rsshub 1.0.0-master.ff1cc11 → 1.0.0-master.ff97759

package/lib/v2/ccreports/index.js

@@ -0,0 +1,41 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+
+const rootUrl = 'https://www.ccreports.com.cn';
+
+module.exports = async (ctx) => {
+    const listData = await got.get(rootUrl);
+    const $ = cheerio.load(listData.data);
+    const list = $('div.index-four-content > div.article-box')
+        .find('div.new-child')
+        .map((_, item) => ({
+            title: $(item).find('p.new-title').text(),
+            link: new URL($(item).find('a').attr('href'), rootUrl).href,
+            author: $(item)
+                .find('p.new-desc')
+                .text()
+                .match(/作者：(.*?)\s/)[1],
+        }))
+        .get();
+
+    const items = await Promise.all(
+        list.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const detailData = await got.get(item.link);
+                const $ = cheerio.load(detailData.data);
+                item.description = $('div.pdbox').html();
+                item.pubDate = timezone(parseDate($('div.newbox > div.newtit > p').text(), 'YYYY-MM-DD HH:mm:ss'), +8);
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: '消费者报道',
+        link: rootUrl,
+        item: items,
+    };
+};

package/lib/v2/ccreports/maintainer.js

@@ -0,0 +1,3 @@
+module.exports = {
+    '/article': ['EsuRt', 'Fatpandac'],
+};

package/lib/v2/ccreports/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'ccreports.com.cn': {
+        _name: '消费者报道',
+        www: [
+            {
+                title: '要闻',
+                docs: 'https://docs.rsshub.app/shopping.html#xiao-fei-zhe-bao-dao-yao-wen',
+                source: ['/'],
+                target: '/ccreports/article',
+            },
+        ],
+    },
+};

package/lib/v2/ccreports/router.js

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

package/lib/v2/cctv/category.js

@@ -0,0 +1,20 @@
+module.exports = async (ctx) => {
+    const { category } = ctx.params;
+    let responseData;
+
+    if (category === 'mzzlbg') {
+        // 每周质量报告
+        const getMzzlbg = require('./utils/mzzlbg');
+        responseData = await getMzzlbg();
+    } else if (category === 'xinwen1j1') {
+        // 新闻1+1
+        const xinwen1j1 = require('./utils/xinwen1j1');
+        responseData = await xinwen1j1(ctx);
+    } else {
+        // 央视新闻
+        const getNews = require('./utils/news');
+        responseData = await getNews(category, ctx);
+    }
+
+    ctx.state.data = responseData;
+};

package/lib/v2/cctv/jx.js

@@ -0,0 +1,51 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+
+module.exports = async (ctx) => {
+    const rootUrl = 'https://photo.cctv.com';
+    const currentUrl = `${rootUrl}/jx/`;
+    const response = await got({
+        method: 'get',
+        url: currentUrl,
+    });
+
+    const $ = cheerio.load(response.data);
+
+    const list = $('.textr a')
+        .slice(0, ctx.query.limit ? parseInt(ctx.query.limit) : 10)
+        .map((_, item) => {
+            item = $(item);
+            return {
+                title: item.text(),
+                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);
+                const date = content('head')
+                    .html()
+                    .match(/publishDate ="(.*) ";/)[1];
+                item.pubDate = date ? parseDate(date, 'YYYYMMDDHHmmss') : null;
+
+                item.description = content('.tujitop').html();
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: '央视网图片《镜象》',
+        link: currentUrl,
+        item: items,
+    };
+};

package/lib/v2/cctv/lm.js

@@ -0,0 +1,78 @@
+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 id = ctx.params.id ?? 'xwzk';
+
+    const rootUrl = 'https://tv.cctv.com';
+    const apiRootUrl = 'https://api.cntv.cn';
+    const vdnRootUrl = 'https://vdn.apps.cntv.cn';
+
+    const currentUrl = `${rootUrl}/lm/${id}/videoset`;
+
+    const titleResponse = await got({
+        method: 'get',
+        url: currentUrl,
+    });
+
+    const $ = cheerio.load(titleResponse.data);
+
+    const topId = titleResponse.data.match(/(TOPC\d{16})/)[1];
+    const apiUrl = `${apiRootUrl}/NewVideo/getVideoListByColumn?id=${topId}&n=20&sort=desc&p=1&mode=0&serviceId=tvcctv`;
+
+    const response = await got({
+        method: 'get',
+        url: apiUrl,
+    });
+
+    const list = response.data.data.list.map((item) => ({
+        url: item.url,
+        guid: item.guid,
+        image: item.image,
+        title: item.title,
+        pubDate: timezone(parseDate(item.time), +8),
+        link: `${vdnRootUrl}/api/getHttpVideoInfo.do?pid=${item.guid}`,
+        description: `<p>${item.brief.replace(/\r\n/g, '</p><p>')}</p>`,
+    }));
+
+    const items = await Promise.all(
+        list.map(
+            async (item) =>
+                await ctx.cache.tryGet(item.link, async () => {
+                    const detailResponse = await got({
+                        method: 'get',
+                        url: item.link,
+                    });
+                    const data = detailResponse.data;
+
+                    item.description += `<video src="${data.hls_url}" controls="controls" poster="${item.image}" width="100%"></video><br>`;
+
+                    for (const c of data.video.chapters) {
+                        item.description += `<video src="${c.url}" controls="controls" poster="${c.image}" width="100%"></video><br>`;
+                    }
+
+                    for (let i = 2; data.video[`chapters${i}`]; i++) {
+                        for (const c of data.video[`chapters${i}`]) {
+                            item.description += `<video src="${c.url}" controls="controls" poster="${c.image}" width="100%"></video><br>`;
+                        }
+                    }
+
+                    item.link = item.url;
+
+                    delete item.url;
+                    delete item.image;
+
+                    return item;
+                })
+        )
+    );
+
+    ctx.state.data = {
+        title: $('title').text(),
+        link: currentUrl,
+        item: items,
+        description: $('meta[name=description]').attr('content'),
+    };
+};

package/lib/v2/cctv/maintainer.js

@@ -0,0 +1,7 @@
+module.exports = {
+    '/lm/:id?': ['nczitzk'],
+    '/photo/jx': ['nczitzk'],
+    '/special/:id?': ['nczitzk'],
+    '/xwlb': ['zengxs'],
+    '/:category': ['idealclover', 'xyqfer'],
+};

package/lib/v2/cctv/radar.js

@@ -0,0 +1,41 @@
+module.exports = {
+    'cctv.com': {
+        _name: '央视新闻',
+        news: [
+            {
+                title: '新闻专题',
+                docs: 'https://docs.rsshub.app/traditional-media.html#yang-shi-xin-wen',
+                source: ['/special/*id'],
+                target: (params) => `/cctv/special/${params.id.split('/')[0]}`,
+            },
+            {
+                title: '专题',
+                docs: 'https://docs.rsshub.app/traditional-media.html#yang-shi-xin-wen',
+                source: ['/:category'],
+                target: '/cctv/:category',
+            },
+        ],
+        photo: [
+            {
+                title: '央视网图片《镜象》',
+                docs: 'https://docs.rsshub.app/traditional-media.html#yang-shi-xin-wen',
+                source: ['/jx', '/'],
+                target: '/cctv/photo/jx',
+            },
+        ],
+        tv: [
+            {
+                title: '新闻联播',
+                docs: 'https://docs.rsshub.app/traditional-media.html#yang-shi-xin-wen',
+                source: ['/lm/xwlb', '/'],
+                target: '/cctv/xwlb',
+            },
+            {
+                title: '栏目',
+                docs: 'https://docs.rsshub.app/traditional-media.html#yang-shi-xin-wen',
+                source: ['/lm/*path', '/'],
+                target: (params) => `/cctv/lm/${params.path.replace('/videoset', '')}`,
+            },
+        ],
+    },
+};

package/lib/v2/cctv/router.js

@@ -0,0 +1,7 @@
+module.exports = (router) => {
+    router.get('/lm/:id?', require('./lm'));
+    router.get('/photo/jx', require('./jx'));
+    router.get('/special/:id?', require('./special'));
+    router.get('/xwlb', require('./xwlb'));
+    router.get('/:category', require('./category'));
+};

package/lib/v2/cctv/special.js

@@ -0,0 +1,110 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+
+module.exports = async (ctx) => {
+    ctx.params.id = ctx.params.id ?? '315';
+
+    let rootUrl;
+
+    if (ctx.params.id === '315') {
+        rootUrl = `https://315.cctv.com`;
+    } else {
+        rootUrl = `https://news.cctv.com/special/${ctx.params.id}/index.shtml`;
+    }
+
+    const response = await got({
+        method: 'get',
+        url: rootUrl,
+    });
+
+    const $ = cheerio.load(response.data);
+    const list = $('h3>a, h4>a')
+        .slice(0, 10)
+        .map((_, item) => {
+            item = $(item);
+            return {
+                title: item.text(),
+                link: item.attr('href'),
+            };
+        })
+        .get();
+
+    const items = await Promise.all(
+        list.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                let date;
+
+                if (item.link.indexOf('photo.cctv.com') > 0) {
+                    const imageLink = item.link.split('.shtml')[0] + '.xml?randomNum=26';
+                    const imageResponse = await got({
+                        method: 'get',
+                        url: imageLink,
+                    });
+                    const images = cheerio.load(imageResponse.data);
+
+                    let description;
+
+                    item.description = '';
+
+                    images('li')
+                        .get()
+                        .forEach((i) => {
+                            i = images(i);
+                            item.description += `<img src="${i.attr('photourl')}">`;
+                            date = i.attr('time').replace(/年/g, '-').replace(/月/g, '-').replace(/日/g, '');
+                            description = i.html();
+                        });
+
+                    item.description += `<p>${description}</p>`;
+                } else {
+                    const detailResponse = await got({
+                        method: 'get',
+                        url: item.link,
+                    });
+                    const content = cheerio.load(detailResponse.data);
+                    const match = detailResponse.data.match(/publishDate ="(.*)";/);
+
+                    if (match === null) {
+                        let dateSpan;
+
+                        if (content('div.info span').html() !== null) {
+                            dateSpan = content('div.info span').text();
+                        } else if (content('div.info1').html() !== null) {
+                            dateSpan = content('div.info1').text().split(' | ')[1];
+                        } else {
+                            dateSpan = content('span.time').text();
+                        }
+
+                        date = timezone(parseDate(dateSpan, 'YYYY年MM月DD日 HH:mm:ss'), +8);
+                    } else {
+                        const dateString = match[1];
+                        date = timezone(parseDate(dateString, 'YYYYMMDDHHmmss'), +8);
+                    }
+
+                    if (content('#content_area').html() !== null) {
+                        item.description = content('#content_area').html();
+                    } else if (content('div.cnt_bd').html() !== null) {
+                        content('div.function,h1,h2').remove();
+                        item.description = content('div.cnt_bd').html();
+                    } else if (content('#text_area').html() !== null) {
+                        item.description = content('#text_area').html();
+                    } else {
+                        item.description = content('div.con').html();
+                    }
+                }
+
+                item.pubDate = date;
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: `${$('title').text().split('_')[0]} - 新闻专题 -  央视网`,
+        link: rootUrl,
+        item: items,
+    };
+};

package/lib/v2/cctv/utils/mzzlbg.js

@@ -0,0 +1,41 @@
+const got = require('@/utils/got');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+
+module.exports = async () => {
+    const url = 'https://api.cntv.cn/video/videolistById?serviceId=cbox&vsid=C10354&em=01&p=1&n=50';
+
+    const response = await got({
+        method: 'get',
+        url,
+    });
+
+    const data = response.data;
+
+    const resultItem = await Promise.all(
+        data.video.map(async (video) => {
+            const url = `https://vdn.apps.cntv.cn/api/getHttpVideoInfo.do?pid=${video.vid}`;
+            const item = {
+                title: video.t,
+                description: video.desc,
+                link: video.url,
+                pubDate: timezone(parseDate(video.ptime), +8),
+            };
+            const { data: videoDetail } = await got({
+                method: 'get',
+                url,
+            });
+
+            item.description += `<br><video src="${videoDetail.hls_url}" controls="controls" poster="${video.img.replace(/\?.+/g, '')}" style="width: 100%"></video>`;
+            return item;
+        })
+    );
+
+    return {
+        title: '每周质量报告',
+        link: 'https://tv.cctv.com/lm/mzzlbg/videoset/index.shtml',
+        description:
+            '《每周质量报告》是CCTV-新闻频道一档以消费者为核心收视人群的新闻专题栏目。创办于2003年，始终致力于产品质量和食品安全领域的调查报道，以打假除劣扶优，推动质量进步为第一诉求，是我国电视新闻界质量新闻领域的旗帜性节目。首播时间：CCTV-新闻周日12:35—12:55。',
+        item: resultItem,
+    };
+};

package/lib/v2/cctv/utils/news.js

@@ -0,0 +1,105 @@
+const got = require('@/utils/got');
+const path = require('path');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+const UA = require('@/utils/rand-user-agent')({ browser: 'mobile safari', os: 'ios', device: 'mobile' });
+
+module.exports = async (category, ctx) => {
+    const url = `https://news.cctv.com/2019/07/gaiban/cmsdatainterface/page/${category}_1.jsonp`;
+
+    const response = await got({
+        method: 'get',
+        url,
+        headers: {
+            Referer: `http://news.cctv.com/${category}`,
+            'User-Agent': UA,
+        },
+    });
+
+    const data = JSON.parse(response.data.slice(category.length + 1, -1));
+    const list = data.data.list;
+    const resultItem = await Promise.all(
+        list.map(({ title, url, focus_date, image }) =>
+            ctx.cache.tryGet(`cctv-news: ${url}`, async () => {
+                const item = {
+                    title,
+                    link: url,
+                    pubDate: timezone(parseDate(focus_date), +8),
+                };
+                const id = path.parse(url).name;
+                const unknownTip = '未知类型，请点击<a href="https://github.com/DIYgod/RSSHub/issues">链接</a>提交issue';
+                let description;
+                let api;
+                let type;
+                let author;
+
+                if (id.startsWith('ART')) {
+                    // 普通新闻
+                    api = `https://api.cntv.cn/Article/getXinwenNextArticleInfo?serviceId=sjnews&id=${id}&t=json`;
+                    type = 'ART';
+                } else if (id.startsWith('PHO')) {
+                    // 图片
+                    api = `https://api.cntv.cn/Article/contentinfo?id=${id}&serviceId=sjnews&t=json`;
+                    type = 'PHO';
+                } else if (id.startsWith('VIDE')) {
+                    // 视频
+                    const vid = path.parse(image).name.split('-')[0];
+                    api = `https://vdn.apps.cntv.cn/api/getHttpVideoInfo.do?pid=${vid}`;
+                    type = 'VIDE';
+                } else {
+                    // 未识别
+                    description = unknownTip;
+                }
+
+                if (api) {
+                    const { data } = await got({
+                        method: 'get',
+                        url: api,
+                        headers: {
+                            'User-Agent': UA,
+                        },
+                    });
+
+                    switch (type) {
+                        case 'ART':
+                            description = data.article_content;
+                            author = data.article_source;
+                            break;
+
+                        case 'PHO':
+                            description = data.photo_album_list.reduce((description, { photo_url, photo_name, photo_brief }) => {
+                                description += `
+                                    <img src=${photo_url} /><br>
+                                    <strong>${photo_name}</strong><br>
+                                    ${photo_brief}<br>
+                                `;
+                                return description;
+                            }, '');
+                            author = data.source;
+                            break;
+
+                        case 'VIDE':
+                            description = `<video src="${data.hls_url}" controls="controls" poster="${image}" style="width: 100%"></video>`;
+                            author = data.article_source;
+                            break;
+
+                        default:
+                            description = unknownTip;
+                    }
+                }
+
+                item.description = description;
+                item.author = author;
+
+                return item;
+            })
+        )
+    );
+
+    return {
+        title: `央视新闻 ${category}`,
+        link: `https://news.cctv.com/${category}`,
+        description: `央视新闻 ${category}`,
+        item: resultItem,
+    };
+};

package/lib/v2/cctv/utils/xinwen1j1.js

@@ -0,0 +1,81 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+//
+// 测试http://localhost:1200/cctv/xinwen1j1
+
+// 完整文章页
+async function load(link) {
+    const res = await got({ method: 'get', url: link });
+    const $ = cheerio.load(res.data);
+    // console.log($('div.image').get().text())
+    // console.log('********')
+    const js_txt = '' + $('script');
+
+    const guid = js_txt.split('guid_Ad_VideoCode = "')[1].split('";')[0];
+    // console.log(guid+' js_txt********')
+    const { data: videoDetail } = await got({
+        method: 'get',
+        url: `http://vdn.apps.cntv.cn/api/getHttpVideoInfo.do?pid=${guid}`,
+    });
+
+    // poster="?" 以后再弄
+    // <video src="${videoDetail.hls_url}" controls="controls"  poster="?"  style="width: 100%"></video>
+    const description = `${$('meta[name=description]').attr('content')}<br><video src="${videoDetail.hls_url}" controls="controls"  style="width: 100%"></video>`;
+
+    // 提取内容
+    return { description };
+}
+
+// 同步 列表
+const ProcessFeed = (data, ctx) =>
+    Promise.all(
+        data.list.map(async (data) => {
+            const title = data.title;
+            const itemUrl = data.url;
+            const itemDate = parseDate(data.focus_date);
+
+            // 列表上提取到的信息
+            const single = {
+                title,
+                link: itemUrl,
+                author: '央视新闻',
+                guid: itemUrl,
+                pubDate: itemDate,
+            };
+
+            // 使用tryGet方法从缓存获取内容。
+            // 当缓存中无法获取到链接内容的时候，则使用load方法加载文章内容。
+            const other = await ctx.cache.tryGet(itemUrl, () => load(itemUrl));
+
+            // 合并解析后的结果集作为该篇文章最终的输出结果
+            return { ...single, ...other };
+        })
+    );
+module.exports = async (ctx) => {
+    const baseUrl = 'https://api.cntv.cn/NewVideo/getVideoListByColumn?id=TOPC1451559066181661&n=20&sort=desc&p=1&mode=0&serviceId=tvcctv';
+    // 获取要处理的页面
+    const res = await got({
+        method: 'get',
+        url: baseUrl,
+        headers: {
+            Referer: 'https://tv.cctv.com/lm/xinwen1j1/videoset/index.shtml#&Type=0',
+        },
+    });
+
+    const data = res.data.data;
+
+    // $('div.image').slice(0,3).find('img').attr('src')
+    // $('div.image').slice(0,3).find('a').attr('href').split('?')[0]
+
+    // 处理所有的页面
+    const result = await ProcessFeed(data, ctx);
+
+    return {
+        title: '《新闻1+1》- 中央电视台新闻频道',
+        link: baseUrl,
+        description:
+            '《新闻1+1》是中央电视台新闻频道的一档访谈类栏目。周一至周五，从时事政策、公共话题、突发事件等大型选题中选取当天最新、最热、最快的新闻话题，还原新闻全貌、解读事件真相，力求以精度、纯度和锐度为新闻导向，呈现最质朴的新闻。首播：CCTV-新闻：周一至周五21:30—21:55；重播：次日01:30，04:30。',
+        item: result,
+    };
+};

package/lib/v2/cctv/xwlb.js

@@ -0,0 +1,50 @@
+const got = require('@/utils/got');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+const cheerio = require('cheerio');
+const dayjs = require('dayjs');
+const customParseFormat = require('dayjs/plugin/customParseFormat');
+dayjs.extend(customParseFormat);
+
+module.exports = async (ctx) => {
+    const res = await got({ method: 'get', url: 'https://tv.cctv.com/lm/xwlb/' });
+    const $ = cheerio.load(res.data);
+    // 解析最新一期新闻联播的日期
+    const latestDate = dayjs($('.rilititle p').text(), 'YYYY-MM-DD');
+    const count = [];
+    for (let i = 0; i < 20; i++) {
+        count.push(i);
+    }
+    const resultItems = await Promise.all(
+        count.map(async (i) => {
+            const newsDate = latestDate.subtract(i, 'days').hour(19);
+            const url = `https://tv.cctv.com/lm/xwlb/day/${newsDate.format('YYYYMMDD')}.shtml`;
+            const item = {
+                title: `新闻联播 ${newsDate.format('YYYY/MM/DD')}`,
+                link: url,
+                pubDate: timezone(parseDate(newsDate), +8),
+                description: await ctx.cache.tryGet(url, async () => {
+                    const res = await got(url);
+                    const content = cheerio.load(res.data);
+                    const list = new Array();
+                    content('body li').map((i, e) => {
+                        e = content(e);
+                        const href = e.find('a').attr('href');
+                        const title = e.find('a').attr('title');
+                        const dur = e.find('span').text();
+                        list.push(`<a href="${href}">${title} ⏱${dur}</a>`);
+                        return i;
+                    });
+                    return list.join('<br/>\n');
+                }),
+            };
+            return item;
+        })
+    );
+
+    ctx.state.data = {
+        title: 'CCTV 新闻联播',
+        link: 'http://tv.cctv.com/lm/xwlb/',
+        item: resultItems,
+    };
+};