rsshub 1.0.0-master.f78ad03 → 1.0.0-master.f792555

package/lib/v2/nua/gra.js

@@ -1,56 +1,21 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-const { parseDate } = require('@/utils/parse-date');
-const timezone = require('@/utils/timezone');
+const util = require('./utils');
 const baseUrl = 'https://grad.nua.edu.cn';
 
 module.exports = async (ctx) => {
     const type = ctx.params.type;
     const newsUrl = `${baseUrl}/${type}/list.htm`;
+    const listName = 'li.list_item';
+    const artiContent = '.read';
+    const listDate = '.Article_PublishDate';
+    const webPageName = '.col_title';
 
-    const response = await got(newsUrl, {
-        https: {
-            rejectUnauthorized: false,
-        },
-    });
-
-    const data = response.data;
-    const $ = cheerio.load(data);
-    const pagename = $('.col_title').text();
-
-    const items = $('li.list_item')
-        .toArray()
-        .map((item) => {
-            item = $(item);
-            return {
-                link: baseUrl + item.find('a').attr('href'),
-                title: item.find('a').attr('title'),
-                pubDate: timezone(parseDate(item.find('.Article_PublishDate').first().text(), 'YYYY-MM-DD'), +8),
-            };
-        });
-
-    const results = await Promise.all(
-        items.map((item) =>
-            ctx.cache.tryGet(item.link, async () => {
-                const result = await got(item.link, {
-                    https: {
-                        rejectUnauthorized: false,
-                    },
-                });
-                const $ = cheerio.load(result.data);
-
-                item.author = $('.arti_publisher').text();
-                item.description = $('.read').html();
-
-                return item;
-            })
-        )
-    );
+    const items = await util.ProcessList(newsUrl, baseUrl, listName, listDate, webPageName);
+    const results = await util.ProcessFeed(items[0], artiContent, ctx);
 
     ctx.state.data = {
-        title: 'NUA-研究生处-' + pagename,
+        title: 'NUA-研究生处-' + items[1],
         link: `${baseUrl}/${type}/list.htm`,
-        description: '南京艺术学院研究生处' + pagename,
+        description: '南京艺术学院 研究生处 ' + items[1],
         item: results,
     };
 };

package/lib/v2/nua/index.js

@@ -0,0 +1,22 @@
+const util = require('./utils');
+
+module.exports = async (ctx) => {
+    const type = ctx.params.type;
+
+    const baseUrl = 'https://www.nua.edu.cn';
+    const newsUrl = `${baseUrl}/${type}/list.htm`;
+    const listName = 'li.news';
+    const listDate = '.news_meta';
+    const webPageName = '.col_title';
+
+    const artiContent = '.read';
+    const items = await util.ProcessList(newsUrl, baseUrl, listName, listDate, webPageName);
+    const results = await util.ProcessFeed(items[0], artiContent, ctx);
+
+    ctx.state.data = {
+        title: 'NUA-' + items[1],
+        link: newsUrl,
+        description: '南京艺术学院 ' + items[1],
+        item: results,
+    };
+};

package/lib/v2/nua/lib.js

@@ -0,0 +1,39 @@
+const util = require('./utils');
+const baseUrl = 'https://lib.nua.edu.cn';
+
+module.exports = async (ctx) => {
+    const type = ctx.params.type;
+    let webPageName;
+
+    switch (type) {
+        case 'xwdt':
+            webPageName = '.wp_column.column-1.selected';
+            break;
+        case 'djdt':
+            webPageName = '.wp_column.column-2.selected';
+            break;
+        case 'zydt':
+            webPageName = '.wp_column.column-3.selected';
+            break;
+        case 'fwdt':
+            webPageName = '.wp_column.column-4.selected';
+            break;
+        default:
+            throw Error(`暂不支持对${type}的订阅`);
+    }
+
+    const newsUrl = `${baseUrl}/${type}/list.htm`;
+    const listName = 'div.news_con';
+    const artiContent = '.wp_articlecontent';
+    const listDate = '.news_date';
+
+    const items = await util.ProcessList(newsUrl, baseUrl, listName, listDate, webPageName);
+    const results = await util.ProcessFeed(items[0], artiContent, ctx);
+
+    ctx.state.data = {
+        title: 'NUA-图书馆-' + items[1],
+        link: `${baseUrl}/${type}/list.htm`,
+        description: '南京艺术学院 图书馆 ' + items[1],
+        item: results,
+    };
+};

package/lib/v2/nua/maintainer.js

@@ -1,3 +1,7 @@
 module.exports = {
+    '/dc/:type': ['evnydd0sf'],
     '/gra/:type': ['evnydd0sf'],
+    '/index/:type': ['evnydd0sf'],
+    '/lib/:type': ['evnydd0sf'],
+    '/sxw/:type': ['evnydd0sf'],
 };

package/lib/v2/nua/radar.js

@@ -1,6 +1,30 @@
 module.exports = {
     'nua.edu.cn': {
         _name: '南京艺术学院',
+        index: [
+            {
+                title: '官网信息',
+                docs: 'https://docs.rsshub.app/university.html#nan-jing-yi-shu-xue-yuan',
+                source: ['/:type/list.htm'],
+                target: '/nua/index/:type',
+            },
+        ],
+        sxw: [
+            {
+                title: '双馨网',
+                docs: 'https://docs.rsshub.app/university.html#nan-jing-yi-shu-xue-yuan',
+                source: ['/:type/list.htm'],
+                target: '/nua/sxw/:type',
+            },
+        ],
+        dc: [
+            {
+                title: '设计学院',
+                docs: 'https://docs.rsshub.app/university.html#nan-jing-yi-shu-xue-yuan',
+                source: ['/:type/list.htm'],
+                target: '/nua/dc/:type',
+            },
+        ],
         grad: [
             {
                 title: '研究生处',
@@ -9,5 +33,13 @@ module.exports = {
                 target: '/nua/gra/:type',
             },
         ],
+        lib: [
+            {
+                title: '图书馆',
+                docs: 'https://docs.rsshub.app/university.html#nan-jing-yi-shu-xue-yuan',
+                source: ['/:type/list.htm'],
+                target: '/nua/lib/:type',
+            },
+        ],
     },
 };

package/lib/v2/nua/router.js

@@ -1,3 +1,7 @@
 module.exports = function (router) {
+    router.get('/dc/:type', require('./dc'));
     router.get('/gra/:type', require('./gra'));
+    router.get('/index/:type', require('./index'));
+    router.get('/lib/:type', require('./lib'));
+    router.get('/sxw/:type', require('./sxw'));
 };

package/lib/v2/nua/sxw.js

@@ -0,0 +1,22 @@
+const util = require('./utils');
+
+module.exports = async (ctx) => {
+    const type = ctx.params.type;
+
+    const baseUrl = 'https://sxw.nua.edu.cn';
+    const newsUrl = `${baseUrl}/${type}/list.htm`;
+    const listName = 'li.list_item';
+    const listDate = '.Article_PublishDate';
+    const webPageName = '.Column_Anchor';
+
+    const artiContent = '.read';
+    const items = await util.ProcessList(newsUrl, baseUrl, listName, listDate, webPageName);
+    const results = await util.ProcessFeed(items[0], artiContent, ctx);
+
+    ctx.state.data = {
+        title: 'NUA-双馨网-' + items[1],
+        link: newsUrl,
+        description: '南京艺术学院 双馨网 ' + items[1],
+        item: results,
+    };
+};

package/lib/v2/nua/utils.js

@@ -0,0 +1,75 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+const { fetchArticle } = require('@/utils/wechat-mp');
+
+const pageType = (href) => {
+    if (!href.startsWith('http')) {
+        return 'in-nua';
+    }
+    const url = new URL(href);
+    if (url.hostname === 'mp.weixin.qq.com') {
+        return 'wechat-mp';
+    } else if (url.hostname === 'www.nua.edu.cn') {
+        return 'nua';
+    } else {
+        return 'unknown';
+    }
+};
+
+function arti_link(text, href) {
+    return `<a href="${href}">${text}</a>`;
+}
+
+async function ProcessList(newsUrl, baseUrl, listName, listDate, webPageName) {
+    const result = await got(newsUrl, { https: { rejectUnauthorized: false } });
+    const $ = cheerio.load(result.data);
+
+    const pageName = $(webPageName).text().trim();
+
+    const items = $(listName)
+        .toArray()
+        .map((item) => {
+            const href = $(item).find('a').attr('href');
+            const type = pageType(href);
+
+            return {
+                link: type === 'in-nua' ? baseUrl + href : href,
+                title: $(item).find('a').attr('title'),
+                pubDate: timezone(parseDate($(item).find(listDate).first().text(), 'YYYY-MM-DD'), +8),
+                type,
+            };
+        });
+
+    return [items, pageName];
+}
+
+const ProcessFeed = async (items, artiContent, ctx) =>
+    await Promise.all(
+        items.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                switch (item.type) {
+                    case 'in-nua':
+                    case 'nua': {
+                        const result = await got(item.link, { https: { rejectUnauthorized: false } });
+                        const $ = cheerio.load(result.data);
+                        item.author = $('.arti_publisher').text() + '  ' + $('.arti_views').text();
+                        item.description = $(artiContent).html();
+                        return item;
+                    }
+                    case 'wechat-mp':
+                        return fetchArticle(ctx, item.link);
+                    case 'unknown':
+                    default:
+                        item.description = `暂不支持解析该内容，请点击 ${arti_link('原文', item.link)}`;
+                        return item;
+                }
+            })
+        )
+    );
+
+module.exports = {
+    ProcessList,
+    ProcessFeed,
+};

package/lib/v2/nuaa/college/cae.js

@@ -0,0 +1,79 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const getCookie = require('../utils/pypasswaf');
+const host = 'https://cae.nuaa.edu.cn/';
+
+const map = new Map([
+    ['zhxw', { title: '综合新闻 | 南京航空航天大学自动化学院', suffix: '5399/list.htm' }],
+    ['dwxz', { title: '党委行政 | 南京航空航天大学自动化学院', suffix: '13266/list.htm' }],
+    ['rshz', { title: '人事/合作 | 南京航空航天大学自动化学院', suffix: '13267/list.htm' }],
+    ['yjs', { title: '研究生培养 | 南京航空航天大学自动化学院', suffix: '13271/list.htm' }],
+    ['bks', { title: '本科生培养 | 南京航空航天大学自动化学院', suffix: '13270/list.htm' }],
+    ['xsgz', { title: '学生工作 | 南京航空航天大学自动化学院', suffix: '13268/list.htm' }],
+    ['tzgg', { title: '通知公告 | 南京航空航天大学自动化学院', suffix: '13264/list.htm' }],
+    ['xsxx', { title: '学术信息 | 南京航空航天大学自动化学院', suffix: '13265/list.htm' }],
+    ['dbgg', { title: '答辩公告 | 南京航空航天大学自动化学院', suffix: 'dbgg/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(host);
+    const gotConfig = {
+        headers: {
+            cookie,
+        },
+    };
+    const response = await got(link, gotConfig);
+    const $ = cheerio.load(response.data);
+
+    const list = $('#wp_news_w6 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(async (info) => {
+            const title = info.title || 'tzgg';
+            const date = info.date;
+            const itemUrl = new URL(info.link, host).href;
+            let description = title + '<br><a href="' + itemUrl + '" target="_blank">查看原文</a>';
+
+            if (getDescription) {
+                description = await ctx.cache.tryGet(itemUrl, async () => {
+                    const arr = itemUrl.split('.');
+                    const pageType = arr[arr.length - 1];
+                    if (pageType === 'htm' || pageType === 'html') {
+                        const response = await got(itemUrl, gotConfig);
+                        const $ = cheerio.load(response.data);
+                        return $('.wp_articlecontent').html() + '<br><hr /><a href="' + itemUrl + '" target="_blank">查看原文</a>';
+                    }
+                });
+            }
+
+            return {
+                title,
+                link: itemUrl,
+                description,
+                pubDate: parseDate(date),
+            };
+        })
+    );
+
+    ctx.state.data = {
+        title: map.get(type).title,
+        link,
+        description: '南京航空航天大学自动化学院RSS',
+        item: out,
+    };
+};

package/lib/v2/nuaa/college/cs.js

@@ -0,0 +1,78 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const getCookie = require('../utils/pypasswaf');
+const host = 'https://cs.nuaa.edu.cn/';
+
+const map = new Map([
+    ['tzgg', { title: '通知公告 | 南京航空航天大学计算机科学与技术学院', suffix: 'tzgg/list.htm' }],
+    ['rdxw', { title: '热点新闻 | 南京航空航天大学计算机科学与技术学院', suffix: '10846/list.htm' }],
+    ['xkky', { title: '学科科研 | 南京航空航天大学计算机科学与技术学院', suffix: '10849/list.htm' }],
+    ['be', { title: '本科生培养 | 南京航空航天大学计算机科学与技术学院', suffix: '10850/list.htm' }],
+    ['me', { title: '研究生培养 | 南京航空航天大学计算机科学与技术学院', suffix: '10851/list.htm' }],
+    ['jxdt', { title: '教学动态 | 南京航空航天大学计算机科学与技术学院', suffix: '1977/list.htm' }],
+    ['xsgz', { title: '学生工作 | 南京航空航天大学计算机科学与技术学院', 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(host);
+    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(async (info) => {
+            const title = info.title || 'tzgg';
+            const date = info.date;
+            const itemUrl = new URL(info.link, host).href;
+            let description = title + '<br><a href="' + itemUrl + '" target="_blank">查看原文</a>';
+
+            if (getDescription) {
+                description = await ctx.cache.tryGet(itemUrl, async () => {
+                    const arr = itemUrl.split('.');
+                    const pageType = arr[arr.length - 1];
+                    if (pageType === 'htm' || pageType === 'html') {
+                        const response = await got(itemUrl, gotConfig);
+                        const $ = cheerio.load(response.data);
+                        return $('.wp_articlecontent').html() + '<br><hr /><a href="' + itemUrl + '" target="_blank">查看原文</a>';
+                    }
+                });
+            }
+
+            return {
+                title,
+                link: itemUrl,
+                description,
+                pubDate: parseDate(date),
+            };
+        })
+    );
+
+    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,
+    };
+};

package/lib/v2/nuaa/jwc/jwc.js

@@ -5,11 +5,11 @@ 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' }],
+    ['tzgg', { title: '通知公告 | 南京航空航天大学教务处', suffix: '8222/list.htm' }],
+    ['jxfw', { title: '教学服务 | 南京航空航天大学教务处', suffix: '8230/list.htm' }],
+    ['xspy', { title: '学生培养 | 南京航空航天大学教务处', suffix: '8231/list.htm' }],
+    ['jxjs', { title: '教学建设 | 南京航空航天大学教务处', suffix: '8232/list.htm' }],
+    ['jxzy', { title: '教学资源 | 南京航空航天大学教务处', suffix: '8233/list.htm' }],
 ]);
 
 module.exports = async (ctx) => {
@@ -18,7 +18,7 @@ module.exports = async (ctx) => {
     const getDescription = Boolean(ctx.params.getDescription) || false;
 
     const link = new URL(suffix, host).href;
-    const cookie = await getCookie();
+    const cookie = await getCookie(host);
     const gotConfig = {
         headers: {
             cookie,
@@ -40,36 +40,30 @@ module.exports = async (ctx) => {
         .get();
 
     const out = await Promise.all(
-        list.map((info) => {
+        list.map(async (info) => {
             const title = info.title || 'tzgg';
             const date = info.date;
             const itemUrl = new URL(info.link, host).href;
+            let description = title + '<br><a href="' + itemUrl + '" target="_blank">查看原文</a>';
 
-            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 (getDescription) {
+                description = await ctx.cache.tryGet(itemUrl, async () => {
+                    const arr = itemUrl.split('.');
+                    const pageType = arr[arr.length - 1];
                     if (pageType === 'htm' || pageType === 'html') {
                         const response = await got(itemUrl, gotConfig);
                         const $ = cheerio.load(response.data);
-                        description = $('.wp_articlecontent').html();
+                        return $('.wp_articlecontent').html() + '<br><hr /><a href="' + itemUrl + '" target="_blank">查看原文</a>';
                     }
-                }
+                });
+            }
 
-                const single = {
-                    title,
-                    link: itemUrl,
-                    description,
-                    pubDate: parseDate(date),
-                };
-                return single;
-            });
+            return {
+                title,
+                link: itemUrl,
+                description,
+                pubDate: parseDate(date),
+            };
         })
     );
 

package/lib/v2/nuaa/maintainer.js

@@ -1,5 +1,6 @@
 module.exports = {
-    '/cs/:type/:getDescription?': ['LogicJake', 'Seiry', 'qrzbing'],
-    '/jwc/:type/:getDescription?': ['arcosx', 'Seiry', 'qrzbing'],
-    '/yjsy/:type?': ['junfengP', 'Seiry'],
+    '/cae/:type/:getDescription?': ['Xm798'],
+    '/cs/:type/:getDescription?': ['LogicJake', 'Seiry', 'qrzbing', 'Xm798'],
+    '/jwc/:type/:getDescription?': ['arcosx', 'Seiry', 'qrzbing', 'Xm798'],
+    '/yjsy/:type/:getDescription?': ['junfengP', 'Seiry', 'Xm798'],
 };

package/lib/v2/nuaa/radar.js

@@ -7,6 +7,12 @@ module.exports = {
                 docs: 'https://docs.rsshub.app/university.html#nan-jing-hang-kong-hang-tian-da-xue',
             },
         ],
+        cae: [
+            {
+                title: '自动化学院',
+                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',

package/lib/v2/nuaa/router.js

@@ -1,5 +1,6 @@
 module.exports = (router) => {
-    router.get('/cs/:type/:getDescription?', require('./cs/index'));
+    router.get('/cae/:type/:getDescription?', require('./college/cae'));
+    router.get('/cs/:type/:getDescription?', require('./college/cs'));
     router.get('/jwc/:type/:getDescription?', require('./jwc/jwc'));
-    router.get('/yjsy/:type?', require('./yjsy/yjsy'));
+    router.get('/yjsy/:type/:getDescription?', require('./yjsy/yjsy'));
 };

package/lib/v2/nuaa/utils/pypasswaf.js

@@ -1,10 +1,10 @@
-const host = 'http://aao.nuaa.edu.cn/';
+const puppeteerUtils = require('@/utils/puppeteer-utils');
 
 /**
  * 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.headers.cookie`即可
  */
-module.exports = async function getCookie() {
+module.exports = async function getCookie(host) {
     const browser = await require('@/utils/puppeteer')({ stealth: true });
     const page = await browser.newPage();
     await page.setRequestInterception(true);
@@ -16,8 +16,7 @@ module.exports = async function getCookie() {
         waitUntil: 'networkidle0',
     });
 
-    let cookie = await page.cookies();
-    cookie = cookie.map(({ name, value }) => `${name}=${value}`).join('; ');
+    const cookie = await puppeteerUtils.getCookies(page);
     await browser.close();
     return cookie;
 };