *BT = argTy->getAs<BuiltinType>()) {
363         // The types are perfectly matched?
364         switch (BT->getKind()) {
365         default:
366           break;
367         case BuiltinType::Char_S:
368         case BuiltinType::SChar:
369         case BuiltinType::UChar:
370         case BuiltinType::Char_U:
371             return Match;
372         case BuiltinType::Bool:
373           if (!Ptr)
374             return Match;
375           break;
376         }
377         // "Partially matched" because of promotions?
378         if (!Ptr) {
379           switch (BT->getKind()) {
380           default:
381             break;
382           case BuiltinType::Int:
383           case BuiltinType::UInt:
384             return MatchPromotion;
385           case BuiltinType::Short:
386           case BuiltinType::UShort:
387           case BuiltinType::WChar_S:
388           case BuiltinType::WChar_U:
389             return NoMatchPromotionTypeConfusion;
390           }
391         }
392       }
393       return NoMatch;
394     }
395 
396     case SpecificTy: {
397       if (const EnumType *ETy = argTy->getAs<EnumType>()) {
398         // If the enum is incomplete we know nothing about the underlying type.
399         // Assume that it's 'int'. Do not use the underlying type for a scoped
400         // enumeration as that needs an exact match.
401         if (!ETy->getDecl()->isComplete())
402           argTy = C.IntTy;
403         else if (ETy->isUnscopedEnumerationType())
404           argTy = ETy->getDecl()->getIntegerType();
405       }
406 
407       if (argTy->isSaturatedFixedPointType())
408         argTy = C.getCorrespondingUnsaturatedType(argTy);
409 
410       argTy = C.getCanonicalType(argTy).getUnqualifiedType();
411 
412       if (T == argTy)
413         return Match;
414       if (const auto *BT = argTy->getAs<BuiltinType>()) {
415         // Check if the only difference between them is signed vs unsigned
416         // if true, return match signedness.
417         switch (BT->getKind()) {
418           default:
419             break;
420           case BuiltinType::Bool:
421             if (Ptr && (T == C.UnsignedCharTy || T == C.SignedCharTy))
422               return NoMatch;
423             [[fallthrough]];
424           case BuiltinType::Char_S:
425           case BuiltinType::SChar:
426             if (T == C.UnsignedShortTy || T == C.ShortTy)
427               return NoMatchTypeConfusion;
428             if (T == C.UnsignedCharTy)
429               return NoMatchSignedness;
430             if (T == C.SignedCharTy)
431               return Match;
432             break;
433           case BuiltinType::Char_U:
434           case BuiltinType::UChar:
435             if (T == C.UnsignedShortTy || T == C.ShortTy)
436               return NoMatchTypeConfusion;
437             if (T == C.UnsignedCharTy)
438               return Match;
439             if (T == C.SignedCharTy)
440               return NoMatchSignedness;
441             break;
442           case BuiltinType::Short:
443             if (T == C.UnsignedShortTy)
444               return NoMatchSignedness;
445             break;
446           case BuiltinType::UShort:
447             if (T == C.ShortTy)
448               return NoMatchSignedness;
449             break;
450           case BuiltinType::Int:
451             if (T == C.UnsignedIntTy)
452               return NoMatchSignedness;
453             break;
454           case BuiltinType::UInt:
455             if (T == C.IntTy)
456               return NoMatchSignedness;
457             break;
458           case BuiltinType::Long:
459             if (T == C.UnsignedLongTy)
460               return NoMatchSignedness;
461             break;
462           case BuiltinType::ULong:
463             if (T == C.LongTy)
464               return NoMatchSignedness;
465             break;
466           case BuiltinType::LongLong:
467             if (T == C.UnsignedLongLongTy)
468               return NoMatchSignedness;
469             break;
470           case BuiltinType::ULongLong:
471             if (T == C.LongLongTy)
472               return NoMatchSignedness;
473             break;
474           }
475           // "Partially matched" because of promotions?
476           if (!Ptr) {
477             switch (BT->getKind()) {
478             default:
479               break;
480             case BuiltinType::Bool:
481               if (T == C.IntTy || T == C.UnsignedIntTy)
482                 return MatchPromotion;
483               break;
484             case BuiltinType::Int:
485             case BuiltinType::UInt:
486               if (T == C.SignedCharTy || T == C.UnsignedCharTy ||
487                   T == C.ShortTy || T == C.UnsignedShortTy || T == C.WCharTy ||
488                   T == C.WideCharTy)
489                 return MatchPromotion;
490               break;
491             case BuiltinType::Char_U:
492               if (T == C.UnsignedIntTy)
493                 return MatchPromotion;
494               if (T == C.UnsignedShortTy)
495                 return NoMatchPromotionTypeConfusion;
496               break;
497             case BuiltinType::Char_S:
498               if (T == C.IntTy)
499                 return MatchPromotion;
500               if (T == C.ShortTy)
501                 return NoMatchPromotionTypeConfusion;
502               break;
503             case BuiltinType::Half:
504             case BuiltinType::Float:
505               if (T == C.DoubleTy)
506                 return MatchPromotion;
507               break;
508             case BuiltinType::Short:
509             case BuiltinType::UShort:
510               if (T == C.SignedCharTy || T == C.UnsignedCharTy)
511                 return NoMatchPromotionTypeConfusion;
512               break;
513             case BuiltinType::WChar_U:
514             case BuiltinType::WChar_S:
515               if (T != C.WCharTy && T != C.WideCharTy)
516                 return NoMatchPromotionTypeConfusion;
517             }
518           }
519       }
520       return NoMatch;
521     }
522 
523     case CStrTy: {
524       const PointerType *PT = argTy->getAs<PointerType>();
525       if (!PT)
526         return NoMatch;
527       QualType pointeeTy = PT->getPointeeType();
528       if (const BuiltinType *BT = pointeeTy->getAs<BuiltinType>())
529         switch (BT->getKind()) {
530           case BuiltinType::Char_U:
531           case BuiltinType::UChar:
532           case BuiltinType::Char_S:
533           case BuiltinType::SChar:
534             return Match;
535           default:
536             break;
537         }
538 
539       return NoMatch;
540     }
541 
542     case WCStrTy: {
543       const PointerType *PT = argTy->getAs<PointerType>();
544       if (!PT)
545         return NoMatch;
546       QualType pointeeTy =
547         C.getCanonicalType(PT->getPointeeType()).getUnqualifiedType();
548       return pointeeTy == C.getWideCharType() ? Match : NoMatch;
549     }
550 
551     case WIntTy: {
552       QualType WInt = C.getCanonicalType(C.getWIntType()).getUnqualifiedType();
553 
554       if (C.getCanonicalType(argTy).getUnqualifiedType() == WInt)
555         return Match;
556 
557       QualType PromoArg = C.isPromotableIntegerType(argTy)
558                               ? C.getPromotedIntegerType(argTy)
559                               : argTy;
560       PromoArg = C.getCanonicalType(PromoArg).getUnqualifiedType();
561 
562       // If the promoted argument is the corresponding signed type of the
563       // wint_t type, then it should match.
564       if (PromoArg->hasSignedIntegerRepresentation() &&
565           C.getCorrespondingUnsignedType(PromoArg) == WInt)
566         return Match;
567 
568       return WInt == PromoArg ? Match : NoMatch;
569     }
570 
571     case CPointerTy:
572       if (argTy->isVoidPointerType()) {
573         return Match;
574       } if (argTy->isPointerType() || argTy->isObjCObjectPointerType() ||
575             argTy->isBlockPointerType() || argTy->isNullPtrType()) {
576         return NoMatchPedantic;
577       } else {
578         return NoMatch;
579       }
580 
581     case ObjCPointerTy: {
582       if (argTy->getAs<ObjCObjectPointerType>() ||
583           argTy->getAs<BlockPointerType>())
584         return Match;
585 
586       // Handle implicit toll-free bridging.
587       if (const PointerType *PT = argTy->getAs<PointerType>()) {
588         // Things such as CFTypeRef are really just opaque pointers
589         // to C structs representing CF types that can often be bridged
590         // to Objective-C objects.  Since the compiler doesn't know which
591         // structs can be toll-free bridged, we just accept them all.
592         QualType pointee = PT->getPointeeType();
593         if (pointee->getAsStructureType() || pointee->isVoidType())
594           return Match;
595       }
596       return NoMatch;
597     }
598   }
599 
600   llvm_unreachable("Invalid ArgType Kind!");
601 }
602 
makeVectorType(ASTContext & C,unsigned NumElts) const603 ArgType ArgType::makeVectorType(ASTContext &C, unsigned NumElts) const {
604   // Check for valid vector element types.
605   if (T.isNull())
606     return ArgType::Invalid();
607 
608   QualType Vec = C.getExtVectorType(T, NumElts);
609   return ArgType(Vec, Name);
610 }
611 
getRepresentativeType(ASTContext & C) const612 QualType ArgType::getRepresentativeType(ASTContext &C) const {
613   QualType Res;
614   switch (K) {
615     case InvalidTy:
616       llvm_unreachable("No representative type for Invalid ArgType");
617     case UnknownTy:
618       llvm_unreachable("No representative type for Unknown ArgType");
619     case AnyCharTy:
620       Res = C.CharTy;
621       break;
622     case SpecificTy:
623       Res = T;
624       break;
625     case CStrTy:
626       Res = C.getPointerType(C.CharTy);
627       break;
628     case WCStrTy:
629       Res = C.getPointerType(C.getWideCharType());
630       break;
631     case ObjCPointerTy:
632       Res = C.ObjCBuiltinIdTy;
633       break;
634     case CPointerTy:
635       Res = C.VoidPtrTy;
636       break;
637     case WIntTy: {
638       Res = C.getWIntType();
639       break;
640     }
641   }
642 
643   if (Ptr)
644     Res = C.getPointerType(Res);
645   return Res;
646 }
647 
getRepresentativeTypeName(ASTContext & C) const648 std::string ArgType::getRepresentativeTypeName(ASTContext &C) const {
649   std::string S = getRepresentativeType(C).getAsString(C.getPrintingPolicy());
650 
651   std::string Alias;
652   if (Name) {
653     // Use a specific name for this type, e.g. "size_t".
654     Alias = Name;
655     if (Ptr) {
656       // If ArgType is actually a pointer to T, append an asterisk.
657       Alias += (Alias[Alias.size()-1] == '*') ? "*" : " *";
658     }
659     // If Alias is the same as the underlying type, e.g. wchar_t, then drop it.
660     if (S == Alias)
661       Alias.clear();
662   }
663 
664   if (!Alias.empty())
665     return std::string("'") + Alias + "' (aka '" + S + "')";
666   return std::string("'") + S + "'";
667 }
668 
669 
670 //===----------------------------------------------------------------------===//
671 // Methods on OptionalAmount.
672 //===----------------------------------------------------------------------===//
673 
674 ArgType
getArgType(ASTContext & Ctx) const675 analyze_format_string::OptionalAmount::getArgType(ASTContext &Ctx) const {
676   return Ctx.IntTy;
677 }
678 
679 //===----------------------------------------------------------------------===//
680 // Methods on LengthModifier.
681 //===----------------------------------------------------------------------===//
682 
683 const char *
toString() const684 analyze_format_string::LengthModifier::toString() const {
685   switch (kind) {
686   case AsChar:
687     return "hh";
688   case AsShort:
689     return "h";
690   case AsShortLong:
691     return "hl";
692   case AsLong: // or AsWideChar
693     return "l";
694   case AsLongLong:
695     return "ll";
696   case AsQuad:
697     return "q";
698   case AsIntMax:
699     return "j";
700   case AsSizeT:
701     return "z";
702   case AsPtrDiff:
703     return "t";
704   case AsInt32:
705     return "I32";
706   case AsInt3264:
707     return "I";
708   case AsInt64:
709     return "I64";
710   case AsLongDouble:
711     return "L";
712   case AsAllocate:
713     return "a";
714   case AsMAllocate:
715     return "m";
716   case AsWide:
717     return "w";
718   case None:
719     return "";
720   }
721   return nullptr;
722 }
723 
724 //===----------------------------------------------------------------------===//
725 // Methods on ConversionSpecifier.
726 //===----------------------------------------------------------------------===//
727 
toString() const728 const char *ConversionSpecifier::toString() const {
729   switch (kind) {
730   case bArg: return "b";
731   case BArg: return "B";
732   case dArg: return "d";
733   case DArg: return "D";
734   case iArg: return "i";
735   case oArg: return "o";
736   case OArg: return "O";
737   case uArg: return "u";
738   case UArg: return "U";
739   case xArg: return "x";
740   case XArg: return "X";
741   case fArg: return "f";
742   case FArg: return "F";
743   case eArg: return "e";
744   case EArg: return "E";
745   case gArg: return "g";
746   case GArg: return "G";
747   case aArg: return "a";
748   case AArg: return "A";
749   case cArg: return "c";
750   case sArg: return "s";
751   case pArg: return "p";
752   case PArg:
753     return "P";
754   case nArg: return "n";
755   case PercentArg:  return "%";
756   case ScanListArg: return "[";
757   case InvalidSpecifier: return nullptr;
758 
759   // POSIX unicode extensions.
760   case CArg: return "C";
761   case SArg: return "S";
762 
763   // Objective-C specific specifiers.
764   case ObjCObjArg: return "@";
765 
766   // FreeBSD kernel specific specifiers.
767   case FreeBSDbArg: return "b";
768   case FreeBSDDArg: return "D";
769   case FreeBSDrArg: return "r";
770   case FreeBSDyArg: return "y";
771 
772   // GlibC specific specifiers.
773   case PrintErrno: return "m";
774 
775   // MS specific specifiers.
776   case ZArg: return "Z";
777 
778   // ISO/IEC TR 18037 (fixed-point) specific specifiers.
779   case rArg:
780     return "r";
781   case RArg:
782     return "R";
783   case kArg:
784     return "k";
785   case KArg:
786     return "K";
787   }
788   return nullptr;
789 }
790 
791 std::optional<ConversionSpecifier>
getStandardSpecifier() const792 ConversionSpecifier::getStandardSpecifier() const {
793   ConversionSpecifier::Kind NewKind;
794 
795   switch (getKind()) {
796   default:
797     return std::nullopt;
798   case DArg:
799     NewKind = dArg;
800     break;
801   case UArg:
802     NewKind = uArg;
803     break;
804   case OArg:
805     NewKind = oArg;
806     break;
807   }
808 
809   ConversionSpecifier FixedCS(*this);
810   FixedCS.setKind(NewKind);
811   return FixedCS;
812 }
813 
814 //===----------------------------------------------------------------------===//
815 // Methods on OptionalAmount.
816 //===----------------------------------------------------------------------===//
817 
toString(raw_ostream & os) const818 void OptionalAmount::toString(raw_ostream &os) const {
819   switch (hs) {
820   case Invalid:
821   case NotSpecified:
822     return;
823   case Arg:
824     if (UsesDotPrefix)
825         os << ".";
826     if (usesPositionalArg())
827       os << "*" << getPositionalArgIndex() << "$";
828     else
829       os << "*";
830     break;
831   case Constant:
832     if (UsesDotPrefix)
833         os << ".";
834     os << amt;
835     break;
836   }
837 }
838 
hasValidLengthModifier(const TargetInfo & Target,const LangOptions & LO) const839 bool FormatSpecifier::hasValidLengthModifier(const TargetInfo &Target,
840                                              const LangOptions &LO) const {
841   switch (LM.getKind()) {
842     case LengthModifier::None:
843       return true;
844 
845     // Handle most integer flags
846     case LengthModifier::AsShort:
847       // Length modifier only applies to FP vectors.
848       if (LO.OpenCL && CS.isDoubleArg())
849         return !VectorNumElts.isInvalid();
850 
851       if (CS.isFixedPointArg())
852         return true;
853 
854       if (Target.getTriple().isOSMSVCRT()) {
855         switch (CS.getKind()) {
856           case ConversionSpecifier::cArg:
857           case ConversionSpecifier::CArg:
858           case ConversionSpecifier::sArg:
859           case ConversionSpecifier::SArg:
860           case ConversionSpecifier::ZArg:
861             return true;
862           default:
863             break;
864         }
865       }
866       [[fallthrough]];
867     case LengthModifier::AsChar:
868     case LengthModifier::AsLongLong:
869     case LengthModifier::AsQuad:
870     case LengthModifier::AsIntMax:
871     case LengthModifier::AsSizeT:
872     case LengthModifier::AsPtrDiff:
873       switch (CS.getKind()) {
874         case ConversionSpecifier::bArg:
875         case ConversionSpecifier::BArg:
876         case ConversionSpecifier::dArg:
877         case ConversionSpecifier::DArg:
878         case ConversionSpecifier::iArg:
879         case ConversionSpecifier::oArg:
880         case ConversionSpecifier::OArg:
881         case ConversionSpecifier::uArg:
882         case ConversionSpecifier::UArg:
883         case ConversionSpecifier::xArg:
884         case ConversionSpecifier::XArg:
885         case ConversionSpecifier::nArg:
886           return true;
887         case ConversionSpecifier::FreeBSDrArg:
888         case ConversionSpecifier::FreeBSDyArg:
889           return Target.getTriple().isOSFreeBSD() || Target.getTriple().isPS();
890         default:
891           return false;
892       }
893 
894     case LengthModifier::AsShortLong:
895       return LO.OpenCL && !VectorNumElts.isInvalid();
896 
897     // Handle 'l' flag
898     case LengthModifier::AsLong: // or AsWideChar
899       if (CS.isDoubleArg()) {
900         // Invalid for OpenCL FP scalars.
901         if (LO.OpenCL && VectorNumElts.isInvalid())
902           return false;
903         return true;
904       }
905 
906       if (CS.isFixedPointArg())
907         return true;
908 
909       switch (CS.getKind()) {
910         case ConversionSpecifier::bArg:
911         case ConversionSpecifier::BArg:
912         case ConversionSpecifier::dArg:
913         case ConversionSpecifier::DArg:
914         case ConversionSpecifier::iArg:
915         case ConversionSpecifier::oArg:
916         case ConversionSpecifier::OArg:
917         case ConversionSpecifier::uArg:
918         case ConversionSpecifier::UArg:
919         case ConversionSpecifier::xArg:
920         case ConversionSpecifier::XArg:
921         case ConversionSpecifier::nArg:
922         case ConversionSpecifier::cArg:
923         case ConversionSpecifier::sArg:
924         case ConversionSpecifier::ScanListArg:
925         case ConversionSpecifier::ZArg:
926           return true;
927         case ConversionSpecifier::FreeBSDrArg:
928         case ConversionSpecifier::FreeBSDyArg:
929           return Target.getTriple().isOSFreeBSD() || Target.getTriple().isPS();
930         default:
931           return false;
932       }
933 
934     case LengthModifier::AsLongDouble:
935       switch (CS.getKind()) {
936         case ConversionSpecifier::aArg:
937         case ConversionSpecifier::AArg:
938         case ConversionSpecifier::fArg:
939         case ConversionSpecifier::FArg:
940         case ConversionSpecifier::eArg:
941         case ConversionSpecifier::EArg:
942         case ConversionSpecifier::gArg:
943         case ConversionSpecifier::GArg:
944           return true;
945         // GNU libc extension.
946         case ConversionSpecifier::dArg:
947         case ConversionSpecifier::iArg:
948         case ConversionSpecifier::oArg:
949         case ConversionSpecifier::uArg:
950         case ConversionSpecifier::xArg:
951         case ConversionSpecifier::XArg:
952           return !Target.getTriple().isOSDarwin() &&
953                  !Target.getTriple().isOSWindows();
954         default:
955           return false;
956       }
957 
958     case LengthModifier::AsAllocate:
959       switch (CS.getKind()) {
960         case ConversionSpecifier::sArg:
961         case ConversionSpecifier::SArg:
962         case ConversionSpecifier::ScanListArg:
963           return true;
964         default:
965           return false;
966       }
967 
968     case LengthModifier::AsMAllocate:
969       switch (CS.getKind()) {
970         case ConversionSpecifier::cArg:
971         case ConversionSpecifier::CArg:
972         case ConversionSpecifier::sArg:
973         case ConversionSpecifier::SArg:
974         case ConversionSpecifier::ScanListArg:
975           return true;
976         default:
977           return false;
978       }
979     case LengthModifier::AsInt32:
980     case LengthModifier::AsInt3264:
981     case LengthModifier::AsInt64:
982       switch (CS.getKind()) {
983         case ConversionSpecifier::dArg:
984         case ConversionSpecifier::iArg:
985         case ConversionSpecifier::oArg:
986         case ConversionSpecifier::uArg:
987         case ConversionSpecifier::xArg:
988         case ConversionSpecifier::XArg:
989           return Target.getTriple().isOSMSVCRT();
990         default:
991           return false;
992       }
993     case LengthModifier::AsWide:
994       switch (CS.getKind()) {
995         case ConversionSpecifier::cArg:
996         case ConversionSpecifier::CArg:
997         case ConversionSpecifier::sArg:
998         case ConversionSpecifier::SArg:
999         case ConversionSpecifier::ZArg:
1000           return Target.getTriple().isOSMSVCRT();
1001         default:
1002           return false;
1003       }
1004   }
1005   llvm_unreachable("Invalid LengthModifier Kind!");
1006 }
1007 
hasStandardLengthModifier() const1008 bool FormatSpecifier::hasStandardLengthModifier() const {
1009   switch (LM.getKind()) {
1010     case LengthModifier::None:
1011     case LengthModifier::AsChar:
1012     case LengthModifier::AsShort:
1013     case LengthModifier::AsLong:
1014     case LengthModifier::AsLongLong:
1015     case LengthModifier::AsIntMax:
1016     case LengthModifier::AsSizeT:
1017     case LengthModifier::AsPtrDiff:
1018     case LengthModifier::AsLongDouble:
1019       return true;
1020     case LengthModifier::AsAllocate:
1021     case LengthModifier::AsMAllocate:
1022     case LengthModifier::AsQuad:
1023     case LengthModifier::AsInt32:
1024     case LengthModifier::AsInt3264:
1025     case LengthModifier::AsInt64:
1026     case LengthModifier::AsWide:
1027     case LengthModifier::AsShortLong: // ???
1028       return false;
1029   }
1030   llvm_unreachable("Invalid LengthModifier Kind!");
1031 }
1032 
hasStandardConversionSpecifier(const LangOptions & LangOpt) const1033 bool FormatSpecifier::hasStandardConversionSpecifier(
1034     const LangOptions &LangOpt) const {
1035   switch (CS.getKind()) {
1036     case ConversionSpecifier::bArg:
1037     case ConversionSpecifier::BArg:
1038     case ConversionSpecifier::cArg:
1039     case ConversionSpecifier::dArg:
1040     case ConversionSpecifier::iArg:
1041     case ConversionSpecifier::oArg:
1042     case ConversionSpecifier::uArg:
1043     case ConversionSpecifier::xArg:
1044     case ConversionSpecifier::XArg:
1045     case ConversionSpecifier::fArg:
1046     case ConversionSpecifier::FArg:
1047     case ConversionSpecifier::eArg:
1048     case ConversionSpecifier::EArg:
1049     case ConversionSpecifier::gArg:
1050     case ConversionSpecifier::GArg:
1051     case ConversionSpecifier::aArg:
1052     case ConversionSpecifier::AArg:
1053     case ConversionSpecifier::sArg:
1054     case ConversionSpecifier::pArg:
1055     case ConversionSpecifier::nArg:
1056     case ConversionSpecifier::ObjCObjArg:
1057     case ConversionSpecifier::ScanListArg:
1058     case ConversionSpecifier::PercentArg:
1059     case ConversionSpecifier::PArg:
1060       return true;
1061     case ConversionSpecifier::CArg:
1062     case ConversionSpecifier::SArg:
1063       return LangOpt.ObjC;
1064     case ConversionSpecifier::InvalidSpecifier:
1065     case ConversionSpecifier::FreeBSDbArg:
1066     case ConversionSpecifier::FreeBSDDArg:
1067     case ConversionSpecifier::FreeBSDrArg:
1068     case ConversionSpecifier::FreeBSDyArg:
1069     case ConversionSpecifier::PrintErrno:
1070     case ConversionSpecifier::DArg:
1071     case ConversionSpecifier::OArg:
1072     case ConversionSpecifier::UArg:
1073     case ConversionSpecifier::ZArg:
1074       return false;
1075     case ConversionSpecifier::rArg:
1076     case ConversionSpecifier::RArg:
1077     case ConversionSpecifier::kArg:
1078     case ConversionSpecifier::KArg:
1079       return LangOpt.FixedPoint;
1080   }
1081   llvm_unreachable("Invalid ConversionSpecifier Kind!");
1082 }
1083 
hasStandardLengthConversionCombination() const1084 bool FormatSpecifier::hasStandardLengthConversionCombination() const {
1085   if (LM.getKind() == LengthModifier::AsLongDouble) {
1086     switch(CS.getKind()) {
1087         case ConversionSpecifier::dArg:
1088         case ConversionSpecifier::iArg:
1089         case ConversionSpecifier::oArg:
1090         case ConversionSpecifier::uArg:
1091         case ConversionSpecifier::xArg:
1092         case ConversionSpecifier::XArg:
1093           return false;
1094         default:
1095           return true;
1096     }
1097   }
1098   return true;
1099 }
1100 
1101 std::optional<LengthModifier>
getCorrectedLengthModifier() const1102 FormatSpecifier::getCorrectedLengthModifier() const {
1103   if (CS.isAnyIntArg() || CS.getKind() == ConversionSpecifier::nArg) {
1104     if (LM.getKind() == LengthModifier::AsLongDouble ||
1105         LM.getKind() == LengthModifier::AsQuad) {
1106       LengthModifier FixedLM(LM);
1107       FixedLM.setKind(LengthModifier::AsLongLong);
1108       return FixedLM;
1109     }
1110   }
1111 
1112   return std::nullopt;
1113 }
1114 
namedTypeToLengthModifier(QualType QT,LengthModifier & LM)1115 bool FormatSpecifier::namedTypeToLengthModifier(QualType QT,
1116                                                 LengthModifier &LM) {
1117   for (/**/; const auto *TT = QT->getAs<TypedefType>();
1118        QT = TT->getDecl()->getUnderlyingType()) {
1119     const TypedefNameDecl *Typedef = TT->getDecl();
1120     const IdentifierInfo *Identifier = Typedef->getIdentifier();
1121     if (Identifier->getName() == "size_t") {
1122       LM.setKind(LengthModifier::AsSizeT);
1123       return true;
1124     } else if (Identifier->getName() == "ssize_t") {
1125       // Not C99, but common in Unix.
1126       LM.setKind(LengthModifier::AsSizeT);
1127       return true;
1128     } else if (Identifier->getName() == "intmax_t") {
1129       LM.setKind(LengthModifier::AsIntMax);
1130       return true;
1131     } else if (Identifier->getName() == "uintmax_t") {
1132       LM.setKind(LengthModifier::AsIntMax);
1133       return true;
1134     } else if (Identifier->getName() == "ptrdiff_t") {
1135       LM.setKind(LengthModifier::AsPtrDiff);
1136       return true;
1137     }
1138   }
1139   return false;
1140 }
1141