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

package/lib/v2/hrbust/utils.js

@@ -0,0 +1,71 @@
+const cheerio = require('cheerio');
+const got = require('@/utils/got');
+const { art } = require('@/utils/render');
+const path = require('path');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+
+// const base = 'http://hrbust.edu.cn';
+const jwzxBase = 'http://jwzx.hrbust.edu.cn/homepage/';
+
+const columnIdBase = (id) => (id ? `${jwzxBase}infoArticleList.do?columnId=${id}` : `${jwzxBase}infoArticleList.do?columnId=354`);
+
+const renderDesc = (desc) =>
+    art(path.join(__dirname, 'templates/description.art'), {
+        desc,
+    });
+
+const detailPage = (e, cache) =>
+    cache.tryGet(e.detailPage, async () => {
+        const result = await got(e.detailPage);
+        const $ = cheerio.load(result.data);
+        const title = $('div#article h2').text().trim();
+        const desc =
+            $('div.body').html() &&
+            $('div.body')
+                .html()
+                .replace(/style="(.*?)"/g, '')
+                .trim();
+        const pubDate = timezone(
+            parseDate(
+                $('div#articleInfo ul li')
+                    .first()
+                    .text()
+                    .replace(/发布日期：/g, '')
+                    .trim()
+            ),
+            +8
+        );
+        return {
+            title: e.title || title,
+            description: renderDesc(desc),
+            link: e.detailPage,
+            pubDate: e.pubDate || pubDate,
+        };
+    });
+
+const fetchAllArticle = (data, base) => {
+    const $ = cheerio.load(data);
+    const article = $('.articleList li div');
+    const info = article
+        .map((i, e) => {
+            const c = cheerio.load(e);
+            const r = {
+                title: c('a').first().text().trim(),
+                detailPage: new URL(c('a').attr('href'), base).href,
+                pubDate: timezone(parseDate(c('span').first().text().trim()), +8),
+            };
+            return r;
+        })
+        .get();
+    return info;
+};
+
+module.exports = {
+    // BASE: base,
+    JWZXBASE: jwzxBase,
+    columnIdBase,
+    fetchAllArticle,
+    detailPage,
+    renderDesc,
+};

package/lib/v2/huanqiu/index.js

@@ -1,6 +1,7 @@
 const got = require('@/utils/got');
 const cheerio = require('cheerio');
 const { parseDate } = require('@/utils/parse-date');
+const { isValidHost } = require('@/utils/valid-host');
 
 function getKeysRecursive(dic, key, attr, array) {
     Object.values(dic).forEach((v) => {
@@ -15,6 +16,10 @@ function getKeysRecursive(dic, key, attr, array) {
 
 module.exports = async (ctx) => {
     const category = ctx.params.category ?? 'china';
+    if (!isValidHost(category)) {
+        throw Error('Invalid category');
+    }
+
     const host = 'https://' + category + '.huanqiu.com';
 
     const resp = await got({

package/lib/v2/hunanpea/maintainer.js

@@ -0,0 +1,3 @@
+module.exports = {
+    '/rsks/:guid': ['TonyRL'],
+};

package/lib/v2/hunanpea/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'hunanpea.com': {
+        _name: '湖南人事考试网',
+        rsks: [
+            {
+                title: '公告',
+                docs: 'https://docs.rsshub.app/study.html#hu-nan-ren-shi-kao-shi-wang',
+                source: ['/Category/:guid/ArticlesByCategory.do'],
+                target: '/hunanpea/rsks/:guid',
+            },
+        ],
+    },
+};

package/lib/v2/hunanpea/router.js

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

package/lib/v2/hunanpea/rsks.js

@@ -0,0 +1,42 @@
+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 baseUrl = 'http://rsks.hunanpea.com';
+    const { guid } = ctx.params;
+    const link = `${baseUrl}/Category/${guid}/ArticlesByCategory.do?PageIndex=1`;
+    const { data: response } = await got(link);
+
+    const $ = cheerio.load(response);
+    const list = $('#column_content > ul > li')
+        .toArray()
+        .map((item) => {
+            item = $(item);
+            return {
+                title: item.find('a').attr('title'),
+                link: `${baseUrl}${item.find('a').attr('href').replace('ArticleDetail.do', 'InternalArticleDetail.do?')}`,
+                pubDate: timezone(parseDate(item.find('em').text()), +8),
+            };
+        });
+
+    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 = $('.content_area').html();
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: `${$('.sitemap h2').text()} - ${$('head title').text()}`,
+        link,
+        item: items,
+    };
+};

package/lib/v2/huxiu/article.js

@@ -1,23 +1,26 @@
 const got = require('@/utils/got');
-const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
 const utils = require('./utils');
 
 module.exports = async (ctx) => {
-    const link = 'https://www.huxiu.com/article';
-    const response = await got({
-        method: 'get',
-        url: link,
+    const link = `${utils.baseUrl}/article/`;
+    const { data } = await got.post(`${utils.articleApi}/web/article/articleList`, {
         headers: {
             Referer: link,
         },
+        form: {
+            platform: 'www',
+            pagesize: ctx.query.limit ? parseInt(ctx.query.limit) : 22,
+        },
     });
 
-    const $ = cheerio.load(response.data);
-
-    const list = $('.article-item-wrap > a')
-        .slice(0, 10)
-        .get()
-        .map((e) => $(e).attr('href'));
+    const list = data.data.dataList.map((item) => ({
+        title: item.title,
+        link: `${utils.baseUrl}/article/${item.aid}.html`,
+        description: item.summary,
+        pubDate: parseDate(item.dateline, 'X'),
+        author: item.user_info.username,
+    }));
 
     const items = await utils.ProcessFeed(list, ctx.cache);
 

package/lib/v2/huxiu/author.js

@@ -1,28 +1,14 @@
 const got = require('@/utils/got');
-const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
 const utils = require('./utils');
 
 module.exports = async (ctx) => {
     const { id } = ctx.params;
-    const link = `https://www.huxiu.com/member/${id}/article.html`;
-
-    const response = await got({
-        method: 'get',
-        url: link,
-        headers: {
-            Referer: link,
-        },
-    });
-
-    const $ = cheerio.load(response.data);
-    const author = $('.author-name').text().trim();
+    const link = `${utils.baseUrl}/member/${id}/article.html`;
 
     const articleList = await got({
         method: 'post',
-        url: 'https://account-api.huxiu.com/web/article/articleList',
-        headers: {
-            Referer: 'https://www.huxiu.com/',
-        },
+        url: `${utils.accountApi}/web/article/articleList`,
         form: {
             platform: 'www',
             uid: id,
@@ -31,15 +17,20 @@ module.exports = async (ctx) => {
         },
     }).json();
 
-    const list = articleList.data.datalist.slice(0, 10).map((item) => item.url);
+    const list = articleList.data.datalist.map((item) => ({
+        title: item.title,
+        link: `${utils.baseUrl}/article/${item.aid}.html`,
+        description: item.summary,
+        pubDate: parseDate(item.time),
+    }));
 
     const items = await utils.ProcessFeed(list, ctx.cache);
 
-    const authorInfo = `虎嗅网 - ${author}`;
+    const authorInfo = `虎嗅网 - ${items[0].author}`;
+
     ctx.state.data = {
         title: authorInfo,
         link,
-        description: authorInfo,
         item: items,
     };
 };

package/lib/v2/huxiu/briefColumn.js

@@ -1,51 +1,41 @@
 const got = require('@/utils/got');
-const cheerio = require('cheerio');
-const { parseDate, parseRelativeDate } = require('@/utils/parse-date');
+const { parseDate } = require('@/utils/parse-date');
+const { baseUrl, briefApi, ProcessFeed } = require('./utils');
 
 module.exports = async (ctx) => {
     const { id } = ctx.params;
-    const link = `https://api-brief.huxiu.com/briefColumn/getContentListByCategoryId`;
-    const response = await got
+    const link = `${briefApi}/briefColumn/getContentListByCategoryId`;
+    const { data: response } = await got
         .post(link, {
-            headers: {
-                'Content-Type': 'application/x-www-form-urlencoded',
+            form: {
+                platform: 'www',
+                brief_column_id: id,
             },
-            body: `platform=www&brief_column_id=${id}&last_id=`,
         })
         .json();
-
-    const list = response.data.datalist;
-
-    const items = await Promise.all(
-        list.map((item) => {
-            const link = `https://www.huxiu.com/brief/${item.brief_id}`;
-
-            return ctx.cache.tryGet(link, async () => {
-                const detailResponse = await got(link);
-                const $ = cheerio.load(detailResponse.data);
-
-                const description = $('#js-part').html();
-
-                return {
-                    title: item.title,
-                    pubDate: item.format_publish_time.includes('前') ? parseRelativeDate(item.format_publish_time) : parseDate(item.format_publish_time),
-                    description,
-                    link,
-                };
-            });
-        })
-    );
-
-    const title = async () => {
-        const res = await got(`https://www.huxiu.com/briefColumn/${id}.html`);
-        const $ = cheerio.load(res.data);
-
-        return $('#top > div:nth-child(1) > div > div.brief-content__name').text();
-    };
+    const {
+        data: { data: briefDetail },
+    } = await got.post(`${briefApi}/briefColumn/detail`, {
+        form: {
+            platform: 'www',
+            brief_column_id: id,
+        },
+    });
+
+    const list = response.datalist.map((item) => ({
+        title: item.title,
+        link: `${baseUrl}/brief/${item.brief_id}`,
+        description: item.preface,
+        pubDate: parseDate(item.publish_time, 'X'),
+    }));
+
+    const items = await ProcessFeed(list, ctx.cache);
 
     ctx.state.data = {
-        title: `虎嗅 - ${await title()}`,
-        link: `https://www.huxiu.com/briefColumn/${id}.html`,
+        title: `虎嗅 - ${briefDetail.name}`,
+        description: briefDetail.summary,
+        image: briefDetail.head_img,
+        link: `${baseUrl}/briefColumn/${id}.html`,
         item: items,
     };
 };

package/lib/v2/huxiu/collection.js

@@ -1,34 +1,39 @@
 const got = require('@/utils/got');
 const cheerio = require('cheerio');
 const utils = require('./utils');
+const { parseDate } = require('@/utils/parse-date');
 
 module.exports = async (ctx) => {
     const { id } = ctx.params;
-    const link = `https://www.huxiu.com/collection/${id}.html`;
-    const response = await got({
-        method: 'get',
-        url: link,
-        headers: {
-            Referer: link,
+    const link = `${utils.baseUrl}/collection/${id}.html`;
+    const { data: response } = await got(link, {
+        https: {
+            rejectUnauthorized: false,
         },
     });
 
-    const data = response.data;
-    const $ = cheerio.load(data);
+    const $ = cheerio.load(response);
 
-    const list = $('.collection-lirt-item > a')
-        .slice(0, 5)
-        .get()
-        .map((e) => $(e).attr('href'));
+    const initialState = utils.parseInitialState($);
+
+    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 items = await utils.ProcessFeed(list, ctx.cache);
 
-    const info = `虎嗅 - ${$('h1').text()}`;
-    const description = $('p').text();
+    const info = `虎嗅 - ${collectionDetail.name}`;
+
     ctx.state.data = {
         title: info,
+        description: collectionDetail.summary,
+        image: collectionDetail.icon,
         link,
-        description,
         item: items,
     };
 };

package/lib/v2/huxiu/moment.js

@@ -2,11 +2,10 @@ 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');
 
 module.exports = async (ctx) => {
-    const rootUrl = 'https://www.huxiu.com';
     const currentUrl = `${rootUrl}/moment`;
-    const apiRootUrl = 'https://moment-api.huxiu.com';
     const apiUrl = `${apiRootUrl}/web-v2/moment/feed`;
 
     const response = await got({

package/lib/v2/huxiu/search.js

@@ -3,18 +3,30 @@ const utils = require('./utils');
 
 module.exports = async (ctx) => {
     const { keyword } = ctx.params;
-    const url = `https://search-api.huxiu.com/api/article?platform=www&s=${encodeURIComponent(keyword)}&sort=&page=1&pagesize=20`;
-    const link = 'https://www.huxiu.com';
+    const url = `${utils.searchApi}/api/article`;
+    const link = utils.baseUrl;
 
-    const response = await got({
-        method: 'post',
-        url,
+    const { data } = await got.post(url, {
         headers: {
-            Referer: link,
+            Referer: utils.baseUrl,
+        },
+        searchParams: {
+            platform: 'www',
+            s: keyword,
+            sort: '',
+            page: 1,
+            pagesize: 20,
+            appid: 'hx_search',
+            ...utils.generateSignature(),
         },
     });
 
-    const list = response.data.data.datalist.slice(0, 10).map((d) => `${link}/article/${d.aid}.html`);
+    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 utils.ProcessFeed(list, ctx.cache);
 

package/lib/v2/huxiu/tag.js

@@ -1,23 +1,30 @@
 const got = require('@/utils/got');
 const cheerio = require('cheerio');
 const utils = require('./utils');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
 
 module.exports = async (ctx) => {
     const { id } = ctx.params;
-    const link = `https://www.huxiu.com/tags/${id}.html`;
-    const response = await got({
-        method: 'get',
-        url: link,
-        headers: {
-            Referer: link,
+    const link = `${utils.baseUrl}/tags/${id}.html`;
+    const { data } = await got(link, {
+        https: {
+            rejectUnauthorized: false,
         },
     });
 
-    const data = response.data;
     const $ = cheerio.load(data);
-    const list = $('.related-article li a')
-        .get()
-        .map((e) => $(e).attr('href'));
+    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 utils.ProcessFeed(list, ctx.cache);
 

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

@@ -0,0 +1,22 @@
+{{ if brief.preface }}{{@ brief.preface }}<br>{{ /if }}
+
+{{ if brief.audio_info }}
+<figure>
+    <figcaption>{{ brief.audio_title }}</figcaption>
+    <audio controls src="{{ brief.audio_info.audio_path }}"></audio>
+</figure>
+<br>
+{{ /if }}
+
+{{ if brief.part_list }}{{ each brief.part_list part }}
+    {{ if part.title }}<h2>{{ part.title }}</h2>{{ /if }}
+    {{ if part.content_info }}{{ each part.content_info content }}
+        {{ if content.type === 'text' }}{{@ content.text }}{{ /if }}
+        {{ if content.type === 'video' }}
+            <video controls poster="{{ content.video.poster }}"><source src="{{ content.video.url }}" type="video/mp4"></video>
+        {{ /if }}
+        {{ if content.type === 'image'}}{{ /if }}
+    {{ /each }}<br>{{ /if }}
+{{ /each }}{{ /if }}
+
+{{ if brief.peroration }}{{ brief.peroration }}{{ /if }}

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

@@ -0,0 +1,3 @@
+{{ if img }}
+<img src="{{ img }}">
+{{ /if }}

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

@@ -0,0 +1,7 @@
+{{ if video_info }}
+<video controls poster="{{ video_info.cover }}" width="{{ video_info.width }}" height="{{ video_info.height }}">
+{{ if video_info.fhd_link }}<source src="{{ video_info.fhd_link }}" type="video/mp4">{{ /if }}
+{{ if video_info.hd_link }}<source src="{{ video_info.hd_link }}" type="video/mp4">{{ /if }}
+{{ if video_info.sd_link }}<source src="{{ video_info.sd_link }}" type="video/mp4">{{ /if }}
+</video>
+{{ /if }}