1 //===- MicrosoftDemangle.cpp ----------------------------------------------===//
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 // This file defines a demangler for MSVC-style mangled symbols.
10 //
11 // This file has no dependencies on the rest of LLVM so that it can be
12 // easily reused in other programs such as libcxxabi.
13 //
14 //===----------------------------------------------------------------------===//
15 
16 #include "llvm/Demangle/MicrosoftDemangle.h"
17 #include "llvm/Demangle/Demangle.h"
18 #include "llvm/Demangle/MicrosoftDemangleNodes.h"
19 
20 #include "llvm/Demangle/DemangleConfig.h"
21 #include "llvm/Demangle/StringView.h"
22 #include "llvm/Demangle/Utility.h"
23 
24 #include <array>
25 #include <cctype>
26 #include <cstdio>
27 #include <tuple>
28 
29 using namespace llvm;
30 using namespace ms_demangle;
31 
32 static bool startsWithDigit(StringView S) {
33   return !S.empty() && std::isdigit(S.front());
34 }
35 
36 
37 struct NodeList {
38   Node *N = nullptr;
39   NodeList *Next = nullptr;
40 };
41 
42 static bool isMemberPointer(StringView MangledName, bool &Error) {
43   Error = false;
44   switch (MangledName.popFront()) {
45   case '$':
46     // This is probably an rvalue reference (e.g. $$Q), and you cannot have an
47     // rvalue reference to a member.
48     return false;
49   case 'A':
50     // 'A' indicates a reference, and you cannot have a reference to a member
51     // function or member.
52     return false;
53   case 'P':
54   case 'Q':
55   case 'R':
56   case 'S':
57     // These 4 values indicate some kind of pointer, but we still don't know
58     // what.
59     break;
60   default:
61     Error = true;
62     return false;
63   }
64 
65   // If it starts with a number, then 6 indicates a non-member function
66   // pointer, and 8 indicates a member function pointer.
67   if (startsWithDigit(MangledName)) {
68     if (MangledName[0] != '6' && MangledName[0] != '8') {
69       Error = true;
70       return false;
71     }
72     return (MangledName[0] == '8');
73   }
74 
75   // Remove ext qualifiers since those can appear on either type and are
76   // therefore not indicative.
77   MangledName.consumeFront('E'); // 64-bit
78   MangledName.consumeFront('I'); // restrict
79   MangledName.consumeFront('F'); // unaligned
80 
81   if (MangledName.empty()) {
82     Error = true;
83     return false;
84   }
85 
86   // The next value should be either ABCD (non-member) or QRST (member).
87   switch (MangledName.front()) {
88   case 'A':
89   case 'B':
90   case 'C':
91   case 'D':
92     return false;
93   case 'Q':
94   case 'R':
95   case 'S':
96   case 'T':
97     return true;
98   default:
99     Error = true;
100     return false;
101   }
102 }
103 
104 static SpecialIntrinsicKind
105 consumeSpecialIntrinsicKind(StringView &MangledName) {
106   if (MangledName.consumeFront("?_7"))
107     return SpecialIntrinsicKind::Vftable;
108   if (MangledName.consumeFront("?_8"))
109     return SpecialIntrinsicKind::Vbtable;
110   if (MangledName.consumeFront("?_9"))
111     return SpecialIntrinsicKind::VcallThunk;
112   if (MangledName.consumeFront("?_A"))
113     return SpecialIntrinsicKind::Typeof;
114   if (MangledName.consumeFront("?_B"))
115     return SpecialIntrinsicKind::LocalStaticGuard;
116   if (MangledName.consumeFront("?_C"))
117     return SpecialIntrinsicKind::StringLiteralSymbol;
118   if (MangledName.consumeFront("?_P"))
119     return SpecialIntrinsicKind::UdtReturning;
120   if (MangledName.consumeFront("?_R0"))
121     return SpecialIntrinsicKind::RttiTypeDescriptor;
122   if (MangledName.consumeFront("?_R1"))
123     return SpecialIntrinsicKind::RttiBaseClassDescriptor;
124   if (MangledName.consumeFront("?_R2"))
125     return SpecialIntrinsicKind::RttiBaseClassArray;
126   if (MangledName.consumeFront("?_R3"))
127     return SpecialIntrinsicKind::RttiClassHierarchyDescriptor;
128   if (MangledName.consumeFront("?_R4"))
129     return SpecialIntrinsicKind::RttiCompleteObjLocator;
130   if (MangledName.consumeFront("?_S"))
131     return SpecialIntrinsicKind::LocalVftable;
132   if (MangledName.consumeFront("?__E"))
133     return SpecialIntrinsicKind::DynamicInitializer;
134   if (MangledName.consumeFront("?__F"))
135     return SpecialIntrinsicKind::DynamicAtexitDestructor;
136   if (MangledName.consumeFront("?__J"))
137     return SpecialIntrinsicKind::LocalStaticThreadGuard;
138   return SpecialIntrinsicKind::None;
139 }
140 
141 static bool startsWithLocalScopePattern(StringView S) {
142   if (!S.consumeFront('?'))
143     return false;
144   if (S.size() < 2)
145     return false;
146 
147   size_t End = S.find('?');
148   if (End == StringView::npos)
149     return false;
150   StringView Candidate = S.substr(0, End);
151   if (Candidate.empty())
152     return false;
153 
154   // \?[0-9]\?
155   // ?@? is the discriminator 0.
156   if (Candidate.size() == 1)
157     return Candidate[0] == '@' || (Candidate[0] >= '0' && Candidate[0] <= '9');
158 
159   // If it's not 0-9, then it's an encoded number terminated with an @
160   if (Candidate.back() != '@')
161     return false;
162   Candidate = Candidate.dropBack();
163 
164   // An encoded number starts with B-P and all subsequent digits are in A-P.
165   // Note that the reason the first digit cannot be A is two fold.  First, it
166   // would create an ambiguity with ?A which delimits the beginning of an
167   // anonymous namespace.  Second, A represents 0, and you don't start a multi
168   // digit number with a leading 0.  Presumably the anonymous namespace
169   // ambiguity is also why single digit encoded numbers use 0-9 rather than A-J.
170   if (Candidate[0] < 'B' || Candidate[0] > 'P')
171     return false;
172   Candidate = Candidate.dropFront();
173   while (!Candidate.empty()) {
174     if (Candidate[0] < 'A' || Candidate[0] > 'P')
175       return false;
176     Candidate = Candidate.dropFront();
177   }
178 
179   return true;
180 }
181 
182 static bool isTagType(StringView S) {
183   switch (S.front()) {
184   case 'T': // union
185   case 'U': // struct
186   case 'V': // class
187   case 'W': // enum
188     return true;
189   }
190   return false;
191 }
192 
193 static bool isCustomType(StringView S) { return S[0] == '?'; }
194 
195 static bool isPointerType(StringView S) {
196   if (S.startsWith("$$Q")) // foo &&
197     return true;
198 
199   switch (S.front()) {
200   case 'A': // foo &
201   case 'P': // foo *
202   case 'Q': // foo *const
203   case 'R': // foo *volatile
204   case 'S': // foo *const volatile
205     return true;
206   }
207   return false;
208 }
209 
210 static bool isArrayType(StringView S) { return S[0] == 'Y'; }
211 
212 static bool isFunctionType(StringView S) {
213   return S.startsWith("$$A8@@") || S.startsWith("$$A6");
214 }
215 
216 static FunctionRefQualifier
217 demangleFunctionRefQualifier(StringView &MangledName) {
218   if (MangledName.consumeFront('G'))
219     return FunctionRefQualifier::Reference;
220   else if (MangledName.consumeFront('H'))
221     return FunctionRefQualifier::RValueReference;
222   return FunctionRefQualifier::None;
223 }
224 
225 static std::pair<Qualifiers, PointerAffinity>
226 demanglePointerCVQualifiers(StringView &MangledName) {
227   if (MangledName.consumeFront("$$Q"))
228     return std::make_pair(Q_None, PointerAffinity::RValueReference);
229 
230   switch (MangledName.popFront()) {
231   case 'A':
232     return std::make_pair(Q_None, PointerAffinity::Reference);
233   case 'P':
234     return std::make_pair(Q_None, PointerAffinity::Pointer);
235   case 'Q':
236     return std::make_pair(Q_Const, PointerAffinity::Pointer);
237   case 'R':
238     return std::make_pair(Q_Volatile, PointerAffinity::Pointer);
239   case 'S':
240     return std::make_pair(Qualifiers(Q_Const | Q_Volatile),
241                           PointerAffinity::Pointer);
242   default:
243     assert(false && "Ty is not a pointer type!");
244   }
245   return std::make_pair(Q_None, PointerAffinity::Pointer);
246 }
247 
248 StringView Demangler::copyString(StringView Borrowed) {
249   char *Stable = Arena.allocUnalignedBuffer(Borrowed.size() + 1);
250   std::strcpy(Stable, Borrowed.begin());
251 
252   return {Stable, Borrowed.size()};
253 }
254 
255 SpecialTableSymbolNode *
256 Demangler::demangleSpecialTableSymbolNode(StringView &MangledName,
257                                           SpecialIntrinsicKind K) {
258   NamedIdentifierNode *NI = Arena.alloc<NamedIdentifierNode>();
259   switch (K) {
260   case SpecialIntrinsicKind::Vftable:
261     NI->Name = "`vftable'";
262     break;
263   case SpecialIntrinsicKind::Vbtable:
264     NI->Name = "`vbtable'";
265     break;
266   case SpecialIntrinsicKind::LocalVftable:
267     NI->Name = "`local vftable'";
268     break;
269   case SpecialIntrinsicKind::RttiCompleteObjLocator:
270     NI->Name = "`RTTI Complete Object Locator'";
271     break;
272   default:
273     DEMANGLE_UNREACHABLE;
274   }
275   QualifiedNameNode *QN = demangleNameScopeChain(MangledName, NI);
276   SpecialTableSymbolNode *STSN = Arena.alloc<SpecialTableSymbolNode>();
277   STSN->Name = QN;
278   bool IsMember = false;
279   if (MangledName.empty()) {
280     Error = true;
281     return nullptr;
282   }
283   char Front = MangledName.popFront();
284   if (Front != '6' && Front != '7') {
285     Error = true;
286     return nullptr;
287   }
288 
289   std::tie(STSN->Quals, IsMember) = demangleQualifiers(MangledName);
290   if (!MangledName.consumeFront('@'))
291     STSN->TargetName = demangleFullyQualifiedTypeName(MangledName);
292   return STSN;
293 }
294 
295 LocalStaticGuardVariableNode *
296 Demangler::demangleLocalStaticGuard(StringView &MangledName) {
297   LocalStaticGuardIdentifierNode *LSGI =
298       Arena.alloc<LocalStaticGuardIdentifierNode>();
299   QualifiedNameNode *QN = demangleNameScopeChain(MangledName, LSGI);
300   LocalStaticGuardVariableNode *LSGVN =
301       Arena.alloc<LocalStaticGuardVariableNode>();
302   LSGVN->Name = QN;
303 
304   if (MangledName.consumeFront("4IA"))
305     LSGVN->IsVisible = false;
306   else if (MangledName.consumeFront("5"))
307     LSGVN->IsVisible = true;
308   else {
309     Error = true;
310     return nullptr;
311   }
312 
313   if (!MangledName.empty())
314     LSGI->ScopeIndex = demangleUnsigned(MangledName);
315   return LSGVN;
316 }
317 
318 static NamedIdentifierNode *synthesizeNamedIdentifier(ArenaAllocator &Arena,
319                                                       StringView Name) {
320   NamedIdentifierNode *Id = Arena.alloc<NamedIdentifierNode>();
321   Id->Name = Name;
322   return Id;
323 }
324 
325 static QualifiedNameNode *synthesizeQualifiedName(ArenaAllocator &Arena,
326                                                   IdentifierNode *Identifier) {
327   QualifiedNameNode *QN = Arena.alloc<QualifiedNameNode>();
328   QN->Components = Arena.alloc<NodeArrayNode>();
329   QN->Components->Count = 1;
330   QN->Components->Nodes = Arena.allocArray<Node *>(1);
331   QN->Components->Nodes[0] = Identifier;
332   return QN;
333 }
334 
335 static QualifiedNameNode *synthesizeQualifiedName(ArenaAllocator &Arena,
336                                                   StringView Name) {
337   NamedIdentifierNode *Id = synthesizeNamedIdentifier(Arena, Name);
338   return synthesizeQualifiedName(Arena, Id);
339 }
340 
341 static VariableSymbolNode *synthesizeVariable(ArenaAllocator &Arena,
342                                               TypeNode *Type,
343                                               StringView VariableName) {
344   VariableSymbolNode *VSN = Arena.alloc<VariableSymbolNode>();
345   VSN->Type = Type;
346   VSN->Name = synthesizeQualifiedName(Arena, VariableName);
347   return VSN;
348 }
349 
350 VariableSymbolNode *Demangler::demangleUntypedVariable(
351     ArenaAllocator &Arena, StringView &MangledName, StringView VariableName) {
352   NamedIdentifierNode *NI = synthesizeNamedIdentifier(Arena, VariableName);
353   QualifiedNameNode *QN = demangleNameScopeChain(MangledName, NI);
354   VariableSymbolNode *VSN = Arena.alloc<VariableSymbolNode>();
355   VSN->Name = QN;
356   if (MangledName.consumeFront("8"))
357     return VSN;
358 
359   Error = true;
360   return nullptr;
361 }
362 
363 VariableSymbolNode *
364 Demangler::demangleRttiBaseClassDescriptorNode(ArenaAllocator &Arena,
365                                                StringView &MangledName) {
366   RttiBaseClassDescriptorNode *RBCDN =
367       Arena.alloc<RttiBaseClassDescriptorNode>();
368   RBCDN->NVOffset = demangleUnsigned(MangledName);
369   RBCDN->VBPtrOffset = demangleSigned(MangledName);
370   RBCDN->VBTableOffset = demangleUnsigned(MangledName);
371   RBCDN->Flags = demangleUnsigned(MangledName);
372   if (Error)
373     return nullptr;
374 
375   VariableSymbolNode *VSN = Arena.alloc<VariableSymbolNode>();
376   VSN->Name = demangleNameScopeChain(MangledName, RBCDN);
377   MangledName.consumeFront('8');
378   return VSN;
379 }
380 
381 FunctionSymbolNode *Demangler::demangleInitFiniStub(StringView &MangledName,
382                                                     bool IsDestructor) {
383   DynamicStructorIdentifierNode *DSIN =
384       Arena.alloc<DynamicStructorIdentifierNode>();
385   DSIN->IsDestructor = IsDestructor;
386 
387   bool IsKnownStaticDataMember = false;
388   if (MangledName.consumeFront('?'))
389     IsKnownStaticDataMember = true;
390 
391   SymbolNode *Symbol = demangleDeclarator(MangledName);
392   if (Error)
393     return nullptr;
394 
395   FunctionSymbolNode *FSN = nullptr;
396 
397   if (Symbol->kind() == NodeKind::VariableSymbol) {
398     DSIN->Variable = static_cast<VariableSymbolNode *>(Symbol);
399 
400     // Older versions of clang mangled this type of symbol incorrectly.  They
401     // would omit the leading ? and they would only emit a single @ at the end.
402     // The correct mangling is a leading ? and 2 trailing @ signs.  Handle
403     // both cases.
404     int AtCount = IsKnownStaticDataMember ? 2 : 1;
405     for (int I = 0; I < AtCount; ++I) {
406       if (MangledName.consumeFront('@'))
407         continue;
408       Error = true;
409       return nullptr;
410     }
411 
412     FSN = demangleFunctionEncoding(MangledName);
413     if (FSN)
414       FSN->Name = synthesizeQualifiedName(Arena, DSIN);
415   } else {
416     if (IsKnownStaticDataMember) {
417       // This was supposed to be a static data member, but we got a function.
418       Error = true;
419       return nullptr;
420     }
421 
422     FSN = static_cast<FunctionSymbolNode *>(Symbol);
423     DSIN->Name = Symbol->Name;
424     FSN->Name = synthesizeQualifiedName(Arena, DSIN);
425   }
426 
427   return FSN;
428 }
429 
430 SymbolNode *Demangler::demangleSpecialIntrinsic(StringView &MangledName) {
431   SpecialIntrinsicKind SIK = consumeSpecialIntrinsicKind(MangledName);
432   if (SIK == SpecialIntrinsicKind::None)
433     return nullptr;
434 
435   switch (SIK) {
436   case SpecialIntrinsicKind::StringLiteralSymbol:
437     return demangleStringLiteral(MangledName);
438   case SpecialIntrinsicKind::Vftable:
439   case SpecialIntrinsicKind::Vbtable:
440   case SpecialIntrinsicKind::LocalVftable:
441   case SpecialIntrinsicKind::RttiCompleteObjLocator:
442     return demangleSpecialTableSymbolNode(MangledName, SIK);
443   case SpecialIntrinsicKind::VcallThunk:
444     return demangleVcallThunkNode(MangledName);
445   case SpecialIntrinsicKind::LocalStaticGuard:
446     return demangleLocalStaticGuard(MangledName);
447   case SpecialIntrinsicKind::RttiTypeDescriptor: {
448     TypeNode *T = demangleType(MangledName, QualifierMangleMode::Result);
449     if (Error)
450       break;
451     if (!MangledName.consumeFront("@8"))
452       break;
453     if (!MangledName.empty())
454       break;
455     return synthesizeVariable(Arena, T, "`RTTI Type Descriptor'");
456   }
457   case SpecialIntrinsicKind::RttiBaseClassArray:
458     return demangleUntypedVariable(Arena, MangledName,
459                                    "`RTTI Base Class Array'");
460   case SpecialIntrinsicKind::RttiClassHierarchyDescriptor:
461     return demangleUntypedVariable(Arena, MangledName,
462                                    "`RTTI Class Hierarchy Descriptor'");
463   case SpecialIntrinsicKind::RttiBaseClassDescriptor:
464     return demangleRttiBaseClassDescriptorNode(Arena, MangledName);
465   case SpecialIntrinsicKind::DynamicInitializer:
466     return demangleInitFiniStub(MangledName, false);
467   case SpecialIntrinsicKind::DynamicAtexitDestructor:
468     return demangleInitFiniStub(MangledName, true);
469   default:
470     break;
471   }
472   Error = true;
473   return nullptr;
474 }
475 
476 IdentifierNode *
477 Demangler::demangleFunctionIdentifierCode(StringView &MangledName) {
478   assert(MangledName.startsWith('?'));
479   MangledName = MangledName.dropFront();
480   if (MangledName.empty()) {
481     Error = true;
482     return nullptr;
483   }
484 
485   if (MangledName.consumeFront("__"))
486     return demangleFunctionIdentifierCode(
487         MangledName, FunctionIdentifierCodeGroup::DoubleUnder);
488   if (MangledName.consumeFront("_"))
489     return demangleFunctionIdentifierCode(MangledName,
490                                           FunctionIdentifierCodeGroup::Under);
491   return demangleFunctionIdentifierCode(MangledName,
492                                         FunctionIdentifierCodeGroup::Basic);
493 }
494 
495 StructorIdentifierNode *
496 Demangler::demangleStructorIdentifier(StringView &MangledName,
497                                       bool IsDestructor) {
498   StructorIdentifierNode *N = Arena.alloc<StructorIdentifierNode>();
499   N->IsDestructor = IsDestructor;
500   return N;
501 }
502 
503 ConversionOperatorIdentifierNode *
504 Demangler::demangleConversionOperatorIdentifier(StringView &MangledName) {
505   ConversionOperatorIdentifierNode *N =
506       Arena.alloc<ConversionOperatorIdentifierNode>();
507   return N;
508 }
509 
510 LiteralOperatorIdentifierNode *
511 Demangler::demangleLiteralOperatorIdentifier(StringView &MangledName) {
512   LiteralOperatorIdentifierNode *N =
513       Arena.alloc<LiteralOperatorIdentifierNode>();
514   N->Name = demangleSimpleString(MangledName, /*Memorize=*/false);
515   return N;
516 }
517 
518 IntrinsicFunctionKind
519 Demangler::translateIntrinsicFunctionCode(char CH,
520                                           FunctionIdentifierCodeGroup Group) {
521   using IFK = IntrinsicFunctionKind;
522   if (!(CH >= '0' && CH <= '9') && !(CH >= 'A' && CH <= 'Z')) {
523     Error = true;
524     return IFK::None;
525   }
526 
527   // Not all ? identifiers are intrinsics *functions*.  This function only maps
528   // operator codes for the special functions, all others are handled elsewhere,
529   // hence the IFK::None entries in the table.
530   static IFK Basic[36] = {
531       IFK::None,             // ?0 # Foo::Foo()
532       IFK::None,             // ?1 # Foo::~Foo()
533       IFK::New,              // ?2 # operator new
534       IFK::Delete,           // ?3 # operator delete
535       IFK::Assign,           // ?4 # operator=
536       IFK::RightShift,       // ?5 # operator>>
537       IFK::LeftShift,        // ?6 # operator<<
538       IFK::LogicalNot,       // ?7 # operator!
539       IFK::Equals,           // ?8 # operator==
540       IFK::NotEquals,        // ?9 # operator!=
541       IFK::ArraySubscript,   // ?A # operator[]
542       IFK::None,             // ?B # Foo::operator <type>()
543       IFK::Pointer,          // ?C # operator->
544       IFK::Dereference,      // ?D # operator*
545       IFK::Increment,        // ?E # operator++
546       IFK::Decrement,        // ?F # operator--
547       IFK::Minus,            // ?G # operator-
548       IFK::Plus,             // ?H # operator+
549       IFK::BitwiseAnd,       // ?I # operator&
550       IFK::MemberPointer,    // ?J # operator->*
551       IFK::Divide,           // ?K # operator/
552       IFK::Modulus,          // ?L # operator%
553       IFK::LessThan,         // ?M operator<
554       IFK::LessThanEqual,    // ?N operator<=
555       IFK::GreaterThan,      // ?O operator>
556       IFK::GreaterThanEqual, // ?P operator>=
557       IFK::Comma,            // ?Q operator,
558       IFK::Parens,           // ?R operator()
559       IFK::BitwiseNot,       // ?S operator~
560       IFK::BitwiseXor,       // ?T operator^
561       IFK::BitwiseOr,        // ?U operator|
562       IFK::LogicalAnd,       // ?V operator&&
563       IFK::LogicalOr,        // ?W operator||
564       IFK::TimesEqual,       // ?X operator*=
565       IFK::PlusEqual,        // ?Y operator+=
566       IFK::MinusEqual,       // ?Z operator-=
567   };
568   static IFK Under[36] = {
569       IFK::DivEqual,           // ?_0 operator/=
570       IFK::ModEqual,           // ?_1 operator%=
571       IFK::RshEqual,           // ?_2 operator>>=
572       IFK::LshEqual,           // ?_3 operator<<=
573       IFK::BitwiseAndEqual,    // ?_4 operator&=
574       IFK::BitwiseOrEqual,     // ?_5 operator|=
575       IFK::BitwiseXorEqual,    // ?_6 operator^=
576       IFK::None,               // ?_7 # vftable
577       IFK::None,               // ?_8 # vbtable
578       IFK::None,               // ?_9 # vcall
579       IFK::None,               // ?_A # typeof
580       IFK::None,               // ?_B # local static guard
581       IFK::None,               // ?_C # string literal
582       IFK::VbaseDtor,          // ?_D # vbase destructor
583       IFK::VecDelDtor,         // ?_E # vector deleting destructor
584       IFK::DefaultCtorClosure, // ?_F # default constructor closure
585       IFK::ScalarDelDtor,      // ?_G # scalar deleting destructor
586       IFK::VecCtorIter,        // ?_H # vector constructor iterator
587       IFK::VecDtorIter,        // ?_I # vector destructor iterator
588       IFK::VecVbaseCtorIter,   // ?_J # vector vbase constructor iterator
589       IFK::VdispMap,           // ?_K # virtual displacement map
590       IFK::EHVecCtorIter,      // ?_L # eh vector constructor iterator
591       IFK::EHVecDtorIter,      // ?_M # eh vector destructor iterator
592       IFK::EHVecVbaseCtorIter, // ?_N # eh vector vbase constructor iterator
593       IFK::CopyCtorClosure,    // ?_O # copy constructor closure
594       IFK::None,               // ?_P<name> # udt returning <name>
595       IFK::None,               // ?_Q # <unknown>
596       IFK::None,               // ?_R0 - ?_R4 # RTTI Codes
597       IFK::None,               // ?_S # local vftable
598       IFK::LocalVftableCtorClosure, // ?_T # local vftable constructor closure
599       IFK::ArrayNew,                // ?_U operator new[]
600       IFK::ArrayDelete,             // ?_V operator delete[]
601       IFK::None,                    // ?_W <unused>
602       IFK::None,                    // ?_X <unused>
603       IFK::None,                    // ?_Y <unused>
604       IFK::None,                    // ?_Z <unused>
605   };
606   static IFK DoubleUnder[36] = {
607       IFK::None,                       // ?__0 <unused>
608       IFK::None,                       // ?__1 <unused>
609       IFK::None,                       // ?__2 <unused>
610       IFK::None,                       // ?__3 <unused>
611       IFK::None,                       // ?__4 <unused>
612       IFK::None,                       // ?__5 <unused>
613       IFK::None,                       // ?__6 <unused>
614       IFK::None,                       // ?__7 <unused>
615       IFK::None,                       // ?__8 <unused>
616       IFK::None,                       // ?__9 <unused>
617       IFK::ManVectorCtorIter,          // ?__A managed vector ctor iterator
618       IFK::ManVectorDtorIter,          // ?__B managed vector dtor iterator
619       IFK::EHVectorCopyCtorIter,       // ?__C EH vector copy ctor iterator
620       IFK::EHVectorVbaseCopyCtorIter,  // ?__D EH vector vbase copy ctor iter
621       IFK::None,                       // ?__E dynamic initializer for `T'
622       IFK::None,                       // ?__F dynamic atexit destructor for `T'
623       IFK::VectorCopyCtorIter,         // ?__G vector copy constructor iter
624       IFK::VectorVbaseCopyCtorIter,    // ?__H vector vbase copy ctor iter
625       IFK::ManVectorVbaseCopyCtorIter, // ?__I managed vector vbase copy ctor
626                                        // iter
627       IFK::None,                       // ?__J local static thread guard
628       IFK::None,                       // ?__K operator ""_name
629       IFK::CoAwait,                    // ?__L operator co_await
630       IFK::Spaceship,                  // ?__M operator<=>
631       IFK::None,                       // ?__N <unused>
632       IFK::None,                       // ?__O <unused>
633       IFK::None,                       // ?__P <unused>
634       IFK::None,                       // ?__Q <unused>
635       IFK::None,                       // ?__R <unused>
636       IFK::None,                       // ?__S <unused>
637       IFK::None,                       // ?__T <unused>
638       IFK::None,                       // ?__U <unused>
639       IFK::None,                       // ?__V <unused>
640       IFK::None,                       // ?__W <unused>
641       IFK::None,                       // ?__X <unused>
642       IFK::None,                       // ?__Y <unused>
643       IFK::None,                       // ?__Z <unused>
644   };
645 
646   int Index = (CH >= '0' && CH <= '9') ? (CH - '0') : (CH - 'A' + 10);
647   switch (Group) {
648   case FunctionIdentifierCodeGroup::Basic:
649     return Basic[Index];
650   case FunctionIdentifierCodeGroup::Under:
651     return Under[Index];
652   case FunctionIdentifierCodeGroup::DoubleUnder:
653     return DoubleUnder[Index];
654   }
655   DEMANGLE_UNREACHABLE;
656 }
657 
658 IdentifierNode *
659 Demangler::demangleFunctionIdentifierCode(StringView &MangledName,
660                                           FunctionIdentifierCodeGroup Group) {
661   if (MangledName.empty()) {
662     Error = true;
663     return nullptr;
664   }
665   switch (Group) {
666   case FunctionIdentifierCodeGroup::Basic:
667     switch (char CH = MangledName.popFront()) {
668     case '0':
669     case '1':
670       return demangleStructorIdentifier(MangledName, CH == '1');
671     case 'B':
672       return demangleConversionOperatorIdentifier(MangledName);
673     default:
674       return Arena.alloc<IntrinsicFunctionIdentifierNode>(
675           translateIntrinsicFunctionCode(CH, Group));
676     }
677     break;
678   case FunctionIdentifierCodeGroup::Under:
679     return Arena.alloc<IntrinsicFunctionIdentifierNode>(
680         translateIntrinsicFunctionCode(MangledName.popFront(), Group));
681   case FunctionIdentifierCodeGroup::DoubleUnder:
682     switch (char CH = MangledName.popFront()) {
683     case 'K':
684       return demangleLiteralOperatorIdentifier(MangledName);
685     default:
686       return Arena.alloc<IntrinsicFunctionIdentifierNode>(
687           translateIntrinsicFunctionCode(CH, Group));
688     }
689   }
690 
691   DEMANGLE_UNREACHABLE;
692 }
693 
694 SymbolNode *Demangler::demangleEncodedSymbol(StringView &MangledName,
695                                              QualifiedNameNode *Name) {
696   if (MangledName.empty()) {
697     Error = true;
698     return nullptr;
699   }
700 
701   // Read a variable.
702   switch (MangledName.front()) {
703   case '0':
704   case '1':
705   case '2':
706   case '3':
707   case '4': {
708     StorageClass SC = demangleVariableStorageClass(MangledName);
709     return demangleVariableEncoding(MangledName, SC);
710   }
711   }
712   FunctionSymbolNode *FSN = demangleFunctionEncoding(MangledName);
713 
714   IdentifierNode *UQN = Name->getUnqualifiedIdentifier();
715   if (UQN->kind() == NodeKind::ConversionOperatorIdentifier) {
716     ConversionOperatorIdentifierNode *COIN =
717         static_cast<ConversionOperatorIdentifierNode *>(UQN);
718     if (FSN)
719       COIN->TargetType = FSN->Signature->ReturnType;
720   }
721   return FSN;
722 }
723 
724 SymbolNode *Demangler::demangleDeclarator(StringView &MangledName) {
725   // What follows is a main symbol name. This may include namespaces or class
726   // back references.
727   QualifiedNameNode *QN = demangleFullyQualifiedSymbolName(MangledName);
728   if (Error)
729     return nullptr;
730 
731   SymbolNode *Symbol = demangleEncodedSymbol(MangledName, QN);
732   if (Error)
733     return nullptr;
734   Symbol->Name = QN;
735 
736   IdentifierNode *UQN = QN->getUnqualifiedIdentifier();
737   if (UQN->kind() == NodeKind::ConversionOperatorIdentifier) {
738     ConversionOperatorIdentifierNode *COIN =
739         static_cast<ConversionOperatorIdentifierNode *>(UQN);
740     if (!COIN->TargetType) {
741       Error = true;
742       return nullptr;
743     }
744   }
745   return Symbol;
746 }
747 
748 // Parser entry point.
749 SymbolNode *Demangler::parse(StringView &MangledName) {
750   // We can't demangle MD5 names, just output them as-is.
751   // Also, MSVC-style mangled symbols must start with '?'.
752   if (MangledName.startsWith("??@")) {
753     // This is an MD5 mangled name.  We can't demangle it, just return the
754     // mangled name.
755     SymbolNode *S = Arena.alloc<SymbolNode>(NodeKind::Md5Symbol);
756     S->Name = synthesizeQualifiedName(Arena, MangledName);
757     return S;
758   }
759 
760   if (!MangledName.startsWith('?')) {
761     Error = true;
762     return nullptr;
763   }
764 
765   MangledName.consumeFront('?');
766 
767   // ?$ is a template instantiation, but all other names that start with ? are
768   // operators / special names.
769   if (SymbolNode *SI = demangleSpecialIntrinsic(MangledName))
770     return SI;
771 
772   return demangleDeclarator(MangledName);
773 }
774 
775 TagTypeNode *Demangler::parseTagUniqueName(StringView &MangledName) {
776   if (!MangledName.consumeFront(".?A"))
777     return nullptr;
778   MangledName.consumeFront(".?A");
779   if (MangledName.empty())
780     return nullptr;
781 
782   return demangleClassType(MangledName);
783 }
784 
785 // <type-encoding> ::= <storage-class> <variable-type>
786 // <storage-class> ::= 0  # private static member
787 //                 ::= 1  # protected static member
788 //                 ::= 2  # public static member
789 //                 ::= 3  # global
790 //                 ::= 4  # static local
791 
792 VariableSymbolNode *Demangler::demangleVariableEncoding(StringView &MangledName,
793                                                         StorageClass SC) {
794   VariableSymbolNode *VSN = Arena.alloc<VariableSymbolNode>();
795 
796   VSN->Type = demangleType(MangledName, QualifierMangleMode::Drop);
797   VSN->SC = SC;
798 
799   if (Error)
800     return nullptr;
801 
802   // <variable-type> ::= <type> <cvr-qualifiers>
803   //                 ::= <type> <pointee-cvr-qualifiers> # pointers, references
804   switch (VSN->Type->kind()) {
805   case NodeKind::PointerType: {
806     PointerTypeNode *PTN = static_cast<PointerTypeNode *>(VSN->Type);
807 
808     Qualifiers ExtraChildQuals = Q_None;
809     PTN->Quals = Qualifiers(VSN->Type->Quals |
810                             demanglePointerExtQualifiers(MangledName));
811 
812     bool IsMember = false;
813     std::tie(ExtraChildQuals, IsMember) = demangleQualifiers(MangledName);
814 
815     if (PTN->ClassParent) {
816       QualifiedNameNode *BackRefName =
817           demangleFullyQualifiedTypeName(MangledName);
818       (void)BackRefName;
819     }
820     PTN->Pointee->Quals = Qualifiers(PTN->Pointee->Quals | ExtraChildQuals);
821 
822     break;
823   }
824   default:
825     VSN->Type->Quals = demangleQualifiers(MangledName).first;
826     break;
827   }
828 
829   return VSN;
830 }
831 
832 // Sometimes numbers are encoded in mangled symbols. For example,
833 // "int (*x)[20]" is a valid C type (x is a pointer to an array of
834 // length 20), so we need some way to embed numbers as part of symbols.
835 // This function parses it.
836 //
837 // <number>               ::= [?] <non-negative integer>
838 //
839 // <non-negative integer> ::= <decimal digit> # when 1 <= Number <= 10
840 //                        ::= <hex digit>+ @  # when Number == 0 or >= 10
841 //
842 // <hex-digit>            ::= [A-P]           # A = 0, B = 1, ...
843 std::pair<uint64_t, bool> Demangler::demangleNumber(StringView &MangledName) {
844   bool IsNegative = MangledName.consumeFront('?');
845 
846   if (startsWithDigit(MangledName)) {
847     uint64_t Ret = MangledName[0] - '0' + 1;
848     MangledName = MangledName.dropFront(1);
849     return {Ret, IsNegative};
850   }
851 
852   uint64_t Ret = 0;
853   for (size_t i = 0; i < MangledName.size(); ++i) {
854     char C = MangledName[i];
855     if (C == '@') {
856       MangledName = MangledName.dropFront(i + 1);
857       return {Ret, IsNegative};
858     }
859     if ('A' <= C && C <= 'P') {
860       Ret = (Ret << 4) + (C - 'A');
861       continue;
862     }
863     break;
864   }
865 
866   Error = true;
867   return {0ULL, false};
868 }
869 
870 uint64_t Demangler::demangleUnsigned(StringView &MangledName) {
871   bool IsNegative = false;
872   uint64_t Number = 0;
873   std::tie(Number, IsNegative) = demangleNumber(MangledName);
874   if (IsNegative)
875     Error = true;
876   return Number;
877 }
878 
879 int64_t Demangler::demangleSigned(StringView &MangledName) {
880   bool IsNegative = false;
881   uint64_t Number = 0;
882   std::tie(Number, IsNegative) = demangleNumber(MangledName);
883   if (Number > INT64_MAX)
884     Error = true;
885   int64_t I = static_cast<int64_t>(Number);
886   return IsNegative ? -I : I;
887 }
888 
889 // First 10 strings can be referenced by special BackReferences ?0, ?1, ..., ?9.
890 // Memorize it.
891 void Demangler::memorizeString(StringView S) {
892   if (Backrefs.NamesCount >= BackrefContext::Max)
893     return;
894   for (size_t i = 0; i < Backrefs.NamesCount; ++i)
895     if (S == Backrefs.Names[i]->Name)
896       return;
897   NamedIdentifierNode *N = Arena.alloc<NamedIdentifierNode>();
898   N->Name = S;
899   Backrefs.Names[Backrefs.NamesCount++] = N;
900 }
901 
902 NamedIdentifierNode *Demangler::demangleBackRefName(StringView &MangledName) {
903   assert(startsWithDigit(MangledName));
904 
905   size_t I = MangledName[0] - '0';
906   if (I >= Backrefs.NamesCount) {
907     Error = true;
908     return nullptr;
909   }
910 
911   MangledName = MangledName.dropFront();
912   return Backrefs.Names[I];
913 }
914 
915 void Demangler::memorizeIdentifier(IdentifierNode *Identifier) {
916   // Render this class template name into a string buffer so that we can
917   // memorize it for the purpose of back-referencing.
918   OutputStream OS;
919   if (!initializeOutputStream(nullptr, nullptr, OS, 1024))
920     // FIXME: Propagate out-of-memory as an error?
921     std::terminate();
922   Identifier->output(OS, OF_Default);
923   OS << '\0';
924   char *Name = OS.getBuffer();
925 
926   StringView Owned = copyString(Name);
927   memorizeString(Owned);
928   std::free(Name);
929 }
930 
931 IdentifierNode *
932 Demangler::demangleTemplateInstantiationName(StringView &MangledName,
933                                              NameBackrefBehavior NBB) {
934   assert(MangledName.startsWith("?$"));
935   MangledName.consumeFront("?$");
936 
937   BackrefContext OuterContext;
938   std::swap(OuterContext, Backrefs);
939 
940   IdentifierNode *Identifier =
941       demangleUnqualifiedSymbolName(MangledName, NBB_Simple);
942   if (!Error)
943     Identifier->TemplateParams = demangleTemplateParameterList(MangledName);
944 
945   std::swap(OuterContext, Backrefs);
946   if (Error)
947     return nullptr;
948 
949   if (NBB & NBB_Template) {
950     // NBB_Template is only set for types and non-leaf names ("a::" in "a::b").
951     // Structors and conversion operators only makes sense in a leaf name, so
952     // reject them in NBB_Template contexts.
953     if (Identifier->kind() == NodeKind::ConversionOperatorIdentifier ||
954         Identifier->kind() == NodeKind::StructorIdentifier) {
955       Error = true;
956       return nullptr;
957     }
958 
959     memorizeIdentifier(Identifier);
960   }
961 
962   return Identifier;
963 }
964 
965 NamedIdentifierNode *Demangler::demangleSimpleName(StringView &MangledName,
966                                                    bool Memorize) {
967   StringView S = demangleSimpleString(MangledName, Memorize);
968   if (Error)
969     return nullptr;
970 
971   NamedIdentifierNode *Name = Arena.alloc<NamedIdentifierNode>();
972   Name->Name = S;
973   return Name;
974 }
975 
976 static bool isRebasedHexDigit(char C) { return (C >= 'A' && C <= 'P'); }
977 
978 static uint8_t rebasedHexDigitToNumber(char C) {
979   assert(isRebasedHexDigit(C));
980   return (C <= 'J') ? (C - 'A') : (10 + C - 'K');
981 }
982 
983 uint8_t Demangler::demangleCharLiteral(StringView &MangledName) {
984   assert(!MangledName.empty());
985   if (!MangledName.startsWith('?'))
986     return MangledName.popFront();
987 
988   MangledName = MangledName.dropFront();
989   if (MangledName.empty())
990     goto CharLiteralError;
991 
992   if (MangledName.consumeFront('$')) {
993     // Two hex digits
994     if (MangledName.size() < 2)
995       goto CharLiteralError;
996     StringView Nibbles = MangledName.substr(0, 2);
997     if (!isRebasedHexDigit(Nibbles[0]) || !isRebasedHexDigit(Nibbles[1]))
998       goto CharLiteralError;
999     // Don't append the null terminator.
1000     uint8_t C1 = rebasedHexDigitToNumber(Nibbles[0]);
1001     uint8_t C2 = rebasedHexDigitToNumber(Nibbles[1]);
1002     MangledName = MangledName.dropFront(2);
1003     return (C1 << 4) | C2;
1004   }
1005 
1006   if (startsWithDigit(MangledName)) {
1007     const char *Lookup = ",/\\:. \n\t'-";
1008     char C = Lookup[MangledName[0] - '0'];
1009     MangledName = MangledName.dropFront();
1010     return C;
1011   }
1012 
1013   if (MangledName[0] >= 'a' && MangledName[0] <= 'z') {
1014     char Lookup[26] = {'\xE1', '\xE2', '\xE3', '\xE4', '\xE5', '\xE6', '\xE7',
1015                        '\xE8', '\xE9', '\xEA', '\xEB', '\xEC', '\xED', '\xEE',
1016                        '\xEF', '\xF0', '\xF1', '\xF2', '\xF3', '\xF4', '\xF5',
1017                        '\xF6', '\xF7', '\xF8', '\xF9', '\xFA'};
1018     char C = Lookup[MangledName[0] - 'a'];
1019     MangledName = MangledName.dropFront();
1020     return C;
1021   }
1022 
1023   if (MangledName[0] >= 'A' && MangledName[0] <= 'Z') {
1024     char Lookup[26] = {'\xC1', '\xC2', '\xC3', '\xC4', '\xC5', '\xC6', '\xC7',
1025                        '\xC8', '\xC9', '\xCA', '\xCB', '\xCC', '\xCD', '\xCE',
1026                        '\xCF', '\xD0', '\xD1', '\xD2', '\xD3', '\xD4', '\xD5',
1027                        '\xD6', '\xD7', '\xD8', '\xD9', '\xDA'};
1028     char C = Lookup[MangledName[0] - 'A'];
1029     MangledName = MangledName.dropFront();
1030     return C;
1031   }
1032 
1033 CharLiteralError:
1034   Error = true;
1035   return '\0';
1036 }
1037 
1038 wchar_t Demangler::demangleWcharLiteral(StringView &MangledName) {
1039   uint8_t C1, C2;
1040 
1041   C1 = demangleCharLiteral(MangledName);
1042   if (Error || MangledName.empty())
1043     goto WCharLiteralError;
1044   C2 = demangleCharLiteral(MangledName);
1045   if (Error)
1046     goto WCharLiteralError;
1047 
1048   return ((wchar_t)C1 << 8) | (wchar_t)C2;
1049 
1050 WCharLiteralError:
1051   Error = true;
1052   return L'\0';
1053 }
1054 
1055 static void writeHexDigit(char *Buffer, uint8_t Digit) {
1056   assert(Digit <= 15);
1057   *Buffer = (Digit < 10) ? ('0' + Digit) : ('A' + Digit - 10);
1058 }
1059 
1060 static void outputHex(OutputStream &OS, unsigned C) {
1061   if (C == 0) {
1062     OS << "\\x00";
1063     return;
1064   }
1065   // It's easier to do the math if we can work from right to left, but we need
1066   // to print the numbers from left to right.  So render this into a temporary
1067   // buffer first, then output the temporary buffer.  Each byte is of the form
1068   // \xAB, which means that each byte needs 4 characters.  Since there are at
1069   // most 4 bytes, we need a 4*4+1 = 17 character temporary buffer.
1070   char TempBuffer[17];
1071 
1072   ::memset(TempBuffer, 0, sizeof(TempBuffer));
1073   constexpr int MaxPos = sizeof(TempBuffer) - 1;
1074 
1075   int Pos = MaxPos - 1; // TempBuffer[MaxPos] is the terminating \0.
1076   while (C != 0) {
1077     for (int I = 0; I < 2; ++I) {
1078       writeHexDigit(&TempBuffer[Pos--], C % 16);
1079       C /= 16;
1080     }
1081   }
1082   TempBuffer[Pos--] = 'x';
1083   assert(Pos >= 0);
1084   TempBuffer[Pos--] = '\\';
1085   OS << StringView(&TempBuffer[Pos + 1]);
1086 }
1087 
1088 static void outputEscapedChar(OutputStream &OS, unsigned C) {
1089   switch (C) {
1090   case '\0': // nul
1091     OS << "\\0";
1092     return;
1093   case '\'': // single quote
1094     OS << "\\\'";
1095     return;
1096   case '\"': // double quote
1097     OS << "\\\"";
1098     return;
1099   case '\\': // backslash
1100     OS << "\\\\";
1101     return;
1102   case '\a': // bell
1103     OS << "\\a";
1104     return;
1105   case '\b': // backspace
1106     OS << "\\b";
1107     return;
1108   case '\f': // form feed
1109     OS << "\\f";
1110     return;
1111   case '\n': // new line
1112     OS << "\\n";
1113     return;
1114   case '\r': // carriage return
1115     OS << "\\r";
1116     return;
1117   case '\t': // tab
1118     OS << "\\t";
1119     return;
1120   case '\v': // vertical tab
1121     OS << "\\v";
1122     return;
1123   default:
1124     break;
1125   }
1126 
1127   if (C > 0x1F && C < 0x7F) {
1128     // Standard ascii char.
1129     OS << (char)C;
1130     return;
1131   }
1132 
1133   outputHex(OS, C);
1134 }
1135 
1136 static unsigned countTrailingNullBytes(const uint8_t *StringBytes, int Length) {
1137   const uint8_t *End = StringBytes + Length - 1;
1138   unsigned Count = 0;
1139   while (Length > 0 && *End == 0) {
1140     --Length;
1141     --End;
1142     ++Count;
1143   }
1144   return Count;
1145 }
1146 
1147 static unsigned countEmbeddedNulls(const uint8_t *StringBytes,
1148                                    unsigned Length) {
1149   unsigned Result = 0;
1150   for (unsigned I = 0; I < Length; ++I) {
1151     if (*StringBytes++ == 0)
1152       ++Result;
1153   }
1154   return Result;
1155 }
1156 
1157 static unsigned guessCharByteSize(const uint8_t *StringBytes, unsigned NumChars,
1158                                   unsigned NumBytes) {
1159   assert(NumBytes > 0);
1160 
1161   // If the number of bytes is odd, this is guaranteed to be a char string.
1162   if (NumBytes % 2 == 1)
1163     return 1;
1164 
1165   // All strings can encode at most 32 bytes of data.  If it's less than that,
1166   // then we encoded the entire string.  In this case we check for a 1-byte,
1167   // 2-byte, or 4-byte null terminator.
1168   if (NumBytes < 32) {
1169     unsigned TrailingNulls = countTrailingNullBytes(StringBytes, NumChars);
1170     if (TrailingNulls >= 4 && NumBytes % 4 == 0)
1171       return 4;
1172     if (TrailingNulls >= 2)
1173       return 2;
1174     return 1;
1175   }
1176 
1177   // The whole string was not able to be encoded.  Try to look at embedded null
1178   // terminators to guess.  The heuristic is that we count all embedded null
1179   // terminators.  If more than 2/3 are null, it's a char32.  If more than 1/3
1180   // are null, it's a char16.  Otherwise it's a char8.  This obviously isn't
1181   // perfect and is biased towards languages that have ascii alphabets, but this
1182   // was always going to be best effort since the encoding is lossy.
1183   unsigned Nulls = countEmbeddedNulls(StringBytes, NumChars);
1184   if (Nulls >= 2 * NumChars / 3 && NumBytes % 4 == 0)
1185     return 4;
1186   if (Nulls >= NumChars / 3)
1187     return 2;
1188   return 1;
1189 }
1190 
1191 static unsigned decodeMultiByteChar(const uint8_t *StringBytes,
1192                                     unsigned CharIndex, unsigned CharBytes) {
1193   assert(CharBytes == 1 || CharBytes == 2 || CharBytes == 4);
1194   unsigned Offset = CharIndex * CharBytes;
1195   unsigned Result = 0;
1196   StringBytes = StringBytes + Offset;
1197   for (unsigned I = 0; I < CharBytes; ++I) {
1198     unsigned C = static_cast<unsigned>(StringBytes[I]);
1199     Result |= C << (8 * I);
1200   }
1201   return Result;
1202 }
1203 
1204 FunctionSymbolNode *Demangler::demangleVcallThunkNode(StringView &MangledName) {
1205   FunctionSymbolNode *FSN = Arena.alloc<FunctionSymbolNode>();
1206   VcallThunkIdentifierNode *VTIN = Arena.alloc<VcallThunkIdentifierNode>();
1207   FSN->Signature = Arena.alloc<ThunkSignatureNode>();
1208   FSN->Signature->FunctionClass = FC_NoParameterList;
1209 
1210   FSN->Name = demangleNameScopeChain(MangledName, VTIN);
1211   if (!Error)
1212     Error = !MangledName.consumeFront("$B");
1213   if (!Error)
1214     VTIN->OffsetInVTable = demangleUnsigned(MangledName);
1215   if (!Error)
1216     Error = !MangledName.consumeFront('A');
1217   if (!Error)
1218     FSN->Signature->CallConvention = demangleCallingConvention(MangledName);
1219   return (Error) ? nullptr : FSN;
1220 }
1221 
1222 EncodedStringLiteralNode *
1223 Demangler::demangleStringLiteral(StringView &MangledName) {
1224   // This function uses goto, so declare all variables up front.
1225   OutputStream OS;
1226   StringView CRC;
1227   uint64_t StringByteSize;
1228   bool IsWcharT = false;
1229   bool IsNegative = false;
1230   size_t CrcEndPos = 0;
1231   char *ResultBuffer = nullptr;
1232 
1233   EncodedStringLiteralNode *Result = Arena.alloc<EncodedStringLiteralNode>();
1234 
1235   // Must happen before the first `goto StringLiteralError`.
1236   if (!initializeOutputStream(nullptr, nullptr, OS, 1024))
1237     // FIXME: Propagate out-of-memory as an error?
1238     std::terminate();
1239 
1240   // Prefix indicating the beginning of a string literal
1241   if (!MangledName.consumeFront("@_"))
1242     goto StringLiteralError;
1243   if (MangledName.empty())
1244     goto StringLiteralError;
1245 
1246   // Char Type (regular or wchar_t)
1247   switch (MangledName.popFront()) {
1248   case '1':
1249     IsWcharT = true;
1250     DEMANGLE_FALLTHROUGH;
1251   case '0':
1252     break;
1253   default:
1254     goto StringLiteralError;
1255   }
1256 
1257   // Encoded Length
1258   std::tie(StringByteSize, IsNegative) = demangleNumber(MangledName);
1259   if (Error || IsNegative || StringByteSize < (IsWcharT ? 2 : 1))
1260     goto StringLiteralError;
1261 
1262   // CRC 32 (always 8 characters plus a terminator)
1263   CrcEndPos = MangledName.find('@');
1264   if (CrcEndPos == StringView::npos)
1265     goto StringLiteralError;
1266   CRC = MangledName.substr(0, CrcEndPos);
1267   MangledName = MangledName.dropFront(CrcEndPos + 1);
1268   if (MangledName.empty())
1269     goto StringLiteralError;
1270 
1271   if (IsWcharT) {
1272     Result->Char = CharKind::Wchar;
1273     if (StringByteSize > 64)
1274       Result->IsTruncated = true;
1275 
1276     while (!MangledName.consumeFront('@')) {
1277       if (MangledName.size() < 2)
1278         goto StringLiteralError;
1279       wchar_t W = demangleWcharLiteral(MangledName);
1280       if (StringByteSize != 2 || Result->IsTruncated)
1281         outputEscapedChar(OS, W);
1282       StringByteSize -= 2;
1283       if (Error)
1284         goto StringLiteralError;
1285     }
1286   } else {
1287     // The max byte length is actually 32, but some compilers mangled strings
1288     // incorrectly, so we have to assume it can go higher.
1289     constexpr unsigned MaxStringByteLength = 32 * 4;
1290     uint8_t StringBytes[MaxStringByteLength];
1291 
1292     unsigned BytesDecoded = 0;
1293     while (!MangledName.consumeFront('@')) {
1294       if (MangledName.size() < 1 || BytesDecoded >= MaxStringByteLength)
1295         goto StringLiteralError;
1296       StringBytes[BytesDecoded++] = demangleCharLiteral(MangledName);
1297     }
1298 
1299     if (StringByteSize > BytesDecoded)
1300       Result->IsTruncated = true;
1301 
1302     unsigned CharBytes =
1303         guessCharByteSize(StringBytes, BytesDecoded, StringByteSize);
1304     assert(StringByteSize % CharBytes == 0);
1305     switch (CharBytes) {
1306     case 1:
1307       Result->Char = CharKind::Char;
1308       break;
1309     case 2:
1310       Result->Char = CharKind::Char16;
1311       break;
1312     case 4:
1313       Result->Char = CharKind::Char32;
1314       break;
1315     default:
1316       DEMANGLE_UNREACHABLE;
1317     }
1318     const unsigned NumChars = BytesDecoded / CharBytes;
1319     for (unsigned CharIndex = 0; CharIndex < NumChars; ++CharIndex) {
1320       unsigned NextChar =
1321           decodeMultiByteChar(StringBytes, CharIndex, CharBytes);
1322       if (CharIndex + 1 < NumChars || Result->IsTruncated)
1323         outputEscapedChar(OS, NextChar);
1324     }
1325   }
1326 
1327   OS << '\0';
1328   ResultBuffer = OS.getBuffer();
1329   Result->DecodedString = copyString(ResultBuffer);
1330   std::free(ResultBuffer);
1331   return Result;
1332 
1333 StringLiteralError:
1334   Error = true;
1335   std::free(OS.getBuffer());
1336   return nullptr;
1337 }
1338 
1339 // Returns MangledName's prefix before the first '@', or an error if
1340 // MangledName contains no '@' or the prefix has length 0.
1341 StringView Demangler::demangleSimpleString(StringView &MangledName,
1342                                            bool Memorize) {
1343   StringView S;
1344   for (size_t i = 0; i < MangledName.size(); ++i) {
1345     if (MangledName[i] != '@')
1346       continue;
1347     if (i == 0)
1348       break;
1349     S = MangledName.substr(0, i);
1350     MangledName = MangledName.dropFront(i + 1);
1351 
1352     if (Memorize)
1353       memorizeString(S);
1354     return S;
1355   }
1356 
1357   Error = true;
1358   return {};
1359 }
1360 
1361 NamedIdentifierNode *
1362 Demangler::demangleAnonymousNamespaceName(StringView &MangledName) {
1363   assert(MangledName.startsWith("?A"));
1364   MangledName.consumeFront("?A");
1365 
1366   NamedIdentifierNode *Node = Arena.alloc<NamedIdentifierNode>();
1367   Node->Name = "`anonymous namespace'";
1368   size_t EndPos = MangledName.find('@');
1369   if (EndPos == StringView::npos) {
1370     Error = true;
1371     return nullptr;
1372   }
1373   StringView NamespaceKey = MangledName.substr(0, EndPos);
1374   memorizeString(NamespaceKey);
1375   MangledName = MangledName.substr(EndPos + 1);
1376   return Node;
1377 }
1378 
1379 NamedIdentifierNode *
1380 Demangler::demangleLocallyScopedNamePiece(StringView &MangledName) {
1381   assert(startsWithLocalScopePattern(MangledName));
1382 
1383   NamedIdentifierNode *Identifier = Arena.alloc<NamedIdentifierNode>();
1384   MangledName.consumeFront('?');
1385   uint64_t Number = 0;
1386   bool IsNegative = false;
1387   std::tie(Number, IsNegative) = demangleNumber(MangledName);
1388   assert(!IsNegative);
1389 
1390   // One ? to terminate the number
1391   MangledName.consumeFront('?');
1392 
1393   assert(!Error);
1394   Node *Scope = parse(MangledName);
1395   if (Error)
1396     return nullptr;
1397 
1398   // Render the parent symbol's name into a buffer.
1399   OutputStream OS;
1400   if (!initializeOutputStream(nullptr, nullptr, OS, 1024))
1401     // FIXME: Propagate out-of-memory as an error?
1402     std::terminate();
1403   OS << '`';
1404   Scope->output(OS, OF_Default);
1405   OS << '\'';
1406   OS << "::`" << Number << "'";
1407   OS << '\0';
1408   char *Result = OS.getBuffer();
1409   Identifier->Name = copyString(Result);
1410   std::free(Result);
1411   return Identifier;
1412 }
1413 
1414 // Parses a type name in the form of A@B@C@@ which represents C::B::A.
1415 QualifiedNameNode *
1416 Demangler::demangleFullyQualifiedTypeName(StringView &MangledName) {
1417   IdentifierNode *Identifier =
1418       demangleUnqualifiedTypeName(MangledName, /*Memorize=*/true);
1419   if (Error)
1420     return nullptr;
1421   assert(Identifier);
1422 
1423   QualifiedNameNode *QN = demangleNameScopeChain(MangledName, Identifier);
1424   if (Error)
1425     return nullptr;
1426   assert(QN);
1427   return QN;
1428 }
1429 
1430 // Parses a symbol name in the form of A@B@C@@ which represents C::B::A.
1431 // Symbol names have slightly different rules regarding what can appear
1432 // so we separate out the implementations for flexibility.
1433 QualifiedNameNode *
1434 Demangler::demangleFullyQualifiedSymbolName(StringView &MangledName) {
1435   // This is the final component of a symbol name (i.e. the leftmost component
1436   // of a mangled name.  Since the only possible template instantiation that
1437   // can appear in this context is a function template, and since those are
1438   // not saved for the purposes of name backreferences, only backref simple
1439   // names.
1440   IdentifierNode *Identifier =
1441       demangleUnqualifiedSymbolName(MangledName, NBB_Simple);
1442   if (Error)
1443     return nullptr;
1444 
1445   QualifiedNameNode *QN = demangleNameScopeChain(MangledName, Identifier);
1446   if (Error)
1447     return nullptr;
1448 
1449   if (Identifier->kind() == NodeKind::StructorIdentifier) {
1450     if (QN->Components->Count < 2) {
1451       Error = true;
1452       return nullptr;
1453     }
1454     StructorIdentifierNode *SIN =
1455         static_cast<StructorIdentifierNode *>(Identifier);
1456     Node *ClassNode = QN->Components->Nodes[QN->Components->Count - 2];
1457     SIN->Class = static_cast<IdentifierNode *>(ClassNode);
1458   }
1459   assert(QN);
1460   return QN;
1461 }
1462 
1463 IdentifierNode *Demangler::demangleUnqualifiedTypeName(StringView &MangledName,
1464                                                        bool Memorize) {
1465   // An inner-most name can be a back-reference, because a fully-qualified name
1466   // (e.g. Scope + Inner) can contain other fully qualified names inside of
1467   // them (for example template parameters), and these nested parameters can
1468   // refer to previously mangled types.
1469   if (startsWithDigit(MangledName))
1470     return demangleBackRefName(MangledName);
1471 
1472   if (MangledName.startsWith("?$"))
1473     return demangleTemplateInstantiationName(MangledName, NBB_Template);
1474 
1475   return demangleSimpleName(MangledName, Memorize);
1476 }
1477 
1478 IdentifierNode *
1479 Demangler::demangleUnqualifiedSymbolName(StringView &MangledName,
1480                                          NameBackrefBehavior NBB) {
1481   if (startsWithDigit(MangledName))
1482     return demangleBackRefName(MangledName);
1483   if (MangledName.startsWith("?$"))
1484     return demangleTemplateInstantiationName(MangledName, NBB);
1485   if (MangledName.startsWith('?'))
1486     return demangleFunctionIdentifierCode(MangledName);
1487   return demangleSimpleName(MangledName, /*Memorize=*/(NBB & NBB_Simple) != 0);
1488 }
1489 
1490 IdentifierNode *Demangler::demangleNameScopePiece(StringView &MangledName) {
1491   if (startsWithDigit(MangledName))
1492     return demangleBackRefName(MangledName);
1493 
1494   if (MangledName.startsWith("?$"))
1495     return demangleTemplateInstantiationName(MangledName, NBB_Template);
1496 
1497   if (MangledName.startsWith("?A"))
1498     return demangleAnonymousNamespaceName(MangledName);
1499 
1500   if (startsWithLocalScopePattern(MangledName))
1501     return demangleLocallyScopedNamePiece(MangledName);
1502 
1503   return demangleSimpleName(MangledName, /*Memorize=*/true);
1504 }
1505 
1506 static NodeArrayNode *nodeListToNodeArray(ArenaAllocator &Arena, NodeList *Head,
1507                                           size_t Count) {
1508   NodeArrayNode *N = Arena.alloc<NodeArrayNode>();
1509   N->Count = Count;
1510   N->Nodes = Arena.allocArray<Node *>(Count);
1511   for (size_t I = 0; I < Count; ++I) {
1512     N->Nodes[I] = Head->N;
1513     Head = Head->Next;
1514   }
1515   return N;
1516 }
1517 
1518 QualifiedNameNode *
1519 Demangler::demangleNameScopeChain(StringView &MangledName,
1520                                   IdentifierNode *UnqualifiedName) {
1521   NodeList *Head = Arena.alloc<NodeList>();
1522 
1523   Head->N = UnqualifiedName;
1524 
1525   size_t Count = 1;
1526   while (!MangledName.consumeFront("@")) {
1527     ++Count;
1528     NodeList *NewHead = Arena.alloc<NodeList>();
1529     NewHead->Next = Head;
1530     Head = NewHead;
1531 
1532     if (MangledName.empty()) {
1533       Error = true;
1534       return nullptr;
1535     }
1536 
1537     assert(!Error);
1538     IdentifierNode *Elem = demangleNameScopePiece(MangledName);
1539     if (Error)
1540       return nullptr;
1541 
1542     Head->N = Elem;
1543   }
1544 
1545   QualifiedNameNode *QN = Arena.alloc<QualifiedNameNode>();
1546   QN->Components = nodeListToNodeArray(Arena, Head, Count);
1547   return QN;
1548 }
1549 
1550 FuncClass Demangler::demangleFunctionClass(StringView &MangledName) {
1551   switch (MangledName.popFront()) {
1552   case '9':
1553     return FuncClass(FC_ExternC | FC_NoParameterList);
1554   case 'A':
1555     return FC_Private;
1556   case 'B':
1557     return FuncClass(FC_Private | FC_Far);
1558   case 'C':
1559     return FuncClass(FC_Private | FC_Static);
1560   case 'D':
1561     return FuncClass(FC_Private | FC_Static);
1562   case 'E':
1563     return FuncClass(FC_Private | FC_Virtual);
1564   case 'F':
1565     return FuncClass(FC_Private | FC_Virtual);
1566   case 'G':
1567     return FuncClass(FC_Private | FC_StaticThisAdjust);
1568   case 'H':
1569     return FuncClass(FC_Private | FC_StaticThisAdjust | FC_Far);
1570   case 'I':
1571     return FuncClass(FC_Protected);
1572   case 'J':
1573     return FuncClass(FC_Protected | FC_Far);
1574   case 'K':
1575     return FuncClass(FC_Protected | FC_Static);
1576   case 'L':
1577     return FuncClass(FC_Protected | FC_Static | FC_Far);
1578   case 'M':
1579     return FuncClass(FC_Protected | FC_Virtual);
1580   case 'N':
1581     return FuncClass(FC_Protected | FC_Virtual | FC_Far);
1582   case 'O':
1583     return FuncClass(FC_Protected | FC_Virtual | FC_StaticThisAdjust);
1584   case 'P':
1585     return FuncClass(FC_Protected | FC_Virtual | FC_StaticThisAdjust | FC_Far);
1586   case 'Q':
1587     return FuncClass(FC_Public);
1588   case 'R':
1589     return FuncClass(FC_Public | FC_Far);
1590   case 'S':
1591     return FuncClass(FC_Public | FC_Static);
1592   case 'T':
1593     return FuncClass(FC_Public | FC_Static | FC_Far);
1594   case 'U':
1595     return FuncClass(FC_Public | FC_Virtual);
1596   case 'V':
1597     return FuncClass(FC_Public | FC_Virtual | FC_Far);
1598   case 'W':
1599     return FuncClass(FC_Public | FC_Virtual | FC_StaticThisAdjust);
1600   case 'X':
1601     return FuncClass(FC_Public | FC_Virtual | FC_StaticThisAdjust | FC_Far);
1602   case 'Y':
1603     return FuncClass(FC_Global);
1604   case 'Z':
1605     return FuncClass(FC_Global | FC_Far);
1606   case '$': {
1607     FuncClass VFlag = FC_VirtualThisAdjust;
1608     if (MangledName.consumeFront('R'))
1609       VFlag = FuncClass(VFlag | FC_VirtualThisAdjustEx);
1610     if (MangledName.empty())
1611       break;
1612     switch (MangledName.popFront()) {
1613     case '0':
1614       return FuncClass(FC_Private | FC_Virtual | VFlag);
1615     case '1':
1616       return FuncClass(FC_Private | FC_Virtual | VFlag | FC_Far);
1617     case '2':
1618       return FuncClass(FC_Protected | FC_Virtual | VFlag);
1619     case '3':
1620       return FuncClass(FC_Protected | FC_Virtual | VFlag | FC_Far);
1621     case '4':
1622       return FuncClass(FC_Public | FC_Virtual | VFlag);
1623     case '5':
1624       return FuncClass(FC_Public | FC_Virtual | VFlag | FC_Far);
1625     }
1626   }
1627   }
1628 
1629   Error = true;
1630   return FC_Public;
1631 }
1632 
1633 CallingConv Demangler::demangleCallingConvention(StringView &MangledName) {
1634   if (MangledName.empty()) {
1635     Error = true;
1636     return CallingConv::None;
1637   }
1638 
1639   switch (MangledName.popFront()) {
1640   case 'A':
1641   case 'B':
1642     return CallingConv::Cdecl;
1643   case 'C':
1644   case 'D':
1645     return CallingConv::Pascal;
1646   case 'E':
1647   case 'F':
1648     return CallingConv::Thiscall;
1649   case 'G':
1650   case 'H':
1651     return CallingConv::Stdcall;
1652   case 'I':
1653   case 'J':
1654     return CallingConv::Fastcall;
1655   case 'M':
1656   case 'N':
1657     return CallingConv::Clrcall;
1658   case 'O':
1659   case 'P':
1660     return CallingConv::Eabi;
1661   case 'Q':
1662     return CallingConv::Vectorcall;
1663   }
1664 
1665   return CallingConv::None;
1666 }
1667 
1668 StorageClass Demangler::demangleVariableStorageClass(StringView &MangledName) {
1669   assert(std::isdigit(MangledName.front()));
1670 
1671   switch (MangledName.popFront()) {
1672   case '0':
1673     return StorageClass::PrivateStatic;
1674   case '1':
1675     return StorageClass::ProtectedStatic;
1676   case '2':
1677     return StorageClass::PublicStatic;
1678   case '3':
1679     return StorageClass::Global;
1680   case '4':
1681     return StorageClass::FunctionLocalStatic;
1682   }
1683   Error = true;
1684   return StorageClass::None;
1685 }
1686 
1687 std::pair<Qualifiers, bool>
1688 Demangler::demangleQualifiers(StringView &MangledName) {
1689   if (MangledName.empty()) {
1690     Error = true;
1691     return std::make_pair(Q_None, false);
1692   }
1693 
1694   switch (MangledName.popFront()) {
1695   // Member qualifiers
1696   case 'Q':
1697     return std::make_pair(Q_None, true);
1698   case 'R':
1699     return std::make_pair(Q_Const, true);
1700   case 'S':
1701     return std::make_pair(Q_Volatile, true);
1702   case 'T':
1703     return std::make_pair(Qualifiers(Q_Const | Q_Volatile), true);
1704   // Non-Member qualifiers
1705   case 'A':
1706     return std::make_pair(Q_None, false);
1707   case 'B':
1708     return std::make_pair(Q_Const, false);
1709   case 'C':
1710     return std::make_pair(Q_Volatile, false);
1711   case 'D':
1712     return std::make_pair(Qualifiers(Q_Const | Q_Volatile), false);
1713   }
1714   Error = true;
1715   return std::make_pair(Q_None, false);
1716 }
1717 
1718 // <variable-type> ::= <type> <cvr-qualifiers>
1719 //                 ::= <type> <pointee-cvr-qualifiers> # pointers, references
1720 TypeNode *Demangler::demangleType(StringView &MangledName,
1721                                   QualifierMangleMode QMM) {
1722   Qualifiers Quals = Q_None;
1723   bool IsMember = false;
1724   if (QMM == QualifierMangleMode::Mangle) {
1725     std::tie(Quals, IsMember) = demangleQualifiers(MangledName);
1726   } else if (QMM == QualifierMangleMode::Result) {
1727     if (MangledName.consumeFront('?'))
1728       std::tie(Quals, IsMember) = demangleQualifiers(MangledName);
1729   }
1730 
1731   if (MangledName.empty()) {
1732     Error = true;
1733     return nullptr;
1734   }
1735 
1736   TypeNode *Ty = nullptr;
1737   if (isTagType(MangledName))
1738     Ty = demangleClassType(MangledName);
1739   else if (isPointerType(MangledName)) {
1740     if (isMemberPointer(MangledName, Error))
1741       Ty = demangleMemberPointerType(MangledName);
1742     else if (!Error)
1743       Ty = demanglePointerType(MangledName);
1744     else
1745       return nullptr;
1746   } else if (isArrayType(MangledName))
1747     Ty = demangleArrayType(MangledName);
1748   else if (isFunctionType(MangledName)) {
1749     if (MangledName.consumeFront("$$A8@@"))
1750       Ty = demangleFunctionType(MangledName, true);
1751     else {
1752       assert(MangledName.startsWith("$$A6"));
1753       MangledName.consumeFront("$$A6");
1754       Ty = demangleFunctionType(MangledName, false);
1755     }
1756   } else if (isCustomType(MangledName)) {
1757     Ty = demangleCustomType(MangledName);
1758   } else {
1759     Ty = demanglePrimitiveType(MangledName);
1760   }
1761 
1762   if (!Ty || Error)
1763     return Ty;
1764   Ty->Quals = Qualifiers(Ty->Quals | Quals);
1765   return Ty;
1766 }
1767 
1768 bool Demangler::demangleThrowSpecification(StringView &MangledName) {
1769   if (MangledName.consumeFront("_E"))
1770     return true;
1771   if (MangledName.consumeFront('Z'))
1772     return false;
1773 
1774   Error = true;
1775   return false;
1776 }
1777 
1778 FunctionSignatureNode *Demangler::demangleFunctionType(StringView &MangledName,
1779                                                        bool HasThisQuals) {
1780   FunctionSignatureNode *FTy = Arena.alloc<FunctionSignatureNode>();
1781 
1782   if (HasThisQuals) {
1783     FTy->Quals = demanglePointerExtQualifiers(MangledName);
1784     FTy->RefQualifier = demangleFunctionRefQualifier(MangledName);
1785     FTy->Quals = Qualifiers(FTy->Quals | demangleQualifiers(MangledName).first);
1786   }
1787 
1788   // Fields that appear on both member and non-member functions.
1789   FTy->CallConvention = demangleCallingConvention(MangledName);
1790 
1791   // <return-type> ::= <type>
1792   //               ::= @ # structors (they have no declared return type)
1793   bool IsStructor = MangledName.consumeFront('@');
1794   if (!IsStructor)
1795     FTy->ReturnType = demangleType(MangledName, QualifierMangleMode::Result);
1796 
1797   FTy->Params = demangleFunctionParameterList(MangledName);
1798 
1799   FTy->IsNoexcept = demangleThrowSpecification(MangledName);
1800 
1801   return FTy;
1802 }
1803 
1804 FunctionSymbolNode *
1805 Demangler::demangleFunctionEncoding(StringView &MangledName) {
1806   FuncClass ExtraFlags = FC_None;
1807   if (MangledName.consumeFront("$$J0"))
1808     ExtraFlags = FC_ExternC;
1809 
1810   if (MangledName.empty()) {
1811     Error = true;
1812     return nullptr;
1813   }
1814 
1815   FuncClass FC = demangleFunctionClass(MangledName);
1816   FC = FuncClass(ExtraFlags | FC);
1817 
1818   FunctionSignatureNode *FSN = nullptr;
1819   ThunkSignatureNode *TTN = nullptr;
1820   if (FC & FC_StaticThisAdjust) {
1821     TTN = Arena.alloc<ThunkSignatureNode>();
1822     TTN->ThisAdjust.StaticOffset = demangleSigned(MangledName);
1823   } else if (FC & FC_VirtualThisAdjust) {
1824     TTN = Arena.alloc<ThunkSignatureNode>();
1825     if (FC & FC_VirtualThisAdjustEx) {
1826       TTN->ThisAdjust.VBPtrOffset = demangleSigned(MangledName);
1827       TTN->ThisAdjust.VBOffsetOffset = demangleSigned(MangledName);
1828     }
1829     TTN->ThisAdjust.VtordispOffset = demangleSigned(MangledName);
1830     TTN->ThisAdjust.StaticOffset = demangleSigned(MangledName);
1831   }
1832 
1833   if (FC & FC_NoParameterList) {
1834     // This is an extern "C" function whose full signature hasn't been mangled.
1835     // This happens when we need to mangle a local symbol inside of an extern
1836     // "C" function.
1837     FSN = Arena.alloc<FunctionSignatureNode>();
1838   } else {
1839     bool HasThisQuals = !(FC & (FC_Global | FC_Static));
1840     FSN = demangleFunctionType(MangledName, HasThisQuals);
1841   }
1842 
1843   if (Error)
1844     return nullptr;
1845 
1846   if (TTN) {
1847     *static_cast<FunctionSignatureNode *>(TTN) = *FSN;
1848     FSN = TTN;
1849   }
1850   FSN->FunctionClass = FC;
1851 
1852   FunctionSymbolNode *Symbol = Arena.alloc<FunctionSymbolNode>();
1853   Symbol->Signature = FSN;
1854   return Symbol;
1855 }
1856 
1857 CustomTypeNode *Demangler::demangleCustomType(StringView &MangledName) {
1858   assert(MangledName.startsWith('?'));
1859   MangledName.popFront();
1860 
1861   CustomTypeNode *CTN = Arena.alloc<CustomTypeNode>();
1862   CTN->Identifier = demangleUnqualifiedTypeName(MangledName, /*Memorize=*/true);
1863   if (!MangledName.consumeFront('@'))
1864     Error = true;
1865   if (Error)
1866     return nullptr;
1867   return CTN;
1868 }
1869 
1870 // Reads a primitive type.
1871 PrimitiveTypeNode *Demangler::demanglePrimitiveType(StringView &MangledName) {
1872   if (MangledName.consumeFront("$$T"))
1873     return Arena.alloc<PrimitiveTypeNode>(PrimitiveKind::Nullptr);
1874 
1875   switch (MangledName.popFront()) {
1876   case 'X':
1877     return Arena.alloc<PrimitiveTypeNode>(PrimitiveKind::Void);
1878   case 'D':
1879     return Arena.alloc<PrimitiveTypeNode>(PrimitiveKind::Char);
1880   case 'C':
1881     return Arena.alloc<PrimitiveTypeNode>(PrimitiveKind::Schar);
1882   case 'E':
1883     return Arena.alloc<PrimitiveTypeNode>(PrimitiveKind::Uchar);
1884   case 'F':
1885     return Arena.alloc<PrimitiveTypeNode>(PrimitiveKind::Short);
1886   case 'G':
1887     return Arena.alloc<PrimitiveTypeNode>(PrimitiveKind::Ushort);
1888   case 'H':
1889     return Arena.alloc<PrimitiveTypeNode>(PrimitiveKind::Int);
1890   case 'I':
1891     return Arena.alloc<PrimitiveTypeNode>(PrimitiveKind::Uint);
1892   case 'J':
1893     return Arena.alloc<PrimitiveTypeNode>(PrimitiveKind::Long);
1894   case 'K':
1895     return Arena.alloc<PrimitiveTypeNode>(PrimitiveKind::Ulong);
1896   case 'M':
1897     return Arena.alloc<PrimitiveTypeNode>(PrimitiveKind::Float);
1898   case 'N':
1899     return Arena.alloc<PrimitiveTypeNode>(PrimitiveKind::Double);
1900   case 'O':
1901     return Arena.alloc<PrimitiveTypeNode>(PrimitiveKind::Ldouble);
1902   case '_': {
1903     if (MangledName.empty()) {
1904       Error = true;
1905       return nullptr;
1906     }
1907     switch (MangledName.popFront()) {
1908     case 'N':
1909       return Arena.alloc<PrimitiveTypeNode>(PrimitiveKind::Bool);
1910     case 'J':
1911       return Arena.alloc<PrimitiveTypeNode>(PrimitiveKind::Int64);
1912     case 'K':
1913       return Arena.alloc<PrimitiveTypeNode>(PrimitiveKind::Uint64);
1914     case 'W':
1915       return Arena.alloc<PrimitiveTypeNode>(PrimitiveKind::Wchar);
1916     case 'S':
1917       return Arena.alloc<PrimitiveTypeNode>(PrimitiveKind::Char16);
1918     case 'U':
1919       return Arena.alloc<PrimitiveTypeNode>(PrimitiveKind::Char32);
1920     }
1921     break;
1922   }
1923   }
1924   Error = true;
1925   return nullptr;
1926 }
1927 
1928 TagTypeNode *Demangler::demangleClassType(StringView &MangledName) {
1929   TagTypeNode *TT = nullptr;
1930 
1931   switch (MangledName.popFront()) {
1932   case 'T':
1933     TT = Arena.alloc<TagTypeNode>(TagKind::Union);
1934     break;
1935   case 'U':
1936     TT = Arena.alloc<TagTypeNode>(TagKind::Struct);
1937     break;
1938   case 'V':
1939     TT = Arena.alloc<TagTypeNode>(TagKind::Class);
1940     break;
1941   case 'W':
1942     if (!MangledName.consumeFront('4')) {
1943       Error = true;
1944       return nullptr;
1945     }
1946     TT = Arena.alloc<TagTypeNode>(TagKind::Enum);
1947     break;
1948   default:
1949     assert(false);
1950   }
1951 
1952   TT->QualifiedName = demangleFullyQualifiedTypeName(MangledName);
1953   return TT;
1954 }
1955 
1956 // <pointer-type> ::= E? <pointer-cvr-qualifiers> <ext-qualifiers> <type>
1957 //                       # the E is required for 64-bit non-static pointers
1958 PointerTypeNode *Demangler::demanglePointerType(StringView &MangledName) {
1959   PointerTypeNode *Pointer = Arena.alloc<PointerTypeNode>();
1960 
1961   std::tie(Pointer->Quals, Pointer->Affinity) =
1962       demanglePointerCVQualifiers(MangledName);
1963 
1964   if (MangledName.consumeFront("6")) {
1965     Pointer->Pointee = demangleFunctionType(MangledName, false);
1966     return Pointer;
1967   }
1968 
1969   Qualifiers ExtQuals = demanglePointerExtQualifiers(MangledName);
1970   Pointer->Quals = Qualifiers(Pointer->Quals | ExtQuals);
1971 
1972   Pointer->Pointee = demangleType(MangledName, QualifierMangleMode::Mangle);
1973   return Pointer;
1974 }
1975 
1976 PointerTypeNode *Demangler::demangleMemberPointerType(StringView &MangledName) {
1977   PointerTypeNode *Pointer = Arena.alloc<PointerTypeNode>();
1978 
1979   std::tie(Pointer->Quals, Pointer->Affinity) =
1980       demanglePointerCVQualifiers(MangledName);
1981   assert(Pointer->Affinity == PointerAffinity::Pointer);
1982 
1983   Qualifiers ExtQuals = demanglePointerExtQualifiers(MangledName);
1984   Pointer->Quals = Qualifiers(Pointer->Quals | ExtQuals);
1985 
1986   // isMemberPointer() only returns true if there is at least one character
1987   // after the qualifiers.
1988   if (MangledName.consumeFront("8")) {
1989     Pointer->ClassParent = demangleFullyQualifiedTypeName(MangledName);
1990     Pointer->Pointee = demangleFunctionType(MangledName, true);
1991   } else {
1992     Qualifiers PointeeQuals = Q_None;
1993     bool IsMember = false;
1994     std::tie(PointeeQuals, IsMember) = demangleQualifiers(MangledName);
1995     assert(IsMember || Error);
1996     Pointer->ClassParent = demangleFullyQualifiedTypeName(MangledName);
1997 
1998     Pointer->Pointee = demangleType(MangledName, QualifierMangleMode::Drop);
1999     if (Pointer->Pointee)
2000       Pointer->Pointee->Quals = PointeeQuals;
2001   }
2002 
2003   return Pointer;
2004 }
2005 
2006 Qualifiers Demangler::demanglePointerExtQualifiers(StringView &MangledName) {
2007   Qualifiers Quals = Q_None;
2008   if (MangledName.consumeFront('E'))
2009     Quals = Qualifiers(Quals | Q_Pointer64);
2010   if (MangledName.consumeFront('I'))
2011     Quals = Qualifiers(Quals | Q_Restrict);
2012   if (MangledName.consumeFront('F'))
2013     Quals = Qualifiers(Quals | Q_Unaligned);
2014 
2015   return Quals;
2016 }
2017 
2018 ArrayTypeNode *Demangler::demangleArrayType(StringView &MangledName) {
2019   assert(MangledName.front() == 'Y');
2020   MangledName.popFront();
2021 
2022   uint64_t Rank = 0;
2023   bool IsNegative = false;
2024   std::tie(Rank, IsNegative) = demangleNumber(MangledName);
2025   if (IsNegative || Rank == 0) {
2026     Error = true;
2027     return nullptr;
2028   }
2029 
2030   ArrayTypeNode *ATy = Arena.alloc<ArrayTypeNode>();
2031   NodeList *Head = Arena.alloc<NodeList>();
2032   NodeList *Tail = Head;
2033 
2034   for (uint64_t I = 0; I < Rank; ++I) {
2035     uint64_t D = 0;
2036     std::tie(D, IsNegative) = demangleNumber(MangledName);
2037     if (Error || IsNegative) {
2038       Error = true;
2039       return nullptr;
2040     }
2041     Tail->N = Arena.alloc<IntegerLiteralNode>(D, IsNegative);
2042     if (I + 1 < Rank) {
2043       Tail->Next = Arena.alloc<NodeList>();
2044       Tail = Tail->Next;
2045     }
2046   }
2047   ATy->Dimensions = nodeListToNodeArray(Arena, Head, Rank);
2048 
2049   if (MangledName.consumeFront("$$C")) {
2050     bool IsMember = false;
2051     std::tie(ATy->Quals, IsMember) = demangleQualifiers(MangledName);
2052     if (IsMember) {
2053       Error = true;
2054       return nullptr;
2055     }
2056   }
2057 
2058   ATy->ElementType = demangleType(MangledName, QualifierMangleMode::Drop);
2059   return ATy;
2060 }
2061 
2062 // Reads a function or a template parameters.
2063 NodeArrayNode *
2064 Demangler::demangleFunctionParameterList(StringView &MangledName) {
2065   // Empty parameter list.
2066   if (MangledName.consumeFront('X'))
2067     return nullptr;
2068 
2069   NodeList *Head = Arena.alloc<NodeList>();
2070   NodeList **Current = &Head;
2071   size_t Count = 0;
2072   while (!Error && !MangledName.startsWith('@') &&
2073          !MangledName.startsWith('Z')) {
2074     ++Count;
2075 
2076     if (startsWithDigit(MangledName)) {
2077       size_t N = MangledName[0] - '0';
2078       if (N >= Backrefs.FunctionParamCount) {
2079         Error = true;
2080         return nullptr;
2081       }
2082       MangledName = MangledName.dropFront();
2083 
2084       *Current = Arena.alloc<NodeList>();
2085       (*Current)->N = Backrefs.FunctionParams[N];
2086       Current = &(*Current)->Next;
2087       continue;
2088     }
2089 
2090     size_t OldSize = MangledName.size();
2091 
2092     *Current = Arena.alloc<NodeList>();
2093     TypeNode *TN = demangleType(MangledName, QualifierMangleMode::Drop);
2094     if (!TN || Error)
2095       return nullptr;
2096 
2097     (*Current)->N = TN;
2098 
2099     size_t CharsConsumed = OldSize - MangledName.size();
2100     assert(CharsConsumed != 0);
2101 
2102     // Single-letter types are ignored for backreferences because memorizing
2103     // them doesn't save anything.
2104     if (Backrefs.FunctionParamCount <= 9 && CharsConsumed > 1)
2105       Backrefs.FunctionParams[Backrefs.FunctionParamCount++] = TN;
2106 
2107     Current = &(*Current)->Next;
2108   }
2109 
2110   if (Error)
2111     return nullptr;
2112 
2113   NodeArrayNode *NA = nodeListToNodeArray(Arena, Head, Count);
2114   // A non-empty parameter list is terminated by either 'Z' (variadic) parameter
2115   // list or '@' (non variadic).  Careful not to consume "@Z", as in that case
2116   // the following Z could be a throw specifier.
2117   if (MangledName.consumeFront('@'))
2118     return NA;
2119 
2120   if (MangledName.consumeFront('Z')) {
2121     // This is a variadic parameter list.  We probably need a variadic node to
2122     // append to the end.
2123     return NA;
2124   }
2125 
2126   Error = true;
2127   return nullptr;
2128 }
2129 
2130 NodeArrayNode *
2131 Demangler::demangleTemplateParameterList(StringView &MangledName) {
2132   NodeList *Head;
2133   NodeList **Current = &Head;
2134   size_t Count = 0;
2135 
2136   while (!Error && !MangledName.startsWith('@')) {
2137     if (MangledName.consumeFront("$S") || MangledName.consumeFront("$$V") ||
2138         MangledName.consumeFront("$$$V") || MangledName.consumeFront("$$Z")) {
2139       // parameter pack separator
2140       continue;
2141     }
2142 
2143     ++Count;
2144 
2145     // Template parameter lists don't participate in back-referencing.
2146     *Current = Arena.alloc<NodeList>();
2147 
2148     NodeList &TP = **Current;
2149 
2150     TemplateParameterReferenceNode *TPRN = nullptr;
2151     if (MangledName.consumeFront("$$Y")) {
2152       // Template alias
2153       TP.N = demangleFullyQualifiedTypeName(MangledName);
2154     } else if (MangledName.consumeFront("$$B")) {
2155       // Array
2156       TP.N = demangleType(MangledName, QualifierMangleMode::Drop);
2157     } else if (MangledName.consumeFront("$$C")) {
2158       // Type has qualifiers.
2159       TP.N = demangleType(MangledName, QualifierMangleMode::Mangle);
2160     } else if (MangledName.startsWith("$1") || MangledName.startsWith("$H") ||
2161                MangledName.startsWith("$I") || MangledName.startsWith("$J")) {
2162       // Pointer to member
2163       TP.N = TPRN = Arena.alloc<TemplateParameterReferenceNode>();
2164       TPRN->IsMemberPointer = true;
2165 
2166       MangledName = MangledName.dropFront();
2167       // 1 - single inheritance       <name>
2168       // H - multiple inheritance     <name> <number>
2169       // I - virtual inheritance      <name> <number> <number> <number>
2170       // J - unspecified inheritance  <name> <number> <number> <number>
2171       char InheritanceSpecifier = MangledName.popFront();
2172       SymbolNode *S = nullptr;
2173       if (MangledName.startsWith('?')) {
2174         S = parse(MangledName);
2175         if (Error)
2176           return nullptr;
2177         memorizeIdentifier(S->Name->getUnqualifiedIdentifier());
2178       }
2179 
2180       switch (InheritanceSpecifier) {
2181       case 'J':
2182         TPRN->ThunkOffsets[TPRN->ThunkOffsetCount++] =
2183             demangleSigned(MangledName);
2184         DEMANGLE_FALLTHROUGH;
2185       case 'I':
2186         TPRN->ThunkOffsets[TPRN->ThunkOffsetCount++] =
2187             demangleSigned(MangledName);
2188         DEMANGLE_FALLTHROUGH;
2189       case 'H':
2190         TPRN->ThunkOffsets[TPRN->ThunkOffsetCount++] =
2191             demangleSigned(MangledName);
2192         DEMANGLE_FALLTHROUGH;
2193       case '1':
2194         break;
2195       default:
2196         Error = true;
2197         break;
2198       }
2199       TPRN->Affinity = PointerAffinity::Pointer;
2200       TPRN->Symbol = S;
2201     } else if (MangledName.startsWith("$E?")) {
2202       MangledName.consumeFront("$E");
2203       // Reference to symbol
2204       TP.N = TPRN = Arena.alloc<TemplateParameterReferenceNode>();
2205       TPRN->Symbol = parse(MangledName);
2206       TPRN->Affinity = PointerAffinity::Reference;
2207     } else if (MangledName.startsWith("$F") || MangledName.startsWith("$G")) {
2208       TP.N = TPRN = Arena.alloc<TemplateParameterReferenceNode>();
2209 
2210       // Data member pointer.
2211       MangledName = MangledName.dropFront();
2212       char InheritanceSpecifier = MangledName.popFront();
2213 
2214       switch (InheritanceSpecifier) {
2215       case 'G':
2216         TPRN->ThunkOffsets[TPRN->ThunkOffsetCount++] =
2217             demangleSigned(MangledName);
2218         DEMANGLE_FALLTHROUGH;
2219       case 'F':
2220         TPRN->ThunkOffsets[TPRN->ThunkOffsetCount++] =
2221             demangleSigned(MangledName);
2222         TPRN->ThunkOffsets[TPRN->ThunkOffsetCount++] =
2223             demangleSigned(MangledName);
2224         DEMANGLE_FALLTHROUGH;
2225       case '0':
2226         break;
2227       default:
2228         Error = true;
2229         break;
2230       }
2231       TPRN->IsMemberPointer = true;
2232 
2233     } else if (MangledName.consumeFront("$0")) {
2234       // Integral non-type template parameter
2235       bool IsNegative = false;
2236       uint64_t Value = 0;
2237       std::tie(Value, IsNegative) = demangleNumber(MangledName);
2238 
2239       TP.N = Arena.alloc<IntegerLiteralNode>(Value, IsNegative);
2240     } else {
2241       TP.N = demangleType(MangledName, QualifierMangleMode::Drop);
2242     }
2243     if (Error)
2244       return nullptr;
2245 
2246     Current = &TP.Next;
2247   }
2248 
2249   if (Error)
2250     return nullptr;
2251 
2252   // Template parameter lists cannot be variadic, so it can only be terminated
2253   // by @.
2254   if (MangledName.consumeFront('@'))
2255     return nodeListToNodeArray(Arena, Head, Count);
2256   Error = true;
2257   return nullptr;
2258 }
2259 
2260 void Demangler::dumpBackReferences() {
2261   std::printf("%d function parameter backreferences\n",
2262               (int)Backrefs.FunctionParamCount);
2263 
2264   // Create an output stream so we can render each type.
2265   OutputStream OS;
2266   if (!initializeOutputStream(nullptr, nullptr, OS, 1024))
2267     std::terminate();
2268   for (size_t I = 0; I < Backrefs.FunctionParamCount; ++I) {
2269     OS.setCurrentPosition(0);
2270 
2271     TypeNode *T = Backrefs.FunctionParams[I];
2272     T->output(OS, OF_Default);
2273 
2274     std::printf("  [%d] - %.*s\n", (int)I, (int)OS.getCurrentPosition(),
2275                 OS.getBuffer());
2276   }
2277   std::free(OS.getBuffer());
2278 
2279   if (Backrefs.FunctionParamCount > 0)
2280     std::printf("\n");
2281   std::printf("%d name backreferences\n", (int)Backrefs.NamesCount);
2282   for (size_t I = 0; I < Backrefs.NamesCount; ++I) {
2283     std::printf("  [%d] - %.*s\n", (int)I, (int)Backrefs.Names[I]->Name.size(),
2284                 Backrefs.Names[I]->Name.begin());
2285   }
2286   if (Backrefs.NamesCount > 0)
2287     std::printf("\n");
2288 }
2289 
2290 char *llvm::microsoftDemangle(const char *MangledName, char *Buf, size_t *N,
2291                               int *Status, MSDemangleFlags Flags) {
2292   int InternalStatus = demangle_success;
2293   Demangler D;
2294   OutputStream S;
2295 
2296   StringView Name{MangledName};
2297   SymbolNode *AST = D.parse(Name);
2298 
2299   if (Flags & MSDF_DumpBackrefs)
2300     D.dumpBackReferences();
2301 
2302   if (D.Error)
2303     InternalStatus = demangle_invalid_mangled_name;
2304   else if (!initializeOutputStream(Buf, N, S, 1024))
2305     InternalStatus = demangle_memory_alloc_failure;
2306   else {
2307     AST->output(S, OF_Default);
2308     S += '\0';
2309     if (N != nullptr)
2310       *N = S.getCurrentPosition();
2311     Buf = S.getBuffer();
2312   }
2313 
2314   if (Status)
2315     *Status = InternalStatus;
2316   return InternalStatus == demangle_success ? Buf : nullptr;
2317 }
2318