rsshub 1.0.0-master.fe69819 → 1.0.0-master.fe775f5

package/lib/v2/shmeea/self-study.js

@@ -0,0 +1,61 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+
+async function load_detail(list, cache) {
+    return await Promise.all(
+        list.map(async (item) => {
+            const notice_item = cheerio.load(item);
+            const href = notice_item('a').attr('href');
+            const url = 'http://www.shmeea.edu.cn' + href;
+            if (href[0] !== '/') {
+                return {
+                    title: notice_item('a').attr('title'),
+                    description: `<a href="${href}" >附件</a>`,
+                    link: href,
+                };
+            }
+            return await cache.tryGet(url, async () => {
+                const detail_response = await got({
+                    method: 'get',
+                    url,
+                    headers: {
+                        Referer: 'http://www.shmeea.edu.cn/page/04000/index.html',
+                        Host: 'www.shmeea.edu.cn',
+                    },
+                });
+                const detail = cheerio.load(detail_response.data);
+                return {
+                    title: notice_item('a').attr('title'),
+                    description: detail('.Article_content').html(),
+                    link: url,
+                    pubDate: timezone(parseDate(detail('.PBtime').text(), 'YYYY-MM-DD HH:mm:ss'), +8),
+                };
+            });
+        })
+    );
+}
+
+module.exports = async (ctx) => {
+    const response = await got({
+        method: 'get',
+        url: 'http://www.shmeea.edu.cn/page/04000/index.html',
+        headers: {
+            Host: 'www.shmeea.edu.cn',
+        },
+    });
+
+    const data = response.data;
+
+    const $ = cheerio.load(data);
+    const list = $('#main > div.container > div > div.span9 > div.page-he  > div  > ul > li').get();
+
+    const detail = await load_detail(list, ctx.cache);
+
+    ctx.state.data = {
+        title: '上海自学考试 - 通知公告',
+        link: 'http://www.shmeea.edu.cn/page/04000/index.html',
+        item: detail,
+    };
+};

package/lib/v2/sohu/maintainer.js

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

package/lib/{routes → v2}/sohu/mp.js

@@ -1,12 +1,12 @@
 const got = require('@/utils/got');
 const cheerio = require('cheerio');
-const date = require('@/utils/date');
+const { parseDate } = require('@/utils/parse-date');
 
 module.exports = async (ctx) => {
     const { id } = ctx.params;
+    const authorArticleAPI = `https://v2.sohu.com/author-page-api/author-articles/pc/${id}`;
 
-    const response = await got.get(`https://v2.sohu.com/author-page-api/author-articles/pc/${id}`);
-
+    const response = await got.get(authorArticleAPI);
     const list = response.data.data.pcArticleVOS.splice(0, 10);
     let author, link;
 
@@ -23,9 +23,10 @@ module.exports = async (ctx) => {
             const $ = cheerio.load(response);
 
             if (!author) {
-                const meta = $('#article-container h4 > a');
-                author = meta.text();
-                link = meta.attr('href').split('==')[0];
+                const meta = $('span[data-role="original-link"]');
+                author = meta.find('a').text();
+                // can't get author's link on server, so use the RSSHub link
+                // link = meta.attr('href').split('==')[0];
             }
 
             const article = $('#mp-editor');
@@ -38,11 +39,11 @@ module.exports = async (ctx) => {
                 title: e.title,
                 link: e.link,
                 description: article.html(),
-                pubDate: date(e.publicTimeStr, 8),
+                pubDate: parseDate(e.publicTime),
                 author,
             };
 
-            return Promise.resolve(single);
+            return single;
         })
     );
 

package/lib/v2/sohu/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'sohu.com': {
+        _name: '搜狐',
+        '.': [
+            {
+                title: '搜狐号',
+                docs: 'https://docs.rsshub.app/new-media.html#sou-hu-hao',
+                source: ['/a/:id'],
+                target: (params) => `/sohu/mp/${params.id.split('_')[1]}`,
+            },
+        ],
+    },
+};

package/lib/v2/sohu/router.js

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

package/lib/v2/soundofhope/channel.js

@@ -0,0 +1,42 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+
+const host = 'https://www.soundofhope.org';
+
+module.exports = async (ctx) => {
+    const channel = ctx.params.channel;
+    const id = ctx.params.id;
+    const url = `${host}/${channel}/${id}`;
+
+    const response = await got(url);
+    const $ = cheerio.load(response.data);
+    const title = $('div.left > nav').text().split('/').slice(1).join('');
+    const list = $('div.item')
+        .map((_, item) => ({
+            title: $(item).find('div.title').text(),
+            link: new URL($(item).find('a').attr('href'), host).href,
+        }))
+        .get();
+
+    const items = await Promise.all(
+        list.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const detailResponse = await got(item.link);
+                const content = cheerio.load(detailResponse.data);
+
+                item.description = content('div.Content__Wrapper-sc-1bvya0-0').html();
+                item.pubDate = timezone(parseDate(content('div.date').text(), 'YYYY.M.D HH:mm'), -8);
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: `希望之声 - ${title}`,
+        link: url,
+        item: items,
+    };
+};

package/lib/v2/soundofhope/maintainer.js

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

package/lib/v2/soundofhope/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'soundofhope.org': {
+        _name: '希望之声',
+        '.': [
+            {
+                title: '频道',
+                docs: 'https://docs.rsshub.app/traditional-media.html#xi-wang-zhi-sheng',
+                source: ['/:channel/:id'],
+                target: '/soundofhope/:channel/:id',
+            },
+        ],
+    },
+};

package/lib/v2/soundofhope/router.js

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

package/lib/v2/sputniknews/index.js

@@ -0,0 +1,93 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+
+const languages = {
+    english: 'https://sputniknews.com',
+    spanish: 'https://mundo.sputniknews.com',
+    german: 'https://snanews.de',
+    french: 'https://fr.sputniknews.com',
+    greek: 'https://sputniknews.gr',
+    italian: 'https://it.sputniknews.com',
+    czech: 'https://cz.sputniknews.com',
+    polish: 'https://pl.sputniknews.com',
+    serbian: 'https://rs.sputniknews.com',
+    latvian: 'https://sputniknewslv.com',
+    lithuanian: 'https://lt.sputniknews.com',
+    moldavian: 'https://md.sputniknews.com',
+    belarusian: 'https://bel.sputnik.by',
+    armenian: 'https://armeniasputnik.am',
+    abkhaz: 'https://sputnik-abkhazia.info',
+    ssetian: 'https://sputnik-ossetia.com',
+    georgian: 'https://sputnik-georgia.com',
+    azerbaijani: 'https://sputnik.az',
+    arabic: 'https://arabic.sputniknews.com',
+    turkish: 'https://tr.sputniknews.com',
+    persian: 'https://ir.sputniknews.com',
+    dari: 'https://af.sputniknews.com',
+    kazakh: 'https://sputniknews.kz',
+    kyrgyz: 'https://kg.sputniknews.com',
+    uzbek: 'https://oz.sputniknews-uz.com',
+    tajik: 'https://sputnik-tj.com',
+    vietnamese: 'https://vn.sputniknews.com',
+    japanese: 'https://jp.sputniknews.com',
+    chinese: 'http://sputniknews.cn',
+    portuguese: 'https://br.sputniknews.com',
+};
+
+module.exports = async (ctx) => {
+    const category = ctx.params.category ?? 'news';
+    const language = ctx.params.language ?? 'english';
+
+    const rootUrl = languages[language];
+    const currentUrl = `${rootUrl}/services/${category}/more.html`;
+
+    const response = await got({
+        method: 'get',
+        url: currentUrl,
+    });
+
+    const $ = cheerio.load(response.data);
+
+    let items = $('.list__title')
+        .toArray()
+        .map((item) => {
+            item = $(item);
+
+            return {
+                title: item.text(),
+                link: `${rootUrl}${item.attr('href')}`,
+            };
+        });
+
+    items = await Promise.all(
+        items.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const detailResponse = await got({
+                    method: 'get',
+                    url: item.link,
+                });
+
+                const content = cheerio.load(detailResponse.data);
+
+                item.category = content('.tag__text')
+                    .toArray()
+                    .map((tag) => content(tag).text());
+
+                content('.article__meta, .article__title, .article__info, .article__quote-bg, .article__google-news, .article__footer, .m-buy, .photoview__ext-link').remove();
+                content('div[data-type="article"]').remove();
+
+                item.description = content('.article').html();
+                item.pubDate = parseDate(content('div[itemprop="datePublished"]').attr('data-unixtime'));
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: `${category} - Sputnik News`,
+        link: `${rootUrl}/${category}`,
+        item: items,
+    };
+};

package/lib/v2/sputniknews/maintainer.js

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

package/lib/v2/sputniknews/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'sputniknews.cn': {
+        _name: '俄罗斯卫星通讯社',
+        '.': [
+            {
+                title: '分类',
+                docs: 'https://docs.rsshub.app/traditional-media.html#e-luo-si-wei-xing-tong-xun-she-fen-lei',
+                source: ['/:category', '/'],
+                target: '/sputniknews/:category/:language',
+            },
+        ],
+    },
+};

package/lib/v2/sputniknews/router.js

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

package/lib/{routes/universities/sysu/sdcs.js → v2/sysu/cse.js}

@@ -4,9 +4,9 @@ const cheerio = require('cheerio');
 module.exports = async (ctx) => {
     const response = await got({
         method: 'get',
-        url: 'http://sdcs.sysu.edu.cn/',
+        url: 'http://cse.sysu.edu.cn/',
         headers: {
-            Referer: `http://sdcs.sysu.edu.cn/`,
+            Referer: `http://cse.sysu.edu.cn/`,
         },
     });
     const $ = cheerio.load(response.data);
@@ -95,7 +95,7 @@ module.exports = async (ctx) => {
 
     ctx.state.data = {
         title: `中山大学 - 数据科学与计算机学院`,
-        link: `http://sdcs.sysu.edu.cn`,
+        link: `http://cse.sysu.edu.cn`,
         description: `中山大学 - 数据科学与计算机学院`,
         language: `zh-cn`,
         item: item_data,

package/lib/v2/sysu/maintainer.js

@@ -0,0 +1,3 @@
+module.exports = {
+    '/sysu/cse': ['Neutrino3316', 'MegrezZhu', 'nczitzk'],
+};

package/lib/v2/sysu/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'sysu.edu.cn': {
+        _name: '中山大学',
+        cse: [
+            {
+                title: '计算机学院（软件学院）',
+                docs: 'https://docs.rsshub.app/universities.html#zhong-shan-da-xue-ji-suan-ji-xue-yuan',
+                source: ['/'],
+                target: '/sysu/cse',
+            },
+        ],
+    },
+};

package/lib/v2/sysu/router.js

@@ -0,0 +1,3 @@
+module.exports = function (router) {
+    router.get('/cse', require('./cse'));
+};