[Serialization] Split doc serialization out of the Serializer class

Extract the common utilities into a SerializerBase class, and make a
new DocSerializer.

Author: jrose-apple

lib/Serialization/Serialization.cpp

@@ -771,38 +771,35 @@ NormalConformanceID Serializer::addConformanceRef(
 }
 
 /// Record the name of a block.
-static void emitBlockID(llvm::BitstreamWriter &out, unsigned ID,
-                        StringRef name,
-                        SmallVectorImpl<unsigned char> &nameBuffer) {
+void SerializerBase::emitBlockID(unsigned ID, StringRef name,
+                                 SmallVectorImpl<unsigned char> &nameBuffer) {
   SmallVector<unsigned, 1> idBuffer;
   idBuffer.push_back(ID);
-  out.EmitRecord(llvm::bitc::BLOCKINFO_CODE_SETBID, idBuffer);
+  Out.EmitRecord(llvm::bitc::BLOCKINFO_CODE_SETBID, idBuffer);
 
   // Emit the block name if present.
   if (name.empty())
     return;
   nameBuffer.resize(name.size());
   memcpy(nameBuffer.data(), name.data(), name.size());
-  out.EmitRecord(llvm::bitc::BLOCKINFO_CODE_BLOCKNAME, nameBuffer);
+  Out.EmitRecord(llvm::bitc::BLOCKINFO_CODE_BLOCKNAME, nameBuffer);
 }
 
-/// Record the name of a record within a block.
-static void emitRecordID(llvm::BitstreamWriter &out, unsigned ID,
-                         StringRef name,
-                         SmallVectorImpl<unsigned char> &nameBuffer) {
+void SerializerBase::emitRecordID(unsigned ID, StringRef name,
+                                  SmallVectorImpl<unsigned char> &nameBuffer) {
   assert(ID < 256 && "can't fit record ID in next to name");
   nameBuffer.resize(name.size()+1);
   nameBuffer[0] = ID;
   memcpy(nameBuffer.data()+1, name.data(), name.size());
-  out.EmitRecord(llvm::bitc::BLOCKINFO_CODE_SETRECORDNAME, nameBuffer);
+  Out.EmitRecord(llvm::bitc::BLOCKINFO_CODE_SETRECORDNAME, nameBuffer);
 }
 
 void Serializer::writeBlockInfoBlock() {
   BCBlockRAII restoreBlock(Out, llvm::bitc::BLOCKINFO_BLOCK_ID, 2);
 
   SmallVector<unsigned char, 64> nameBuffer;
-#define BLOCK(X) emitBlockID(Out, X ## _ID, #X, nameBuffer)
-#define BLOCK_RECORD(K, X) emitRecordID(Out, K::X, #X, nameBuffer)
+#define BLOCK(X) emitBlockID(X ## _ID, #X, nameBuffer)
+#define BLOCK_RECORD(K, X) emitRecordID(K::X, #X, nameBuffer)
 
   BLOCK(MODULE_BLOCK);
 
@@ -891,7 +888,7 @@ void Serializer::writeBlockInfoBlock() {
   BLOCK_RECORD(sil_block, SIL_TWO_OPERANDS_EXTRA_ATTR);
 
   // These layouts can exist in both decl blocks and sil blocks.
-#define BLOCK_RECORD_WITH_NAMESPACE(K, X) emitRecordID(Out, X, #X, nameBuffer)
+#define BLOCK_RECORD_WITH_NAMESPACE(K, X) emitRecordID(X, #X, nameBuffer)
   BLOCK_RECORD_WITH_NAMESPACE(sil_block,
                               decls_block::INVALID_PROTOCOL_CONFORMANCE);
   BLOCK_RECORD_WITH_NAMESPACE(sil_block,
@@ -932,28 +929,6 @@ void Serializer::writeBlockInfoBlock() {
 #undef BLOCK_RECORD
 }
 
-void Serializer::writeDocBlockInfoBlock() {
-  BCBlockRAII restoreBlock(Out, llvm::bitc::BLOCKINFO_BLOCK_ID, 2);
-
-  SmallVector<unsigned char, 64> nameBuffer;
-#define BLOCK(X) emitBlockID(Out, X ## _ID, #X, nameBuffer)
-#define BLOCK_RECORD(K, X) emitRecordID(Out, K::X, #X, nameBuffer)
-
-  BLOCK(MODULE_DOC_BLOCK);
-
-  BLOCK(CONTROL_BLOCK);
-  BLOCK_RECORD(control_block, METADATA);
-  BLOCK_RECORD(control_block, MODULE_NAME);
-  BLOCK_RECORD(control_block, TARGET);
-
-  BLOCK(COMMENT_BLOCK);
-  BLOCK_RECORD(comment_block, DECL_COMMENTS);
-  BLOCK_RECORD(comment_block, GROUP_NAMES);
-
-#undef BLOCK
-#undef BLOCK_RECORD
-}
-
 void Serializer::writeHeader(const SerializationOptions &options) {
   {
     BCBlockRAII restoreBlock(Out, CONTROL_BLOCK_ID, 3);
@@ -4686,22 +4661,20 @@ void Serializer::writeAST(ModuleOrSourceFile DC,
   }
 }
 
-void Serializer::writeToStream(raw_ostream &os) {
+void SerializerBase::writeToStream(raw_ostream &os) {
   os.write(Buffer.data(), Buffer.size());
   os.flush();
 }
 
-Serializer::Serializer(ArrayRef<unsigned char> signature,
-                       ModuleOrSourceFile DC) {
+SerializerBase::SerializerBase(ArrayRef<unsigned char> signature,
+                               ModuleOrSourceFile DC) {
   for (unsigned char byte : signature)
     Out.Emit(byte, 8);
 
   this->M = getModule(DC);
   this->SF = DC.dyn_cast<SourceFile *>();
 }
 
-Serializer::~Serializer() = default;
-
 void Serializer::writeToStream(raw_ostream &os, ModuleOrSourceFile DC,
                                const SILModule *SILMod,
                                const SerializationOptions &options) {
@@ -4748,8 +4721,7 @@ void swift::serialize(ModuleOrSourceFile DC,
                          options.DocOutputPath,
                          [&](raw_ostream &out) {
       SharedTimer timer("Serialization, swiftdoc");
-      Serializer::writeDocToStream(out, DC, options.GroupInfoPath,
-                                   getContext(DC));
+      writeDocToStream(out, DC, options.GroupInfoPath);
       return false;
     });
   }

lib/Serialization/Serialization.h

@@ -34,7 +34,8 @@ namespace serialization {
 
 using FilenamesTy = ArrayRef<std::string>;
 
-class Serializer {
+class SerializerBase {
+protected:
   SmallVector<char, 0> Buffer;
   llvm::BitstreamWriter Out{Buffer};
 
@@ -50,6 +51,21 @@ class Serializer {
   /// serialized. Any other decls will be cross-referenced instead.
   const SourceFile *SF = nullptr;
 
+  /// Record the name of a block.
+  void emitBlockID(unsigned ID, StringRef name,
+                   SmallVectorImpl<unsigned char> &nameBuffer);
+
+  /// Record the name of a record within a block.
+  void emitRecordID(unsigned ID, StringRef name,
+                    SmallVectorImpl<unsigned char> &nameBuffer);
+
+  void writeToStream(raw_ostream &os);
+
+public:
+  SerializerBase(ArrayRef<unsigned char> signature, ModuleOrSourceFile DC);
+};
+
+class Serializer : public SerializerBase {
 public:
   /// Stores a declaration or a type to be written to the AST file.
   ///
@@ -313,16 +329,10 @@ class Serializer {
   /// Writes the BLOCKINFO block for the serialized module file.
   void writeBlockInfoBlock();
 
-  /// Writes the BLOCKINFO block for the module documentation file.
-  void writeDocBlockInfoBlock();
-
   /// Writes the Swift module file header and name, plus metadata determining
   /// if the module can be loaded.
   void writeHeader(const SerializationOptions &options = {});
 
-  /// Writes the Swift doc module file header and name.
-  void writeDocHeader();
-
   /// Writes the dependencies used to build this module: its imported
   /// modules and its source files.
   void writeInputBlock(const SerializationOptions &options);
@@ -439,21 +449,15 @@ class Serializer {
   void writeAST(ModuleOrSourceFile DC,
                 bool enableNestedTypeLookupTable);
 
-  void writeToStream(raw_ostream &os);
-
-  Serializer(ArrayRef<unsigned char> signature, ModuleOrSourceFile DC);
-  ~Serializer();
+  using SerializerBase::SerializerBase;
+  using SerializerBase::writeToStream;
 
 public:
   /// Serialize a module to the given stream.
   static void writeToStream(raw_ostream &os, ModuleOrSourceFile DC,
                             const SILModule *M,
                             const SerializationOptions &options);
 
-  /// Serialize module documentation to the given stream.
-  static void writeDocToStream(raw_ostream &os, ModuleOrSourceFile DC,
-                               StringRef GroupInfoPath, ASTContext &Ctx);
-
   /// Records the use of the given Type.
   ///
   /// The Type will be scheduled for serialization if necessary.
@@ -562,6 +566,10 @@ class Serializer {
   void writeGenericRequirements(ArrayRef<Requirement> requirements,
                                 const std::array<unsigned, 256> &abbrCodes);
 };
+
+/// Serialize module documentation to the given stream.
+void writeDocToStream(raw_ostream &os, ModuleOrSourceFile DC,
+                      StringRef GroupInfoPath);
 } // end namespace serialization
 } // end namespace swift
 #endif

lib/Serialization/SerializeDoc.cpp

@@ -38,7 +38,9 @@ using pFileNameToGroupNameMap = std::unique_ptr<FileNameToGroupNameMap>;
 namespace {
 class YamlGroupInputParser {
   StringRef RecordPath;
-  std::string Separator = "/";
+  static constexpr const char * const Separator = "/";
+
+  // FIXME: This isn't thread-safe.
   static llvm::StringMap<pFileNameToGroupNameMap> AllMaps;
 
   bool parseRoot(FileNameToGroupNameMap &Map, llvm::yaml::Node *Root,
@@ -136,7 +138,7 @@ llvm::StringMap<pFileNameToGroupNameMap> YamlGroupInputParser::AllMaps;
 
 class DeclGroupNameContext {
   struct GroupNameCollector {
-    const std::string NullGroupName = "";
+    static const StringLiteral NullGroupName;
     const bool Enable;
     GroupNameCollector(bool Enable) : Enable(Enable) {}
     virtual ~GroupNameCollector() = default;
@@ -163,7 +165,7 @@ class DeclGroupNameContext {
       if (!PathOp.hasValue())
         return NullGroupName;
       StringRef FullPath =
-        VD->getASTContext().SourceMgr.getIdentifierForBuffer(PathOp.getValue());
+          Ctx.SourceMgr.getIdentifierForBuffer(PathOp.getValue());
       if (!pMap) {
         YamlGroupInputParser Parser(RecordPath);
         if (!Parser.parse()) {
@@ -209,6 +211,9 @@ class DeclGroupNameContext {
   }
 };
 
+const StringLiteral
+DeclGroupNameContext::GroupNameCollector::NullGroupName = "";
+
 struct DeclCommentTableData {
   StringRef Brief;
   RawComment Raw;
@@ -276,6 +281,42 @@ class DeclCommentTableInfo {
   }
 };
 
+class DocSerializer : public SerializerBase {
+public:
+  using SerializerBase::SerializerBase;
+  using SerializerBase::writeToStream;
+
+  using SerializerBase::Out;
+  using SerializerBase::M;
+  using SerializerBase::SF;
+
+  /// Writes the BLOCKINFO block for the module documentation file.
+  void writeDocBlockInfoBlock() {
+    BCBlockRAII restoreBlock(Out, llvm::bitc::BLOCKINFO_BLOCK_ID, 2);
+
+    SmallVector<unsigned char, 64> nameBuffer;
+#define BLOCK(X) emitBlockID(X ## _ID, #X, nameBuffer)
+#define BLOCK_RECORD(K, X) emitRecordID(K::X, #X, nameBuffer)
+
+    BLOCK(MODULE_DOC_BLOCK);
+
+    BLOCK(CONTROL_BLOCK);
+    BLOCK_RECORD(control_block, METADATA);
+    BLOCK_RECORD(control_block, MODULE_NAME);
+    BLOCK_RECORD(control_block, TARGET);
+
+    BLOCK(COMMENT_BLOCK);
+    BLOCK_RECORD(comment_block, DECL_COMMENTS);
+    BLOCK_RECORD(comment_block, GROUP_NAMES);
+
+#undef BLOCK
+#undef BLOCK_RECORD
+  }
+
+  /// Writes the Swift doc module file header and name.
+  void writeDocHeader();
+};
+
 } // end anonymous namespace
 
 static void writeGroupNames(const comment_block::GroupNamesLayout &GroupNames,
@@ -422,7 +463,7 @@ static void writeDeclCommentTable(
   DeclCommentList.emit(scratch, tableOffset, hashTableBlob);
 }
 
-void Serializer::writeDocHeader() {
+void DocSerializer::writeDocHeader() {
   {
     BCBlockRAII restoreBlock(Out, CONTROL_BLOCK_ID, 3);
     control_block::ModuleNameLayout ModuleName(Out);
@@ -440,9 +481,9 @@ void Serializer::writeDocHeader() {
   }
 }
 
-void Serializer::writeDocToStream(raw_ostream &os, ModuleOrSourceFile DC,
-                                  StringRef GroupInfoPath, ASTContext &Ctx) {
-  Serializer S{MODULE_DOC_SIGNATURE, DC};
+void serialization::writeDocToStream(raw_ostream &os, ModuleOrSourceFile DC,
+                                     StringRef GroupInfoPath) {
+  DocSerializer S{MODULE_DOC_SIGNATURE, DC};
   // FIXME: This is only really needed for debugging. We don't actually use it.
   S.writeDocBlockInfoBlock();
 
@@ -451,7 +492,7 @@ void Serializer::writeDocToStream(raw_ostream &os, ModuleOrSourceFile DC,
     S.writeDocHeader();
     {
       BCBlockRAII restoreBlock(S.Out, COMMENT_BLOCK_ID, 4);
-      DeclGroupNameContext GroupContext(GroupInfoPath, Ctx);
+      DeclGroupNameContext GroupContext(GroupInfoPath, S.M->getASTContext());
       comment_block::DeclCommentListLayout DeclCommentList(S.Out);
       writeDeclCommentTable(DeclCommentList, S.SF, S.M, GroupContext);
       comment_block::GroupNamesLayout GroupNames(S.Out);