rsshub 1.0.0-master.f7f8b7a → 1.0.0-master.f812e91

package/lib/routes/facebook/page.js

@@ -1,128 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-
-const fetchPageHtml = (linkPath, cacheKey, cache) => {
-    const url = `https://mbasic.facebook.com${linkPath}`;
-
-    return cache.tryGet(cacheKey, async () => {
-        const { data: html } = await got.get(url);
-        return html;
-    });
-};
-
-const parseStoryPage = async (linkPath, cache) => {
-    const { searchParams: q } = new URL('https://mbasic.facebook.com' + linkPath);
-    const storyFbId = q.get('story_fbid');
-    const storyId = q.get('id');
-    const cacheKey = `story/${storyFbId}/${storyId}`;
-
-    const html = await fetchPageHtml(linkPath, cacheKey, cache);
-    const $ = cheerio.load(html);
-
-    const url = `https://www.facebook.com/story.php?story_fbid=${storyFbId}&id=${storyId}`;
-    const $story = $('#m_story_permalink_view').eq(0);
-    const $box = $story.find('div > div > div > div').eq(0);
-    const $header = $box.find('header').eq(0);
-    const $content = $box.find('div > div').eq(0);
-    $content.find('a[href^="https://lm.facebook.com/l.php"]').each((_, ele) => {
-        const link = $(ele);
-        const originalLink = new URL(link.attr('href')).searchParams.get('u');
-        if (originalLink) {
-            link.attr('href', decodeURIComponent(originalLink));
-        }
-    });
-    const $attach = $story.find('div > div > div > div:nth-child(3)').eq(0);
-
-    const attachLinkList = $attach
-        .find('a')
-        .toArray()
-        .map((a) => $(a).attr('href'));
-    const isAttachAreImageSet = attachLinkList.filter((link) => new RegExp('/photos/').test(link)).length === attachLinkList.length;
-    const title = $header.find('h3').text();
-
-    const content = $content.html();
-
-    let images = [];
-    if (isAttachAreImageSet) {
-        images = await Promise.all(attachLinkList.map((link) => parsePhotoPage(link, cache)));
-    }
-
-    return {
-        url,
-        title,
-        content,
-        images,
-    };
-};
-
-const parsePhotoPage = async (linkPath, cache) => {
-    const { pathname } = new URL('https://mbasic.facebook.com' + linkPath);
-    const cacheKey = `photos${pathname}`;
-
-    const html = await fetchPageHtml(linkPath, cacheKey, cache);
-    const $ = cheerio.load(html);
-
-    const title = $('#MPhotoContent div.msg > a > strong').first().text();
-    const url = `https://www.facebook.com${pathname}`;
-    const $content = $('#MPhotoContent div.msg > div');
-    const content = $content.html();
-    const image = $('#MPhotoContent div.desc.attachment > span > div > span > a[target=_blank].sec').attr('href');
-
-    return {
-        title,
-        url,
-        content,
-        image,
-    };
-};
-
-module.exports = async (ctx) => {
-    const { id } = ctx.params;
-    const pageId = encodeURIComponent(id);
-    const linkPath = `/${pageId}`;
-
-    const html = await fetchPageHtml(linkPath, pageId, ctx.cache);
-    const $ = cheerio.load(html);
-
-    const itemLinks = $('footer > div:nth-child(2) > a:nth-child(1)')
-        .toArray()
-        .map((a) => $(a).attr('href'));
-
-    const items = await Promise.all(
-        itemLinks.map(async (itemLink) => {
-            if (new RegExp(`^/.+/photos/`).test(itemLink)) {
-                const data = await parsePhotoPage(itemLink, ctx.cache);
-                return {
-                    title: data.title,
-                    link: data.url,
-                    description: `<img src="${data.image}"><br>${data.content}`,
-                };
-            }
-            if (new RegExp(`^/story.php`).test(itemLink)) {
-                const data = await parseStoryPage(itemLink, ctx.cache);
-                const isSingleImageStory = data.images.length === 1;
-                const isEmptyImageList = data.images.length === 0;
-
-                let desc = '';
-                desc += data.images.map((image) => `<img src="${image.image}"><br>${image.content}`).join('<br>');
-                if (!isSingleImageStory) {
-                    !isEmptyImageList && (desc += '<br>');
-                    desc += data.content;
-                }
-
-                return {
-                    title: data.title,
-                    link: data.url,
-                    description: desc,
-                };
-            }
-        })
-    );
-
-    ctx.state.data = {
-        title: $('#m-timeline-cover-section h1 span').text(),
-        link: `https://www.facebook.com/${pageId}`,
-        description: $('#sub_profile_pic_content>div>div:nth-child(3) div>span').find('br').replaceWith('\n').text(),
-        item: items.filter((item) => !!item),
-    };
-};

package/lib/routes/feixuew/index.js

@@ -1,63 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-
-module.exports = async (ctx) => {
-    let currentUrl;
-    let query;
-
-    const rootUrl = 'https://www.feixuew.com';
-
-    ctx.params.id = ctx.params.id || 'latest';
-
-    if (ctx.params.id === 'latest') {
-        currentUrl = rootUrl;
-    } else {
-        currentUrl = `${rootUrl}/sort/${ctx.params.id}`;
-    }
-
-    const response = await got({
-        method: 'get',
-        url: currentUrl,
-    });
-
-    const $ = cheerio.load(response.data);
-
-    if (ctx.params.id === 'latest') {
-        query = $('ul.ul-new li a').slice(4, 10);
-    } else {
-        query = $('a.sjwu').slice(0, 10);
-    }
-
-    const list = query
-        .map((_, item) => {
-            item = $(item);
-            return {
-                title: item.attr('title'),
-                link: item.attr('href'),
-            };
-        })
-        .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);
-
-                item.description = content('div.info-con').html();
-                item.pubDate = new Date(content('p.f-sgray span').eq(1).text().replace('发布时间：', '') + ' GMT+8').toUTCString();
-
-                return item;
-            })
-        )
-    );
-
-    ctx.state.data = {
-        title: `飞雪娱乐网 - ${$('title').text().split('-')[0]}`,
-        link: currentUrl,
-        item: items,
-    };
-};

package/lib/routes/fjnews/fznews.js

@@ -1,55 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-const url = require('url');
-
-let host = 'http://news.fznews.com.cn/fzxw/';
-let citytitle = '福州新闻';
-
-module.exports = async (ctx) => {
-    if (ctx.params.city === 'fj') {
-        host = 'http://news.fznews.com.cn/dsxw/';
-        citytitle = '福建新闻';
-    } else {
-        host = 'http://news.fznews.com.cn/fzxw/';
-        citytitle = '福州新闻';
-    }
-
-    const response = await got.get(host);
-    const $ = cheerio.load(response.data);
-    const list = $('.li-h32f14 li:not(.clearfix)').slice(0, ctx.params.limit).get();
-    const out = await Promise.all(
-        list.map(async (item) => {
-            const $ = cheerio.load(item);
-            const title = $('a').text();
-            const itemUrl = url.resolve(host, $('a').attr('href'));
-
-            const cache = await ctx.cache.get(itemUrl);
-            if (cache) {
-                return Promise.resolve(JSON.parse(cache));
-            }
-
-            const res = await got.get(itemUrl);
-            const capture = cheerio.load(res.data);
-            const contents = capture('.zhengwen').html();
-            const time = capture('div.laiyuan div.left').text().substring(0, 19);
-            const author = capture('.laiyuan').text().replace('分享到：', '');
-
-            const single = {
-                title,
-                link: itemUrl,
-                guid: itemUrl,
-                author,
-                description: contents,
-                pubDate: new Date(time).toUTCString(),
-            };
-
-            return Promise.resolve(single);
-        })
-    );
-
-    ctx.state.data = {
-        title: citytitle,
-        link: host,
-        item: out,
-    };
-};

package/lib/routes/fjnews/jjnews.js

@@ -1,43 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-const url = require('url');
-const host = 'http://tt.m.jxnews.com.cn/jjtop/';
-
-module.exports = async (ctx) => {
-    const response = await got.get(host);
-    const $ = cheerio.load(response.data);
-    const list = $('li.col-xs-12').get();
-    const out = await Promise.all(
-        list.map(async (item) => {
-            const $ = cheerio.load(item);
-            const title = $('.list-text-style').text();
-            const itemUrl = url.resolve(host, $('a').attr('href'));
-
-            const cache = await ctx.cache.get(itemUrl);
-            if (cache) {
-                return Promise.resolve(JSON.parse(cache));
-            }
-
-            const res = await got.get(itemUrl);
-            const capture = cheerio.load(res.data);
-            const contents = capture('content').html();
-            const time = capture('div.col-xs-8').text().substring(0, 19);
-
-            const single = {
-                title,
-                link: itemUrl,
-                guid: itemUrl,
-                description: contents,
-                pubDate: new Date(time).toUTCString(),
-            };
-
-            return Promise.resolve(single);
-        })
-    );
-
-    ctx.state.data = {
-        title: '九江新闻',
-        link: host,
-        item: out,
-    };
-};

package/lib/routes/fulinian/index.js

@@ -1,50 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-
-const rootUrl = 'https://www.fulinian.com/';
-
-module.exports = async (ctx) => {
-    ctx.params.caty = ctx.params.caty || '';
-
-    const currentUrl = `${rootUrl}/${ctx.params.caty}`;
-    const response = await got({
-        method: 'get',
-        url: currentUrl,
-    });
-
-    const $ = cheerio.load(response.data);
-    const list = $('article.excerpt')
-        .slice(0, 10)
-        .map((_, item) => {
-            item = $(item);
-            const a = item.find('header h2 a');
-            return {
-                title: a.text(),
-                link: a.attr('href'),
-            };
-        })
-        .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);
-
-                item.description = content('article.article-content').html();
-                item.pubDate = new Date(content('div.article-meta span.item').eq(0).text() + ' GMT+8').toUTCString();
-
-                return item;
-            })
-        )
-    );
-
-    ctx.state.data = {
-        title: `福利年 - ${$('h1').text()}`,
-        link: currentUrl,
-        item: items,
-    };
-};

package/lib/routes/gab/explore.js

@@ -1,31 +0,0 @@
-const cheerio = require('cheerio');
-const got = require('@/utils/got');
-
-module.exports = async (ctx) => {
-    const sort = ctx.params.sort || 'hot';
-
-    let baseURL = 'https://gab.com/api/v1/timelines/explore?page=1&sort_by=';
-    baseURL += sort === 'hot' ? 'hot' : 'top_today';
-
-    const response = await got.get(baseURL);
-
-    const result = await Promise.all(
-        response.data.map((item) => {
-            const media = item.media_attachments.map((item) => `<img src='${item.url}'>`).join('<br>');
-            const $ = cheerio.load(item.content);
-
-            return {
-                title: $.text() || 'Untitled',
-                description: item.content + (media && '<br>' + media),
-                link: item.url || baseURL,
-                pubDate: new Date(item.created_at).toISOString(),
-            };
-        })
-    );
-
-    ctx.state.data = {
-        title: `Gab - Popular posts`,
-        link: baseURL,
-        item: result,
-    };
-};

package/lib/routes/gamegrape/index.js

@@ -1,73 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-
-function getNum(str) {
-    const result = str.match(/(\d{1,2}).*/);
-    if (result) {
-        return parseInt(result[1]);
-    }
-}
-function resolveRelativeTime(relativeTime) {
-    const result = /\S* · (\d{1,2}天)?(\d{1,2}小时)?(\d{1,2}分钟)?/.exec(relativeTime);
-    let day, hour, min;
-    if (result[1]) {
-        day = getNum(result[1]);
-    }
-    if (result[2]) {
-        hour = getNum(result[2]);
-    }
-    if (result[3]) {
-        min = getNum(result[3]);
-    }
-    return (((day || 0) * 24 + (hour || 0)) * 60 + (min || 0)) * 60 * 1000;
-}
-
-module.exports = async (ctx) => {
-    const id = ctx.params.id || '';
-
-    const rootUrl = 'http://youxiputao.com';
-    const currentUrl = `${rootUrl}/article${id ? `/index/id/${id}` : ''}`;
-    const response = await got({
-        method: 'get',
-        url: currentUrl,
-    });
-
-    const $ = cheerio.load(response.data);
-
-    const list = $('.news-info-box')
-        .slice(0, 15)
-        .map((_, item) => {
-            item = $(item);
-            const a = item.find('a[title]');
-
-            return {
-                title: a.text(),
-                link: `${rootUrl}${a.attr('href')}`,
-                pubDate: Date.now() - resolveRelativeTime(item.find('.pull-right').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);
-
-                item.description = content('.info').html();
-                item.author = content('.users-info h4').text();
-
-                return item;
-            })
-        )
-    );
-
-    ctx.state.data = {
-        title: $('title').text(),
-        link: currentUrl,
-        item: items,
-    };
-};

package/lib/routes/gvm/index.js

@@ -1,94 +0,0 @@
-const got = require('got');
-const cheerio = require('cheerio');
-
-module.exports = async (ctx) => {
-    const category = ctx.params.category || 'newest';
-    let suffix = '/newest';
-
-    const options = {
-        newest: '最新文章',
-        recommend: '你可能會喜歡',
-        opinion: '名家專欄',
-        topic: '專題',
-        news: '時事熱點',
-        politics: '政治',
-        society: '社會',
-        figure: '人物報導',
-        world: '國際',
-        world_focus: '全球焦點',
-        cross_strait_politics: '兩岸',
-        money: '金融理財',
-        investment: '投資理財',
-        insurance: '保險規劃',
-        retire: '退休理財',
-        fintech: '金融Fintech',
-        real_estate: '房地產',
-        economy: '總體經濟',
-        tech: '科技',
-        tech_trend: '科技趨勢',
-        energy: '能源',
-        business: '產經',
-        industry: '傳產',
-        service: '消費服務',
-        medical: '生技醫藥',
-        family_business_succession: '傳承轉型',
-        startup: '創業新創',
-        management: '管理',
-        agriculture: '農業',
-        education: '教育',
-        higher_education: '高教',
-        technological: '技職',
-        parent: '親子教育',
-        world_education: '國際文教',
-        sports: '體育',
-        life: '好享生活',
-        art: '時尚設計',
-        self_growth: '心靈成長',
-        film: '藝文影視',
-        travel: '旅遊',
-        environment: '環境生態',
-        health: '健康',
-        food: '美食',
-        career: '職場生涯',
-        survey: '調查',
-        county: '縣市',
-        csr: 'CSR',
-    };
-
-    if (category !== 'newest' && category !== 'recommend') {
-        suffix = `/category/${category}`;
-    }
-
-    const response = await got({
-        method: 'get',
-        url: 'https://www.gvm.com.tw' + suffix,
-    });
-
-    const $ = cheerio.load(response.body);
-
-    const articles = $('#article_list .article-list-item .article-list-item__intro')
-        .map((index, ele) => ({
-            title: $('a', ele).text(),
-            link: $('a', ele).attr('href'),
-            pubDate: new Date($('.time', ele).text()),
-            author: $('.author', ele).text(),
-        }))
-        .get();
-
-    const item = await Promise.all(
-        articles.map((item) =>
-            ctx.cache.tryGet(item.link, async () => {
-                const res = await got.get(item.link);
-                const content = cheerio.load(res.body);
-                item.description = content('.article-content').html();
-                return item;
-            })
-        )
-    );
-
-    ctx.state.data = {
-        title: `遠見 - ${options[category]}`,
-        link: 'https://www.gvm.com.tw' + suffix,
-        item,
-    };
-};

package/lib/routes/hkcd/pdf.js

@@ -1,30 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-const iconv = require('iconv-lite');
-
-module.exports = async (ctx) => {
-    const rootUrl = 'http://hk.hkcd.com';
-    const currentUrl = `${rootUrl}/pdf/index.htm`;
-    const response = await got({
-        method: 'get',
-        url: currentUrl,
-        responseType: 'buffer',
-    });
-
-    const data = iconv.decode(response.data, 'gb2312');
-    const date = data.match(/<td align="right"><font color="#FFFFFF">(.*)/)[1];
-    const $ = cheerio.load(data.replace('<table width="177" border="0" cellspacing="1" cellpadding="0"', '<table id="content" width="177" border="0" cellspacing="1" cellpadding="0"'));
-
-    ctx.state.data = {
-        title: '香港商报PDF版',
-        link: currentUrl,
-        item: [
-            {
-                title: date,
-                link: currentUrl,
-                description: $('#content').html(),
-                pubDate: new Date(date.replace(/年|月/g, '-').replace(/日/, '')).toUTCString(),
-            },
-        ],
-    };
-};

package/lib/routes/hkcnews/news.js

@@ -1,63 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-
-const titles = {
-    '': '眾聞',
-    13: '經濟',
-    15: '社會',
-    14: '生活',
-    12: '政治',
-    16: '國際',
-    20: '台灣',
-    21: '人物',
-    19: '中國',
-};
-
-module.exports = async (ctx) => {
-    const category = ctx.params.category || '';
-
-    const rootUrl = 'https://www.hkcnews.com';
-    const apiUrl = `${rootUrl}/data/newsposts${category === '' ? '' : `/${category}`}?page=1`;
-    const response = await got({
-        method: 'get',
-        url: apiUrl,
-    });
-
-    const list = response.data.items.map((item) => {
-        const $ = cheerio.load(item);
-        const news = $('.article-block-body a');
-        const date = $('.line').eq(1).text().split('.').reverse().join('-');
-
-        return {
-            title: news.text(),
-            link: `${rootUrl}${news.attr('href')}`,
-            pubDate: new Date(date.length === 8 ? `${new Date().getFullYear().toString().slice(0, 2)}${date}` : date).toUTCString(),
-        };
-    });
-
-    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);
-
-                item.description = content('.article-content').html();
-                // substring 2 times
-                // from https://hkcnews.com/article/47878/loreal歐萊雅-薇婭-李佳琦-47884/薇婭、李佳琦幫loreal帶貨引價格爭議-loreal道歉
-                // to https://hkcnews.com/article/47878
-                item.guid = item.link.substring(0, item.link.lastIndexOf('/')).substring(0, item.link.substring(0, item.link.lastIndexOf('/')).lastIndexOf('/'));
-
-                return item;
-            })
-        )
-    );
-
-    ctx.state.data = {
-        title: `眾新聞 - ${titles[category]}`,
-        link: `${rootUrl}/news${category === '' ? '' : `/${category}/${titles[category]}`}`,
-        item: items,
-    };
-};