Improved shebang detection code

src/source_base.cpp

@@ -38,24 +38,6 @@ Glib::RefPtr<Gsv::StyleSchemeManager> Source::StyleSchemeManager::get_default()
   return instance;
 }
 
-static Glib::RefPtr<Gsv::Language> guess_shebang(const boost::filesystem::path &file_path, Glib::RefPtr<Gsv::LanguageManager> &language_manager) {
-  std::ifstream input(file_path.string(), std::ios::binary);
-  std::string line;
-  if(!std::getline(input, line) || line.empty() || line == "#!")
-    return {};
-
-  if(line.find("perl") < line.size())
-    return language_manager->get_language("perl");
-  else if(line.find("python3") < line.size())
-    return language_manager->get_language("python3");
-  else if(line.find("python") < line.size())
-    return language_manager->get_language("python");
-  else if(line.find("sh") < line.size())
-    return language_manager->get_language("sh");
-
-  return {};
-}
-
 Glib::RefPtr<Gsv::Language> Source::guess_language(const boost::filesystem::path &file_path) {
   auto language_manager = LanguageManager::get_default();
   bool result_uncertain = false;
@@ -96,11 +78,31 @@ Glib::RefPtr<Gsv::Language> Source::guess_language(const boost::filesystem::path
       language = language_manager->get_language("cpphdr");
     else {
       std::ifstream input(file_path.string(), std::ios::binary);
-      std::string tag(5, '\0');
-      if(input && input.read(&tag[0], static_cast<std::streamsize>(tag.size())) && tag == "<?xml")
+      std::string first_chars(2, '\0');
+      if(input && input.read(&first_chars[0], static_cast<std::streamsize>(first_chars.size()))) {
+        if(first_chars == "<?") {
+          std::string next_chars(3, '\0');
+          if(input.read(&next_chars[0], static_cast<std::streamsize>(next_chars.size())) && next_chars == "xml")
             language = language_manager->get_language("xml");
-      else if(input && starts_with(tag, "#!"))
-        language = guess_shebang(file_path, language_manager);
+        }
+        else if(first_chars == "#!") {
+          boost::filesystem::path interpreter;
+          input >> interpreter;
+          auto interpreter_filename = interpreter.filename().string();
+          if(interpreter_filename == "env")
+            input >> interpreter_filename;
+          if(interpreter_filename == "perl")
+            return language_manager->get_language("perl");
+          else if(interpreter_filename == "python3")
+            return language_manager->get_language("python3");
+          else if(interpreter_filename == "python")
+            return language_manager->get_language("python");
+          else if(interpreter_filename == "sh" || interpreter_filename == "bash")
+            return language_manager->get_language("sh");
+          else if(interpreter_filename == "node")
+            return language_manager->get_language("js");
+        }
+      }
     }
   }
   else if(language->get_id() == "cuda") {