feat: add 澎湃新闻列表 (#5768)

Co-authored-by: Henry Wang <hi@henry.wang>

docs/traditional-media.md

@@ -463,11 +463,21 @@ category 对应的关键词有
 
 ### 首页头条
 
-<Route author="HenryQW" example="/thepaper/featured" path="/thepaper/featured"/>
+<Route author="HenryQW nczitzk" example="/thepaper/featured" path="/thepaper/featured"/>
 
 ### 频道
 
-<Route author="xyqfer" example="/thepaper/channel/27224" path="/thepaper/channel/:id" :paramsDesc="['频道 id']"/>
+<Route author="xyqfer nczitzk" example="/thepaper/channel/27224" path="/thepaper/channel/:id" :paramsDesc="['频道 id，可在频道页 URL 中找到']">
+
+| 视频  | 时事  | 财经  | 思想  | 澎湃号 | 生活  |
+| ----- | ----- | ----- | ----- | ------ | ----- |
+| 26916 | 25950 | 25951 | 25952 | 36079  | 25953 |
+
+</Route>
+
+### 列表
+
+<Route author="nczitzk" example="/thepaper/list/25457" path="/thepaper/list/:id" :paramsDesc="['列表 id，可在列表页 URL 中找到']"/>
 
 ### 澎湃美数组作品集
 
@@ -478,7 +488,6 @@ category 对应的关键词有
 | 2    | 4    | 3      | 453      |
 
 </Route>
-
 ## 齐鲁晚报
 
 ### 新闻

lib/router.js

@@ -896,6 +896,7 @@ router.get('/embassy/:country/:city?', require('./routes/embassy/index'));
 // 澎湃新闻
 router.get('/thepaper/featured', require('./routes/thepaper/featured'));
 router.get('/thepaper/channel/:id', require('./routes/thepaper/channel'));
+router.get('/thepaper/list/:id', require('./routes/thepaper/list'));
 
 // 澎湃美数课
 router.get('/thepaper/839studio', require('./routes/thepaper/839studio/studio.js'));

lib/routes/thepaper/channel.js

@@ -2,12 +2,14 @@ const utils = require('./utils');
 
 module.exports = async (ctx) => {
     const { id } = ctx.params;
-    const link = `https://m.thepaper.cn/channel_${id}`;
-    const items = await utils.ProcessFeed(link, ctx);
+    const rootUrl = 'https://m.thepaper.cn';
+
+    const link = id === '26916' ? `${rootUrl}/channel_26916` : `${rootUrl}/list_page.jsp?nodeid=${id}&isList=0&pageidx=1`;
+    const items = await utils.ProcessFeed(id === '26916' ? '.list_video_item' : '.list_item_infor', link, ctx);
 
     ctx.state.data = {
         title: `澎湃新闻频道 - ${id}`,
-        link,
+        link: `${rootUrl}/channel_${id}`,
         item: items,
     };
 };

lib/routes/thepaper/featured.js

@@ -1,12 +1,14 @@
 const utils = require('./utils');
 
 module.exports = async (ctx) => {
-    const link = 'https://m.thepaper.cn/';
-    const items = await utils.ProcessFeed(link, ctx);
+    const rootUrl = 'https://m.thepaper.cn';
+
+    const link = `${rootUrl}/list_page.jsp?&nodeid=25949&isList=1&pageidx=1`;
+    const items = await utils.ProcessFeed('.list_item_infor', link, ctx);
 
     ctx.state.data = {
         title: '澎湃新闻 - 首页头条',
-        link,
+        link: `${rootUrl}/channel_25949`,
         item: items,
     };
 };

lib/routes/thepaper/list.js

@@ -0,0 +1,15 @@
+const utils = require('./utils');
+
+module.exports = async (ctx) => {
+    const { id } = ctx.params;
+    const rootUrl = 'https://m.thepaper.cn';
+
+    const link = `${rootUrl}/list_page.jsp?nodeid=${id}&isList=1&pageidx=1`;
+    const items = await utils.ProcessFeed('.list_item_infor', link, ctx);
+
+    ctx.state.data = {
+        title: `澎湃新闻列表 - ${id}`,
+        link: `${rootUrl}/list_${id}`,
+        item: items,
+    };
+};

lib/routes/thepaper/utils.js

@@ -3,17 +3,16 @@ const date = require('@/utils/date');
 const got = require('@/utils/got');
 
 module.exports = {
-    ProcessFeed: async (link, ctx) => {
+    ProcessFeed: async (query, link, ctx) => {
         const res = await got.get(link);
 
         const $ = cheerio.load(res.data);
-        const list = $('.list_item').slice(0, 10).get();
+        const list = $(query).slice(0, 10).get();
 
         return await Promise.all(
             list.map(async (item) => {
                 const $ = cheerio.load(item);
-                const itemUrlID = $(item).find('.list_item_title a').attr('href');
-                const itemUrl = `https://m.thepaper.cn/${itemUrlID}`;
+                const itemUrl = `https://m.thepaper.cn/${$(item).find('a').eq(0).attr('href')}`;
                 const cache = await ctx.cache.get(itemUrl);
                 if (cache) {
                     return Promise.resolve(JSON.parse(cache));
@@ -22,14 +21,25 @@ module.exports = {
                 const res = await got.get(itemUrl);
                 const content = cheerio.load(res.data);
 
-                content('.contheight').replaceWith('<br>');
+                let description, pubDate;
+
+                if (content('div.news_video_msg').length > 0) {
+                    description = content('#vdetail_sum').html();
+                    pubDate = content('div.news_video_msg').html().split('&#xA0;&#xA0;')[0];
+                } else if (content('#slider_wrapper_ul').length > 0) {
+                    description = '';
+                    pubDate = new Date(date($(item).find('div.list_item_extra span').eq(1).text())).toUTCString();
+                } else {
+                    description = content('div.newsdetail_content').html();
+                    pubDate = new Date(content('div.date').text().trim().split('来源：')[0].trim() + ' GMT+8').toUTCString();
+                }
 
                 const single = {
-                    title: content('.news_video_name').text() || content('.title').text(),
+                    title: content('title').text(),
                     link: itemUrl,
-                    description: content('#vdetail_sum_p').html() || content('.news_part_limit div').html(),
-                    pubDate: date(content('.date').text().trim()),
-                    author: content('.author').text(),
+                    description,
+                    pubDate,
+                    author: content('div.author').text(),
                 };
                 return Promise.resolve(single);
             })