[clang-tools-extra][ExtractAPI] create clang-symbolgraph-merger

Create a clang tool to merge all the JSON symbolgraph emited by --emit-symbol-graph or -extract-api options into one unified JSON symbolgraph file.

Differential Revision: https://reviews.llvm.org/D158646

clang-tools-extra/CMakeLists.txt

@@ -13,6 +13,7 @@ if(CLANG_INCLUDE_TESTS)
   endif()
 endif()
 
+add_subdirectory(clang-symbolgraph-merger)
 add_subdirectory(clang-apply-replacements)
 add_subdirectory(clang-reorder-fields)
 add_subdirectory(modularize)

clang-tools-extra/clang-symbolgraph-merger/CMakeLists.txt

@@ -0,0 +1,3 @@
+include_directories(include)
+add_subdirectory(lib)
+add_subdirectory(tool)

clang-tools-extra/clang-symbolgraph-merger/include/clang-symbolgraph-merger/SymbolGraph.h

@@ -0,0 +1,48 @@
+#ifndef SYMBOLGRAPH_H
+#define SYMBOLGRAPH_H
+
+#include "clang/Basic/LangStandard.h"
+#include "clang/ExtractAPI/API.h"
+#include "clang/ExtractAPI/AvailabilityInfo.h"
+#include "clang/ExtractAPI/DeclarationFragments.h"
+#include "llvm/ADT/DenseMap.h"
+#include "llvm/ADT/StringRef.h"
+#include "llvm/Support/JSON.h"
+#include <memory>
+#include <vector>
+
+namespace sgmerger {
+
+// see https://github.com/apple/swift-docc-symbolkit/bdob/main/openapi.yaml
+struct SymbolGraph {
+
+  struct Symbol {
+    Symbol(const llvm::json::Object &SymbolObj);
+
+    llvm::json::Object SymbolObj;
+    std::string AccessLevel;
+    clang::extractapi::APIRecord::RecordKind Kind;
+    clang::extractapi::DeclarationFragments DeclFragments;
+    clang::extractapi::FunctionSignature FunctionSign;
+    std::string Name;
+    std::string USR;
+    clang::extractapi::AvailabilitySet Availabilities;
+    clang::extractapi::DocComment Comments;
+    clang::extractapi::RecordLocation Location;
+    clang::extractapi::DeclarationFragments SubHeadings;
+
+    // underlying type in case of Typedef
+    clang::extractapi::SymbolReference UnderLyingType;
+  };
+
+  SymbolGraph(const llvm::StringRef JSON);
+  llvm::json::Object SymbolGraphObject;
+  llvm::json::Object Metadata;
+  llvm::json::Object Module;
+  std::vector<Symbol> Symbols;
+  llvm::json::Array Relationships;
+};
+
+} // namespace sgmerger
+
+#endif /* SYMBOLGRAPH_H */

clang-tools-extra/clang-symbolgraph-merger/include/clang-symbolgraph-merger/SymbolGraphMerger.h

@@ -0,0 +1,45 @@
+#ifndef SYMBOLGRAPHMERGER_H
+#define SYMBOLGRAPHMERGER_H
+
+#include "clang-symbolgraph-merger/SymbolGraph.h"
+#include "clang-symbolgraph-merger/SymbolGraphVisitor.h"
+#include "clang/Basic/LangStandard.h"
+#include "clang/ExtractAPI/API.h"
+#include "llvm/ADT/DenseMap.h"
+#include "llvm/ADT/StringRef.h"
+#include "llvm/TargetParser/Triple.h"
+#include <memory>
+
+namespace sgmerger {
+
+using SymbolMap = llvm::DenseMap<llvm::StringRef, const SymbolGraph::Symbol *>;
+
+class SymbolGraphMerger : public SymbolGraphVisitor<SymbolGraphMerger> {
+public:
+  SymbolGraphMerger(const clang::SmallVector<SymbolGraph> &SymbolGraphs,
+                    const std::string &ProductName = "")
+      : ProductName(ProductName), Lang(clang::Language::Unknown),
+        SymbolGraphs(SymbolGraphs) {}
+  bool merge();
+  bool visitMetadata(const llvm::json::Object &Metadata);
+  bool visitModule(const llvm::json::Object &Module);
+  bool visitSymbol(const SymbolGraph::Symbol &Symbol);
+  bool visitRelationship(const llvm::json::Object &Relationship);
+
+private:
+  std::string Generator;
+
+  // stuff required to construct the APISet
+  std::string ProductName;
+  llvm::Triple Target;
+  clang::Language Lang;
+
+  SymbolMap PendingSymbols;
+  SymbolMap VisitedSymbols;
+
+  const clang::SmallVector<SymbolGraph> &SymbolGraphs;
+};
+
+} // namespace sgmerger
+
+#endif /* SYMBOLGRAPHMERGER_H */

clang-tools-extra/clang-symbolgraph-merger/include/clang-symbolgraph-merger/SymbolGraphVisitor.h

@@ -0,0 +1,68 @@
+#ifndef SYMBOLGRAPHVISITOR_H
+#define SYMBOLGRAPHVISITOR_H
+
+#include "clang-symbolgraph-merger/SymbolGraph.h"
+#include "clang/ExtractAPI/API.h"
+#include "llvm/ADT/StringRef.h"
+#include "llvm/Support/JSON.h"
+#include <memory>
+
+namespace sgmerger {
+
+// Visits a symbol graph obbect and record the extracted info to API
+template <typename Derived> class SymbolGraphVisitor {
+public:
+  bool traverseSymbolGraph(const SymbolGraph &SG) {
+    bool Success = true;
+    Success = (getDerived()->visitMetadata(SG.Metadata) &&
+               getDerived()->visitModule(SG.Module) &&
+               getDerived()->traverseSymbols(SG.Symbols) &&
+               getDerived()->traverseRelationships(SG.Relationships));
+
+    return Success;
+  }
+
+  bool traverseSymbols(const std::vector<SymbolGraph::Symbol> &Symbols) {
+    bool Success = true;
+    for (const auto &Symbol : Symbols)
+      Success = getDerived()->visitSymbol(Symbol);
+    return Success;
+  }
+
+  bool traverseRelationships(const llvm::json::Array &Relationships) {
+    bool Success = true;
+    for (const auto &RelValue : Relationships) {
+      if (const auto *RelObj = RelValue.getAsObject())
+        Success = getDerived()->visitRelationship(*RelObj);
+    }
+    return Success;
+  }
+
+  bool visitMetadata(const llvm::json::Object &Metadata);
+  bool visitModule(const llvm::json::Object &Module);
+  bool visitSymbol(const SymbolGraph::Symbol &Symbol);
+  bool visitRelationship(const llvm::json::Object &Relationship);
+
+  std::unique_ptr<clang::extractapi::APISet> getAPISet() {
+    return std::move(API);
+  }
+
+protected:
+  std::unique_ptr<clang::extractapi::APISet> API;
+
+public:
+  SymbolGraphVisitor(const SymbolGraphVisitor &) = delete;
+  SymbolGraphVisitor(SymbolGraphVisitor &&) = delete;
+  SymbolGraphVisitor &operator=(const SymbolGraphVisitor &) = delete;
+  SymbolGraphVisitor &operator=(SymbolGraphVisitor &&) = delete;
+
+protected:
+  SymbolGraphVisitor() : API(nullptr) {}
+  ~SymbolGraphVisitor() = default;
+
+  Derived *getDerived() { return static_cast<Derived *>(this); };
+};
+
+} // namespace sgmerger
+
+#endif /* SYMBOLGRAPHVISITOR_H */

clang-tools-extra/clang-symbolgraph-merger/lib/CMakeLists.txt

@@ -0,0 +1,14 @@
+set(LLVM_LINK_COMPONENTS Support)
+
+add_clang_library(clangSymbolGraphMerger
+  SymbolGraphMerger.cpp
+  SymbolGraph.cpp
+  )
+
+clang_target_link_libraries(clangSymbolGraphMerger
+  PRIVATE
+  clangBasic
+  clangToolingCore
+  clangToolingInclusions
+  clangExtractAPI
+)