feat(route): fix NUAA 教务处/计算机科学与技术学院信息 (#9361)

* feat(route): fix NUAA 教务处/计算机科学与技术学院信息

* refactor: migrate to v2

* fix: deprecated url.resolve

docs/university.md

@@ -1467,23 +1467,23 @@ jsjxy.hbut.edu.cn 证书链不全，自建 RSSHub 可设置环境变量 NODE_TLS
 
 ## 南京航空航天大学
 
-### 教务通知
+### 教务处
 
-<Route author="arcosx Seiry" example="/nuaa/jwc/default" path="/nuaa/jwc/:type" :paramsDesc="['分类名']">
+<Route author="arcosx Seiry qrzbing" example="/nuaa/jwc/tzgg" path="/nuaa/jwc/:type/:getDescription?" :paramsDesc="['分类名', '是否获取描述']">
 
-| 教学服务          | 教学建设 | 学生培养 | 教学资源 |
-| ------------- | ---- | ---- | ---- |
-| jxfw(default) | jxjs | xspy | jxzy |
+| 通知公告 | 教学服务 | 教学建设 | 学生培养 | 教学资源 |
+| ---- | ---- | ---- | ---- | ---- |
+| tzgg | jxfw | jxjs | xspy | jxzy |
 
 </Route>
 
 ### 计算机科学与技术学院
 
-<Route author="LogicJake Seiry" example="/nuaa/cs/kydt" path="/nuaa/cs/:type?" :paramsDesc="['分类名']"/>
+<Route author="LogicJake Seiry qrzbing" example="/nuaa/cs/jxdt" path="/nuaa/cs/:type/:getDescription?" :paramsDesc="['分类名', '是否获取描述']">
 
-| 通知公告 | 新闻动态 | 科研动态 | 教学动态 | 学生工作 | 招生信息 | 就业信息 |
-| ---- | ---- | ---- | ---- | ---- | ---- | ---- |
-| tzgg | xwdt | kydt | jxdt | xsgz | zsxx | jyxx |
+| 通知公告 | 热点新闻 | 学科科研 | 教学动态 | 本科生培养 | 研究生培养 | 学生工作 |
+| ---- | ---- | ---- | ---- | ----- | ----- | ---- |
+| tzgg | rdxw | xkky | jxdt | be    | me    | xsgz |
 
 </Route>
 

lib/router.js

@@ -519,10 +519,10 @@ router.get('/seu/cse/:type?', lazyloadRouteHandler('./routes/universities/seu/cs
 // 南京工业大学
 router.get('/njtech/jwc', lazyloadRouteHandler('./routes/universities/njtech/jwc'));
 
-// 南京航空航天大学
-router.get('/nuaa/jwc/:type?', lazyloadRouteHandler('./routes/universities/nuaa/jwc/jwc'));
-router.get('/nuaa/cs/:type?', lazyloadRouteHandler('./routes/universities/nuaa/cs/index'));
-router.get('/nuaa/yjsy/:type?', lazyloadRouteHandler('./routes/universities/nuaa/yjsy/yjsy'));
+// 南京航空航天大学 migrated to v2
+// router.get('/nuaa/jwc/:type/:getDescription?', lazyloadRouteHandler('./routes/universities/nuaa/jwc/jwc'));
+// router.get('/nuaa/cs/:type/:getDescription?', lazyloadRouteHandler('./routes/universities/nuaa/cs/index'));
+// router.get('/nuaa/yjsy/:type?', lazyloadRouteHandler('./routes/universities/nuaa/yjsy/yjsy'));
 
 // 河海大学
 router.get('/hhu/libNews', lazyloadRouteHandler('./routes/universities/hhu/libNews'));

lib/routes/universities/nuaa/cs/index.js

@@ -1,84 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-const url = require('url');
-const getCookie = require('../utils/pypasswaf');
-const host = 'http://cs.nuaa.edu.cn/';
-
-const map = new Map([
-    ['tzgg', { title: '南京航空航天大学计算机科学与技术学院 -- 通知公告', suffix: '1995/list.htm' }],
-    ['xwdt', { title: '南京航空航天大学计算机科学与技术学院 -- 新闻动态', suffix: '1997/list.htm' }],
-    ['kydt', { title: '南京航空航天大学计算机科学与技术学院 -- 科研动态', suffix: '1975/list.htm' }],
-    ['jxdt', { title: '南京航空航天大学计算机科学与技术学院 -- 教学动态', suffix: '1977/list.htm' }],
-    ['xsgz', { title: '南京航空航天大学计算机科学与技术学院 -- 学生工作', suffix: '1959/list.htm' }],
-    ['zsxx', { title: '南京航空航天大学计算机科学与技术学院 -- 招生信息', suffix: '1993/list.htm' }],
-    ['jyxx', { title: '南京航空航天大学计算机科学与技术学院 -- 就业信息', suffix: '1994/list.htm' }],
-]);
-
-module.exports = async (ctx) => {
-    const type = ctx.params.type || 'tzgg';
-    const suffix = map.get(type).suffix;
-
-    const link = url.resolve(host, suffix);
-    const cookie = await getCookie();
-    const gotConfig = {
-        headers: {
-            cookie,
-        },
-    };
-    const response = await got.get(link, gotConfig);
-    const $ = cheerio.load(response.data);
-
-    const list = $('#news_list > ul > li')
-        .slice(0, 10)
-        .map(function () {
-            const info = {
-                title: $(this).find('a').attr('title'),
-                link: $(this).find('a').attr('href'),
-                date: $(this).find('span').text(),
-            };
-            return info;
-        })
-        .get();
-
-    const out = await Promise.all(
-        list.map(async (info) => {
-            const title = info.title || 'tzgg';
-            const date = info.date;
-            const itemUrl = url.resolve(host, info.link);
-
-            const cache = await ctx.cache.get(itemUrl);
-            if (cache) {
-                return Promise.resolve(JSON.parse(cache));
-            }
-
-            const arr = itemUrl.split('.');
-            const pageType = arr[arr.length - 1];
-
-            let description = itemUrl;
-            if (pageType === 'htm' || pageType === 'html') {
-                const response = await got.get(itemUrl, gotConfig);
-                const $ = cheerio.load(response.data);
-                description = $('.wp_articlecontent')
-                    .html()
-                    .replace(/src="\//g, `src="${url.resolve(host, '.')}`)
-                    .trim();
-            }
-
-            const single = {
-                title,
-                link: itemUrl,
-                description,
-                pubDate: new Date(date).toUTCString(),
-            };
-            ctx.cache.set(itemUrl, JSON.stringify(single));
-            return Promise.resolve(single);
-        })
-    );
-
-    ctx.state.data = {
-        title: map.get(type).title,
-        link,
-        description: '南京航空航天大学计算机科学与技术学院RSS',
-        item: out,
-    };
-};

lib/routes/universities/nuaa/jwc/jwc.js

@@ -1,105 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-const url = require('url');
-const host = 'http://aao.nuaa.edu.cn/';
-// https is not allowed in nuaa;interesting ha
-
-const map = {
-    default: 8230,
-    jxfw: 8230, // 教学服务
-    xspy: 8231, // 学生培养
-    jxjs: 8232, // 教学建设
-    jxzy: 8233, // 教学资源
-};
-
-async function load(link, cookie, ctx) {
-    const cache = await ctx.cache.get(link);
-    if (cache) {
-        return JSON.parse(cache);
-    }
-    const response = await got.get(link, {
-        headers: {
-            cookie,
-        },
-    });
-    const $ = cheerio.load(response.data);
-    const pubDate = new Date(
-        $('.release-time')
-            .text()
-            .slice(-10)
-            .match(/\d{4}-\d{2}-\d{2}/)
-    ).toUTCString();
-    const images = $('img');
-    for (let k = 0; k < images.length; k++) {
-        $(images[k]).replaceWith(`<img src="${url.resolve(host, $(images[k]).attr('src'))}" />`);
-    }
-    const description = $('.wp_articlecontent').html();
-    const result = { pubDate, description };
-    ctx.cache.set(link, JSON.stringify(result));
-
-    return result;
-}
-
-module.exports = async (ctx) => {
-    const browser = await require('@/utils/puppeteer')();
-    const type = ctx.params.type || 'default';
-    const listUrl = `${host}${map[type]}/list.htm`;
-
-    const page = await browser.newPage();
-    await page.setUserAgent('Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/74.0.3729.169 Safari/537.36');
-    await page.evaluateOnNewDocument(() => {
-        // eslint-disable-next-line
-        Object.defineProperty(navigator, 'webdriver', { get: () => false });
-    });
-    // 南航的新waf现在采用cookie验证的方法防止爬虫，并且会对chrome headless进行检测
-
-    await page.goto(listUrl, {
-        waitUntil: 'networkidle0',
-    });
-
-    let msg;
-    try {
-        // data = await page.$('ul.right-ul')
-        msg = await page.$$eval('ul.right-ul > li', (es) =>
-            es.map((e) => {
-                const html = e.innerHTML;
-
-                let [, title] = html.match(/<a.+?>(.+)<\/a>/);
-                if (title.match(/font/)) {
-                    [, title] = title.match(/>(.+)</);
-                }
-                const [, href] = html.match(/href="(.+?)"/);
-                // const [, time] = html.match(/-time">(.+?)</);
-                return {
-                    title,
-                    link: href,
-                    guid: href,
-                };
-            })
-        );
-    } catch (e) {
-        // debugger
-    }
-
-    let cookie = await page.cookies();
-    cookie = cookie.map(({ name, value }) => `${name}=${value}`).join('; ');
-    browser.close();
-    msg = await Promise.all(
-        msg.map(async (e) => {
-            const link = url.resolve(host, e.link);
-            return {
-                ...e,
-                link,
-                guid: link,
-                ...(await load(link, cookie, ctx)),
-            };
-        })
-    );
-
-    ctx.state.data = {
-        title: '南航教务',
-        link: host,
-        description: '南航教务RSS',
-        item: msg,
-    };
-};

lib/v2/nuaa/cs/index.js

@@ -0,0 +1,87 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const getCookie = require('../utils/pypasswaf');
+const host = 'http://cs.nuaa.edu.cn/';
+
+const map = new Map([
+    ['tzgg', { title: '南京航空航天大学计算机科学与技术学院 -- 通知公告', suffix: 'tzgg/list.htm' }],
+    ['rdxw', { title: '南京航空航天大学计算机科学与技术学院 -- 热点新闻', suffix: '10846/list.htm' }],
+    ['xkky', { title: '南京航空航天大学计算机科学与技术学院 -- 学科科研', suffix: '10849/list.htm' }],
+    ['be', { title: '南京航空航天大学计算机科学与技术学院 -- 本科生培养', suffix: '10850/list.htm' }],
+    ['me', { title: '南京航空航天大学计算机科学与技术学院 -- 研究生培养', suffix: '10851/list.htm' }],
+    ['jxdt', { title: '南京航空航天大学计算机科学与技术学院 -- 教学动态', suffix: '1977/list.htm' }],
+    ['xsgz', { title: '南京航空航天大学计算机科学与技术学院 -- 学生工作', suffix: '1959/list.htm' }],
+]);
+
+module.exports = async (ctx) => {
+    const type = ctx.params.type;
+    const getDescription = Boolean(ctx.params.getDescription) || false;
+    const suffix = map.get(type).suffix;
+
+    const link = new URL(suffix, host).href;
+    const cookie = await getCookie();
+    const gotConfig = {
+        headers: {
+            cookie,
+        },
+    };
+    const response = await got(link, gotConfig);
+    const $ = cheerio.load(response.data);
+
+    const list = $('#news_list ul li')
+        .slice(0, Math.min(parseInt($('.per_count', '#wp_paging_w6').text()), parseInt($('.all_count', '#wp_paging_w6').slice(1).text())))
+        .map(function () {
+            const info = {
+                title: $(this).find('a').attr('title'),
+                link: $(this).find('a').attr('href'),
+                date: $(this).find('span').text(),
+            };
+            return info;
+        })
+        .get();
+
+    const out = await Promise.all(
+        list.map((info) => {
+            const title = info.title || 'tzgg';
+            const date = info.date;
+            const itemUrl = new URL(info.link, host).href;
+
+            return ctx.cache.tryGet(itemUrl, async () => {
+                const arr = itemUrl.split('.');
+                const pageType = arr[arr.length - 1];
+
+                // 南航新 WAF 过于敏感
+                // 目前 description 需要遍历页面，会被 WAF 拦截导致无法输出
+                // 考虑换一种获取 description 的方式或者将标题当作 title。
+                let description = title;
+                if (getDescription) {
+                    description = itemUrl;
+                    if (pageType === 'htm' || pageType === 'html') {
+                        const response = await got.get(itemUrl, gotConfig);
+                        const $ = cheerio.load(response.data);
+                        description = $('.wp_articlecontent')
+                            .html()
+                            .replace(/src="\//g, `src="${new URL('.', host).href}`)
+                            .trim();
+                    }
+                }
+
+                const single = {
+                    title,
+                    link: itemUrl,
+                    description,
+                    pubDate: parseDate(date),
+                };
+                return single;
+            });
+        })
+    );
+
+    ctx.state.data = {
+        title: map.get(type).title,
+        link,
+        description: '南京航空航天大学计算机科学与技术学院RSS',
+        item: out,
+    };
+};

lib/v2/nuaa/jwc/jwc.js

@@ -0,0 +1,82 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const getCookie = require('../utils/pypasswaf');
+const host = 'http://aao.nuaa.edu.cn/';
+
+const map = new Map([
+    ['tzgg', { title: '南京航空航天大学教务处 -- 通知公告', suffix: '8222/list.htm' }],
+    ['jxfw', { title: '南京航空航天大学教务处 -- 教学服务', suffix: '8230/list.htm' }],
+    ['xspy', { title: '南京航空航天大学教务处 -- 学生培养', suffix: '8231/list.htm' }],
+    ['jxjs', { title: '南京航空航天大学教务处 -- 教学建设', suffix: '8232/list.htm' }],
+    ['jxzy', { title: '南京航空航天大学教务处 -- 教学资源', suffix: '8233/list.htm' }],
+]);
+
+module.exports = async (ctx) => {
+    const type = ctx.params.type;
+    const suffix = map.get(type).suffix;
+    const getDescription = Boolean(ctx.params.getDescription) || false;
+
+    const link = new URL(suffix, host).href;
+    const cookie = await getCookie();
+    const gotConfig = {
+        headers: {
+            cookie,
+        },
+    };
+    const response = await got(link, gotConfig);
+    const $ = cheerio.load(response.data);
+
+    const list = $('#wp_news_w8 ul li')
+        .slice(0, 10)
+        .map(function () {
+            const info = {
+                title: $(this).find('a').text(),
+                link: $(this).find('a').attr('href'),
+                date: $(this).find('span').text(),
+            };
+            return info;
+        })
+        .get();
+
+    const out = await Promise.all(
+        list.map((info) => {
+            const title = info.title || 'tzgg';
+            const date = info.date;
+            const itemUrl = new URL(info.link, host).href;
+
+            return ctx.cache.tryGet(itemUrl, async () => {
+                const arr = itemUrl.split('.');
+                const pageType = arr[arr.length - 1];
+
+                // 南航新 WAF 过于敏感
+                // 目前 description 需要遍历页面，会被 WAF 拦截导致无法输出
+                // 考虑换一种获取 description 的方式或者将标题当作 title。
+                let description = title;
+                if (getDescription) {
+                    description = itemUrl;
+                    if (pageType === 'htm' || pageType === 'html') {
+                        const response = await got(itemUrl, gotConfig);
+                        const $ = cheerio.load(response.data);
+                        description = $('.wp_articlecontent').html();
+                    }
+                }
+
+                const single = {
+                    title,
+                    link: itemUrl,
+                    description,
+                    pubDate: parseDate(date),
+                };
+                return single;
+            });
+        })
+    );
+
+    ctx.state.data = {
+        title: map.get(type).title,
+        link,
+        description: '南京航空航天大学教务处RSS',
+        item: out,
+    };
+};

lib/v2/nuaa/maintainer.js

@@ -0,0 +1,5 @@
+module.exports = {
+    '/cs/:type/:getDescription?': ['LogicJake', 'Seiry', 'qrzbing'],
+    '/jwc/:type/:getDescription?': ['arcosx', 'Seiry', 'qrzbing'],
+    '/yjsy/:type?': ['junfengP', 'Seiry'],
+};

lib/v2/nuaa/radar.js

@@ -0,0 +1,23 @@
+module.exports = {
+    'nuaa.edu.cn': {
+        _name: '南京航空航天大学',
+        aao: [
+            {
+                title: '教务处',
+                docs: 'https://docs.rsshub.app/university.html#nan-jing-hang-kong-hang-tian-da-xue',
+            },
+        ],
+        cs: [
+            {
+                title: '计算机科学与技术学院',
+                docs: 'https://docs.rsshub.app/university.html#nan-jing-hang-kong-hang-tian-da-xue',
+            },
+        ],
+        'www.graduate': [
+            {
+                title: '研究生院',
+                docs: 'https://docs.rsshub.app/university.html#nan-jing-hang-kong-hang-tian-da-xue',
+            },
+        ],
+    },
+};

lib/v2/nuaa/router.js

@@ -0,0 +1,5 @@
+module.exports = (router) => {
+    router.get('/cs/:type/:getDescription?', require('./cs/index'));
+    router.get('/jwc/:type/:getDescription?', require('./jwc/jwc'));
+    router.get('/yjsy/:type?', require('./yjsy/yjsy'));
+};

lib/v2/nuaa/utils/pypasswaf.js

@@ -2,7 +2,7 @@ const host = 'http://aao.nuaa.edu.cn/';
 
 /**
  * async function 获取cookie
- * @desc 返回一个可用的cookie，使用 `got` 发起请求的时候，传入到`options.header.cookie`即可
+ * @desc 返回一个可用的cookie，使用 `got` 发起请求的时候，传入到`options.headers.cookie`即可
  */
 module.exports = async function getCookie() {
     const browser = await require('@/utils/puppeteer')();

lib/v2/nuaa/yjsy/yjsy.js

@@ -1,8 +1,7 @@
 const got = require('@/utils/got');
 const cheerio = require('cheerio');
-const url = require('url');
 const host = 'http://www.graduate.nuaa.edu.cn/';
-const formatPubDate = require('@/utils/date.js');
+const { parseDate } = require('@/utils/parse-date');
 
 const map = {
     latest: '2146/list.htm',
@@ -12,24 +11,20 @@ const map = {
     xxfw: '2147/list.htm',
 };
 
-async function load(link, ctx) {
-    const cache = await ctx.cache.get(link);
-    if (cache) {
-        return cache;
-    }
-    const response = await got.get(host + link);
-    const $ = cheerio.load(response.data);
-    const images = $('img');
-    for (let k = 0; k < images.length; k++) {
-        $(images[k]).replaceWith(`<img src="${url.resolve(host, $(images[k]).attr('src'))}" />`);
-    }
-    const description = $('.wp_articlecontent').html();
-    ctx.cache.set(link, description);
-    return { description };
-}
+const load = (link, ctx) =>
+    ctx.cache.tryGet(link, async () => {
+        const response = await got(host + link);
+        const $ = cheerio.load(response.data);
+        const images = $('img');
+        for (let k = 0; k < images.length; k++) {
+            $(images[k]).replaceWith(`<img src="${new URL($(images[k]).attr('src'), host).href}" />`);
+        }
+        const description = $('.wp_articlecontent').html();
+        return { description };
+    });
 
 module.exports = async (ctx) => {
-    const type = ctx.params.type || 'latest';
+    const type = ctx.params.type ?? 'latest';
     const response = await got({
         method: 'get',
         url: host + map[type],
@@ -43,12 +38,11 @@ module.exports = async (ctx) => {
             const itemUrl = a.attr('href');
             const single = {
                 title: a.text(),
-                link: url.resolve(host, itemUrl),
-                guid: url.resolve(host, itemUrl),
-                pubDate: formatPubDate(t.text()),
+                link: new URL(itemUrl, host).href,
+                pubDate: parseDate(t.text(), 'YYYY-MM-DD'),
             };
             const other = await load(itemUrl, ctx);
-            return Promise.resolve(Object.assign({}, single, other));
+            return { ...single, ...other };
         })
     );