feat(route): add 聯合新聞網 (#10654)

* feat(route): add 聯合新聞網

* fix(route): concatenate the url correctly

* feat(route): fetch link name

* feat(route): description uses HTML instead of raw text

* feat(route): remove `from` search params from link

* feat(route): improve feed description readability

* fix(route): article was not properly parsed

* refactor(route): improve feed fetching

docs/traditional-media.md

@@ -1376,6 +1376,18 @@ category 对应的关键词有
 
 <Route author="shunf4" example="/zaobao/interactive-graphics" path="/zaobao/interactive-graphics" />
 
+## 聯合新聞網
+
+### 即時新聞
+
+<Route author="miles170" example="/udn/news/breakingnews/99" path="/udn/news/breakingnews/:id" :paramsDesc="['类别']" radar="1">
+
+| 0  | 1  | 2  | 3  | 4  | 5  | 6  | 7  | 8  | 9  | 11 | 12 | 13 | 99  |
+| -- | -- | -- | -- | -- | -- | -- | -- | -- | -- | -- | -- | -- | --- |
+| 精選 | 要聞 | 社會 | 地方 | 兩岸 | 國際 | 財經 | 運動 | 娛樂 | 生活 | 股市 | 文教 | 數位 | 不分類 |
+
+</Route>
+
 ## 路透社
 
 ::: warning 迁移说明

lib/v2/udn/breaking-news.js

@@ -0,0 +1,85 @@
+const got = require('@/utils/got');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+const cheerio = require('cheerio');
+const { art } = require('@/utils/render');
+const path = require('path');
+
+module.exports = async (ctx) => {
+    const id = ctx.params.id;
+    const link = `/news/breaknews/1/${id}#breaknews`;
+    const name = await getLinkName(ctx, link);
+    const rootUrl = 'https://udn.com';
+    const response = await got(`${rootUrl}/api/more?page=1&channelId=1&cate_id=${id}&type=breaknews`);
+    const items = await Promise.all(
+        response.data.lists.map((item) => {
+            let link = item.titleLink.indexOf('http://') === 0 || item.titleLink.indexOf('https://') === 0 ? item.titleLink : `${rootUrl}${item.titleLink}`;
+            const linkUrl = new URL(link);
+            // cleanup query paramter
+            linkUrl.query = linkUrl.search = '';
+            link = linkUrl.toString();
+            return ctx.cache.tryGet(link, async () => {
+                let result = await got(link);
+                // VIP article requires redirection
+                // e.g. https://udn.com/news/story/7331/6576320
+                const vip = result.data.match(/<script language=javascript>window\.location\.href="(https?:\/\/[^"]+")/);
+                if (vip) {
+                    result = await got(vip[1]);
+                }
+                const $ = cheerio.load(result.data);
+                const metadata = $('script[type="application/ld+json"]').eq(0).text().trim();
+                const data = metadata.startsWith('[') ? JSON.parse(metadata)[0] : JSON.parse(metadata);
+                // e.g. https://udn.com/news/story/7331/6576320
+                const content = $('.article-content__editor');
+                // e.g. https://udn.com/news/story/7240/6576424
+                const body = $('.article-body__editor');
+                let description = '';
+                if (data.image) {
+                    description += art(path.join(__dirname, 'templates/figure.art'), {
+                        src: data.image.contentUrl,
+                        alt: data.image.name,
+                    });
+                }
+                if (content.length) {
+                    description += content.html();
+                } else if (body.length) {
+                    description += body.html();
+                }
+                return {
+                    title: item.title,
+                    author: data.author.name,
+                    description,
+                    pubDate: timezone(parseDate(item.time.date, 'YYYY-MM-DD HH:mm'), +8),
+                    category: [data.articleSelection, ...data.keywords.split(',')],
+                    link,
+                };
+            });
+        })
+    );
+
+    ctx.state.data = {
+        title: `即時${name} - 聯合新聞網`,
+        link: `${rootUrl}${link}`,
+        description: 'udn.com 提供即時新聞以及豐富的政治、社會、地方、兩岸、國際、財經、數位、運動、NBA、娛樂、生活、健康、旅遊新聞，以最即時、多元的內容，滿足行動世代的需求',
+        item: items,
+    };
+};
+
+const getLinkName = async (ctx, link) => {
+    const url = 'https://udn.com/news/breaknews';
+    const links = await ctx.cache.tryGet(url, async () => {
+        const result = await got(url);
+        const $ = cheerio.load(result.data);
+        const data = $('.cate-list__subheader a')
+            .get()
+            .map((item) => {
+                item = $(item);
+                return { [item.attr('href')]: item.text().trim() };
+            });
+        return Object.assign({}, ...data);
+    });
+    if (link in links) {
+        return links[link];
+    }
+    return '列表';
+};

lib/v2/udn/maintainer.js

@@ -0,0 +1,3 @@
+module.exports = {
+    '/news/breakingnews/:id': ['miles170'],
+};

lib/v2/udn/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'udn.com': {
+        _name: '聯合新聞網',
+        '.': [
+            {
+                title: '即時新聞',
+                docs: 'https://docs.rsshub.app/new-media.html#lian-he-xin-wen-wang-ji-shi-xin-wen',
+                source: '/news/breaknews/1/:id',
+                target: '/udn/news/breakingnews/:id',
+            },
+        ],
+    },
+};

lib/v2/udn/router.js

@@ -0,0 +1,3 @@
+module.exports = (router) => {
+    router.get('/news/breakingnews/:id', require('./breaking-news'));
+};

lib/v2/udn/templates/figure.art

@@ -0,0 +1,4 @@
+<figure>
+<picture><img src={{ src }} alt={{ alt }}></picture>
+<figurecaption>{{ alt }}</figurecaption>
+</figure>