fix(route): google news (#11662)

docs/en/new-media.md

@@ -340,7 +340,7 @@ Country
 
 ### News
 
-<RouteEn author="zoenglinghou" example="/google/news/Headlines/hl=en-US&gl=US&ceid=US:en" path="/google/news/:category/:locale" :paramsDesc="['Category Title', 'locales, could be found behind `?`, including `hl`, `gl`, and `ceid` as parameters']"/>
+<RouteEn author="zoenglinghou" example="/google/news/Top stories/hl=en-US&gl=US&ceid=US:en" path="/google/news/:category/:locale" :paramsDesc="['Category Title', 'locales, could be found behind `?`, including `hl`, `gl`, and `ceid` as parameters']"/>
 
 ## Grub Street
 

docs/new-media.md

@@ -2260,7 +2260,7 @@ Type 栏目：
 
 ### 新闻
 
-<Route author="zoenglinghou" example="/google/news/要闻/hl=zh-CN&gl=CN&ceid=CN:zh-Hans" path="/google/news/:category/:locale" :paramsDesc="['子分类标题', '地区语言设置，在地址栏 `?` 后，包含 `hl`，`gl`，以及 `ceid` 参数']"/>
+<Route author="zoenglinghou" example="/google/news/焦点新闻/hl=zh-CN&gl=CN&ceid=CN:zh-Hans" path="/google/news/:category/:locale" :paramsDesc="['子分类标题', '地区语言设置，在地址栏 `?` 后，包含 `hl`，`gl`，以及 `ceid` 参数']"/>
 
 ## 观察者网
 

lib/router.js

@@ -2455,7 +2455,7 @@ router.get('/cowlevel/element/:id', lazyloadRouteHandler('./routes/cowlevel/elem
 // router.get('/2048/bbs/:fid', lazyloadRouteHandler('./routes/2048/bbs'));
 
 // Google News
-router.get('/google/news/:category/:locale', lazyloadRouteHandler('./routes/google/news'));
+// router.get('/google/news/:category/:locale', lazyloadRouteHandler('./routes/google/news'));
 
 // 虛詞
 router.get('/p-articles/section/:section', lazyloadRouteHandler('./routes/p-articles/section'));

lib/v2/google/maintainer.js

@@ -3,6 +3,7 @@ module.exports = {
     '/citations/:id': ['KellyHwong'],
     '/doodles/:language?': ['xyqfer'],
     '/fonts/:sort?': ['Fatpandac'],
+    '/news/:category/:locale': ['zoenglinghou'],
     '/scholar/:query': ['HenryQW'],
     '/sites/:id': ['hoilc'],
     '/sites/recentChanges/:id': ['nczitzk'],

lib/v2/google/news.js

@@ -1,84 +1,66 @@
 const got = require('@/utils/got');
 const cheerio = require('cheerio');
-const url = require('url');
+const { parseDate } = require('@/utils/parse-date');
+const { art } = require('@/utils/render');
+const path = require('path');
+
+const baseUrl = 'https://news.google.com';
 
 module.exports = async (ctx) => {
     const category = ctx.params.category;
     const locale = ctx.params.locale;
 
-    const category_url = await ctx.cache.tryGet([locale, category], async () => {
-        const front_page = await got({
-            method: 'get',
-            url: `https://news.google.com/?${locale}`,
-        });
-
-        const front_data = front_page.data;
+    const categoryUrls = await ctx.cache.tryGet(`google:news:${locale}`, async () => {
+        const { data: front_data } = await got(`${baseUrl}/?${locale}`);
 
         const $ = cheerio.load(front_data);
-        const category_list = $('a.wmzpFf.yETrXb');
-
-        const category_text = [];
-        category_list.each((index, item) => {
-            category_text.push($(item).text());
-        });
-
-        return url.resolve('https://news.google.com', category_list.eq(category_text.indexOf(category)).attr('href'));
+        return [
+            ...$('a.brSCsc')
+                .toArray()
+                .slice(3) // skip Home, For you and Following
+                .map((item) => {
+                    item = $(item);
+                    return {
+                        category: item.text(),
+                        url: new URL(item.attr('href'), baseUrl).href,
+                    };
+                }),
+            ...$('a.aqvwYd') // Home
+                .toArray()
+                .map((item) => {
+                    item = $(item);
+                    return {
+                        category: item.text(),
+                        url: new URL(item.attr('href'), baseUrl).href,
+                    };
+                }),
+        ];
     });
+    const categoryUrl = categoryUrls.find((item) => item.category === category).url;
 
-    const response = await got({
-        method: 'get',
-        url: category_url,
-    });
-
-    const data = response.data;
+    const { data } = await got(categoryUrl);
     const $ = cheerio.load(data);
-    const list = $('div.xrnccd');
 
-    let itemPicUrl;
-    let meta;
-    let description;
-    let link;
+    const list = [...$('.UwIKyb'), ...$('.IBr9hb'), ...$('.IFHyqb')]; // 3 rows of news, 3-rows-wide news, single row news
+
+    const items = list.map((item) => {
+        item = $(item);
+        const title = item.find('h4').text();
+        return {
+            title,
+            description: art(path.join(__dirname, 'templates/news.art'), {
+                img: item.find('img.Quavad').attr('src'),
+                brief: title,
+            }),
+            pubDate: parseDate(item.find('time').attr('datetime')),
+            author: item.find('.oovtQ').text(),
+            link: new URL(item.find('a.WwrzSb').first().attr('href'), baseUrl).href,
+        };
+    });
 
     ctx.state.data = {
         title: $('title').text(),
-        link: category_url,
-        item:
-            list &&
-            list
-                .map((index, item) => {
-                    item = $(item);
-                    itemPicUrl = item.find('img').attr('src');
-
-                    meta = item.find('div.SVJrMe').first();
-
-                    description = item
-                        .find('h4.ipQwMb')
-                        .removeAttr('class')
-                        .each((index, item) => {
-                            $(item)
-                                .children()
-                                .removeAttr('class')
-                                .attr('href', url.resolve('https://news.google.com', $(item).children().attr('href')));
-                        })
-                        .toString();
-
-                    link = url.resolve('https://news.google.com', item.find('a.VDXfz').first().attr('href'));
-
-                    if (itemPicUrl !== undefined) {
-                        description = `${description}<img src="${itemPicUrl}">`;
-                    }
-                    if (description === null) {
-                        description = link;
-                    }
-
-                    return {
-                        title: item.find('h3').children().text(),
-                        description,
-                        pubDate: meta.find('time').first().attr('datetime'),
-                        author: meta.find('a.wEwyrc').first().text(),
-                        link,
-                    };
-                })
-                .get(),
+        link: categoryUrl,
+        item: items,
     };
 };

lib/v2/google/router.js

@@ -3,6 +3,7 @@ module.exports = function (router) {
     router.get('/citations/:id', require('./citations'));
     router.get('/doodles/:language?', require('./doodles'));
     router.get('/fonts/:sort?', require('./fonts'));
+    router.get('/news/:category/:locale', require('./news'));
     router.get('/scholar/:query', require('./scholar'));
     router.get('/sites/recentChanges/:id', require('./sitesRecentChanges'));
     router.get('/sites/:id', require('./sites'));

lib/v2/google/templates/news.art

@@ -0,0 +1,5 @@
+{{ if img }}
+<img src="{{ img }}">
+<br>
+{{ /if }}
+{{ brief }}