Index language as enum instead of string

src/indexer.cc

@@ -996,21 +996,23 @@ void indexDeclaration(CXClientData client_data, const CXIdxDeclInfo* decl) {
     return;
 
 
-  // update the file language
-  // TODO: only do this when |is_first_ownership| in ConsumeFile is true
-  switch (clang_getCursorLanguage(decl->cursor)) {
-  case CXLanguage_C:
-    db->language = "c";
-    break;
-  case CXLanguage_CPlusPlus:
-    db->language = "cpp";
-    break;
-  case CXLanguage_ObjC:
-    db->language = "objectivec";
-    break;
-  case CXLanguage_Invalid:
-    db->language = "invalid";
-    break;
+  // The language of this declaration
+  LanguageId declLang = [decl] () {
+      switch (clang_getCursorLanguage(decl->cursor)) {
+      case CXLanguage_C:
+        return LanguageId::C;
+      case CXLanguage_CPlusPlus:
+        return LanguageId::Cpp;
+      case CXLanguage_ObjC:
+        return LanguageId::ObjC;
+      default:
+        return LanguageId::Unknown;
+      };
+  } ();
+
+  // Only update the file language if the new language is "greater" than the old
+  if (declLang > db->language) {
+    db->language = declLang;
   }
 
   NamespaceHelper* ns = &param->ns;

src/indexer.h

@@ -464,6 +464,18 @@ struct IndexInclude {
   std::string resolved_path;
 };
 
+// Used to identify the language at a file level. The ordering is important, as
+// a file previously identified as `C`, will be changed to `Cpp` if it
+// encounters a c++ declaration.
+enum class LanguageId {
+  Unknown = 0,
+  C = 1,
+  Cpp = 2,
+  ObjC = 3
+};
+
+MAKE_REFLECT_TYPE_PROXY(LanguageId, std::underlying_type<LanguageId>::type);
+
 struct IndexFile {
   IdCache id_cache;
 
@@ -473,9 +485,7 @@ struct IndexFile {
   std::string path;
   std::vector<std::string> args;
   int64_t last_modification_time = 0;
-  // markdown compatible language identifier.
-  // "cpp", "c", "objectivec", or invalid"
-  std::string language;
+  LanguageId language = LanguageId::Unknown;
 
   // The path to the translation unit cc file which caused the creation of this
   // IndexFile. When parsing a translation unit we generate many IndexFile

src/query.cc

@@ -171,10 +171,23 @@ void CompareGroups(std::vector<T>& previous_data,
 QueryFile::Def BuildFileDef(const IdMap& id_map, const IndexFile& indexed) {
   QueryFile::Def def;
   def.path = indexed.path;
-  def.language = indexed.language;
   def.includes = indexed.includes;
   def.inactive_regions = indexed.skipped_by_preprocessor;
 
+  // Convert enum to markdown compatible strings
+  def.language = [indexed] () {
+      switch (indexed.language) {
+      case LanguageId::C:
+        return "c";
+      case LanguageId::Cpp:
+        return "cpp";
+      case LanguageId::ObjC:
+        return "objectivec";
+      default:
+        return "";
+      }
+    } ();
+
   auto add_outline = [&def, &id_map](SymbolIdx idx, Range range) {
     def.outline.push_back(SymbolRef(idx, id_map.ToQuery(range)));
   };