From e9f9d0e560f7b8fde091bb8e9b6b860d95146ff6 Mon Sep 17 00:00:00 2001
From: xaoxuu <git@xaoxuu.com>
Date: Tue, 29 Nov 2022 21:54:15 +0800
Subject: [PATCH] [opt] search

---
 _config.yml                        |   7 +-
 layout/_partial/sidebar/index.ejs  |   2 +-
 layout/_partial/widgets/search.ejs |  18 ++-
 scripts/generators/search.js       | 180 +++++++++++++++--------------
 source/js/search/local-search.js   |  25 +---
 5 files changed, 115 insertions(+), 117 deletions(-)

diff --git a/_config.yml b/_config.yml
index 6b4b047..9ce251e 100755
--- a/_config.yml
+++ b/_config.yml
@@ -63,10 +63,11 @@ article:
 
 search:
   service: local_search # local_search, todo...
-  local_search:
+  local_search: # 在 front-matter 中设置 indexing:false 来避免被搜索索引
     field: all # post, page, all
-    path: /search.json
-    content: true
+    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/layout/_partial/sidebar/index.ejs b/layout/_partial/sidebar/index.ejs
index 1fd13b2..cbfd48c 100755
--- a/layout/_partial/sidebar/index.ejs
+++ b/layout/_partial/sidebar/index.ejs
@@ -83,7 +83,7 @@ function layoutWidgets() {
       if (name in theme.data.widgets) {
         Object.assign(widget, theme.data.widgets[name])
       }
-      if (typeof w == 'object' && w.override) {
+      if (typeof w == 'object' && (w.override || w.layout)) {
         Object.assign(widget, w)
       }
       if (widget && widget.layout) {
diff --git a/layout/_partial/widgets/search.ejs b/layout/_partial/widgets/search.ejs
index 438f5ef..3fbfce4 100644
--- a/layout/_partial/widgets/search.ejs
+++ b/layout/_partial/widgets/search.ejs
@@ -5,15 +5,23 @@ function layoutDiv() {
   el += '<div class="search-wrapper" id="search">'
   el += '<form class="search-form">'
   var filter = ''
-  if (item.filter == 'auto' && page.layout == 'wiki') {
-    let matches = page.path.match(/(.*?)\/(.*?)\//i)
-    if (matches?.length > 0) {
-      filter = matches[0]
+  if (item.filter == 'auto') {
+    if (page.layout == 'wiki') {
+      let matches = page.path.match(/(.*?)\/(.*?)\//i)
+      if (matches?.length > 0) {
+        filter = matches[0]
+      }
+    }
+  } else {
+    if (item.filter?.length > 0) {
+      filter = item.filter
     }
   }
   el += '<input type="text" class="search-input" id="search-input"'
   if (filter.length > 0) {
-    filter = '/' + filter
+    if (!filter.startsWith('/')) {
+      filter = '/' + filter
+    }
     el += ' data-filter="' + filter + '"'
     el += ' placeholder="' + (item.placeholder || __('search.search_in', filter)) + '">'
   } else {
diff --git a/scripts/generators/search.js b/scripts/generators/search.js
index 12e682e..b578483 100644
--- a/scripts/generators/search.js
+++ b/scripts/generators/search.js
@@ -3,101 +3,105 @@
  */
 
 hexo.extend.generator.register('search_json_generator', function (locals) {
-  if (this.theme.config.search.service != 'local_search') {
-    return {}
-  }
-  var config = this.config
-  const { local_search } = this.theme.config.search
-
-  var searchfield = local_search.field
-  var content = local_search.content
+  if (this.theme.config.search.service != 'local_search') { return {} }
+  const { root } = this.config
+  const { local_search: cfg } = this.theme.config.search
+  cfg.sort = '-date'
+  cfg.field = cfg.field?.trim()
 
   var posts, pages
-
-  if (searchfield.trim() != '') {
-      searchfield = searchfield.trim()
-      if (searchfield == 'post'){
-          posts = locals.posts.sort('-date')
-      } else if (searchfield == 'page') {
-          pages = locals.pages
-      } else {
-          posts = locals.posts.sort('-date')
-          pages = locals.pages
-      }
+  if (cfg.field == 'post') {
+    posts = locals.posts?.filter(p => p.content?.length > 0).sort(cfg.sort)
+  } else if (cfg.field == 'page') {
+    pages = locals.pages?.filter(p => p.content?.length > 0)
   } else {
-      posts = locals.posts.sort('-date')
+    posts = locals.posts?.filter(p => p.content?.length > 0).sort(cfg.sort)
+    pages = locals.pages?.filter(p => p.content?.length > 0)
   }
 
-  var res = new Array() 
-  var index = 0
-  
-  if (posts) {     
-      posts.each(function(post) {
-          if (post.indexing != undefined && !post.indexing) return
-          var temp_post = new Object() 
-          if (post.title) { 
-              temp_post.title = post.title 
-          }
-          if (post.path) { 
-              temp_post.url = config.root + post.path 
-          } 
-          if (content != false && post._content) { 
-              temp_post.content = post._content 
-          } 
-          if (post.tags && post.tags.length > 0) { 
-              var tags = []
-              post.tags.forEach(function (tag) {
-                  tags.push(tag.name)
-              }) 
-              temp_post.tags = tags 
-          } 
-          if (post.categories && post.categories.length > 0) { 
-              var categories = []
-              post.categories.forEach(function (cate) {
-                  categories.push(cate.name)
-              }) 
-              temp_post.categories = categories 
-          } 
-          res[index] = temp_post  
-          index += 1 
+  var res = new Array()
+
+  function generateJson(post) {
+    var temp_post = new Object()
+    if (post.title) {
+      temp_post.title = post.title.trim()
+    }
+    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, '')
+        // 部分HTML标签
+        content = content.replace(/<iframe[\s|\S]+iframe>/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, '')
+        }
+        // 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, '')
+      }
+      temp_post.content = content.trim()
+    }
+    if (post.tags && post.tags.length > 0) {
+      var tags = []
+      post.tags.forEach(function (tag) {
+        tags.push(tag.name)
+      })
+      temp_post.tags = tags
+    }
+    if (post.categories && post.categories.length > 0) {
+      var categories = []
+      post.categories.forEach(function (cate) {
+        categories.push(cate.name)
+      })
+      temp_post.categories = categories
+    }
+    return temp_post
+  }
+
+  if (posts) {
+    posts.each(function(post) {
+      if (post.indexing == false) return
+      let temp_post = generateJson(post)
+      res.push(temp_post)
     }) 
   } 
-  if (pages) { 
-      pages.each(function(page) {
-          if (page.indexing != undefined && !page.indexing) return
-          var temp_page = new Object() 
-          if (page.title) { 
-              temp_page.title = page.title 
-          } 
-          if (page.path) { 
-              temp_page.url = config.root + page.path 
-          } 
-          if (content != false && page._content) { 
-              temp_page.content = page._content 
-          } 
-          if (page.tags && page.tags.length > 0) { 
-              var tags = new Array() 
-              var tag_index = 0 
-              page.tags.each(function (tag) {
-                  tags[tag_index] = tag.name 
-              }) 
-              temp_page.tags = tags 
-          } 
-          if (page.categories && page.categories.length > 0) {
-              temp_page.categories = []
-              (page.categories.each || page.categories.forEach)(function (item) {
-                  temp_page.categories.push(item)
-              })
-          } 
-          res[index] = temp_page  
-          index += 1 
-      }) 
-  } 
-
-  var json = JSON.stringify(res)
-
+  if (pages) {
+    pages.each(function(page) {
+      if (page.indexing == false) return
+      let temp_post = generateJson(page)
+      res.push(temp_post)
+    })
+  }
   return {
-      path: local_search.path,
-      data: json
+    path: cfg.path,
+    data: JSON.stringify(res)
   }
 })
\ No newline at end of file
diff --git a/source/js/search/local-search.js b/source/js/search/local-search.js
index 89c38da..91739c6 100644
--- a/source/js/search/local-search.js
+++ b/source/js/search/local-search.js
@@ -24,20 +24,6 @@
 /*exported searchFunc*/
 var searchFunc = function(path, filter, searchId, contentId) {
 
-  function stripHtml(html) {
-    html = html.replace(/<style([\s\S]*?)<\/style>/gi, "");
-    html = html.replace(/<script([\s\S]*?)<\/script>/gi, "");
-    html = html.replace(/<figure([\s\S]*?)<\/figure>/gi, "");
-    html = html.replace(/<\/div>/ig, "\n");
-    html = html.replace(/<\/li>/ig, "\n");
-    html = html.replace(/<li>/ig, "  *  ");
-    html = html.replace(/<\/ul>/ig, "\n");
-    html = html.replace(/<\/p>/ig, "\n");
-    html = html.replace(/<br\s*[\/]?>/gi, "\n");
-    html = html.replace(/<[^>]+>/ig, "");
-    return html;
-  }
-
   function getAllCombinations(keywords) {
     var i, j, result = [];
 
@@ -68,18 +54,17 @@ 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 (!data.title || data.title.trim() === "") {
-            return;
-          }
-          if (filter && !data.url.includes(filter)) {
+          if (filter && !data.path.includes(filter)) {
             return;
           }
           var dataTitle = data.title.trim().toLowerCase();
           var dataTitleLowerCase = dataTitle.toLowerCase();
-          var dataContent = stripHtml(data.content.trim());
+          var dataContent = data.content;
           var dataContentLowerCase = dataContent.toLowerCase();
-          var dataUrl = data.url;
+          var dataUrl = data.path;
           var indexTitle = -1;
           var indexContent = -1;
           var firstOccur = -1;