source: src/SymTab/Demangle.cc@ 6b419ce

Last change on this file since 6b419ce was c649d2c, checked in by Andrew Beach <ajbeach@…>, 23 months ago

Updated the demangler to also demangle operator names. You can also pass a filename to demangle now. There is also a test, but I don't really know what good tests are for the demangler.
  • Property mode set to 100644
File size: 20.3 KB
Line 
1//
2// Cforall Version 1.0.0 Copyright (C) 2018 University of Waterloo
3//
4// The contents of this file are covered under the licence agreement in the
5// file "LICENCE" distributed with Cforall.
6//
7// Demangle.cc -- Convert a mangled name into a human readable name.
8//
9// Author : Rob Schluntz
10// Created On : Thu Jul 19 12:52:41 2018
11// Last Modified By : Andrew Beach
12// Last Modified On : Mon Nov 6 15:59:00 2023
13// Update Count : 12
14//
15
16#include <algorithm>
17#include <sstream>
18
19#include "CodeGen/GenType.h"
20#include "CodeGen/OperatorTable.h"
21#include "Common/PassVisitor.h"
22#include "Common/utility.h" // isPrefix
23#include "Mangler.h"
24#include "SynTree/Type.h"
25#include "SynTree/Declaration.h"
26
27#define DEBUG
28#ifdef DEBUG
29#define PRINT(x) x
30#else
31#define PRINT(x) {}
32#endif
33
34namespace {
35 struct GenType : public WithVisitorRef<GenType>, public WithShortCircuiting {
36 std::string typeString;
37 GenType( const std::string &typeString );
38
39 void previsit( BaseSyntaxNode * );
40 void postvisit( BaseSyntaxNode * );
41
42 void postvisit( FunctionType * funcType );
43 void postvisit( VoidType * voidType );
44 void postvisit( BasicType * basicType );
45 void postvisit( PointerType * pointerType );
46 void postvisit( ArrayType * arrayType );
47 void postvisit( ReferenceType * refType );
48 void postvisit( StructInstType * structInst );
49 void postvisit( UnionInstType * unionInst );
50 void postvisit( EnumInstType * enumInst );
51 void postvisit( TypeInstType * typeInst );
52 void postvisit( TupleType * tupleType );
53 void postvisit( VarArgsType * varArgsType );
54 void postvisit( ZeroType * zeroType );
55 void postvisit( OneType * oneType );
56 void postvisit( GlobalScopeType * globalType );
57 void postvisit( QualifiedType * qualType );
58
59 private:
60 void handleQualifiers( Type *type );
61 std::string handleGeneric( ReferenceToType * refType );
62 void genArray( const Type::Qualifiers &qualifiers, Type *base, Expression *dimension, bool isVarLen, bool isStatic );
63 };
64
65 std::string genDemangleType( Type * type, const std::string & baseString ) {
66 PassVisitor<GenType> gt( baseString );
67 assert( type );
68 type->accept( gt );
69 return gt.pass.typeString;
70 }
71
72 GenType::GenType( const std::string &typeString ) : typeString( typeString ) {}
73
74 // *** BaseSyntaxNode
75 void GenType::previsit( BaseSyntaxNode * ) {
76 // turn off automatic recursion for all nodes, to allow each visitor to
77 // precisely control the order in which its children are visited.
78 visit_children = false;
79 }
80
81 void GenType::postvisit( BaseSyntaxNode * node ) {
82 std::stringstream ss;
83 node->print( ss );
84 assertf( false, "Unhandled node reached in GenType: %s", ss.str().c_str() );
85 }
86
87 void GenType::postvisit( VoidType * voidType ) {
88 typeString = "void " + typeString;
89 handleQualifiers( voidType );
90 }
91
92 void GenType::postvisit( BasicType * basicType ) {
93 BasicType::Kind kind = basicType->kind;
94 assert( 0 <= kind && kind < BasicType::NUMBER_OF_BASIC_TYPES );
95 typeString = std::string( BasicType::typeNames[kind] ) + " " + typeString;
96 handleQualifiers( basicType );
97 }
98
99 void GenType::genArray( const Type::Qualifiers & qualifiers, Type * base, Expression *dimension, bool isVarLen, bool ) {
100 std::ostringstream os;
101 if ( typeString != "" ) {
102 if ( typeString[ 0 ] == '*' ) {
103 os << "(" << typeString << ")";
104 } else {
105 os << typeString;
106 } // if
107 } // if
108 os << "[";
109
110 if ( qualifiers.is_const ) {
111 os << "const ";
112 } // if
113 if ( qualifiers.is_volatile ) {
114 os << "volatile ";
115 } // if
116 if ( qualifiers.is_restrict ) {
117 os << "__restrict ";
118 } // if
119 if ( qualifiers.is_atomic ) {
120 os << "_Atomic ";
121 } // if
122 if ( dimension != 0 ) {
123 // TODO: ???
124 // PassVisitor<CodeGenerator> cg( os, pretty, genC, lineMarks );
125 // dimension->accept( cg );
126 } else if ( isVarLen ) {
127 // no dimension expression on a VLA means it came in with the * token
128 os << "*";
129 } // if
130 os << "]";
131
132 typeString = os.str();
133
134 base->accept( *visitor );
135 }
136
137 void GenType::postvisit( PointerType * pointerType ) {
138 assert( pointerType->base != 0);
139 if ( pointerType->get_isStatic() || pointerType->get_isVarLen() || pointerType->dimension ) {
140 assert(false);
141 genArray( pointerType->get_qualifiers(), pointerType->base, pointerType->dimension, pointerType->get_isVarLen(), pointerType->get_isStatic() );
142 } else {
143 handleQualifiers( pointerType );
144 if ( typeString[ 0 ] == '?' ) {
145 typeString = "* " + typeString;
146 } else {
147 typeString = "*" + typeString;
148 } // if
149 pointerType->base->accept( *visitor );
150 } // if
151 }
152
153 void GenType::postvisit( ArrayType * arrayType ) {
154 genArray( arrayType->get_qualifiers(), arrayType->base, arrayType->dimension, arrayType->get_isVarLen(), arrayType->get_isStatic() );
155 }
156
157 void GenType::postvisit( ReferenceType * refType ) {
158 assert( false );
159 assert( refType->base != 0);
160 handleQualifiers( refType );
161 typeString = "&" + typeString;
162 refType->base->accept( *visitor );
163 }
164
165 void GenType::postvisit( FunctionType * funcType ) {
166 std::ostringstream os;
167
168 if ( typeString != "" ) {
169 if ( typeString[0] == '*' ) {
170 os << "(" << typeString << ")";
171 } else {
172 os << typeString;
173 } // if
174 } // if
175
176 /************* parameters ***************/
177 const std::list<DeclarationWithType *> &pars = funcType->parameters;
178
179 if ( pars.empty() ) {
180 if ( funcType->get_isVarArgs() ) {
181 os << "()";
182 } else {
183 os << "(void)";
184 } // if
185 } else {
186 os << "(" ;
187
188 unsigned int i = 0;
189 for (DeclarationWithType * p : pars) {
190 os << genDemangleType( p->get_type(), "" );
191 if (++i != pars.size()) os << ", ";
192 }
193
194 if ( funcType->get_isVarArgs() ) {
195 os << ", ...";
196 } // if
197 os << ")";
198 } // if
199
200 typeString = os.str();
201
202 if ( funcType->returnVals.size() == 0 ) {
203 typeString += ": void";
204 } else {
205 typeString += ": " + genDemangleType(funcType->returnVals.front()->get_type(), "");
206 } // if
207
208 // add forall
209 if( ! funcType->forall.empty() ) {
210 std::ostringstream os;
211 os << "forall(";
212 unsigned int i = 0;
213 for ( auto td : funcType->forall ) {
214 os << td->typeString() << " " << td->name;
215 if (! td->assertions.empty()) {
216 os << " | { ";
217 unsigned int j = 0;
218 for (DeclarationWithType * assert : td->assertions) {
219 os << genDemangleType(assert->get_type(), assert->name);
220 if (++j != td->assertions.size()) os << ", ";
221 }
222 os << "}";
223 }
224 if (++i != funcType->forall.size()) os << ", ";
225 }
226 os << ")";
227 typeString = typeString + " -> " + os.str();
228 }
229 }
230
231 std::string GenType::handleGeneric( ReferenceToType * refType ) {
232 if ( ! refType->parameters.empty() ) {
233 std::ostringstream os;
234 // TODO: ???
235 // PassVisitor<CodeGenerator> cg( os, pretty, genC, lineMarks );
236 os << "(";
237 // cg.pass.genCommaList( refType->parameters.begin(), refType->parameters.end() );
238 os << ") ";
239 return os.str();
240 }
241 return "";
242 }
243
244 void GenType::postvisit( StructInstType * structInst ) {
245 typeString = "struct " + structInst->name + handleGeneric( structInst ) + " " + typeString;
246 handleQualifiers( structInst );
247 }
248
249 void GenType::postvisit( UnionInstType * unionInst ) {
250 typeString = "union " + unionInst->name + handleGeneric( unionInst ) + " " + typeString;
251 handleQualifiers( unionInst );
252 }
253
254 void GenType::postvisit( EnumInstType * enumInst ) {
255 typeString = "enum " + enumInst->name + " " + typeString;
256 handleQualifiers( enumInst );
257 }
258
259 void GenType::postvisit( TypeInstType * typeInst ) {
260 typeString = typeInst->name + " " + typeString;
261 handleQualifiers( typeInst );
262 }
263
264 void GenType::postvisit( TupleType * tupleType ) {
265 unsigned int i = 0;
266 std::ostringstream os;
267 os << "[";
268 for ( Type * t : *tupleType ) {
269 i++;
270 os << genDemangleType( t, "" ) << (i == tupleType->size() ? "" : ", ");
271 }
272 os << "] ";
273 typeString = os.str() + typeString;
274 }
275
276 void GenType::postvisit( VarArgsType * varArgsType ) {
277 typeString = "__builtin_va_list " + typeString;
278 handleQualifiers( varArgsType );
279 }
280
281 void GenType::postvisit( ZeroType * zeroType ) {
282 // ideally these wouldn't hit codegen at all, but should be safe to make them ints
283 typeString = "zero_t " + typeString;
284 handleQualifiers( zeroType );
285 }
286
287 void GenType::postvisit( OneType * oneType ) {
288 // ideally these wouldn't hit codegen at all, but should be safe to make them ints
289 typeString = "one_t " + typeString;
290 handleQualifiers( oneType );
291 }
292
293 void GenType::postvisit( GlobalScopeType * globalType ) {
294 handleQualifiers( globalType );
295 }
296
297 void GenType::postvisit( QualifiedType * qualType ) {
298 std::ostringstream os;
299 os << genDemangleType( qualType->parent, "" ) << "." << genDemangleType( qualType->child, "" ) << typeString;
300 typeString = os.str();
301 handleQualifiers( qualType );
302 }
303
304 void GenType::handleQualifiers( Type * type ) {
305 if ( type->get_const() ) {
306 typeString = "const " + typeString;
307 } // if
308 if ( type->get_volatile() ) {
309 typeString = "volatile " + typeString;
310 } // if
311 if ( type->get_restrict() ) {
312 typeString = "__restrict " + typeString;
313 } // if
314 if ( type->get_atomic() ) {
315 typeString = "_Atomic " + typeString;
316 } // if
317 }
318}
319
320
321namespace SymTab {
322 namespace Mangler {
323 namespace {
324 struct StringView {
325 private:
326 std::string str;
327 size_t idx = 0;
328 // typedef Type * (StringView::*parser)(Type::Qualifiers);
329 typedef std::function<Type * (Type::Qualifiers)> parser;
330 std::vector<std::pair<std::string, parser>> parsers;
331 public:
332 StringView(const std::string & str);
333
334 bool done() const { return idx >= str.size(); }
335 char cur() const { assert(! done()); return str[idx]; }
336
337 bool expect(char ch) { return str[idx++] == ch; }
338 void next(size_t inc = 1) { idx += inc; }
339
340 /// determines if `pref` is a prefix of `str`
341 bool isPrefix(const std::string & pref);
342 bool extractNumber(size_t & out);
343 bool extractName(std::string & out);
344 bool stripMangleName(std::string & name);
345
346 Type * parseFunction(Type::Qualifiers tq);
347 Type * parseTuple(Type::Qualifiers tq);
348 Type * parseVoid(Type::Qualifiers tq);
349 Type * parsePointer(Type::Qualifiers tq);
350 Type * parseArray(Type::Qualifiers tq);
351 Type * parseStruct(Type::Qualifiers tq);
352 Type * parseUnion(Type::Qualifiers tq);
353 Type * parseEnum(Type::Qualifiers tq);
354 Type * parseType(Type::Qualifiers tq);
355
356 Type * parseType();
357 bool parse(std::string & name, Type *& type);
358 };
359
360 StringView::StringView(const std::string & str) : str(str) {
361 // basic types
362 for (size_t k = 0; k < BasicType::NUMBER_OF_BASIC_TYPES; ++k) {
363 parsers.emplace_back(Encoding::basicTypes[k], [k](Type::Qualifiers tq) {
364 PRINT( std::cerr << "basic type: " << k << std::endl; )
365 return new BasicType(tq, (BasicType::Kind)k);
366 });
367 }
368 // type variable types
369 for (size_t k = 0; k < TypeDecl::NUMBER_OF_KINDS; ++k) {
370 static const std::string typeVariableNames[] = { "DT", "DST", "OT", "FT", "TT", "ALT", };
371 static_assert(
372 sizeof(typeVariableNames)/sizeof(typeVariableNames[0]) == TypeDecl::NUMBER_OF_KINDS,
373 "Each type variable kind should have a demangle name prefix"
374 );
375 parsers.emplace_back(Encoding::typeVariables[k], [k, this](Type::Qualifiers tq) -> TypeInstType * {
376 PRINT( std::cerr << "type variable type: " << k << std::endl; )
377 size_t N;
378 if (! extractNumber(N)) return nullptr;
379 return new TypeInstType(tq, toString(typeVariableNames[k], N), (TypeDecl::Kind)k != TypeDecl::Ftype);
380 });
381 }
382 // everything else
383 parsers.emplace_back(Encoding::void_t, [this](Type::Qualifiers tq) { return parseVoid(tq); });
384 parsers.emplace_back(Encoding::function, [this](Type::Qualifiers tq) { return parseFunction(tq); });
385 parsers.emplace_back(Encoding::pointer, [this](Type::Qualifiers tq) { return parsePointer(tq); });
386 parsers.emplace_back(Encoding::array, [this](Type::Qualifiers tq) { return parseArray(tq); });
387 parsers.emplace_back(Encoding::tuple, [this](Type::Qualifiers tq) { return parseTuple(tq); });
388 parsers.emplace_back(Encoding::struct_t, [this](Type::Qualifiers tq) { return parseStruct(tq); });
389 parsers.emplace_back(Encoding::union_t, [this](Type::Qualifiers tq) { return parseUnion(tq); });
390 parsers.emplace_back(Encoding::enum_t, [this](Type::Qualifiers tq) { return parseEnum(tq); });
391 parsers.emplace_back(Encoding::type, [this](Type::Qualifiers tq) { return parseType(tq); });
392 parsers.emplace_back(Encoding::zero, [](Type::Qualifiers tq) { return new ZeroType(tq); });
393 parsers.emplace_back(Encoding::one, [](Type::Qualifiers tq) { return new OneType(tq); });
394 }
395
396 bool StringView::extractNumber(size_t & out) {
397 std::stringstream numss;
398 if (idx >= str.size()) return false;
399 while (isdigit(str[idx])) {
400 numss << str[idx];
401 ++idx;
402 if (idx == str.size()) break;
403 }
404 if (! (numss >> out)) return false;
405 PRINT( std::cerr << "extractNumber success: " << out << std::endl; )
406 return true;
407 }
408
409 bool StringView::extractName(std::string & out) {
410 size_t len;
411 if (! extractNumber(len)) return false;
412 if (idx+len > str.size()) return false;
413 out = str.substr(idx, len);
414 idx += len;
415 PRINT( std::cerr << "extractName success: " << out << std::endl; )
416 return true;
417 }
418
419 bool StringView::isPrefix(const std::string & pref) {
420 // if ( pref.size() > str.size()-idx ) return false;
421 // auto its = std::mismatch( pref.begin(), pref.end(), std::next(str.begin(), idx) );
422 // if (its.first == pref.end()) {
423 // idx += pref.size();
424 // return true;
425 // }
426
427 // This update is untested because there are no tests for this code.
428 if ( ::isPrefix( str, pref, idx ) ) {
429 idx += pref.size();
430 return true;
431 }
432 return false;
433 }
434
435 // strips __NAME__cfa__TYPE_N, where N is [0-9]+: returns str is a match is found, returns empty string otherwise
436 bool StringView::stripMangleName(std::string & name) {
437 PRINT( std::cerr << "====== " << str.size() << " " << str << std::endl; )
438 if (str.size() < 2+Encoding::manglePrefix.size()) return false; // +2 for at least _1 suffix
439 if ( ! isPrefix(Encoding::manglePrefix) || ! isdigit(str.back() ) ) return false;
440
441 // get name
442 if (! extractName(name)) return false;
443
444 // find bounds for type
445 PRINT( std::cerr << idx << " " << str.size() << std::endl; )
446 PRINT( std::cerr << "[");
447 while (isdigit(str.back())) {
448 PRINT(std::cerr << ".");
449 str.pop_back();
450 if (str.size() <= idx) return false;
451 }
452 PRINT( std::cerr << "]" << std::endl );
453 if (str.back() != '_') return false;
454 str.pop_back();
455 PRINT( std::cerr << str.size() << " " << name << " " << str.substr(idx) << std::endl; )
456 return str.size() > idx;
457 }
458
459 Type * StringView::parseFunction(Type::Qualifiers tq) {
460 PRINT( std::cerr << "function..." << std::endl; )
461 if (done()) return nullptr;
462 FunctionType * ftype = new FunctionType( tq, false );
463 std::unique_ptr<Type> manager(ftype);
464 Type * retVal = parseType();
465 if (! retVal) return nullptr;
466 PRINT( std::cerr << "with return type: " << retVal << std::endl; )
467 ftype->returnVals.push_back(ObjectDecl::newObject("", retVal, nullptr));
468 if (done() || ! expect('_')) return nullptr;
469 while (! done()) {
470 PRINT( std::cerr << "got ch: " << cur() << std::endl; )
471 if (cur() == '_') return manager.release();
472 Type * param = parseType();
473 if (! param) return nullptr;
474 PRINT( std::cerr << "with parameter : " << param << std::endl; )
475 ftype->parameters.push_back(ObjectDecl::newObject("", param, nullptr));
476 }
477 return nullptr;
478 }
479
480 Type * StringView::parseTuple(Type::Qualifiers tq) {
481 PRINT( std::cerr << "tuple..." << std::endl; )
482 std::list< Type * > types;
483 size_t ncomponents;
484 if (! extractNumber(ncomponents)) return nullptr;
485 for (size_t i = 0; i < ncomponents; ++i) {
486 // TODO: delete all on return
487 if (done()) return nullptr;
488 PRINT( std::cerr << "got ch: " << cur() << std::endl; )
489 Type * t = parseType();
490 if (! t) return nullptr;
491 PRINT( std::cerr << "with type : " << t << std::endl; )
492 types.push_back(t);
493 }
494 return new TupleType( tq, types );
495 }
496
497 Type * StringView::parseVoid(Type::Qualifiers tq) {
498 return new VoidType( tq );
499 }
500
501 Type * StringView::parsePointer(Type::Qualifiers tq) {
502 PRINT( std::cerr << "pointer..." << std::endl; )
503 Type * t = parseType();
504 if (! t) return nullptr;
505 return new PointerType( tq, t );
506 }
507
508 Type * StringView::parseArray(Type::Qualifiers tq) {
509 PRINT( std::cerr << "array..." << std::endl; )
510 size_t length;
511 if (! extractNumber(length)) return nullptr;
512 Type * t = parseType();
513 if (! t) return nullptr;
514 return new ArrayType( tq, t, new ConstantExpr( Constant::from_ulong(length) ), false, false );
515 }
516
517 Type * StringView::parseStruct(Type::Qualifiers tq) {
518 PRINT( std::cerr << "struct..." << std::endl; )
519 std::string name;
520 if (! extractName(name)) return nullptr;
521 return new StructInstType(tq, name);
522 }
523
524 Type * StringView::parseUnion(Type::Qualifiers tq) {
525 PRINT( std::cerr << "union..." << std::endl; )
526 std::string name;
527 if (! extractName(name)) return nullptr;
528 return new UnionInstType(tq, name);
529 }
530
531 Type * StringView::parseEnum(Type::Qualifiers tq) {
532 PRINT( std::cerr << "enum..." << std::endl; )
533 std::string name;
534 if (! extractName(name)) return nullptr;
535 return new EnumInstType(tq, name);
536 }
537
538 Type * StringView::parseType(Type::Qualifiers tq) {
539 PRINT( std::cerr << "type..." << std::endl; )
540 std::string name;
541 if (! extractName(name)) return nullptr;
542 PRINT( std::cerr << "typename..." << name << std::endl; )
543 return new TypeInstType(tq, name, false);
544 }
545
546 Type * StringView::parseType() {
547 if (done()) return nullptr;
548
549 std::list<TypeDecl *> forall;
550 if (isPrefix(Encoding::forall)) {
551 PRINT( std::cerr << "polymorphic with..." << std::endl; )
552 size_t dcount, fcount, vcount, acount;
553 if (! extractNumber(dcount)) return nullptr;
554 PRINT( std::cerr << dcount << " dtypes" << std::endl; )
555 if (! expect('_')) return nullptr;
556 if (! extractNumber(fcount)) return nullptr;
557 PRINT( std::cerr << fcount << " ftypes" << std::endl; )
558 if (! expect('_')) return nullptr;
559 if (! extractNumber(vcount)) return nullptr;
560 PRINT( std::cerr << vcount << " ttypes" << std::endl; )
561 if (! expect('_')) return nullptr;
562 if (! extractNumber(acount)) return nullptr;
563 PRINT( std::cerr << acount << " assertions" << std::endl; )
564 if (! expect('_')) return nullptr;
565 for (size_t i = 0; i < acount; ++i) {
566 // TODO: need to recursively parse assertions, but for now just return nullptr so that
567 // demangler does not crash if there are assertions
568 return nullptr;
569 }
570 if (! expect('_')) return nullptr;
571 }
572
573 // qualifiers
574 Type::Qualifiers tq;
575 while (true) {
576 auto qual = std::find_if(Encoding::qualifiers.begin(), Encoding::qualifiers.end(), [this](decltype(Encoding::qualifiers)::value_type val) {
577 return isPrefix(val.second);
578 });
579 if (qual == Encoding::qualifiers.end()) break;
580 tq |= qual->first;
581 }
582
583 // find the correct type parser and use it
584 auto iter = std::find_if(parsers.begin(), parsers.end(), [this](std::pair<std::string, parser> & p) {
585 return isPrefix(p.first);
586 });
587 assertf(iter != parsers.end(), "Unhandled type letter: %c at index: %zd", cur(), idx);
588 Type * ret = iter->second(tq);
589 if (! ret) return nullptr;
590 ret->forall = std::move(forall);
591 return ret;
592 }
593
594 bool StringView::parse(std::string & name, Type *& type) {
595 if (! stripMangleName(name)) return false;
596 PRINT( std::cerr << "stripped name: " << name << std::endl; )
597 Type * t = parseType();
598 if (! t) return false;
599 type = t;
600 return true;
601 }
602
603 std::string demangle(const std::string & mangleName) {
604 SymTab::Mangler::StringView view(mangleName);
605 std::string name;
606 Type * type = nullptr;
607 if (! view.parse(name, type)) return mangleName;
608 auto info = CodeGen::operatorLookupByOutput(name);
609 if (info) name = info->inputName;
610 std::unique_ptr<Type> manager(type);
611 return genDemangleType(type, name);
612 }
613 } // namespace
614 } // namespace Mangler
615} // namespace SymTab
616
617extern "C" {
618 char * cforall_demangle(const char * mangleName, int option __attribute__((unused))) {
619 const std::string & demangleName = SymTab::Mangler::demangle(mangleName);
620 return strdup(demangleName.c_str());
621 }
622}
623
624// Local Variables: //
625// tab-width: 4 //
626// mode: c++ //
627// compile-command: "make install" //
628// End: //
Note: See TracBrowser for help on using the repository browser.