From e875690c079fad6054796a68af373f7ac0298bcb Mon Sep 17 00:00:00 2001
From: KeiLongW <wong525305@gmail.com>
Date: Thu, 16 Jan 2020 14:15:40 +0800
Subject: [PATCH] feat: Add Yahoo news (#3720)

---
 docs/en/traditional-media.md   | 18 ++++++++++++++
 docs/traditional-media.md      | 18 ++++++++++++++
 lib/router.js                  |  3 +++
 lib/routes/yahoo-news/index.js | 44 ++++++++++++++++++++++++++++++++++
 4 files changed, 83 insertions(+)
 create mode 100755 lib/routes/yahoo-news/index.js

diff --git a/docs/en/traditional-media.md b/docs/en/traditional-media.md
index 98a3bb388e..02891c97a1 100644
--- a/docs/en/traditional-media.md
+++ b/docs/en/traditional-media.md
@@ -81,3 +81,21 @@ Provides a better reading experience (full text articles) over the official one.
 | (空)               | dual            | en      | dual-traditionalchinese               | traditionalchinese  |
 
 </RouteEn>
+
+## Yahoo
+
+### News
+
+<Route author="KeiLongW" example="/yahoo-news/hk/world" path="/yahoo-news/:region/:category?" :paramsDesc="['Region','Category']">
+
+`Region`
+| Hong Kong | Taiwan | US |
+| -- | -- | -- |
+| hk | tw | en |
+
+`Category`
+| All | World | Business | Entertainment | Sports | Health |
+| -- | -- | -- | -- | -- | -- |
+| (Empty) | world | business | entertainment | sports | health |
+
+</Route>
diff --git a/docs/traditional-media.md b/docs/traditional-media.md
index 1099c28d7b..58c51cb692 100644
--- a/docs/traditional-media.md
+++ b/docs/traditional-media.md
@@ -96,6 +96,24 @@ Solidot 提供的 feed:
 
 <Route author="xyqfer" example="/the-economist/gre-vocabulary" path="/the-economist/gre-vocabulary" />
 
+## Yahoo
+
+### 新聞
+
+<Route author="KeiLongW" example="/yahoo-news/hk/world" path="/yahoo-news/:region/:category?" :paramsDesc="['地区','类别']">
+
+`地区`
+| 香港 | 台灣 | 美國 |
+| -- | -- | -- |
+| hk | tw | en |
+
+`类別`
+| 新聞總集 | 兩岸國際 | 財經 | 娛樂 | 體育 | 健康 |
+| -- | -- | -- | -- | -- | -- |
+| (空) | world | business | entertainment | sports | health |
+
+</Route>
+
 ## 半月谈
 
 ### 板块
diff --git a/lib/router.js b/lib/router.js
index 3c2cc5fbd6..ef58964b96 100644
--- a/lib/router.js
+++ b/lib/router.js
@@ -2101,6 +2101,9 @@ router.get('/mlog-club/projects', require('./routes/mlog-club/projects'));
 // Chrome 网上应用店
 router.get('/chrome/webstore/extensions/:id', require('./routes/chrome/extensions'));
 
+// yahoo
+router.get('/yahoo-news/:region/:category?', require('./routes/yahoo-news/index'));
+
 // 白鲸出海
 router.get('/baijing', require('./routes/baijing'));
 
diff --git a/lib/routes/yahoo-news/index.js b/lib/routes/yahoo-news/index.js
new file mode 100755
index 0000000000..9875cc7aa6
--- /dev/null
+++ b/lib/routes/yahoo-news/index.js
@@ -0,0 +1,44 @@
+const got = require('@/utils/got');
+const parser = require('@/utils/rss-parser');
+const cheerio = require('cheerio');
+
+module.exports = async (ctx) => {
+    const region = ctx.params.region === 'en' ? '' : ctx.params.region.toLowerCase() + '.';
+    const category = ctx.params.category ? ctx.params.category.toLowerCase() : '';
+    const rssUrl = `https://${region}news.yahoo.com/rss/${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 items = await Promise.all(
+        filteredItems.map(async (item) => {
+            const description = await ctx.cache.tryGet(item.link, async () => {
+                const response = await got({
+                    method: 'get',
+                    url: item.link,
+                });
+
+                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'));
+                $descriptionRoot('div.content-root').append($('article p').removeAttr('content'));
+                return $descriptionRoot.html();
+            });
+            const single = {
+                title: item.title,
+                description,
+                pubDate: item.pubDate,
+                link: item.link,
+            };
+            return Promise.resolve(single);
+        })
+    );
+
+    ctx.state.data = {
+        title,
+        link,
+        description,
+        item: items,
+    };
+};