xref: /freebsd-src/contrib/llvm-project/clang/lib/AST/FormatString.cpp (revision cb14a3fe5122c879eae1fb480ed7ce82a699ddb6) 
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.
29 FormatStringHandler::~FormatStringHandler() {}
30 
31 //===----------------------------------------------------------------------===//
32 // Functions for parsing format strings components in both printf and
33 // scanf format strings.
34 //===----------------------------------------------------------------------===//
35 
36 OptionalAmount
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
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
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
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
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
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
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 
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
324 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       argTy = C.getCanonicalType(argTy).getUnqualifiedType();
407 
408       if (T == argTy)
409         return Match;
410       if (const auto *BT = argTy->getAs<BuiltinType>()) {
411         // Check if the only difference between them is signed vs unsigned
412         // if true, we consider they are compatible.
413         switch (BT->getKind()) {
414           default:
415             break;
416           case BuiltinType::Bool:
417             if (Ptr && (T == C.UnsignedCharTy || T == C.SignedCharTy))
418               return NoMatch;
419             [[fallthrough]];
420           case BuiltinType::Char_S:
421           case BuiltinType::SChar:
422           case BuiltinType::Char_U:
423           case BuiltinType::UChar:
424             if (T == C.UnsignedShortTy || T == C.ShortTy)
425               return NoMatchTypeConfusion;
426             if (T == C.UnsignedCharTy || T == C.SignedCharTy)
427               return Match;
428             break;
429           case BuiltinType::Short:
430             if (T == C.UnsignedShortTy)
431               return Match;
432             break;
433           case BuiltinType::UShort:
434             if (T == C.ShortTy)
435               return Match;
436             break;
437           case BuiltinType::Int:
438             if (T == C.UnsignedIntTy)
439               return Match;
440             break;
441           case BuiltinType::UInt:
442             if (T == C.IntTy)
443               return Match;
444             break;
445           case BuiltinType::Long:
446             if (T == C.UnsignedLongTy)
447               return Match;
448             break;
449           case BuiltinType::ULong:
450             if (T == C.LongTy)
451               return Match;
452             break;
453           case BuiltinType::LongLong:
454             if (T == C.UnsignedLongLongTy)
455               return Match;
456             break;
457           case BuiltinType::ULongLong:
458             if (T == C.LongLongTy)
459               return Match;
460             break;
461           }
462           // "Partially matched" because of promotions?
463           if (!Ptr) {
464             switch (BT->getKind()) {
465             default:
466               break;
467             case BuiltinType::Bool:
468               if (T == C.IntTy || T == C.UnsignedIntTy)
469                 return MatchPromotion;
470               break;
471             case BuiltinType::Int:
472             case BuiltinType::UInt:
473               if (T == C.SignedCharTy || T == C.UnsignedCharTy ||
474                   T == C.ShortTy || T == C.UnsignedShortTy || T == C.WCharTy ||
475                   T == C.WideCharTy)
476                 return MatchPromotion;
477               break;
478             case BuiltinType::Char_U:
479               if (T == C.UnsignedIntTy)
480                 return MatchPromotion;
481               if (T == C.UnsignedShortTy)
482                 return NoMatchPromotionTypeConfusion;
483               break;
484             case BuiltinType::Char_S:
485               if (T == C.IntTy)
486                 return MatchPromotion;
487               if (T == C.ShortTy)
488                 return NoMatchPromotionTypeConfusion;
489               break;
490             case BuiltinType::Half:
491             case BuiltinType::Float16:
492             case BuiltinType::Float:
493               if (T == C.DoubleTy)
494                 return MatchPromotion;
495               break;
496             case BuiltinType::Short:
497             case BuiltinType::UShort:
498               if (T == C.SignedCharTy || T == C.UnsignedCharTy)
499                 return NoMatchPromotionTypeConfusion;
500               break;
501             case BuiltinType::WChar_U:
502             case BuiltinType::WChar_S:
503               if (T != C.WCharTy && T != C.WideCharTy)
504                 return NoMatchPromotionTypeConfusion;
505             }
506           }
507       }
508       return NoMatch;
509     }
510 
511     case CStrTy: {
512       const PointerType *PT = argTy->getAs<PointerType>();
513       if (!PT)
514         return NoMatch;
515       QualType pointeeTy = PT->getPointeeType();
516       if (const BuiltinType *BT = pointeeTy->getAs<BuiltinType>())
517         switch (BT->getKind()) {
518           case BuiltinType::Char_U:
519           case BuiltinType::UChar:
520           case BuiltinType::Char_S:
521           case BuiltinType::SChar:
522             return Match;
523           default:
524             break;
525         }
526 
527       return NoMatch;
528     }
529 
530     case WCStrTy: {
531       const PointerType *PT = argTy->getAs<PointerType>();
532       if (!PT)
533         return NoMatch;
534       QualType pointeeTy =
535         C.getCanonicalType(PT->getPointeeType()).getUnqualifiedType();
536       return pointeeTy == C.getWideCharType() ? Match : NoMatch;
537     }
538 
539     case WIntTy: {
540       QualType WInt = C.getCanonicalType(C.getWIntType()).getUnqualifiedType();
541 
542       if (C.getCanonicalType(argTy).getUnqualifiedType() == WInt)
543         return Match;
544 
545       QualType PromoArg = C.isPromotableIntegerType(argTy)
546                               ? C.getPromotedIntegerType(argTy)
547                               : argTy;
548       PromoArg = C.getCanonicalType(PromoArg).getUnqualifiedType();
549 
550       // If the promoted argument is the corresponding signed type of the
551       // wint_t type, then it should match.
552       if (PromoArg->hasSignedIntegerRepresentation() &&
553           C.getCorrespondingUnsignedType(PromoArg) == WInt)
554         return Match;
555 
556       return WInt == PromoArg ? Match : NoMatch;
557     }
558 
559     case CPointerTy:
560       if (argTy->isVoidPointerType()) {
561         return Match;
562       } if (argTy->isPointerType() || argTy->isObjCObjectPointerType() ||
563             argTy->isBlockPointerType() || argTy->isNullPtrType()) {
564         return NoMatchPedantic;
565       } else {
566         return NoMatch;
567       }
568 
569     case ObjCPointerTy: {
570       if (argTy->getAs<ObjCObjectPointerType>() ||
571           argTy->getAs<BlockPointerType>())
572         return Match;
573 
574       // Handle implicit toll-free bridging.
575       if (const PointerType *PT = argTy->getAs<PointerType>()) {
576         // Things such as CFTypeRef are really just opaque pointers
577         // to C structs representing CF types that can often be bridged
578         // to Objective-C objects.  Since the compiler doesn't know which
579         // structs can be toll-free bridged, we just accept them all.
580         QualType pointee = PT->getPointeeType();
581         if (pointee->getAsStructureType() || pointee->isVoidType())
582           return Match;
583       }
584       return NoMatch;
585     }
586   }
587 
588   llvm_unreachable("Invalid ArgType Kind!");
589 }
590 
591 ArgType ArgType::makeVectorType(ASTContext &C, unsigned NumElts) const {
592   // Check for valid vector element types.
593   if (T.isNull())
594     return ArgType::Invalid();
595 
596   QualType Vec = C.getExtVectorType(T, NumElts);
597   return ArgType(Vec, Name);
598 }
599 
600 QualType ArgType::getRepresentativeType(ASTContext &C) const {
601   QualType Res;
602   switch (K) {
603     case InvalidTy:
604       llvm_unreachable("No representative type for Invalid ArgType");
605     case UnknownTy:
606       llvm_unreachable("No representative type for Unknown ArgType");
607     case AnyCharTy:
608       Res = C.CharTy;
609       break;
610     case SpecificTy:
611       Res = T;
612       break;
613     case CStrTy:
614       Res = C.getPointerType(C.CharTy);
615       break;
616     case WCStrTy:
617       Res = C.getPointerType(C.getWideCharType());
618       break;
619     case ObjCPointerTy:
620       Res = C.ObjCBuiltinIdTy;
621       break;
622     case CPointerTy:
623       Res = C.VoidPtrTy;
624       break;
625     case WIntTy: {
626       Res = C.getWIntType();
627       break;
628     }
629   }
630 
631   if (Ptr)
632     Res = C.getPointerType(Res);
633   return Res;
634 }
635 
636 std::string ArgType::getRepresentativeTypeName(ASTContext &C) const {
637   std::string S = getRepresentativeType(C).getAsString(C.getPrintingPolicy());
638 
639   std::string Alias;
640   if (Name) {
641     // Use a specific name for this type, e.g. "size_t".
642     Alias = Name;
643     if (Ptr) {
644       // If ArgType is actually a pointer to T, append an asterisk.
645       Alias += (Alias[Alias.size()-1] == '*') ? "*" : " *";
646     }
647     // If Alias is the same as the underlying type, e.g. wchar_t, then drop it.
648     if (S == Alias)
649       Alias.clear();
650   }
651 
652   if (!Alias.empty())
653     return std::string("'") + Alias + "' (aka '" + S + "')";
654   return std::string("'") + S + "'";
655 }
656 
657 
658 //===----------------------------------------------------------------------===//
659 // Methods on OptionalAmount.
660 //===----------------------------------------------------------------------===//
661 
662 ArgType
663 analyze_format_string::OptionalAmount::getArgType(ASTContext &Ctx) const {
664   return Ctx.IntTy;
665 }
666 
667 //===----------------------------------------------------------------------===//
668 // Methods on LengthModifier.
669 //===----------------------------------------------------------------------===//
670 
671 const char *
672 analyze_format_string::LengthModifier::toString() const {
673   switch (kind) {
674   case AsChar:
675     return "hh";
676   case AsShort:
677     return "h";
678   case AsShortLong:
679     return "hl";
680   case AsLong: // or AsWideChar
681     return "l";
682   case AsLongLong:
683     return "ll";
684   case AsQuad:
685     return "q";
686   case AsIntMax:
687     return "j";
688   case AsSizeT:
689     return "z";
690   case AsPtrDiff:
691     return "t";
692   case AsInt32:
693     return "I32";
694   case AsInt3264:
695     return "I";
696   case AsInt64:
697     return "I64";
698   case AsLongDouble:
699     return "L";
700   case AsAllocate:
701     return "a";
702   case AsMAllocate:
703     return "m";
704   case AsWide:
705     return "w";
706   case None:
707     return "";
708   }
709   return nullptr;
710 }
711 
712 //===----------------------------------------------------------------------===//
713 // Methods on ConversionSpecifier.
714 //===----------------------------------------------------------------------===//
715 
716 const char *ConversionSpecifier::toString() const {
717   switch (kind) {
718   case bArg: return "b";
719   case BArg: return "B";
720   case dArg: return "d";
721   case DArg: return "D";
722   case iArg: return "i";
723   case oArg: return "o";
724   case OArg: return "O";
725   case uArg: return "u";
726   case UArg: return "U";
727   case xArg: return "x";
728   case XArg: return "X";
729   case fArg: return "f";
730   case FArg: return "F";
731   case eArg: return "e";
732   case EArg: return "E";
733   case gArg: return "g";
734   case GArg: return "G";
735   case aArg: return "a";
736   case AArg: return "A";
737   case cArg: return "c";
738   case sArg: return "s";
739   case pArg: return "p";
740   case PArg:
741     return "P";
742   case nArg: return "n";
743   case PercentArg:  return "%";
744   case ScanListArg: return "[";
745   case InvalidSpecifier: return nullptr;
746 
747   // POSIX unicode extensions.
748   case CArg: return "C";
749   case SArg: return "S";
750 
751   // Objective-C specific specifiers.
752   case ObjCObjArg: return "@";
753 
754   // FreeBSD kernel specific specifiers.
755   case FreeBSDbArg: return "b";
756   case FreeBSDDArg: return "D";
757   case FreeBSDrArg: return "r";
758   case FreeBSDyArg: return "y";
759 
760   // GlibC specific specifiers.
761   case PrintErrno: return "m";
762 
763   // MS specific specifiers.
764   case ZArg: return "Z";
765   }
766   return nullptr;
767 }
768 
769 std::optional<ConversionSpecifier>
770 ConversionSpecifier::getStandardSpecifier() const {
771   ConversionSpecifier::Kind NewKind;
772 
773   switch (getKind()) {
774   default:
775     return std::nullopt;
776   case DArg:
777     NewKind = dArg;
778     break;
779   case UArg:
780     NewKind = uArg;
781     break;
782   case OArg:
783     NewKind = oArg;
784     break;
785   }
786 
787   ConversionSpecifier FixedCS(*this);
788   FixedCS.setKind(NewKind);
789   return FixedCS;
790 }
791 
792 //===----------------------------------------------------------------------===//
793 // Methods on OptionalAmount.
794 //===----------------------------------------------------------------------===//
795 
796 void OptionalAmount::toString(raw_ostream &os) const {
797   switch (hs) {
798   case Invalid:
799   case NotSpecified:
800     return;
801   case Arg:
802     if (UsesDotPrefix)
803         os << ".";
804     if (usesPositionalArg())
805       os << "*" << getPositionalArgIndex() << "$";
806     else
807       os << "*";
808     break;
809   case Constant:
810     if (UsesDotPrefix)
811         os << ".";
812     os << amt;
813     break;
814   }
815 }
816 
817 bool FormatSpecifier::hasValidLengthModifier(const TargetInfo &Target,
818                                              const LangOptions &LO) const {
819   switch (LM.getKind()) {
820     case LengthModifier::None:
821       return true;
822 
823     // Handle most integer flags
824     case LengthModifier::AsShort:
825       // Length modifier only applies to FP vectors.
826       if (LO.OpenCL && CS.isDoubleArg())
827         return !VectorNumElts.isInvalid();
828 
829       if (Target.getTriple().isOSMSVCRT()) {
830         switch (CS.getKind()) {
831           case ConversionSpecifier::cArg:
832           case ConversionSpecifier::CArg:
833           case ConversionSpecifier::sArg:
834           case ConversionSpecifier::SArg:
835           case ConversionSpecifier::ZArg:
836             return true;
837           default:
838             break;
839         }
840       }
841       [[fallthrough]];
842     case LengthModifier::AsChar:
843     case LengthModifier::AsLongLong:
844     case LengthModifier::AsQuad:
845     case LengthModifier::AsIntMax:
846     case LengthModifier::AsSizeT:
847     case LengthModifier::AsPtrDiff:
848       switch (CS.getKind()) {
849         case ConversionSpecifier::bArg:
850         case ConversionSpecifier::BArg:
851         case ConversionSpecifier::dArg:
852         case ConversionSpecifier::DArg:
853         case ConversionSpecifier::iArg:
854         case ConversionSpecifier::oArg:
855         case ConversionSpecifier::OArg:
856         case ConversionSpecifier::uArg:
857         case ConversionSpecifier::UArg:
858         case ConversionSpecifier::xArg:
859         case ConversionSpecifier::XArg:
860         case ConversionSpecifier::nArg:
861           return true;
862         case ConversionSpecifier::FreeBSDrArg:
863         case ConversionSpecifier::FreeBSDyArg:
864           return Target.getTriple().isOSFreeBSD() || Target.getTriple().isPS();
865         default:
866           return false;
867       }
868 
869     case LengthModifier::AsShortLong:
870       return LO.OpenCL && !VectorNumElts.isInvalid();
871 
872     // Handle 'l' flag
873     case LengthModifier::AsLong: // or AsWideChar
874       if (CS.isDoubleArg()) {
875         // Invalid for OpenCL FP scalars.
876         if (LO.OpenCL && VectorNumElts.isInvalid())
877           return false;
878         return true;
879       }
880 
881       switch (CS.getKind()) {
882         case ConversionSpecifier::bArg:
883         case ConversionSpecifier::BArg:
884         case ConversionSpecifier::dArg:
885         case ConversionSpecifier::DArg:
886         case ConversionSpecifier::iArg:
887         case ConversionSpecifier::oArg:
888         case ConversionSpecifier::OArg:
889         case ConversionSpecifier::uArg:
890         case ConversionSpecifier::UArg:
891         case ConversionSpecifier::xArg:
892         case ConversionSpecifier::XArg:
893         case ConversionSpecifier::nArg:
894         case ConversionSpecifier::cArg:
895         case ConversionSpecifier::sArg:
896         case ConversionSpecifier::ScanListArg:
897         case ConversionSpecifier::ZArg:
898           return true;
899         case ConversionSpecifier::FreeBSDrArg:
900         case ConversionSpecifier::FreeBSDyArg:
901           return Target.getTriple().isOSFreeBSD() || Target.getTriple().isPS();
902         default:
903           return false;
904       }
905 
906     case LengthModifier::AsLongDouble:
907       switch (CS.getKind()) {
908         case ConversionSpecifier::aArg:
909         case ConversionSpecifier::AArg:
910         case ConversionSpecifier::fArg:
911         case ConversionSpecifier::FArg:
912         case ConversionSpecifier::eArg:
913         case ConversionSpecifier::EArg:
914         case ConversionSpecifier::gArg:
915         case ConversionSpecifier::GArg:
916           return true;
917         // GNU libc extension.
918         case ConversionSpecifier::dArg:
919         case ConversionSpecifier::iArg:
920         case ConversionSpecifier::oArg:
921         case ConversionSpecifier::uArg:
922         case ConversionSpecifier::xArg:
923         case ConversionSpecifier::XArg:
924           return !Target.getTriple().isOSDarwin() &&
925                  !Target.getTriple().isOSWindows();
926         default:
927           return false;
928       }
929 
930     case LengthModifier::AsAllocate:
931       switch (CS.getKind()) {
932         case ConversionSpecifier::sArg:
933         case ConversionSpecifier::SArg:
934         case ConversionSpecifier::ScanListArg:
935           return true;
936         default:
937           return false;
938       }
939 
940     case LengthModifier::AsMAllocate:
941       switch (CS.getKind()) {
942         case ConversionSpecifier::cArg:
943         case ConversionSpecifier::CArg:
944         case ConversionSpecifier::sArg:
945         case ConversionSpecifier::SArg:
946         case ConversionSpecifier::ScanListArg:
947           return true;
948         default:
949           return false;
950       }
951     case LengthModifier::AsInt32:
952     case LengthModifier::AsInt3264:
953     case LengthModifier::AsInt64:
954       switch (CS.getKind()) {
955         case ConversionSpecifier::dArg:
956         case ConversionSpecifier::iArg:
957         case ConversionSpecifier::oArg:
958         case ConversionSpecifier::uArg:
959         case ConversionSpecifier::xArg:
960         case ConversionSpecifier::XArg:
961           return Target.getTriple().isOSMSVCRT();
962         default:
963           return false;
964       }
965     case LengthModifier::AsWide:
966       switch (CS.getKind()) {
967         case ConversionSpecifier::cArg:
968         case ConversionSpecifier::CArg:
969         case ConversionSpecifier::sArg:
970         case ConversionSpecifier::SArg:
971         case ConversionSpecifier::ZArg:
972           return Target.getTriple().isOSMSVCRT();
973         default:
974           return false;
975       }
976   }
977   llvm_unreachable("Invalid LengthModifier Kind!");
978 }
979 
980 bool FormatSpecifier::hasStandardLengthModifier() const {
981   switch (LM.getKind()) {
982     case LengthModifier::None:
983     case LengthModifier::AsChar:
984     case LengthModifier::AsShort:
985     case LengthModifier::AsLong:
986     case LengthModifier::AsLongLong:
987     case LengthModifier::AsIntMax:
988     case LengthModifier::AsSizeT:
989     case LengthModifier::AsPtrDiff:
990     case LengthModifier::AsLongDouble:
991       return true;
992     case LengthModifier::AsAllocate:
993     case LengthModifier::AsMAllocate:
994     case LengthModifier::AsQuad:
995     case LengthModifier::AsInt32:
996     case LengthModifier::AsInt3264:
997     case LengthModifier::AsInt64:
998     case LengthModifier::AsWide:
999     case LengthModifier::AsShortLong: // ???
1000       return false;
1001   }
1002   llvm_unreachable("Invalid LengthModifier Kind!");
1003 }
1004 
1005 bool FormatSpecifier::hasStandardConversionSpecifier(
1006     const LangOptions &LangOpt) const {
1007   switch (CS.getKind()) {
1008     case ConversionSpecifier::bArg:
1009     case ConversionSpecifier::BArg:
1010     case ConversionSpecifier::cArg:
1011     case ConversionSpecifier::dArg:
1012     case ConversionSpecifier::iArg:
1013     case ConversionSpecifier::oArg:
1014     case ConversionSpecifier::uArg:
1015     case ConversionSpecifier::xArg:
1016     case ConversionSpecifier::XArg:
1017     case ConversionSpecifier::fArg:
1018     case ConversionSpecifier::FArg:
1019     case ConversionSpecifier::eArg:
1020     case ConversionSpecifier::EArg:
1021     case ConversionSpecifier::gArg:
1022     case ConversionSpecifier::GArg:
1023     case ConversionSpecifier::aArg:
1024     case ConversionSpecifier::AArg:
1025     case ConversionSpecifier::sArg:
1026     case ConversionSpecifier::pArg:
1027     case ConversionSpecifier::nArg:
1028     case ConversionSpecifier::ObjCObjArg:
1029     case ConversionSpecifier::ScanListArg:
1030     case ConversionSpecifier::PercentArg:
1031     case ConversionSpecifier::PArg:
1032       return true;
1033     case ConversionSpecifier::CArg:
1034     case ConversionSpecifier::SArg:
1035       return LangOpt.ObjC;
1036     case ConversionSpecifier::InvalidSpecifier:
1037     case ConversionSpecifier::FreeBSDbArg:
1038     case ConversionSpecifier::FreeBSDDArg:
1039     case ConversionSpecifier::FreeBSDrArg:
1040     case ConversionSpecifier::FreeBSDyArg:
1041     case ConversionSpecifier::PrintErrno:
1042     case ConversionSpecifier::DArg:
1043     case ConversionSpecifier::OArg:
1044     case ConversionSpecifier::UArg:
1045     case ConversionSpecifier::ZArg:
1046       return false;
1047   }
1048   llvm_unreachable("Invalid ConversionSpecifier Kind!");
1049 }
1050 
1051 bool FormatSpecifier::hasStandardLengthConversionCombination() const {
1052   if (LM.getKind() == LengthModifier::AsLongDouble) {
1053     switch(CS.getKind()) {
1054         case ConversionSpecifier::dArg:
1055         case ConversionSpecifier::iArg:
1056         case ConversionSpecifier::oArg:
1057         case ConversionSpecifier::uArg:
1058         case ConversionSpecifier::xArg:
1059         case ConversionSpecifier::XArg:
1060           return false;
1061         default:
1062           return true;
1063     }
1064   }
1065   return true;
1066 }
1067 
1068 std::optional<LengthModifier>
1069 FormatSpecifier::getCorrectedLengthModifier() const {
1070   if (CS.isAnyIntArg() || CS.getKind() == ConversionSpecifier::nArg) {
1071     if (LM.getKind() == LengthModifier::AsLongDouble ||
1072         LM.getKind() == LengthModifier::AsQuad) {
1073       LengthModifier FixedLM(LM);
1074       FixedLM.setKind(LengthModifier::AsLongLong);
1075       return FixedLM;
1076     }
1077   }
1078 
1079   return std::nullopt;
1080 }
1081 
1082 bool FormatSpecifier::namedTypeToLengthModifier(QualType QT,
1083                                                 LengthModifier &LM) {
1084   for (/**/; const auto *TT = QT->getAs<TypedefType>();
1085        QT = TT->getDecl()->getUnderlyingType()) {
1086     const TypedefNameDecl *Typedef = TT->getDecl();
1087     const IdentifierInfo *Identifier = Typedef->getIdentifier();
1088     if (Identifier->getName() == "size_t") {
1089       LM.setKind(LengthModifier::AsSizeT);
1090       return true;
1091     } else if (Identifier->getName() == "ssize_t") {
1092       // Not C99, but common in Unix.
1093       LM.setKind(LengthModifier::AsSizeT);
1094       return true;
1095     } else if (Identifier->getName() == "intmax_t") {
1096       LM.setKind(LengthModifier::AsIntMax);
1097       return true;
1098     } else if (Identifier->getName() == "uintmax_t") {
1099       LM.setKind(LengthModifier::AsIntMax);
1100       return true;
1101     } else if (Identifier->getName() == "ptrdiff_t") {
1102       LM.setKind(LengthModifier::AsPtrDiff);
1103       return true;
1104     }
1105   }
1106   return false;
1107 }
1108