Merge pull request #654 from DIYgod/master

[pull] master from diygod:master

docs/en/new-media.md

@@ -434,6 +434,16 @@ Provides a better reading experience (full text articles) over the official one.
 
 </RouteEn>
 
+## Kuwait Local
+
+### Latest News
+
+<RouteEn author="TonyRL" example="/kuwaitlocal" path="/kuwaitlocal" radar="1" rssbud="1"/>
+
+### Categorised News
+
+<RouteEn author="TonyRL" example="/kuwaitlocal/article" path="/kuwaitlocal/:category?" :paramsDesc="['Category name, can be found in URL, `latest` by default']" radar="1" rssbud="1"/>
+
 ## Letterboxd
 
 ### User diary

docs/en/traditional-media.md

@@ -160,6 +160,18 @@ Generates full-text feeds that the official feed doesn't provide.
 
 </RouteEn>
 
+## CNBC
+
+### Full article RSS
+
+<RouteEn author="TonyRL" example="/cnbc/rss" path="/cnbc/rss/:id?" :paramsDesc="['Channel ID, can be found in Official RSS URL, `100003114` (Top News) by default']">
+
+Provides a better reading experience (full articles) over the official ones.
+
+Support all channels, refer to [CNBC RSS feeds](https://www.cnbc.com/rss-feeds/).
+
+</RouteEn>
+
 ## Deutsche Welle
 
 ### News

docs/programming.md

@@ -720,7 +720,7 @@ GitHub 官方也提供了一些 RSS:
 
 ### 分类订阅
 
-<Route author="qwertyuiop6" example="/anquanke/week" path="/anquanke/:category" :paramsDesc="['分类订阅']" radar="1" rssbud="1">
+<Route author="qwertyuiop6" example="/anquanke/week" path="/anquanke/:category/:fulltext?" :paramsDesc="['分类订阅', '是否获取全文，如需获取全文参数传入 `quanwen` 或 `fulltext`']" radar="1" rssbud="1">
 
 | 360 网络安全周报 | 活动       | 知识        | 资讯   | 招聘  | 工具   |
 | ---------- | -------- | --------- | ---- | --- | ---- |

docs/traditional-media.md

@@ -141,6 +141,18 @@ pageClass: routes
 
 </Route>
 
+## CNBC
+
+### 全文 RSS
+
+<Route author="TonyRL" example="/cnbc/rss" path="/cnbc/rss/:id?" :paramsDesc="['频道 ID，可在官方频道 RSS URL 中找到，留空为 `100003114` (Top News)']">
+
+通过提取文章全文，以提供比官方源更佳的阅读体验。
+
+支持所有频道，频道名称见 [官方频道 RSS](https://www.cnbc.com/rss-feeds/)。
+
+</Route>
+
 ## Deutsche Welle 德国之声
 
 ### 新闻

lib/v2/anquanke/category.js

@@ -1,21 +1,35 @@
 const got = require('@/utils/got');
+const cheerio = require('cheerio');
 const { parseDate } = require('@/utils/parse-date');
 const timezone = require('@/utils/timezone');
 
 module.exports = async (ctx) => {
     const api = 'https://api.anquanke.com/data/v1/posts?size=10&page=1&category=';
     const type = ctx.params.category;
+    const fulltext = ctx.params.fulltext;
     const host = 'https://www.anquanke.com';
     const res = await got(`${api}${type}`);
     const dataArray = res.data.data;
 
-    const items = dataArray.map((item) => ({
-        title: item.title,
-        description: item.desc,
-        pubDate: timezone(parseDate(item.date), +8),
-        link: `${host}/${type === 'week' ? 'week' : 'post'}/id/${item.id}`,
-        author: item.author.nickname,
-    }));
+    const items = await Promise.all(
+        dataArray.map(async (item) => {
+            const art_url = `${host}/${type === 'week' ? 'week' : 'post'}/id/${item.id}`;
+            return {
+                title: item.title,
+                description:
+                    fulltext === 'fulltext' || fulltext === 'quanwen'
+                        ? await ctx.cache.tryGet(art_url, async () => {
+                              const { data: res } = await got(art_url);
+                              const content = cheerio.load(res);
+                              return content('#js-article').html();
+                          })
+                        : item.desc,
+                pubDate: timezone(parseDate(item.date), +8),
+                link: art_url,
+                author: item.author.nickname,
+            };
+        })
+    );
 
     ctx.state.data = {
         title: `安全客-${dataArray[0].category_name}`,

lib/v2/anquanke/maintainer.js

@@ -1,4 +1,4 @@
 module.exports = {
     // '/vul': ['qwertyuiop6'],
-    '/:category': ['qwertyuiop6'],
+    '/:category/:fulltext?': ['qwertyuiop6'],
 };

lib/v2/anquanke/router.js

@@ -1,4 +1,4 @@
 module.exports = (router) => {
     // router.get('/vul', require('./vul')); // 404
-    router.get('/:category', require('./category'));
+    router.get('/:category/:fulltext?', require('./category'));
 };

lib/v2/cnbc/maintainer.js

@@ -0,0 +1,3 @@
+module.exports = {
+    '/rss/:id?': ['TonyRL'],
+};

lib/v2/cnbc/radar.js

@@ -0,0 +1,21 @@
+module.exports = {
+    'cnbc.com': {
+        _name: 'CNBC',
+        search: [
+            {
+                title: '全文 RSS',
+                docs: 'https://docs.rsshub.app/traditional-media.html#cnbc',
+                source: ['/rs/search/combinedcms/view.xml'],
+                target: (_, url) => `/cnbc/rss/${new URL(url).searchParams.get('id')}`,
+            },
+        ],
+        www: [
+            {
+                title: '全文 RSS',
+                docs: 'https://docs.rsshub.app/traditional-media.html#cnbc',
+                source: ['/id/:id/device/rss/rss.html'],
+                target: '/cnbc/rss/:id',
+            },
+        ],
+    },
+};

lib/v2/cnbc/router.js

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

lib/v2/cnbc/rss.js

@@ -0,0 +1,54 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const parser = require('@/utils/rss-parser');
+
+module.exports = async (ctx) => {
+    const { id = '100003114' } = ctx.params;
+    const feed = await parser.parseURL(`https://search.cnbc.com/rs/search/combinedcms/view.xml?partnerId=wrss01&id=${id}`);
+
+    const items = await Promise.all(
+        feed.items
+            .filter((i) => !i.link.startsWith('https://www.cnbc.com/select/'))
+            .map((item) =>
+                ctx.cache.tryGet(item.link, async () => {
+                    const { data: response } = await got(item.link);
+                    const $ = cheerio.load(response);
+
+                    delete item.content;
+                    delete item.contentSnippet;
+                    delete item.isoDate;
+
+                    item.description = '';
+                    if ($('.RenderKeyPoints-keyPoints').length) {
+                        $('.RenderKeyPoints-keyPoints').html();
+                    }
+                    if ($('.FeaturedContent-articleBody').length) {
+                        item.description += $('.FeaturedContent-articleBody').html();
+                    }
+                    if ($('.ArticleBody-articleBody').length) {
+                        item.description += $('.ArticleBody-articleBody').html();
+                    }
+                    if ($('.LiveBlogBody-articleBody').length) {
+                        item.description += $('.LiveBlogBody-articleBody').html();
+                    }
+                    if ($('.ClipPlayer-clipPlayer').length) {
+                        item.description += $('.ClipPlayer-clipPlayer').html();
+                    }
+
+                    const meta = JSON.parse($('[type=application/ld+json]').last().text());
+                    item.author = meta.author ? (meta.author.name ? meta.author.name : meta.author.map((a) => a.name).join(', ')) : null;
+                    item.category = meta.keywords;
+
+                    return item;
+                })
+            )
+    );
+
+    ctx.state.data = {
+        title: feed.title,
+        link: feed.link,
+        description: feed.description,
+        item: items,
+        language: feed.language,
+    };
+};

lib/v2/kuwaitlocal/index.js

@@ -0,0 +1,47 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+
+const baseUrl = 'https://kuwaitlocal.com';
+
+module.exports = async (ctx) => {
+    const { category = 'latest' } = ctx.params;
+    const url = `${baseUrl}/news/${category === 'latest' ? category : `categories/${category}`}`;
+
+    const { data: response } = await got(url);
+    const $ = cheerio.load(response);
+    const list = $('.news_list a')
+        .toArray()
+        .map((item) => {
+            item = $(item);
+            return {
+                title: item.text(),
+                link: baseUrl + item.attr('href'),
+            };
+        });
+
+    const items = await Promise.all(
+        list.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const { data: response } = await got(item.link);
+                const $ = cheerio.load(response);
+
+                item.pubDate = parseDate($('.date_icon').next().text());
+                $('[id^=div-gpt-ad], .pad_10_0, .hide_desktop').remove();
+                item.description = $('.mob_pad_view').html();
+                item.category = $('.news_tags a')
+                    .toArray()
+                    .map((item) => $(item).text());
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: $('head title').text().trim(),
+        description: $('head meta[name="description"]').attr('content').trim(),
+        link: url,
+        item: items,
+        language: 'en',
+    };
+};

lib/v2/kuwaitlocal/maintainer.js

@@ -0,0 +1,4 @@
+module.exports = {
+    '/': ['TonyRL'],
+    '/:category?': ['TonyRL'],
+};

lib/v2/kuwaitlocal/radar.js

@@ -0,0 +1,19 @@
+module.exports = {
+    'kuwaitlocal.com': {
+        _name: 'Kuwait Local',
+        '.': [
+            {
+                title: 'Latest News',
+                docs: 'https://docs.rsshub.app/en/new-media.html#kuwait-local',
+                source: ['/news/latest', '/news', '/'],
+                target: '/kuwaitlocal',
+            },
+            {
+                title: 'Categorised News',
+                docs: 'https://docs.rsshub.app/en/new-media.html#kuwait-local',
+                source: ['/news/categories/:category'],
+                target: '/kuwaitlocal/:category',
+            },
+        ],
+    },
+};

lib/v2/kuwaitlocal/router.js

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

lib/v2/wallstreetcn/live.js

@@ -1,5 +1,7 @@
 const got = require('@/utils/got');
 const { parseDate } = require('@/utils/parse-date');
+const { art } = require('@/utils/render');
+const path = require('path');
 
 const titles = {
     global: '要闻',
@@ -30,9 +32,13 @@ module.exports = async (ctx) => {
         .map((item) => ({
             link: item.uri,
             title: item.title || item.content_text,
-            description: item.content + (item.content_more ?? ''),
             pubDate: parseDate(item.display_time * 1000),
             author: item.author?.display_name ?? '',
+            description: art(path.join(__dirname, 'templates/description.art'), {
+                description: item.content,
+                more: item.content_more,
+                images: item.images,
+            }),
         }));
 
     ctx.state.data = {

lib/v2/wallstreetcn/templates/description.art

@@ -0,0 +1,11 @@
+{{ if description }}
+{{@ description }}
+{{ /if }}
+{{ if more }}
+{{@ more }}
+{{ /if }}
+{{ if images }}
+{{ each images image }}
+<img src="{{ image.uri }}" width="{{ image.width }}" height="{{ image.height }}">
+{{ /each }}
+{{ /if }}