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         case BuiltinType::Bool:
372           return Match;
373         }
374         // "Partially matched" because of promotions?
375         if (!Ptr) {
376           switch (BT->getKind()) {
377           default:
378             break;
379           case BuiltinType::Int:
380           case BuiltinType::UInt:
381             return MatchPromotion;
382           case BuiltinType::Short:
383           case BuiltinType::UShort:
384           case BuiltinType::WChar_S:
385           case BuiltinType::WChar_U:
386             return NoMatchPromotionTypeConfusion;
387           }
388         }
389       }
390       return NoMatch;
391     }
392 
393     case SpecificTy: {
394       if (const EnumType *ETy = argTy->getAs<EnumType>()) {
395         // If the enum is incomplete we know nothing about the underlying type.
396         // Assume that it's 'int'. Do not use the underlying type for a scoped
397         // enumeration as that needs an exact match.
398         if (!ETy->getDecl()->isComplete())
399           argTy = C.IntTy;
400         else if (ETy->isUnscopedEnumerationType())
401           argTy = ETy->getDecl()->getIntegerType();
402       }
403       argTy = C.getCanonicalType(argTy).getUnqualifiedType();
404 
405       if (T == argTy)
406         return Match;
407       if (const auto *BT = argTy->getAs<BuiltinType>()) {
408         // Check if the only difference between them is signed vs unsigned
409         // if true, we consider they are compatible.
410         switch (BT->getKind()) {
411           default:
412             break;
413           case BuiltinType::Char_S:
414           case BuiltinType::SChar:
415           case BuiltinType::Char_U:
416           case BuiltinType::UChar:
417           case BuiltinType::Bool:
418             if (T == C.UnsignedShortTy || T == C.ShortTy)
419               return NoMatchTypeConfusion;
420             if (T == C.UnsignedCharTy || T == C.SignedCharTy)
421               return Match;
422             break;
423           case BuiltinType::Short:
424             if (T == C.UnsignedShortTy)
425               return Match;
426             break;
427           case BuiltinType::UShort:
428             if (T == C.ShortTy)
429               return Match;
430             break;
431           case BuiltinType::Int:
432             if (T == C.UnsignedIntTy)
433               return Match;
434             break;
435           case BuiltinType::UInt:
436             if (T == C.IntTy)
437               return Match;
438             break;
439           case BuiltinType::Long:
440             if (T == C.UnsignedLongTy)
441               return Match;
442             break;
443           case BuiltinType::ULong:
444             if (T == C.LongTy)
445               return Match;
446             break;
447           case BuiltinType::LongLong:
448             if (T == C.UnsignedLongLongTy)
449               return Match;
450             break;
451           case BuiltinType::ULongLong:
452             if (T == C.LongLongTy)
453               return Match;
454             break;
455           }
456           // "Partially matched" because of promotions?
457           if (!Ptr) {
458             switch (BT->getKind()) {
459             default:
460               break;
461             case BuiltinType::Int:
462             case BuiltinType::UInt:
463               if (T == C.SignedCharTy || T == C.UnsignedCharTy ||
464                   T == C.ShortTy || T == C.UnsignedShortTy || T == C.WCharTy ||
465                   T == C.WideCharTy)
466                 return MatchPromotion;
467               break;
468             case BuiltinType::Short:
469             case BuiltinType::UShort:
470               if (T == C.SignedCharTy || T == C.UnsignedCharTy)
471                 return NoMatchPromotionTypeConfusion;
472               break;
473             case BuiltinType::WChar_U:
474             case BuiltinType::WChar_S:
475               if (T != C.WCharTy && T != C.WideCharTy)
476                 return NoMatchPromotionTypeConfusion;
477             }
478           }
479       }
480       return NoMatch;
481     }
482 
483     case CStrTy: {
484       const PointerType *PT = argTy->getAs<PointerType>();
485       if (!PT)
486         return NoMatch;
487       QualType pointeeTy = PT->getPointeeType();
488       if (const BuiltinType *BT = pointeeTy->getAs<BuiltinType>())
489         switch (BT->getKind()) {
490           case BuiltinType::Char_U:
491           case BuiltinType::UChar:
492           case BuiltinType::Char_S:
493           case BuiltinType::SChar:
494             return Match;
495           default:
496             break;
497         }
498 
499       return NoMatch;
500     }
501 
502     case WCStrTy: {
503       const PointerType *PT = argTy->getAs<PointerType>();
504       if (!PT)
505         return NoMatch;
506       QualType pointeeTy =
507         C.getCanonicalType(PT->getPointeeType()).getUnqualifiedType();
508       return pointeeTy == C.getWideCharType() ? Match : NoMatch;
509     }
510 
511     case WIntTy: {
512       QualType WInt = C.getCanonicalType(C.getWIntType()).getUnqualifiedType();
513 
514       if (C.getCanonicalType(argTy).getUnqualifiedType() == WInt)
515         return Match;
516 
517       QualType PromoArg = C.isPromotableIntegerType(argTy)
518                               ? C.getPromotedIntegerType(argTy)
519                               : argTy;
520       PromoArg = C.getCanonicalType(PromoArg).getUnqualifiedType();
521 
522       // If the promoted argument is the corresponding signed type of the
523       // wint_t type, then it should match.
524       if (PromoArg->hasSignedIntegerRepresentation() &&
525           C.getCorrespondingUnsignedType(PromoArg) == WInt)
526         return Match;
527 
528       return WInt == PromoArg ? Match : NoMatch;
529     }
530 
531     case CPointerTy:
532       if (argTy->isVoidPointerType()) {
533         return Match;
534       } if (argTy->isPointerType() || argTy->isObjCObjectPointerType() ||
535             argTy->isBlockPointerType() || argTy->isNullPtrType()) {
536         return NoMatchPedantic;
537       } else {
538         return NoMatch;
539       }
540 
541     case ObjCPointerTy: {
542       if (argTy->getAs<ObjCObjectPointerType>() ||
543           argTy->getAs<BlockPointerType>())
544         return Match;
545 
546       // Handle implicit toll-free bridging.
547       if (const PointerType *PT = argTy->getAs<PointerType>()) {
548         // Things such as CFTypeRef are really just opaque pointers
549         // to C structs representing CF types that can often be bridged
550         // to Objective-C objects.  Since the compiler doesn't know which
551         // structs can be toll-free bridged, we just accept them all.
552         QualType pointee = PT->getPointeeType();
553         if (pointee->getAsStructureType() || pointee->isVoidType())
554           return Match;
555       }
556       return NoMatch;
557     }
558   }
559 
560   llvm_unreachable("Invalid ArgType Kind!");
561 }
562 
563 ArgType ArgType::makeVectorType(ASTContext &C, unsigned NumElts) const {
564   // Check for valid vector element types.
565   if (T.isNull())
566     return ArgType::Invalid();
567 
568   QualType Vec = C.getExtVectorType(T, NumElts);
569   return ArgType(Vec, Name);
570 }
571 
572 QualType ArgType::getRepresentativeType(ASTContext &C) const {
573   QualType Res;
574   switch (K) {
575     case InvalidTy:
576       llvm_unreachable("No representative type for Invalid ArgType");
577     case UnknownTy:
578       llvm_unreachable("No representative type for Unknown ArgType");
579     case AnyCharTy:
580       Res = C.CharTy;
581       break;
582     case SpecificTy:
583       Res = T;
584       break;
585     case CStrTy:
586       Res = C.getPointerType(C.CharTy);
587       break;
588     case WCStrTy:
589       Res = C.getPointerType(C.getWideCharType());
590       break;
591     case ObjCPointerTy:
592       Res = C.ObjCBuiltinIdTy;
593       break;
594     case CPointerTy:
595       Res = C.VoidPtrTy;
596       break;
597     case WIntTy: {
598       Res = C.getWIntType();
599       break;
600     }
601   }
602 
603   if (Ptr)
604     Res = C.getPointerType(Res);
605   return Res;
606 }
607 
608 std::string ArgType::getRepresentativeTypeName(ASTContext &C) const {
609   std::string S = getRepresentativeType(C).getAsString(C.getPrintingPolicy());
610 
611   std::string Alias;
612   if (Name) {
613     // Use a specific name for this type, e.g. "size_t".
614     Alias = Name;
615     if (Ptr) {
616       // If ArgType is actually a pointer to T, append an asterisk.
617       Alias += (Alias[Alias.size()-1] == '*') ? "*" : " *";
618     }
619     // If Alias is the same as the underlying type, e.g. wchar_t, then drop it.
620     if (S == Alias)
621       Alias.clear();
622   }
623 
624   if (!Alias.empty())
625     return std::string("'") + Alias + "' (aka '" + S + "')";
626   return std::string("'") + S + "'";
627 }
628 
629 
630 //===----------------------------------------------------------------------===//
631 // Methods on OptionalAmount.
632 //===----------------------------------------------------------------------===//
633 
634 ArgType
635 analyze_format_string::OptionalAmount::getArgType(ASTContext &Ctx) const {
636   return Ctx.IntTy;
637 }
638 
639 //===----------------------------------------------------------------------===//
640 // Methods on LengthModifier.
641 //===----------------------------------------------------------------------===//
642 
643 const char *
644 analyze_format_string::LengthModifier::toString() const {
645   switch (kind) {
646   case AsChar:
647     return "hh";
648   case AsShort:
649     return "h";
650   case AsShortLong:
651     return "hl";
652   case AsLong: // or AsWideChar
653     return "l";
654   case AsLongLong:
655     return "ll";
656   case AsQuad:
657     return "q";
658   case AsIntMax:
659     return "j";
660   case AsSizeT:
661     return "z";
662   case AsPtrDiff:
663     return "t";
664   case AsInt32:
665     return "I32";
666   case AsInt3264:
667     return "I";
668   case AsInt64:
669     return "I64";
670   case AsLongDouble:
671     return "L";
672   case AsAllocate:
673     return "a";
674   case AsMAllocate:
675     return "m";
676   case AsWide:
677     return "w";
678   case None:
679     return "";
680   }
681   return nullptr;
682 }
683 
684 //===----------------------------------------------------------------------===//
685 // Methods on ConversionSpecifier.
686 //===----------------------------------------------------------------------===//
687 
688 const char *ConversionSpecifier::toString() const {
689   switch (kind) {
690   case bArg: return "b";
691   case BArg: return "B";
692   case dArg: return "d";
693   case DArg: return "D";
694   case iArg: return "i";
695   case oArg: return "o";
696   case OArg: return "O";
697   case uArg: return "u";
698   case UArg: return "U";
699   case xArg: return "x";
700   case XArg: return "X";
701   case fArg: return "f";
702   case FArg: return "F";
703   case eArg: return "e";
704   case EArg: return "E";
705   case gArg: return "g";
706   case GArg: return "G";
707   case aArg: return "a";
708   case AArg: return "A";
709   case cArg: return "c";
710   case sArg: return "s";
711   case pArg: return "p";
712   case PArg:
713     return "P";
714   case nArg: return "n";
715   case PercentArg:  return "%";
716   case ScanListArg: return "[";
717   case InvalidSpecifier: return nullptr;
718 
719   // POSIX unicode extensions.
720   case CArg: return "C";
721   case SArg: return "S";
722 
723   // Objective-C specific specifiers.
724   case ObjCObjArg: return "@";
725 
726   // FreeBSD kernel specific specifiers.
727   case FreeBSDbArg: return "b";
728   case FreeBSDDArg: return "D";
729   case FreeBSDrArg: return "r";
730   case FreeBSDyArg: return "y";
731 
732   // GlibC specific specifiers.
733   case PrintErrno: return "m";
734 
735   // MS specific specifiers.
736   case ZArg: return "Z";
737   }
738   return nullptr;
739 }
740 
741 std::optional<ConversionSpecifier>
742 ConversionSpecifier::getStandardSpecifier() const {
743   ConversionSpecifier::Kind NewKind;
744 
745   switch (getKind()) {
746   default:
747     return std::nullopt;
748   case DArg:
749     NewKind = dArg;
750     break;
751   case UArg:
752     NewKind = uArg;
753     break;
754   case OArg:
755     NewKind = oArg;
756     break;
757   }
758 
759   ConversionSpecifier FixedCS(*this);
760   FixedCS.setKind(NewKind);
761   return FixedCS;
762 }
763 
764 //===----------------------------------------------------------------------===//
765 // Methods on OptionalAmount.
766 //===----------------------------------------------------------------------===//
767 
768 void OptionalAmount::toString(raw_ostream &os) const {
769   switch (hs) {
770   case Invalid:
771   case NotSpecified:
772     return;
773   case Arg:
774     if (UsesDotPrefix)
775         os << ".";
776     if (usesPositionalArg())
777       os << "*" << getPositionalArgIndex() << "$";
778     else
779       os << "*";
780     break;
781   case Constant:
782     if (UsesDotPrefix)
783         os << ".";
784     os << amt;
785     break;
786   }
787 }
788 
789 bool FormatSpecifier::hasValidLengthModifier(const TargetInfo &Target,
790                                              const LangOptions &LO) const {
791   switch (LM.getKind()) {
792     case LengthModifier::None:
793       return true;
794 
795     // Handle most integer flags
796     case LengthModifier::AsShort:
797       // Length modifier only applies to FP vectors.
798       if (LO.OpenCL && CS.isDoubleArg())
799         return !VectorNumElts.isInvalid();
800 
801       if (Target.getTriple().isOSMSVCRT()) {
802         switch (CS.getKind()) {
803           case ConversionSpecifier::cArg:
804           case ConversionSpecifier::CArg:
805           case ConversionSpecifier::sArg:
806           case ConversionSpecifier::SArg:
807           case ConversionSpecifier::ZArg:
808             return true;
809           default:
810             break;
811         }
812       }
813       [[fallthrough]];
814     case LengthModifier::AsChar:
815     case LengthModifier::AsLongLong:
816     case LengthModifier::AsQuad:
817     case LengthModifier::AsIntMax:
818     case LengthModifier::AsSizeT:
819     case LengthModifier::AsPtrDiff:
820       switch (CS.getKind()) {
821         case ConversionSpecifier::bArg:
822         case ConversionSpecifier::BArg:
823         case ConversionSpecifier::dArg:
824         case ConversionSpecifier::DArg:
825         case ConversionSpecifier::iArg:
826         case ConversionSpecifier::oArg:
827         case ConversionSpecifier::OArg:
828         case ConversionSpecifier::uArg:
829         case ConversionSpecifier::UArg:
830         case ConversionSpecifier::xArg:
831         case ConversionSpecifier::XArg:
832         case ConversionSpecifier::nArg:
833           return true;
834         case ConversionSpecifier::FreeBSDrArg:
835         case ConversionSpecifier::FreeBSDyArg:
836           return Target.getTriple().isOSFreeBSD() || Target.getTriple().isPS();
837         default:
838           return false;
839       }
840 
841     case LengthModifier::AsShortLong:
842       return LO.OpenCL && !VectorNumElts.isInvalid();
843 
844     // Handle 'l' flag
845     case LengthModifier::AsLong: // or AsWideChar
846       if (CS.isDoubleArg()) {
847         // Invalid for OpenCL FP scalars.
848         if (LO.OpenCL && VectorNumElts.isInvalid())
849           return false;
850         return true;
851       }
852 
853       switch (CS.getKind()) {
854         case ConversionSpecifier::bArg:
855         case ConversionSpecifier::BArg:
856         case ConversionSpecifier::dArg:
857         case ConversionSpecifier::DArg:
858         case ConversionSpecifier::iArg:
859         case ConversionSpecifier::oArg:
860         case ConversionSpecifier::OArg:
861         case ConversionSpecifier::uArg:
862         case ConversionSpecifier::UArg:
863         case ConversionSpecifier::xArg:
864         case ConversionSpecifier::XArg:
865         case ConversionSpecifier::nArg:
866         case ConversionSpecifier::cArg:
867         case ConversionSpecifier::sArg:
868         case ConversionSpecifier::ScanListArg:
869         case ConversionSpecifier::ZArg:
870           return true;
871         case ConversionSpecifier::FreeBSDrArg:
872         case ConversionSpecifier::FreeBSDyArg:
873           return Target.getTriple().isOSFreeBSD() || Target.getTriple().isPS();
874         default:
875           return false;
876       }
877 
878     case LengthModifier::AsLongDouble:
879       switch (CS.getKind()) {
880         case ConversionSpecifier::aArg:
881         case ConversionSpecifier::AArg:
882         case ConversionSpecifier::fArg:
883         case ConversionSpecifier::FArg:
884         case ConversionSpecifier::eArg:
885         case ConversionSpecifier::EArg:
886         case ConversionSpecifier::gArg:
887         case ConversionSpecifier::GArg:
888           return true;
889         // GNU libc extension.
890         case ConversionSpecifier::dArg:
891         case ConversionSpecifier::iArg:
892         case ConversionSpecifier::oArg:
893         case ConversionSpecifier::uArg:
894         case ConversionSpecifier::xArg:
895         case ConversionSpecifier::XArg:
896           return !Target.getTriple().isOSDarwin() &&
897                  !Target.getTriple().isOSWindows();
898         default:
899           return false;
900       }
901 
902     case LengthModifier::AsAllocate:
903       switch (CS.getKind()) {
904         case ConversionSpecifier::sArg:
905         case ConversionSpecifier::SArg:
906         case ConversionSpecifier::ScanListArg:
907           return true;
908         default:
909           return false;
910       }
911 
912     case LengthModifier::AsMAllocate:
913       switch (CS.getKind()) {
914         case ConversionSpecifier::cArg:
915         case ConversionSpecifier::CArg:
916         case ConversionSpecifier::sArg:
917         case ConversionSpecifier::SArg:
918         case ConversionSpecifier::ScanListArg:
919           return true;
920         default:
921           return false;
922       }
923     case LengthModifier::AsInt32:
924     case LengthModifier::AsInt3264:
925     case LengthModifier::AsInt64:
926       switch (CS.getKind()) {
927         case ConversionSpecifier::dArg:
928         case ConversionSpecifier::iArg:
929         case ConversionSpecifier::oArg:
930         case ConversionSpecifier::uArg:
931         case ConversionSpecifier::xArg:
932         case ConversionSpecifier::XArg:
933           return Target.getTriple().isOSMSVCRT();
934         default:
935           return false;
936       }
937     case LengthModifier::AsWide:
938       switch (CS.getKind()) {
939         case ConversionSpecifier::cArg:
940         case ConversionSpecifier::CArg:
941         case ConversionSpecifier::sArg:
942         case ConversionSpecifier::SArg:
943         case ConversionSpecifier::ZArg:
944           return Target.getTriple().isOSMSVCRT();
945         default:
946           return false;
947       }
948   }
949   llvm_unreachable("Invalid LengthModifier Kind!");
950 }
951 
952 bool FormatSpecifier::hasStandardLengthModifier() const {
953   switch (LM.getKind()) {
954     case LengthModifier::None:
955     case LengthModifier::AsChar:
956     case LengthModifier::AsShort:
957     case LengthModifier::AsLong:
958     case LengthModifier::AsLongLong:
959     case LengthModifier::AsIntMax:
960     case LengthModifier::AsSizeT:
961     case LengthModifier::AsPtrDiff:
962     case LengthModifier::AsLongDouble:
963       return true;
964     case LengthModifier::AsAllocate:
965     case LengthModifier::AsMAllocate:
966     case LengthModifier::AsQuad:
967     case LengthModifier::AsInt32:
968     case LengthModifier::AsInt3264:
969     case LengthModifier::AsInt64:
970     case LengthModifier::AsWide:
971     case LengthModifier::AsShortLong: // ???
972       return false;
973   }
974   llvm_unreachable("Invalid LengthModifier Kind!");
975 }
976 
977 bool FormatSpecifier::hasStandardConversionSpecifier(
978     const LangOptions &LangOpt) const {
979   switch (CS.getKind()) {
980     case ConversionSpecifier::bArg:
981     case ConversionSpecifier::BArg:
982     case ConversionSpecifier::cArg:
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     case ConversionSpecifier::fArg:
990     case ConversionSpecifier::FArg:
991     case ConversionSpecifier::eArg:
992     case ConversionSpecifier::EArg:
993     case ConversionSpecifier::gArg:
994     case ConversionSpecifier::GArg:
995     case ConversionSpecifier::aArg:
996     case ConversionSpecifier::AArg:
997     case ConversionSpecifier::sArg:
998     case ConversionSpecifier::pArg:
999     case ConversionSpecifier::nArg:
1000     case ConversionSpecifier::ObjCObjArg:
1001     case ConversionSpecifier::ScanListArg:
1002     case ConversionSpecifier::PercentArg:
1003     case ConversionSpecifier::PArg:
1004       return true;
1005     case ConversionSpecifier::CArg:
1006     case ConversionSpecifier::SArg:
1007       return LangOpt.ObjC;
1008     case ConversionSpecifier::InvalidSpecifier:
1009     case ConversionSpecifier::FreeBSDbArg:
1010     case ConversionSpecifier::FreeBSDDArg:
1011     case ConversionSpecifier::FreeBSDrArg:
1012     case ConversionSpecifier::FreeBSDyArg:
1013     case ConversionSpecifier::PrintErrno:
1014     case ConversionSpecifier::DArg:
1015     case ConversionSpecifier::OArg:
1016     case ConversionSpecifier::UArg:
1017     case ConversionSpecifier::ZArg:
1018       return false;
1019   }
1020   llvm_unreachable("Invalid ConversionSpecifier Kind!");
1021 }
1022 
1023 bool FormatSpecifier::hasStandardLengthConversionCombination() const {
1024   if (LM.getKind() == LengthModifier::AsLongDouble) {
1025     switch(CS.getKind()) {
1026         case ConversionSpecifier::dArg:
1027         case ConversionSpecifier::iArg:
1028         case ConversionSpecifier::oArg:
1029         case ConversionSpecifier::uArg:
1030         case ConversionSpecifier::xArg:
1031         case ConversionSpecifier::XArg:
1032           return false;
1033         default:
1034           return true;
1035     }
1036   }
1037   return true;
1038 }
1039 
1040 std::optional<LengthModifier>
1041 FormatSpecifier::getCorrectedLengthModifier() const {
1042   if (CS.isAnyIntArg() || CS.getKind() == ConversionSpecifier::nArg) {
1043     if (LM.getKind() == LengthModifier::AsLongDouble ||
1044         LM.getKind() == LengthModifier::AsQuad) {
1045       LengthModifier FixedLM(LM);
1046       FixedLM.setKind(LengthModifier::AsLongLong);
1047       return FixedLM;
1048     }
1049   }
1050 
1051   return std::nullopt;
1052 }
1053 
1054 bool FormatSpecifier::namedTypeToLengthModifier(QualType QT,
1055                                                 LengthModifier &LM) {
1056   for (/**/; const auto *TT = QT->getAs<TypedefType>();
1057        QT = TT->getDecl()->getUnderlyingType()) {
1058     const TypedefNameDecl *Typedef = TT->getDecl();
1059     const IdentifierInfo *Identifier = Typedef->getIdentifier();
1060     if (Identifier->getName() == "size_t") {
1061       LM.setKind(LengthModifier::AsSizeT);
1062       return true;
1063     } else if (Identifier->getName() == "ssize_t") {
1064       // Not C99, but common in Unix.
1065       LM.setKind(LengthModifier::AsSizeT);
1066       return true;
1067     } else if (Identifier->getName() == "intmax_t") {
1068       LM.setKind(LengthModifier::AsIntMax);
1069       return true;
1070     } else if (Identifier->getName() == "uintmax_t") {
1071       LM.setKind(LengthModifier::AsIntMax);
1072       return true;
1073     } else if (Identifier->getName() == "ptrdiff_t") {
1074       LM.setKind(LengthModifier::AsPtrDiff);
1075       return true;
1076     }
1077   }
1078   return false;
1079 }
1080