From ec80154f2ff0bda74cda0f3e72793cb4a9423eb8 Mon Sep 17 00:00:00 2001
From: sgqy <sgqy@users.noreply.github.com>
Date: Sat, 3 Nov 2018 10:57:31 +0800
Subject: [PATCH] fix [solidot] to show content image (#1053)

---
 routes/solidot/_article.js | 46 +++++++++++++++++++++++++++++
 routes/solidot/main.js     | 59 ++++++--------------------------------
 2 files changed, 54 insertions(+), 51 deletions(-)
 create mode 100644 routes/solidot/_article.js

diff --git a/routes/solidot/_article.js b/routes/solidot/_article.js
new file mode 100644
index 000000000..9759063cf
--- /dev/null
+++ b/routes/solidot/_article.js
@@ -0,0 +1,46 @@
+const axios = require('../../utils/axios'); // get web content
+const cheerio = require('cheerio'); // html parser
+
+const domain = 'https://www.solidot.org';
+
+module.exports = async function get_article(url) {
+    if (/^\/.*$/.test(url)) {
+        url = domain + url;
+    }
+    const response = await axios({
+        method: 'get',
+        url: url,
+        'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:62.0) Gecko/20100101 Firefox/62.0',
+    });
+    const data = response.data;
+    const $ = cheerio.load(data);
+
+    const date_raw = $('div.talk_time')
+        .clone()
+        .children()
+        .remove()
+        .end()
+        .text();
+    const date_str_zh = date_raw.replace(/^[^`]*发表于(.*分)[^`]*$/g, '$1'); // use [^`] to match \n
+    const date_str = date_str_zh
+        .replace(/[年月]/g, '-')
+        .replace(/时/g, ':')
+        .replace(/[日分]/g, '');
+    let date = new Date(date_str);
+    date.setHours(date.getHours() - 8);
+    date = new Date(Date.UTC(date.getFullYear(), date.getMonth(), date.getDate(), date.getHours(), date.getMinutes(), date.getSeconds()));
+
+    const item = {
+        title: $('div.bg_htit > h2').text(),
+        pubDate: date.toUTCString(),
+        author: $('div.talk_time > b')
+            .text()
+            .replace(/^来自(.*)部门$/g, '$1'),
+        link: url,
+        description: $('div.p_mainnew')
+            .html()
+            .replace(/href="\//g, 'href="' + domain + '/'),
+        category: $('div.icon_float > a').attr('title'),
+    };
+    return item;
+};
diff --git a/routes/solidot/main.js b/routes/solidot/main.js
index 298c89035..039f10818 100644
--- a/routes/solidot/main.js
+++ b/routes/solidot/main.js
@@ -5,6 +5,7 @@
 
 const axios = require('../../utils/axios'); // get web content
 const cheerio = require('cheerio'); // html parser
+const get_article = require('./_article');
 
 const base_url = 'https://$type$.solidot.org';
 module.exports = async (ctx) => {
@@ -19,60 +20,16 @@ module.exports = async (ctx) => {
     const data = response.data; // content is html format
     const $ = cheerio.load(data);
 
-    // get content
-    const a = $('div.block_m');
-
-    const msg_list = [];
+    // get urls
+    const a = $('div.block_m').find('div.bg_htit > h2 > a');
+    const urls = [];
     for (let i = 0; i < a.length; ++i) {
-        const title = $(a[i])
-            .find('div.bg_htit > h2 > a')
-            .text();
-        const link =
-            list_url +
-            $(a[i])
-                .find('div.bg_htit > h2 > a')
-                .attr('href');
-        const content = $(a[i])
-            .find('div.p_mainnew')
-            .html()
-            .replace(/href="\//g, 'href="' + list_url);
-        const author = $(a[i])
-            .find('div.talk_time > b')
-            .text()
-            .replace(/^来自/, '');
-        const category = $(a[i])
-            .find('div.icon_float > a')
-            .attr('title');
-        const cat_img = $(a[i])
-            .find('div.icon_float > a')
-            .html();
-        const date_raw = $(a[i])
-            .find('div.talk_time')
-            .clone()
-            .children()
-            .remove()
-            .end()
-            .text();
-        const date_str_zh = date_raw.replace(/^[^`]*发表于(.*分)[^`]*$/g, '$1'); // use [^`] to match \n
-        const date_str = date_str_zh
-            .replace(/[年月]/g, '-')
-            .replace(/时/g, ':')
-            .replace(/[日分]/g, '');
-        let date = new Date(date_str);
-        date.setHours(date.getHours() - 8);
-        date = new Date(Date.UTC(date.getFullYear(), date.getMonth(), date.getDate(), date.getHours(), date.getMinutes(), date.getSeconds()));
-        const item = {
-            title: title,
-            author: author,
-            pubDate: date.toUTCString(),
-            category: category, // not supported: 2018-10-29
-            link: link,
-            description: content + '<br/><br/>' + cat_img,
-        };
-
-        msg_list.push(item);
+        urls.push($(a[i]).attr('href'));
     }
 
+    // get articles
+    const msg_list = await Promise.all(urls.map((u) => get_article(u)));
+
     // feed the data
     ctx.state.data = {
         title: '奇客的资讯，重要的东西',