rsshub 1.0.0-master.fe99193 → 1.0.0-master.fe9c82a

package/lib/v2/hellogithub/volume.js

@@ -1,62 +1,46 @@
 const got = require('@/utils/got');
-const { parseDate } = require('@/utils/parse-date');
 const { art } = require('@/utils/render');
 const path = require('path');
 const md = require('markdown-it')({
     html: true,
 });
+const cheerio = require('cheerio');
 
 art.defaults.imports.render = function (string) {
     return md.render(string);
 };
 
 module.exports = async (ctx) => {
-    const limit = ctx.query.limit ? parseInt(ctx.query.limit) : 30;
-
     const rootUrl = 'https://hellogithub.com';
-    const apiRootUrl = 'https://api.hellogithub.com';
-    const currentUrl = `${rootUrl}/periodical/volume/`;
-    const apiUrl = `${apiRootUrl}/v1/volume/all/`;
+    const apiUrl = 'https://api.hellogithub.com/v1/periodical/';
 
-    const buildResponse = await got({
+    const periodicalResponse = await got({
         method: 'get',
-        url: rootUrl,
+        url: apiUrl,
     });
-
-    const buildId = buildResponse.data.match(/"buildId":"(.*?)",/)[1];
-
-    const response = await got({
+    const current = periodicalResponse.data.volumes[0].num;
+    const currentUrl = `${rootUrl}/periodical/volume/${current}`;
+    const buildResponse = await got({
         method: 'get',
-        url: apiUrl,
+        url: currentUrl,
     });
 
-    let items = response.data.data.slice(0, limit).map((item) => ({
-        guid: item.num,
-        title: `No.${item.num}`,
-        link: `${rootUrl}/periodical/volume/${item.guid}`,
-    }));
-
-    items = await Promise.all(
-        items.map((item) =>
-            ctx.cache.tryGet(item.link, async () => {
-                const detailUrl = `${rootUrl}/_next/data/${buildId}/periodical/volume/${item.guid}.json`;
-
-                const detailResponse = await got({
-                    method: 'get',
-                    url: detailUrl,
-                });
-
-                const data = detailResponse.data;
-
-                item.pubDate = parseDate(data.pageProps.volume.publish_at);
-                item.description = art(path.join(__dirname, 'templates/volume.art'), {
-                    data: data.pageProps.volume.data,
-                });
-
-                return item;
-            })
-        )
-    );
+    const $ = cheerio.load(buildResponse.data);
+
+    const text = $('#__NEXT_DATA__').text();
+    const response = JSON.parse(text);
+    const data = response.props;
+    const id = data.pageProps.volume.current_num;
+
+    const items = [
+        {
+            title: `No.${id}`,
+            link: `${rootUrl}/periodical/volume/${id}`,
+            description: art(path.join(__dirname, 'templates/volume.art'), {
+                data: data.pageProps.volume.data,
+            }),
+        },
+    ];
 
     ctx.state.data = {
         title: 'HelloGithub - 月刊',

package/lib/{routes/universities → v2}/hitsz/article.js

@@ -1,34 +1,32 @@
 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 host = 'http://www.hitsz.edu.cn';
-    const category = ctx.params.category || 'id-74';
+    const host = 'https://www.hitsz.edu.cn';
+    const category = ctx.params.category ?? 'id-77';
+    const link = `${host}/article/${category}.html`;
 
-    const response = await got.get(`${host}/article/${category}.html`);
+    const response = await got(link);
     const $ = cheerio.load(response.data);
     const category_name = $('div.title_page').text().trim();
 
     const lists = $('.mainside_news ul li')
-        .map((_, el) => ({
+        .toArray()
+        .map((el) => ({
             title: $('a', el).text().trim(),
             link: `${host}${$('a', el).attr('href')}`,
-            pubDate: new Date($('span[class=date]', el).text()).toUTCString(),
-        }))
-        .get();
+            pubDate: timezone(parseDate($('span[class=date]', el).text()), 8),
+        }));
 
     const items = await Promise.all(
         lists.map((item) =>
             ctx.cache.tryGet(item.link, async () => {
                 const response = await got.get(item.link);
                 const $ = cheerio.load(response.data);
-                item.description =
-                    $('div.detail').html() &&
-                    $('div.detail')
-                        .html()
-                        .replace(/src="\//g, `src="${new URL('.', host).toString()}"`)
-                        .replace(/href="\//g, `href="${new URL('.', host).toString()}"`)
-                        .trim();
+                item.description = $('div.edittext').html().trim();
+                item.pubDate = timezone(parseDate($('.item').first().text().replace('发布时间：', '')), 8);
                 return item;
             })
         )
@@ -36,7 +34,7 @@ module.exports = async (ctx) => {
 
     ctx.state.data = {
         title: '哈尔滨工业大学（深圳）-' + category_name,
-        link: `${host}/article/${category}.html`,
+        link,
         description: '哈尔滨工业大学（深圳）-' + category_name,
         item: items,
     };

package/lib/v2/hitsz/maintainer.js

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

package/lib/v2/hitsz/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'hitsz.edu.cn': {
+        _name: '哈尔滨工业大学（深圳）',
+        '.': [
+            {
+                title: '新闻中心',
+                docs: 'https://docs.rsshub.app/university.html#ha-er-bin-gong-ye-da-xue-shen-zhen',
+                source: ['/article/:category?', '/subject/:category?'],
+                target: (params) => `/hitsz/article/${params.category.replace('.html', '')}`,
+            },
+        ],
+    },
+};

package/lib/v2/hitsz/router.js

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

package/lib/v2/hk01/utils.js

@@ -9,8 +9,8 @@ const apiRootUrl = 'https://web-data.api.hk01.com';
 module.exports = {
     rootUrl,
     apiRootUrl,
-    ProcessItems: async (items, limit, tryGet) =>
-        await Promise.all(
+    ProcessItems: (items, limit, tryGet) =>
+        Promise.all(
             items
                 .filter((item) => item.type !== 2)
                 .slice(0, limit ? parseInt(limit) : 50)

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/hljucm/maintainer.js

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

package/lib/v2/hljucm/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'hljucm.net': {
+        _name: '黑龙江中医药大学',
+        yjsy: [
+            {
+                title: '研究生院',
+                docs: 'https://docs.rsshub.app/university.html#hei-long-jiang-zhong-yi-yao-da-xue',
+                source: ['/index/:category', '/index'],
+                target: (params) => `/hljucm/yjsy/${params.category.replace('.htm', '')}`,
+            },
+        ],
+    },
+};

package/lib/v2/hljucm/router.js

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

package/lib/v2/hljucm/yjsy.js

@@ -0,0 +1,56 @@
+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 category = ctx.params.category ?? 'xwdt';
+
+    const rootUrl = 'https://yjsy.hljucm.net';
+    const currentUrl = `${rootUrl}/index/${category}.htm`;
+
+    const response = await got(currentUrl);
+
+    const $ = cheerio.load(response.data);
+
+    const list = $('.postlist a')
+        .toArray()
+        .map((item) => {
+            item = $(item);
+
+            return {
+                title: item.attr('title'),
+                link: new URL(item.attr('href'), currentUrl).href,
+            };
+        });
+
+    const items = await Promise.all(
+        list.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const detailResponse = await got(item.link);
+
+                const content = cheerio.load(detailResponse.data);
+
+                item.description = content('#vsb_newscontent').html();
+                item.pubDate = timezone(parseDate(content('.timestyle56043').text()), +8);
+
+                const files = detailResponse.data.match(/<span>附件【<a href="(.*)"><span>(.*)<\/span><\/a>】<\/span>/g);
+
+                if (files) {
+                    for (const file of files) {
+                        const { link, name } = file.match(/【<a href="(?<link>.*)"><span>(?<name>.*)<\/span><\/a>】/).groups;
+                        item.description += `<a href="${link}">${name}</a>`;
+                    }
+                }
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: $('title').text(),
+        link: currentUrl,
+        item: items,
+    };
+};

package/lib/v2/hotukdeals/hottest.js

@@ -0,0 +1,25 @@
+const got = require('@/utils/got');
+const { JSDOM } = require('jsdom');
+
+module.exports = async (ctx) => {
+    const data = await got.get(`https://www.hotukdeals.com/`, {
+        headers: {
+            Referer: `https://www.hotukdeals.com/`,
+        },
+    });
+
+    const dom = new JSDOM(data.data, {
+        runScripts: 'dangerously',
+    });
+    const threads = dom.window.__INITIAL_STATE__.widgets.hottestWidget.threads;
+
+    ctx.state.data = {
+        title: `hotukdeals hottest`,
+        link: `https://www.hotukdeals.com/`,
+        item: threads.map((item) => ({
+            title: item.title,
+            description: `<img src="https://images.hotukdeals.com/${item.mainImage.path}/${item.mainImage.name}/re/768x768/qt/60/${item.mainImage.name}.jpg"><br>${item.temperature}° ${item.title}<br>${item.displayPrice}`,
+            link: item.url,
+        })),
+    };
+};

package/lib/v2/hotukdeals/maintainer.js

@@ -1,3 +1,4 @@
 module.exports = {
     '/:type': ['DIYgod'],
+    '/hottest': ['DIYgod'],
 };

package/lib/v2/hotukdeals/radar.js

@@ -26,6 +26,12 @@ module.exports = {
                 source: ['/'],
                 target: '/hotukdeals/discussed',
             },
+            {
+                title: 'hottest',
+                docs: 'https://docs.rsshub.app/shopping.html#hotukdeals',
+                source: ['/'],
+                target: '/hotukdeals/hottest',
+            },
         ],
     },
 };

package/lib/v2/hotukdeals/router.js

@@ -1,3 +1,4 @@
 module.exports = function (router) {
+    router.get('/hottest', require('./hottest'));
     router.get('/:type', require('./index'));
 };