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

package/lib/v2/douyin/hashtag.js

@@ -1,4 +1,3 @@
-const cheerio = require('cheerio');
 const { parseDate } = require('@/utils/parse-date');
 const { art } = require('@/utils/render');
 const config = require('@/config').value;
@@ -23,19 +22,26 @@ module.exports = async (ctx) => {
             const browser = await require('@/utils/puppeteer')();
             const page = await browser.newPage();
             await page.setRequestInterception(true);
+            let awemeList = '';
             page.on('request', (request) => {
-                request.resourceType() === 'document' ? request.continue() : request.abort();
+                request.resourceType() === 'document' || request.resourceType() === 'script' || request.resourceType() === 'xhr' ? request.continue() : request.abort();
+            });
+            page.on('response', async (response) => {
+                const request = response.request();
+                if (request.url().includes('/v1/web/challenge/aweme')) {
+                    awemeList = await response.json();
+                }
             });
             await page.goto(tagUrl, {
-                waitUntil: 'domcontentloaded',
+                waitUntil: 'networkidle2',
             });
             await page.waitForSelector('#RENDER_DATA');
-            const html = await page.evaluate(() => document.documentElement.innerHTML);
+            const html = await page.evaluate(() => document.getElementById('RENDER_DATA').innerText);
             await browser.close();
 
-            const $ = cheerio.load(html);
-            const renderData = JSON.parse(decodeURIComponent($('script#RENDER_DATA').text()));
-            const dataKey = Object.keys(renderData).find((key) => renderData[key].topicDetail && renderData[key].defaultData);
+            const renderData = JSON.parse(decodeURIComponent(html));
+            const dataKey = Object.keys(renderData).find((key) => renderData[key].topicDetail);
+            renderData[dataKey].defaultData = awemeList.aweme_list;
             return renderData[dataKey];
         },
         config.cache.routeExpire,
@@ -49,7 +55,7 @@ module.exports = async (ctx) => {
     const posts = tagData.defaultData;
     const items = posts.map((post) => {
         // parse video
-        let videoList = post.video && post.video.bitRateList && post.video.bitRateList.map((item) => resolveUrl(item.playApi));
+        let videoList = post.video && post.video.bit_rate && post.video.bit_rate.map((bit_rate) => resolveUrl(bit_rate.play_addr.url_list[0]));
         if (relay) {
             videoList = videoList.map((item) => proxyVideo(item, relay));
         }
@@ -57,14 +63,9 @@ module.exports = async (ctx) => {
         duration = duration && duration / 1000;
         let img;
         // if (!embed) {
-        //     img = post.video && post.video.dynamicCover; // dynamic cover (webp)
+        //     img = post.video && post.video.dynamic_cover && post.video.dynamic_cover.url_list[post.video.dynamic_cover.url_list.length - 1]; // dynamic cover (webp)
         // }
-        img =
-            img ||
-            (post.video &&
-                ((post.video.coverUrlList && post.video.coverUrlList[post.video.coverUrlList.length - 1]) || // HD
-                    post.video.originCover || // LD
-                    post.video.cover)); // even more LD
+        img = img || (post.video && post.video.origin_cover && post.video.origin_cover.url_list[post.video.origin_cover.url_list.length - 1]);
         img = img && resolveUrl(img);
 
         // render description
@@ -76,10 +77,10 @@ module.exports = async (ctx) => {
         return {
             title: post.desc,
             description,
-            link: `https://www.douyin.com/video/${post.awemeId}`,
-            pubDate: parseDate(post.createTime * 1000),
-            category: post.textExtra.map((extra) => extra.hashtagName),
-            author: post.authorInfo.nickname,
+            link: `https://www.douyin.com/video/${post.aweme_id}`,
+            pubDate: parseDate(post.create_time, 'X'),
+            category: post.text_extra.map((extra) => extra.hashtag_name),
+            author: post.author.nickname,
         };
     });
 

package/lib/v2/douyin/live.js

@@ -1,5 +1,5 @@
 const config = require('@/config').value;
-const { getOriginAvatar } = require('./utils');
+const { getOriginAvatar, universalGet } = require('./utils');
 
 module.exports = async (ctx) => {
     const { rid } = ctx.params;
@@ -9,26 +9,7 @@ module.exports = async (ctx) => {
 
     const pageUrl = `https://live.douyin.com/${rid}`;
 
-    const renderData = await ctx.cache.tryGet(
-        `douyin:live:${rid}`,
-        async () => {
-            const browser = await require('@/utils/puppeteer')();
-            const page = await browser.newPage();
-            await page.setRequestInterception(true);
-            page.on('request', (request) => {
-                request.resourceType() === 'document' ? request.continue() : request.abort();
-            });
-            await page.goto(pageUrl, {
-                waitUntil: 'domcontentloaded',
-            });
-            const data = JSON.parse(decodeURIComponent(await page.evaluate(() => document.querySelector('#RENDER_DATA').innerText)));
-            browser.close();
-
-            return data;
-        },
-        config.cache.routeExpire,
-        false
-    );
+    const renderData = await ctx.cache.tryGet(`douyin:live:${rid}`, () => universalGet(pageUrl, 'live', ctx), config.cache.routeExpire, false);
 
     if (renderData[Object.keys(renderData).find((k) => renderData[k].status_code)].status_code !== '0') {
         throw `Status code ${renderData[Object.keys(renderData).find((k) => renderData[k].status_code)].status_code}`;

package/lib/v2/douyin/user.js

@@ -1,9 +1,8 @@
-const cheerio = require('cheerio');
 const { parseDate } = require('@/utils/parse-date');
 const { art } = require('@/utils/render');
 const config = require('@/config').value;
 const { fallback, queryToBoolean } = require('@/utils/readable-social');
-const { templates, resolveUrl, proxyVideo, getOriginAvatar } = require('./utils');
+const { templates, resolveUrl, proxyVideo, getOriginAvatar, universalGet } = require('./utils');
 
 module.exports = async (ctx) => {
     const uid = ctx.params.uid;
@@ -20,27 +19,13 @@ module.exports = async (ctx) => {
     const pageData = await ctx.cache.tryGet(
         `douyin:user:${uid}`,
         async () => {
-            const browser = await require('@/utils/puppeteer')();
-            const page = await browser.newPage();
-            await page.setRequestInterception(true);
-            page.on('request', (request) => {
-                request.resourceType() === 'document' ? request.continue() : request.abort();
-            });
-            await page.goto(pageUrl, {
-                waitUntil: 'domcontentloaded',
-            });
-            const html = await page.evaluate(() => document.documentElement.innerHTML);
-            await browser.close();
-
-            const $ = cheerio.load(html);
-            const renderData = JSON.parse(decodeURIComponent($('script#RENDER_DATA').html()));
+            const renderData = await universalGet(pageUrl, 'user', ctx);
             const dataKey = Object.keys(renderData).filter((key) => renderData[key].user && renderData[key].post)[0];
             return renderData[dataKey];
         },
         config.cache.routeExpire,
         false
     );
-
     const userInfo = pageData.user.user;
     const userNickName = userInfo.nickname;
     const userDescription = userInfo.desc;

package/lib/v2/douyin/utils.js

@@ -1,4 +1,8 @@
 const path = require('path');
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const logger = require('@/utils/logger');
+const config = require('@/config').value;
 
 const templates = {
     desc: path.join(__dirname, 'templates/desc.art'),
@@ -42,9 +46,90 @@ const getOriginAvatar = (url) =>
         .replace(/^(.*\.douyinpic\.com\/).*(\/aweme-avatar\/)([^?]*)(\?.*)?$/, '$1origin$2$3')
         .replace(/~\w+_\d+x\d+/g, '');
 
+const extractRenderData = (html) => {
+    if (!html) {
+        throw Error('Empty html. The request may be filtered by WAF.');
+    }
+
+    const $ = cheerio.load(html);
+    const renderData = JSON.parse(decodeURIComponent($('script#RENDER_DATA').html()));
+    if (!renderData) {
+        const title = $('title').text();
+        if (title.includes('验证')) {
+            throw Error(title);
+        } else if ($('#captcha_container').length > 0) {
+            throw Error('Captcha required.');
+        }
+        throw Error('Failed to get render data.');
+    } else if (renderData.isSpider) {
+        throw Error('The request was considered to be from a spider.');
+    } else {
+        return renderData;
+    }
+};
+
+const gotGet = async (url, ua) => {
+    const response = await got(url, {
+        headers: {
+            'User-Agent': ua, // magic!
+        },
+    });
+    return extractRenderData(response.data);
+};
+
+const puppeteerGet = async (pageUrl) => {
+    const browser = await require('@/utils/puppeteer')();
+    const page = await browser.newPage();
+    await page.setRequestInterception(true);
+    page.on('request', (request) => {
+        request.resourceType() === 'document' ? request.continue() : request.abort();
+    });
+    await page.goto(pageUrl, {
+        waitUntil: 'domcontentloaded',
+    });
+    const html = await page.evaluate(() => document.documentElement.innerHTML);
+    await browser.close();
+    return extractRenderData(html);
+};
+
+const universalGet = async (url, route, ctx) => {
+    const cacheKey = `douyin:utils:universalGet:gotFirst:${route}`;
+    const gotFirst = await ctx.cache.tryGet(
+        cacheKey,
+        () => '1',
+        config.cache.contentExpire,
+        false // no refresh now
+    );
+
+    const ua = route === 'live' ? 'Mozilla/5.0 (compatible; Googlebot/2.1; +http://www.google.com/bot.html)' : 'curl/7.86.0';
+
+    let data;
+    if (gotFirst === '1') {
+        try {
+            data = await gotGet(url, ua);
+            ctx.cache.set(cacheKey, '1', config.cache.contentExpire); // refresh if success
+        } catch (e) {
+            logger.warn('Failed to get data with got, trying puppeteer: ' + e.message);
+            data = await puppeteerGet(url);
+            ctx.cache.set(cacheKey, '0', config.cache.contentExpire); // only set if success
+        }
+    } else {
+        try {
+            data = await puppeteerGet(url);
+            ctx.cache.set(cacheKey, '0', config.cache.contentExpire); // refresh if success
+        } catch (e) {
+            logger.warn('Failed to get data with puppeteer, trying got: ' + e.message);
+            data = await gotGet(url, ua);
+            ctx.cache.set(cacheKey, '1', config.cache.contentExpire); // only set if success
+        }
+    }
+    return data;
+};
+
 module.exports = {
     templates,
     resolveUrl,
     proxyVideo,
     getOriginAvatar,
+    universalGet,
 };

package/lib/v2/dtcj/datainsight.js

@@ -28,20 +28,19 @@ module.exports = async (ctx) => {
         .get();
 
     const items = await Promise.all(
-        list.map(
-            async (item) =>
-                await ctx.cache.tryGet(item.link, async () => {
-                    const detailResponse = await got({
-                        method: 'get',
-                        url: item.link,
-                    });
-                    const content = cheerio.load(detailResponse.data);
-
-                    item.description = content('.content-3mNFyi').html();
-                    item.pubDate = parseDate(detailResponse.data.match(/"date":"(\d{4}-\d{2}-\d{2}T\d{2}:\d{2}:\d{2}.\d{3}\+\d{2}:\d{2})","thumbnail_url":/)[1]);
-
-                    return item;
-                })
+        list.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const detailResponse = await got({
+                    method: 'get',
+                    url: item.link,
+                });
+                const content = cheerio.load(detailResponse.data);
+
+                item.description = content('.content-3mNFyi').html();
+                item.pubDate = parseDate(detailResponse.data.match(/"date":"(\d{4}-\d{2}-\d{2}T\d{2}:\d{2}:\d{2}.\d{3}\+\d{2}:\d{2})","thumbnail_url":/)[1]);
+
+                return item;
+            })
         )
     );
 

package/lib/v2/dut/index.js

@@ -3,9 +3,13 @@ const cheerio = require('cheerio');
 const { parseDate } = require('@/utils/parse-date');
 const defaults = require('./defaults');
 const shortcuts = require('./shortcuts');
+const { isValidHost } = require('@/utils/valid-host');
 
 module.exports = async (ctx) => {
     const site = ctx.params[0] ?? 'news';
+    if (!isValidHost(site)) {
+        throw Error('Invalid site');
+    }
 
     let items;
     let category = ctx.params[1] ?? (defaults.hasOwnProperty(site) ? defaults[site] : '');

package/lib/v2/eagle/blog.js

@@ -1,16 +1,20 @@
 const got = require('@/utils/got');
 const cheerio = require('cheerio');
 const { parseDate } = require('@/utils/parse-date');
-
+const { isValidHost } = require('@/utils/valid-host');
 const cateList = ['all', 'design-resources', 'learn-design', 'inside-eagle'];
 
 module.exports = async (ctx) => {
     let cate = ctx.params.cate ?? 'all';
     let language = ctx.params.language ?? 'cn';
-    if (cateList.indexOf(cate) === -1) {
+    if (!isValidHost(cate) || !isValidHost(language)) {
+        throw Error('Invalid host');
+    }
+    if (!cateList.includes(cate)) {
         language = cate;
         cate = 'all';
     }
+
     const host = `https://${language}.eagle.cool`;
     const url = `${host}/blog/${cate === 'all' ? '' : cate}`;
 

package/lib/v2/eastday/24.js

@@ -59,21 +59,20 @@ module.exports = async (ctx) => {
                         links.push(item.link.replace(/\.html/, `-${i}.html`));
                     }
 
-                    links.forEach(
-                        async (link) =>
-                            await ctx.cache.tryGet(link, async () => {
-                                const pageResponse = await got({
-                                    method: 'get',
-                                    url: link,
-                                });
-                                const subContent = cheerio.load(pageResponse.data);
-
-                                subContent('img').each(function () {
-                                    subContent(this).attr('src', subContent(this).attr('data-url'));
-                                });
-
-                                item.description += subContent('#J-contain_detail_cnt').html();
-                            })
+                    links.forEach((link) =>
+                        ctx.cache.tryGet(link, async () => {
+                            const pageResponse = await got({
+                                method: 'get',
+                                url: link,
+                            });
+                            const subContent = cheerio.load(pageResponse.data);
+
+                            subContent('img').each(function () {
+                                subContent(this).attr('src', subContent(this).attr('data-url'));
+                            });
+
+                            item.description += subContent('#J-contain_detail_cnt').html();
+                        })
                     );
                 }
 

package/lib/v2/{acm-ecnu → ecnu}/contest.js

@@ -9,10 +9,7 @@ module.exports = async (ctx) => {
     const rootUrl = 'https://acm.ecnu.edu.cn';
     const currentUrl = `${rootUrl}/contest/`;
 
-    const response = await got({
-        method: 'get',
-        url: currentUrl,
-    });
+    const response = await got(currentUrl);
 
     const $ = cheerio.load(response.data);
     const $trList = $('div > div > table > tbody > tr');

package/lib/v2/ecnu/maintainer.js

@@ -0,0 +1,4 @@
+module.exports = {
+    '/acm/contest/:category?': ['a180285'],
+    '/yjs': ['shengmaosu'],
+};

package/lib/v2/ecnu/radar.js

@@ -0,0 +1,30 @@
+module.exports = {
+    'ecnu.edu.cn': {
+        _name: '华东师范大学',
+        acm: [
+            {
+                title: '所有比赛列表',
+                docs: 'https://docs.rsshub.app/university.html#hua-dong-shi-fan-da-xue',
+                source: ['/contest/', '/'],
+                target: '/ecnu/acm/contest/',
+            },
+            {
+                title: '仅公开比赛列表',
+                docs: 'https://docs.rsshub.app/university.html#hua-dong-shi-fan-da-xue',
+                source: ['/contest/', '/'],
+                target: '/ecnu/acm/contest/public',
+            },
+        ],
+    },
+    'kaoyan.com': {
+        _name: '华东师范大学',
+        yz: [
+            {
+                title: '研究生院',
+                docs: 'https://docs.rsshub.app/university.html#hua-dong-shi-fan-da-xue',
+                source: ['/ecnu/tiaoji', '/'],
+                target: '/ecnu/yjs',
+            },
+        ],
+    },
+};

package/lib/v2/ecnu/router.js

@@ -0,0 +1,4 @@
+module.exports = function (router) {
+    router.get('/acm/contest/:category?', require('./contest'));
+    router.get('/yjs', require('./yjs'));
+};

package/lib/v2/ecnu/yjs.js

@@ -0,0 +1,37 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+
+module.exports = async (ctx) => {
+    const link = 'https://yz.kaoyan.com/ecnu/tiaoji/';
+    const response = await got(link);
+    const $ = cheerio.load(response.data);
+    const list = $('.subList li')
+        .toArray()
+        .map((item) => {
+            item = $(item);
+            return {
+                title: item.find('li a').text(),
+                link: item.find('li a').attr('href').replace('http:', 'https:'),
+            };
+        });
+
+    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 = $('.articleCon').html();
+                item.pubDate = parseDate($('.outer_utime').text());
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: '华东师范大学研究生院',
+        link,
+        description: '华东师范大学研究生调剂信息',
+        item: items,
+    };
+};

package/lib/v2/economist/espresso.js

@@ -15,36 +15,66 @@ module.exports = async (ctx) => {
         false
     );
 
-    const gobbets = $('._gobbets ._gobbet p')
-        .toArray()
-        .map((item) => {
-            item = $(item);
-            return {
-                title: item.text(),
-                description: item.html(),
-                link,
-                guid: item.text(),
-            };
-        });
-
-    const articles = $('._articles ._article')
-        .toArray()
-        .map((item) => {
-            item = $(item);
-            const title = item.find('h3').text();
-            item.find('h3').remove();
-            return {
-                title,
-                description: item.html(),
-                link,
-                guid: item.text(),
-            };
-        });
+    const nextData = JSON.parse($('script#__NEXT_DATA__').text());
+    const parts = nextData.props.pageProps.content.hasPart.parts[0].hasPart.parts.filter((part) => part.type.includes('Article'));
+
+    const renderHTML = (node) => {
+        let el;
+        switch (node.type) {
+            case 'tag':
+                el = $(`<${node.name}>`);
+                for (const name in node.attribs) {
+                    el.attr(name, node.attribs[name]);
+                }
+                if (node.children) {
+                    el.append(node.children.map(renderHTML));
+                }
+                return el;
+            case 'text':
+                return node.data;
+            default:
+                el = $('<div>');
+                el.append(node.map(renderHTML));
+                return el;
+        }
+    };
+
+    const gobbetPart = parts.shift();
+    const gobbets = gobbetPart.text.map((node, index) => {
+        const html = renderHTML(node).html();
+        return {
+            guid: `${gobbetPart.id}/${index}`,
+            link,
+            title: gobbetPart.headline,
+            pubDate: gobbetPart.datePublished,
+            description: html,
+        };
+    });
+
+    const articles = parts.map((part) => {
+        const image = part.image.main;
+        const imgNode = {
+            type: 'tag',
+            name: 'img',
+            attribs: { width: image.width, height: image.height, src: image.url.canonical },
+        };
+
+        const html = renderHTML([imgNode, ...part.text]).html();
+
+        return {
+            link,
+            guid: part.id,
+            title: part.headline,
+            pubDate: part.datePublished,
+            description: html,
+        };
+    });
 
     ctx.state.data = {
         title: $('head title').text(),
         link,
         description: $('meta[property="og:description"]').attr('content'),
+        language: 'en-gb',
         item: [...gobbets, ...articles],
     };
 };

package/lib/v2/economist/full.js

@@ -2,8 +2,8 @@ const parser = require('@/utils/rss-parser');
 const got = require('@/utils/got');
 const cheerio = require('cheerio');
 
-const getArticleDetail = async (link, ctx) =>
-    await ctx.cache.tryGet(link, async () => {
+const getArticleDetail = (link, ctx) =>
+    ctx.cache.tryGet(link, async () => {
         const response = await got(link);
         const $ = cheerio.load(response.data);
         $('div.article-audio-player__center-tooltip').remove();

package/lib/v2/economist/global-business-review.js

@@ -0,0 +1,86 @@
+const got = require('@/utils/got');
+
+const ALLOW_LANGUAGE = {
+    en: 'en_GB',
+    cn: 'zh_CN',
+    tw: 'zh_TW',
+};
+
+const TITLE = {
+    en: 'The Economist Global Business Review',
+    cn: '经济学人 · 商论',
+    tw: '經濟學人 · 商論',
+};
+
+const DESCRIPTION = {
+    en: 'The Economist Global Business Review is a new bilingual digital app from the editors of The Economist Group.',
+    cn: '《经济学人·商论》是《经济学人》2015年5月推出的旗下中英双语APP，萃取《经济学人》在商业、金融、科技等领域的精华文章，为中国读者呈现全球视角的深度分析，并鼓励中国的读者批判性地思考中国和全球重大议题。',
+    tw: '《經濟學人·商論》是經濟學人集團官方中英雙語電子APP，萃取《經濟學人》在商業、金融、科技等領域的精華文章，為中國讀者呈現全球視角的深度分析，並鼓勵中國的讀者批判性地思考中國和全球的重大議題。',
+};
+
+const parseContent = function (item, article_id, language) {
+    if (item.type === 'paragraph') {
+        return parseParagraph(item.data, language);
+    } else if (item.type === 'image') {
+        return parseImage(item.data, article_id, language);
+    } else if (item.type === 'subtitle') {
+        return parseParagraph(item.data, language, 'h3');
+    }
+};
+
+const parseTitle = function (data, language) {
+    const content = Object.assign({}, ...data.map((x) => ({ [x.lang]: x.text })));
+    return language.map((item) => content[ALLOW_LANGUAGE[item]]).join('');
+};
+
+const parseParagraph = function (data, language, type = 'p') {
+    const content = Object.assign({}, ...data.map((x) => ({ [x.lang]: x.text })));
+    return `<div>${language.map((item) => `<${type}>${content[ALLOW_LANGUAGE[item]]}</${type}>`).join('')}</div>`;
+};
+
+const parseImage = function (data, article_id, language) {
+    const content = Object.assign({}, ...data.map((x) => ({ [x.lang]: x.image_path })));
+    return `<div><img src="https://businessreviewglobal-cdn.com/article_images/${article_id}/${encodeURIComponent(content[ALLOW_LANGUAGE[language[0]]])}"/></div>`;
+};
+
+const getArticleDetail = (article_id, language, ctx) => {
+    const link = `https://api.hummingbird.businessreview.global/api/article/index?id=${article_id}`;
+    return ctx.cache.tryGet(`${link}:${language.join('-')}`, async () => {
+        const response = await got({
+            method: 'get',
+            url: link,
+        });
+        const data = response.data.body;
+        let content = '';
+        content += parseParagraph(data.rubric, language);
+        content += data.content.map((item) => parseContent(item, article_id, language)).join('');
+        return content;
+    });
+};
+
+module.exports = async (ctx) => {
+    const language = (ctx.params.language && ctx.params.language.split('-').filter((item, index, arr) => ALLOW_LANGUAGE[item] && arr.indexOf(item, 0) === index)) || ['cn', 'en'];
+    const main_language = language[0];
+
+    const response = await got({
+        method: 'get',
+        url: 'https://api.hummingbird.businessreview.global/api/toc/get_articles',
+    });
+
+    const items = await Promise.all(
+        response.data.articles.new.slice(0, ctx.query.limit ? parseInt(ctx.query.limit) : 10).map(async (item) => ({
+            title: parseTitle(item.body.title, [main_language]),
+            description: await getArticleDetail(item.article_id, language, ctx),
+            category: parseTitle(item.body.fly_title, [main_language]),
+            link: `https://www.businessreview.global/latest/${item.article_id}`,
+            pubDate: item.publication_date,
+        }))
+    );
+
+    ctx.state.data = {
+        title: TITLE[main_language],
+        link: 'https://www.businessreview.global',
+        description: DESCRIPTION[main_language],
+        item: items,
+    };
+};

package/lib/v2/economist/maintainer.js

@@ -2,5 +2,6 @@ module.exports = {
     '/download': ['nczitzk'],
     '/espresso': ['TonyRL'],
     '/gre-vocabulary': ['xyqfer'],
+    '/global-business-review/:language?': ['prnake'],
     '/:endpoint': ['ImSingee'],
 };