fix(route): fix invalid region and refactor to v2 (#11777)

* fix(route): fix invalid region and refactor to v2

* refactor: change route path

---------

docs/en/traditional-media.md

@@ -680,7 +680,7 @@ Provide full article RSS for WSJ topics.
 
 ### News
 
-<RouteEn author="KeiLongW" example="/yahoo-news/hk/world" path="/yahoo-news/:region/:category?" :paramsDesc="['Region','Category']">
+<RouteEn author="KeiLongW" example="/yahoo/news/hk/world" path="/yahoo/news/:region/:category?" :paramsDesc="['Region','Category']">
 
 `Region`
 | Hong Kong | Taiwan | US |

docs/traditional-media.md

@@ -530,7 +530,7 @@ Solidot 提供的 feed:
 
 ### 新聞
 
-<Route author="KeiLongW" example="/yahoo-news/hk/world" path="/yahoo-news/:region/:category?" :paramsDesc="['地区','类别']">
+<Route author="KeiLongW" example="/yahoo/news/hk/world" path="/yahoo/news/:region/:category?" :paramsDesc="['地区','类别']" radar="1">
 
 `地区`
 

lib/router.js

@@ -2315,7 +2315,7 @@ router.get('/chrome/webstore/extensions/:id', lazyloadRouteHandler('./routes/chr
 router.get('/rthk-news/:lang/:category', lazyloadRouteHandler('./routes/rthk-news/index'));
 
 // yahoo
-router.get('/yahoo-news/:region/:category?', lazyloadRouteHandler('./routes/yahoo-news/index'));
+// router.get('/yahoo-news/:region/:category?', lazyloadRouteHandler('./routes/yahoo-news/index'));
 
 // Yahoo!テレビ
 router.get('/yahoo-jp-tv/:query', lazyloadRouteHandler('./routes/yahoo-jp-tv/index'));

lib/v2/yahoo/maintainer.js

@@ -0,0 +1,3 @@
+module.exports = {
+    '/news/:region/:category?': ['KeiLongW'],
+};

lib/v2/yahoo/news/index.js

@@ -4,17 +4,17 @@ const cheerio = require('cheerio');
 const { isValidHost } = require('@/utils/valid-host');
 
 module.exports = async (ctx) => {
-    const region = ctx.params.region === 'en' ? '' : ctx.params.region.toLowerCase() + '.';
-    if (!isValidHost(region)) {
+    const region = ctx.params.region === 'en' ? '' : ctx.params.region.toLowerCase();
+    if (!isValidHost(region) && region !== '') {
         throw Error('Invalid region');
     }
     const category = ctx.params.category ? ctx.params.category.toLowerCase() : '';
-    const rssUrl = `https://${region}news.yahoo.com/rss/${category}`;
+    const rssUrl = `https://${region ? region + '.' : ''}news.yahoo.com/rss/${category}`;
     const feed = await parser.parseURL(rssUrl);
     const title = feed.title;
     const link = feed.link;
     const description = feed.description;
-    const filteredItems = feed.items.filter((item) => !item.link.includes('promotions') && item.link.includes('yahoo.com'));
+    const filteredItems = feed.items.filter((item) => !item.link.includes('promotions') && new URL(item.link).hostname.match(/.*\.yahoo\.com$/));
     const items = await Promise.all(
         filteredItems.map(async (item) => {
             const description = await ctx.cache.tryGet(item.link, async () => {
@@ -24,10 +24,8 @@ module.exports = async (ctx) => {
                 });
 
                 const $ = cheerio.load(response.data);
-                const $descriptionRoot = cheerio.load('<div><div class="img-root"></div><div class="content-root"></div></div>');
-                $descriptionRoot('div.img-root').append($('article img:not(header img)'));
-                $descriptionRoot('div.content-root').append($('article p').removeAttr('content'));
-                return $descriptionRoot.html();
+                $('.caas-content-byline-wrapper, .caas-xray-wrapper, .caas-header, .caas-readmore').remove();
+                return $('.caas-content-wrapper').html();
             });
             const single = {
                 title: item.title,
@@ -35,7 +33,7 @@ module.exports = async (ctx) => {
                 pubDate: item.pubDate,
                 link: item.link,
             };
-            return Promise.resolve(single);
+            return single;
         })
     );
 

lib/v2/yahoo/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'yahoo.com': {
+        _name: 'Yahoo',
+        '.': [
+            {
+                title: '新闻',
+                docs: 'https://docs.rsshub.app/traditional-media.html#yahoo',
+                source: ['/'],
+                target: '/yahoo/news/:region/:category?',
+            },
+        ],
+    },
+};

lib/v2/yahoo/router.js

@@ -0,0 +1,3 @@
+module.exports = function (router) {
+    router.get('/news/:region/:category?', require('./news/index'));
+};