对电影天堂增加全文抓取 (#1697)

let list = $('.co_content8 table tr').get();
list.splice(0, 1);
原因如下：
页面内含有两个 .co_content8 table
仅第一个 table内 第一个 tr 元素是广告

const list = $('.co_content8 table tr:not(:first-child)').get();
会丢失第二个table的第一个tr元素

lib/routes/dytt/index.js

@@ -1,6 +1,24 @@
 const axios = require('../../utils/axios');
 const cheerio = require('cheerio');
 const iconv = require('iconv-lite');
+
+async function load(link, ctx) {
+    const cache = await ctx.cache.get(link);
+    if (cache) {
+        return cache;
+    }
+    const response = await axios.get(link, {
+        responseType: 'arraybuffer',
+    });
+    response.data = iconv.decode(response.data, 'gb2312');
+
+    const $ = cheerio.load(response.data);
+
+    const description = $('div#Zoom').html();
+    await ctx.cache.set(link, description, 24 * 60 * 60);
+    return description;
+}
+
 module.exports = async (ctx) => {
     const response = await axios.get('http://www.dytt8.net', {
         responseType: 'arraybuffer',
@@ -9,25 +27,34 @@ module.exports = async (ctx) => {
 
     const $ = cheerio.load(response.data);
     const list = $('.co_content8 table tr').get();
+    // 页面含有2个.co_content8 table
+    // 仅第一个table内第一个tr元素是广告连接
+    // 去除该广告连接
+    list.splice(0, 1);
+    // const list = $('.co_content8 table tr:not(:first-child)').get();
+    const process = await Promise.all(
+        list.map(async (item) => {
+            const link = $(item).find('a:nth-of-type(2)');
+            const itemUrl = 'http://www.dytt8.net' + link.attr('href');
+            const other = await load(itemUrl, ctx);
+            return {
+                title: link.text(),
+                description: other,
+                pubDate: new Date(
+                    $(item)
+                        .find('font')
+                        .text()
+                ).toUTCString(),
+                link: itemUrl,
+            };
+        })
+    );
+
     const data = {
         title: '电影天堂',
         link: 'http://www.dytt8.net',
         description: '电影天堂RSS',
-        item: list
-            .map((item) => {
-                const link = $(item).find('a:nth-of-type(2)');
-                return {
-                    title: link.text(),
-                    description: link.text(),
-                    pubDate: new Date(
-                        $(item)
-                            .find('font')
-                            .text()
-                    ).toUTCString(),
-                    link: 'http://www.dytt8.net' + link.attr('href'),
-                };
-            })
-            .slice(1),
+        item: process,
     };
 
     ctx.state.data = data;