rsshub 1.0.0-master.f78ad03 → 1.0.0-master.f796ff1

package/lib/routes/bbc/index.js

@@ -85,6 +85,7 @@ module.exports = async (ctx) => {
     ctx.state.data = {
         title,
         link,
+        image: 'https://www.bbc.com/favicon.ico',
         description: title,
         item: items,
     };

package/lib/routes/cell/cover.js

@@ -87,7 +87,7 @@ module.exports = async (ctx) => {
 
             const single = {
                 title: `${pageMeta.issueName} | ${pageMeta.issueNum} (${pageMeta.issueDate})`,
-                author: '@yech1990',
+                author: '@y9c',
                 description: contents,
                 link: address,
                 guid: address,

package/lib/routes/sankakucomplex/post.js

@@ -7,10 +7,7 @@ module.exports = async (ctx) => {
         url: 'https://idol.sankakucomplex.com/post/atom',
     });
 
-    const $ = cheerio.load(response.data, {
-        normalizeWhitespace: true,
-        xmlMode: true,
-    });
+    const $ = cheerio.load(response.data);
 
     ctx.state.data = {
         title: 'Posts | Idol Complex - Idol & Cosplay Images',

package/lib/routes/v2ex/topics.js

@@ -1,14 +1,10 @@
 const got = require('@/utils/got');
+const { parseDate } = require('@/utils/parse-date');
 
 module.exports = async (ctx) => {
     const type = ctx.params.type;
 
-    const response = await got({
-        method: 'get',
-        url: `https://www.v2ex.com/api/topics/${type}.json`,
-    });
-
-    const data = response.data;
+    const { data } = await got(`https://www.v2ex.com/api/topics/${type}.json`);
 
     let title;
     if (type === 'hot') {
@@ -25,10 +21,10 @@ module.exports = async (ctx) => {
             title: item.title,
             description: `${item.member.username}: ${item.content_rendered}`,
             content: { text: item.content, html: item.content_rendered },
-            pubDate: new Date(item.created * 1000).toUTCString(),
-            guid: item.id,
+            pubDate: parseDate(item.created, 'X'),
             link: item.url,
             author: item.member.username,
+            comments: item.replies,
         })),
     };
 };

package/lib/utils/parse-date.js

@@ -70,19 +70,19 @@ const patterns = [
     },
     {
         unit: 'days',
-        regExp: /(\d+)(?:天|日|day(?:s)?)/,
+        regExp: /(\d+)(?:天|日|d(?:ay)?(?:s)?)/,
     },
     {
         unit: 'hours',
-        regExp: /(\d+)(?:(?:个|個)?(?:(?:小)?时|時|点|點)|h(?:ou)?r(?:s)?)/,
+        regExp: /(\d+)(?:(?:个|個)?(?:(?:小)?时|時|点|點)|h(?:(?:ou)?r)?(?:s)?)/,
     },
     {
         unit: 'minutes',
-        regExp: /(\d+)(?:分(?:钟|鐘)?|min(?:ute)?(?:s)?)/,
+        regExp: /(\d+)(?:分(?:钟|鐘)?|m(?:in(?:ute)?)?(?:s)?)/,
     },
     {
         unit: 'seconds',
-        regExp: /(\d+)(?:秒(?:钟|鐘)?|sec(?:ond)?(?:s)?)/,
+        regExp: /(\d+)(?:秒(?:钟|鐘)?|s(?:ec(?:ond)?)?(?:s)?)/,
     },
 ];
 

package/lib/utils/puppeteer.js

@@ -1,5 +1,6 @@
 const config = require('@/config').value;
 let puppeteer = require('puppeteer');
+const proxyChain = require('proxy-chain');
 
 const options = {
     args: ['--no-sandbox', '--disable-setuid-sandbox', '--disable-infobars', '--window-position=0,0', '--ignore-certificate-errors', '--ignore-certificate-errors-spki-list', `--user-agent=${config.ua}`],
@@ -7,6 +8,13 @@ const options = {
     ignoreHTTPSErrors: true,
 };
 
+let proxyUri;
+if (config.proxyUri && typeof config.proxyUri === 'string') {
+    proxyUri = config.proxyUri;
+} else if (config.proxy && config.proxy.protocol && config.proxy.host && config.proxy.port) {
+    proxyUri = `${config.proxy.protocol}://${config.proxy.host}:${config.proxy.port}`;
+}
+
 /**
  * @param {Object} extraOptions
  * @param {boolean} extraOptions.stealth - Use puppeteer-extra-plugin-stealth
@@ -41,6 +49,9 @@ module.exports = async (extraOptions = {}) => {
         puppeteer.use(require('puppeteer-extra-plugin-stealth')());
     }
     let browser;
+    if (proxyUri) {
+        options.args.push(`--proxy-server=${await proxyChain.anonymizeProxy(proxyUri)}`);
+    }
     if (config.puppeteerWSEndpoint) {
         browser = await puppeteer.connect({
             browserWSEndpoint: config.puppeteerWSEndpoint,

package/lib/utils/request-wrapper.js

@@ -73,11 +73,17 @@ const requestWrapper = (url, options) => {
     if (agent && new RegExp(config.proxy.url_regex).test(url)) {
         let agentResult;
         try {
-            agentResult = agent[(options.protocol || url.match(/(https?:)/)[1]).slice(0, -1)];
+            agentResult = agent[(options.protocol || url.match(/(^https?:)/)[1]).slice(0, -1)];
         } catch (error) {
             agentResult = null;
         }
-        options.agent = agentResult;
+        try {
+            if (new URL(url).host !== new URL(config.proxyUri).host) {
+                options.agent = agentResult;
+            }
+        } catch (error) {
+            options.agent = agentResult;
+        }
 
         if (config.proxy.auth) {
             if (!options.headers) {

package/lib/utils/wechat-mp.js

@@ -7,6 +7,8 @@
  * lib/v2/gzh360
  * lib/v2/pku/nsd/gd
  * lib/v2/sdu/cs
+ * lib/v2/nua/utils
+ * lib/v2/hrbeu
  *
  * If your new route is not in the above folders, please add it to the list.
  *

package/lib/v2/18comic/utils.js

@@ -24,7 +24,7 @@ module.exports = {
 
                 return {
                     title: item.text().trim(),
-                    link: `${rootUrl}${item.prev().prev().attr('href')}`,
+                    link: `${rootUrl}${item.prev().find('a').attr('href')}`,
                 };
             });
 
@@ -52,7 +52,7 @@ module.exports = {
                         .join(', ');
                     item.description = `<p>${content('#intro-block .p-t-5').text()}</p><img src="${content('.img_zoom_img img')
                         .toArray()
-                        .map((image) => content(image).attr('src'))
+                        .map((image) => content(image).attr('data-original'))
                         .join('"><img src="')}">`;
 
                     return item;

package/lib/v2/35photo/genre.js

@@ -4,6 +4,7 @@ module.exports = async (ctx) => {
     const id = ctx.params.id;
 
     const currentUrl = `${rootUrl}/genre_${id}`;
+    const apiUrl = `${rootUrl}/show_block.php?type=getNextPageData&page=genre&community_id=${id}`;
 
-    ctx.state.data = await ProcessItems(currentUrl);
+    ctx.state.data = await ProcessItems(currentUrl, apiUrl);
 };

package/lib/v2/35photo/utils.js

@@ -5,13 +5,28 @@ const rootUrl = 'https://35photo.pro';
 
 module.exports = {
     rootUrl,
-    ProcessItems: async (currentUrl) => {
-        const response = await got({
+    ProcessItems: async (currentUrl, apiUrl) => {
+        let $, response;
+
+        response = await got({
             method: 'get',
             url: currentUrl,
         });
 
-        const $ = cheerio.load(response.data);
+        const data = response.data;
+
+        $ = cheerio.load(data);
+
+        const title = $('h1').text();
+
+        if (apiUrl) {
+            response = await got({
+                method: 'get',
+                url: `${apiUrl}&lastId=${$('div.countLike').last().attr('photo-id')}`,
+            });
+
+            $ = cheerio.load(data + response.data.data);
+        }
 
         const items = $('.item')
             .toArray()
@@ -34,7 +49,7 @@ module.exports = {
             });
 
         return {
-            title: `${$('h1').text()} - 35PHOTO`,
+            title: `${title} - 35PHOTO`,
             link: currentUrl,
             item: items,
         };

package/lib/v2/36kr/index.js

@@ -27,16 +27,19 @@ module.exports = async (ctx) => {
 
     const data = JSON.parse(response.data.match(/"itemList":(\[.*?\])/)[1]);
 
-    let items = data.slice(0, ctx.query.limit ? parseInt(ctx.query.limit) : 30).map((item) => {
-        item = item.templateMaterial ?? item;
-        return {
-            title: item.widgetTitle.replace(/<\/?em>/g, ''),
-            author: item.author,
-            pubDate: parseDate(item.publishTime),
-            link: `${rootUrl}/${path === '/newsflashes' ? 'newsflashes' : 'p'}/${item.itemId}`,
-            description: item.widgetContent ?? item.content,
-        };
-    });
+    let items = data
+        .slice(0, ctx.query.limit ? parseInt(ctx.query.limit) : 30)
+        .filter((item) => item.itemType !== 0)
+        .map((item) => {
+            item = item.templateMaterial ?? item;
+            return {
+                title: item.widgetTitle.replace(/<\/?em>/g, ''),
+                author: item.author,
+                pubDate: parseDate(item.publishTime),
+                link: `${rootUrl}/${path === '/newsflashes' ? 'newsflashes' : 'p'}/${item.itemId}`,
+                description: item.widgetContent ?? item.content,
+            };
+        });
 
     if (!/^\/(search|newsflashes)/.test(path)) {
         items = await Promise.all(
@@ -47,20 +50,22 @@ module.exports = async (ctx) => {
                         url: item.link,
                     });
 
-                    const key = CryptoJS.enc.Utf8.parse('efabccee-b754-4c');
-                    const cipherText = detailResponse.data.match(/{"state":"(.*)","isEncrypt":true}/)[1];
-
-                    const content = JSON.parse(
-                        CryptoJS.AES.decrypt(cipherText, key, {
-                            mode: CryptoJS.mode.ECB,
-                            padding: CryptoJS.pad.Pkcs7,
-                        })
-                            .toString(CryptoJS.enc.Utf8)
-                            .toString()
-                    ).articleDetail.articleDetailData.data;
-
-                    item.description = content.widgetContent;
-
+                    const cipherTextList = detailResponse.data.match(/{"state":"(.*)","isEncrypt":true}/) ?? [];
+                    if (cipherTextList.length !== 0) {
+                        const key = CryptoJS.enc.Utf8.parse('efabccee-b754-4c');
+                        const content = JSON.parse(
+                            CryptoJS.AES.decrypt(cipherTextList[1], key, {
+                                mode: CryptoJS.mode.ECB,
+                                padding: CryptoJS.pad.Pkcs7,
+                            })
+                                .toString(CryptoJS.enc.Utf8)
+                                .toString()
+                        ).articleDetail.articleDetailData.data;
+                        item.description = content.widgetContent;
+                    } else {
+                        const $ = cheerio.load(detailResponse.body);
+                        item.description = $('div.articleDetailContent').html();
+                    }
                     return item;
                 })
             )

package/lib/v2/8world/index.js

@@ -3,10 +3,10 @@ const cheerio = require('cheerio');
 const { parseDate } = require('@/utils/parse-date');
 
 module.exports = async (ctx) => {
-    const category = ctx.params.category ?? 'realtime';
+    const path = ctx.path === '/' ? '/realtime' : ctx.path;
 
     const rootUrl = 'https://www.8world.com';
-    const currentUrl = `${rootUrl}/${category}`;
+    const currentUrl = `${rootUrl}${path}`;
 
     const response = await got({
         method: 'get',
@@ -15,19 +15,19 @@ module.exports = async (ctx) => {
 
     const $ = cheerio.load(response.data);
 
-    const list = $('div[data-column="Two-Third"] .article-title .article-link')
-        .map((_, item) => {
+    let items = $('div[data-column="Two-Third"] .article-title .article-link')
+        .toArray()
+        .map((item) => {
             item = $(item);
 
             return {
                 title: item.text(),
                 link: `${rootUrl}${item.attr('href')}`,
             };
-        })
-        .get();
+        });
 
-    const items = await Promise.all(
-        list.map((item) =>
+    items = await Promise.all(
+        items.map((item) =>
             ctx.cache.tryGet(item.link, async () => {
                 const detailResponse = await got({
                     method: 'get',

package/lib/v2/8world/maintainer.js

@@ -1,3 +1,4 @@
 module.exports = {
     '/:category?': ['nczitzk'],
+    '/topic/:id': ['nczitzk'],
 };

package/lib/v2/8world/radar.js

@@ -8,6 +8,12 @@ module.exports = {
                 source: ['/:category', '/'],
                 target: '/8world/:category?',
             },
+            {
+                title: '标签',
+                docs: 'https://docs.rsshub.app/new-media.html#_8-shi-jie-biao-qian',
+                source: ['/topic/:id', '/'],
+                target: '/8world/topic/:id',
+            },
         ],
     },
 };

package/lib/v2/8world/router.js

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

package/lib/v2/aicaijing/index.js

@@ -0,0 +1,73 @@
+const got = require('@/utils/got');
+const { parseDate } = require('@/utils/parse-date');
+const { art } = require('@/utils/render');
+const path = require('path');
+
+module.exports = async (ctx) => {
+    const category = ctx.params.category ?? 'latest';
+    const id = ctx.params.id ?? 14;
+
+    const titles = {
+        14: '热点 - 最新',
+        5: '热点 - 科技',
+        9: '热点 - 消费',
+        7: '热点 - 出行',
+        13: '热点 - 文娱',
+        10: '热点 - 教育',
+        25: '热点 - 地产',
+        11: '热点 - 更多',
+        28: '深度 - 出行',
+        29: '深度 - 科技',
+        31: '深度 - 消费',
+        33: '深度 - 教育',
+        34: '深度 - 更多',
+        8: '深度 - 地产',
+        6: '深度 - 文娱',
+    };
+
+    const categories = {
+        latest: {
+            url: '',
+            title: '最新文章',
+        },
+        recommend: {
+            url: '&isRecommend=true',
+            title: '推荐资讯',
+        },
+        cover: {
+            url: '&position=1',
+            title: '封面文章',
+        },
+        information: {
+            url: `&categoryId=${id}`,
+            title: titles[id],
+        },
+    };
+
+    const rootUrl = 'https://www.aicaijing.com.cn';
+    const apiRootUrl = 'https://api.aicaijing.com.cn';
+    const apiUrl = `${apiRootUrl}/article/detail/list?size=${ctx.query.limit ?? 50}&page=1${categories[category].url}`;
+
+    const response = await got({
+        method: 'get',
+        url: apiUrl,
+    });
+
+    const items = response.data.data.items.map((item) => ({
+        title: item.title,
+        link: `${rootUrl}/article/${item.articleId}`,
+        author: item.userInfo.nickname,
+        pubDate: parseDate(item.createTime),
+        category: [item.category.name].concat(item.tags.map((t) => t.name)),
+        description: art(path.join(__dirname, 'templates/description.art'), {
+            image: item.cover,
+            description: item.content,
+        }),
+    }));
+
+    ctx.state.data = {
+        title: `AI 财经社 - ${categories[category].title}`,
+        link: rootUrl,
+        item: items,
+    };
+};

package/lib/v2/aicaijing/maintainer.js

@@ -0,0 +1,6 @@
+module.exports = {
+    '/cover': ['nczitzk'],
+    '/information/:id?': ['nczitzk'],
+    '/latest': ['nczitzk'],
+    '/recommend': ['nczitzk'],
+};

package/lib/v2/aicaijing/radar.js

@@ -0,0 +1,31 @@
+module.exports = {
+    'aicaijing.com': {
+        _name: 'AI 财经社',
+        www: [
+            {
+                title: '最新文章',
+                docs: 'https://docs.rsshub.app/finance.html#ai-cai-jing-she-zui-xin-wen-zhang',
+                source: ['/'],
+                target: '/aicaijing/latest',
+            },
+            {
+                title: '封面文章',
+                docs: 'https://docs.rsshub.app/finance.html#ai-cai-jing-she-feng-mian-wen-zhang',
+                source: ['/'],
+                target: '/aicaijing/cover',
+            },
+            {
+                title: '推荐资讯',
+                docs: 'https://docs.rsshub.app/finance.html#ai-cai-jing-she-tui-jian-zi-xun',
+                source: ['/'],
+                target: '/aicaijing/recommend',
+            },
+            {
+                title: '热点 & 深度',
+                docs: 'https://docs.rsshub.app/finance.html#ai-cai-jing-she-re-dian-shen-du',
+                source: ['/information/:id', '/'],
+                target: '/aicaijing/information/:id?',
+            },
+        ],
+    },
+};

package/lib/v2/aicaijing/router.js

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

package/lib/v2/aicaijing/templates/description.art

@@ -0,0 +1,4 @@
+{{ if image }}
+<img src="{{ image }}">
+{{ /if }}
+{{@ description }}

package/lib/v2/aiea/index.js

@@ -0,0 +1,39 @@
+const buildData = require('@/utils/common-config');
+
+module.exports = async (ctx) => {
+    const link = 'http://www.aiea.org/0504';
+    const period = ctx.params.period ?? '';
+
+    let nth_child = 'n';
+    switch (period) {
+        case 'upcoming':
+            nth_child = '1';
+            break;
+
+        case 'past':
+            nth_child = '2';
+            break;
+
+        case 'both':
+            nth_child = 'n';
+            break;
+
+        default:
+            break;
+    }
+
+    ctx.state.data = await buildData({
+        link,
+        url: link,
+        title: `%title%`,
+        params: {
+            title: 'AIEA Seminars',
+        },
+        item: {
+            item: `.seminar-contents .seminar-partWrap:nth-child(${nth_child}) > .seminar-list`,
+            title: `$('.seminar-list-title > span').text()`,
+            link: `$('a[href^="/0504"]').attr('href')`,
+            description: `$('.seminar-list .txt > .title').text()`,
+        },
+    });
+};

package/lib/v2/aiea/maintainer.js

@@ -0,0 +1,3 @@
+module.exports = {
+    '/seminars/:period': ['zxx-457'],
+};

package/lib/v2/aiea/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'aiea.org': {
+        _name: 'Asian Innovation and Entrepreneurship Association',
+        www: [
+            {
+                title: 'Seminar Series',
+                docs: 'https://docs.rsshub.app/study.html#AIEA-Seminar-Series',
+                source: ['/0504', '/'],
+                target: '/aiea/seminars/upcoming',
+            },
+        ],
+    },
+};

package/lib/v2/aiea/router.js

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

package/lib/v2/aliyundrive/files.js

@@ -0,0 +1,63 @@
+const got = require('@/utils/got');
+
+module.exports = async (ctx) => {
+    const { share_id, parent_file_id } = ctx.params;
+    const url = `https://www.aliyundrive.com/s/${share_id}${parent_file_id ? `/folder/${parent_file_id}` : ''}`;
+
+    const shareRes = await got.post('https://api.aliyundrive.com/adrive/v3/share_link/get_share_by_anonymous', {
+        params: {
+            share_id,
+        },
+        json: {
+            share_id,
+        },
+        headers: {
+            referer: 'https://www.aliyundrive.com/',
+            origin: 'https://www.aliyundrive.com',
+            'x-canary': 'client=web,app=share,version=v2.3.1',
+        },
+    });
+
+    const tokenRes = await got.post('https://api.aliyundrive.com/v2/share_link/get_share_token', {
+        headers: {
+            referer: 'https://www.aliyundrive.com/',
+            origin: 'https://www.aliyundrive.com',
+            'x-canary': 'client=web,app=share,version=v2.3.1',
+        },
+        json: {
+            share_id,
+        },
+    });
+    const share_token = tokenRes.data.share_token;
+
+    const listRes = await got.post('https://api.aliyundrive.com/v2/file/list', {
+        headers: {
+            referer: 'https://www.aliyundrive.com/',
+            origin: 'https://www.aliyundrive.com',
+            'x-canary': 'client=web,app=share,version=v2.3.1',
+            'x-share-token': share_token,
+        },
+        json: {
+            limit: 100,
+            order_by: 'created_at',
+            order_direction: 'DESC',
+            parent_file_id: parent_file_id || 'root',
+            share_id,
+        },
+    });
+
+    const result = listRes.data.items.map((item) => ({
+        title: item.name,
+        description: item.name + (item.thumbnail ? `<img src="${item.thumbnail}">` : ''),
+        link: url,
+        pubDate: item.created_at,
+        updated: item.updated_at,
+        guid: item.file_id,
+    }));
+
+    ctx.state.data = {
+        title: `${shareRes.data.display_name || `${share_id}${parent_file_id ? `-${parent_file_id}` : ''}`}-阿里云盘`,
+        link: url,
+        item: result,
+    };
+};

package/lib/v2/aliyundrive/maintainer.js

@@ -0,0 +1,3 @@
+module.exports = {
+    '/files/:share_id/:parent_file_id?': ['DIYgod'],
+};

package/lib/v2/aliyundrive/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'aliyundrive.com': {
+        _name: '阿里云盘',
+        www: [
+            {
+                title: '文件列表',
+                docs: 'https://docs.rsshub.app/programming.html#a-li-yun-pan',
+                source: ['/s/:share_id', '/s/:share_id/folder/:parent_file_id'],
+                target: '/aliyundrive/files/:share_id/:parent_file_id?',
+            },
+        ],
+    },
+};

package/lib/v2/aliyundrive/router.js

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