feat: add support for BUPT portal (#3290)

docs/university.md

@@ -104,6 +104,16 @@ pageClass: routes
 
 <Route author="RicardoMing" example="/bupt/grs" path="/bupt/grs" />
 
+### 信息门户
+
+<Route author="RicardoMing" example="/bupt/portal" path="/bupt/portal" />
+
+::: warning 注意
+信息门户的通知需要通过统一身份认证后才能获取，因此需要在校园网或校园 VPN 环境下自建。
+
+设置环境变量: `BUPT_USERNAME` 用户名为学号， `BUPT_PASSWORD` 统一身份认证的密码。
+:::
+
 ## 常州大学
 
 ### 教务处

lib/config.js

@@ -98,6 +98,10 @@ const calculateValue = () => {
         chuiniu: {
             member: envs.CHUINIU_MEMBER,
         },
+        bupt: {
+            username: envs.BUPT_USERNAME,
+            password: envs.BUPT_PASSWORD,
+        },
     };
 };
 calculateValue();

lib/router.js

@@ -1309,6 +1309,7 @@ router.get('/21caijing/channel/:name', require('./routes/21caijing/channel'));
 // 北京邮电大学
 router.get('/bupt/yz/:type', require('./routes/universities/bupt/yz'));
 router.get('/bupt/grs', require('./routes/universities/bupt/grs'));
+router.get('/bupt/portal', require('./routes/universities/bupt/portal'));
 
 // VOCUS 方格子
 router.get('/vocus/publication/:id', require('./routes/vocus/publication'));

lib/routes/universities/bupt/portal.js

@@ -0,0 +1,172 @@
+const config = require('@/config').value;
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const url = require('url');
+
+function isToday(time) {
+    return new Date().getTime() - new Date(time).getTime() < 86400000;
+}
+
+function getPubDate(time) {
+    return isToday(time) ? new Date() : new Date(time + ' 08:00:00');
+}
+
+const base = 'http://my.bupt.edu.cn';
+const sourceTimezoneOffset = -8;
+
+let portalCookie = null;
+let authCookie = null;
+let castgc = null;
+
+module.exports = async (ctx) => {
+    if (!config.bupt || !config.bupt.username || !config.bupt.password) {
+        throw 'BUPT Portal RSS is disabled due to the lack of <a href="https://docs.rsshub.app/university.html#bei-jing-you-dian-da-xue">relevant config</a>';
+    }
+
+    const reqUrl = url.resolve(base, '/index.portal?.pn=p1778');
+    let reqRes = await got({
+        method: 'get',
+        followRedirect: false,
+        url: reqUrl,
+        headers: {
+            cookie: portalCookie,
+        },
+    });
+
+    // Login
+    if (reqRes.statusCode === 302) {
+        if (reqRes.headers['set-cookie'] === undefined) {
+            portalCookie = null;
+            throw 'Can not obtain portalCookie';
+        }
+        portalCookie = reqRes.headers['set-cookie'].toString().match(/JSESSIONID=.{37}/)[0];
+
+        const authRes = await got({
+            method: 'get',
+            followRedirect: false,
+            url: reqRes.headers.location,
+            headers: {
+                cookie: `${authCookie}; ${castgc}`,
+            },
+        });
+        if (authRes.statusCode === 200) {
+            authCookie = authRes.headers['set-cookie'].toString().match(/JSESSIONID=.{37}/)[0];
+            const authPage = cheerio.load(authRes.data);
+
+            const loginRes = await got({
+                method: 'post',
+                followRedirect: false,
+                url: reqRes.headers.location,
+                headers: {
+                    cookie: authCookie,
+                    referer: reqRes.headers.location,
+                },
+                form: true,
+                data: {
+                    username: config.bupt.username,
+                    password: config.bupt.password,
+                    lt: authPage('[name=lt]').attr('value'),
+                    execution: authPage('[name=execution]').attr('value'),
+                    _eventId: authPage('[name=_eventId]').attr('value'),
+                    rmShown: 1,
+                },
+            });
+            if (loginRes.headers['set-cookie'] === undefined) {
+                authCookie = null;
+                castgc = null;
+                throw 'Can not obtain castgc';
+            }
+            castgc = loginRes.headers['set-cookie'].toString().match(/CASTGC=.{85}/)[0];
+
+            await got({
+                method: 'get',
+                followRedirect: false,
+                url: loginRes.headers.location,
+                headers: {
+                    cookie: portalCookie,
+                },
+            });
+        } else if (authRes.statusCode === 302) {
+            await got({
+                method: 'get',
+                followRedirect: false,
+                url: authRes.headers.location,
+                headers: {
+                    cookie: portalCookie,
+                },
+            });
+        } else {
+            throw 'BUPT auth login failed';
+        }
+
+        reqRes = await got({
+            method: 'get',
+            followRedirect: false,
+            url: reqUrl,
+            headers: {
+                cookie: portalCookie,
+            },
+        });
+    }
+
+    const $ = cheerio.load(reqRes.data);
+    const list = $('.newslist li').get();
+    const out = await Promise.all(
+        list.map(async (i) => {
+            const item = $(i);
+            const itemUrl = url.resolve(
+                base,
+                $(item)
+                    .find('a')
+                    .attr('href')
+            );
+            const cache = await ctx.cache.get(itemUrl);
+            if (cache) {
+                return Promise.resolve(JSON.parse(cache));
+            }
+
+            const title = $(item)
+                .find('a')
+                .text();
+            const author = $(item)
+                .find('.author')
+                .text();
+            const time = getPubDate(
+                $(item)
+                    .find('.time')
+                    .text()
+            );
+            time.setTime(time.getTime() + (sourceTimezoneOffset - time.getTimezoneOffset() / 60) * 60 * 60 * 1000);
+
+            const itemResponse = await got({
+                method: 'get',
+                url: itemUrl,
+                headers: {
+                    cookie: portalCookie,
+                },
+            });
+            const itemElement = cheerio.load(itemResponse.data);
+            // Remove useless print button
+            itemElement('table.Noprint').remove();
+            const description = itemElement('.singleexpert').html();
+
+            const single = {
+                title: title,
+                author: author,
+                description: description,
+                pubDate: time.toUTCString(),
+                link: itemUrl,
+                guid: itemUrl,
+            };
+
+            ctx.cache.set(itemUrl, JSON.stringify(single));
+            return Promise.resolve(single);
+        })
+    );
+
+    ctx.state.data = {
+        title: '北京邮电大学信息门户',
+        link: reqUrl,
+        item: out,
+    };
+};