rsshub 1.0.0-master.fb2a407 → 1.0.0-master.fc5a2f6

package/lib/v2/sciencenet/user.js

@@ -0,0 +1,65 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const iconv = require('iconv-lite');
+const timezone = require('@/utils/timezone');
+const { parseDate } = require('@/utils/parse-date');
+
+module.exports = async (ctx) => {
+    const id = ctx.params.id;
+
+    const rootUrl = 'https://blog.sciencenet.cn';
+    const currentUrl = `${rootUrl}/u/${id}`;
+
+    let response = await got({
+        method: 'get',
+        url: currentUrl,
+        responseType: 'buffer',
+    });
+
+    let $ = cheerio.load(iconv.decode(response.data, 'gbk'));
+
+    response = await got({
+        method: 'get',
+        url: $('.xg1 a').eq(1).attr('href'),
+    });
+
+    $ = cheerio.load(response.data);
+
+    let items = $('item')
+        .slice(0, ctx.query.limit ? parseInt(ctx.query.limit) : 50)
+        .toArray()
+        .map((item) => {
+            item = $(item);
+
+            return {
+                title: item.find('title').text(),
+                link: item.find('guid').text(),
+            };
+        });
+
+    items = await Promise.all(
+        items.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const detailResponse = await got({
+                    method: 'get',
+                    url: item.link,
+                    responseType: 'buffer',
+                });
+
+                const content = cheerio.load(iconv.decode(detailResponse.data, 'gbk'));
+
+                item.author = content('.xs2').text();
+                item.description = content('#blog_article').html();
+                item.pubDate = timezone(parseDate(content('.xg1').eq(5).text()), +8);
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: `科学网 - ${items[0].author}的博文`,
+        link: currentUrl,
+        item: items,
+    };
+};

package/lib/v2/sdu/cmse.js

@@ -0,0 +1,59 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+
+const host = 'http://www.cmse.sdu.edu.cn/';
+const typelist = ['通知公告', '学院新闻', '本科生教育', '研究生教育', '学术动态'];
+const urlList = ['zxzx/tzgg.htm', 'zxzx/xyxw.htm', 'zxzx/bksjy.htm', 'zxzx/yjsjy.htm', 'zxzx/xsdt.htm'];
+
+module.exports = async (ctx) => {
+    const type = ctx.params.type ? parseInt(ctx.params.type) : 0;
+    const link = new URL(urlList[type], host).href;
+    const response = await got(link);
+
+    const $ = cheerio.load(response.data);
+
+    let item = $('.article_list li')
+        .map((_, e) => {
+            e = $(e);
+            const a = e.find('a');
+            return {
+                title: a.text().trim(),
+                link: a.attr('href'),
+                pubDate: parseDate(e.find('.date').text(), 'YYYY/MM/DD'),
+            };
+        })
+        .get();
+
+    item = await Promise.all(
+        item
+            .filter((e) => e.link.startsWith('../info'))
+            .map((item) => {
+                item.link = new URL(item.link.slice('3'), host).href;
+                return ctx.cache.tryGet(item.link, async () => {
+                    const response = await got(item.link);
+                    const $ = cheerio.load(response.data);
+
+                    item.title = $('.contentTitle').text();
+                    item.author =
+                        $('.contentTitle2')
+                            .find('span')
+                            .eq(1)
+                            .text()
+                            .trim()
+                            .match(/作者：(.*)/)[1] || '山东大学材料科学与工程学院';
+                    $('.contentTitle, .contentTitle2').remove();
+                    item.description = $('.content_detail').html();
+
+                    return item;
+                });
+            })
+    );
+
+    ctx.state.data = {
+        title: `山东大学材料科学与工程学院${typelist[type]}`,
+        description: $('title').text(),
+        link,
+        item,
+    };
+};

package/lib/v2/sdu/cs.js

@@ -0,0 +1,57 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const { finishArticleItem } = require('@/utils/wechat-mp');
+
+const host = 'https://www.cs.sdu.edu.cn/';
+const typelist = ['学院公告', '学术报告', '科技简讯'];
+const urlList = ['xygg.htm', 'xsbg.htm', 'kjjx.htm'];
+
+module.exports = async (ctx) => {
+    const type = ctx.params.type ? parseInt(ctx.params.type) : 0;
+    const link = new URL(urlList[type], host).href;
+
+    const response = await got(link);
+
+    const $ = cheerio.load(response.data);
+
+    let item = $('.dqlb ul li')
+        .map((_, e) => {
+            e = $(e);
+            const a = e.find('a');
+            return {
+                title: a.text().trim(),
+                link: a.attr('href').startsWith('info/') ? host + a.attr('href') : a.attr('href'),
+                pubDate: parseDate(e.find('.fr').text().trim(), 'YYYY-MM-DD'),
+            };
+        })
+        .get();
+
+    item = await Promise.all(
+        item.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                if (new URL(item.link).hostname === 'mp.weixin.qq.com') {
+                    return finishArticleItem(ctx, item);
+                } else if (new URL(item.link).hostname !== 'www.cs.sdu.edu.cn') {
+                    return item;
+                }
+                const response = await got(item.link);
+                const $ = cheerio.load(response.data);
+
+                item.title = $('.xqnr_tit h2').text().trim();
+                item.author = $('.xqnr_tit span').eq(1).text().trim().replace('编辑：', '') || '山东大学计算机科学与技术学院';
+                $('.xqnr_tit').remove();
+                item.description = $('form[name=_newscontent_fromname]').html();
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: `山东大学计算机科学与技术学院${typelist[type]}通知`,
+        description: $('title').text(),
+        link,
+        item,
+    };
+};

package/lib/v2/sdu/data.js

@@ -0,0 +1,110 @@
+module.exports = {
+    wh: {
+        news: {
+            name: '山东大学（威海）新闻网',
+            route: '/news',
+            source: ['/*path', '/'],
+            titlePrefix: '(威海)新闻网|',
+            docs: 'https://docs.rsshub.app/university.html#shan-dong-da-xue-wei-hai',
+            getTarget(url) {
+                return this.route + '/' + url.replace(/\.htm$/, '');
+            },
+            url: 'https://xinwen.wh.sdu.edu.cn/',
+            columns: {
+                xyyw: {
+                    name: '校园要闻',
+                    url: 'xyyw.htm',
+                },
+                xsdt: {
+                    name: '学生动态',
+                    url: 'xsdt.htm',
+                },
+                zhxw: {
+                    name: '综合新闻',
+                    url: 'zhxw.htm',
+                },
+                sdsd: {
+                    name: '山大视点',
+                    url: 'sdsd.htm',
+                },
+                jjxy: {
+                    name: '菁菁校园',
+                    url: 'jjxy.htm',
+                },
+                xyjx: {
+                    name: '校园简讯',
+                    url: 'xyjx.htm',
+                },
+                mjzc: {
+                    name: '玛珈之窗',
+                    url: 'mjzc.htm',
+                },
+                rdzt: {
+                    name: '热点专题',
+                    url: 'rdzt.htm',
+                },
+                mtsj: {
+                    name: '媒体视角',
+                    url: 'mtsj.htm',
+                },
+                gjsy: {
+                    name: '高教视野',
+                    url: 'gjsy.htm',
+                },
+                llxx: {
+                    name: '理论学习',
+                    url: 'llxx.htm',
+                },
+            },
+        },
+        jwc: {
+            name: '山东大学（威海）教务处',
+            route: '/jwc',
+            source: ['/*path', '/'],
+            titlePrefix: '(威海)教务处|',
+            docs: 'https://docs.rsshub.app/university.html#shan-dong-da-xue-wei-hai',
+            getTarget(url) {
+                return this.route + '/' + url.replace(/\.htm$/, '');
+            },
+            url: 'https://jwc.wh.sdu.edu.cn/',
+            columns: {
+                gzzd: {
+                    name: '规章制度',
+                    url: 'gzzd.htm',
+                },
+                zyjs: {
+                    name: '专业建设',
+                    url: 'zyjs.htm',
+                },
+                sjjx: {
+                    name: '实践教学',
+                    url: 'sjjx.htm',
+                },
+                zbfc: {
+                    name: '支部风采',
+                    url: 'zbfc.htm',
+                },
+                fwzn: {
+                    name: '服务指南',
+                    url: 'fwzn.htm',
+                },
+                jwyw: {
+                    name: '教务要闻',
+                    url: 'jwyw.htm',
+                },
+                gztz: {
+                    name: '工作通知',
+                    url: 'gztz.htm',
+                },
+                jwjb: {
+                    name: '教务简报',
+                    url: 'jwjb.htm',
+                },
+                cyxz: {
+                    name: '常用下载',
+                    url: 'cyxz.htm',
+                },
+            },
+        },
+    },
+};

package/lib/v2/sdu/epe.js

@@ -0,0 +1,55 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+
+const host = 'https://www.epe.sdu.edu.cn/';
+const typelist = ['学院动态', '通知公告', '学术论坛'];
+const urlList = ['zxzx/xydt.htm', 'zxzx/tzgg.htm', 'zxzx/xslt.htm'];
+
+module.exports = async (ctx) => {
+    const type = ctx.params.type ? parseInt(ctx.params.type) : 0;
+    const link = new URL(urlList[type], host).href;
+    const response = await got(link);
+
+    const $ = cheerio.load(response.data);
+
+    let item = $('#page_right_main li a')
+        .map((_, e) => {
+            e = $(e);
+            return {
+                title: e.attr('title'),
+                link: e.attr('href'),
+            };
+        })
+        .get();
+
+    item = await Promise.all(
+        item
+            .filter((e) => e.link.startsWith('../info'))
+            .map((item) => {
+                item.link = new URL(item.link.slice('3'), host).href;
+                return ctx.cache.tryGet(item.link, async () => {
+                    const response = await got(item.link);
+                    const $ = cheerio.load(response.data);
+
+                    const info = $('#show_info').text().split(/\s{4}/);
+                    const date = info[0].split('：')[1];
+
+                    item.title = $('#show_title').text().trim();
+                    item.author = info[1].replace('编辑：', '') || '山东大学能源与动力工程学院';
+                    item.description = $('#show_content').html();
+                    item.pubDate = timezone(parseDate(date), +8);
+
+                    return item;
+                });
+            })
+    );
+
+    ctx.state.data = {
+        title: `山东大学能源与动力工程学院${typelist[type]}`,
+        description: $('title').text(),
+        link,
+        item,
+    };
+};

package/lib/v2/sdu/extractor/index.js

@@ -0,0 +1,15 @@
+module.exports = async (link, ctx) => {
+    if (link.startsWith('https://xinwen.wh.sdu.edu.cn/')) {
+        return await require('./wh/news')(link, ctx);
+    }
+    if (link.startsWith('https://www.view.sdu.edu.cn/')) {
+        return await require('./view')(link, ctx);
+    }
+    if (link.startsWith('https://www.sdrj.sdu.edu.cn/')) {
+        return await require('./sdrj')(link, ctx);
+    }
+    if (link.startsWith('https://jwc.wh.sdu.edu.cn/')) {
+        return await require('./wh/jwc')(link, ctx);
+    }
+    return {};
+};

package/lib/v2/sdu/extractor/sdrj.js

@@ -0,0 +1,21 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+
+module.exports = async (link, ctx) =>
+    await ctx.cache.tryGet(link, async () => {
+        let content, author, exactDate;
+        try {
+            const result = await got(link);
+            const $ = cheerio.load(result.data);
+            content = $('#vsb_content').html();
+            author = $("form[name='_newscontent_fromname'] > h1").text();
+            const exactDateLine = $("form[name='_newscontent_fromname'] > p.info").text().trim();
+            const exactDateText = exactDateLine.match(/^发布时间：(?<date>\d+\/\d+\/\d+\s\d{2}:\d{2}:\d{2})/).groups.date;
+            exactDate = timezone(parseDate(exactDateText, 'YYYY/MM/DD HH:mm:ss'), +8);
+            return { description: content, author, exactDate };
+        } catch (e) {
+            return { description: content, author, exactDate };
+        }
+    });

package/lib/v2/sdu/extractor/view.js

@@ -0,0 +1,21 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+
+module.exports = async (link, ctx) =>
+    await ctx.cache.tryGet(link, async () => {
+        let content, author, exactDate;
+        try {
+            const result = await got(link);
+            const $ = cheerio.load(result.data);
+            content = $('#vsb_content').html();
+            author = $("form[name='_newscontent_fromname'] > div > p:last-of-type").text();
+            const exactDateLine = $('.news_tit > p:last-child').text();
+            const exactDateText = exactDateLine.match(/^发布日期：(?<date>\d+年\d+月\d+日\s\d{2}:\d{2})/).groups.date;
+            exactDate = timezone(parseDate(exactDateText, 'YYYY年MM月DD日 HH:mm'), +8);
+            return { description: content, author, exactDate };
+        } catch (e) {
+            return { description: content, author, exactDate };
+        }
+    });

package/lib/v2/sdu/extractor/wh/jwc.js

@@ -0,0 +1,24 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+
+module.exports = async (link, ctx) =>
+    await ctx.cache.tryGet(link, async () => {
+        let content, exactDate;
+        try {
+            const result = await got(link);
+            const $ = cheerio.load(result.data);
+            const form = $("form[name='_newscontent_fromname']");
+            form.find('h2').remove();
+            const exactDateElement = form.find('#author');
+            const exactDateStr = exactDateElement.text();
+            exactDateElement.remove();
+            content = form.html();
+            const exactDateText = exactDateStr.match(/^创建时间：(?<date>\d{4}-\d{2}-\d{2}\s\d{2}:\d{2}:\d{2})/).groups.date;
+            exactDate = timezone(parseDate(exactDateText, 'YYYY-MM-DD HH:mm:ss'), +8);
+            return { description: content, exactDate };
+        } catch (e) {
+            return { description: content, exactDate };
+        }
+    });

package/lib/v2/sdu/extractor/wh/news.js

@@ -0,0 +1,21 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+
+module.exports = async (link, ctx) =>
+    await ctx.cache.tryGet(link, async () => {
+        let content, author, exactDate;
+        try {
+            const result = await got(link);
+            const $ = cheerio.load(result.data);
+            content = $('#main-content').html();
+            author = $('#source').text();
+            const exactDateLine = $('.news_tit > p:last-child').text();
+            const exactDateText = exactDateLine.match(/^发布日期：(?<date>\d+年\d+月\d+日\s\d{2}:\d{2})/).groups.date;
+            exactDate = timezone(parseDate(exactDateText, 'YYYY年MM月DD日 HH:mm'), +8);
+            return { description: content, author, exactDate };
+        } catch (e) {
+            return { description: content, author, exactDate };
+        }
+    });

package/lib/v2/sdu/maintainer.js

@@ -0,0 +1,9 @@
+module.exports = {
+    '/cmse/:type?': ['Ji4n1ng'],
+    '/cs/:type?': ['Ji4n1ng'],
+    '/epe/:type?': ['Ji4n1ng'],
+    '/mech/:type?': ['Ji4n1ng'],
+    '/sc/:type?': ['Ji4n1ng'],
+    '/wh/news/:column?': ['kxxt'],
+    '/wh/jwc/:column?': ['kxxt'],
+};

package/lib/v2/sdu/mech.js

@@ -0,0 +1,60 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+
+const typelist = ['通知公告', '院所新闻', '教学信息', '学术动态', '学院简报'];
+const urlList = ['xwdt/tzgg.htm', 'xwdt/ysxw.htm', 'xwdt/jxxx.htm', 'xwdt/xsdt.htm', 'xwdt/xyjb.htm'];
+const host = 'https://www.mech.sdu.edu.cn/';
+
+module.exports = async (ctx) => {
+    const type = ctx.params.type ? parseInt(ctx.params.type) : 0;
+    const link = new URL(urlList[type], host).href;
+    const response = await got(link);
+
+    const $ = cheerio.load(response.data);
+
+    let item = $('#page_list li a')
+        .slice(0, 1)
+        .map((_, e) => {
+            e = $(e);
+            return {
+                title: e.attr('title'),
+                link: e.attr('href'),
+            };
+        })
+        .get();
+
+    item = await Promise.all(
+        item
+            .filter((e) => e.link.startsWith('../info') || e.link.startsWith('https://www.rd.sdu.edu.cn/'))
+            .map((item) => {
+                const isFromMech = item.link.startsWith('../info');
+                if (isFromMech) {
+                    item.link = new URL(item.link.slice('3'), host).href;
+                }
+                return ctx.cache.tryGet(item.link, async () => {
+                    const response = await got(item.link);
+                    const $ = cheerio.load(response.data);
+
+                    const info = $('#show_info').text().split(/\s{4}/);
+                    const date = info[0].split('：')[1];
+
+                    item.title = $('#show_title').text().trim();
+                    item.author = info[1].replace('作者：', '') || '山东大学机械工程学院';
+                    $('#show_title, #show_info').remove();
+                    item.description = $('form[name=_newscontent_fromname] div').html();
+                    item.pubDate = timezone(parseDate(date), +8);
+
+                    return item;
+                });
+            })
+    );
+
+    ctx.state.data = {
+        title: `山东大学机械工程学院${typelist[type]}`,
+        description: $('title').text(),
+        link,
+        item,
+    };
+};