feat: 添加微信公众号文章话题 Tag (#5639)

assets/radar-rules.js

@@ -1829,6 +1829,12 @@
                 source: '/mp/homepage',
                 target: (params, url) => `/wechat/mp/homepage/${new URL(url).searchParams.get('__biz')}/${new URL(url).searchParams.get('hid')}/${new URL(url).searchParams.get('cid') ? new URL(url).searchParams.get('cid') : ''}`,
             },
+            {
+                title: '微信公众号话题',
+                docs: 'https://docs.rsshub.app/new-media.html#wei-xin-gong-zhong-hao-wen-zhang-hua-ti-tag',
+                source: '/mp/appmsgalbum',
+                target: (params, url) => `/wechat/mp/msgalbum/${new URL(url).searchParams.get('__biz')}/${new URL(url).searchParams.get('album_id')}`,
+            },
         ],
         egame: [
             {

docs/new-media.md

@@ -1271,6 +1271,14 @@ area 分区选项
 
 </Route>
 
+### 公众号文章话题 Tag
+
+<Route author="MisteryMonster" example="/wechat/mp/msgalbum/MzA3MDM3NjE5NQ==/1375870284640911361" path="/wechat/mp/msgalbum/:biz/:aid" :paramsDesc="['公众号id', 'Tag id', ]" radar="1" anticrawler="1">
+
+一些公众号（如看理想）会在微信文章里添加 Tag ，点入 Tag 的链接如 <https://mp.weixin.qq.com/mp/appmsgalbum?__biz=MzA3MDM3NjE5NQ==&action=getalbum&album_id=1375870284640911361>，其中`biz` 为 `MzA3MDM3NjE5NQ==`，`aid` 为 `1375870284640911361`。
+
+</Route>
+
 ### 公众平台系统公告栏目
 
 <Route author="xyqfer" example="/wechat/announce" path="/wechat/announce" />

lib/router.js

@@ -502,6 +502,7 @@ router.get('/wechat/uread/:userid', require('./routes/tencent/wechat/uread'));
 router.get('/wechat/ershicimi/:id', require('./routes/tencent/wechat/ershcimi'));
 router.get('/wechat/wjdn/:id', require('./routes/tencent/wechat/wjdn'));
 router.get('/wechat/mp/homepage/:biz/:hid/:cid?', require('./routes/tencent/wechat/mp'));
+router.get('/wechat/mp/msgalbum/:biz/:aid', require('./routes/tencent/wechat/msgalbum'));
 
 // All the Flight Deals
 router.get('/atfd/:locations/:nearby?', require('./routes/atfd/index'));

lib/routes/tencent/wechat/msgalbum.js

@@ -0,0 +1,59 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const dayjs = require('dayjs');
+module.exports = async (ctx) => {
+    const { biz, aid } = ctx.params;
+    const aidurl = `&album_id=${aid}`;
+
+    const HTMLresponse = await got({
+        method: 'get',
+        url: `https://mp.weixin.qq.com/mp/appmsgalbum?__biz=${biz}&action=getalbum${aidurl}`,
+    });
+    const $ = cheerio.load(HTMLresponse.data);
+    const list = $('li').get();
+    const mptitle = $('.album__author-name').text() + `|` + $('.album__label-title').text();
+    const articledata = await Promise.all(
+        list.map(async (item) => {
+            const link = $(item).attr('data-link').replace('http://', 'https://');
+            const title = $(item).attr('data-title');
+            const cache = await ctx.cache.get(link);
+            if (cache) {
+                return Promise.resolve(JSON.parse(cache));
+            }
+            const response2 = await got({
+                method: 'get',
+                url: link,
+            });
+            const articleHtml = response2.data;
+            const $2 = cheerio.load(articleHtml);
+            $2('img').removeAttr('src');
+            $2('div#js_profile_qrcode').remove();
+
+            const content = $2('div#js_content.rich_media_content')
+                .html()
+                .replace('iframe/preview.html?width=500&height=375&', 'txp/iframe/player.html?')
+                .replace('<iframe ', '<iframe width="640" height="360"')
+                .replace(/data-src/g, 'src');
+            const author = $2('div#meta_content:not(:last-child)').text();
+            const single = {
+                content,
+                author,
+                link,
+                title,
+            };
+            ctx.cache.set(link, JSON.stringify(single));
+            return Promise.resolve(single);
+        })
+    );
+    ctx.state.data = {
+        title: `${mptitle}`,
+        link: `https://mp.weixin.qq.com/mp/appmsgalbum?__biz=${biz}&action=getalbum${aidurl}`,
+        item: list.map((item, index) => ({
+            title: `${articledata[index].title}`,
+            description: $(item).find('.album__item-img').html() + `<br><br>${articledata[index].content}`,
+            link: `${articledata[index].link}`,
+            author: `${articledata[index].author}`,
+            pubDate: dayjs.unix($(item).find('.js_article_create_time').text()).format(),
+        })),
+    };
+};