rsshub 1.0.0-master.fcaf427 → 1.0.0-master.fcb92f0

package/lib/routes/universities/shmtu/www.js

@@ -1,61 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-const url = require('url');
-const host = 'http://www.shmtu.edu.cn';
-
-async function load(link) {
-    const response = await got.get(link);
-    const $ = cheerio.load(response.data);
-
-    const description = $('meta[name="description"]').attr('content');
-
-    return { description };
-}
-
-const ProcessFeed = (list, caches) =>
-    Promise.all(
-        list.map(async (item) => {
-            const $ = cheerio.load(item);
-
-            const itemUrl = url.resolve(host, $('.title', item).find('a').attr('href'));
-            const category = $('.department', item).text().trim();
-
-            const single = {
-                title: $('.title', item).text().trim(),
-                link: itemUrl,
-                guid: itemUrl,
-                pubDate: new Date($('.date', item).find('span').find('span').attr('content')).toUTCString(),
-                category,
-                author: category,
-            };
-
-            const other = await caches.tryGet(itemUrl, () => load(itemUrl));
-
-            return Promise.resolve(Object.assign({}, single, other));
-        })
-    );
-
-module.exports = async (ctx) => {
-    const type = ctx.params.type;
-    const info = type === 'notes' ? '通知公告' : '学术讲座';
-
-    const response = await got({
-        method: 'get',
-        url: host + `/${type}`,
-        headers: {
-            Referer: host,
-        },
-    });
-
-    const $ = cheerio.load(response.data);
-    const list = $('tr', 'tbody').slice(0, 10).get();
-
-    const result = await ProcessFeed(list, ctx.cache);
-
-    ctx.state.data = {
-        title: `上海海事大学 ${info}`,
-        link: host + `/${type}`,
-        description: '上海海事大学 官网信息',
-        item: result,
-    };
-};

package/lib/routes/universities/sustech/bidding.js

@@ -1,34 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-
-module.exports = async (ctx) => {
-    const response = await got({
-        method: 'get',
-        url: 'http://biddingoffice.sustech.edu.cn/',
-    });
-
-    const data = response.data;
-
-    const $ = cheerio.load(data);
-
-    const list = $('.index-wrap.index-2 ul li');
-
-    ctx.state.data = {
-        title: '南方科技大学采购与招标管理部',
-        link: 'http://biddingoffice.sustech.edu.cn/',
-        item:
-            list &&
-            list
-                .map((index, item) => {
-                    item = $(item);
-                    const itemPubdate = item.find('li > span').text();
-                    return {
-                        pubDate: itemPubdate,
-                        title: item.find('li > a').text(),
-                        description: item.find('li > a').text(),
-                        link: item.find('li > a').attr('href'),
-                    };
-                })
-                .get(),
-    };
-};

package/lib/routes/universities/sustech/newshub-zh.js

@@ -1,35 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-
-module.exports = async (ctx) => {
-    const response = await got({
-        method: 'get',
-        url: 'https://newshub.sustech.edu.cn/zh/news/',
-    });
-
-    const data = response.data;
-
-    const $ = cheerio.load(data);
-
-    const list = $('.m-newslist ul li');
-
-    ctx.state.data = {
-        title: '南方科技大学新闻网-中文',
-        link: 'https://newshub.sustech.edu.cn/zh/news/',
-        item:
-            list &&
-            list
-                .map((index, item) => {
-                    item = $(item);
-                    const itemPicUrl = String(item.find('a > div.u-pic').attr('style')).replace('background-image:url(', 'https://newshub.sustech.edu.cn/').replace(');', '');
-                    const itemPubdate = item.find('a > div.u-date').text();
-                    return {
-                        pubDate: itemPubdate,
-                        title: item.find('a > div.u-text > div.title.f-clamp').text(),
-                        description: `<img src="${itemPicUrl}" class="type:primaryImage" ><br>${item.find('a > div.u-text > div.details.f-clamp4 > p').text()}`,
-                        link: item.find('a').attr('href'),
-                    };
-                })
-                .get(),
-    };
-};

package/lib/routes/universities/sustyjs/sustyjs.js

@@ -1,24 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-
-module.exports = async (ctx) => {
-    const link = 'https://gs.sustech.edu.cn/tonggao/p/1';
-    const response = await got.get(link);
-    const $ = cheerio.load(response.data);
-    const list = $('.block02 ul li').slice(0, 10);
-
-    ctx.state.data = {
-        title: '南方科技大学研究生院',
-        link,
-        description: '南方科技大学研招网通知公告',
-        item:
-            list &&
-            list
-                .map((index, item) => {
-                    item = $(item);
-
-                    return { title: item.find('li a').text(), description: item.find('li a').text(), link: item.find('li a').attr('href') };
-                })
-                .get(),
-    };
-};

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

@@ -1,51 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-const util = require('./utils');
-
-const map = new Map([
-    [1, { title: '硕士招生 - 深圳大学研究生招生网' }],
-    [2, { title: '博士招生 - 深圳大学研究生招生网' }],
-]);
-
-module.exports = async (ctx) => {
-    let type = Number.parseInt(ctx.params.type);
-    const struct = {
-        1: {
-            selector: {
-                list: '.list',
-                item: 'li',
-                content: '#vsb_content',
-            },
-            url: 'https://yz.szu.edu.cn/sszs/gg.htm',
-        },
-        2: {
-            selector: {
-                list: '.list',
-                item: 'li',
-                content: '#vsb_content',
-            },
-            url: 'https://yz.szu.edu.cn/bszs/gg.htm',
-        },
-    };
-    if (type !== 1 && type !== 2) {
-        // fallback to default
-        type = 1;
-    }
-    const url = struct[type].url;
-
-    const response = await got.get(url);
-    const data = response.data;
-
-    const $ = cheerio.load(data);
-    const list = $(struct[type].selector.list).find(struct[type].selector.item).get();
-
-    const name = $('title').text();
-    const result = await util.ProcessFeed(list, ctx.cache, struct[type]);
-
-    ctx.state.data = {
-        title: map.get(type).title,
-        link: url,
-        description: name,
-        item: result,
-    };
-};

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

@@ -1,78 +0,0 @@
-const cheerio = require('cheerio');
-const got = require('@/utils/got');
-const url = require('url');
-
-const ProcessFeed = (list, cache, current) =>
-    Promise.all(
-        list
-            .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 ($('a').length > 0) {
-                    return true;
-                }
-                return false;
-
-                // return typeof ($('a').attr('href')) !== undefined;
-                // return false;
-            })
-            .map(async (item) => {
-                let $ = cheerio.load(item);
-                const $url = url.resolve(current.url, $('a').attr('href'));
-                const key = $url;
-                // 检查缓存中是否存在该页面
-
-                const value = await cache.get(key);
-                if (value) {
-                    // 查询返回未过期缓存
-                    return JSON.parse(value);
-                } else {
-                    // 加载新闻内容页面
-                    const response = await got.get($url);
-
-                    const data = response.data;
-                    $ = cheerio.load(data); // 使用 cheerio 加载返回的 HTML
-
-                    // 还原图片地址
-                    $(`${current.selector.content} img`).each((index, elem) => {
-                        const $elem = $(elem);
-                        const src = $elem.attr('src');
-                        if (src) {
-                            $elem.attr('src', url.resolve(current.url, src));
-                        }
-                    });
-
-                    // 还原链接地址
-                    $(`${current.selector.content} a`).each((index, elem) => {
-                        const $elem = $(elem);
-                        const src = $elem.attr('href');
-                        if (src) {
-                            $elem.attr('href', url.resolve(current.url, src));
-                        }
-                    });
-
-                    // 去除样式
-                    $('img, div, span, p, table, td, tr').removeAttr('style');
-                    $('style, script').remove();
-
-                    const title = $('h2').text();
-
-                    const single = {
-                        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
-                        author: '深圳大学研究生招生网',
-                        guid: $url, // 文章唯一标识
-                    };
-                    // 将内容写入缓存
-                    cache.set(key, JSON.stringify(single)); // 缓存时间为24h
-                    // 返回列表上提取到的信息
-                    return single;
-                }
-            })
-    );
-
-module.exports = {
-    ProcessFeed,
-};

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

@@ -1,24 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-
-module.exports = async (ctx) => {
-    const link = 'https://yz.szu.edu.cn/sszs/gg.htm';
-    const response = await got.get(link);
-    const $ = cheerio.load(response.data);
-    const list = $('.list li').slice(0, 10);
-
-    ctx.state.data = {
-        title: '深圳大学',
-        link,
-        description: '深圳大学研招网通知公告',
-        item:
-            list &&
-            list
-                .map((index, item) => {
-                    item = $(item);
-
-                    return { title: item.find('li A').text(), description: item.find('li A').text(), link: item.find('li A').attr('href') };
-                })
-                .get(),
-    };
-};

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

@@ -1,34 +0,0 @@
-const got = require('@/utils/got'); // get web content
-const cheerio = require('cheerio'); // html parser
-
-const domain = 'http://sse.tongji.edu.cn';
-
-module.exports = async function get_article(url) {
-    if (/^\/.*$/.test(url)) {
-        url = domain + url;
-    }
-    const response = await got({
-        method: 'get',
-        url,
-    });
-    const data = response.data;
-
-    const $ = cheerio.load(data);
-    const title = $('div.view-title').text();
-    const pub_date_raw = $('div.view-info').text();
-    const pub_date_parse = /\d{2,4}-\d{1,2}-\d{1,2}/.exec(pub_date_raw);
-    const pub_date = new Date(pub_date_parse[0]).toUTCString();
-    const content = $('div.view-cnt')
-        .html()
-        .replace(/<!\[CDATA\[/g, '<!--')
-        .replace(/\]\]>/g, '-->')
-        .replace(/href="\//g, 'href="' + domain + '/');
-
-    const item = {
-        title,
-        pubDate: pub_date,
-        link: url,
-        description: content,
-    };
-    return item;
-};

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

@@ -1,41 +0,0 @@
-// Warning: The author still knows nothing about javascript!
-
-// params:
-// type: notification type
-
-const got = require('@/utils/got'); // get web content
-const cheerio = require('cheerio'); // html parser
-const get_article = require('./_article');
-
-const base_url = 'http://sse.tongji.edu.cn/data/list/$type$';
-module.exports = async (ctx) => {
-    const type = ctx.params.type || 'xwdt';
-
-    const list_url = base_url.replace('$type$', type);
-    const response = await got({
-        method: 'get',
-        url: list_url,
-    });
-    const data = response.data; // content is html format
-    const $ = cheerio.load(data);
-
-    // get urls
-    const detail_urls = [];
-
-    const a = $('ul.data-list').find('a').slice(0, 10);
-
-    for (let i = 0; i < a.length; ++i) {
-        const tmp = $(a[i]).attr('href');
-        detail_urls.push(tmp);
-    }
-
-    // get articles
-    const article_list = await Promise.all(detail_urls.map((url) => get_article(url)));
-
-    // feed the data
-    ctx.state.data = {
-        title: '同济大学软件学院',
-        link: list_url,
-        item: article_list,
-    };
-};

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

@@ -1,23 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-
-module.exports = async (ctx) => {
-    const link = 'http://yz.tongji.edu.cn/zsxw/ggtz.htm';
-    const response = await got.get(link);
-    const $ = cheerio.load(response.data);
-    const list = $('.list_main_content li').slice(0, 10);
-
-    ctx.state.data = {
-        title: '同济大学研究生院',
-        link,
-        description: '同济大学研究生院通知公告',
-        item:
-            list &&
-            list
-                .map((index, item) => {
-                    item = $(item);
-                    return { title: item.find('li a').text(), description: item.find('li a').text(), link: item.find('li a').attr('href') };
-                })
-                .get(),
-    };
-};

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

@@ -1,27 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-
-module.exports = async (ctx) => {
-    const link = 'http://www.tynu.edu.cn/index/xyxw.htm';
-    const response = await got.get(link);
-    const $ = cheerio.load(response.data);
-
-    const list = $('.winstyle197529 a[class="c197529"]');
-
-    ctx.state.data = {
-        title: '太原师范学院校园新闻',
-        link,
-        item:
-            list &&
-            list
-                .map((index, item) => {
-                    item = $(item);
-                    return {
-                        title: item.text(),
-                        description: item.text(),
-                        link: item.attr('href'),
-                    };
-                })
-                .get(),
-    };
-};

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

@@ -1,113 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-
-// 获取含有列表的页面
-async function loadTypePage(link) {
-    const Response = await got({
-        method: 'get',
-        url: link,
-    });
-    const data = Response.data;
-    return { data };
-}
-// 同步获取含有列表的页面
-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, () => 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));
-        })
-    );
-
-// 完整文章页
-async function load(link) {
-    const response = await got.get(link);
-    const $ = cheerio.load(response.data);
-
-    // 将style去掉
-    const reg = /style\s*?=\s*?([‘"])[\s\S]*?\1/gi;
-    const description = $('div.v_news_content').html().replace(reg, '');
-
-    // 提取内容
-    return { description };
-}
-
-// 同步完整文章页
-const ProcessFeed = (list, caches) => {
-    const host = 'http://www.auto.uestc.edu.cn/';
-
-    return Promise.all(
-        list.map(async (item) => {
-            const $ = cheerio.load(item);
-            const title = $('h6 a').text();
-
-            // 还原相对链接为绝对链接
-            const itemUrl = $('h6 a').attr('href').replace('..', host);
-            const itemDate = new Date($('span').text().replace(/-/g, '/')).toUTCString();
-
-            // 列表上提取到的信息
-            const single = {
-                title,
-                link: itemUrl,
-                author: '成电自动化',
-                guid: itemUrl,
-                pubDate: itemDate,
-            };
-
-            // 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, () => 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));
-        })
-    );
-};
-
-// 通知公告 http://www.auto.uestc.edu.cn/index/tzgg1.htm
-// 学术看板 http://www.auto.uestc.edu.cn/index/xskb1.htm
-// 焦点新闻 http://www.auto.uestc.edu.cn/index/jdxw.htm
-// 综合新闻 http://www.auto.uestc.edu.cn/index/zhxw1.htm
-
-// 测试：http://localhost:1200/uestc/auto/tzgg1+xskb1
-module.exports = async (ctx) => {
-    const baseUrl = 'http://www.auto.uestc.edu.cn/';
-    const baseIndexUrl = 'http://www.auto.uestc.edu.cn/index/';
-
-    const type = ctx.params.type || 'tzgg1';
-    const allType = type.split('+');
-
-    const listLink = [];
-    for (const idx in allType) {
-        listLink.push(baseIndexUrl + allType[idx] + '.htm');
-    }
-    const Pages = await ProcessListPage(listLink, ctx.cache);
-
-    // 合并要处理的link
-    let headName = '自动化';
-    let list = null;
-    for (const idx in Pages) {
-        const $ = cheerio.load(Pages[idx].data);
-        headName = headName + '+' + $('title').text().split('-')[0];
-        if (list === null) {
-            list = $('dl.clearfix');
-        } else {
-            list = $('dl.clearfix').append(list);
-        }
-    }
-    list = list.get();
-
-    // 处理所有的页面
-    const result = await ProcessFeed(list, ctx.cache);
-
-    ctx.state.data = {
-        title: headName,
-        link: baseUrl,
-        description: '电子科技大学 - ' + headName,
-        item: result,
-    };
-};

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

@@ -1,40 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-
-const rootlink = 'http://cqe.uestc.edu.cn/';
-const mapUrl = {
-    hdyg: 'xwgg/hdyg.htm',
-    tz: 'xwgg/tzgg.htm',
-    kctz: 'kcjs/tzgg.htm',
-    lrxb: 'ldljy/tzgg.htm',
-};
-
-const mapTitle = {
-    hdyg: '成电文化素质教育中心-活动预告',
-    tz: '成电文化素质教育中心-通知',
-    kctz: '成电文化素质教育中心-课程通知',
-    lrxb: '立人班选拔',
-};
-
-module.exports = async (ctx) => {
-    const type = ctx.params.type || 'hdyg';
-    const response = await got({
-        method: 'get',
-        url: rootlink + mapUrl[type],
-    });
-
-    const data = response.data;
-    const $ = cheerio.load(data);
-
-    ctx.state.data = {
-        title: mapTitle[type],
-        link: rootlink + mapUrl[type],
-        item: $('div.Newslist>ul>li')
-            .map((_, item) => ({
-                title: $(item).find('a').attr('title'),
-                link: $(item).find('a').attr('href').replace('..', rootlink),
-                pubDate: $(item).find('span').text().substr(1, 10),
-            }))
-            .get(),
-    };
-};