rsshub 1.0.0-master.f607442 → 1.0.0-master.f641a2a

package/lib/v2/segmentfault/user.js

@@ -1,7 +1,7 @@
 const got = require('@/utils/got');
 const cheerio = require('cheerio');
 const { parseDate } = require('@/utils/parse-date');
-
+const { acw_sc__v2 } = require('./utils');
 const host = 'https://segmentfault.com';
 
 module.exports = async (ctx) => {
@@ -19,10 +19,16 @@ module.exports = async (ctx) => {
         author,
     }));
 
+    const acwScV2Cookie = await acw_sc__v2(list[0].link, ctx.cache.tryGet);
+
     const items = await Promise.all(
         list.map((item) =>
             ctx.cache.tryGet(item.link, async () => {
-                const response = await got(item.link);
+                const response = await got(item.link, {
+                    headers: {
+                        cookie: `acw_sc__v2=${acwScV2Cookie};`,
+                    },
+                });
                 const content = cheerio.load(response.data);
 
                 item.description = content('article')

package/lib/v2/segmentfault/utils.js

@@ -0,0 +1,34 @@
+const zlib = require('zlib');
+const got = require('@/utils/got');
+const config = require('@/config').value;
+const { getAcwScV2ByArg1 } = require('@/v2/5eplay/utils');
+
+const acw_sc__v2 = (link, tryGet) =>
+    tryGet(
+        'segmentfault:acw_sc__v2',
+        async () => {
+            const response = await got(link, {
+                decompress: false,
+            });
+
+            const unzipData = zlib.createUnzip();
+            unzipData.write(response.body);
+
+            let acw_sc__v2 = '';
+            for await (const data of unzipData) {
+                const strData = data.toString();
+                const matches = strData.match(/var arg1='(.*?)';/);
+                if (matches) {
+                    acw_sc__v2 = getAcwScV2ByArg1(matches[1]);
+                    break;
+                }
+            }
+            return acw_sc__v2;
+        },
+        config.cache.routeExpire,
+        false
+    );
+
+module.exports = {
+    acw_sc__v2,
+};

package/lib/v2/sehuatang/maintainer.js

@@ -1,6 +1,7 @@
 module.exports = {
     '/bt/:subforumid?': ['qiwihui', 'junfengP', 'nczitzk'],
     '/picture/:subforumid': ['qiwihui', 'junfengP', 'nczitzk'],
+    '/user/:uid': ['JamYiz'],
     '/:subforumid?/:type?': ['qiwihui', 'junfengP', 'nczitzk'],
     '/:subforumid?': ['qiwihui', 'junfengP', 'nczitzk'],
     '/': ['qiwihui', 'junfengP', 'nczitzk'],

package/lib/v2/sehuatang/radar.js

@@ -4,7 +4,7 @@ module.exports = {
         '.': [
             {
                 title: '分区帖子',
-                docs: 'https://docs.rsshub.app/multimedia.html#se-hua-tang-fen-qu-tie-zi',
+                docs: 'https://docs.rsshub.app/routes/multimedia#se-hua-tang-fen-qu-tie-zi',
                 source: ['/:category', '/'],
                 target: (params, url) => {
                     const theUrl = new URL(url);

package/lib/v2/sehuatang/router.js

@@ -1,6 +1,7 @@
 module.exports = function (router) {
     router.get('/bt/:subforumid?', require('./index'));
     router.get('/picture/:subforumid', require('./index'));
+    router.get('/user/:uid', require('./user'));
     router.get('/:subforumid?/:type?', require('./index'));
     router.get('/:subforumid?', require('./index'));
     router.get('', require('./index'));

package/lib/v2/sehuatang/user.js

@@ -0,0 +1,113 @@
+// 导入必要的模组
+const got = require('@/utils/got'); // 自订的 got
+const cheerio = require('cheerio'); // 可以使用类似 jQuery 的 API HTML 解析器
+const { parseDate } = require('@/utils/parse-date');
+const config = require('@/config').value;
+
+const baseUrl = 'https://sehuatang.org/';
+
+module.exports = async (ctx) => {
+    if (!config.sehuatang.cookie) {
+        throw 'Sehuatang RSS is disabled due to the lack of <a href="https://docs.rsshub.app/install/#pei-zhi-bu-fen-rss-mo-kuai-pei-zhi">relevant config</a>';
+    }
+    // 从Url参数中获取uid
+    const uid = ctx.params.uid;
+    const link = `${baseUrl}home.php?mod=space&uid=${uid}&do=thread&view=me&from=space`;
+
+    const response = await got(link, {
+        headers: {
+            Cookie: config.sehuatang.cookie,
+            'Accept-Encoding': 'gzip, deflate, br',
+            'Accept-Language': 'zh-CN,zh;q=0.9,en;q=0.8',
+        },
+    });
+
+    const $ = cheerio.load(response.data);
+
+    const list = $('#delform tr:not(.th)')
+        .slice(0, ctx.query.limit ? parseInt(ctx.query.limit) : 25)
+        .toArray()
+        .map((item) => {
+            item = $(item);
+            const a = item.find('th>a').first();
+            const category = item.find('.xg1').first().text();
+            return {
+                title: `[${category}] ${a.text()}`,
+                // `link` 需要一个绝对 URL，但 `a.attr('href')` 返回一个相对 URL。
+                link: baseUrl + a.attr('href'),
+                // pubDate: '',
+                author: $('.mt').first().text(),
+            };
+        });
+
+    const out = await Promise.all(
+        list.map((info) =>
+            ctx.cache.tryGet(info.link, async () => {
+                const response = await got(info.link, {
+                    headers: {
+                        Cookie: config.sehuatang.cookie,
+                        'Accept-Encoding': 'gzip, deflate, br',
+                        'Accept-Language': 'zh-CN,zh;q=0.9,en;q=0.8',
+                    },
+                });
+
+                const $ = cheerio.load(response.data);
+                const postMessage = $("[id^='postmessage']").slice(0, 1);
+                const images = $(postMessage).find('img');
+                for (const image of images) {
+                    const file = $(image).attr('file');
+                    if (!file || file === 'undefined') {
+                        $(image).replaceWith('');
+                    } else {
+                        $(image).replaceWith($(`<img src="${file}">`));
+                    }
+                }
+                // if postMessage does not have any images, try to parse image url from `.pattl`
+                if (images.length === 0) {
+                    const pattl = $('.pattl');
+                    const pattlImages = $(pattl).find('img');
+                    for (const pattlImage of pattlImages) {
+                        const file = $(pattlImage).attr('file');
+                        if (!file || file === 'undefined') {
+                            $(pattlImage).replaceWith('');
+                        } else {
+                            $(pattlImage).replaceWith($(`<img src="${file}" />`));
+                        }
+                    }
+                    postMessage.append($(pattl));
+                }
+                $('em[onclick]').remove();
+
+                info.description = (postMessage.html() || '抓取原帖失败').replace(/ignore_js_op/g, 'div');
+
+                const dateString = $('.authi em').first().text();
+                const datestampString = dateString.split(' ')[1];
+                const timestampString = dateString.split(' ')[2];
+                const datetimeString = `${datestampString} ${timestampString}`;
+                const timestamp = new Date(datetimeString).getTime();
+
+                info.pubDate = $('.authi em span').length > 0 ? parseDate($('.authi em span').attr('title')) : parseDate(timestamp);
+
+                const magnet = postMessage.find('div.blockcode li').first().text();
+                const isMag = magnet.startsWith('magnet');
+                const torrent = postMessage.find('p.attnm a').attr('href');
+
+                const hasEnclosureUrl = isMag || torrent !== undefined;
+                if (hasEnclosureUrl) {
+                    const enclosureUrl = isMag ? magnet : new URL(torrent, baseUrl).href;
+                    info.enclosure_url = enclosureUrl;
+                    info.enclosure_type = isMag ? 'application/x-bittorrent' : 'application/octet-stream';
+                }
+
+                return info;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        // 在此处输出您的 RSS
+        title: `${$('.mt').text()}的帖子-色花堂`,
+        link,
+        item: out,
+    };
+};

package/lib/v2/sensortower/radar.js

@@ -4,7 +4,7 @@ module.exports = {
         '.': [
             {
                 title: 'Blog',
-                docs: 'https://docs.rsshub.app/new-media.html#sensor-tower-blog',
+                docs: 'https://docs.rsshub.app/routes/new-media#sensor-tower-blog',
                 source: ['/blog', '/zh-CN/blog', '/ja/blog', '/ko/blog', '/'],
                 target: '/sensortower/blog',
             },

package/lib/v2/setn/index.js

@@ -0,0 +1,100 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+
+const defaultRootUrl = 'https://www.setn.com';
+
+const rootUrls = {
+    娛樂: 'https://star.setn.com',
+    健康: 'https://health.setn.com',
+    旅遊: 'https://travel.setn.com',
+    富房網: 'https://fuhouse.setn.com',
+    女孩: 'https://watch.setn.com',
+};
+
+const ids = {
+    即時: '',
+    熱門: 0,
+    政治: 6,
+    社會: 41,
+    國際: 5,
+    兩岸: 68,
+    生活: 4,
+    運動: 34,
+    地方: 97,
+    財經: 2,
+    名家: 9,
+    新奇: 42,
+    科技: 7,
+    汽車: 12,
+    寵物: 47,
+    HOT焦點: 31,
+};
+
+const getCurrentUrl = (category) => {
+    const rootUrl = rootUrls.hasOwnProperty(category) ? rootUrls[category] : defaultRootUrl;
+
+    if (ids.hasOwnProperty(category)) {
+        return `${rootUrl}/ViewAll.aspx${ids[category] === '' ? '' : `?PageGroupID=${ids[category]}`}`;
+    }
+
+    return `${rootUrl}/viewall`;
+};
+
+module.exports = async (ctx) => {
+    const category = ctx.params.category ?? '即時';
+    const limit = ctx.query.limit ? parseInt(ctx.query.limit) : 42;
+
+    const currentUrl = getCurrentUrl(category);
+
+    const response = await got({
+        method: 'get',
+        url: currentUrl,
+    });
+
+    const $ = cheerio.load(response.data);
+
+    let items = $('#NewsList, .newsList, .hotNewsList')
+        .find('.newsItems, .st-news, .all_three_list, div.title-word')
+        .slice(0, limit)
+        .toArray()
+        .map((item) => {
+            item = $(item);
+
+            const a = item.find('a').last();
+            const link = a.attr('href').replace(/(\?|&)utm_campaign=.*/g, '');
+
+            return {
+                title: a.text(),
+                link: /^http/.test(link) ? link : `${rootUrls.hasOwnProperty(category) ? rootUrls[category] : defaultRootUrl}${link}`,
+            };
+        });
+
+    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);
+
+                content('#gad_setn_innity_oop_1x1').remove();
+
+                item.author = content('meta[property="author"]').attr('content');
+                item.category = [content('meta[property="article:section"]').attr('content'), ...content('meta[name="news_keywords"]').attr('content').split(',')];
+                item.pubDate = parseDate(content('meta[property="article:published_time"]').attr('content'));
+                item.description = content('article, .content-p').html();
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: `三立新聞網 - ${category}`,
+        link: currentUrl,
+        item: items,
+    };
+};

package/lib/v2/setn/maintainer.js

@@ -0,0 +1,3 @@
+module.exports = {
+    '/:category?': ['nczitzk'],
+};

package/lib/v2/setn/radar.js

@@ -0,0 +1,53 @@
+module.exports = {
+    'setn.com': {
+        _name: '三立新聞網',
+        '.': [
+            {
+                title: '新聞',
+                docs: 'https://docs.rsshub.app/routes/traditional-media#san-li-xin-wen-wang-xin-wen',
+                source: ['/ViewAll.aspx', '/'],
+                target: '/setn',
+            },
+        ],
+        star: [
+            {
+                title: '娛樂新聞',
+                docs: 'https://docs.rsshub.app/routes/traditional-media#san-li-xin-wen-wang-xin-wen',
+                source: ['/viewall', '/'],
+                target: '/setn/娛樂',
+            },
+        ],
+        health: [
+            {
+                title: '健康新聞',
+                docs: 'https://docs.rsshub.app/routes/traditional-media#san-li-xin-wen-wang-xin-wen',
+                source: ['/viewall', '/'],
+                target: '/setn/健康',
+            },
+        ],
+        travel: [
+            {
+                title: '旅遊新聞',
+                docs: 'https://docs.rsshub.app/routes/traditional-media#san-li-xin-wen-wang-xin-wen',
+                source: ['/viewall', '/'],
+                target: '/setn/旅遊',
+            },
+        ],
+        fuhouse: [
+            {
+                title: '富房網新聞',
+                docs: 'https://docs.rsshub.app/routes/traditional-media#san-li-xin-wen-wang-xin-wen',
+                source: ['/viewall', '/'],
+                target: '/setn/富房網',
+            },
+        ],
+        watch: [
+            {
+                title: '女孩新聞',
+                docs: 'https://docs.rsshub.app/routes/traditional-media#san-li-xin-wen-wang-xin-wen',
+                source: ['/viewall', '/'],
+                target: '/setn/女孩',
+            },
+        ],
+    },
+};

package/lib/v2/setn/router.js

@@ -0,0 +1,3 @@
+module.exports = function (router) {
+    router.get('/:category?', require('./index'));
+};

package/lib/v2/seu/cse/index.js

@@ -0,0 +1,63 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+
+module.exports = async (ctx) => {
+    const host = 'https://cse.seu.edu.cn';
+
+    const map = {
+        22535: { initial: 'xyxw' },
+        22536: { initial: 'tzgg' },
+        22538: { initial: 'jwxx' },
+        22537: { initial: 'jyxx' },
+        22539: { initial: 'xgsw' },
+    };
+
+    const { type = 22535 } = ctx.params;
+    const id = type.length === 4 ? Object.keys(map).find((key) => map[key].initial === type) : parseInt(type); // backward compatible
+    const link = new URL(`${id}/list.htm`, host).href;
+
+    const { data: response } = await got(link);
+    const $ = cheerio.load(response);
+
+    const list = $('.news_list .news')
+        .toArray()
+        .map((e) => {
+            e = $(e);
+            const a = e.find('.news_title a');
+            return {
+                title: a.attr('title'),
+                link: new URL(a.attr('href'), host).href,
+                pubDate: parseDate(e.find('.news_meta').text()),
+            };
+        });
+
+    const out = await Promise.all(
+        list.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                let response;
+                try {
+                    response = await got(item.link);
+                } catch (error) {
+                    // intranet
+                    if (error.response.url.startsWith('https://newids.seu.edu.cn/')) {
+                        return item;
+                    }
+                    throw error;
+                }
+                const $ = cheerio.load(response.data);
+
+                item.description = $('div.wp_articlecontent').html();
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        link,
+        title: `${$('meta[name=keywords]').attr('content')}${$('meta[name=description]').attr('content')} -- ${$('head title').text()}`,
+        description: '东南大学计算机技术与工程学院RSS',
+        item: out,
+    };
+};

package/lib/v2/seu/maintainer.js

@@ -0,0 +1,6 @@
+module.exports = {
+    '/cse/:type?': ['LogicJake'],
+    '/radio/academic': ['HenryQW'],
+    '/yjs': ['Denkiyohou'],
+    '/yzb/:type': ['fuzy112'],
+};

package/lib/v2/seu/radar.js

@@ -0,0 +1,37 @@
+module.exports = {
+    'seu.edu.cn': {
+        _name: '东南大学',
+        cse: [
+            {
+                title: '计算机技术与工程学院',
+                docs: 'https://docs.rsshub.app/routes/university#dong-nan-da-xue',
+                source: ['/:type/list.htm', '/'],
+                target: '/seu/cse/:type?',
+            },
+        ],
+        radio: [
+            {
+                title: '信息科学与工程学院学术活动',
+                docs: 'https://docs.rsshub.app/routes/university#dong-nan-da-xue',
+                source: ['/_s29/15986/list.psp', '/'],
+                target: '/seu/radio/academic',
+            },
+        ],
+        seugs: [
+            {
+                title: '研究生院全部公告',
+                docs: 'https://docs.rsshub.app/routes/university#dong-nan-da-xue-yan-jiu-sheng-yuan-quan-bu-gong-gao',
+                source: ['/26671/list.htm', '/'],
+                target: '/seu/yjs',
+            },
+        ],
+        yzb: [
+            {
+                title: '研究生招生网通知公告',
+                docs: 'https://docs.rsshub.app/routes/university#dong-nan-da-xue',
+                source: ['/:type/list.htm'],
+                target: '/seu/yzb/:type',
+            },
+        ],
+    },
+};

package/lib/v2/seu/radio/academic.js

@@ -0,0 +1,43 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+
+module.exports = async (ctx) => {
+    const host = 'https://radio.seu.edu.cn';
+    const link = new URL('_s29/15986/list.psp', host).href;
+    const response = await got(link);
+
+    const $ = cheerio.load(response.data);
+
+    const list = $('.list_item')
+        .toArray()
+        .map((e) => {
+            e = $(e);
+            const a = e.find('.Article_Title a');
+            return {
+                title: a.attr('title'),
+                link: new URL(a.attr('href'), host).href,
+                pubDate: parseDate(e.find('.Article_PublishDate').text()),
+            };
+        });
+
+    const out = await Promise.all(
+        list.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const response = await got(item.link);
+                const $ = cheerio.load(response.data);
+
+                item.author = $('.arti_publisher').text().replace('发布者：', '');
+                item.description = $('.wp_articlecontent').html();
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: '东南大学信息科学与工程学院 -- 学术活动',
+        link,
+        item: out,
+    };
+};

package/lib/v2/seu/router.js

@@ -0,0 +1,6 @@
+module.exports = (router) => {
+    router.get('/cse/:type?', require('./cse'));
+    router.get('/radio/academic', require('./radio/academic'));
+    router.get('/yjs', require('./yjs'));
+    router.get('/yzb/:type', require('./yzb'));
+};

package/lib/v2/seu/yjs.js

@@ -0,0 +1,37 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+
+const url = 'https://seugs.seu.edu.cn/26671/list.htm';
+
+module.exports = async (ctx) => {
+    const response = await got(url);
+    const $ = cheerio.load(response.data);
+    const list = $('.news')
+        .toArray()
+        .map((element) => {
+            const info = {
+                title: $(element).find('span.news_title > a').attr('title'),
+                link: `https://seugs.seu.edu.cn${$(element).find('span.news_title > a').attr('href')}`,
+                date: $(element).find('span.news_meta').text(),
+            };
+            return info;
+        });
+
+    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 = $('.wp_articlecontent').html();
+                item.pubDate = new Date(item.date).toUTCString();
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: '东南大学研究生公告',
+        link: url,
+        item: items,
+    };
+};

package/lib/v2/seu/yzb/index.js

@@ -0,0 +1,51 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+
+module.exports = async (ctx) => {
+    const host = 'https://yzb.seu.edu.cn';
+
+    const map = {
+        6676: { id: 1 },
+        6677: { id: 2 },
+        6679: { id: 3 },
+    };
+
+    const { type } = ctx.params;
+    const id = type.length === 1 ? Object.keys(map).find((key) => map[key].id === parseInt(type)) : parseInt(type); // backward compatible
+    const url = new URL(`${id}/list.htm`, host).href;
+
+    const { data: response } = await got(url);
+    const $ = cheerio.load(response);
+
+    const list = $('#wp_news_w3 td tbody tr')
+        .toArray()
+        .map((elem) => {
+            elem = $(elem);
+            const a = elem.find('td a');
+            return {
+                title: a.attr('title'),
+                link: new URL(a.attr('href'), host).href,
+                pubDate: parseDate(elem.find('td div').text()),
+            };
+        });
+
+    const items = await Promise.all(
+        list.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const { data: response } = await got(item.link);
+                const $ = cheerio.load(response);
+
+                item.description = $('.Article_Content').html();
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        link: url,
+        title: `东南大学研究生招生网 -- ${$('head title').text()}`,
+        item: items,
+    };
+};

package/lib/v2/shcstheatre/maintainer.js

@@ -0,0 +1,3 @@
+module.exports = {
+    '/programs': ['fuzy112'],
+};

package/lib/v2/shcstheatre/programs.js

@@ -0,0 +1,39 @@
+const got = require('@/utils/got');
+const { art } = require('@/utils/render');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+const cheerio = require('cheerio');
+const path = require('path');
+
+module.exports = async (ctx) => {
+    const url = 'https://www.shcstheatre.com/Program/programList.aspx';
+    const res = await got.get(url);
+    const $ = cheerio.load(res.data);
+    const items = await Promise.all(
+        $('#datarow .program-name a').map((_, elem) => {
+            const link = new URL($(elem).attr('href'), url);
+            return ctx.cache.tryGet(link.toString(), async () => {
+                const id = link.searchParams.get('id');
+                const res2 = await got.post('https://www.shcstheatre.com/webapi.ashx?op=GettblprogramCache', {
+                    headers: { 'Content-Type': 'application/x-www-form-urlencoded; charset=UTF-8' },
+                    form: { id },
+                });
+                const data = res2.data.data.tblprogram[0];
+                return {
+                    title: data.SCS_WEB_BRIEFNAME,
+                    link: link.toString(),
+                    description: art(path.join(__dirname, 'templates/description.art'), data),
+                    pubDate: timezone(parseDate(data.SJ_DATE_PC), +8),
+                };
+            });
+        })
+    );
+    const image = $('.menu-logo img').attr('src');
+
+    ctx.state.data = {
+        title: '上海文化广场 - 节目列表',
+        link: url,
+        image,
+        item: items,
+    };
+};