Merge pull request #29414 from bitjammer/acgarland/rdar-58853310-interface-language

SymbolGraph: Serialize interface language

Author: bitjammer

lib/SymbolGraphGen/Edge.cpp

@@ -25,12 +25,14 @@ void Edge::serialize(llvm::json::OStream &OS) const {
 
     // In case a dependent module isn't available, serialize a fallback name.
     auto TargetModuleName = Target->getModuleContext()->getName().str();
+
     if (TargetModuleName != Walker->M.getName().str()) {
-      auto TargetSymbolIdentifier = Walker->getSymbolIdentifier(Target);
-      auto TargetComponents = TargetSymbolIdentifier.SimpleComponents;
+      SmallVector<SmallString<32>, 8> TargetPathComponents;
+      Walker->getPathComponents(Target, TargetPathComponents);
+
       SmallString<128> Scratch(TargetModuleName);
-      for (auto it = TargetComponents.begin();
-           it != TargetComponents.end(); ++it) {
+      for (auto it = TargetPathComponents.begin();
+           it != TargetPathComponents.end(); ++it) {
         Scratch.push_back('.');
         Scratch.append(*it);
       }

lib/SymbolGraphGen/Symbol.cpp

@@ -76,15 +76,30 @@ void Symbol::serializeKind(llvm::json::OStream &OS) const {
 
 void Symbol::serializeIdentifier(SymbolGraphASTWalker &Walker,
                                  llvm::json::OStream &OS) const {
-  AttributeRAII A("identifier", OS);
-  Walker.getSymbolIdentifier(VD).serialize(OS);
+  OS.attributeObject("identifier", [&](){
+    OS.attribute("precise", Walker.getUSR(VD));
+    OS.attribute("interfaceLanguage", "swift");
+  });
+}
+
+void Symbol::serializePathComponents(SymbolGraphASTWalker &Walker,
+                                     llvm::json::OStream &OS) const {
+  OS.attributeArray("pathComponents", [&](){
+    SmallVector<SmallString<32>, 8> PathComponents;
+    Walker.getPathComponents(VD, PathComponents);
+    for (auto Component : PathComponents) {
+      OS.value(Component);
+    }
+  });
 }
 
 void Symbol::serializeNames(SymbolGraphASTWalker &Walker,
                             llvm::json::OStream &OS) const {
   OS.attributeObject("names", [&](){
-    auto Identifier = Walker.getSymbolIdentifier(VD);
-    OS.attribute("title", Identifier.SimpleComponents.back());
+    SmallVector<SmallString<32>, 8> PathComponents;
+    Walker.getPathComponents(VD, PathComponents);
+    
+    OS.attribute("title", PathComponents.back());
     // "navigator": null
     Walker.serializeSubheadingDeclarationFragments("subheading", VD, OS);
     // "prose": null
@@ -370,6 +385,7 @@ void Symbol::serialize(SymbolGraphASTWalker &Walker,
   OS.object([&](){
     serializeKind(OS);
     serializeIdentifier(Walker, OS);
+    serializePathComponents(Walker, OS);
     serializeNames(Walker, OS);
     serializeDocComment(Walker, OS);
 

lib/SymbolGraphGen/Symbol.h

@@ -24,46 +24,6 @@ namespace symbolgraphgen {
 struct AvailabilityDomain;
 struct SymbolGraphASTWalker;
 
-/**
- An identifier for a symbol that provides a globally unique identifier suitable for
- internal lookups and a locally unique path for human use, such as a URL.
- */
-struct SymbolIdentifier {
-  /**
-   A string that uniquely identifies a symbol within a module in the event of
-   ambiguities. A precise identifier need not be human readable.
-   */
-  StringRef PreciseIdentifier;
-  
-  /**
-   The components for a "fully qualified" identifier.
-   */
-  ArrayRef<StringRef> SimpleComponents;
-
-  SymbolIdentifier(llvm::StringRef PreciseIdentifier,
-                   ArrayRef<StringRef> SimpleComponents)
-    : PreciseIdentifier(PreciseIdentifier),
-    SimpleComponents(SimpleComponents) {
-      assert(!PreciseIdentifier.empty());
-    }
-
-  void serialize(llvm::json::OStream &OS) const {
-    OS.object([&](){
-      OS.attribute("precise", PreciseIdentifier);
-      OS.attributeArray("simpleComponents", [&](){
-        for (auto Component : SimpleComponents) {
-          OS.value(Component);
-        }
-      });
-    });
-  }
-  
-  bool operator==(const SymbolIdentifier &Other) const {
-    return PreciseIdentifier == Other.PreciseIdentifier &&
-      SimpleComponents == Other.SimpleComponents;
-  }
-};
-
 /// A symbol from a module: a node in a graph.
 struct Symbol {
   const ValueDecl *VD;
@@ -76,6 +36,9 @@ struct Symbol {
   void serializeIdentifier(SymbolGraphASTWalker &Walker,
                            llvm::json::OStream &OS) const;
 
+  void serializePathComponents(SymbolGraphASTWalker &Walker,
+                               llvm::json::OStream &OS) const;
+
   void serializeNames(SymbolGraphASTWalker &Walker,
                       llvm::json::OStream &OS) const;
 

lib/SymbolGraphGen/SymbolGraphASTWalker.cpp

@@ -149,24 +149,15 @@ StringRef SymbolGraphASTWalker::getUSR(const ValueDecl *VD) {
   return USR;
 }
 
-SymbolIdentifier
-SymbolGraphASTWalker::getSymbolIdentifier(const ValueDecl *VD) {
-  // Look in the symbol identifier cache for this declartion.
-  auto Found = SymbolIdentifierCache.find(VD);
-  if (Found != SymbolIdentifierCache.end()) {
-    return Found->getSecond();
-  }
-
-  // Not found; need to build a symbol identifier and add it to the cache.
-  auto PreciseIdentifier = getUSR(VD);
-  llvm::SmallVector<llvm::StringRef, 4> SimpleIdentifierChain;
-  
+void
+SymbolGraphASTWalker::getPathComponents(const ValueDecl *VD,
+                                        SmallVectorImpl<SmallString<32>> &Components) {
   // Collect the spellings of the fully qualified identifier components.
   auto Decl = VD;
   while (Decl && !isa<ModuleDecl>(Decl)) {
     SmallString<32> Scratch;
     Decl->getFullName().getString(Scratch);
-    SimpleIdentifierChain.push_back(Ctx.allocateCopy(Scratch.str()));
+    Components.push_back(Scratch);
     if (const auto *DC = Decl->getDeclContext()) {
       if (const auto *Proto = DC->getExtendedProtocolDecl()) {
         Decl = Proto;
@@ -179,17 +170,9 @@ SymbolGraphASTWalker::getSymbolIdentifier(const ValueDecl *VD) {
       Decl = nullptr;
     }
   }
-  
-  // The list is leaf-to-root, but our list is root-to-leaf, so reverse it.
-  std::reverse(SimpleIdentifierChain.begin(), SimpleIdentifierChain.end());
 
-  SymbolIdentifier Identifier {
-    PreciseIdentifier,
-    Ctx.allocateCopy(llvm::makeArrayRef(SimpleIdentifierChain))
-  };
-
-  SymbolIdentifierCache.insert({VD, Identifier});
-  return Identifier;
+  // The list is leaf-to-root, but our list is root-to-leaf, so reverse it.
+  std::reverse(Components.begin(), Components.end());
 }
 
 PrintOptions SymbolGraphASTWalker::getDeclarationFragmentsPrintOptions() const {

lib/SymbolGraphGen/SymbolGraphASTWalker.h

@@ -27,7 +27,6 @@ class ValueDecl;
 
 namespace symbolgraphgen {
 
-struct SymbolIdentifier;
 struct SymbolGraph;
 struct SymbolGraphOptions;
 
@@ -49,9 +48,6 @@ struct SymbolGraphASTWalker : public SourceEntityWalker {
   /// A context for allocations.
   markup::MarkupContext Ctx;
 
-  /// A cache of identifiers for declarations that may be seen more than once.
-  llvm::DenseMap<const Decl *, SymbolIdentifier> SymbolIdentifierCache;
-
   /// A cache of USRs for declarations.
   llvm::DenseMap<const ValueDecl *, StringRef> USRCache;
 
@@ -72,8 +68,8 @@ struct SymbolGraphASTWalker : public SourceEntityWalker {
   /// Get the USR of a declaration and add it to the local allocator.
   StringRef getUSR(const ValueDecl *VD);
 
-  /// Returns a `SymbolIdentifier` for a given declaration.
-  SymbolIdentifier getSymbolIdentifier(const ValueDecl *VD);
+  /// Returns an array of path components for a declaration.
+  void getPathComponents(const ValueDecl *VD, SmallVectorImpl<SmallString<32>> &Components);
 
   // MARK: - Declaration Fragments