From 5fc71e2f862d7e82f18a3f37709f29865d491923 Mon Sep 17 00:00:00 2001
From: xaoxuu <git@xaoxuu.com>
Date: Mon, 19 Dec 2022 22:32:08 +0800
Subject: [PATCH] [opt] search content

---
 _config.yml                      |  1 -
 scripts/generators/search.js     | 42 ++++++--------------------------
 source/js/search/local-search.js |  7 ++----
 3 files changed, 9 insertions(+), 41 deletions(-)

diff --git a/_config.yml b/_config.yml
index 8dbb59b..9aaa6ec 100755
--- a/_config.yml
+++ b/_config.yml
@@ -79,7 +79,6 @@ search:
     field: all # post, page, all
     path: /search.json # 搜索文件存放位置
     content: true # 是否搜索内容
-    codeblock: true # 是否搜索代码块（需要content: true)
   
 
 ######## Comments ########
diff --git a/scripts/generators/search.js b/scripts/generators/search.js
index 6b8f007..ea6cb15 100644
--- a/scripts/generators/search.js
+++ b/scripts/generators/search.js
@@ -1,6 +1,7 @@
 /**
  * https://github.com/wzpan/hexo-generator-search
  */
+const { stripHTML } = require('hexo-util')
 
 hexo.extend.generator.register('search_json_generator', function (locals) {
   if (this.theme.config.search.service != 'local_search') { return {} }
@@ -29,46 +30,17 @@ hexo.extend.generator.register('search_json_generator', function (locals) {
     if (post.path) {
       temp_post.path = root + post.path
     }
-    if (cfg.content != false && post._content) {
-      var content = post._content.trim()
-      // 过滤掉标签和注释
-      if (content.includes('{%')) {
-        // 需要保留内容的的标签
-        content = content.replace(/{%\s*mark\s*(.*?)\s*%}/g, '$1')
-        content = content.replace(/{%\s*folding\s*(.*?)\s*%}/g, '$1')
-        content = content.replace(/{%\s*copy\s*(.*?)\s*%}/g, '$1')
-        content = content.replace(/{%\s*note\s*(.*?)\s*%}/g, '$1')
-        content = content.replace(/{%\s*kbd\s*(.*?)\s*%}/g, '$1')
-        content = content.replace(/{%\s*emp\s*(.*?)\s*%}/g, '$1')
-        content = content.replace(/{%\s*wavy\s*(.*?)\s*%}/g, '$1')
-        content = content.replace(/{%\s*sub\s*(.*?)\s*%}/g, '$1')
-        content = content.replace(/{%\s*sup\s*(.*?)\s*%}/g, '$1')
-        content = content.replace(/<!--\s*folder(.*?)\s*-->/g, '$1')
-        content = content.replace(/<!--\s*tab(.*?)\s*-->/g, '$1')
-        content = content.replace(/<!--\s*node(.*?)\s*-->/g, '$1')
-        // 不保留内容的标签
-        content = content.replace(/{%\s*(.*?)\s*%}/g, '')
-      }
-      // 注释
-      content = content.replace(/<!--\s*(.*?)\s*-->/g, '')
-      // ## 标题
-      content = content.replace(/[#]{2,} /g, '')
+    if (cfg.content != false && post.content) {
+      var content = stripHTML(post.content).trim()
       // 部分HTML标签
       content = content.replace(/<iframe[\s|\S]+iframe>/g, '')
       content = content.replace(/<hr>/g, '')
       content = content.replace(/<br>/g, '')
-      // 图片
-      content = content.replace(/\!\[(.*?)\]\((.*?)\)/g, '')
-      // 链接
-      content = content.replace(/\[(.*?)\]\((.*?)\)/g, '$1')
-      // 过滤代码块
-      if (cfg.codeblock == false) {
-        content = content.replace(/```([^`]+)```/g, '')
-      }
+      // 换行符换成空格
+      content = content.replace(/\\n/g, ' ')
+      content = content.replace(/\n/g, ' ')
       // 多个连续空格换成单个空格
-      content = content.replace(/[\s]{2,} /g, ' ')
-      // 特殊字符
-      content = content.replace(/[\r|\n]+/g, '')
+      content = content.replace(/[\s]{2,}/g, ' ')
       temp_post.content = content.trim()
     }
     if (post.tags && post.tags.length > 0) {
diff --git a/source/js/search/local-search.js b/source/js/search/local-search.js
index 246c0f6..694d8e2 100644
--- a/source/js/search/local-search.js
+++ b/source/js/search/local-search.js
@@ -54,13 +54,10 @@ var searchFunc = function(path, filter, searchId, contentId) {
         }
         // perform local searching
         datas.forEach(function(data) {
-          if (!data.title?.trim().length) { return }
           if (!data.content?.trim().length) { return }
           var matches = 0;
-          if (filter && !data.path.includes(filter)) {
-            return;
-          }
-          var dataTitle = data.title.trim();
+          if (filter && !data.path.includes(filter)) { return }
+          var dataTitle = data.title?.trim() || 'Untitled';
           var dataTitleLowerCase = dataTitle.toLowerCase();
           var dataContent = data.content;
           var dataContentLowerCase = dataContent.toLowerCase();