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

package/lib/v2/tingshuitz/radar.js

@@ -0,0 +1,68 @@
+module.exports = {
+    'dl.gov.cn': {
+        _name: '停水通知',
+        swj: [
+            {
+                title: '大连市',
+                docs: 'https://docs.rsshub.app/forecast.html#ting-shui-tong-zhi',
+                source: ['/col/col4296/index.html', '/'],
+                target: '/tingshuitz/dalian',
+            },
+        ],
+    },
+    'hzwgc.com': {
+        _name: '停水通知',
+        www: [
+            {
+                title: '杭州市',
+                docs: 'https://docs.rsshub.app/forecast.html#ting-shui-tong-zhi',
+                source: ['/public/stop_the_water', '/'],
+                target: '/tingshuitz/hangzhou',
+            },
+        ],
+    },
+    'jlwater.com': {
+        _name: '停水通知',
+        '.': [
+            {
+                title: '南京市',
+                docs: 'https://docs.rsshub.app/forecast.html#ting-shui-tong-zhi',
+                source: ['/portal/10000013', '/'],
+                target: '/tingshuitz/nanjing',
+            },
+        ],
+    },
+    'whwater.com': {
+        _name: '停水通知',
+        '.': [
+            {
+                title: '武汉市',
+                docs: 'https://docs.rsshub.app/forecast.html#ting-shui-tong-zhi',
+                source: ['/IWater.shtml', '/'],
+                target: '/tingshuitz/wuhan',
+            },
+        ],
+    },
+    'xswater.com': {
+        _name: '停水通知',
+        www: [
+            {
+                title: '萧山区',
+                docs: 'https://docs.rsshub.app/forecast.html#ting-shui-tong-zhi',
+                source: ['/gongshui/channels/227.html', '/'],
+                target: '/tingshuitz/xiaoshan',
+            },
+        ],
+    },
+    'yjsswjt.com': {
+        _name: '停水通知',
+        '.': [
+            {
+                title: '阳江市',
+                docs: 'https://docs.rsshub.app/forecast.html#ting-shui-tong-zhi',
+                source: ['/zxdt_list.jsp', '/'],
+                target: '/tingshuitz/yangjiang',
+            },
+        ],
+    },
+};

package/lib/v2/tingshuitz/router.js

@@ -0,0 +1,11 @@
+module.exports = (router) => {
+    router.get('/dalian', require('./dalian'));
+    router.get('/dongguan', require('./dongguan'));
+    router.get('/guangzhou', require('./guangzhou'));
+    router.get('/hangzhou', require('./hangzhou'));
+    router.get('/nanjing', require('./nanjing'));
+    router.get('/wuhan/:channelId?', require('./wuhan'));
+    router.get('/xian', require('./xian'));
+    router.get('/xiaoshan', require('./xiaoshan'));
+    router.get('/yangjiang', require('./yangjiang'));
+};

package/lib/v2/tingshuitz/wuhan.js

@@ -0,0 +1,32 @@
+const got = require('@/utils/got');
+const { parseDate } = require('@/utils/parse-date');
+const baseUrl = 'https://www.whwater.com';
+
+module.exports = async (ctx) => {
+    const { channelId = 68 } = ctx.params;
+    const response = await got.post('https://manager.whwater.com:8900/website/article/findChannelArticle', {
+        form: {
+            channelId,
+            searchKey: '',
+            thumbnailStatus: 0,
+            topStatus: 0,
+            recommendStatus: 0,
+            page: 1,
+            size: ctx.query.limit ? parseInt(ctx.query.limit) : 30,
+        },
+    });
+
+    const data = response.data.data;
+    const items = data.articleArray.map((item) => ({
+        title: item.title,
+        description: item.content,
+        pubDate: parseDate(item.publishTime),
+        link: `https://${baseUrl}/IPolicyDetails.shtml?id=31&sid=${channelId}${item.articleLink}`,
+    }));
+
+    ctx.state.data = {
+        title: `${data.channelName}通知 - 武汉市水务集团有限公司`,
+        link: `${baseUrl}/IWater.shtml?id=31&sid=48`,
+        item: items,
+    };
+};

package/lib/{routes → v2}/tingshuitz/xian.js

@@ -4,7 +4,7 @@ const iconv = require('iconv-lite');
 
 module.exports = async (ctx) => {
     const url = 'http://www.xazls.com/tsgg/index.htm';
-    const response = await got.get(url, {
+    const response = await got(url, {
         responseType: 'buffer',
     });
 

package/lib/{routes → v2}/tingshuitz/yangjiang.js

@@ -1,20 +1,17 @@
 const got = require('@/utils/got');
 const cheerio = require('cheerio');
-const iconv = require('iconv-lite');
+const { parseDate } = require('@/utils/parse-date');
 
 module.exports = async (ctx) => {
-    const url = 'http://www.yjsswjt.com/zxdt_list.jsp?flbz=7';
-    const response = await got.get(url, {
-        responseType: 'buffer',
-    });
+    const url = 'https://www.yjsswjt.com/zxdt_list.jsp?flbz=7';
+    const response = await got(url);
 
-    const data = response.data;
-    const $ = cheerio.load(iconv.decode(data, 'gb2312'));
+    const $ = cheerio.load(response.data);
     const list = $('div.list_ul_div > ul > li');
 
     ctx.state.data = {
         title: '停水通知 - 阳江市水务集团有限公司',
-        link: 'http://www.yjsswjt.com/zxdt_list.jsp?flbz=7',
+        link: 'https://www.yjsswjt.com/zxdt_list.jsp?flbz=7',
         item: list
             .map((_, el) => {
                 const item = $(el);
@@ -24,6 +21,7 @@ module.exports = async (ctx) => {
                     title: item.find('span').text().trim(),
                     description: item.find('span').text().trim(),
                     link: 'http://www.yjsswjt.com/list.jsp?id=' + id,
+                    pubDate: parseDate(item.find('.datetime').text(), 'YYYY.MM.DD'),
                 };
             })
             .get(),

package/lib/v2/uraaka-joshi/maintainer.js

@@ -0,0 +1,4 @@
+module.exports = {
+    '/': ['SettingDust', 'Halcao'],
+    '/:id': ['SettingDust', 'Halcao'],
+};

package/lib/v2/uraaka-joshi/radar.js

@@ -0,0 +1,19 @@
+module.exports = {
+    'uraaka-joshi.com': {
+        _name: '裏垢女子まとめ',
+        '.': [
+            {
+                title: '主页',
+                docs: 'https://docs-rsshub.pages.dev/other.html#li-gou-nu-zi-まとめ',
+                source: ['/'],
+                target: '/uraaka-joshi',
+            },
+            {
+                title: '用户',
+                docs: 'https://docs-rsshub.pages.dev/other.html#li-gou-nu-zi-まとめ',
+                source: ['/:id'],
+                target: '/uraaka-joshi/:id',
+            },
+        ],
+    },
+};

package/lib/v2/uraaka-joshi/router.js

@@ -0,0 +1,4 @@
+module.exports = (router) => {
+    router.get('/', require('./uraaka-joshi'));
+    router.get('/:id', require('./uraaka-joshi-user'));
+};

package/lib/v2/uraaka-joshi/uraaka-joshi-user.js

@@ -0,0 +1,89 @@
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const config = require('@/config').value;
+
+module.exports = async (ctx) => {
+    const { id } = ctx.params;
+    const link = `https://www.uraaka-joshi.com/user/${id}`;
+
+    const response = await ctx.cache.tryGet(
+        link,
+        async () => {
+            const browser = await require('@/utils/puppeteer')();
+            const page = await browser.newPage();
+
+            await page.goto(link);
+            await page.waitForSelector('#pickup03 .grid-cell');
+            await page.waitForSelector('#pickup04 .grid-cell');
+            await page.waitForSelector('#main-block .grid-cell');
+
+            const html = await page.evaluate(() => document.documentElement.innerHTML);
+            browser.close();
+            return html;
+        },
+        config.cache.routeExpire,
+        false
+    );
+
+    const $ = cheerio.load(response);
+    const items = $('#main-block .grid .grid-cell')
+        .toArray()
+        .map((item) => {
+            item = $(item);
+
+            // remove event and styles
+            item.find('*').removeAttr('onclick');
+            item.find('*').removeAttr('onerror');
+            item.find('*').removeAttr('style');
+
+            // format account style
+            const account = item.find('.account-group-link-row');
+            account.html(account.text());
+
+            // extract video tag from its player
+            item.find('.plyr--video').each((_, player) => {
+                player = $(player);
+
+                const v = player.find('video');
+                player.replaceWith(v);
+                v.attr('poster', 'https:' + v.attr('data-poster'));
+                v.find('source').attr('src', 'https:' + v.find('source').attr('src'));
+            });
+
+            // correct src of img tags
+            item.find('img').each((_, i) => {
+                i = $(i);
+                i.attr('src', 'https:' + i.attr('data-src').split('?resize')[0]);
+                i.removeAttr('data-src');
+            });
+
+            const author = item.find('.account-group').text();
+            const categories = item
+                .find('.hashtag-item .hashtag')
+                .toArray()
+                .map((c) => $(c).text().trim());
+            const link = item.find('.account-group-link-row').attr('href');
+            const date = parseDate(item.find('.profile-char').attr('datetime'));
+            const guid = item.find('a.tap-image').attr('data-tweet-id') || item.find('video[class^="js-player-"]').attr('data-tweet-id');
+
+            item.find('.grow-room').remove();
+            item.find('div.profile-group.mt10.prl2').eq(1).remove();
+
+            return {
+                title: item.find('.profile-text').text(),
+                description: item.html(),
+                link,
+                pubDate: date,
+                guid,
+                category: categories,
+                author,
+            };
+        });
+
+    ctx.state.data = {
+        title: $('title').text(),
+        description: $('meta[name="description"]').attr('content'),
+        link,
+        item: items,
+    };
+};

package/lib/v2/uraaka-joshi/uraaka-joshi.js

@@ -0,0 +1,71 @@
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+
+module.exports = async (ctx) => {
+    const link = `https://www.uraaka-joshi.com/`;
+    const title = `裏垢女子まとめ`;
+
+    const browser = await require('@/utils/puppeteer')();
+
+    const page = await browser.newPage();
+    await page.goto(link);
+
+    await page.waitForSelector('#pickup03 .grid-cell');
+    await page.waitForSelector('#pickup04 .grid-cell');
+    await page.waitForSelector('#main-block .grid-cell');
+
+    const bodyHandle = await page.$('body');
+    const html = await page.evaluate((body) => body.innerHTML, bodyHandle);
+    browser.close();
+    const $ = cheerio.load(html);
+    const list = $('.grid-cell');
+
+    ctx.state.data = {
+        title,
+        link,
+        item:
+            list &&
+            list
+                .map((index, item) => {
+                    item = $(item);
+
+                    // remove event and styles
+                    item.find('*').removeAttr('onclick');
+                    item.find('*').removeAttr('onerror');
+                    item.find('*').removeAttr('style');
+
+                    // format account style
+                    const account = item.find('.account-group-link-row');
+                    account.html(account.text());
+
+                    // extract video tag from its player
+                    item.find('.plyr--video').each((_, player) => {
+                        player = $(player);
+
+                        const video = player.find('video');
+                        player.replaceWith(video);
+                        const poster = video.attr('data-poster');
+                        video.attr('poster', 'https:' + poster);
+
+                        const source = video.find('source');
+                        const src = source.attr('src');
+                        source.attr('src', 'https:' + src);
+                    });
+
+                    // correct src of img tags
+                    item.find('img').each((_, image) => {
+                        const src = $(image).attr('data-src');
+                        $(image).attr('src', 'https:' + src);
+                    });
+
+                    return {
+                        title: item.find('.account-group').text() + ` - ${title}`,
+                        description: item.html(),
+                        link: item.find('.account-group-link-row').attr('href'),
+                        pubDate: parseDate(item.find('.profile-char').attr('datetime')),
+                        guid: item.find('a.tap-image').attr('data-tweet-id') || item.find('video[class^="js-player-"]').attr('data-tweet-id') || parseDate(item.find('.profile-char').attr('datetime')).getTime(),
+                    };
+                })
+                .get(),
+    };
+};

package/lib/v2/ustc/eeis.js

@@ -0,0 +1,59 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+
+const map = new Map([
+    ['tzgg', { title: '中国科学技术大学电子工程与信息科学系 - 通知公告', id: '2702' }],
+    ['xwxx', { title: '中国科学技术大学电子工程与信息科学系 - 新闻信息', id: '2706' }],
+]);
+
+const host = 'https://eeis.ustc.edu.cn';
+
+module.exports = async (ctx) => {
+    const type = ctx.params.type ?? 'tzgg';
+    const info = map.get(type);
+    if (!info) {
+        throw 'invalid type';
+    }
+    const id = info.id;
+
+    const response = await got(`${host}/${id}/list.htm`);
+    const $ = cheerio.load(response.data);
+    const list = $('div[portletmode=simpleList]')
+        .find('div.newslist')
+        .toArray()
+        .map((item) => {
+            item = $(item);
+            const title = item.find('a').attr('title').trim();
+            const link = item.find('a').attr('href').startsWith('/') ? host + item.find('a').attr('href') : item.find('a').attr('href');
+            const pubDate = timezone(parseDate(item.find('.text-secondary').text() + '/' + item.find('.text-primary').text(), 'YYYY/MM/DD'), +8);
+            return {
+                title,
+                pubDate,
+                link,
+            };
+        });
+
+    const items = await Promise.all(
+        list.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                let desc = '';
+                try {
+                    const response = await got(item.link);
+                    desc = cheerio.load(response.data)('div.wp_articlecontent').html();
+                    item.description = desc;
+                } catch (err) {
+                    // intranet only contents
+                }
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: info.title,
+        link: `${host}/${id}/list.htm`,
+        item: items,
+    };
+};

package/lib/v2/ustc/gs.js

@@ -0,0 +1,59 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+
+const map = new Map([
+    ['tzgg', { title: '中国科学技术大学研究生院 - 通知公告', id: '9' }],
+    ['xwdt', { title: '中国科学技术大学研究生院 - 新闻动态', id: '10' }],
+]);
+
+const host = 'https://gradschool.ustc.edu.cn';
+
+module.exports = async (ctx) => {
+    const type = ctx.params.type ?? 'tzgg';
+    const info = map.get(type);
+    if (!info) {
+        throw 'invalid type';
+    }
+    const id = info.id;
+
+    const response = await got(`${host}/column/${id}`);
+    const $ = cheerio.load(response.data);
+    let items = $('div.r-box > ul')
+        .find('li')
+        .toArray()
+        .map((item) => {
+            item = $(item);
+            const title = item.find('a').text().trim();
+            const link = item.find('a').attr('href').startsWith('/article') ? host + item.find('a').attr('href') : item.find('a').attr('href');
+            const pubDate = timezone(parseDate(item.find('time').text(), 'YYYY-MM-DD'), +8);
+            return {
+                title,
+                pubDate,
+                link,
+            };
+        });
+
+    items = await Promise.all(
+        items.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                let desc = '';
+                try {
+                    const response = await got(item.link);
+                    desc = cheerio.load(response.data)('article.article').html();
+                    item.description = desc;
+                } catch (err) {
+                    // intranet only contents
+                }
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: info.title,
+        link: `${host}/column/${id}`,
+        item: items,
+    };
+};

package/lib/v2/ustc/index.js

@@ -0,0 +1,79 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+// const UA = require('@/utils/rand-user-agent')({ browser: 'chrome', os: 'windows', device: 'desktop' });
+
+const notice_type = {
+    jx: { title: '中国科学技术大学 - 教学类通知', url: 'https://ustc.edu.cn/tzgg/jxltz.htm' },
+    ky: { title: '中国科学技术大学 - 科研类通知', url: 'https://ustc.edu.cn/tzgg/kyltz.htm' },
+    gl: { title: '中国科学技术大学 - 管理类通知', url: 'https://ustc.edu.cn/tzgg/glltz.htm' },
+    fw: { title: '中国科学技术大学 - 服务类通知', url: 'https://ustc.edu.cn/tzgg/fwltz.htm' },
+};
+
+// 对防抓的措施
+// function getCookie(ctx) {
+//     const cache_key = `ustc-cookie-${new Date().toLocaleDateString()}`;
+//     return ctx.cache.tryGet(cache_key, async () => {
+//         const { headers } = await got('https://ustc.edu.cn/system/resource/code/datainput.jsp', {
+//             headers: { 'user-agent': UA },
+//         });
+//         const cookie = headers['set-cookie']
+//             .filter((c) => c.match(/(user_trace_token|X_HTTP_TOKEN)/))
+//             .map((c) => c.split(';')[0])
+//             .join('; ');
+//         return cookie;
+//     });
+// }
+
+module.exports = async (ctx) => {
+    const type = ctx.params.type ?? 'gl';
+    // 发起 HTTP GET 请求
+    const response = await got({
+        method: 'get',
+
+        /* headers: {
+            'user-agent': UA,
+            cookie: await getCookie(ctx),
+        }, */
+        url: notice_type[type].url,
+    });
+
+    const $ = cheerio.load(response.data);
+    let items = $('table[portletmode=simpleList] > tbody > tr.light')
+        .map(function () {
+            const child = $(this).children();
+            const info = {
+                title: $(child[1]).find('a').attr('title'),
+                link: $(child[1]).find('a').attr('href').startsWith('../') ? new URL($(child[1]).find('a').attr('href'), notice_type[type].url).href : $(child[1]).find('a').attr('href'),
+                pubDate: timezone(parseDate($(child[2]).text(), 'YYYY-MM-DD'), +8),
+            };
+            return info;
+        })
+        .get();
+
+    items = await Promise.all(
+        items
+            .filter((item) => item.link)
+            .map((item) =>
+                ctx.cache.tryGet(item.link, async () => {
+                    try {
+                        const response = await got(item.link);
+                        const $ = cheerio.load(response.data);
+
+                        item.description = $('div.v_news_content').html();
+                    } catch {
+                        // intranet contents
+                    }
+                    return item;
+                })
+            )
+    );
+
+    ctx.state.data = {
+        title: notice_type[type].title,
+        description: notice_type[type].title,
+        link: notice_type[type].url,
+        item: items,
+    };
+};

package/lib/{routes/universities → v2}/ustc/job.js

@@ -1,4 +1,5 @@
 const got = require('@/utils/got');
+const { parseDate } = require('@/utils/parse-date');
 
 const titles = {
     RecruitList: '专场招聘会',
@@ -8,7 +9,7 @@ const titles = {
 };
 
 module.exports = async (ctx) => {
-    const category = ctx.params.category || 'joblist2';
+    const category = ctx.params.category ?? 'joblist2';
 
     const rand = 0.01234567890123456;
     const apiRootUrl = 'http://ustc.ahbys.com';
@@ -65,26 +66,26 @@ module.exports = async (ctx) => {
             case 'RecruitList':
                 return {
                     title: item.ID,
-                    pubDate: new Date(item.HoldDate),
+                    pubDate: parseDate(item.HoldDate),
                     link: `${rootUrl}/Recruit.html?id=${item.ID}`,
                 };
             case 'Doublechoice':
                 return {
                     ID: item.ID,
                     title: item.Theme,
-                    pubDate: new Date(item.HoldDate),
+                    pubDate: parseDate(item.HoldDate),
                     link: `${rootUrl}/R2.html?id=${item.ID}`,
                 };
             case 'Broadcast':
                 return {
                     title: item.ID,
-                    pubDate: new Date(item.HoldDate),
+                    pubDate: parseDate(item.HoldDate),
                     link: `${rootUrl}/R2.html?id=${item.ID}`,
                 };
             case 'joblist2':
                 return {
                     title: item.JID,
-                    pubDate: new Date(item.UpdateDate),
+                    pubDate: parseDate(item.UpdateDate),
                     link: `${rootUrl}/Job2.html?jid=${item.JID}&cid=${item.CompanyID}`,
                 };
             default: