chore: optimize performance (replace he with entities) (#6497)

oppilate · web-flow · commit dfdb9630b7e8 · 2020-12-22T18:38:10.000Z
diff --git a/lib/middleware/parameter.js b/lib/middleware/parameter.js
@@ -1,4 +1,4 @@
-const he = require('he');
+const entities = require('entities');
 const mercury_parser = require('@postlight/mercury-parser');
 const cheerio = require('cheerio');
 const OpenCC = require('opencc');
@@ -24,14 +24,14 @@ module.exports = async (ctx, next) => {
         ctx.state.data.item = ctx.state.data.item || [];
 
         // decode HTML entities
-        ctx.state.data.title && (ctx.state.data.title = he.decode(ctx.state.data.title + ''));
-        ctx.state.data.description && (ctx.state.data.description = he.decode(ctx.state.data.description + ''));
+        ctx.state.data.title && (ctx.state.data.title = entities.decodeXML(ctx.state.data.title + ''));
+        ctx.state.data.description && (ctx.state.data.description = entities.decodeXML(ctx.state.data.description + ''));
 
         // sort items
         ctx.state.data.item = ctx.state.data.item.sort((a, b) => +new Date(b.pubDate || 0) - +new Date(a.pubDate || 0));
 
-        ctx.state.data.item.forEach((item) => {
-            item.title && (item.title = he.decode(item.title + ''));
+        const handleItem = async (item) => {
+            item.title && (item.title = entities.decodeXML(item.title + ''));
 
             // handle pubDate
             if (item.pubDate) {
@@ -111,9 +111,12 @@ module.exports = async (ctx, next) => {
                         $ele.removeAttr(e);
                     });
                 });
-                item.description = he.decode($('body').html() + '') + (config.suffix || '');
+                item.description = entities.decodeXML($('body').html() + '') + (config.suffix || '');
             }
-        });
+            return item;
+        };
+
+        ctx.state.data.item = await Promise.all(ctx.state.data.item.map(handleItem));
 
         if (ctx.query) {
             // limit
diff --git a/lib/routes/aozora/newbook.js b/lib/routes/aozora/newbook.js
@@ -5,7 +5,7 @@
 
 const got = require('@/utils/got'); // get web content
 const cheerio = require('cheerio'); // html parser
-const he = require('he');
+const entities = require('entities');
 
 const base_url = 'https://www.aozora.gr.jp/index_pages/';
 module.exports = async (ctx) => {
@@ -50,7 +50,7 @@ module.exports = async (ctx) => {
         let title = '';
         let title_sub = '';
         for (let j = 0; j < title_info.length; ++j) {
-            const tmp = he.decode($(title_info[j]).html()); // should convert from escaped to unicode
+            const tmp = entities.decodeXML($(title_info[j]).html()); // should convert from escaped to unicode
             if (tmp.includes('作品名：')) {
                 title = $(title_info[j]).find('td:nth-child(2)').text();
             }
diff --git a/lib/routes/patchwork.kernel.org/comments.js b/lib/routes/patchwork.kernel.org/comments.js
@@ -1,6 +1,6 @@
 const got = require('@/utils/got');
 const cache = require('./cache');
-const he = require('he');
+const entities = require('entities');
 const queryString = require('query-string');
 
 module.exports = async (ctx) => {
@@ -25,7 +25,7 @@ module.exports = async (ctx) => {
         link: host,
         item: data.map((item) => ({
             title: item.subject,
-            description: he.escape(he.escape(item.content)).replace(/\n/g, '<br>'),
+            description: entities.escape(entities.escape(item.content)).replace(/\n/g, '<br>'),
             pubDate: new Date(item.date).toUTCString(),
             link: item.web_url,
         })),
diff --git a/lib/routes/tencent/wechat/wemp.js b/lib/routes/tencent/wechat/wemp.js
@@ -1,5 +1,5 @@
 const got = require('@/utils/got');
-const he = require('he');
+const entities = require('entities');
 const cheerio = require('cheerio');
 const date = require('@/utils/date');
 
@@ -34,7 +34,7 @@ module.exports = async (ctx) => {
                 const matchs = /"(url|temp_url)":"([^"]+mp\.weixin\.qq\.com[^"]+)"/.exec(response);
                 let weixinLink = '';
                 if (matchs && matchs[2]) {
-                    weixinLink = he.unescape(unescape(matchs[2].replace(/\\u/g, '%u')));
+                    weixinLink = entities.decodeXML(unescape(matchs[2].replace(/\\u/g, '%u')));
                 }
 
                 const single = {
diff --git a/package.json b/package.json
@@ -2,13 +2,14 @@
   "name": "rsshub",
   "version": "1.0.0",
   "description": "Make RSS Great Again!",
-  "main": "lib/pkg.js",
+  "main": "lib/workers_index.js",
   "files": [
     "lib"
   ],
   "scripts": {
     "start": "node lib/index.js",
     "dev": "cross-env NODE_ENV=dev nodemon --inspect lib/index.js",
+    "profiling": "NODE_ENV=production node --prof lib/index.js",
     "docs:dev": "vuepress dev docs",
     "docs:build": "vuepress build docs",
     "format": "eslint \"**/*.js\" --fix && node docs/.format/format.js && prettier \"**/*.{js,json}\" --write",
@@ -41,7 +42,6 @@
     "@vuepress/plugin-google-analytics": "1.7.1",
     "@vuepress/plugin-pwa": "1.7.1",
     "cross-env": "7.0.3",
-    "entities": "2.1.0",
     "eslint": "7.16.0",
     "eslint-config-prettier": "7.1.0",
     "eslint-plugin-prettier": "3.3.0",
@@ -78,12 +78,12 @@
     "dayjs": "1.9.7",
     "dotenv": "8.2.0",
     "emailjs-imap-client": "3.1.0",
+    "entities": "2.1.0",
     "etag": "1.8.1",
     "fanfou-sdk": "4.2.0",
     "git-rev-sync": "3.0.1",
     "googleapis": "66.0.0",
     "got": "11.8.1",
-    "he": "1.2.0",
     "https-proxy-agent": "5.0.0",
     "iconv-lite": "0.6.2",
     "instagram-private-api": "1.43.3",
