rsshub 1.0.0-master.f2ed98e → 1.0.0-master.f2eee10

package/lib/v2/hk01/latest.js

@@ -0,0 +1,20 @@
+const got = require('@/utils/got');
+const { rootUrl, apiRootUrl, ProcessItems } = require('./utils');
+
+module.exports = async (ctx) => {
+    const currentUrl = `${rootUrl}/latest`;
+    const apiUrl = `${apiRootUrl}/v2/page/latest`;
+
+    const response = await got({
+        method: 'get',
+        url: apiUrl,
+    });
+
+    const items = await ProcessItems(response.data.items, ctx.query.limit, ctx.cache.tryGet);
+
+    ctx.state.data = {
+        title: '即時 | 香港01',
+        link: currentUrl,
+        item: items,
+    };
+};

package/lib/v2/hk01/maintainer.js

@@ -1,6 +1,7 @@
 module.exports = {
     '/channel/:id': ['hoilc', 'Fatpandac', 'nczitzk'],
     '/hot': ['hoilc', 'Fatpandac', 'nczitzk'],
+    '/latest': ['5upernova-heng'],
     '/issue/:id': ['hoilc', 'Fatpandac', 'nczitzk'],
     '/tag/:id': ['hoilc', 'Fatpandac', 'nczitzk'],
     '/zone/:id': ['hoilc', 'Fatpandac', 'nczitzk'],

package/lib/v2/hk01/radar.js

@@ -4,34 +4,40 @@ module.exports = {
         '.': [
             {
                 title: '热门',
-                docs: 'https://docs.rsshub.app/traditional-media.html#xiang-gang-01-re-men',
+                docs: 'https://docs.rsshub.app/new-media.html#xiang-gang-01-re-men',
                 source: ['/hot', '/'],
                 target: '/hk01/hot',
             },
             {
                 title: '栏目',
-                docs: 'https://docs.rsshub.app/traditional-media.html#xiang-gang-01-lan-mu',
+                docs: 'https://docs.rsshub.app/new-media.html#xiang-gang-01-lan-mu',
                 source: ['/zone/:id', '/'],
                 target: '/hk01/zone/:id?',
             },
             {
                 title: '子栏目',
-                docs: 'https://docs.rsshub.app/traditional-media.html#xiang-gang-01-zi-lan-mu',
+                docs: 'https://docs.rsshub.app/new-media.html#xiang-gang-01-zi-lan-mu',
                 source: ['/channel/:id', '/'],
                 target: '/hk01/channel/:id?',
             },
             {
                 title: '专题',
-                docs: 'https://docs.rsshub.app/traditional-media.html#xiang-gang-01-zhuan-ti',
+                docs: 'https://docs.rsshub.app/new-media.html#xiang-gang-01-zhuan-ti',
                 source: ['/issue/:id', '/'],
                 target: '/hk01/issue/:id?',
             },
             {
                 title: '标签',
-                docs: 'https://docs.rsshub.app/traditional-media.html#xiang-gang-01-biao-qian',
+                docs: 'https://docs.rsshub.app/new-media.html#xiang-gang-01-biao-qian',
                 source: ['/tag/:id', '/'],
                 target: '/hk01/tag/:id?',
             },
+            {
+                title: '即時',
+                docs: 'https://docs.rsshub.app/new-media.html#xiang-gang-01-ji-shi',
+                source: ['/latest', '/'],
+                target: '/hk01/latest',
+            },
         ],
     },
 };

package/lib/v2/hk01/router.js

@@ -2,6 +2,7 @@ module.exports = function (router) {
     router.get('/channel/:id?', require('./channel'));
     router.get('/hot', require('./hot'));
     router.get('/issue/:id?', require('./issue'));
+    router.get('/latest', require('./latest'));
     router.get('/tag/:id?', require('./tag'));
     router.get('/zone/:id?', require('./zone'));
 };

package/lib/v2/hkepc/data.js

@@ -0,0 +1,62 @@
+const baseUrl = 'https://www.hkepc.com';
+
+module.exports = {
+    baseUrl,
+    categoryMap: {
+        price: {
+            url: `${baseUrl}/news`,
+            feedSuffix: ' - 腦場新聞',
+            selector: '#sidebar > div:nth-child(1) > div.content > ul > li',
+        },
+        review: {
+            url: `${baseUrl}/news`,
+            feedSuffix: ' - 新品快遞',
+            selector: '#sidebar > div:nth-child(2) > div.content > ul > li',
+        },
+        coverStory: {
+            url: `${baseUrl}/news`,
+            feedSuffix: ' - 專題報導',
+            selector: '#sidebar > div:nth-child(3) > div.content > ul > li',
+        },
+        news: {
+            url: `${baseUrl}/news`,
+            feedSuffix: ' - 新聞中心',
+            selector: '#sidebar > div:nth-child(4) > div.content > ul > li',
+        },
+        press: {
+            url: `${baseUrl}/news`,
+            feedSuffix: ' - 業界資訊',
+            selector: '#sidebar > div:nth-child(5) > div.content > ul > li',
+        },
+        member: {
+            url: `${baseUrl}/news`,
+            feedSuffix: ' - 會員消息',
+            selector: '#sidebar > div:nth-child(6) > div.content > ul > li',
+        },
+        digital: {
+            url: baseUrl,
+            feedSuffix: ' - 流動數碼',
+            selector: '#contentR5 > div.left > div.article > div.heading',
+        },
+        entertainment: {
+            url: baseUrl,
+            feedSuffix: ' - 生活娛樂',
+            selector: '#contentR5 > div.right > div.article > div.heading',
+        },
+        latest: {
+            url: baseUrl,
+            feedSuffix: ' - 最新消息',
+            selector: 'div .item',
+        },
+        '': {
+            url: baseUrl,
+            feedSuffix: ' - 最新消息',
+            selector: 'div .item',
+        },
+        ocLab: {
+            url: `${baseUrl}/ocLab`,
+            feedSuffix: ' - 超頻領域',
+            selector: '.heading',
+        },
+    },
+};

package/lib/v2/hkepc/index.js

@@ -2,185 +2,84 @@ const got = require('@/utils/got');
 const cheerio = require('cheerio');
 const { parseDate } = require('@/utils/parse-date');
 const timezone = require('@/utils/timezone');
-const path = require('path');
-const { art } = require('@/utils/render');
+const { baseUrl, categoryMap } = require('./data');
 
 module.exports = async (ctx) => {
     const category = ctx.params.category ?? '';
-    const baseUrl = `https://www.hkepc.com`;
 
-    let response;
-    if (category === 'price' || category === 'review' || category === 'coverStory' || category === 'news' || category === 'press' || category === 'member') {
-        response = await got({
-            url: `${baseUrl}/news`,
-            headers: {
-                Referer: baseUrl,
-            },
-        });
-    } else if (category === 'digital' || category === 'entertainment' || category === 'latest' || category === '') {
-        response = await got(baseUrl);
-    } else {
-        // category === ocLab
-        response = await got({
-            url: `${baseUrl}/${category}`,
-            headers: {
-                Referer: baseUrl,
-            },
-        });
-    }
+    const { data: response } = await got(categoryMap[category].url, {
+        headers: {
+            Referer: baseUrl,
+        },
+    });
 
-    const $ = cheerio.load(response.data);
+    const $ = cheerio.load(response);
 
-    let feedTitle = '電腦領域 HKEPC';
-    let items;
-    switch (category) {
-        case 'price':
-            feedTitle += ' - 腦場新聞';
-            items = $('#sidebar > div:nth-child(1) > div.content > ul > li')
-                .find('a')
-                .toArray()
-                .map((item) => ({
-                    title: $(item).text(),
-                    link: baseUrl + $(item).attr('href'),
-                }));
-            break;
-        case 'review':
-            feedTitle += ' - 新品快遞';
-            items = $('#sidebar > div:nth-child(2) > div.content > ul > li')
-                .find('a')
-                .toArray()
-                .map((item) => ({
-                    title: $(item).text(),
-                    link: baseUrl + $(item).attr('href'),
-                }));
-            break;
-        case 'coverStory':
-            feedTitle += ' - 專題報導';
-            items = $('#sidebar > div:nth-child(3) > div.content > ul > li')
-                .find('a')
-                .toArray()
-                .map((item) => ({
-                    title: $(item).text(),
-                    link: baseUrl + $(item).attr('href'),
-                }));
-            break;
-        case 'news':
-            feedTitle += ' - 新聞中心';
-            items = $('#sidebar > div:nth-child(4) > div.content > ul > li')
-                .find('a')
-                .toArray()
-                .map((item) => ({
-                    title: $(item).text(),
-                    link: baseUrl + $(item).attr('href'),
-                }));
-            break;
-        case 'press':
-            feedTitle += ' - 業界資訊';
-            items = $('#sidebar > div:nth-child(5) > div.content > ul > li')
-                .find('a')
-                .toArray()
-                .map((item) => ({
-                    title: $(item).text(),
-                    link: baseUrl + $(item).attr('href'),
-                }));
-            break;
-        case 'member':
-            feedTitle += ' - 會員消息';
-            items = $('#sidebar > div:nth-child(6) > div.content > ul > li')
-                .find('a')
-                .toArray()
-                .map((item) => ({
-                    title: $(item).text(),
-                    link: baseUrl + $(item).attr('href'),
-                }));
-            break;
-        case 'digital':
-            feedTitle += ' - 流動數碼';
-            items = $('#contentR5 > div.left > div.article > div.heading')
-                .find('a')
-                .toArray()
-                .map((item) => ({
-                    title: $(item).text(),
-                    link: baseUrl + $(item).attr('href'),
-                }));
-            break;
-        case 'entertainment':
-            feedTitle += ' - 生活娛樂';
-            items = $('#contentR5 > div.right > div.article > div.heading')
-                .find('a')
-                .toArray()
-                .map((item) => ({
-                    title: $(item).text(),
-                    link: baseUrl + $(item).attr('href'),
-                }));
-            break;
-        case 'latest':
-        case '':
-            feedTitle += ' - 最新消息';
-            items = $('div .item')
-                .find('a')
-                .toArray()
-                .map((item) => ({
-                    title: $(item).text(),
-                    link: baseUrl + $(item).attr('href'),
-                }));
-            break;
-        case 'ocLab':
-            feedTitle += ' - 超頻領域';
-            items = $('.heading')
-                .toArray()
-                .map((item) => ({
-                    title: $(item).text(),
-                    link: baseUrl + $(item).attr('href'),
-                }));
-            break;
-        default:
-            break;
-    }
+    const list = $(categoryMap[category].selector)
+        .find('a')
+        .toArray()
+        .map((item) => ({
+            title: $(item).text(),
+            link: baseUrl + $(item).attr('href'),
+        }));
 
-    items = await Promise.all(
-        items.map((item) =>
+    const items = await Promise.all(
+        list.map((item) =>
             ctx.cache.tryGet(item.link, async () => {
-                const detailResponse = await got({
-                    url: item.link,
+                const { data: response } = await got(item.link, {
                     headers: {
                         Referer: baseUrl,
                     },
                 });
 
-                const content = cheerio.load(detailResponse.data);
+                const $ = cheerio.load(response);
+                const content = $('#view');
+                const nextPages = $('#articleFooter .navigation a')
+                    .toArray()
+                    .map((a) => `${baseUrl}${a.attribs.href}`)
+                    .slice(1);
+
+                if (nextPages.length) {
+                    const pages = await Promise.all(
+                        nextPages.map(async (url) => {
+                            const { data: response } = await got(url, {
+                                headers: {
+                                    referer: item.link,
+                                },
+                            });
+                            const $ = cheerio.load(response);
+                            return $('#view').html();
+                        })
+                    );
+                    content.append(pages);
+                }
 
                 // remove unwanted elements
-                content('#view > div.advertisement').remove();
-                content('div#comments').remove();
-                content('div#share_btn').remove();
+                content.find('#view > div.advertisement').remove();
+                content.find('div#comments').remove();
+                content.find('div#share_btn').remove();
+                content.find('#articleFooter').remove();
 
                 // Non-breaking space U+00A0, ` ` in html
                 // Taken from /caixin/blog.js
-                content('#view > p')
-                    .filter((_, e) => e.children[0].data === String.fromCharCode(160))
+                content
+                    .find('#view > p')
+                    .filter((_, e) => e.children[0]?.data === String.fromCharCode(160))
                     .remove();
 
                 // fix lazyload image
-                content('#view > p > img').each((_, e) => {
-                    e = content(e);
-                    e.after(
-                        art(path.join(__dirname, 'templates/image.art'), {
-                            src: e.attr('rel') ?? e.attr('src'),
-                            alt: e.attr('alt'),
-                        })
-                    );
-                    e.remove();
+                content.find('#view > p > img').each((_, e) => {
+                    if (e.attribs.rel) {
+                        e.attribs.src = e.attribs.rel;
+                    }
                 });
 
-                item.author = content('.newsAuthor').text().trim() ?? content('#articleHead div.author').text().trim();
-                item.category = content('div#relatedArticles div.tags a')
+                item.author = $('.newsAuthor').text().trim() || $('#articleHead div.author').text().trim();
+                item.category = $('div#relatedArticles div.tags a')
                     .toArray()
-                    .map((e) => content(e).text().trim());
-                item.description = art(path.join(__dirname, 'templates/description.art'), {
-                    desc: content('#view').html(),
-                });
-                item.pubDate = timezone(parseDate(content('.publishDate').text()), +8);
+                    .map((e) => $(e).text().trim());
+                item.description = content.html();
+                item.pubDate = timezone(parseDate($('.publishDate').text()), +8);
                 item.guid = item.link.substring(0, item.link.lastIndexOf('/'));
 
                 return item;
@@ -189,15 +88,7 @@ module.exports = async (ctx) => {
     );
 
     ctx.state.data = {
-        title: feedTitle,
-        link: `https://www.hkepc.com/${category}`,
-        description: '電腦領域 HKEPC Hardware - 全港 No.1 PC網站',
-        language: 'zh-hk',
-        item: items,
-    };
-
-    ctx.state.json = {
-        title: feedTitle,
+        title: `電腦領域 HKEPC${categoryMap[category].feedSuffix}`,
         link: `https://www.hkepc.com/${category}`,
         description: '電腦領域 HKEPC Hardware - 全港 No.1 PC網站',
         language: 'zh-hk',

package/lib/v2/hnu/careers.js

@@ -0,0 +1,33 @@
+const got = require('@/utils/got');
+
+module.exports = async (ctx) => {
+    const rootUrl = 'http://scc.hnu.edu.cn/';
+    // 校园招聘的路由为/module/getcareers，但该页面默认仅加载整体框架，具体的招聘信息在页面加载完成后再交易获取。
+    // 具体的交易可以通过f12，网络栏进行查看，项目中仅保留了必须携带的参数。
+    // 其中有用的参数为count，为获取招聘信息的数量。
+    const currentRoute = `${rootUrl}module/getcareers`;
+    const currentUrl = `${currentRoute}?start_page=1&type=inner&day=&count=20&start=1`;
+    const response = await got({
+        method: 'get',
+        url: currentUrl,
+    });
+    const company_list = response.data.data;
+
+    ctx.state.data = {
+        title: '校园招聘',
+        link: currentRoute,
+        item: company_list.map((company_info) => {
+            const ret = {
+                // 标题：公司名称
+                title: company_info.company_name,
+                // 链接：公司招聘简章链接
+                link: `${rootUrl}detail/career?id=${company_info.career_talk_id}`,
+                // 描述：校招教室+时间+专业
+                description: company_info.address + ' - ' + company_info.meet_time + ' - ' + company_info.professionals,
+                // 分类：公司类型（私营、国企）+ 工作城市
+                category: company_info.company_property + ' - ' + company_info.city_name,
+            };
+            return ret;
+        }),
+    };
+};

package/lib/v2/hnu/maintainer.js

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

package/lib/v2/hnu/radar.js

@@ -0,0 +1,12 @@
+module.exports = {
+    'hnu.edu.cn': {
+        _name: '湖南大学',
+        scc: [
+            {
+                title: '校园招聘',
+                docs: 'https://docs.rsshub.app/university.html#hu-nan-da-xue',
+                target: '/hnu/careers',
+            },
+        ],
+    },
+};

package/lib/v2/hnu/router.js

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

package/lib/v2/hrbust/radar.js

@@ -6,7 +6,7 @@ module.exports = {
                 title: '教务处',
                 docs: 'https://docs.rsshub.app/university.html#ha-er-bin-li-gong-da-xue',
                 source: '/homepage/*',
-                target: (params, url) => `/hrbust/jwzx/${url.match(/(?<=columnId=)\d+/)}`,
+                target: (params, url) => `/hrbust/jwzx/${new URL(url).searchParams.get('columnId')}`,
             },
         ],
     },

package/lib/v2/huggingface/daily-papers.js

@@ -0,0 +1,23 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+
+module.exports = async (ctx) => {
+    const { body: response } = await got('https://huggingface.co/papers');
+    const $ = cheerio.load(response);
+    const papers = $('main > div[data-target="DailyPapers"]').data('props');
+
+    const items = papers.dailyPapers.map((item) => ({
+        title: item.title,
+        link: `https://arxiv.org/abs/${item.paper.id}`,
+        description: item.paper.summary.replace(/\n/g, ' '),
+        pubDate: parseDate(item.publishedAt),
+        author: item.paper.authors.map((author) => author.name).join(', '),
+    }));
+
+    ctx.state.data = {
+        title: 'Huggingface Daily Papers',
+        link: 'https://huggingface.co/papers',
+        item: items,
+    };
+};

package/lib/v2/huggingface/maintainer.js

@@ -0,0 +1,3 @@
+module.exports = {
+    '/daily-papers': ['zeyugao'],
+};

package/lib/v2/huggingface/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'huggingface.co': {
+        _name: 'Hugging Face',
+        '.': [
+            {
+                title: 'Daily Papers',
+                docs: 'https://docs.rsshub.app/programming.html#huggingface',
+                source: ['/papers', '/'],
+                target: '/huggingface/daily-papers',
+            },
+        ],
+    },
+};

package/lib/v2/huggingface/router.js

@@ -0,0 +1,3 @@
+module.exports = (router) => {
+    router.get('/daily-papers', require('./daily-papers'));
+};

package/lib/v2/hunanpea/maintainer.js

@@ -0,0 +1,3 @@
+module.exports = {
+    '/rsks/:guid': ['TonyRL'],
+};

package/lib/v2/hunanpea/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'hunanpea.com': {
+        _name: '湖南人事考试网',
+        rsks: [
+            {
+                title: '公告',
+                docs: 'https://docs.rsshub.app/study.html#hu-nan-ren-shi-kao-shi-wang',
+                source: ['/Category/:guid/ArticlesByCategory.do'],
+                target: '/hunanpea/rsks/:guid',
+            },
+        ],
+    },
+};

package/lib/v2/hunanpea/router.js

@@ -0,0 +1,3 @@
+module.exports = (router) => {
+    router.get('/rsks/:guid', require('./rsks'));
+};

package/lib/v2/hunanpea/rsks.js

@@ -0,0 +1,42 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+
+module.exports = async (ctx) => {
+    const baseUrl = 'http://rsks.hunanpea.com';
+    const { guid } = ctx.params;
+    const link = `${baseUrl}/Category/${guid}/ArticlesByCategory.do?PageIndex=1`;
+    const { data: response } = await got(link);
+
+    const $ = cheerio.load(response);
+    const list = $('#column_content > ul > li')
+        .toArray()
+        .map((item) => {
+            item = $(item);
+            return {
+                title: item.find('a').attr('title'),
+                link: `${baseUrl}${item.find('a').attr('href').replace('ArticleDetail.do', 'InternalArticleDetail.do?')}`,
+                pubDate: timezone(parseDate(item.find('em').text()), +8),
+            };
+        });
+
+    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 = $('.content_area').html();
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: `${$('.sitemap h2').text()} - ${$('head title').text()}`,
+        link,
+        item: items,
+    };
+};

package/lib/v2/huoxian/maintainer.js

@@ -0,0 +1,3 @@
+module.exports = {
+    '/zone': ['p7e4'],
+};

package/lib/v2/huoxian/radar.js

@@ -0,0 +1,11 @@
+module.exports = {
+    'huoxian.cn': {
+        _name: '火线',
+        zone: [
+            {
+                title: 'Zone',
+                docs: 'https://docs.rsshub.app/bbs.html#huo-xian',
+            },
+        ],
+    },
+};

package/lib/v2/huoxian/router.js

@@ -0,0 +1,3 @@
+module.exports = (router) => {
+    router.get('/zone', require('./zone'));
+};

package/lib/v2/huoxian/zone.js

@@ -0,0 +1,19 @@
+const got = require('@/utils/got');
+const { parseDate } = require('@/utils/parse-date');
+
+module.exports = async (ctx) => {
+    const { data } = await got('https://zone.huoxian.cn/api/discussions?sort=-createdAt');
+    const items = data.data.map((item) => ({
+        title: item.attributes.title,
+        link: `https://zone.huoxian.cn/d/${item.attributes.slug}`,
+        description: data.included.find((i) => i.id === item.relationships.firstPost.data.id).attributes.contentHtml,
+        pubDate: parseDate(item.attributes.createdAt),
+        author: data.included.find((i) => i.id === item.relationships.user.data.id).attributes.displayName,
+        category: data.included.filter((i) => item.relationships.tags.data.map((t) => t.id).includes(i.id) && i.type === 'tags').map((i) => i.attributes.name),
+    }));
+    ctx.state.data = {
+        title: '火线 Zone-安全攻防社区',
+        link: 'https://zone.huoxian.cn/',
+        item: items,
+    };
+};

package/lib/v2/hust/aia/news.js

@@ -0,0 +1,26 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+
+module.exports = async (ctx) => {
+    const link = 'https://aia.hust.edu.cn/xyxw.htm';
+    const response = await got(link);
+    const $ = cheerio.load(response.data);
+    const list = $('.list li');
+
+    ctx.state.data = {
+        title: '华科人工智能和自动化学院新闻',
+        link,
+        item:
+            list &&
+            list.toArray().map((item) => {
+                item = $(item);
+                return {
+                    title: item.find('a h2').text(),
+                    description: item.find('a div').text() || '华科人工智能和自动化学院新闻',
+                    pubDate: parseDate(item.find('.date3').text(), 'DDYYYY-MM'),
+                    link: new URL(item.find('a').attr('href'), link).href,
+                };
+            }),
+    };
+};