feat(route): wxkol (DIYgod#12010)

* feat(route): wxkol * fix: use async pool
benzking · Mar 2, 2023 · a9d6957 · a9d6957
1 parent 85eab96
commit a9d6957
Show file tree

Hide file tree

Showing 6 changed files with 113 additions and 4 deletions.
diff --git a/docs/new-media.md b/docs/new-media.md
@@ -3940,6 +3940,12 @@ column 为 third 时可选的 category:
 
 <Route author="nczitzk" example="/tam/forecast" path="/tam/forecast"/>
 
+## 微小领
+
+### 微信公众号
+
+<Route author="TonyRL" example="/wxkol/show/3590876722" path="/wxkol/show/:id" :paramsDesc="['公众号 id，可在 URL 找到']" radar="1"/>
+
 ## 微信
 
 ::: tip 提示
@@ -3950,15 +3956,15 @@ column 为 third 时可选的 category:
 
 ### 公众号（CareerEngine 来源）
 
-<Route author="HenryQW" example="/wechat/ce/595a5b14d7164e53908f1606" path="/wechat/ce/:id" :paramsDesc="['公众号 id, 在 [CareerEngine](https://search.careerengine.us/) 搜索公众号，通过 URL 中找到对应的公众号 id']" anticrawler="1"/>
+<Route author="HenryQW" example="/wechat/ce/595a5b14d7164e53908f1606" path="/wechat/ce/:id" :paramsDesc="['公众号 id，在 [CareerEngine](https://search.careerengine.us/) 搜索公众号，通过 URL 中找到对应的公众号 id']" anticrawler="1"/>
 
 ### 公众号（二十次幂来源）
 
-<Route author="sanmmm" example="/wechat/ershicimi/813oxJOl" path="/wechat/ershicimi/:id" :paramsDesc="['公众号id, 打开公众号页, 在 URL 中找到 id']" anticrawler="1"/>
+<Route author="sanmmm" example="/wechat/ershicimi/813oxJOl" path="/wechat/ershicimi/:id" :paramsDesc="['公众号id，打开公众号页，在 URL 中找到 id']" anticrawler="1"/>
 
 ### 公众号（Telegram 频道来源）
 
-<Route author="LogicJake Rongronggg9" example="/wechat/tgchannel/lifeweek" path="/wechat/tgchannel/:id/:mpName?/:searchQueryType?" :paramsDesc="['公众号绑定频道 id', '欲筛选的公众号全名（URL-encoded, 精确匹配），在频道订阅了多个公众号时可选用', '搜索查询类型，见下表']">
+<Route author="LogicJake Rongronggg9" example="/wechat/tgchannel/lifeweek" path="/wechat/tgchannel/:id/:mpName?/:searchQueryType?" :paramsDesc="['公众号绑定频道 id', '欲筛选的公众号全名（URL-encoded，精确匹配），在频道订阅了多个公众号时可选用', '搜索查询类型，见下表']">
 
 | 搜索查询类型 | 将使用的搜索关键字 |            适用于           |
 | :----: | :-------: | :----------------------: |
@@ -3992,6 +3998,10 @@ column 为 third 时可选的 category:
 
 <Route author="TonyRL" example="/wechat/wechat2rss/5b925323244e9737c39285596c53e3a2f4a30774" path="/wechat/wechat2rss/:id" :paramsDesc="['公众号 id，打开 `https://wechat2rss.xlab.app/posts/list/`，在 URL 中找到 id；注意不是公众号页的 id，而是订阅的 id']" radar="1"/>
 
+### 公众号（微小领来源）
+
+见 [#微小领](#wei-xiao-ling)
+
 ### 公众号栏目 (非推送 & 历史消息)
 
 <Route author="MisteryMonster" example="/wechat/mp/homepage/MzA3MDM3NjE5NQ==/16" path="/wechat/mp/homepage/:biz/:hid/:cid?" :paramsDesc="['公众号id', '分页id', '页内栏目']" radar="1" rssbud="1" anticrawler="1">

diff --git a/lib/utils/wechat-mp.js b/lib/utils/wechat-mp.js
@@ -222,7 +222,7 @@ const fetchArticle = (ctx, url, bypassHostCheck = false) => {
 
         let mpName = $('.profile_nickname').first().text();
         mpName = mpName && mpName.trim();
-        return { title, author, description, summary, pubDate, mpName, link: url };
+        return { title, author, description, summary, pubDate, mpName, link: response.url };
     });
 };
 

diff --git a/lib/v2/wxkol/maintainer.js b/lib/v2/wxkol/maintainer.js
@@ -0,0 +1,3 @@
+module.exports = {
+    '/show/:id': ['TonyRL'],
+};
diff --git a/lib/v2/wxkol/radar.js b/lib/v2/wxkol/radar.js
@@ -0,0 +1,13 @@
+module.exports = {
+    'wxkol.com': {
+        _name: '微小领',
+        '.': [
+            {
+                title: '微信公众号',
+                docs: 'https://docs.rsshub.app/new-media.html#wei-xiao-ling',
+                source: ['/show/:id'],
+                target: (params) => `/wxkol/show/${params.id.replace('.html', '')}`,
+            },
+        ],
+    },
+};
diff --git a/lib/v2/wxkol/router.js b/lib/v2/wxkol/router.js
@@ -0,0 +1,3 @@
+module.exports = (router) => {
+    router.get('/show/:id', require('./show'));
+};
diff --git a/lib/v2/wxkol/show.js b/lib/v2/wxkol/show.js
@@ -0,0 +1,80 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { fetchArticle } = require('@/utils/wechat-mp');
+const config = require('@/config').value;
+const asyncPool = require('tiny-async-pool');
+
+module.exports = async (ctx) => {
+    const baseUrl = 'https://www.wxkol.com';
+    const { id } = ctx.params;
+    const url = `${baseUrl}/show/${id}.html`;
+
+    const asyncPoolAll = async (...args) => {
+        const results = [];
+        for await (const result of asyncPool(...args)) {
+            results.push(result);
+        }
+        return results;
+    };
+
+    const feedData = await ctx.cache.tryGet(
+        url,
+        async () => {
+            const { data: response } = await got(url);
+            const $ = cheerio.load(response);
+
+            const list = $('.artlist li')
+                .toArray()
+                .map((item) => {
+                    item = $(item);
+                    const a = item.find('.title a');
+                    return {
+                        title: a.attr('title'),
+                        link: `${baseUrl}${a.attr('href')}`,
+                    };
+                });
+
+            return {
+                feedTitle: $('head title').text(),
+                feedDescription: $('head description').text(),
+                feedImage: $('.main .logo .avatar')
+                    .attr('style')
+                    .match(/url\('(.+)'\)/)[1],
+                feedItem: list,
+            };
+        },
+        config.cache.routeExpire,
+        false
+    );
+
+    const urlList = await asyncPoolAll(2, feedData.feedItem, (item) =>
+        ctx.cache.tryGet(item.link, async () => {
+            const { data: response } = await got(item.link);
+            const $ = cheerio.load(response);
+            item.link = `${baseUrl}${$('.source a').attr('href')}`;
+            return item;
+        })
+    );
+
+    const items = await asyncPoolAll(4, urlList, async (item) => {
+        const { title, author, description, summary, pubDate, mpName, link: itemLink } = await fetchArticle(ctx, item.link, true);
+
+        item.title = title;
+        item.author = author;
+        item.description = description;
+        item.summary = summary;
+        item.pubDate = pubDate;
+        item.author = mpName;
+        item.link = itemLink;
+
+        return item;
+    });
+
+    ctx.state.data = {
+        title: feedData.feedTitle,
+        description: feedData.feedDescription,
+        link: url,
+        image: feedData.feedImage,
+        item: items,
+    };
+};