rsshub 1.0.0-master.fdf137b → 1.0.0-master.fe775f5

package/lib/v2/sicau/router.js

@@ -1,3 +1,5 @@
 module.exports = function (router) {
+    router.get('/dky/:category?', require('./dky'));
+    router.get('/yan/:category?', require('./yan'));
     router.get('/zsjy/:category?', require('./zsjy'));
 };

package/lib/v2/sicau/yan.js

@@ -0,0 +1,53 @@
+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 category = ctx.params.category ?? 'xwgg';
+
+    const rootUrl = 'https://yan.sicau.edu.cn';
+    const currentUrl = `${rootUrl}/index/${category}.htm`;
+    const response = await got({
+        method: 'get',
+        url: currentUrl,
+    });
+
+    const $ = cheerio.load(response.data);
+
+    const list = $('.list-4 a[title]')
+        .slice(0, 10)
+        .map((_, item) => {
+            item = $(item);
+
+            return {
+                title: item.text(),
+                link: `${rootUrl}${item.attr('href').replace(/\.\./, '/')}`,
+            };
+        })
+        .get();
+
+    const items = await Promise.all(
+        list.map(
+            async (item) =>
+                await ctx.cache.tryGet(item.link, async () => {
+                    const detailResponse = await got({
+                        method: 'get',
+                        url: item.link,
+                    });
+                    const content = cheerio.load(detailResponse.data);
+
+                    item.description = content('.v_news_content').html();
+                    item.pubDate = timezone(parseDate(detailResponse.data.match(/发布时间: (\d{4}-\d{2}-\d{2} \d{2}:\d{2}:\d{2})/)[1], 'YYYY-MM-DD HH:mm:ss'), +8);
+
+                    return item;
+                })
+        )
+    );
+
+    ctx.state.data = {
+        title: $('title').text(),
+        link: currentUrl,
+        item: items,
+    };
+};

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/v2/stcn/index.js

@@ -0,0 +1,71 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const timezone = require('@/utils/timezone');
+const { parseDate } = require('@/utils/parse-date');
+
+const shortcuts = {
+    'news/news': 'xw/news',
+    'news/gd': 'gd',
+    'news/sd': 'xw/sd',
+    'news/pl': 'xw/pl',
+    'data/data': 'data',
+};
+
+module.exports = async (ctx) => {
+    let path = ctx.path.replace(/\//, '');
+    if (shortcuts.hasOwnProperty(path)) {
+        path = shortcuts[path];
+    }
+
+    const rootUrl = 'https://www.stcn.com';
+    const currentUrl = `${rootUrl}/${path || 'xw/news'}`;
+
+    const response = await got({
+        method: 'get',
+        url: currentUrl,
+    });
+
+    const match = response.data.match(/var curr_channel_id = "(light_\d+)";/);
+    const $ = cheerio.load(response.data);
+
+    $('.website, .websit').remove();
+
+    let items = $('.box_left, .box_left2, .left2')
+        .find('a[title]')
+        .slice(0, ctx.query.limit ? parseInt(ctx.query.limit) : 20)
+        .toArray()
+        .map((item) => {
+            item = $(item);
+
+            return {
+                title: item.text(),
+                link: item.attr('href').replace(/^\.\./, currentUrl.split('/').slice(0, -1).join('/')).replace(/^\./, currentUrl),
+            };
+        });
+
+    items = await Promise.all(
+        items.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const detailResponse = await got({
+                    method: 'get',
+                    url: item.link,
+                });
+
+                const content = cheerio.load(detailResponse.data);
+
+                item.description = content('.txt_con').html();
+                item.author = content('meta[name="author"]').attr('content');
+                item.category = content('meta[name="keywords"]').attr('content').split(';');
+                item.pubDate = timezone(parseDate(detailResponse.data.match(/publishdate = '(\d{4}-\d{2}-\d{2} \d{2}:\d{2}:\d{2})';/)[1]), +8);
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: `${(match ? $('#' + match[1]).text() : '') || $('.cur').text()} - 证券时报网`,
+        link: currentUrl,
+        item: items,
+    };
+};

package/lib/v2/stcn/maintainer.js

@@ -0,0 +1,6 @@
+module.exports = {
+    '/data/:id?': ['nczitzk'],
+    '/kuaixun/:id?': ['nczitzk'],
+    '/stock/:id?': ['nczitzk'],
+    '/:id?': ['nczitzk'],
+};

package/lib/v2/stcn/radar.js

@@ -0,0 +1,31 @@
+module.exports = {
+    'stcn.com': {
+        _name: '证券时报网',
+        '.': [
+            {
+                title: '要闻',
+                docs: 'https://docs.rsshub.app/finance.html#zheng-quan-shi-bao-wang-yao-wen',
+                source: ['/xw/:id', '/gd', '/'],
+                target: (params, url) => `/stcn${new URL(url).pathname}`,
+            },
+            {
+                title: '快讯',
+                docs: 'https://docs.rsshub.app/finance.html#zheng-quan-shi-bao-wang-kuai-xun',
+                source: ['/kuaixun/:id', '/kuaixun', '/'],
+                target: (params, url) => `/stcn${new URL(url).pathname}`,
+            },
+            {
+                title: '股市',
+                docs: 'https://docs.rsshub.app/finance.html#zheng-quan-shi-bao-wang-gu-shi',
+                source: ['/stock/:id', '/stock', '/'],
+                target: (params, url) => `/stcn${new URL(url).pathname}`,
+            },
+            {
+                title: '数据',
+                docs: 'https://docs.rsshub.app/finance.html#zheng-quan-shi-bao-wang-shu-ju',
+                source: ['/data/:id', '/data', '/'],
+                target: (params, url) => `/stcn${new URL(url).pathname}`,
+            },
+        ],
+    },
+};

package/lib/v2/stcn/router.js

@@ -0,0 +1,3 @@
+module.exports = function (router) {
+    router.get(/([\w\d/-]+)?/, require('./index'));
+};

package/lib/v2/swpu/bgw.js

@@ -0,0 +1,58 @@
+const { joinUrl } = require('./utils');
+const { parseDate } = require('@/utils/parse-date');
+const cheerio = require('cheerio');
+const got = require('@/utils/got');
+const timezone = require('@/utils/timezone');
+
+module.exports = async (ctx) => {
+    const url = `https://www.swpu.edu.cn/bgw2/${ctx.params.code}.htm`;
+
+    const res = await got.get(url);
+    const $ = cheerio.load(res.data);
+
+    const title = $('.title').text();
+
+    // 获取标题、时间及链接
+    const items = [];
+    $('.notice > ul > li > a').each((i, elem) => {
+        items.push({
+            title: $(elem.children[0]).text(),
+            pubDate: timezone(parseDate($(elem.children[1]).text()), +8),
+            link: joinUrl('https://www.swpu.edu.cn', $(elem).attr('href')), // 实际获得连接 "../info/1312/17891.htm"
+        });
+    });
+
+    // 请求全文
+    const out = await Promise.all(
+        items.map(async (item) => {
+            const $ = await ctx.cache.tryGet(item.link, async () => {
+                const res = await got.get(item.link);
+                return cheerio.load(res.data);
+            });
+
+            if ($('title').text().startsWith('系统提示')) {
+                item.author = '系统';
+                item.description = '无权访问';
+            } else {
+                item.author = '办公网';
+                item.description = $('.v_news_content').html();
+                for (const elem of $('.v_news_content p')) {
+                    if ($(elem).css('text-align') === 'right') {
+                        item.author = $(elem).text();
+                        break;
+                    }
+                }
+            }
+
+            return item;
+        })
+    );
+
+    ctx.state.data = {
+        title: `西南石油大学办公网 ${title}`,
+        link: url,
+        description: `西南石油大学办公网 ${title} 列表`,
+        language: 'zh-CN',
+        item: out,
+    };
+};

package/lib/v2/swpu/dean.js

@@ -0,0 +1,59 @@
+const { joinUrl } = require('./utils');
+const { parseDate } = require('@/utils/parse-date');
+const cheerio = require('cheerio');
+const got = require('@/utils/got');
+const timezone = require('@/utils/timezone');
+
+module.exports = async (ctx) => {
+    const url = `https://www.swpu.edu.cn/dean/${ctx.params.code}.htm`;
+
+    const res = await got.get(url);
+    const $ = cheerio.load(res.data);
+
+    let title = $('.r_list > h3').text();
+    title = title.substring(title.indexOf('：') + 1);
+
+    // 获取标题、时间及链接
+    const items = [];
+    $('.r_list > ul > li').each((i, elem) => {
+        items.push({
+            title: $('label:eq(0)', elem).text().trim(),
+            link: joinUrl('https://www.swpu.edu.cn/dean/', $('a', elem).attr('href')),
+        });
+    });
+
+    // 请求全文
+    const out = await Promise.all(
+        items.map(async (item) => {
+            const $ = await ctx.cache.tryGet(item.link, async () => {
+                const res = await got.get(item.link);
+                return cheerio.load(res.data);
+            });
+
+            if ($('title').text().startsWith('系统提示')) {
+                item.author = '系统';
+                item.description = '无权访问';
+            } else {
+                item.author = '教务处';
+                item.description = $('.v_news_content').html();
+                item.pubDate = timezone(parseDate($('#lbDate').text(), '更新时间：YYYY年MM月DD日'), +8);
+                for (const elem of $('.v_news_content p')) {
+                    if ($(elem).css('text-align') === 'right') {
+                        item.author = $(elem).text();
+                        break;
+                    }
+                }
+            }
+
+            return item;
+        })
+    );
+
+    ctx.state.data = {
+        title: `西南石油大学教务处 ${title}`,
+        link: url,
+        description: `西南石油大学教务处 ${title}`,
+        language: 'zh-CN',
+        item: out,
+    };
+};

package/lib/v2/swpu/dxy.js

@@ -0,0 +1,61 @@
+const { joinUrl } = require('./utils');
+const { parseDate } = require('@/utils/parse-date');
+const cheerio = require('cheerio');
+const got = require('@/utils/got');
+const timezone = require('@/utils/timezone');
+
+module.exports = async (ctx) => {
+    // 移除 urltype=tree.TreeTempUrl 虽然也能顺利访问页面，
+    // 但标题会缺失，而且在其他地方定位提取标题也比较麻烦。
+    const url = `https://www.swpu.edu.cn/dxy/list1.jsp?urltype=tree.TreeTempUrl&wbtreeid=${ctx.params.code}`;
+
+    const res = await got.get(url);
+    const $ = cheerio.load(res.data);
+
+    let title = $('title').text();
+    title = title.substring(0, title.indexOf('-'));
+
+    // 获取标题、时间及链接
+    const items = [];
+    $('tr[height="20"]').each((i, elem) => {
+        items.push({
+            title: $('a[title]', elem).text().trim(),
+            pubDate: timezone(parseDate($('td:eq(1)', elem).text(), 'YYYY年MM月DD日'), +8),
+            link: joinUrl('https://www.swpu.edu.cn/dxy/', $('a[title]', elem).attr('href')),
+        });
+    });
+
+    // 请求全文
+    const out = await Promise.all(
+        items.map(async (item) => {
+            const $ = await ctx.cache.tryGet(item.link, async () => {
+                const res = await got.get(item.link);
+                return cheerio.load(res.data);
+            });
+
+            if ($('title').text().startsWith('系统提示')) {
+                item.author = '系统';
+                item.description = '无权访问';
+            } else {
+                item.author = '电气信息学院';
+                item.description = $('.v_news_content').html();
+                for (const elem of $('.v_news_content p')) {
+                    if ($(elem).css('text-align') === 'right') {
+                        item.author = $(elem).text();
+                        break;
+                    }
+                }
+            }
+
+            return item;
+        })
+    );
+
+    ctx.state.data = {
+        title: `西南石油大学电气信息学院 ${title}`,
+        link: url,
+        description: `西南石油大学电气信息学院 ${title}`,
+        language: 'zh-CN',
+        item: out,
+    };
+};

package/lib/v2/swpu/maintainer.js

@@ -0,0 +1,6 @@
+module.exports = {
+    '/bgw/:code': ['CYTMWIA'],
+    '/dean/:code': ['CYTMWIA'],
+    '/dxy/:code': ['CYTMWIA'],
+    '/scs/:code': ['CYTMWIA'],
+};

package/lib/v2/swpu/radar.js

@@ -0,0 +1,31 @@
+module.exports = {
+    'swpu.edu.cn': {
+        _name: '西南石油大学',
+        '.': [
+            {
+                title: '办公网',
+                docs: 'https://docs.rsshub.app/university.html#xi-nan-shi-you-da-xue',
+                source: ['/'],
+                target: '',
+            },
+            {
+                title: '教务处',
+                docs: 'https://docs.rsshub.app/university.html#xi-nan-shi-you-da-xue',
+                source: ['/'],
+                target: '',
+            },
+            {
+                title: '计算机科学学院',
+                docs: 'https://docs.rsshub.app/university.html#xi-nan-shi-you-da-xue',
+                source: ['/'],
+                target: '',
+            },
+            {
+                title: '电气信息学院',
+                docs: 'https://docs.rsshub.app/university.html#xi-nan-shi-you-da-xue',
+                source: ['/'],
+                target: '',
+            },
+        ],
+    },
+};

package/lib/v2/swpu/router.js

@@ -0,0 +1,6 @@
+module.exports = function (router) {
+    router.get('/bgw/:code', require('./bgw'));
+    router.get('/dean/:code', require('./dean'));
+    router.get('/dxy/:code', require('./dxy'));
+    router.get('/scs/:code', require('./scs'));
+};