rsshub 1.0.0-master.fc6953a → 1.0.0-master.fdba2d7

package/lib/routes/universities/szu/yz/utils.js

@@ -2,10 +2,10 @@ const cheerio = require('cheerio');
 const got = require('@/utils/got');
 const url = require('url');
 
-const ProcessFeed = async (list, cache, current) =>
-    await Promise.all(
+const ProcessFeed = (list, cache, current) =>
+    Promise.all(
         list
-            .filter(function (item) {
+            .filter((item) => {
                 // 如果不包含链接说明不是新闻item，如表头的tr
                 const $ = cheerio.load(item);
                 if ($('a').length > 0) {
@@ -58,7 +58,7 @@ const ProcessFeed = async (list, cache, current) =>
                     const title = $('h2').text();
 
                     const single = {
-                        title: title,
+                        title,
                         description: $(current.selector.content).html(),
                         link: $url,
                         pubDate: new Date($('div.ny_fbt').text().substr(6, 16)).toUTCString(), // 混有发表时间和点击量，取出时间

package/lib/routes/universities/szuyjs/szuyjs.js

@@ -9,7 +9,7 @@ module.exports = async (ctx) => {
 
     ctx.state.data = {
         title: '深圳大学',
-        link: link,
+        link,
         description: '深圳大学研招网通知公告',
         item:
             list &&

package/lib/routes/universities/thu/career.js

@@ -1,13 +1,32 @@
 const got = require('@/utils/got');
 const cheerio = require('cheerio');
-const resolve_url = require('url').resolve;
+const { parseDate } = require('@/utils/parse-date');
 
 const base_url = 'https://career.cic.tsinghua.edu.cn/';
+const route = 'xsglxt/f/jyxt/anony/xxfb';
 
 module.exports = async (ctx) => {
-    const page = 1;
+    const list = await fetchPage(1);
 
-    const route = 'xsglxt/f/jyxt/anony/xxfb';
+    const title = '招聘信息';
+    ctx.state.data = {
+        title: '清华大学 - ' + title,
+        link: base_url + route,
+        description: '清华大学学生职业发展指导中心 - ' + title,
+        item: await Promise.all(
+            list.map((item) =>
+                ctx.cache.tryGet(item.link, async () => {
+                    const res = await got.get(item.link);
+                    const content = cheerio.load(res.data);
+                    item.description = content('div.sideleft.left, div.content.teacher').html();
+                    return item;
+                })
+            )
+        ),
+    };
+};
+
+async function fetchPage(page) {
     const response = await got({
         method: 'post',
         url: base_url + route,
@@ -23,18 +42,16 @@ module.exports = async (ctx) => {
     });
 
     const $ = cheerio.load(response.data);
-
-    const title = '招聘信息';
-    ctx.state.data = {
-        title: '清华大学 - ' + title,
-        link: base_url,
-        description: '清华大学学生职业发展指导中心 - ' + title,
-        item: $('#ts_div .list li')
-            .map((_, elem) => ({
-                link: resolve_url(base_url, $('a', elem).attr('ahref')),
-                title: $('a', elem).text(),
-                pubDate: new Date($('span', elem).text()).toUTCString(),
-            }))
-            .get(),
-    };
-};
+    return $('#ts_div .list li')
+        .map((_, item) => {
+            item = $(item);
+            const a = item.find('a');
+            const date = parseDate(item.find('span').text());
+            return {
+                title: a.text(),
+                link: new URL(a.attr('ahref'), base_url),
+                pubDate: date,
+            };
+        })
+        .get();
+}

package/lib/routes/universities/thu/index.js

@@ -93,7 +93,7 @@ module.exports = async (ctx) => {
 
 function get_link(url, alter_url) {
     if (url !== undefined) {
-        return resolve_url(base_url, url);
+        return resolve_url(news_url, url);
     } else {
         return alter_url;
     }

package/lib/routes/universities/tju/sse/_article.js

@@ -9,7 +9,7 @@ module.exports = async function get_article(url) {
     }
     const response = await got({
         method: 'get',
-        url: url,
+        url,
     });
     const data = response.data;
 
@@ -25,7 +25,7 @@ module.exports = async function get_article(url) {
         .replace(/href="\//g, 'href="' + domain + '/');
 
     const item = {
-        title: title,
+        title,
         pubDate: pub_date,
         link: url,
         description: content,

package/lib/routes/universities/tjuyjs/tjuyjs.js

@@ -9,7 +9,7 @@ module.exports = async (ctx) => {
 
     ctx.state.data = {
         title: '同济大学研究生院',
-        link: link,
+        link,
         description: '同济大学研究生院通知公告',
         item:
             list &&

package/lib/routes/universities/tynu/tynu.js

@@ -10,7 +10,7 @@ module.exports = async (ctx) => {
 
     ctx.state.data = {
         title: '太原师范学院校园新闻',
-        link: link,
+        link,
         item:
             list &&
             list

package/lib/routes/universities/uestc/auto.js

@@ -11,13 +11,13 @@ async function loadTypePage(link) {
     return { data };
 }
 // 同步获取含有列表的页面
-const ProcessListPage = async (list, caches) =>
-    await Promise.all(
+const ProcessListPage = (list, caches) =>
+    Promise.all(
         list.map(async (link) => {
             const itemUrl = link;
             // 使用tryGet方法从缓存获取内容。
             // 当缓存中无法获取到链接内容的时候，则使用load方法加载文章内容。
-            const other = await caches.tryGet(itemUrl, async () => await loadTypePage(itemUrl));
+            const other = await caches.tryGet(itemUrl, () => loadTypePage(itemUrl));
             // 合并解析后的结果集作为该篇文章最终的输出结果
             return Promise.resolve(Object.assign({}, other));
         })
@@ -37,10 +37,10 @@ async function load(link) {
 }
 
 // 同步完整文章页
-const ProcessFeed = async (list, caches) => {
+const ProcessFeed = (list, caches) => {
     const host = 'http://www.auto.uestc.edu.cn/';
 
-    return await Promise.all(
+    return Promise.all(
         list.map(async (item) => {
             const $ = cheerio.load(item);
             const title = $('h6 a').text();
@@ -51,7 +51,7 @@ const ProcessFeed = async (list, caches) => {
 
             // 列表上提取到的信息
             const single = {
-                title: title,
+                title,
                 link: itemUrl,
                 author: '成电自动化',
                 guid: itemUrl,
@@ -60,7 +60,7 @@ const ProcessFeed = async (list, caches) => {
 
             // 使用tryGet方法从缓存获取内容。
             // 当缓存中无法获取到链接内容的时候，则使用load方法加载文章内容。
-            const other = await caches.tryGet(itemUrl, async () => await load(itemUrl));
+            const other = await caches.tryGet(itemUrl, () => load(itemUrl));
 
             // 合并解析后的结果集作为该篇文章最终的输出结果
             return Promise.resolve(Object.assign({}, single, other));

package/lib/routes/universities/uestc/cs.js

@@ -11,13 +11,13 @@ async function loadTypePage(link) {
     return { data };
 }
 // 同步获取含有列表的页面
-const ProcessListPage = async (list, caches) =>
-    await Promise.all(
+const ProcessListPage = (list, caches) =>
+    Promise.all(
         list.map(async (link) => {
             const itemUrl = link;
             // 使用tryGet方法从缓存获取内容。
             // 当缓存中无法获取到链接内容的时候，则使用load方法加载文章内容。
-            const other = await caches.tryGet(itemUrl, async () => await loadTypePage(itemUrl));
+            const other = await caches.tryGet(itemUrl, () => loadTypePage(itemUrl));
             // 合并解析后的结果集作为该篇文章最终的输出结果
             return Promise.resolve(Object.assign({}, other));
         })
@@ -36,10 +36,10 @@ async function load(link) {
     return { description };
 }
 
-const ProcessFeed = async (list, caches) => {
+const ProcessFeed = (list, caches) => {
     const host = 'http://www.scse.uestc.edu.cn/';
 
-    return await Promise.all(
+    return Promise.all(
         list.map(async (item) => {
             const $ = cheerio.load(item);
             const title = $('span a').text();
@@ -49,7 +49,7 @@ const ProcessFeed = async (list, caches) => {
 
             // 列表上提取到的信息
             const single = {
-                title: title,
+                title,
                 link: itemUrl,
                 author: '成电计算机',
                 guid: itemUrl,
@@ -58,7 +58,7 @@ const ProcessFeed = async (list, caches) => {
 
             // 使用tryGet方法从缓存获取内容。
             // 当缓存中无法获取到链接内容的时候，则使用load方法加载文章内容。
-            const other = await caches.tryGet(itemUrl, async () => await load(itemUrl));
+            const other = await caches.tryGet(itemUrl, () => load(itemUrl));
 
             // 合并解析后的结果集作为该篇文章最终的输出结果
             return Promise.resolve(Object.assign({}, single, other));

package/lib/routes/universities/uestc/gr.js

@@ -0,0 +1,46 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+
+module.exports = async (ctx) => {
+    const baseUrl = 'https://gr.uestc.edu.cn/tongzhi/';
+    const baseIndexUrl = 'https://gr.uestc.edu.cn';
+
+    const response = await got.get(baseIndexUrl);
+
+    const $ = cheerio.load(response.data);
+
+    const url_list = [];
+    $('[href^="/tongzhi/"]').each((_, item) => {
+        url_list.push(baseIndexUrl + item.attribs.href);
+    });
+
+    const out = await Promise.all(
+        url_list.map(async (news_url) => {
+            const news_detail = await ctx.cache.tryGet(news_url, async () => {
+                const result = await got.get(news_url);
+
+                const $ = cheerio.load(result.data);
+
+                const title = '[' + $('.over').text() + '] ' + $('div.title').text();
+                const author = $('.info').text().split('|')[1].trim().substring(3);
+                const date = new Date($('.info').text().split('|')[0].trim().substring(4)).toUTCString();
+                const description = $('.content').html();
+
+                return {
+                    title,
+                    link: news_url,
+                    author,
+                    pubDate: date,
+                    description,
+                };
+            });
+            return Promise.resolve(news_detail);
+        })
+    );
+
+    ctx.state.data = {
+        title: '电子科技大学研究生院通知公告',
+        link: baseUrl,
+        item: out,
+    };
+};

package/lib/routes/universities/uestc/is.js

@@ -0,0 +1,102 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+
+// 首页 https://www.is.uestc.edu.cn/
+// 通知公告
+// 最新公告无单独页面
+// 院办 https://www.is.uestc.edu.cn/noticesList.do?type=91
+// 组织 https://www.is.uestc.edu.cn/noticesList.do?type=132
+// 学生科 https://www.is.uestc.edu.cn/noticesList.do?type=90
+// 教务科 https://www.is.uestc.edu.cn/noticesList.do?type=93
+// 研管科 https://www.is.uestc.edu.cn/noticesList.do?type=92
+// 实验中心 https://www.is.uestc.edu.cn/noticesList.do?type=97
+// 企业技术服务中心 https://www.is.uestc.edu.cn/noticesList.do?type=185
+// 新工科中心 https://www.is.uestc.edu.cn/noticesList.do?type=184
+// 实习实训办公室 https://www.is.uestc.edu.cn/noticesList.do?type=96
+// 招聘 https://www.is.uestc.edu.cn/noticesList.do?type=137
+// 实习实训 https://www.is.uestc.edu.cn/noticesList.do?type=135
+
+const baseUrl = 'https://www.is.uestc.edu.cn/';
+const listUrl = baseUrl + 'noticesList.do?type=';
+
+const types = {
+    latest: 'latest',
+    yb: 91,
+    zx: 132,
+    xsk: 90,
+    jwk: 93,
+    ygk: 92,
+    syzx: 97,
+    qyjsfwzx: 185,
+    xgkzx: 184,
+    sxsxbgs: 96,
+    zp: 135,
+    sxsx: 137,
+};
+
+const convertDate = (text) => {
+    const date = new Date(text);
+    const now = new Date();
+    if (now.getMonth() > date.getMonth()) {
+        return new Date(`${new Date().getFullYear()}-${text}`);
+    } else {
+        if (now.getMonth() < date.getMonth() && now.getMonth() === date.getMonth() && now.getDate() < date.getDate()) {
+            return new Date(`${new Date().getFullYear() - 1}-${text}`);
+        }
+    }
+};
+
+const getTypeItems = async (type) => {
+    if (type !== 'latest') {
+        const res = await got({
+            method: 'get',
+            url: listUrl + types[type],
+            headers: {
+                Referer: baseUrl,
+            },
+        });
+
+        const $ = cheerio.load(res.data);
+
+        return $('.new-list li a')
+            .slice(0, 10)
+            .map((_, elem) => ({
+                link: baseUrl + elem.attribs.href,
+                title: elem.attribs.title,
+                pubDate: new Date($(elem).children('.new-time').text()).toUTCString(),
+            }))
+            .get();
+    } else {
+        const res = await got({
+            method: 'get',
+            url: baseUrl,
+            headers: {
+                Referer: baseUrl,
+            },
+        });
+
+        const $ = cheerio.load(res.data);
+
+        return $('.annce-cont ul li div')
+            .slice(0, 10)
+            .map((_, elem) => ({
+                link: baseUrl + $(elem).children('a').attr('href'),
+                title: $(elem).children('a').attr('title'),
+                pubDate: convertDate($(elem).children('span').text()),
+            }))
+            .get();
+    }
+};
+
+module.exports = async (ctx) => {
+    const type = ctx.params.type || 'latest';
+    const allTypes = type.split('+');
+
+    const result = await Promise.all(allTypes.map((t) => getTypeItems(t)));
+
+    ctx.state.data = {
+        title: '电子科技大学信软学院通知公告',
+        link: baseUrl,
+        item: result.flat(),
+    };
+};

package/lib/routes/universities/uestc/sice.js

@@ -0,0 +1,64 @@
+const cheerio = require('cheerio');
+
+module.exports = async (ctx) => {
+    const baseIndexUrl = 'https://www.sice.uestc.edu.cn/index.htm';
+    const host = 'https://www.sice.uestc.edu.cn/';
+    const browser = await require('@/utils/puppeteer')();
+    const page = await browser.newPage();
+
+    // 浏览器伪装
+    await page.evaluateOnNewDocument(() => {
+        // 在每个新页面打开前执行以下脚本
+        const newProto = navigator.__proto__;
+        delete newProto.webdriver; // 删除navigator.webdriver字段
+        navigator.__proto__ = newProto;
+        window.chrome = {}; // 添加window.chrome字段，为增加真实性还需向内部填充一些值
+        window.chrome.app = { InstallState: 'hehe', RunningState: 'haha', getDetails: 'xixi', getIsInstalled: 'ohno' };
+        window.chrome.csi = function () {};
+        window.chrome.loadTimes = function () {};
+        window.chrome.runtime = function () {};
+        Object.defineProperty(navigator, 'userAgent', {
+            // userAgent在无头模式下有headless字样，所以需覆写
+            get: () => 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_3) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/81.0.4044.113 Safari/537.36',
+        });
+        Object.defineProperty(navigator, 'plugins', {
+            // 伪装真实的插件信息
+            get: () => [{ description: 'Portable Document Format', filename: 'internal-pdf-viewer', length: 1, name: 'Chrome PDF Plugin' }],
+        });
+        Object.defineProperty(navigator, 'languages', {
+            // 添加语言
+            get: () => ['zh-CN', 'zh', 'en'],
+        });
+        const originalQuery = window.navigator.permissions.query; // notification伪装
+        window.navigator.permissions.query = (parameters) => (parameters.name === 'notifications' ? Promise.resolve({ state: Notification.permission }) : originalQuery(parameters));
+    });
+
+    await page.goto(baseIndexUrl, {
+        waitUntil: 'networkidle0',
+    });
+    const content = await page.content();
+    await browser.close();
+    const $ = cheerio.load(content);
+
+    const out = $('.notice p')
+        .map((index, item) => {
+            item = $(item);
+            let date = new Date(new Date().getFullYear() + '-' + item.find('a.date').text());
+            if (new Date() < date) {
+                date = new Date(new Date().getFullYear() - 1 + '-' + item.find('a.date').text());
+            }
+            return {
+                title: item.find('a[href]').text(),
+                link: host + item.find('a[href]').attr('href'),
+                pubDate: date,
+            };
+        })
+        .get();
+
+    ctx.state.data = {
+        title: '信通通知公告',
+        link: 'https://www.sice.uestc.edu.cn/tzgg/yb.htm',
+        description: '电子科技大学信息与通信工程学院通知公告',
+        item: out,
+    };
+};

package/lib/routes/universities/upc/jsj.js

@@ -59,9 +59,9 @@ module.exports = async (ctx) => {
             const detail_content = $('div.contain > div.main > div.right_cont > article').html();
             // ### 设置 RSS feed item
             const single = {
-                title: title,
+                title,
                 link: singleUrl,
-                // author: author,
+                // author,
                 description: detail_content,
                 // pubDate: updateDate,
             };

package/lib/routes/universities/upc/main.js

@@ -50,9 +50,9 @@ module.exports = async (ctx) => {
             const detail_content = $('div.container > div.main-ny > div.main-ny-cont > div.main-ny-cont-box > div.main-content-box > form').html();
             // ### 设置 RSS feed item
             const single = {
-                title: title,
+                title,
                 link: singleUrl,
-                // author: author,
+                // author,
                 description: detail_content,
                 // pubDate: updateDate,
             };

package/lib/routes/universities/upcyjs/upcyjs.js

@@ -9,7 +9,7 @@ module.exports = async (ctx) => {
 
     ctx.state.data = {
         title: '中国石油大学研究生院',
-        link: link,
+        link,
         description: '中国石油大学研究生院通知公告',
         item:
             list &&

package/lib/routes/universities/usst/jwc.js

@@ -5,7 +5,7 @@ module.exports = async (ctx) => {
     const url = `http://jwc.usst.edu.cn`;
     const response = await got({
         method: 'get',
-        url: url,
+        url,
     });
     const $ = cheerio.load(response.data);
     $('td[width="30px"]').remove();
@@ -40,9 +40,9 @@ module.exports = async (ctx) => {
             const detail_content = $('.wp_articlecontent').html();
             // ### 设置 RSS feed item
             const single = {
-                title: title,
+                title,
                 link: singleUrl,
-                // author: author,
+                // author,
                 description: detail_content,
                 // pubDate: updateDate,
             };