rsshub 1.0.0-master.fd3a269 → 1.0.0-master.fd40700

package/lib/v2/eprice/rss.js

@@ -14,11 +14,6 @@ module.exports = async (ctx) => {
         e.link = e.link.replace(/^http:\/\//i, 'https://');
     });
 
-    const renderDesc = (desc) =>
-        art(path.join(__dirname, 'templates/description.art'), {
-            desc,
-        });
-
     const items = await Promise.all(
         feed.items.map((item) =>
             ctx.cache.tryGet(item.link, async () => {
@@ -34,28 +29,41 @@ module.exports = async (ctx) => {
 
                 // remove unwanted elements
                 $('noscript').remove();
-                $('div.ad-336x280-g').remove();
-                $('div.ad-728x90-g').remove();
-                $('.adsbygoogle').remove();
-                $('.join-eprice-fb').remove();
-                $('div.clear').remove();
-                $('div.text-left').remove();
-                $('div.signature').remove();
                 $('div[id^=dablewidget]').remove();
+                $('.adsbygoogle, .join-eprice-fb, .teads').remove();
+                $('div.ad-336x280-g, div.ad-728x90-g').remove();
+                $('div.clear, div.news-vote, div.text-left, div.signature').remove();
+                $('ul.inner, ul.navigator, ul.infobar').remove();
+                $('iframe[src^="https://www.facebook.com/plugins/like.php"]').remove();
 
                 // extract categories
                 item.category = item.categories;
 
                 // fix lazyload image
+                $('a').each((_, e) => {
+                    e = $(e);
+                    if (e.attr('href') && e.attr('href').endsWith('.jpg')) {
+                        e.after(
+                            art(path.join(__dirname, 'templates/image.art'), {
+                                alt: e.attr('title'),
+                                src: e.attr('href'),
+                                title: e.attr('title'),
+                            })
+                        );
+                        e.remove();
+                    }
+                });
                 $('img').each((_, e) => {
                     e = $(e);
                     if (e.attr('data-original')) {
-                        e.attr('src', e.attr('data-original'));
-                        e.removeAttr('class');
-                        e.removeAttr('data-original');
-                        e.removeAttr('data-title');
-                        e.removeAttr('data-thumbnail');
-                        e.removeAttr('data-url');
+                        e.after(
+                            art(path.join(__dirname, 'templates/image.art'), {
+                                alt: e.attr('title'),
+                                src: e.attr('data-original'),
+                                title: e.attr('title'),
+                            })
+                        );
+                        e.remove();
                     }
                 });
 
@@ -67,7 +75,9 @@ module.exports = async (ctx) => {
                 delete item.enclosure;
                 delete item.isoDate;
 
-                item.description = renderDesc($('div.user-comment-block').html() ?? $('div.content').html());
+                item.description = art(path.join(__dirname, 'templates/description.art'), {
+                    desc: $('div.user-comment-block').html() ?? $('div.content').html() ?? $('li.inner').html(),
+                });
                 item.pubDate = parseDate(item.pubDate);
 
                 return item;

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

@@ -1 +1 @@
-{{ desc }}
+{{@ desc }}

package/lib/v2/eprice/templates/image.art

@@ -0,0 +1 @@
+<img alt="{{ alt }}" src="{{ src }}" title="{{ alt }}">

package/lib/v2/eventernote/maintainer.js

@@ -0,0 +1,3 @@
+module.exports = {
+    '/actors/:name/:id': ['KTachibanaM'],
+};

package/lib/v2/eventernote/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'eventernote.com': {
+        _name: 'Eventernote',
+        www: [
+            {
+                title: '声优活动及演唱会',
+                docs: 'https://docs.rsshub.app/anime.html#eventernote',
+                source: '/actors/:name/:id/events',
+                target: '/eventernote/actors/:name/:id',
+            },
+        ],
+    },
+};

package/lib/v2/eventernote/router.js

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

package/lib/{routes → v2}/ezone/index.js

@@ -1,11 +1,13 @@
 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 || '';
+    const category = ctx.params.category;
 
     const rootUrl = 'https://ezone.ulifestyle.com.hk';
-    const currentUrl = `${rootUrl}/${category || '/getLastestPageRight/latestNews'}`;
+    const currentUrl = `${rootUrl}/${category ?? '/getLastestPageRight/latestNews'}`;
 
     const response = await got({
         method: 'get',
@@ -17,7 +19,6 @@ module.exports = async (ctx) => {
 
     if (category) {
         list = $('.item-grid')
-            .slice(0, 10)
             .map((_, item) => {
                 item = $(item);
                 return {
@@ -29,7 +30,7 @@ module.exports = async (ctx) => {
         list = response.data[0][1].items.map((item) => ({
             title: item.headlines.items[0].name,
             link: `${rootUrl}/article/${item.standardDocumentId}`,
-            pubDate: new Date(item.publishDateStr).toUTCString(),
+            pubDate: timezone(parseDate(item.publishDateStr), +8),
         }));
     }
 
@@ -43,9 +44,16 @@ module.exports = async (ctx) => {
                 const content = cheerio.load(detailResponse.data);
                 const pubDate = detailResponse.data.match(/<span>(\d{2}-\d{2}-\d{4} \d{2}:\d{2})<\/span>/)[1];
 
+                content('a[data-href]').each(function () {
+                    content(this).after(`<img src="${content(this).attr('data-href')}">`);
+                    content(this).remove();
+                });
+
+                content('.gallery-remarks, .overlay-text-v2, .ulvideo').remove();
+
                 item.title = content('h2').eq(0).text();
                 item.description = content('.content').html();
-                item.pubDate = item.pubDate || new Date(`${pubDate.substr(6, 4)}-${pubDate.substr(0, 5)} ${pubDate.substr(11, 5)}`).toUTCString();
+                item.pubDate = item.pubDate ?? timezone(parseDate(`${pubDate.substr(6, 4)}-${pubDate.substr(0, 5)} ${pubDate.substr(11, 5)}`), +8);
 
                 return item;
             })

package/lib/v2/ezone/maintainer.js

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

package/lib/v2/ezone/radar.js

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

package/lib/v2/ezone/router.js

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

package/lib/v2/filmdeepfocus/index.js

@@ -0,0 +1,58 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+
+module.exports = async (ctx) => {
+    const category = ctx.params.category ?? 'page';
+
+    const rootUrl = 'https://www.filmdeepfocus.com';
+    const currentUrl = `${rootUrl}/${category}`;
+
+    const response = await got({
+        method: 'get',
+        url: currentUrl,
+    });
+
+    const $ = cheerio.load(response.data);
+
+    let items = $('.summary-title-link')
+        .slice(0, ctx.query.limit ? parseInt(ctx.query.limit) : 20)
+        .toArray()
+        .map((item) => {
+            item = $(item);
+
+            return {
+                title: item.text(),
+                link: `${rootUrl}${item.attr('href')}`,
+            };
+        });
+
+    items = await Promise.all(
+        items.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const detailResponse = await got({
+                    method: 'get',
+                    url: item.link,
+                });
+
+                const content = cheerio.load(detailResponse.data);
+
+                content('.image-block').each(function () {
+                    content(this).html(`<img src="${content(this).find('img').attr('data-src')}">`);
+                });
+
+                item.description = content('.entry-content').html();
+                item.author = content('meta[itemprop="publisher"]').attr('content');
+                item.pubDate = parseDate(content('meta[itemprop="datePublished"]').attr('content'));
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: $('title').text(),
+        link: currentUrl,
+        item: items,
+    };
+};

package/lib/v2/filmdeepfocus/maintainer.js

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

package/lib/v2/filmdeepfocus/radar.js

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

package/lib/v2/filmdeepfocus/router.js

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

package/lib/v2/firefox/index.js

@@ -0,0 +1,69 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+const { art } = require('@/utils/render');
+const path = require('path');
+
+module.exports = async (ctx) => {
+    const baseUrl = 'https://monitor.firefox.com';
+
+    const response = await got({
+        url: baseUrl + '/breaches',
+        header: {
+            Referer: baseUrl,
+        },
+    });
+
+    const $ = cheerio.load(response.body);
+
+    const list = JSON.parse($('div#breach-array-json').attr('data-breach-array')).breaches.map((item) => ({
+        title: item.Title,
+        link: baseUrl + '/breach-details/' + item.Name,
+        pubDate: timezone(parseDate(item.AddedDate), 0),
+        category: item.DataClasses.split(', '),
+    }));
+
+    const items = await Promise.all(
+        list &&
+            list.map((item) =>
+                ctx.cache.tryGet(
+                    item.link,
+                    async () => {
+                        const article = await got({
+                            url: item.link,
+                            header: {
+                                Referer: baseUrl + '/breaches',
+                            },
+                        });
+                        const content = cheerio.load(article.body);
+
+                        item.description = art(path.join(__dirname, 'templates/description.art'), {
+                            header: content('#breach-detail').html(),
+                            overview: content('section.detail-section.jst-cntr.flx.flx-col').html(),
+                            dataClasses: content('section.detail-section.bg-white.jst-cntr.flx.flx-col').html(),
+                        });
+
+                        return item;
+                    },
+                    31536000
+                )
+            )
+    );
+
+    ctx.state.data = {
+        title: $('title').text(),
+        link: baseUrl,
+        description: String($('head meta[name=description]').attr('content').trim()),
+        item: items,
+        image: $('head meta[property=og:image]').attr('content'),
+    };
+
+    ctx.state.json = {
+        title: $('title').text(),
+        link: baseUrl,
+        description: String($('head meta[name=description]').attr('content').trim()),
+        item: items,
+        image: $('head meta[property=og:image]').attr('content'),
+    };
+};

package/lib/v2/firefox/maintainer.js

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

package/lib/v2/firefox/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'firefox.com': {
+        _name: 'Mozilla',
+        monitor: [
+            {
+                title: 'Firefox Monitor',
+                docs: 'https://docs.rsshub.app/other.html#mozilla',
+                source: ['/', '/breaches'],
+                target: '/firefox/breaches',
+            },
+        ],
+    },
+};

package/lib/v2/firefox/router.js

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

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

@@ -0,0 +1,5 @@
+{{@ header }}
+<br>
+{{@ overview }}
+<br>
+{{@ dataClasses }}

package/lib/v2/fortunechina/index.js

@@ -0,0 +1,70 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate, parseRelativeDate } = require('@/utils/parse-date');
+
+module.exports = async (ctx) => {
+    const category = ctx.params.category ?? '';
+
+    const rootUrl = 'https://www.fortunechina.com';
+    const currentUrl = `${rootUrl}${category ? `/${category}` : ''}`;
+
+    const response = await got({
+        method: 'get',
+        url: currentUrl,
+    });
+
+    const $ = cheerio.load(response.data);
+
+    let items = $('.main')
+        .find('h3 a')
+        .slice(0, ctx.query.limit ? parseInt(ctx.query.limit) : 15)
+        .toArray()
+        .map((item) => {
+            item = $(item);
+
+            const link = item.attr('href');
+
+            return {
+                title: item.text(),
+                link: link.indexOf('http') === 0 ? link : `${currentUrl}/${item.attr('href')}`,
+            };
+        });
+
+    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);
+
+                const spans = content('.mod-info span').text();
+                let matches = spans.match(/(\d{4}年\d{2}月\d{2}日)/);
+                if (matches) {
+                    item.pubDate = parseDate(matches[1].replace(/年|月/g, '-').replace(/日/, ''));
+                } else {
+                    matches = spans.match(/(\d+小时前)/);
+                    if (matches) {
+                        item.pubDate = parseRelativeDate(matches[1]);
+                    }
+                }
+
+                item.author = content('.name').text();
+
+                content('.mod-info, .title').remove();
+
+                item.description = content('#articleContent, .eval-desc').html();
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: category ? $('title').text() : '财富中文网',
+        link: currentUrl,
+        item: items,
+    };
+};

package/lib/v2/fortunechina/maintainer.js

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

package/lib/v2/fortunechina/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'fortunechina.com': {
+        _name: '财富中文网',
+        '.': [
+            {
+                title: '分类',
+                docs: 'https://docs.rsshub.app/new-media.html#cai-fu-zhong-wen-wang-fen-lei',
+                source: ['/:category', '/'],
+                target: '/fortunechina/:category?',
+            },
+        ],
+    },
+};

package/lib/v2/fortunechina/router.js

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

package/lib/{routes → v2}/ft/channel.js

@@ -4,5 +4,6 @@ module.exports = async (ctx) => {
     ctx.state.data = await utils.getData({
         site: ctx.params.language === 'chinese' ? 'www' : 'big5',
         channel: ctx.params.channel,
+        ctx,
     });
 };

package/lib/v2/ft/maintainer.js

@@ -0,0 +1,4 @@
+module.exports = {
+    '/ft/myft/:key': ['HenryQW'],
+    '/ft/:language/:channel?': ['HenryQW', 'xyqfer'],
+};

package/lib/v2/ft/myft.js

@@ -0,0 +1,44 @@
+const got = require('@/utils/got');
+const parser = require('@/utils/rss-parser');
+const cheerio = require('cheerio');
+
+module.exports = async (ctx) => {
+    const ProcessFeed = (content) => {
+        // clean up the article
+        content.find('div.o-share, aside, div.o-ads').remove();
+
+        return content.html();
+    };
+
+    const link = `https://www.ft.com/myft/following/${ctx.params.key}.rss`;
+
+    const feed = await parser.parseURL(link);
+
+    const items = await Promise.all(
+        feed.items.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const response = await got({
+                    method: 'get',
+                    url: item.link,
+                    headers: {
+                        Referer: 'https://www.facebook.com',
+                    },
+                });
+
+                const $ = cheerio.load(response.data);
+
+                item.description = ProcessFeed($('div.article__content-body'));
+                item.author = $('a.n-content-tag--author').text();
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: `FT.com - myFT`,
+        link,
+        description: `FT.com - myFT`,
+        item: items,
+    };
+};

package/lib/v2/ft/radar.js

@@ -0,0 +1,24 @@
+module.exports = {
+    'ftchinese.com': {
+        _name: 'Financial Times',
+        '.': [
+            {
+                title: 'FT 中文网',
+                docs: 'https://docs.rsshub.app/traditional-media.html#financial-times',
+            },
+            {
+                title: 'myFT 个人 RSS',
+                docs: 'https://docs.rsshub.app/traditional-media.html#financial-times',
+            },
+        ],
+    },
+    'ft.com': {
+        _name: 'Financial Times',
+        '.': [
+            {
+                title: 'myFT personal RSS',
+                docs: 'https://docs.rsshub.app/en/traditional-media.html#financial-times',
+            },
+        ],
+    },
+};

package/lib/v2/ft/router.js

@@ -0,0 +1,4 @@
+module.exports = function (router) {
+    router.get('/myft/:key', require('./myft'));
+    router.get('/:language/:channel?', require('./channel'));
+};

package/lib/{routes → v2}/ft/utils.js

@@ -7,21 +7,21 @@ const ProcessFeed = ($, link) => {
     let content = $('div.story-container');
 
     // 处理封面图片
-    content.find('div.story-image > figure').each((i, e) => {
+    content.find('div.story-image > figure').each((_, e) => {
         const src = `https://thumbor.ftacademy.cn/unsafe/1340x754/${e.attribs['data-url']}`;
 
         $(`<img src=${src}>`).insertAfter(content.find('div.story-lead')[0]);
     });
 
     // 付费文章跳转
-    content.find('div#subscribe-now-container').each((i, e) => {
+    content.find('div#subscribe-now-container').each((_, e) => {
         $(`<br/><p>此文章为付费文章，会员<a href='${link}'>请访问网站阅读</a>。</p>`).insertAfter(content.find('div.story-body')[0]);
         $(e).remove();
     });
 
     // 获取作者
     let author = '';
-    content.find('span.story-author > a').each((i, e) => {
+    content.find('span.story-author > a').each((_, e) => {
         author += `${$(e).text()} `;
     });
     author = author.trim();
@@ -31,7 +31,7 @@ const ProcessFeed = ($, link) => {
         .find(
             'div.story-theme, h1.story-headline, div.story-byline, div.mpu-container-instory,script, div#story-action-placeholder, div.copyrightstatement-container, div.clearfloat, div.o-ads, h2.list-title, div.allcomments, div.logincomment, div.nologincomment'
         )
-        .each((i, e) => {
+        .each((_, e) => {
             $(e).remove();
         });
     content = content.html();
@@ -39,7 +39,7 @@ const ProcessFeed = ($, link) => {
     return { content, author, title };
 };
 
-const getData = async ({ site = 'www', channel }) => {
+const getData = async ({ site = 'www', channel, ctx }) => {
     let feed;
 
     if (channel) {
@@ -47,7 +47,7 @@ const getData = async ({ site = 'www', channel }) => {
         channel = channel.split('-').join('/');
 
         try {
-            feed = await parser.parseURL(`http://${site}.ftchinese.com/rss/${channel}`);
+            feed = await parser.parseURL(`https://${site}.ftchinese.com/rss/${channel}`);
         } catch (error) {
             return {
                 title: `FT 中文网 ${channel} 不存在`,
@@ -55,23 +55,23 @@ const getData = async ({ site = 'www', channel }) => {
             };
         }
     } else {
-        feed = await parser.parseURL(`http://${site}.ftchinese.com/rss/feed`);
+        feed = await parser.parseURL(`https://${site}.ftchinese.com/rss/feed`);
     }
 
     const items = await Promise.all(
-        feed.items.splice(0, 10).map(async (item) => {
-            const response = await got.get(`${item.link}?full=y&archive`);
+        feed.items.splice(0, 10).map((item) => {
+            item.link = item.link.replace('http://', 'https://');
+            return ctx.cache.tryGet(item.link, async () => {
+                const response = await got.get(`${item.link}?full=y&archive`);
 
-            const $ = cheerio.load(response.data);
-            const result = ProcessFeed($, item.link);
-            const single = {
-                title: result.title,
-                description: result.content,
-                author: result.author,
-                pubDate: item.pubDate,
-                link: item.link,
-            };
-            return Promise.resolve(single);
+                const $ = cheerio.load(response.data);
+                const result = ProcessFeed($, item.link);
+
+                item.title = result.title;
+                item.description = result.content;
+                item.author = result.author;
+                return item;
+            });
         })
     );