1 //===- ExtractAPI/Serialization/SymbolGraphSerializer.cpp -------*- C++ -*-===//
2 //
3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4 // See https://llvm.org/LICENSE.txt for license information.
5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6 //
7 //===----------------------------------------------------------------------===//
8 ///
9 /// \file
10 /// This file implements the SymbolGraphSerializer.
11 ///
12 //===----------------------------------------------------------------------===//
13
14 #include "clang/ExtractAPI/Serialization/SymbolGraphSerializer.h"
15 #include "clang/Basic/SourceLocation.h"
16 #include "clang/Basic/Version.h"
17 #include "clang/ExtractAPI/API.h"
18 #include "clang/ExtractAPI/DeclarationFragments.h"
19 #include "llvm/ADT/STLExtras.h"
20 #include "llvm/ADT/STLFunctionalExtras.h"
21 #include "llvm/ADT/SmallVector.h"
22 #include "llvm/Support/Compiler.h"
23 #include "llvm/Support/Path.h"
24 #include "llvm/Support/VersionTuple.h"
25 #include "llvm/Support/raw_ostream.h"
26 #include <iterator>
27 #include <optional>
28
29 using namespace clang;
30 using namespace clang::extractapi;
31 using namespace llvm;
32
33 namespace {
34
35 /// Helper function to inject a JSON object \p Obj into another object \p Paren
36 /// at position \p Key.
serializeObject(Object & Paren,StringRef Key,std::optional<Object> && Obj)37 void serializeObject(Object &Paren, StringRef Key,
38 std::optional<Object> &&Obj) {
39 if (Obj)
40 Paren[Key] = std::move(*Obj);
41 }
42
43 /// Helper function to inject a JSON array \p Array into object \p Paren at
44 /// position \p Key.
serializeArray(Object & Paren,StringRef Key,std::optional<Array> && Array)45 void serializeArray(Object &Paren, StringRef Key,
46 std::optional<Array> &&Array) {
47 if (Array)
48 Paren[Key] = std::move(*Array);
49 }
50
51 /// Helper function to inject a JSON array composed of the values in \p C into
52 /// object \p Paren at position \p Key.
53 template <typename ContainerTy>
serializeArray(Object & Paren,StringRef Key,ContainerTy && C)54 void serializeArray(Object &Paren, StringRef Key, ContainerTy &&C) {
55 Paren[Key] = Array(C);
56 }
57
58 /// Serialize a \c VersionTuple \p V with the Symbol Graph semantic version
59 /// format.
60 ///
61 /// A semantic version object contains three numeric fields, representing the
62 /// \c major, \c minor, and \c patch parts of the version tuple.
63 /// For example version tuple 1.0.3 is serialized as:
64 /// \code
65 /// {
66 /// "major" : 1,
67 /// "minor" : 0,
68 /// "patch" : 3
69 /// }
70 /// \endcode
71 ///
72 /// \returns \c std::nullopt if the version \p V is empty, or an \c Object
73 /// containing the semantic version representation of \p V.
serializeSemanticVersion(const VersionTuple & V)74 std::optional<Object> serializeSemanticVersion(const VersionTuple &V) {
75 if (V.empty())
76 return std::nullopt;
77
78 Object Version;
79 Version["major"] = V.getMajor();
80 Version["minor"] = V.getMinor().value_or(0);
81 Version["patch"] = V.getSubminor().value_or(0);
82 return Version;
83 }
84
85 /// Serialize the OS information in the Symbol Graph platform property.
86 ///
87 /// The OS information in Symbol Graph contains the \c name of the OS, and an
88 /// optional \c minimumVersion semantic version field.
serializeOperatingSystem(const Triple & T)89 Object serializeOperatingSystem(const Triple &T) {
90 Object OS;
91 OS["name"] = T.getOSTypeName(T.getOS());
92 serializeObject(OS, "minimumVersion",
93 serializeSemanticVersion(T.getMinimumSupportedOSVersion()));
94 return OS;
95 }
96
97 /// Serialize the platform information in the Symbol Graph module section.
98 ///
99 /// The platform object describes a target platform triple in corresponding
100 /// three fields: \c architecture, \c vendor, and \c operatingSystem.
serializePlatform(const Triple & T)101 Object serializePlatform(const Triple &T) {
102 Object Platform;
103 Platform["architecture"] = T.getArchName();
104 Platform["vendor"] = T.getVendorName();
105
106 if (!T.getEnvironmentName().empty())
107 Platform["environment"] = T.getEnvironmentName();
108
109 Platform["operatingSystem"] = serializeOperatingSystem(T);
110 return Platform;
111 }
112
113 /// Serialize a source position.
serializeSourcePosition(const PresumedLoc & Loc)114 Object serializeSourcePosition(const PresumedLoc &Loc) {
115 assert(Loc.isValid() && "invalid source position");
116
117 Object SourcePosition;
118 SourcePosition["line"] = Loc.getLine() - 1;
119 SourcePosition["character"] = Loc.getColumn() - 1;
120
121 return SourcePosition;
122 }
123
124 /// Serialize a source location in file.
125 ///
126 /// \param Loc The presumed location to serialize.
127 /// \param IncludeFileURI If true, include the file path of \p Loc as a URI.
128 /// Defaults to false.
serializeSourceLocation(const PresumedLoc & Loc,bool IncludeFileURI=false)129 Object serializeSourceLocation(const PresumedLoc &Loc,
130 bool IncludeFileURI = false) {
131 Object SourceLocation;
132 serializeObject(SourceLocation, "position", serializeSourcePosition(Loc));
133
134 if (IncludeFileURI) {
135 std::string FileURI = "file://";
136 // Normalize file path to use forward slashes for the URI.
137 FileURI += sys::path::convert_to_slash(Loc.getFilename());
138 SourceLocation["uri"] = FileURI;
139 }
140
141 return SourceLocation;
142 }
143
144 /// Serialize a source range with begin and end locations.
serializeSourceRange(const PresumedLoc & BeginLoc,const PresumedLoc & EndLoc)145 Object serializeSourceRange(const PresumedLoc &BeginLoc,
146 const PresumedLoc &EndLoc) {
147 Object SourceRange;
148 serializeObject(SourceRange, "start", serializeSourcePosition(BeginLoc));
149 serializeObject(SourceRange, "end", serializeSourcePosition(EndLoc));
150 return SourceRange;
151 }
152
153 /// Serialize the availability attributes of a symbol.
154 ///
155 /// Availability information contains the introduced, deprecated, and obsoleted
156 /// versions of the symbol as semantic versions, if not default.
157 /// Availability information also contains flags to indicate if the symbol is
158 /// unconditionally unavailable or deprecated,
159 /// i.e. \c __attribute__((unavailable)) and \c __attribute__((deprecated)).
160 ///
161 /// \returns \c std::nullopt if the symbol has default availability attributes,
162 /// or an \c Array containing an object with the formatted availability
163 /// information.
serializeAvailability(const AvailabilityInfo & Avail)164 std::optional<Array> serializeAvailability(const AvailabilityInfo &Avail) {
165 if (Avail.isDefault())
166 return std::nullopt;
167
168 Array AvailabilityArray;
169
170 if (Avail.isUnconditionallyDeprecated()) {
171 Object UnconditionallyDeprecated;
172 UnconditionallyDeprecated["domain"] = "*";
173 UnconditionallyDeprecated["isUnconditionallyDeprecated"] = true;
174 AvailabilityArray.emplace_back(std::move(UnconditionallyDeprecated));
175 }
176
177 if (Avail.Domain.str() != "") {
178 Object Availability;
179 Availability["domain"] = Avail.Domain;
180
181 if (Avail.isUnavailable()) {
182 Availability["isUnconditionallyUnavailable"] = true;
183 } else {
184 serializeObject(Availability, "introduced",
185 serializeSemanticVersion(Avail.Introduced));
186 serializeObject(Availability, "deprecated",
187 serializeSemanticVersion(Avail.Deprecated));
188 serializeObject(Availability, "obsoleted",
189 serializeSemanticVersion(Avail.Obsoleted));
190 }
191
192 AvailabilityArray.emplace_back(std::move(Availability));
193 }
194
195 return AvailabilityArray;
196 }
197
198 /// Get the language name string for interface language references.
getLanguageName(Language Lang)199 StringRef getLanguageName(Language Lang) {
200 switch (Lang) {
201 case Language::C:
202 return "c";
203 case Language::ObjC:
204 return "objective-c";
205 case Language::CXX:
206 return "c++";
207 case Language::ObjCXX:
208 return "objective-c++";
209
210 // Unsupported language currently
211 case Language::OpenCL:
212 case Language::OpenCLCXX:
213 case Language::CUDA:
214 case Language::HIP:
215 case Language::HLSL:
216
217 // Languages that the frontend cannot parse and compile
218 case Language::Unknown:
219 case Language::Asm:
220 case Language::LLVM_IR:
221 case Language::CIR:
222 llvm_unreachable("Unsupported language kind");
223 }
224
225 llvm_unreachable("Unhandled language kind");
226 }
227
228 /// Serialize the identifier object as specified by the Symbol Graph format.
229 ///
230 /// The identifier property of a symbol contains the USR for precise and unique
231 /// references, and the interface language name.
serializeIdentifier(const APIRecord & Record,Language Lang)232 Object serializeIdentifier(const APIRecord &Record, Language Lang) {
233 Object Identifier;
234 Identifier["precise"] = Record.USR;
235 Identifier["interfaceLanguage"] = getLanguageName(Lang);
236
237 return Identifier;
238 }
239
240 /// Serialize the documentation comments attached to a symbol, as specified by
241 /// the Symbol Graph format.
242 ///
243 /// The Symbol Graph \c docComment object contains an array of lines. Each line
244 /// represents one line of striped documentation comment, with source range
245 /// information.
246 /// e.g.
247 /// \code
248 /// /// This is a documentation comment
249 /// ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~' First line.
250 /// /// with multiple lines.
251 /// ^~~~~~~~~~~~~~~~~~~~~~~' Second line.
252 /// \endcode
253 ///
254 /// \returns \c std::nullopt if \p Comment is empty, or an \c Object containing
255 /// the formatted lines.
serializeDocComment(const DocComment & Comment)256 std::optional<Object> serializeDocComment(const DocComment &Comment) {
257 if (Comment.empty())
258 return std::nullopt;
259
260 Object DocComment;
261
262 Array LinesArray;
263 for (const auto &CommentLine : Comment) {
264 Object Line;
265 Line["text"] = CommentLine.Text;
266 serializeObject(Line, "range",
267 serializeSourceRange(CommentLine.Begin, CommentLine.End));
268 LinesArray.emplace_back(std::move(Line));
269 }
270
271 serializeArray(DocComment, "lines", std::move(LinesArray));
272
273 return DocComment;
274 }
275
276 /// Serialize the declaration fragments of a symbol.
277 ///
278 /// The Symbol Graph declaration fragments is an array of tagged important
279 /// parts of a symbol's declaration. The fragments sequence can be joined to
280 /// form spans of declaration text, with attached information useful for
281 /// purposes like syntax-highlighting etc. For example:
282 /// \code
283 /// const int pi; -> "declarationFragments" : [
284 /// {
285 /// "kind" : "keyword",
286 /// "spelling" : "const"
287 /// },
288 /// {
289 /// "kind" : "text",
290 /// "spelling" : " "
291 /// },
292 /// {
293 /// "kind" : "typeIdentifier",
294 /// "preciseIdentifier" : "c:I",
295 /// "spelling" : "int"
296 /// },
297 /// {
298 /// "kind" : "text",
299 /// "spelling" : " "
300 /// },
301 /// {
302 /// "kind" : "identifier",
303 /// "spelling" : "pi"
304 /// }
305 /// ]
306 /// \endcode
307 ///
308 /// \returns \c std::nullopt if \p DF is empty, or an \c Array containing the
309 /// formatted declaration fragments array.
310 std::optional<Array>
serializeDeclarationFragments(const DeclarationFragments & DF)311 serializeDeclarationFragments(const DeclarationFragments &DF) {
312 if (DF.getFragments().empty())
313 return std::nullopt;
314
315 Array Fragments;
316 for (const auto &F : DF.getFragments()) {
317 Object Fragment;
318 Fragment["spelling"] = F.Spelling;
319 Fragment["kind"] = DeclarationFragments::getFragmentKindString(F.Kind);
320 if (!F.PreciseIdentifier.empty())
321 Fragment["preciseIdentifier"] = F.PreciseIdentifier;
322 Fragments.emplace_back(std::move(Fragment));
323 }
324
325 return Fragments;
326 }
327
328 /// Serialize the \c names field of a symbol as specified by the Symbol Graph
329 /// format.
330 ///
331 /// The Symbol Graph names field contains multiple representations of a symbol
332 /// that can be used for different applications:
333 /// - \c title : The simple declared name of the symbol;
334 /// - \c subHeading : An array of declaration fragments that provides tags,
335 /// and potentially more tokens (for example the \c +/- symbol for
336 /// Objective-C methods). Can be used as sub-headings for documentation.
serializeNames(const APIRecord * Record)337 Object serializeNames(const APIRecord *Record) {
338 Object Names;
339 Names["title"] = Record->Name;
340
341 serializeArray(Names, "subHeading",
342 serializeDeclarationFragments(Record->SubHeading));
343 DeclarationFragments NavigatorFragments;
344 // The +/- prefix for Objective-C methods is important information, and
345 // should be included in the navigator fragment. The entire subheading is
346 // not included as it can contain too much information for other records.
347 switch (Record->getKind()) {
348 case APIRecord::RK_ObjCClassMethod:
349 NavigatorFragments.append("+ ", DeclarationFragments::FragmentKind::Text,
350 /*PreciseIdentifier*/ "");
351 break;
352 case APIRecord::RK_ObjCInstanceMethod:
353 NavigatorFragments.append("- ", DeclarationFragments::FragmentKind::Text,
354 /*PreciseIdentifier*/ "");
355 break;
356 default:
357 break;
358 }
359
360 NavigatorFragments.append(Record->Name,
361 DeclarationFragments::FragmentKind::Identifier,
362 /*PreciseIdentifier*/ "");
363 serializeArray(Names, "navigator",
364 serializeDeclarationFragments(NavigatorFragments));
365
366 return Names;
367 }
368
serializeSymbolKind(APIRecord::RecordKind RK,Language Lang)369 Object serializeSymbolKind(APIRecord::RecordKind RK, Language Lang) {
370 auto AddLangPrefix = [&Lang](StringRef S) -> std::string {
371 return (getLanguageName(Lang) + "." + S).str();
372 };
373
374 Object Kind;
375 switch (RK) {
376 case APIRecord::RK_Unknown:
377 Kind["identifier"] = AddLangPrefix("unknown");
378 Kind["displayName"] = "Unknown";
379 break;
380 case APIRecord::RK_Namespace:
381 Kind["identifier"] = AddLangPrefix("namespace");
382 Kind["displayName"] = "Namespace";
383 break;
384 case APIRecord::RK_GlobalFunction:
385 Kind["identifier"] = AddLangPrefix("func");
386 Kind["displayName"] = "Function";
387 break;
388 case APIRecord::RK_GlobalFunctionTemplate:
389 Kind["identifier"] = AddLangPrefix("func");
390 Kind["displayName"] = "Function Template";
391 break;
392 case APIRecord::RK_GlobalFunctionTemplateSpecialization:
393 Kind["identifier"] = AddLangPrefix("func");
394 Kind["displayName"] = "Function Template Specialization";
395 break;
396 case APIRecord::RK_GlobalVariableTemplate:
397 Kind["identifier"] = AddLangPrefix("var");
398 Kind["displayName"] = "Global Variable Template";
399 break;
400 case APIRecord::RK_GlobalVariableTemplateSpecialization:
401 Kind["identifier"] = AddLangPrefix("var");
402 Kind["displayName"] = "Global Variable Template Specialization";
403 break;
404 case APIRecord::RK_GlobalVariableTemplatePartialSpecialization:
405 Kind["identifier"] = AddLangPrefix("var");
406 Kind["displayName"] = "Global Variable Template Partial Specialization";
407 break;
408 case APIRecord::RK_GlobalVariable:
409 Kind["identifier"] = AddLangPrefix("var");
410 Kind["displayName"] = "Global Variable";
411 break;
412 case APIRecord::RK_EnumConstant:
413 Kind["identifier"] = AddLangPrefix("enum.case");
414 Kind["displayName"] = "Enumeration Case";
415 break;
416 case APIRecord::RK_Enum:
417 Kind["identifier"] = AddLangPrefix("enum");
418 Kind["displayName"] = "Enumeration";
419 break;
420 case APIRecord::RK_StructField:
421 Kind["identifier"] = AddLangPrefix("property");
422 Kind["displayName"] = "Instance Property";
423 break;
424 case APIRecord::RK_Struct:
425 Kind["identifier"] = AddLangPrefix("struct");
426 Kind["displayName"] = "Structure";
427 break;
428 case APIRecord::RK_UnionField:
429 Kind["identifier"] = AddLangPrefix("property");
430 Kind["displayName"] = "Instance Property";
431 break;
432 case APIRecord::RK_Union:
433 Kind["identifier"] = AddLangPrefix("union");
434 Kind["displayName"] = "Union";
435 break;
436 case APIRecord::RK_CXXField:
437 Kind["identifier"] = AddLangPrefix("property");
438 Kind["displayName"] = "Instance Property";
439 break;
440 case APIRecord::RK_StaticField:
441 Kind["identifier"] = AddLangPrefix("type.property");
442 Kind["displayName"] = "Type Property";
443 break;
444 case APIRecord::RK_ClassTemplate:
445 case APIRecord::RK_ClassTemplateSpecialization:
446 case APIRecord::RK_ClassTemplatePartialSpecialization:
447 case APIRecord::RK_CXXClass:
448 Kind["identifier"] = AddLangPrefix("class");
449 Kind["displayName"] = "Class";
450 break;
451 case APIRecord::RK_CXXMethodTemplate:
452 Kind["identifier"] = AddLangPrefix("method");
453 Kind["displayName"] = "Method Template";
454 break;
455 case APIRecord::RK_CXXMethodTemplateSpecialization:
456 Kind["identifier"] = AddLangPrefix("method");
457 Kind["displayName"] = "Method Template Specialization";
458 break;
459 case APIRecord::RK_CXXFieldTemplate:
460 Kind["identifier"] = AddLangPrefix("property");
461 Kind["displayName"] = "Template Property";
462 break;
463 case APIRecord::RK_Concept:
464 Kind["identifier"] = AddLangPrefix("concept");
465 Kind["displayName"] = "Concept";
466 break;
467 case APIRecord::RK_CXXStaticMethod:
468 Kind["identifier"] = AddLangPrefix("type.method");
469 Kind["displayName"] = "Static Method";
470 break;
471 case APIRecord::RK_CXXInstanceMethod:
472 Kind["identifier"] = AddLangPrefix("method");
473 Kind["displayName"] = "Instance Method";
474 break;
475 case APIRecord::RK_CXXConstructorMethod:
476 Kind["identifier"] = AddLangPrefix("method");
477 Kind["displayName"] = "Constructor";
478 break;
479 case APIRecord::RK_CXXDestructorMethod:
480 Kind["identifier"] = AddLangPrefix("method");
481 Kind["displayName"] = "Destructor";
482 break;
483 case APIRecord::RK_ObjCIvar:
484 Kind["identifier"] = AddLangPrefix("ivar");
485 Kind["displayName"] = "Instance Variable";
486 break;
487 case APIRecord::RK_ObjCInstanceMethod:
488 Kind["identifier"] = AddLangPrefix("method");
489 Kind["displayName"] = "Instance Method";
490 break;
491 case APIRecord::RK_ObjCClassMethod:
492 Kind["identifier"] = AddLangPrefix("type.method");
493 Kind["displayName"] = "Type Method";
494 break;
495 case APIRecord::RK_ObjCInstanceProperty:
496 Kind["identifier"] = AddLangPrefix("property");
497 Kind["displayName"] = "Instance Property";
498 break;
499 case APIRecord::RK_ObjCClassProperty:
500 Kind["identifier"] = AddLangPrefix("type.property");
501 Kind["displayName"] = "Type Property";
502 break;
503 case APIRecord::RK_ObjCInterface:
504 Kind["identifier"] = AddLangPrefix("class");
505 Kind["displayName"] = "Class";
506 break;
507 case APIRecord::RK_ObjCCategory:
508 Kind["identifier"] = AddLangPrefix("class.extension");
509 Kind["displayName"] = "Class Extension";
510 break;
511 case APIRecord::RK_ObjCProtocol:
512 Kind["identifier"] = AddLangPrefix("protocol");
513 Kind["displayName"] = "Protocol";
514 break;
515 case APIRecord::RK_MacroDefinition:
516 Kind["identifier"] = AddLangPrefix("macro");
517 Kind["displayName"] = "Macro";
518 break;
519 case APIRecord::RK_Typedef:
520 Kind["identifier"] = AddLangPrefix("typealias");
521 Kind["displayName"] = "Type Alias";
522 break;
523 default:
524 llvm_unreachable("API Record with uninstantiable kind");
525 }
526
527 return Kind;
528 }
529
530 /// Serialize the symbol kind information.
531 ///
532 /// The Symbol Graph symbol kind property contains a shorthand \c identifier
533 /// which is prefixed by the source language name, useful for tooling to parse
534 /// the kind, and a \c displayName for rendering human-readable names.
serializeSymbolKind(const APIRecord & Record,Language Lang)535 Object serializeSymbolKind(const APIRecord &Record, Language Lang) {
536 return serializeSymbolKind(Record.KindForDisplay, Lang);
537 }
538
539 /// Serialize the function signature field, as specified by the
540 /// Symbol Graph format.
541 ///
542 /// The Symbol Graph function signature property contains two arrays.
543 /// - The \c returns array is the declaration fragments of the return type;
544 /// - The \c parameters array contains names and declaration fragments of the
545 /// parameters.
546 template <typename RecordTy>
serializeFunctionSignatureMixin(Object & Paren,const RecordTy & Record)547 void serializeFunctionSignatureMixin(Object &Paren, const RecordTy &Record) {
548 const auto &FS = Record.Signature;
549 if (FS.empty())
550 return;
551
552 Object Signature;
553 serializeArray(Signature, "returns",
554 serializeDeclarationFragments(FS.getReturnType()));
555
556 Array Parameters;
557 for (const auto &P : FS.getParameters()) {
558 Object Parameter;
559 Parameter["name"] = P.Name;
560 serializeArray(Parameter, "declarationFragments",
561 serializeDeclarationFragments(P.Fragments));
562 Parameters.emplace_back(std::move(Parameter));
563 }
564
565 if (!Parameters.empty())
566 Signature["parameters"] = std::move(Parameters);
567
568 serializeObject(Paren, "functionSignature", std::move(Signature));
569 }
570
571 template <typename RecordTy>
serializeTemplateMixin(Object & Paren,const RecordTy & Record)572 void serializeTemplateMixin(Object &Paren, const RecordTy &Record) {
573 const auto &Template = Record.Templ;
574 if (Template.empty())
575 return;
576
577 Object Generics;
578 Array GenericParameters;
579 for (const auto &Param : Template.getParameters()) {
580 Object Parameter;
581 Parameter["name"] = Param.Name;
582 Parameter["index"] = Param.Index;
583 Parameter["depth"] = Param.Depth;
584 GenericParameters.emplace_back(std::move(Parameter));
585 }
586 if (!GenericParameters.empty())
587 Generics["parameters"] = std::move(GenericParameters);
588
589 Array GenericConstraints;
590 for (const auto &Constr : Template.getConstraints()) {
591 Object Constraint;
592 Constraint["kind"] = Constr.Kind;
593 Constraint["lhs"] = Constr.LHS;
594 Constraint["rhs"] = Constr.RHS;
595 GenericConstraints.emplace_back(std::move(Constraint));
596 }
597
598 if (!GenericConstraints.empty())
599 Generics["constraints"] = std::move(GenericConstraints);
600
601 serializeObject(Paren, "swiftGenerics", Generics);
602 }
603
generateParentContexts(const SmallVectorImpl<SymbolReference> & Parents,Language Lang)604 Array generateParentContexts(const SmallVectorImpl<SymbolReference> &Parents,
605 Language Lang) {
606 Array ParentContexts;
607
608 for (const auto &Parent : Parents) {
609 Object Elem;
610 Elem["usr"] = Parent.USR;
611 Elem["name"] = Parent.Name;
612 if (Parent.Record)
613 Elem["kind"] = serializeSymbolKind(Parent.Record->KindForDisplay,
614 Lang)["identifier"];
615 else
616 Elem["kind"] =
617 serializeSymbolKind(APIRecord::RK_Unknown, Lang)["identifier"];
618 ParentContexts.emplace_back(std::move(Elem));
619 }
620
621 return ParentContexts;
622 }
623
624 /// Walk the records parent information in reverse to generate a hierarchy
625 /// suitable for serialization.
626 SmallVector<SymbolReference, 8>
generateHierarchyFromRecord(const APIRecord * Record)627 generateHierarchyFromRecord(const APIRecord *Record) {
628 SmallVector<SymbolReference, 8> ReverseHierarchy;
629 for (const auto *Current = Record; Current != nullptr;
630 Current = Current->Parent.Record)
631 ReverseHierarchy.emplace_back(Current);
632
633 return SmallVector<SymbolReference, 8>(
634 std::make_move_iterator(ReverseHierarchy.rbegin()),
635 std::make_move_iterator(ReverseHierarchy.rend()));
636 }
637
getHierarchyReference(const APIRecord * Record,const APISet & API)638 SymbolReference getHierarchyReference(const APIRecord *Record,
639 const APISet &API) {
640 // If the parent is a category extended from internal module then we need to
641 // pretend this belongs to the associated interface.
642 if (auto *CategoryRecord = dyn_cast_or_null<ObjCCategoryRecord>(Record)) {
643 return CategoryRecord->Interface;
644 // FIXME: TODO generate path components correctly for categories extending
645 // an external module.
646 }
647
648 return SymbolReference(Record);
649 }
650
651 } // namespace
652
addSymbol(Object && Symbol)653 Object *ExtendedModule::addSymbol(Object &&Symbol) {
654 Symbols.emplace_back(std::move(Symbol));
655 return Symbols.back().getAsObject();
656 }
657
addRelationship(Object && Relationship)658 void ExtendedModule::addRelationship(Object &&Relationship) {
659 Relationships.emplace_back(std::move(Relationship));
660 }
661
662 /// Defines the format version emitted by SymbolGraphSerializer.
663 const VersionTuple SymbolGraphSerializer::FormatVersion{0, 5, 3};
664
serializeMetadata() const665 Object SymbolGraphSerializer::serializeMetadata() const {
666 Object Metadata;
667 serializeObject(Metadata, "formatVersion",
668 serializeSemanticVersion(FormatVersion));
669 Metadata["generator"] = clang::getClangFullVersion();
670 return Metadata;
671 }
672
673 Object
serializeModuleObject(StringRef ModuleName) const674 SymbolGraphSerializer::serializeModuleObject(StringRef ModuleName) const {
675 Object Module;
676 Module["name"] = ModuleName;
677 serializeObject(Module, "platform", serializePlatform(API.getTarget()));
678 return Module;
679 }
680
shouldSkip(const APIRecord * Record) const681 bool SymbolGraphSerializer::shouldSkip(const APIRecord *Record) const {
682 if (!Record)
683 return true;
684
685 // Skip unconditionally unavailable symbols
686 if (Record->Availability.isUnconditionallyUnavailable())
687 return true;
688
689 // Filter out symbols prefixed with an underscored as they are understood to
690 // be symbols clients should not use.
691 if (Record->Name.starts_with("_"))
692 return true;
693
694 // Skip explicitly ignored symbols.
695 if (IgnoresList.shouldIgnore(Record->Name))
696 return true;
697
698 return false;
699 }
700
getModuleForCurrentSymbol()701 ExtendedModule &SymbolGraphSerializer::getModuleForCurrentSymbol() {
702 if (!ForceEmitToMainModule && ModuleForCurrentSymbol)
703 return *ModuleForCurrentSymbol;
704
705 return MainModule;
706 }
707
serializePathComponents(const APIRecord * Record) const708 Array SymbolGraphSerializer::serializePathComponents(
709 const APIRecord *Record) const {
710 return Array(map_range(Hierarchy, [](auto Elt) { return Elt.Name; }));
711 }
712
getRelationshipString(RelationshipKind Kind)713 StringRef SymbolGraphSerializer::getRelationshipString(RelationshipKind Kind) {
714 switch (Kind) {
715 case RelationshipKind::MemberOf:
716 return "memberOf";
717 case RelationshipKind::InheritsFrom:
718 return "inheritsFrom";
719 case RelationshipKind::ConformsTo:
720 return "conformsTo";
721 case RelationshipKind::ExtensionTo:
722 return "extensionTo";
723 }
724 llvm_unreachable("Unhandled relationship kind");
725 }
726
serializeRelationship(RelationshipKind Kind,const SymbolReference & Source,const SymbolReference & Target,ExtendedModule & Into)727 void SymbolGraphSerializer::serializeRelationship(RelationshipKind Kind,
728 const SymbolReference &Source,
729 const SymbolReference &Target,
730 ExtendedModule &Into) {
731 Object Relationship;
732 SmallString<64> TestRelLabel;
733 if (EmitSymbolLabelsForTesting) {
734 llvm::raw_svector_ostream OS(TestRelLabel);
735 OS << SymbolGraphSerializer::getRelationshipString(Kind) << " $ "
736 << Source.USR << " $ ";
737 if (Target.USR.empty())
738 OS << Target.Name;
739 else
740 OS << Target.USR;
741 Relationship["!testRelLabel"] = TestRelLabel;
742 }
743 Relationship["source"] = Source.USR;
744 Relationship["target"] = Target.USR;
745 Relationship["targetFallback"] = Target.Name;
746 Relationship["kind"] = SymbolGraphSerializer::getRelationshipString(Kind);
747
748 if (ForceEmitToMainModule)
749 MainModule.addRelationship(std::move(Relationship));
750 else
751 Into.addRelationship(std::move(Relationship));
752 }
753
getConstraintString(ConstraintKind Kind)754 StringRef SymbolGraphSerializer::getConstraintString(ConstraintKind Kind) {
755 switch (Kind) {
756 case ConstraintKind::Conformance:
757 return "conformance";
758 case ConstraintKind::ConditionalConformance:
759 return "conditionalConformance";
760 }
761 llvm_unreachable("Unhandled constraint kind");
762 }
763
serializeAPIRecord(const APIRecord * Record)764 void SymbolGraphSerializer::serializeAPIRecord(const APIRecord *Record) {
765 Object Obj;
766
767 // If we need symbol labels for testing emit the USR as the value and the key
768 // starts with '!'' to ensure it ends up at the top of the object.
769 if (EmitSymbolLabelsForTesting)
770 Obj["!testLabel"] = Record->USR;
771
772 serializeObject(Obj, "identifier",
773 serializeIdentifier(*Record, API.getLanguage()));
774 serializeObject(Obj, "kind", serializeSymbolKind(*Record, API.getLanguage()));
775 serializeObject(Obj, "names", serializeNames(Record));
776 serializeObject(
777 Obj, "location",
778 serializeSourceLocation(Record->Location, /*IncludeFileURI=*/true));
779 serializeArray(Obj, "availability",
780 serializeAvailability(Record->Availability));
781 serializeObject(Obj, "docComment", serializeDocComment(Record->Comment));
782 serializeArray(Obj, "declarationFragments",
783 serializeDeclarationFragments(Record->Declaration));
784
785 Obj["pathComponents"] = serializePathComponents(Record);
786 Obj["accessLevel"] = Record->Access.getAccess();
787
788 ExtendedModule &Module = getModuleForCurrentSymbol();
789 // If the hierarchy has at least one parent and child.
790 if (Hierarchy.size() >= 2)
791 serializeRelationship(MemberOf, Hierarchy.back(),
792 Hierarchy[Hierarchy.size() - 2], Module);
793
794 CurrentSymbol = Module.addSymbol(std::move(Obj));
795 }
796
traverseAPIRecord(const APIRecord * Record)797 bool SymbolGraphSerializer::traverseAPIRecord(const APIRecord *Record) {
798 if (!Record)
799 return true;
800 if (shouldSkip(Record))
801 return true;
802 Hierarchy.push_back(getHierarchyReference(Record, API));
803 // Defer traversal mechanics to APISetVisitor base implementation
804 auto RetVal = Base::traverseAPIRecord(Record);
805 Hierarchy.pop_back();
806 return RetVal;
807 }
808
visitAPIRecord(const APIRecord * Record)809 bool SymbolGraphSerializer::visitAPIRecord(const APIRecord *Record) {
810 serializeAPIRecord(Record);
811 return true;
812 }
813
visitGlobalFunctionRecord(const GlobalFunctionRecord * Record)814 bool SymbolGraphSerializer::visitGlobalFunctionRecord(
815 const GlobalFunctionRecord *Record) {
816 if (!CurrentSymbol)
817 return true;
818
819 serializeFunctionSignatureMixin(*CurrentSymbol, *Record);
820 return true;
821 }
822
visitCXXClassRecord(const CXXClassRecord * Record)823 bool SymbolGraphSerializer::visitCXXClassRecord(const CXXClassRecord *Record) {
824 if (!CurrentSymbol)
825 return true;
826
827 for (const auto &Base : Record->Bases)
828 serializeRelationship(RelationshipKind::InheritsFrom, Record, Base,
829 getModuleForCurrentSymbol());
830 return true;
831 }
832
visitClassTemplateRecord(const ClassTemplateRecord * Record)833 bool SymbolGraphSerializer::visitClassTemplateRecord(
834 const ClassTemplateRecord *Record) {
835 if (!CurrentSymbol)
836 return true;
837
838 serializeTemplateMixin(*CurrentSymbol, *Record);
839 return true;
840 }
841
visitClassTemplatePartialSpecializationRecord(const ClassTemplatePartialSpecializationRecord * Record)842 bool SymbolGraphSerializer::visitClassTemplatePartialSpecializationRecord(
843 const ClassTemplatePartialSpecializationRecord *Record) {
844 if (!CurrentSymbol)
845 return true;
846
847 serializeTemplateMixin(*CurrentSymbol, *Record);
848 return true;
849 }
850
visitCXXMethodRecord(const CXXMethodRecord * Record)851 bool SymbolGraphSerializer::visitCXXMethodRecord(
852 const CXXMethodRecord *Record) {
853 if (!CurrentSymbol)
854 return true;
855
856 serializeFunctionSignatureMixin(*CurrentSymbol, *Record);
857 return true;
858 }
859
visitCXXMethodTemplateRecord(const CXXMethodTemplateRecord * Record)860 bool SymbolGraphSerializer::visitCXXMethodTemplateRecord(
861 const CXXMethodTemplateRecord *Record) {
862 if (!CurrentSymbol)
863 return true;
864
865 serializeTemplateMixin(*CurrentSymbol, *Record);
866 return true;
867 }
868
visitCXXFieldTemplateRecord(const CXXFieldTemplateRecord * Record)869 bool SymbolGraphSerializer::visitCXXFieldTemplateRecord(
870 const CXXFieldTemplateRecord *Record) {
871 if (!CurrentSymbol)
872 return true;
873
874 serializeTemplateMixin(*CurrentSymbol, *Record);
875 return true;
876 }
877
visitConceptRecord(const ConceptRecord * Record)878 bool SymbolGraphSerializer::visitConceptRecord(const ConceptRecord *Record) {
879 if (!CurrentSymbol)
880 return true;
881
882 serializeTemplateMixin(*CurrentSymbol, *Record);
883 return true;
884 }
885
visitGlobalVariableTemplateRecord(const GlobalVariableTemplateRecord * Record)886 bool SymbolGraphSerializer::visitGlobalVariableTemplateRecord(
887 const GlobalVariableTemplateRecord *Record) {
888 if (!CurrentSymbol)
889 return true;
890
891 serializeTemplateMixin(*CurrentSymbol, *Record);
892 return true;
893 }
894
895 bool SymbolGraphSerializer::
visitGlobalVariableTemplatePartialSpecializationRecord(const GlobalVariableTemplatePartialSpecializationRecord * Record)896 visitGlobalVariableTemplatePartialSpecializationRecord(
897 const GlobalVariableTemplatePartialSpecializationRecord *Record) {
898 if (!CurrentSymbol)
899 return true;
900
901 serializeTemplateMixin(*CurrentSymbol, *Record);
902 return true;
903 }
904
visitGlobalFunctionTemplateRecord(const GlobalFunctionTemplateRecord * Record)905 bool SymbolGraphSerializer::visitGlobalFunctionTemplateRecord(
906 const GlobalFunctionTemplateRecord *Record) {
907 if (!CurrentSymbol)
908 return true;
909
910 serializeTemplateMixin(*CurrentSymbol, *Record);
911 return true;
912 }
913
visitObjCContainerRecord(const ObjCContainerRecord * Record)914 bool SymbolGraphSerializer::visitObjCContainerRecord(
915 const ObjCContainerRecord *Record) {
916 if (!CurrentSymbol)
917 return true;
918
919 for (const auto &Protocol : Record->Protocols)
920 serializeRelationship(ConformsTo, Record, Protocol,
921 getModuleForCurrentSymbol());
922
923 return true;
924 }
925
visitObjCInterfaceRecord(const ObjCInterfaceRecord * Record)926 bool SymbolGraphSerializer::visitObjCInterfaceRecord(
927 const ObjCInterfaceRecord *Record) {
928 if (!CurrentSymbol)
929 return true;
930
931 if (!Record->SuperClass.empty())
932 serializeRelationship(InheritsFrom, Record, Record->SuperClass,
933 getModuleForCurrentSymbol());
934 return true;
935 }
936
traverseObjCCategoryRecord(const ObjCCategoryRecord * Record)937 bool SymbolGraphSerializer::traverseObjCCategoryRecord(
938 const ObjCCategoryRecord *Record) {
939 if (SkipSymbolsInCategoriesToExternalTypes &&
940 !API.findRecordForUSR(Record->Interface.USR))
941 return true;
942
943 auto *CurrentModule = ModuleForCurrentSymbol;
944 if (auto ModuleExtendedByRecord = Record->getExtendedExternalModule())
945 ModuleForCurrentSymbol = &ExtendedModules[*ModuleExtendedByRecord];
946
947 if (!walkUpFromObjCCategoryRecord(Record))
948 return false;
949
950 bool RetVal = traverseRecordContext(Record);
951 ModuleForCurrentSymbol = CurrentModule;
952 return RetVal;
953 }
954
walkUpFromObjCCategoryRecord(const ObjCCategoryRecord * Record)955 bool SymbolGraphSerializer::walkUpFromObjCCategoryRecord(
956 const ObjCCategoryRecord *Record) {
957 return visitObjCCategoryRecord(Record);
958 }
959
visitObjCCategoryRecord(const ObjCCategoryRecord * Record)960 bool SymbolGraphSerializer::visitObjCCategoryRecord(
961 const ObjCCategoryRecord *Record) {
962 // If we need to create a record for the category in the future do so here,
963 // otherwise everything is set up to pretend that the category is in fact the
964 // interface it extends.
965 for (const auto &Protocol : Record->Protocols)
966 serializeRelationship(ConformsTo, Record->Interface, Protocol,
967 getModuleForCurrentSymbol());
968
969 return true;
970 }
971
visitObjCMethodRecord(const ObjCMethodRecord * Record)972 bool SymbolGraphSerializer::visitObjCMethodRecord(
973 const ObjCMethodRecord *Record) {
974 if (!CurrentSymbol)
975 return true;
976
977 serializeFunctionSignatureMixin(*CurrentSymbol, *Record);
978 return true;
979 }
980
visitObjCInstanceVariableRecord(const ObjCInstanceVariableRecord * Record)981 bool SymbolGraphSerializer::visitObjCInstanceVariableRecord(
982 const ObjCInstanceVariableRecord *Record) {
983 // FIXME: serialize ivar access control here.
984 return true;
985 }
986
walkUpFromTypedefRecord(const TypedefRecord * Record)987 bool SymbolGraphSerializer::walkUpFromTypedefRecord(
988 const TypedefRecord *Record) {
989 // Short-circuit walking up the class hierarchy and handle creating typedef
990 // symbol objects manually as there are additional symbol dropping rules to
991 // respect.
992 return visitTypedefRecord(Record);
993 }
994
visitTypedefRecord(const TypedefRecord * Record)995 bool SymbolGraphSerializer::visitTypedefRecord(const TypedefRecord *Record) {
996 // Typedefs of anonymous types have their entries unified with the underlying
997 // type.
998 bool ShouldDrop = Record->UnderlyingType.Name.empty();
999 // enums declared with `NS_OPTION` have a named enum and a named typedef, with
1000 // the same name
1001 ShouldDrop |= (Record->UnderlyingType.Name == Record->Name);
1002 if (ShouldDrop)
1003 return true;
1004
1005 // Create the symbol record if the other symbol droppping rules permit it.
1006 serializeAPIRecord(Record);
1007 if (!CurrentSymbol)
1008 return true;
1009
1010 (*CurrentSymbol)["type"] = Record->UnderlyingType.USR;
1011
1012 return true;
1013 }
1014
serializeSingleRecord(const APIRecord * Record)1015 void SymbolGraphSerializer::serializeSingleRecord(const APIRecord *Record) {
1016 switch (Record->getKind()) {
1017 // dispatch to the relevant walkUpFromMethod
1018 #define CONCRETE_RECORD(CLASS, BASE, KIND) \
1019 case APIRecord::KIND: { \
1020 walkUpFrom##CLASS(static_cast<const CLASS *>(Record)); \
1021 break; \
1022 }
1023 #include "clang/ExtractAPI/APIRecords.inc"
1024 // otherwise fallback on the only behavior we can implement safely.
1025 case APIRecord::RK_Unknown:
1026 visitAPIRecord(Record);
1027 break;
1028 default:
1029 llvm_unreachable("API Record with uninstantiable kind");
1030 }
1031 }
1032
serializeGraph(StringRef ModuleName,ExtendedModule && EM)1033 Object SymbolGraphSerializer::serializeGraph(StringRef ModuleName,
1034 ExtendedModule &&EM) {
1035 Object Root;
1036 serializeObject(Root, "metadata", serializeMetadata());
1037 serializeObject(Root, "module", serializeModuleObject(ModuleName));
1038
1039 Root["symbols"] = std::move(EM.Symbols);
1040 Root["relationships"] = std::move(EM.Relationships);
1041
1042 return Root;
1043 }
1044
serializeGraphToStream(raw_ostream & OS,SymbolGraphSerializerOption Options,StringRef ModuleName,ExtendedModule && EM)1045 void SymbolGraphSerializer::serializeGraphToStream(
1046 raw_ostream &OS, SymbolGraphSerializerOption Options, StringRef ModuleName,
1047 ExtendedModule &&EM) {
1048 Object Root = serializeGraph(ModuleName, std::move(EM));
1049 if (Options.Compact)
1050 OS << formatv("{0}", json::Value(std::move(Root))) << "\n";
1051 else
1052 OS << formatv("{0:2}", json::Value(std::move(Root))) << "\n";
1053 }
1054
serializeMainSymbolGraph(raw_ostream & OS,const APISet & API,const APIIgnoresList & IgnoresList,SymbolGraphSerializerOption Options)1055 void SymbolGraphSerializer::serializeMainSymbolGraph(
1056 raw_ostream &OS, const APISet &API, const APIIgnoresList &IgnoresList,
1057 SymbolGraphSerializerOption Options) {
1058 SymbolGraphSerializer Serializer(
1059 API, IgnoresList, Options.EmitSymbolLabelsForTesting,
1060 /*ForceEmitToMainModule=*/true,
1061 /*SkipSymbolsInCategoriesToExternalTypes=*/true);
1062
1063 Serializer.traverseAPISet();
1064 Serializer.serializeGraphToStream(OS, Options, API.ProductName,
1065 std::move(Serializer.MainModule));
1066 // FIXME: TODO handle extended modules here
1067 }
1068
serializeWithExtensionGraphs(raw_ostream & MainOutput,const APISet & API,const APIIgnoresList & IgnoresList,llvm::function_ref<std::unique_ptr<llvm::raw_pwrite_stream> (Twine BaseName)> CreateOutputStream,SymbolGraphSerializerOption Options)1069 void SymbolGraphSerializer::serializeWithExtensionGraphs(
1070 raw_ostream &MainOutput, const APISet &API,
1071 const APIIgnoresList &IgnoresList,
1072 llvm::function_ref<std::unique_ptr<llvm::raw_pwrite_stream>(Twine BaseName)>
1073 CreateOutputStream,
1074 SymbolGraphSerializerOption Options) {
1075 SymbolGraphSerializer Serializer(API, IgnoresList,
1076 Options.EmitSymbolLabelsForTesting);
1077 Serializer.traverseAPISet();
1078
1079 Serializer.serializeGraphToStream(MainOutput, Options, API.ProductName,
1080 std::move(Serializer.MainModule));
1081
1082 for (auto &ExtensionSGF : Serializer.ExtendedModules) {
1083 if (auto ExtensionOS =
1084 CreateOutputStream(API.ProductName + "@" + ExtensionSGF.getKey()))
1085 Serializer.serializeGraphToStream(*ExtensionOS, Options, API.ProductName,
1086 std::move(ExtensionSGF.getValue()));
1087 }
1088 }
1089
1090 std::optional<Object>
serializeSingleSymbolSGF(StringRef USR,const APISet & API)1091 SymbolGraphSerializer::serializeSingleSymbolSGF(StringRef USR,
1092 const APISet &API) {
1093 APIRecord *Record = API.findRecordForUSR(USR);
1094 if (!Record)
1095 return {};
1096
1097 Object Root;
1098 APIIgnoresList EmptyIgnores;
1099 SymbolGraphSerializer Serializer(API, EmptyIgnores,
1100 /*EmitSymbolLabelsForTesting*/ false,
1101 /*ForceEmitToMainModule*/ true);
1102
1103 // Set up serializer parent chain
1104 Serializer.Hierarchy = generateHierarchyFromRecord(Record);
1105
1106 Serializer.serializeSingleRecord(Record);
1107 serializeObject(Root, "symbolGraph",
1108 Serializer.serializeGraph(API.ProductName,
1109 std::move(Serializer.MainModule)));
1110
1111 Language Lang = API.getLanguage();
1112 serializeArray(Root, "parentContexts",
1113 generateParentContexts(Serializer.Hierarchy, Lang));
1114
1115 Array RelatedSymbols;
1116
1117 for (const auto &Fragment : Record->Declaration.getFragments()) {
1118 // If we don't have a USR there isn't much we can do.
1119 if (Fragment.PreciseIdentifier.empty())
1120 continue;
1121
1122 APIRecord *RelatedRecord = API.findRecordForUSR(Fragment.PreciseIdentifier);
1123
1124 // If we can't find the record let's skip.
1125 if (!RelatedRecord)
1126 continue;
1127
1128 Object RelatedSymbol;
1129 RelatedSymbol["usr"] = RelatedRecord->USR;
1130 RelatedSymbol["declarationLanguage"] = getLanguageName(Lang);
1131 RelatedSymbol["accessLevel"] = RelatedRecord->Access.getAccess();
1132 RelatedSymbol["filePath"] = RelatedRecord->Location.getFilename();
1133 RelatedSymbol["moduleName"] = API.ProductName;
1134 RelatedSymbol["isSystem"] = RelatedRecord->IsFromSystemHeader;
1135
1136 serializeArray(RelatedSymbol, "parentContexts",
1137 generateParentContexts(
1138 generateHierarchyFromRecord(RelatedRecord), Lang));
1139
1140 RelatedSymbols.push_back(std::move(RelatedSymbol));
1141 }
1142
1143 serializeArray(Root, "relatedSymbols", RelatedSymbols);
1144 return Root;
1145 }
1146