rsshub 1.0.0-master.f7cdf8b → 1.0.0-master.f7f8b7a

package/lib/{routes/universities → v2}/harvard/health/blog.js

@@ -1,5 +1,6 @@
 const got = require('@/utils/got');
 const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
 
 module.exports = async (ctx) => {
     const rootUrl = 'https://www.health.harvard.edu';
@@ -12,17 +13,16 @@ module.exports = async (ctx) => {
 
     const $ = cheerio.load(response.data);
 
-    const list = $('.text-2xl')
-        .slice(0, 10)
-        .map((_, item) => {
+    const list = $('.lg\\:text-2xl')
+        .toArray()
+        .map((item) => {
             item = $(item).parent();
 
             return {
                 title: item.text(),
                 link: item.attr('href'),
             };
-        })
-        .get();
+        });
 
     const items = await Promise.all(
         list.map((item) =>
@@ -33,10 +33,11 @@ module.exports = async (ctx) => {
                 });
                 const content = cheerio.load(detailResponse.data);
 
-                const date = item.link.substr(item.link.length - 12, 8);
+                const ldJson = JSON.parse(content('script[type="application/ld+json"]').text())['@graph'].find((i) => i['@type'] === 'Article');
 
                 item.description = content('.content-repository-content').html();
-                item.pubDate = Date.parse(`${date.substr(0, 4)}-${date.substr(4, 2)}-${date.substr(6, 2)}`);
+                item.pubDate = parseDate(ldJson.datePublished);
+                item.author = ldJson.author.name;
 
                 return item;
             })

package/lib/v2/harvard/maintainer.js

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

package/lib/v2/harvard/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'harvard.edu': {
+        _name: 'Harvard University',
+        'www.health': [
+            {
+                title: 'Health Blog',
+                docs: 'https://docs.rsshub.app/routes/new-media#harvard-health-publishing',
+                source: ['/blog'],
+                target: '/harvard/health/blog',
+            },
+        ],
+    },
+};

package/lib/v2/harvard/router.js

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

package/lib/v2/huxiu/briefColumn.js

@@ -1,41 +1,29 @@
 const got = require('@/utils/got');
-const { parseDate } = require('@/utils/parse-date');
-const { baseUrl, briefApi, ProcessFeed } = require('./utils');
+
+const { apiBriefRootUrl, processItems, fetchBriefColumnData } = require('./util');
 
 module.exports = async (ctx) => {
     const { id } = ctx.params;
-    const link = `${briefApi}/briefColumn/getContentListByCategoryId`;
-    const { data: response } = await got
-        .post(link, {
-            form: {
-                platform: 'www',
-                brief_column_id: id,
-            },
-        })
-        .json();
-    const {
-        data: { data: briefDetail },
-    } = await got.post(`${briefApi}/briefColumn/detail`, {
+    const limit = ctx.query.limit ? parseInt(ctx.query.limit, 10) : 20;
+
+    const apiUrl = new URL('briefColumn/getContentListByCategoryId', apiBriefRootUrl).href;
+
+    const { data: response } = await got.post(apiUrl, {
         form: {
             platform: 'www',
             brief_column_id: id,
+            pagesize: limit,
         },
     });
 
-    const list = response.datalist.map((item) => ({
-        title: item.title,
-        link: `${baseUrl}/brief/${item.brief_id}`,
-        description: item.preface,
-        pubDate: parseDate(item.publish_time, 'X'),
-    }));
+    ctx.state.json = response.data.datalist;
+
+    const items = await processItems(response.data.datalist, limit, ctx.cache.tryGet);
 
-    const items = await ProcessFeed(list, ctx.cache);
+    const data = await fetchBriefColumnData(id);
 
     ctx.state.data = {
-        title: `虎嗅 - ${briefDetail.name}`,
-        description: briefDetail.summary,
-        image: briefDetail.head_img,
-        link: `${baseUrl}/briefColumn/${id}.html`,
         item: items,
+        ...data,
     };
 };

package/lib/v2/huxiu/channel.js

@@ -0,0 +1,28 @@
+const got = require('@/utils/got');
+
+const { rootUrl, apiArticleRootUrl, processItems, fetchData } = require('./util');
+
+module.exports = async (ctx) => {
+    const { id } = ctx.params;
+    const limit = ctx.query.limit ? parseInt(ctx.query.limit, 10) : 20;
+
+    const apiUrl = new URL(`web/${id ? 'channel' : 'article'}/articleList`, apiArticleRootUrl).href;
+    const currentUrl = new URL(id ? `channel/${id}.html` : 'article', rootUrl).href;
+
+    const { data: response } = await got.post(apiUrl, {
+        form: {
+            platform: 'www',
+            channel_id: id,
+            pagesize: limit,
+        },
+    });
+
+    const items = await processItems(response.data?.dataList ?? response.data.datalist, limit, ctx.cache.tryGet);
+
+    const data = await fetchData(currentUrl);
+
+    ctx.state.data = {
+        item: items,
+        ...data,
+    };
+};

package/lib/v2/huxiu/club.js

@@ -0,0 +1,30 @@
+const got = require('@/utils/got');
+
+const { apiBriefRootUrl, processItems, fetchClubData } = require('./util');
+
+module.exports = async (ctx) => {
+    const { id } = ctx.params;
+    const limit = ctx.query.limit ? parseInt(ctx.query.limit, 10) : 20;
+
+    const apiUrl = new URL('club/briefList', apiBriefRootUrl).href;
+
+    const { data, briefColumnId } = await fetchClubData(id);
+
+    const { data: response } = await got.post(apiUrl, {
+        form: {
+            platform: 'www',
+            club_id: id,
+            brief_column_id: briefColumnId,
+            pagesize: limit,
+        },
+    });
+
+    ctx.state.json = response.data.datalist;
+
+    const items = await processItems(response.data.datalist, limit, ctx.cache.tryGet);
+
+    ctx.state.data = {
+        item: items,
+        ...data,
+    };
+};

package/lib/v2/huxiu/collection.js

@@ -1,39 +1,27 @@
 const got = require('@/utils/got');
-const cheerio = require('cheerio');
-const utils = require('./utils');
-const { parseDate } = require('@/utils/parse-date');
+
+const { rootUrl, apiArticleRootUrl, processItems, fetchData } = require('./util');
 
 module.exports = async (ctx) => {
     const { id } = ctx.params;
-    const link = `${utils.baseUrl}/collection/${id}.html`;
-    const { data: response } = await got(link, {
-        https: {
-            rejectUnauthorized: false,
-        },
-    });
+    const limit = ctx.query.limit ? parseInt(ctx.query.limit, 10) : 10;
 
-    const $ = cheerio.load(response);
+    const apiUrl = new URL('web/collection/articleList', apiArticleRootUrl).href;
+    const currentUrl = new URL(`collection/${id}.html`, rootUrl).href;
 
-    const initialState = utils.parseInitialState($);
-
-    const { collectionDetail } = initialState.collectionDetail;
-    const list = collectionDetail.article_list.datalist.map((e) => ({
-        title: e.title,
-        link: `${utils.baseUrl}/article/${e.aid}.html`,
-        description: e.summary,
-        pubDate: parseDate(e.dateline, 'X'),
-        author: e.user_info.username,
-    }));
+    const { data: response } = await got.post(apiUrl, {
+        form: {
+            platform: 'www',
+            collection_id: id,
+        },
+    });
 
-    const items = await utils.ProcessFeed(list, ctx.cache);
+    const items = await processItems(response.data.datalist, limit, ctx.cache.tryGet);
 
-    const info = `虎嗅 - ${collectionDetail.name}`;
+    const data = await fetchData(currentUrl);
 
     ctx.state.data = {
-        title: info,
-        description: collectionDetail.summary,
-        image: collectionDetail.icon,
-        link,
         item: items,
+        ...data,
     };
 };

package/lib/v2/huxiu/maintainer.js

@@ -1,8 +1,11 @@
 module.exports = {
-    '/article': ['HenryQW'],
-    '/author/:id': ['HenryQW'],
-    '/collection/:id': ['AlexdanerZe'],
+    '/article': ['HenryQW', 'nczitzk'],
+    '/briefcolumn/:id': ['Fatpandac', 'nczitzk'],
+    '/channel/:id?': ['nczitzk'],
+    '/club/:id': ['nczitzk'],
+    '/collection/:id': ['AlexdanerZe', 'nczitzk'],
+    '/member/:id/:type?': ['HenryQW', 'nczitzk'],
     '/moment': ['nczitzk'],
-    '/tag/:id': ['xyqfer', 'HenryQW'],
-    '/search/:keyword': ['xyqfer', 'HenryQW'],
+    '/search/:keyword': ['xyqfer', 'HenryQW', 'nczitzk'],
+    '/tag/:id': ['xyqfer', 'HenryQW', 'nczitzk'],
 };

package/lib/v2/huxiu/member.js

@@ -0,0 +1,27 @@
+const got = require('@/utils/got');
+
+const { rootUrl, apiMemberRootUrl, processItems, fetchData } = require('./util');
+
+module.exports = async (ctx) => {
+    const { id, type = 'article' } = ctx.params;
+    const limit = ctx.query.limit ? parseInt(ctx.query.limit, 10) : 10;
+
+    const apiUrl = new URL(`web/${type}/${type}List`, apiMemberRootUrl).href;
+    const currentUrl = new URL(`member/${id}${type === 'article' ? '' : `/${type}`}.html`, rootUrl).href;
+
+    const { data: response } = await got.post(apiUrl, {
+        form: {
+            platform: 'www',
+            uid: id,
+        },
+    });
+
+    const items = await processItems(response.data.datalist, limit, ctx.cache.tryGet);
+
+    const data = await fetchData(currentUrl);
+
+    ctx.state.data = {
+        item: items,
+        ...data,
+    };
+};

package/lib/v2/huxiu/moment.js

@@ -1,39 +1,25 @@
 const got = require('@/utils/got');
-const { parseDate } = require('@/utils/parse-date');
-const { art } = require('@/utils/render');
-const path = require('path');
-const { baseUrl: rootUrl, momentApi: apiRootUrl } = require('./utils');
+
+const { rootUrl, apiMomentRootUrl, processItems, fetchData } = require('./util');
 
 module.exports = async (ctx) => {
-    const currentUrl = `${rootUrl}/moment`;
-    const apiUrl = `${apiRootUrl}/web-v2/moment/feed`;
+    const limit = ctx.query.limit ? parseInt(ctx.query.limit, 10) : 20;
+
+    const apiUrl = new URL('web-v2/moment/feed', apiMomentRootUrl).href;
+    const currentUrl = new URL('moment', rootUrl).href;
 
-    const response = await got({
-        method: 'post',
-        url: apiUrl,
+    const { data: response } = await got.post(apiUrl, {
         form: {
-            last_dateline: parseInt(new Date().getTime() / 1000),
             platform: 'www',
-            is_ai: 0,
         },
     });
 
-    const items = response.data.data.moment_list.datalist[0].datalist.map((item) => ({
-        title: item.content,
-        link: item.share_url,
-        author: item.user_info.username,
-        pubDate: parseDate(item.publish_time * 1000),
-        description: art(path.join(__dirname, 'templates/moment.art'), {
-            description: item.content,
-            images: item.img_urls,
-            video: item.video,
-            comments: item.comment,
-        }),
-    }));
+    const items = await processItems(response.data.moment_list.datalist[0].datalist, limit, ctx.cache.tryGet);
+
+    const data = await fetchData(currentUrl);
 
     ctx.state.data = {
-        title: '虎嗅 - 24小时',
-        link: currentUrl,
         item: items,
+        ...data,
     };
 };

package/lib/v2/huxiu/radar.js

@@ -3,44 +3,52 @@ module.exports = {
         _name: '虎嗅',
         '.': [
             {
-                title: '首页资讯',
-                docs: 'https://docs.rsshub.app/routes/new-media#hu-xiu',
-                source: ['/article', '/'],
+                title: '资讯',
+                docs: 'https://docs.rsshub.app/routes/new-media#hu-xiu-zi-xun',
+                source: ['/article'],
                 target: '/huxiu/article',
             },
+            {
+                title: '频道',
+                docs: 'https://docs.rsshub.app/routes/new-media#hu-xiu-pin-dao',
+                source: ['/channel'],
+                target: (params) => `/huxiu/channel/${params.id.replace(/\.html$/, '')}`,
+            },
             {
                 title: '24小时',
-                docs: 'https://docs.rsshub.app/routes/new-media#hu-xiu',
-                source: ['/moment', '/'],
+                docs: 'https://docs.rsshub.app/routes/new-media#hu-xiu-24-xiao-shi',
+                source: ['/moment'],
                 target: '/huxiu/moment',
             },
             {
                 title: '标签',
-                docs: 'https://docs.rsshub.app/routes/new-media#hu-xiu',
+                docs: 'https://docs.rsshub.app/routes/new-media#hu-xiu-biao-qian',
                 source: ['/tags/:id'],
-                target: (params) => `/huxiu/tag/${params.id.replace('.html', '')}`,
+                target: (params) => `/huxiu/tag/${params.id.replace(/\.html$/, '')}`,
             },
             {
                 title: '搜索',
-                docs: 'https://docs.rsshub.app/routes/new-media#hu-xiu',
+                docs: 'https://docs.rsshub.app/routes/new-media#hu-xiu-sou-suo',
+                source: ['/'],
+                target: '/huxiu/search/:keyword',
             },
             {
                 title: '作者',
-                docs: 'https://docs.rsshub.app/routes/new-media#hu-xiu',
-                source: ['/member/:id/*', '/'],
-                target: '/huxiu/author/:id',
+                docs: 'https://docs.rsshub.app/routes/new-media#hu-xiu-zuo-zhe',
+                source: ['/member/:id/:type'],
+                target: (params) => `/huxiu/member/${params.id}/${params.type.replace(/\.html$/, '')}`,
             },
             {
                 title: '文集',
-                docs: 'https://docs.rsshub.app/routes/new-media#hu-xiu',
-                source: ['/collection/:id', '/'],
-                target: (params) => `/huxiu/collection/${params.id.replace('.html', '')}`,
+                docs: 'https://docs.rsshub.app/routes/new-media#hu-xiu-wen-ji',
+                source: ['/collection/:id'],
+                target: (params) => `/huxiu/collection/${params.id.replace(/\.html$/, '')}`,
             },
             {
                 title: '简报',
-                docs: 'https://docs.rsshub.app/routes/new-media#hu-xiu',
+                docs: 'https://docs.rsshub.app/routes/new-media#hu-xiu-jian-bao',
                 source: ['/briefColumn/:id', '/'],
-                target: (params) => `/huxiu/briefcolumn/${params.id.replace('.html', '')}`,
+                target: (params) => `/huxiu/briefcolumn/${params.id.replace(/\.html$/, '')}`,
             },
         ],
     },

package/lib/v2/huxiu/router.js

@@ -1,9 +1,12 @@
 module.exports = (router) => {
-    router.get('/article', require('./article'));
-    router.get('/author/:id', require('./author'));
+    router.get('/article', require('./channel'));
+    router.get('/author/:id/:type?', require('./member'));
+    router.get('/briefcolumn/:id', require('./briefColumn'));
+    router.get('/channel/:id?', require('./channel'));
+    router.get('/club/:id', require('./club'));
     router.get('/collection/:id', require('./collection'));
+    router.get('/member/:id/:type?', require('./member'));
     router.get('/moment', require('./moment'));
-    router.get('/tag/:id', require('./tag'));
     router.get('/search/:keyword', require('./search'));
-    router.get('/briefcolumn/:id', require('./briefColumn'));
+    router.get('/tag/:id', require('./tag'));
 };

package/lib/v2/huxiu/search.js

@@ -1,40 +1,35 @@
 const got = require('@/utils/got');
-const utils = require('./utils');
+
+const { rootUrl, apiSearchRootUrl, generateSignature, processItems, fetchData } = require('./util');
 
 module.exports = async (ctx) => {
     const { keyword } = ctx.params;
-    const url = `${utils.searchApi}/api/article`;
-    const link = utils.baseUrl;
+    const limit = ctx.query.limit ? parseInt(ctx.query.limit, 10) : 20;
 
-    const { data } = await got.post(url, {
-        headers: {
-            Referer: utils.baseUrl,
-        },
+    const apiUrl = new URL('api/article', apiSearchRootUrl).href;
+    const currentUrl = rootUrl;
+
+    const { data: response } = await got.post(apiUrl, {
         searchParams: {
             platform: 'www',
             s: keyword,
             sort: '',
             page: 1,
-            pagesize: 20,
-            appid: 'hx_search',
-            ...utils.generateSignature(),
+            pagesize: limit,
+            appid: 'hx_search_202303',
+            ...generateSignature(),
         },
     });
 
-    const list = data.data.datalist.map((d) => ({
-        title: d.title,
-        link: d.url.includes('article') ? `${d.url}.html` : d.url,
-        description: d.content,
-        author: d.author,
-    }));
+    const items = await processItems(response.data.datalist, limit, ctx.cache.tryGet);
+
+    const data = await fetchData(currentUrl);
+    data.title = `${keyword}-搜索结果-${data.title}`;
 
-    const items = await utils.ProcessFeed(list, ctx.cache);
+    ctx.state.json = response.data.datalist;
 
-    const info = `虎嗅网 - ${keyword}`;
     ctx.state.data = {
-        title: info,
-        link,
-        description: info,
         item: items,
+        ...data,
     };
 };

package/lib/v2/huxiu/tag.js

@@ -1,38 +1,26 @@
 const got = require('@/utils/got');
-const cheerio = require('cheerio');
-const utils = require('./utils');
-const { parseDate } = require('@/utils/parse-date');
-const timezone = require('@/utils/timezone');
+
+const { rootUrl, processItems, fetchData } = require('./util');
 
 module.exports = async (ctx) => {
     const { id } = ctx.params;
-    const link = `${utils.baseUrl}/tags/${id}.html`;
-    const { data } = await got(link, {
-        https: {
-            rejectUnauthorized: false,
+    const limit = ctx.query.limit ? parseInt(ctx.query.limit, 10) : 10;
+
+    const apiUrl = new URL('v2_action/tag_article_list', rootUrl).href;
+    const currentUrl = new URL(`tags/${id}.html`, rootUrl).href;
+
+    const { data: response } = await got.post(apiUrl, {
+        form: {
+            tag_id: id,
         },
     });
 
-    const $ = cheerio.load(data);
-    const list = $('.related-article li')
-        .toArray()
-        .map((e) => {
-            e = $(e);
-            const a = e.find('a');
-            return {
-                title: a.text(),
-                link: `${utils.baseUrl}${a.attr('href')}`,
-                pubDate: timezone(parseDate(e.find('.time').text()), 8),
-            };
-        });
+    const items = await processItems(response.data, limit, ctx.cache.tryGet);
 
-    const items = await utils.ProcessFeed(list, ctx.cache);
+    const data = await fetchData(currentUrl);
 
-    const info = `虎嗅 - ${$('h1').text()}`;
     ctx.state.data = {
-        title: info,
-        link,
-        description: info,
         item: items,
+        ...data,
     };
 };

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

@@ -0,0 +1,46 @@
+{{ if (!video || !video.src) && image?.src }}
+  <figure>
+    <img 
+      {{ if image.width }}
+        width={{ image.width }}
+      {{ /if }}
+      {{ if image.height }}
+        height={{ image.height }}
+      {{ /if }}
+    src="{{ image.src }}">
+  </figure>
+{{ /if }}
+
+{{ if audio?.src }}
+  <audio controls>
+    <source src="{{ audio.src }}" type="{{ audio.type }}">
+  </audio>
+{{ /if }}
+
+{{ if video?.src }}
+  <video
+    {{ set poster = video.poster || image?.src }}
+    {{ if poster }}
+      poster="{{ poster }}"
+    {{ /if }}
+    controls>
+    <source
+      src="{{ video.src }}"
+      type="{{ video.type }}">
+    <object data="{{ video.src }}">
+      <embed src="{{ video.src }}">
+    </object>
+  </video>
+{{ /if }}
+
+{{ if preface }}
+  {{@ preface }}
+{{ /if }}
+
+{{ if summary }}
+  {{@ summary }}
+{{ /if }}
+
+{{ if description }}
+  {{@ description }}
+{{ /if }}