feat: Caixin latest (#6539)

docs/traditional-media.md

@@ -311,6 +311,14 @@ Category 列表：
 
 <Route author="EsuRt"  example="/caixin/article" path="/caixin/article"/>
 
+### 最新文章
+
+<Route author="tpnonthealps" example="/caixin/latest" path="/caixin/latest">
+
+说明：此 RSS feed 会自动抓取财新网的最新文章，但不包含 FM 及视频内容。
+
+</Route>
+
 ## 第一财经
 
 ### 直播区

lib/router.js

@@ -3766,4 +3766,7 @@ router.get('/sciencenet/blog/:type?/:time?/:sort?', require('./routes/sciencenet
 // DailyArt
 router.get('/dailyart/:language?', require('./routes/dailyart/index'));
 
+// Caixin Latest
+router.get('/caixin/latest', require('./routes/caixin/latest'));
+
 module.exports = router;

lib/routes/caixin/latest.js

@@ -0,0 +1,71 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+
+module.exports = async (ctx) => {
+    const li_r = await got({
+        method: 'get',
+        url: 'http://finance.caixin.com/2020-11-03/101622309.html',
+    });
+
+    const $ = cheerio.load(li_r.data);
+    const list = $('div.columnBox a[name="new_artical"]~ul.list li');
+
+    const tasks = [];
+
+    list.map((_index, li) => $(li).find('a').first().attr('href'))
+        .filter((_index, link) => link.indexOf('fm.caixin.com') === -1 && link.indexOf('video.caixin.com') === -1) // content filter
+        .each((_index, link) => {
+            const entry = ctx.cache.tryGet(link, async () => {
+                const entry_r = await got.get(link);
+                const $ = cheerio.load(entry_r.data);
+                // title
+                const h1 = $('div#conTit h1').text();
+                // desc items
+                const subhead = $('div#subhead.subhead').text();
+                let pic_url = $('img.cx-img-loader').attr('src');
+                if (pic_url === undefined) {
+                    pic_url = $('img.cx-img-loader').attr('data-src');
+                }
+                const pic_alt = $('dl.media_pic dd').text();
+                const content = $('div#Main_Content_Val.text').html();
+
+                /*
+                const [, count] =
+                    $('a.box_titlecontent.cost-uibtn')
+                        .text()
+                        .match(/本文共计(\d+)字/) || [];
+                if (count !== 0 && count !== undefined) {
+                    content = `${content}<blockquote><p>此乃财新通收费文章，全文共计${count}字。</p></blockquote>`;
+                }
+                */
+                // desc
+                let desc;
+                if (pic_url === undefined) {
+                    desc = `<blockquote><p>${subhead}</p></blockquote>${content}`;
+                } else {
+                    desc = `<blockquote><p>${subhead}</p></blockquote><img src="${pic_url}" alt="${pic_alt}">${content}`;
+                }
+                // time
+                const [, year, month, day, hour, minute] = $('div#artInfo.artInfo')
+                    .text()
+                    .match(/(\d+)年(\d+)月(\d+)日 *(\d+):(\d+)/);
+                const time = new Date(`${year}-${month}-${day}T${hour}:${minute}:00+0800`).toUTCString();
+
+                return {
+                    title: h1,
+                    description: desc,
+                    pubDate: time,
+                    link: link,
+                };
+            });
+            tasks.push(entry);
+        });
+
+    const rss = await Promise.all(tasks);
+
+    ctx.state.data = {
+        title: '财新网 - 最新文章',
+        link: 'http://www.caixin.com/',
+        item: rss,
+    };
+};