feat(route): add kyodonews (#9435)

Signed-off-by: Rongrong <15956627+Rongronggg9@users.noreply.github.com>

docs/traditional-media.md

@@ -827,6 +827,16 @@ Type 栏目：
 
 <Route author="nczitzk" example="/pts/dailynews" path="/pts/dailynews"/>
 
+## 共同网
+
+### 最新报道
+
+<Route author="Rongronggg9" example="/kyodonews" path="/kyodonews/:language?/:keyword?" :paramsDesc="['语言: `china` = 简体中文 (默认), `tchina` = 繁體中文', '关键词']">  
+
+`keyword` 为关键词，由于共同网有许多关键词并不在主页列出，此处不一一列举，可从关键词页的 URL 的最后一级路径中提取。如 `日中关系` 的关键词页 URL 为 `https://china.kyodonews.net/news/japan-china_relationship`, 则将 `japan-china_relationship` 填入 `keyword`。特别地，当填入 `rss` 时，将从共同网官方 RSS 中抓取文章；略去时，将从首页抓取最新报道 (注意：首页更新可能比官方 RSS 稍慢)。
+
+</Route>
+
 ## 国际金融报栏目
 
 ### 栏目

lib/v2/kyodonews/index.js

@@ -0,0 +1,88 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const timezone = require('@/utils/timezone');
+const { parseDate } = require('@/utils/parse-date');
+
+module.exports = async (ctx) => {
+    const language = ctx.params.language ?? 'china';
+    const keyword = ctx.params.keyword === 'RSS' ? 'rss' : ctx.params.keyword ?? '';
+
+    // raise error for invalid languages
+    if (!['china', 'tchina'].includes(language)) {
+        throw Error('Invalid language');
+    }
+
+    const rootUrl = `https://${language}.kyodonews.net`;
+    const currentUrl = `${rootUrl}/${keyword ? (keyword === 'rss' ? 'rss/news.xml' : `news/${keyword}`) : ''}`;
+
+    let response;
+    try {
+        response = await got(currentUrl);
+    } catch (e) {
+        throw e.response && e.response.statusCode === 404 ? new Error('Invalid keyword') : e;
+    }
+
+    const $ = cheerio.load(response.data, { xmlMode: keyword === 'rss' });
+
+    let title, description, items;
+    if (keyword === 'rss') {
+        title = $('channel > title').text();
+        description = $('channel > description').text();
+        items = $('item')
+            .map((_, item) => {
+                const $item = $(item);
+                const link = $item.find('link').text();
+                // const pubDate = $item.find('pubDate').text();
+                return {
+                    link,
+                    // pubDate,  // no need to normalize because it's from a valid RSS feed
+                };
+            })
+            .get();
+    } else {
+        title = $('head > title').text();
+        description = $('meta[name="description"]').attr('content');
+        items = $('div.sec-latest > ul > li')
+            .map((_, item) => {
+                item = $(item);
+                const link = item.find('a').attr('href');
+                return {
+                    link: `${link.startsWith('http') ? '' : rootUrl}${link}`,
+                };
+            })
+            .get();
+    }
+
+    items = await Promise.all(
+        items.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const detailResponse = await got({
+                    method: 'get',
+                    url: item.link,
+                });
+
+                const $ = cheerio.load(detailResponse.data);
+                item.title = $('head > title').text();
+                item.author = $('meta[name="author"]').attr('content');
+                item.description = $('div.article-body')
+                    .html()
+                    .replace(/（完）(?=<\/p>\s*$)/m, '');
+
+                const pubDate_match = $('script[type="application/ld+json"]')
+                    .html()
+                    .match(/"datePublished":"([\d\s-:]*?)"/);
+                if (pubDate_match) {
+                    item.pubDate = timezone(parseDate(pubDate_match[1]), 9);
+                }
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title,
+        description,
+        link: currentUrl,
+        item: items,
+    };
+};

lib/v2/kyodonews/maintainer.js

@@ -0,0 +1,3 @@
+module.exports = {
+    '/:language?/:keyword?': ['Rongronggg9'],
+};

lib/v2/kyodonews/radar.js

@@ -0,0 +1,33 @@
+module.exports = {
+    'kyodonews.net': {
+        _name: '共同网',
+        china: [
+            {
+                title: '最新报道',
+                docs: 'https://docs.rsshub.app/traditional-media.html#gong-tong-wang-zui-xin-bao-dao',
+                source: '/',
+                target: '/kyodonews/china',
+            },
+            {
+                title: '关键词',
+                docs: 'https://docs.rsshub.app/traditional-media.html#gong-tong-wang-zui-xin-bao-dao',
+                source: '/news/:keyword',
+                target: '/kyodonews/china/:keyword?',
+            },
+        ],
+        tchina: [
+            {
+                title: '最新報道',
+                docs: 'https://docs.rsshub.app/traditional-media.html#gong-tong-wang-zui-xin-bao-dao',
+                source: '/',
+                target: '/kyodonews/tchina/:keyword?',
+            },
+            {
+                title: '關鍵詞',
+                docs: 'https://docs.rsshub.app/traditional-media.html#gong-tong-wang-zui-xin-bao-dao',
+                source: '/news/:keyword',
+                target: '/kyodonews/tchina/:keyword?',
+            },
+        ],
+    },
+};

lib/v2/kyodonews/router.js

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