rsshub 1.0.0-master.fe50f9c → 1.0.0-master.fe560d2

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/{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/full.js

@@ -7,7 +7,7 @@ const getArticleDetail = async (link, ctx) =>
         const response = await got(link);
         const $ = cheerio.load(response.data);
         $('div.article-audio-player__center-tooltip').remove();
-        const nextData = JSON.parse($('head script[type="application/ld+json"]').text());
+        const nextData = JSON.parse($('head script[type="application/ld+json"]').first().text());
 
         const article = ($('figure[class^=css-]').first().parent().parent().html() || '') + $('p.article__body-text').parent().html();
         const categories = nextData.keywords?.map((k) => k);

package/lib/v2/ehentai/ehapi.js

@@ -50,15 +50,27 @@ async function parsePage(cache, data, get_bittorrent = false, embed_thumb = fals
     // "l" for Compact
     // "e" for Extended
     // "t" for Thumbnail
-    const layout = $('div#dms option[selected]').attr('value');
-
-    let galleries;
-    if (layout !== 't') {
-        galleries = $('table[class^="itg glt"] tbody');
-    } else {
-        // Thumbnail layout has different page structure
-        // resolve it here
-        galleries = $('div[class^="itg gld"]');
+    let layout = 't';
+
+    // "itg gld" for Thumbnail
+    let galleries = $('div[class^="itg gld"]');
+    // "itg gltm" for Minimal or Minimal+
+    if (galleries.length <= 0) {
+        galleries = $('table[class^="itg gltm"] tbody');
+        layout = 'm';
+    }
+    // "itg gltc" for Compact
+    if (galleries.length <= 0) {
+        galleries = $('table[class^="itg gltc"] tbody');
+        layout = 'l';
+    }
+    // "itg glte" for Extended
+    if (galleries.length <= 0) {
+        galleries = $('table[class^="itg glte"] tbody');
+        layout = 'e';
+    }
+    if (galleries.length <= 0) {
+        return [];
     }
 
     async function parseElement(cache, element) {
@@ -108,12 +120,11 @@ async function parsePage(cache, data, get_bittorrent = false, embed_thumb = fals
             if ('le'.includes(layout)) {
                 // artist tags will only show in Compact or Extended layout
                 // get artist names as author
-                const artist = $(el)
+                item.author = $(el)
                     .find('div.gt[title^="artist:"]')
                     .toArray()
-                    .map((i, tag) => $(tag).text())
+                    .map((tag) => $(tag).text())
                     .join(' / ');
-                item.author = artist;
             }
             return item;
         }

package/lib/v2/fantia/maintainer.js

@@ -0,0 +1,4 @@
+module.exports = {
+    '/search/:type?/:caty?/:peroid?/:order?/:rating?/:keyword?': ['nczitzk'],
+    '/user/:id': ['nczitzk'],
+};

package/lib/v2/fantia/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'fantia.jp': {
+        _name: 'Fantia',
+        '.': [
+            {
+                title: '用户投稿',
+                docs: 'https://docs.rsshub.app/picture.html#fantia',
+                source: ['/fanclubs/:id'],
+                target: '/fantia/user/:id',
+            },
+        ],
+    },
+};

package/lib/v2/fantia/router.js

@@ -0,0 +1,4 @@
+module.exports = (router) => {
+    router.get('/search/:type?/:caty?/:period?/:order?/:rating?/:keyword?', require('./search'));
+    router.get('/user/:id', require('./user'));
+};

package/lib/{routes → v2}/fantia/search.js

@@ -1,4 +1,6 @@
 const got = require('@/utils/got');
+const { parseDate } = require('@/utils/parse-date');
+const Config = require('@/config').value.fantia;
 
 module.exports = async (ctx) => {
     const type = ctx.params.type || 'posts';
@@ -6,15 +8,18 @@ module.exports = async (ctx) => {
     const order = ctx.params.order || 'updater';
     const rating = ctx.params.rating || 'all';
     const keyword = ctx.params.keyword || '';
-    const peroid = ctx.params.peroid || '';
+    const period = ctx.params.period || '';
 
     const rootUrl = 'https://fantia.jp';
-    const apiUrl = `${rootUrl}/api/v1/search/${type}?keyword=${keyword}&peroid=${peroid}&brand_type=0&category=${caty === 'all' ? '' : caty}&order=${order}${
+    const apiUrl = `${rootUrl}/api/v1/search/${type}?keyword=${keyword}&peroid=${period}&brand_type=0&category=${caty === 'all' ? '' : caty}&order=${order}${
         rating === 'all' ? '' : rating === 'general' ? '&rating=general' : '&adult=1'
     }&per_page=30`;
     const response = await got({
         method: 'get',
         url: apiUrl,
+        headers: {
+            Cookie: Config.cookies ? Config.cookies : '',
+        },
     });
 
     let items = {};
@@ -32,7 +37,7 @@ module.exports = async (ctx) => {
             items = response.data.posts.map((item) => ({
                 title: item.title,
                 link: `${rootUrl}/posts/${item.id}`,
-                pubDate: new Date(item.posted_at).toUTCString(),
+                pubDate: parseDate(item.posted_at),
                 author: item.fanclub.fanclub_name_with_creator_name,
                 description: `${item.comment ? `<p>${item.comment}</p>` : ''}<img src="${item.thumb ? item.thumb.main : item.thumb_micro}">`,
             }));

package/lib/{routes → v2}/fantia/user.js

@@ -1,4 +1,6 @@
 const got = require('@/utils/got');
+const { parseDate } = require('@/utils/parse-date');
+const Config = require('@/config').value.fantia;
 
 module.exports = async (ctx) => {
     const rootUrl = 'https://fantia.jp';
@@ -6,25 +8,28 @@ module.exports = async (ctx) => {
     const response = await got({
         method: 'get',
         url: userUrl,
+        headers: {
+            Cookie: Config.cookies ? Config.cookies : '',
+        },
     });
-
     const list = response.data.fanclub.recent_posts.map((item) => ({
         title: item.title,
         link: `${rootUrl}/api/v1/posts/${item.id}`,
         description: `<p>${item.comment}</p>`,
-        pubDate: new Date(item.posted_at).toUTCString(),
+        pubDate: parseDate(item.posted_at),
     }));
-
     const items = await Promise.all(
         list.map((item) =>
             ctx.cache.tryGet(item.link, async () => {
                 const contentResponse = await got({
                     method: 'get',
                     url: item.link,
+                    headers: {
+                        Cookie: Config.cookies ? Config.cookies : '',
+                    },
                 });
-
                 item.link = item.link.replace('api/v1/', '');
-                item.description += `<img src="${contentResponse.data.post.thumb.large}">`;
+                item.description += `<img src="${contentResponse.data.post?.thumb?.large ? contentResponse.data.post?.thumb?.large : contentResponse.data.post.thumb_micro}">`;
 
                 return item;
             })

package/lib/v2/fastbull/express-news.js

@@ -20,7 +20,7 @@ module.exports = async (ctx) => {
 
             return {
                 title: item.find('.title_name').text(),
-                pubDate: parseDate(item.attr('data-date')),
+                pubDate: parseDate(parseInt(item.attr('data-date'))),
                 link: `${rootUrl}${item.find('.title_name').attr('href')}`,
             };
         });

package/lib/v2/fastbull/news.js

@@ -24,8 +24,8 @@ module.exports = async (ctx) => {
                 title: item.find('.title').text(),
                 link: `${rootUrl}${item.attr('href')}`,
                 author: item.find('.resource').text(),
-                pubDate: parseDate(item.find('.new_time').attr('data-date')),
                 description: item.find('.tips').text(),
+                pubDate: parseDate(parseInt(item.find('.new_time').attr('data-date'))),
             };
         });
 

package/lib/v2/finviz/maintainer.js

@@ -0,0 +1,4 @@
+module.exports = {
+    '/news/:ticker': ['HenryQW'],
+    '/:category?': ['nczitzk'],
+};

package/lib/v2/finviz/news.js

@@ -0,0 +1,45 @@
+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 ?? 'news';
+
+    const rootUrl = 'https://finviz.com';
+    const currentUrl = `${rootUrl}/news.ashx`;
+
+    const response = await got({
+        method: 'get',
+        url: currentUrl,
+    });
+
+    const $ = cheerio.load(response.data);
+
+    const items = $('.table-fixed')
+        .eq(category === 'blog' ? 1 : 0)
+        .find('.nn')
+        .toArray()
+        .map((item) => {
+            item = $(item);
+
+            const a = item.find('.nn-tab-link');
+
+            return {
+                title: a.text(),
+                link: a.attr('href'),
+                pubDate: timezone(parseDate(item.find('.nn-date').text(), ['MMM-DD', 'HH:mmA']), -4),
+                description:
+                    item
+                        .find('td[data-boxover]')
+                        .attr('data-boxover')
+                        ?.match(/<td class='news_tooltip-tab'>([\s\S]*)<\/td>/)[1] ?? '',
+            };
+        });
+
+    ctx.state.data = {
+        title: `finviz - ${category}`,
+        link: currentUrl,
+        item: items,
+    };
+};

package/lib/{routes/finviz/news.js → v2/finviz/quote.js}

@@ -1,13 +1,13 @@
 const got = require('@/utils/got');
 const cheerio = require('cheerio');
-const dateParser = require('@/utils/dateParser');
+const { parseDate } = require('@/utils/parse-date');
 
 module.exports = async (ctx) => {
     const link = `https://finviz.com/quote.ashx?t=${ctx.params.ticker}`;
-    const response = await got.get(link);
+    const response = await got(link);
 
     const $ = cheerio.load(response.body);
-    const data = $('table.fullview-news-outer tr').slice(0, 10);
+    const data = $('table.fullview-news-outer tr');
 
     let dateRow = '';
     const item = await Promise.all(
@@ -21,8 +21,8 @@ module.exports = async (ctx) => {
                 }
                 return {
                     title: $(e).find('a').text(),
-                    pubDate: dateParser(date, 'MMM-DD-YY HH:mmA'),
-                    authors: $(e).find('span').text(),
+                    pubDate: parseDate(date, 'MMM-DD-YY HH:mmA'),
+                    author: $(e).find('span').text(),
                     link: $(e).find('a').attr('href'),
                 };
             })

package/lib/v2/finviz/radar.js

@@ -0,0 +1,19 @@
+module.exports = {
+    'finviz.com': {
+        _name: 'finviz',
+        '.': [
+            {
+                title: 'News',
+                docs: 'https://docs.rsshub.app/finance.html#finviz-news',
+                source: ['/news.ashx', '/'],
+                target: '/finviz/:category?',
+            },
+            {
+                title: 'US Stock News',
+                docs: 'https://docs.rsshub.app/finance.html#finviz-mei-gu-gu-piao-xin-wen',
+                source: ['/quote.ashx', '/'],
+                target: '/finviz/news/:category?',
+            },
+        ],
+    },
+};

package/lib/v2/finviz/router.js

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