feat(route): cw (#10758)

* feat(route): cw

* fix: typo

* fix: use custom got

docs/traditional-media.md

@@ -1795,6 +1795,38 @@ category 对应的关键词有
 
 <Route author="TonyRL" example="/taiwannews/hot" path="/taiwannews/hot/:lang?" :paramsDesc="['語言，`en` 或 `zh`，默認 `en`']" radar="1" rssbud="1"/>
 
+## 天下雜誌
+
+### 最新上線
+
+<Route author="TonyRL" example="/cw/today" path="/cw/today" radar="1" rssbud="1"/>
+
+### 主頻道
+
+<Route author="TonyRL" example="/cw/master/8" path="/cw/master/:channel" :paramsDesc="['主頻道 ID，可在 URL 中找到']" radar="1" rssbud="1">
+
+| 主頻道名稱 | 主頻道 ID |
+| ----- | ------ |
+| 財經    | 8      |
+| 產業    | 7      |
+| 國際    | 9      |
+| 管理    | 10     |
+| 環境    | 12     |
+| 教育    | 13     |
+| 人物    | 14     |
+| 政治社會  | 77     |
+| 調查排行  | 15     |
+| 健康關係  | 79     |
+| 時尚品味  | 11     |
+| 運動生活  | 103    |
+| 重磅外媒  | 16     |
+
+</Route>
+
+### 子頻道
+
+<Route author="TonyRL" example="/cw/sub/615" path="/cw/sub/:channel" :paramsDesc="['子頻道 ID，可在 URL 中找到']" radar="1" rssbud="1"/>
+
 ## 卫报 The Guardian
 
 通过提取文章全文，以提供比官方源更佳的阅读体验。

lib/v2/cw/maintainer.js

@@ -0,0 +1,5 @@
+module.exports = {
+    '/master/:channel': ['TonyRL'],
+    '/sub/:channel': ['TonyRL'],
+    '/today': ['TonyRL'],
+};

lib/v2/cw/master.js

@@ -0,0 +1,28 @@
+const cheerio = require('cheerio');
+const { baseUrl, cookieJar, got, parseList, parseItems } = require('./utils');
+
+module.exports = async (ctx) => {
+    const { channel } = ctx.params;
+    const pageUrl = `${baseUrl}/masterChannel.action`;
+    const { data: response } = await got(pageUrl, {
+        headers: {
+            Referer: baseUrl,
+        },
+        cookieJar,
+        searchParams: {
+            idMasterChannel: channel,
+        },
+    });
+    const $ = cheerio.load(response);
+
+    const list = parseList($, ctx.query.limit ? Number(ctx.query.limit) : 100);
+    const items = await parseItems(list, ctx.cache.tryGet);
+
+    ctx.state.data = {
+        title: $('head title').text(),
+        description: $('meta[name=description]').attr('content'),
+        image: `${baseUrl}/assets_new/img/fbshare.jpg`,
+        language: $('meta[property="og:locale"]').attr('content'),
+        item: items,
+    };
+};

lib/v2/cw/radar.js

@@ -0,0 +1,25 @@
+module.exports = {
+    'cw.com.tw': {
+        _name: '天下雜誌',
+        '.': [
+            {
+                title: '最新上線',
+                docs: 'https://docs.rsshub.app/traditional-media.html#tian-xia-za-zhi',
+                source: ['/today', '/'],
+                target: '/cw/today',
+            },
+            {
+                title: '主頻道',
+                docs: 'https://docs.rsshub.app/traditional-media.html#tian-xia-za-zhi',
+                source: ['/masterChannel.action'],
+                target: (_, url) => `/cw/master/${new URL(url).searchParams.get('idMasterChannel')}`,
+            },
+            {
+                title: '子頻道',
+                docs: 'https://docs.rsshub.app/traditional-media.html#tian-xia-za-zhi',
+                source: ['/subchannel.action'],
+                target: (_, url) => `/cw/sub/${new URL(url).searchParams.get('idSubChannel')}`,
+            },
+        ],
+    },
+};

lib/v2/cw/router.js

@@ -0,0 +1,5 @@
+module.exports = (router) => {
+    router.get('/master/:channel', require('./master'));
+    router.get('/sub/:channel', require('./sub'));
+    router.get('/today', require('./today'));
+};

lib/v2/cw/sub.js

@@ -0,0 +1,28 @@
+const cheerio = require('cheerio');
+const { baseUrl, cookieJar, got, parseList, parseItems } = require('./utils');
+
+module.exports = async (ctx) => {
+    const { channel } = ctx.params;
+    const pageUrl = `${baseUrl}/subchannel.action`;
+    const { data: response } = await got(pageUrl, {
+        headers: {
+            Referer: baseUrl,
+        },
+        cookieJar,
+        searchParams: {
+            idSubChannel: channel,
+        },
+    });
+    const $ = cheerio.load(response);
+
+    const list = parseList($, ctx.query.limit ? Number(ctx.query.limit) : 100);
+    const items = await parseItems(list, ctx.cache.tryGet);
+
+    ctx.state.data = {
+        title: $('head title').text(),
+        description: $('meta[name=description]').attr('content'),
+        image: `${baseUrl}/assets_new/img/fbshare.jpg`,
+        language: $('meta[property="og:locale"]').attr('content'),
+        item: items,
+    };
+};

lib/v2/cw/today.js

@@ -0,0 +1,24 @@
+const cheerio = require('cheerio');
+const { baseUrl, cookieJar, got, parseList, parseItems } = require('./utils');
+
+module.exports = async (ctx) => {
+    const pageUrl = `${baseUrl}/today`;
+    const { data: response } = await got(pageUrl, {
+        headers: {
+            Referer: baseUrl,
+        },
+        cookieJar,
+    });
+    const $ = cheerio.load(response);
+
+    const list = parseList(response, ctx.query.limit ? Number(ctx.query.limit) : 100);
+    const items = await parseItems(list, ctx.cache.tryGet);
+
+    ctx.state.data = {
+        title: $('head title').text(),
+        description: $('meta[name=description]').attr('content'),
+        image: `${baseUrl}/assets_new/img/fbshare.jpg`,
+        language: $('meta[property="og:locale"]').attr('content'),
+        item: items,
+    };
+};

lib/v2/cw/utils.js

@@ -0,0 +1,63 @@
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const { CookieJar } = require('tough-cookie');
+const cookieJar = new CookieJar();
+const config = require('@/config').value;
+
+const baseUrl = 'https://www.cw.com.tw';
+
+const got = require('@/utils/got').extend({
+    headers: {
+        'User-Agent': config.trueUA,
+    },
+});
+
+const parseList = ($, limit) =>
+    $('.caption')
+        .toArray()
+        .map((item) => {
+            item = $(item);
+            return {
+                title: item.find('h3').text(),
+                link: item.find('h3 a').attr('href'),
+                pubDate: parseDate(item.find('time').text()),
+            };
+        })
+        .slice(0, limit);
+
+const parseItems = (list, tryGet) =>
+    Promise.all(
+        list.map((item) =>
+            tryGet(item.link, async () => {
+                const { data: response } = await got(item.link, {
+                    cookieJar,
+                });
+
+                const $ = cheerio.load(response);
+                const meta = JSON.parse($('head script[type="application/ld+json"]').eq(0).text());
+                $('.article__head .breadcrumb, .article__head h1, .article__provideViews, .ad').remove();
+                $('img.lazyload').each((_, img) => {
+                    if (img.attribs['data-src']) {
+                        img.attribs.src = img.attribs['data-src'];
+                        delete img.attribs['data-src'];
+                    }
+                });
+
+                item.title = $('head title').text();
+                item.category = $('meta[name=keywords]').attr('content').split(',');
+                item.pubDate = parseDate(meta.datePublished);
+                item.author = meta.author.name.replace(',', ' ') || meta.publisher.name;
+                item.description = $('.article__head .container').html() + $('.article__content').html();
+
+                return item;
+            })
+        )
+    );
+
+module.exports = {
+    baseUrl,
+    cookieJar,
+    got,
+    parseList,
+    parseItems,
+};