rsshub 1.0.0-master.fc91df6 → 1.0.0-master.fd40700

package/lib/v2/mygopen/index.js

@@ -0,0 +1,29 @@
+const got = require('@/utils/got');
+const { parseDate } = require('@/utils/parse-date');
+
+module.exports = async (ctx) => {
+    const label = ctx.params.label ?? '';
+
+    const rootUrl = 'https://www.mygopen.com';
+    const currentUrl = `${rootUrl}${label ? `/search/label/${label}` : ''}`;
+    const apiUrl = `${rootUrl}/feeds/posts/default${label ? `/-/${label}` : ''}?alt=json-in-script&max-results=${ctx.query.limit ? parseInt(ctx.query.limit) : 50}`;
+
+    const response = await got({
+        method: 'get',
+        url: apiUrl,
+    });
+
+    const items = JSON.parse(response.data.match(/gdata\.io\.handleScriptLoaded\((.*)\);/)[1]).feed.entry.map((item) => ({
+        title: item.title.$t,
+        description: item.content.$t,
+        pubDate: parseDate(item.published.$t),
+        link: item.link.pop().href,
+    }));
+
+    ctx.state.data = {
+        title: `MyGoPen${label ? `: ${label}` : ''}`,
+        link: currentUrl,
+        item: items,
+        description: '詐騙與謠言頻傳的年代，「MyGoPen｜這是假消息」提醒網路使用者隨時要用謹慎懷疑的態度面對網路上的消息。',
+    };
+};

package/lib/v2/mygopen/maintainer.js

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

package/lib/v2/mygopen/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'mygopen.com': {
+        _name: 'MyGoPen',
+        '.': [
+            {
+                title: '分類',
+                docs: 'https://docs.rsshub.app/new-media.html#mygopen-fen-lei',
+                source: ['/search/label/:label', '/'],
+                target: '/mygopen/:label?',
+            },
+        ],
+    },
+};

package/lib/v2/mygopen/router.js

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

package/lib/v2/neea/jlpt.js

@@ -0,0 +1,52 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+
+module.exports = async (ctx) => {
+    const rootUrl = 'https://news.neea.cn';
+    const currentUrl = `${rootUrl}/JLPT/1/newslist.htm`;
+
+    const response = await got({
+        method: 'get',
+        url: currentUrl,
+    });
+
+    const $ = cheerio.load(response.data);
+
+    let items = $('a')
+        .toArray()
+        .map((item) => {
+            item = $(item);
+
+            const matches = item.text().match(/(\d{4}-\d{2}-\d{2})/);
+
+            return {
+                title: item.text(),
+                link: `${rootUrl}/JLPT/1/${item.attr('href')}`,
+                pubDate: matches ? parseDate(matches[1]) : '',
+            };
+        });
+
+    items = await Promise.all(
+        items.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('.dvContent').html();
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: '重要通知 - 教育部考试中心日本语能力测试',
+        link: currentUrl,
+        item: items,
+    };
+};

package/lib/v2/neea/maintainer.js

@@ -0,0 +1,4 @@
+module.exports = {
+    '/jlpt': ['nczitzk'],
+    '/:type?': ['SunShinenny'],
+};

package/lib/v2/neea/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'neea.cn': {
+        _name: '中国教育考试网',
+        jlpt: [
+            {
+                title: '教育部考试中心日本语能力测试重要通知',
+                docs: 'https://docs.rsshub.app/study.html#jiao-yu-bu-kao-shi-zhong-xin-ri-ben-yu-neng-li-ce-shi-zhong-yao-tong-zhi',
+                source: ['/'],
+                target: '/neea/jlpt',
+            },
+        ],
+    },
+};

package/lib/v2/neea/router.js

@@ -0,0 +1,4 @@
+module.exports = function (router) {
+    router.get('/jlpt', require('./jlpt'));
+    router.get('/:type?', require('./index'));
+};

package/lib/v2/netease/maintainer.js

@@ -0,0 +1,5 @@
+module.exports = {
+    '/renjian/:category?': ['nczitzk'],
+    '/today/:need_content?': ['nczitzk'],
+    '/news/rank/:category?/:type?/:time?': ['nczitzk'],
+};

package/lib/v2/netease/radar.js

@@ -0,0 +1,29 @@
+module.exports = {
+    '163.com': {
+        _name: '网易',
+        renjian: [
+            {
+                title: '人间',
+                docs: 'https://docs.rsshub.app/new-media.html#wang-yi-xin-wen-ren-jian',
+                source: ['/:category', '/'],
+                target: '/netease/renjian/:category?',
+            },
+        ],
+        news: [
+            {
+                title: '排行榜',
+                docs: 'https://docs.rsshub.app/new-media.html#wang-yi-xin-wen-pai-hang-bang',
+                source: ['/:category', '/'],
+                target: '/netease/news/rank/:category?/:type?/:time?',
+            },
+        ],
+        m: [
+            {
+                title: '今日关注',
+                docs: 'https://docs.rsshub.app/new-media.html#wang-yi-xin-wen-jin-ri-guan-zhu',
+                source: ['/'],
+                target: '/netease/today/:need_content?',
+            },
+        ],
+    },
+};

package/lib/{routes/netease/news → v2/netease}/rank.js

@@ -1,9 +1,9 @@
-const url = require('url');
 const got = require('@/utils/got');
 const cheerio = require('cheerio');
 const iconv = require('iconv-lite');
+const { parseDate } = require('@/utils/parse-date');
 
-const rootUrl = 'http://news.163.com/';
+const rootUrl = 'http://news.163.com';
 
 const config = {
     whole: {
@@ -56,7 +56,7 @@ const config = {
     },
 };
 
-const time = {
+const timeRange = {
     hour: {
         index: 0,
         title: '1小时',
@@ -76,34 +76,32 @@ const time = {
 };
 
 module.exports = async (ctx) => {
-    ctx.params.category = ctx.params.category ? ctx.params.category : 'whole';
-    ctx.params.type = ctx.params.type ? ctx.params.type : 'click';
-    ctx.params.time = ctx.params.time ? ctx.params.time : 'day';
+    const category = ctx.params.category || 'whole';
+    const type = ctx.params.type || 'click';
+    const time = ctx.params.time || 'day';
 
-    const cfg = config[ctx.params.category];
+    const cfg = config[category];
     if (!cfg) {
         throw Error('Bad category. See <a href="https://docs.rsshub.app/new-media.html#wang-yi-xin-wen-pai-hang-bang">docs</a>');
-    } else if (
-        (ctx.params.category !== 'whole' && ctx.params.type === 'click' && ctx.params.time === 'month') ||
-        (ctx.params.category === 'whole' && ctx.params.type === 'click' && ctx.params.time === 'hour') ||
-        (ctx.params.type === 'follow' && ctx.params.time === 'hour')
-    ) {
-        throw Error('Bad time range. See <a href="https://docs.rsshub.app/new-media.html#wang-yi-xin-wen-pai-hang-bang">docs</a>');
+    } else if ((category !== 'whole' && type === 'click' && time === 'month') || (category === 'whole' && type === 'click' && time === 'hour') || (type === 'follow' && time === 'hour')) {
+        throw Error('Bad timeRange range. See <a href="https://docs.rsshub.app/new-media.html#wang-yi-xin-wen-pai-hang-bang">docs</a>');
     }
 
-    const currentUrl = ctx.params.category === 'money' ? cfg.link : url.resolve(rootUrl, cfg.link);
+    const currentUrl = category === 'money' ? cfg.link : `${rootUrl}${cfg.link}`;
     const response = await got({
         method: 'get',
         url: currentUrl,
         responseType: 'buffer',
     });
+
     const $ = cheerio.load(iconv.decode(response.data, 'gbk'));
+
     const list = $('div.tabContents')
-        .eq(time[ctx.params.time].index + (ctx.params.category === 'whole' ? (ctx.params.type === 'click' ? -1 : 2) : ctx.params.type === 'click' ? 0 : 2))
+        .eq(timeRange[time].index + (category === 'whole' ? (type === 'click' ? -1 : 2) : type === 'click' ? 0 : 2))
         .find('table tbody tr td a')
-        .slice(0, 15)
         .map((_, item) => {
             item = $(item);
+
             return {
                 link: item.attr('href'),
             };
@@ -116,6 +114,7 @@ module.exports = async (ctx) => {
                 try {
                     const category = item.link.split('.163.com')[0].split('//').pop().split('.').pop();
                     const link = `https://3g.163.com/${category}/article/${item.link.split('/').pop()}`;
+
                     const detailResponse = await got({
                         method: 'get',
                         url: link,
@@ -128,19 +127,27 @@ module.exports = async (ctx) => {
                     });
 
                     item.title = content('meta[property="og:title"]').attr('content').replace('_手机网易网', '');
-                    item.pubDate = content('meta[property="og:release_date"]').attr('content');
+                    item.pubDate = parseDate(content('meta[property="og:release_date"]').attr('content'));
                     item.description = content('.content').html();
-
-                    return item;
                 } catch (err) {
                     return Promise.resolve('');
                 }
+
+                return item;
             })
         )
     );
 
+    for (let i = 0; i < items.length; i++) {
+        if (items[i] === undefined) {
+            items.splice(i, 1);
+        } else if (items[i].title === undefined) {
+            items.splice(i, 1);
+        }
+    }
+
     ctx.state.data = {
-        title: `网易新闻${time[ctx.params.time].title}${ctx.params.type === 'click' ? '点击' : '跟帖'}榜 - ${cfg.title}`,
+        title: `网易新闻${timeRange[time].title}${type === 'click' ? '点击' : '跟帖'}榜 - ${cfg.title}`,
         link: currentUrl,
         item: items,
     };

package/lib/v2/netease/renjian.js

@@ -0,0 +1,74 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const iconv = require('iconv-lite');
+const { parseDate } = require('@/utils/parse-date');
+
+const titles = {
+    texie: '特写',
+    jishi: '记事',
+    daxie: '大写',
+    haodu: '好读',
+    kanke: '看客',
+};
+
+module.exports = async (ctx) => {
+    const category = ctx.params.category ?? 'texie';
+
+    const rootUrl = 'http://renjian.163.com';
+    const currentUrl = `${rootUrl}/special/renjian_${category}`;
+
+    const response = await got({
+        method: 'get',
+        url: currentUrl,
+        responseType: 'buffer',
+    });
+
+    const data = iconv.decode(response.data, 'gbk');
+
+    let items = {};
+
+    const urls = data.match(/url:"(.*)",/g);
+
+    if (urls) {
+        items = urls.slice(0, ctx.query.limit ? parseInt(ctx.query.limit) : 15).map((item) => ({
+            link: item.match(/url:"(.*)",/)[1],
+        }));
+    } else {
+        const $ = cheerio.load(data);
+
+        items = $('.article h3 a')
+            .slice(0, ctx.query.limit ? parseInt(ctx.query.limit) : 15)
+            .map((_, item) => {
+                item = $(item);
+                return {
+                    link: item.attr('href'),
+                };
+            })
+            .get();
+    }
+
+    items = await Promise.all(
+        items.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const detailResponse = await got({
+                    method: 'get',
+                    url: item.link,
+                });
+                const content = cheerio.load(detailResponse.data);
+
+                item.title = content('h1').text();
+                item.author = content('.author_txt name').text();
+                item.description = content('#endText').html() ?? content('#content').html();
+                item.pubDate = parseDate(content('.pub_time').text() ?? content('.post_info').text().split('来源:')[0].trim());
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: `人间 - ${titles[category]} - 网易新闻`,
+        link: currentUrl,
+        item: items,
+    };
+};

package/lib/v2/netease/router.js

@@ -0,0 +1,5 @@
+module.exports = function (router) {
+    router.get('/renjian/:category?', require('./renjian'));
+    router.get('/today/:need_content?', require('./today'));
+    router.get('/news/rank/:category?/:type?/:time?', require('./rank'));
+};

package/lib/v2/netease/today.js

@@ -0,0 +1,55 @@
+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 needContent = /t|y/i.test(ctx.params.need_content ?? 'true');
+
+    const rootUrl = 'https://gw.m.163.com';
+    const currentUrl = `${rootUrl}/nc/api/v1/feed/static/normal-list?start=0&tid=T1573700340788&size=${ctx.query.limit ?? (needContent ? 30 : 200)}`;
+
+    const response = await got({
+        method: 'get',
+        url: currentUrl,
+    });
+
+    let items = response.data.data.items.map((item) => ({
+        title: item.title,
+        author: item.source,
+        pubDate: timezone(parseDate(item.ptime), +8),
+        description: `<p>${item.digest}</p><img src="${item.imgsrc}">`,
+        link: item.url || `https://c.m.163.com/news/a/${item.docid}.html`,
+    }));
+
+    if (needContent) {
+        items = await Promise.all(
+            items.map((item) =>
+                ctx.cache.tryGet(item.link, async () => {
+                    const detailResponse = await got({
+                        method: 'get',
+                        url: item.link,
+                    });
+
+                    const content = cheerio.load(detailResponse.data);
+
+                    content('.bot_word').remove();
+
+                    content('figure[data-echo]').each(function () {
+                        content(this).html(`<img src="${content(this).attr('data-echo')}">`);
+                    });
+
+                    item.description = content('.content, article').html();
+
+                    return item;
+                })
+            )
+        );
+    }
+
+    ctx.state.data = {
+        title: '今日关注 - 网易新闻',
+        link: 'https://wp.m.163.com/163/html/newsapp/todayFocus/index.html',
+        item: items,
+    };
+};

package/lib/v2/news/maintainer.js

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

package/lib/v2/news/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'news.cn': {
+        _name: '新华网',
+        '.': [
+            {
+                title: '新华社新闻',
+                docs: 'https://docs.rsshub.app/new-media.html#xin-hua-wang-xin-hua-she-xin-wen',
+                source: ['/'],
+                target: '/news/whxw',
+            },
+        ],
+    },
+};

package/lib/v2/news/router.js

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

package/lib/{routes → v2}/news/whxw.js

@@ -13,39 +13,47 @@ module.exports = async (ctx) => {
 
     const $ = cheerio.load(response.data);
 
-    const list = $('h3 a')
-        .slice(0, 35)
-        .map((_, item) => {
+    let items = $('h3 a')
+        .toArray()
+        .map((item) => {
             item = $(item);
             return {
                 title: item.text(),
                 link: 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 () => {
                 let detailResponse;
+
                 try {
                     detailResponse = await got(item.link);
                 } catch (error) {
                     item.status = 404;
                 }
+
                 if (item.status !== 404) {
                     const content = cheerio.load(detailResponse.data);
                     const date = content('.header-time.left').text();
                     const author =
-                        content('.editor').text() ||
+                        content('.editor').text() ??
                         content('.p-jc')
                             .text()
-                            .replace(/[\r\n]/g, '');
-                    item.author = author.match(/责任编辑.(.*)/)[1].trim();
+                            .replace(/[\r\n]/g, '')
+                            .replace('： ', ':')
+                            .trim();
+
+                    item.author = author.split(':').pop().replace('】', '');
+
                     content('#articleEdit').remove();
+
                     item.description = content('#detail').html();
                     item.pubDate = timezone(parseDate(date, 'YYYYMM/DD HH:mm:ss'), +8);
                 }
+
+                delete item.status;
                 return item;
             })
         )

package/lib/v2/newsmarket/index.js

@@ -0,0 +1,63 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+
+module.exports = async (ctx) => {
+    const category = ctx.params.category ?? '';
+
+    const rootUrl = 'https://www.newsmarket.com.tw';
+    const currentUrl = `${rootUrl}${category ? `/blog/category/${category}` : ''}`;
+
+    const response = await got({
+        method: 'get',
+        url: currentUrl,
+    });
+
+    const $ = cheerio.load(response.data);
+
+    const list = $('.title a')
+        .slice(0, ctx.query.limit ? parseInt(ctx.query.limit) : 20)
+        .map((_, item) => {
+            item = $(item);
+
+            return {
+                title: item.text(),
+                link: item.attr('href'),
+                pubDate: parseDate(),
+            };
+        })
+        .get();
+
+    const items = await Promise.all(
+        list.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const detailResponse = await got({
+                    method: 'get',
+                    url: item.link,
+                });
+
+                const content = cheerio.load(detailResponse.data);
+
+                content('figure img').each(function () {
+                    content(this)
+                        .parent()
+                        .html(`<img src="${content(this).attr('data-src')}">`);
+                });
+
+                content('.inline-post').remove();
+
+                item.author = content('.author-name').text();
+                item.description = content('.entry-content').html();
+                item.pubDate = parseDate(detailResponse.data.match(/"datePublished":"(.*)","dateModified"/)[1]);
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: $('title').text(),
+        link: currentUrl,
+        item: items,
+    };
+};

package/lib/v2/newsmarket/maintainer.js

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

package/lib/v2/newsmarket/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'newsmarket.com.tw': {
+        _name: '上下游News&Market',
+        '.': [
+            {
+                title: '分類',
+                docs: 'https://docs.rsshub.app/new-media.html#shang-xia-you-news-market',
+                source: ['/blog/category/:category', '/'],
+                target: '/newsmarket/:category?',
+            },
+        ],
+    },
+};

package/lib/v2/newsmarket/router.js

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

package/lib/v2/ngocn2/index.js

@@ -0,0 +1,56 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+
+module.exports = async (ctx) => {
+    const category = ctx.params.category ?? 'article';
+
+    const rootUrl = 'https://ngocn2.org';
+    const currentUrl = `${rootUrl}/${category}`;
+
+    const response = await got({
+        method: 'get',
+        url: currentUrl,
+    });
+
+    const $ = cheerio.load(response.data);
+
+    const list = $('.articleroll__article a')
+        .map((_, item) => {
+            item = $(item);
+
+            return {
+                title: item.find('.title').text(),
+                link: `${rootUrl}${item.attr('href')}`,
+                pubDate: parseDate(item.find('.meta').text()),
+            };
+        })
+        .get();
+
+    const items = await Promise.all(
+        list.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const detailResponse = await got({
+                    method: 'get',
+                    url: item.link,
+                });
+
+                const content = cheerio.load(detailResponse.data);
+
+                content('.gatsby-resp-image-link').each(function () {
+                    content(this).html(`<img src="${content(this).find('img').attr('src')}">`);
+                });
+
+                item.description = content('.article__content').html();
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: `${$('.sectitle__content').text()} - NGOCN`,
+        link: currentUrl,
+        item: items,
+    };
+};

package/lib/v2/ngocn2/maintainer.js

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

package/lib/v2/ngocn2/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'ngocn2.org': {
+        _name: 'NGOCN',
+        '.': [
+            {
+                title: '分类',
+                docs: 'https://docs.rsshub.app/new-media.html#ngocn2-fen-lei',
+                source: ['/'],
+                target: '/ngocn2/:category?',
+            },
+        ],
+    },
+};

package/lib/v2/ngocn2/router.js

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