rsshub 1.0.0-master.fdbb6fa → 1.0.0-master.fdc59d2

package/lib/v2/nuaa/yjsy/yjsy.js

@@ -1,55 +1,74 @@
 const got = require('@/utils/got');
 const cheerio = require('cheerio');
-const host = 'http://www.graduate.nuaa.edu.cn/';
 const { parseDate } = require('@/utils/parse-date');
+const getCookie = require('../utils/pypasswaf');
+const host = 'http://www.graduate.nuaa.edu.cn/';
 
-const map = {
-    latest: '2146/list.htm',
-    yyxw: 'yyxw/list.htm',
-    sjwj: 'sjwj/list.htm',
-    glwj: '2017/list.htm',
-    xxfw: '2147/list.htm',
-};
-
-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 };
-    });
+const map = new Map([
+    ['tzgg', { title: '通知公告 | 南京航空航天大学研究生院', suffix: '2145/list.htm' }],
+    ['xwdt', { title: '新闻动态 | 南京航空航天大学研究生院', suffix: '13276/list.htm' }],
+    ['xsxx', { title: '学术信息 | 南京航空航天大学研究生院', suffix: '13277/list.htm' }],
+    ['ssfc', { title: '师生风采 | 南京航空航天大学研究生院', suffix: '13278/list.htm' }],
+]);
 
 module.exports = async (ctx) => {
-    const type = ctx.params.type ?? 'latest';
-    const response = await got({
-        method: 'get',
-        url: host + map[type],
-    });
+    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 = $('#news_list ul.news_ul > li').slice(0, 10).get();
-    const process = await Promise.all(
-        list.map(async (item) => {
-            const a = $(item).find('a');
-            const t = $(item).find('span');
-            const itemUrl = a.attr('href');
-            const single = {
-                title: a.text(),
-                link: new URL(itemUrl, host).href,
-                pubDate: parseDate(t.text(), 'YYYY-MM-DD'),
+
+    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),
             };
-            const other = await load(itemUrl, ctx);
-            return { ...single, ...other };
         })
     );
 
     ctx.state.data = {
-        title: '南航研究生院',
-        link: host,
-        description: '南航研究生院RSS',
-        item: process,
+        title: map.get(type).title,
+        link,
+        description: '南京航空航天大学研究生院RSS',
+        item: out,
     };
 };

package/lib/{routes/universities → v2}/nuist/bulletin.js

@@ -1,6 +1,6 @@
 const got = require('@/utils/got');
 const cheerio = require('cheerio');
-const url = require('url');
+const { parseDate } = require('@/utils/parse-date');
 
 const baseTitle = '南信大信息公告栏';
 const baseUrl = 'https://bulletin.nuist.edu.cn';
@@ -25,9 +25,9 @@ module.exports = async (ctx) => {
     const category = map.hasOwnProperty(ctx.params.category) ? ctx.params.category : '791';
     const link = `${baseUrl}/${category}/list.htm`;
 
-    const response = await got.get(link);
+    const response = await got(link);
     const $ = cheerio.load(response.data);
-    const list = $('.news_list').find('.news').slice(0, 10);
+    const list = $('.news_list').find('.news');
 
     ctx.state.data = {
         title: baseTitle + (category === '791' ? '' : ':' + map[category]),
@@ -42,18 +42,18 @@ module.exports = async (ctx) => {
                         title: itemXsTitle.text(),
                         author: item.find('.xs_bgr').text(),
                         category: '学术报告',
-                        pubDate: new Date(item.find('.xs_date').text()).toUTCString(),
-                        link: url.resolve(baseUrl, itemXsTitle.attr('href')),
+                        pubDate: parseDate(item.find('.xs_date').text()),
+                        link: new URL(itemXsTitle.attr('href'), baseUrl).href,
                     };
                 }
 
                 const itemTitle = item.find('.news_title');
                 return {
-                    title: [itemTitle.find('.zdtb img').length > 0 ? '[顶]' : '', itemTitle.find('.btt').text()].join(''),
+                    title: [itemTitle.find('.zdtb img').length > 0 ? '[顶]' : '', itemTitle.find('.btt').text()].join(' '),
                     author: item.find('.news_org').text(),
                     category: itemTitle.find('.wjj').text(),
-                    pubDate: new Date(item.find('.news_date').text()).toUTCString(),
-                    link: url.resolve(baseUrl, itemTitle.find('.btt a').attr('href')),
+                    pubDate: parseDate(item.find('.news_date').text()),
+                    link: new URL(itemTitle.find('.btt a').attr('href'), baseUrl).href,
                 };
             })
             .get(),

package/lib/v2/nuist/cas.js

@@ -0,0 +1,71 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+
+const baseTitle = 'NUIST AS（南信大大气科学学院）';
+const baseUrl = 'https://cas.nuist.edu.cn';
+
+module.exports = async (ctx) => {
+    const { category = 'xxgg' } = ctx.params;
+    const link = `${baseUrl}/index/${category}.htm`;
+
+    const response = await got(link);
+    const $ = cheerio.load(response.data);
+    const list = $('#ctl00_ctl00_body_NewsList')
+        .find('tr.gridline')
+        .toArray()
+        .map((item) => {
+            item = $(item);
+            return {
+                title: item.find('.Title').text().trim(),
+                link: new URL(item.find('.Title').attr('href'), baseUrl).href,
+                pubDate: parseDate(
+                    item
+                        .find('td')
+                        .eq(2)
+                        .text()
+                        .replace(/[[\]]+/g, ''),
+                    'YYYY.MM.DD'
+                ),
+            };
+        });
+
+    const items = await Promise.all(
+        list.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                let response;
+                try {
+                    response = await got(item.link);
+                    const $ = cheerio.load(response.data);
+
+                    const authorMatch = $('.zzxx')
+                        .text()
+                        .match(/作者:(.*) 发布时间/);
+                    item.author = authorMatch ? authorMatch[1].trim() : null;
+                    item.pubDate = timezone(
+                        parseDate(
+                            item
+                                .find('.zzxx')
+                                .text()
+                                .match(/发布时间:(\d{4}\/\d{2}\/\d{2} \d{2}:\d{2}:\d{2})/)[1]
+                        ),
+                        +8
+                    );
+                    item.description = $('#vsb_content').html();
+                } catch (e) {
+                    // intranet
+                }
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: baseTitle + '：' + $('.dqlm').find('a').eq(1).text(),
+        description: $('meta[name=description]').attr('content'),
+        link,
+        item: items,
+    };
+};

package/lib/v2/nuist/jwc.js

@@ -0,0 +1,43 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+
+const baseTitle = '南京信息工程大学-教务处';
+const baseUrl = 'https://jwc.nuist.edu.cn';
+
+module.exports = async (ctx) => {
+    const { category = 'jxyw' } = ctx.params;
+    const link = `${baseUrl}/${category === 'jxyw' || category === 'xyjx' ? 'index' : 'xxtz'}/${category}.htm`;
+
+    const response = await got(link);
+    const $ = cheerio.load(response.data);
+    const list = $('.main_list ul li')
+        .toArray()
+        .map((item) => {
+            item = $(item);
+            return {
+                title: item.find('a').contents().first().text(),
+                link: new URL(item.find('a').attr('href'), baseUrl).href,
+                pubDate: parseDate(item.find('.date').text()),
+            };
+        });
+
+    const items = await Promise.all(
+        list.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const response = await got(item.link);
+                const $ = cheerio.load(response.data);
+
+                item.description = $('#vsb_content').html();
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: baseTitle + '：' + $('.dqwz').find('a').eq(1).text(),
+        link,
+        item: items,
+    };
+};

package/lib/v2/nuist/library/lib.js

@@ -0,0 +1,43 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const baseTitle = '南京信息工程大学图书馆通知';
+const baseUrl = 'https://lib.nuist.edu.cn';
+
+module.exports = async (ctx) => {
+    const link = baseUrl + '/index/tzgg.htm';
+
+    const response = await got(link);
+    const $ = cheerio.load(response.data);
+    const list = $('.list2 li')
+        .toArray()
+        .map((item) => {
+            item = $(item);
+
+            return {
+                title: item.find('a').attr('title'),
+                category: '通知',
+                link: new URL(item.find('a').attr('href'), link).href,
+            };
+        });
+
+    const items = await Promise.all(
+        list.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const response = await got(item.link);
+                const $ = cheerio.load(response.data);
+
+                item.description = $('#vsb_content').html();
+                item.pubDate = parseDate($('.date').find('span').eq(0).text(), 'YYYY年MM月DD日');
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: baseTitle,
+        link,
+        item: items,
+    };
+};

package/lib/v2/nuist/maintainer.js

@@ -0,0 +1,10 @@
+module.exports = {
+    '/bulletin/:category?': ['gylidian'],
+    '/cas/:category?': ['gylidian'],
+    '/jwc/:category?': ['gylidian'],
+    '/lib': ['gylidian'],
+    '/scs/:category?': ['gylidian'],
+    '/sese/:category?': ['gylidian'],
+    '/xgc': ['gylidian'],
+    '/yjs/:path+': ['gylidian', 'nczitzk'],
+};

package/lib/v2/nuist/radar.js

@@ -0,0 +1,69 @@
+module.exports = {
+    'nuist.edu.cn': {
+        _name: '南京信息工程大学',
+        bulletin: [
+            {
+                title: '信息公告栏',
+                docs: 'https://docs.rsshub.app/university.html#nan-jing-xin-xi-gong-cheng-da-xue',
+                source: ['/:category/list.htm'],
+                target: '/nuist/bulletin/:category',
+            },
+        ],
+        cas: [
+            {
+                title: '大气科学学院',
+                docs: 'https://docs.rsshub.app/university.html#nan-jing-xin-xi-gong-cheng-da-xue',
+                source: ['/index/:category'],
+                target: (params) => `/nuist/cas/${params.category.replace('.htm', '')}`,
+            },
+        ],
+        jwc: [
+            {
+                title: '教务处',
+                docs: 'https://docs.rsshub.app/university.html#nan-jing-xin-xi-gong-cheng-da-xue',
+                source: ['/index/:category', '/xxtz/:category'],
+                target: (params) => `/nuist/jwc/${params.category.replace('.htm', '')}`,
+            },
+        ],
+        lib: [
+            {
+                title: '图书馆',
+                docs: 'https://docs.rsshub.app/university.html#nan-jing-xin-xi-gong-cheng-da-xue',
+                source: ['/', '/index/tzgg.htm'],
+                target: '/nuist/lib',
+            },
+        ],
+        scs: [
+            {
+                title: '计软院',
+                docs: 'https://docs.rsshub.app/university.html#nan-jing-xin-xi-gong-cheng-da-xue',
+                source: ['/:category/list.htm'],
+                target: '/nuist/scs/:category',
+            },
+        ],
+        sese: [
+            {
+                title: '环科院',
+                docs: 'https://docs.rsshub.app/university.html#nan-jing-xin-xi-gong-cheng-da-xue',
+                source: ['/:category'],
+                target: (params) => `/nuist/sese/${params.category.replace('.htm', '')}`,
+            },
+        ],
+        xgc: [
+            {
+                title: '学生工作处',
+                docs: 'https://docs.rsshub.app/university.html#nan-jing-xin-xi-gong-cheng-da-xue',
+                source: ['/', '/419/list.htm'],
+                target: '/nuist/xgc',
+            },
+        ],
+        yjs: [
+            {
+                title: '研究生院学科建设处',
+                docs: 'https://docs.rsshub.app/university.html#nan-jing-xin-xi-gong-cheng-da-xue',
+                source: ['/'],
+                target: '/nuist/jwc/:path+',
+            },
+        ],
+    },
+};

package/lib/v2/nuist/router.js

@@ -0,0 +1,10 @@
+module.exports = function (router) {
+    router.get('/bulletin/:category?', require('./bulletin'));
+    router.get('/cas/:category?', require('./cas'));
+    router.get('/jwc/:category?', require('./jwc'));
+    router.get('/lib', require('./library/lib'));
+    router.get('/scs/:category?', require('./scs'));
+    router.get('/sese/:category?', require('./sese'));
+    router.get('/xgc', require('./xgc'));
+    router.get(/yjs\/([\w/-]+)?/, require('./yjs'));
+};

package/lib/v2/nuist/scs.js

@@ -0,0 +1,49 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const baseTitle = 'NUIST CS（南信大计软院）';
+const baseUrl = 'https://scs.nuist.edu.cn';
+
+module.exports = async (ctx) => {
+    const { category = 'xwkx' } = ctx.params;
+    const link = `${baseUrl}/${category}/list.htm`;
+
+    const response = await got(link);
+    const $ = cheerio.load(response.data);
+
+    const list = $('.newsList ul')
+        .eq(0)
+        .find('li')
+        .toArray()
+        .map((item) => {
+            item = $(item);
+            return {
+                title: item.find('a').attr('title'),
+                link: new URL(item.find('a').attr('href'), baseUrl).href,
+                pubDate: parseDate(item.find('.newsDate').text()),
+            };
+        });
+
+    const items = await Promise.all(
+        list.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const response = await got(item.link);
+                const $ = cheerio.load(response.data);
+
+                const authorMatch = $('.newsTitleAddDate')
+                    .text()
+                    .match(/发布者：(.*)发布时间/);
+                item.author = authorMatch ? authorMatch[1].trim() : null;
+                item.description = $('.newsContent').html();
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: baseTitle + '：' + $('head title').text(),
+        description: $('meta[name=description]').attr('content'),
+        link,
+        item: items,
+    };
+};

package/lib/v2/nuist/sese.js

@@ -0,0 +1,45 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+
+const baseTitle = 'NUIST ESE（南信大环科院）';
+const baseUrl = 'https://sese.nuist.edu.cn';
+
+module.exports = async (ctx) => {
+    const { category = 'tzgg1' } = ctx.params;
+    const link = `${baseUrl}/${category}.htm`;
+
+    const response = await got(link);
+    const $ = cheerio.load(response.data);
+    const list = $('#ctl00_ctl00_mainbody_rightbody_listcontent_NewsList .gridline')
+        .toArray()
+        .map((item) => {
+            item = $(item);
+            return {
+                title: item.find('a').eq(1).text(),
+                link: new URL(item.find('a').eq(1).attr('href'), baseUrl).href,
+                category: item.find('a').eq(0).text(),
+                pubDate: parseDate($(item).find('.gridlinedate').text(), 'YYYY年MM月DD日'),
+            };
+        });
+
+    const items = await Promise.all(
+        list.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const response = await got(item.link);
+                const $ = cheerio.load(response.data);
+
+                item.description = $('#vsb_content_6').html();
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: baseTitle + '：' + $('.lmtitle').text(),
+        description: $('meta[name=description]').attr('content'),
+        link,
+        item: items,
+    };
+};

package/lib/v2/nuist/xgc.js

@@ -0,0 +1,49 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+
+const baseTitle = '南信大学生工作处';
+const baseUrl = 'https://xgc.nuist.edu.cn';
+
+module.exports = async (ctx) => {
+    const link = baseUrl + '/419/list.htm';
+
+    const response = await got(link);
+    const $ = cheerio.load(response.data);
+    const list = $('.wp_article_list .list_item')
+        .toArray()
+        .map((item) => {
+            item = $(item);
+            return {
+                title: item.find('a').attr('title'),
+                link: new URL(item.find('a').attr('href'), baseUrl).href,
+                pubDate: parseDate(item.find('.Article_PublishDate').text(), 'YYYY-MM-DD'),
+            };
+        });
+
+    const items = await Promise.all(
+        list.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                let response;
+                try {
+                    response = await got(item.link);
+                    const $ = cheerio.load(response.data);
+
+                    const articleInfo = $('.arti_metas');
+
+                    item.author = articleInfo.find('.arti_publisher').text().replace('作者：', '');
+                    item.description = $('.entry').html();
+                } catch (e) {
+                    // intranet
+                }
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: baseTitle,
+        link,
+        item: items.filter((x) => x),
+    };
+};

package/lib/v2/nuist/yjs.js

@@ -0,0 +1,60 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const timezone = require('@/utils/timezone');
+const { parseDate } = require('@/utils/parse-date');
+
+module.exports = async (ctx) => {
+    const path = ctx.path === '/yjs/' ? 'index/tzgg' : ctx.path.replace(/^\/yjs\//, '');
+
+    const rootUrl = 'https://yjs.nuist.edu.cn';
+    const currentUrl = `${rootUrl}/${path}.htm`;
+
+    const response = await got({
+        method: 'get',
+        url: currentUrl,
+    });
+
+    const $ = cheerio.load(response.data);
+
+    let items = $('.gridlinediv')
+        .toArray()
+        .map((item) => {
+            item = $(item);
+
+            const a = item.find('a').last();
+
+            return {
+                title: a.text(),
+                link: new URL(a.attr('href'), currentUrl).href,
+                pubDate: parseDate(item.next().text(), 'YYYY年MM月DD日'),
+            };
+        });
+
+    items = await Promise.all(
+        items.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const detailResponse = await got({
+                    method: 'get',
+                    url: item.link,
+                });
+
+                const content = cheerio.load(detailResponse.data);
+
+                const timeMatches = content('.bar')
+                    .text()
+                    .match(/(\d{4}年\d{2}月\d{2}日 \d{2}:\d{2})/);
+
+                item.description = content('.v_news_content').html();
+                item.pubDate = timeMatches ? timezone(parseDate(timeMatches[1], 'YYYY年MM月DD日 HH:mm'), +8) : item.pubDate;
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: $('title').text(),
+        link: currentUrl,
+        item: items,
+    };
+};