rsshub 1.0.0-master.fbcba24 → 1.0.0-master.fc648b5

package/lib/routes/gov/moe/moe.js

@@ -1,84 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-const logger = require('@/utils/logger');
-const liburl = require('url');
-
-const moeUrl = 'http://www.moe.gov.cn/';
-const typesIdMap = [
-    { type: 'policy_anal', id: 'tt_con2', name: '政策解读' },
-    { type: 'newest_file', id: 'nine_con1', name: '最新文件' },
-    { type: 'notice', id: 'nine_con2', name: '公告公示' },
-    { type: 'edu_ministry_news', id: 'nine_con3', name: '教育部简报' },
-];
-
-module.exports = async (ctx) => {
-    const type = ctx.params.type;
-    let id = '';
-    let name = '';
-
-    for (let i = 0; i < typesIdMap.length; i++) {
-        if (typesIdMap[i].type === type) {
-            id = typesIdMap[i].id;
-            name = typesIdMap[i].name;
-        }
-    }
-
-    if (id === '') {
-        logger.error('The given type not found.');
-        return;
-    }
-
-    const response = await got({
-        method: 'get',
-        url: moeUrl,
-        Host: 'www.moe.gov.cn',
-    });
-
-    const $ = cheerio.load(response.data);
-    const newsLis = $('div#' + id + '>ul>li');
-
-    ctx.state.data = {
-        title: name,
-        link: moeUrl,
-        item: await Promise.all(
-            newsLis
-                .map(async (_, item) => {
-                    item = $(item);
-
-                    const firstA = item.find('a');
-                    const itemUrl = liburl.resolve(moeUrl, firstA.attr('href'));
-
-                    const infos = await ctx.cache.tryGet(itemUrl, async () => {
-                        const res = {};
-                        const response = await got({
-                            method: 'get',
-                            url: itemUrl,
-                            Host: 'www.moe.gov.cn',
-                        });
-                        const data = cheerio.load(response.data);
-
-                        if (itemUrl.indexOf('www.gov.cn') > -1) {
-                            res.description = data('div.pages_content > p').html();
-                            res.title = data('div.article > h1').text();
-                        } else if (itemUrl.indexOf('srcsite') > -1) {
-                            res.description = data('div#xxgk_content_div').html();
-                            res.title = data('div#xxgk_content_redheadbg > h1').text();
-                        } else if (itemUrl.indexOf('jyb_') > -1) {
-                            res.description = data('div.TRS_Editor').html();
-                            res.title = data('div#content_body > h1').text();
-                        }
-
-                        return res;
-                    });
-
-                    return {
-                        title: infos.title,
-                        description: infos.description,
-                        link: itemUrl,
-                        pubDate: `${new Date().getFullYear()}-${item.find('span').text()}`,
-                    };
-                })
-                .get()
-        ),
-    };
-};

package/lib/routes/guancha/headline.js

@@ -1,48 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-
-module.exports = async (ctx) => {
-    const rootUrl = 'https://www.guancha.cn';
-    const currentUrl = `${rootUrl}/GuanChaZheTouTiao/list_1.shtml`;
-    const response = await got({
-        method: 'get',
-        url: currentUrl,
-    });
-
-    const $ = cheerio.load(response.data);
-
-    const list = $('.headline-list li .content-headline h3 a')
-        .map((_, item) => {
-            item = $(item);
-            return {
-                title: item.text(),
-                link: `${rootUrl}${item.attr('href')}`,
-                description: item.parent().next().html(),
-                pubDate: new Date(item.parents('div').eq(0).find('span').text()).toUTCString(),
-            };
-        })
-        .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('.all-txt').html();
-
-                return item;
-            })
-        )
-    );
-
-    ctx.state.data = {
-        title: '观察者网 - 头条',
-        link: currentUrl,
-        item: items,
-    };
-};

package/lib/routes/guancha/index.js

@@ -1,132 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-const date = require('@/utils/date');
-
-const config = {
-    review: { title: '评论 & 研究', query: '.module-news-main' },
-    story: { title: '要闻', query: '.img-List' },
-    fengwen: { title: '风闻', query: '.fengwen-list' },
-    redian: { title: '热点' },
-    gundong: { title: '滚动' },
-    all: { title: '全部' },
-    home: { title: '首页' },
-    others: { title: '热点 & 滚动' },
-};
-
-module.exports = async (ctx) => {
-    const total = 10;
-    const caty = ctx.params.caty || 'all';
-    const rootUrl = 'https://www.guancha.cn';
-
-    let newsList = [],
-        redianList = [],
-        gundongList = [];
-
-    // 'review', 'story' and 'fengwen' come from homepage.
-
-    if (caty === 'review' || caty === 'story' || caty === 'fengwen' || caty === 'all' || caty === 'home') {
-        const response = await got({
-            method: 'get',
-            url: rootUrl,
-        });
-
-        const $ = cheerio.load(response.data);
-
-        const fetchPost = (slice) =>
-            slice
-                .find('h4.module-title a')
-
-                // Filter some blank links which lead to no contents but 'https://user.guancha.cn'.
-
-                .filter((_, item) => $(item).attr('href') !== 'https://user.guancha.cn')
-                .map((_, item) => {
-                    item = $(item);
-                    const link = item.attr('href');
-                    return {
-                        title: item.text(),
-                        link: (link.indexOf('http') === 0 ? '' : rootUrl) + link,
-                    };
-                })
-                .get();
-
-        if (caty === 'all' || caty === 'home') {
-            newsList = fetchPost($(config.review.query))
-                .slice(0, total / 3)
-                .concat(fetchPost($(config.story.query)).slice(0, total / 3), fetchPost($(config.fengwen.query)).slice(0, total / 3));
-        } else {
-            newsList = fetchPost($(config[caty].query)).slice(0, total);
-        }
-    }
-
-    // 'redian' and 'gundong' come from api.
-
-    if (caty === 'redian' || caty === 'all' || caty === 'others') {
-        const response = await got({
-            method: 'get',
-            url: `${rootUrl}/api/redian.htm`,
-        });
-        redianList = response.data.items
-            .map((item) => ({
-                title: item.TITLE,
-                link: `${rootUrl}${item.HTTP_URL}`,
-            }))
-            .slice(0, caty === 'all' ? total / 3 : total);
-    }
-    if (caty === 'gundong' || caty === 'all' || caty === 'others') {
-        const response = await got({
-            method: 'get',
-            url: `${rootUrl}/api/gundong.htm`,
-        });
-        gundongList = response.data.items
-            .map((item) => ({
-                title: item.TITLE,
-                link: `${rootUrl}${item.HTTP_URL}`,
-            }))
-            .slice(0, caty === 'all' ? total / 3 : total);
-    }
-
-    const items = await Promise.all(
-        newsList.concat(redianList, gundongList).map((item) =>
-            ctx.cache.tryGet(item.link, async () => {
-                let detailResponse = await got({
-                    method: 'get',
-                    url: item.link,
-                });
-
-                // Some links to posts in 'fengwen' must be redirected in order to fetch content.
-                // eg. https://www.guancha.cn/politics/2020_10_23_569021.shtml
-                //  => https://user.guancha.cn/main/content?id=399176
-
-                const jumpMatch = detailResponse.data.match(/user.guancha.cn\/main\/content\?id=(.*)";/);
-                if (jumpMatch !== null) {
-                    item.link = `https://user.guancha.cn/main/content?id=${jumpMatch[1]}`;
-                    detailResponse = await got({
-                        method: 'get',
-                        url: item.link,
-                    });
-                }
-
-                const content = cheerio.load(detailResponse.data);
-
-                const dateMatch = detailResponse.data.match(/"pubDate": "(.*)"/);
-                if (dateMatch === null) {
-                    // PubDates of posts in 'fengwen' are in an informal format.
-
-                    item.pubDate = date(content('.time1').text(), 8);
-                } else {
-                    item.pubDate = new Date(dateMatch[1]).toUTCString();
-                }
-
-                item.description = content('.all-txt').html() || content('.article-txt-content').html();
-
-                return item;
-            })
-        )
-    );
-
-    ctx.state.data = {
-        title: `观察者网 - ${config[caty].title}`,
-        link: rootUrl,
-        item: items,
-    };
-};

package/lib/routes/guancha/member.js

@@ -1,71 +0,0 @@
-const got = require('@/utils/got');
-
-const title = {
-    recommend: '精选',
-    books: '观书堂',
-    courses: '在线课',
-    huodongs: '观学院',
-};
-
-module.exports = async (ctx) => {
-    const caty = ctx.params.caty || 'recommend';
-
-    const rootUrl = 'https://member.guancha.cn';
-    const apiUrl = `${rootUrl}/zaixianke/home`;
-    const response = await got({
-        method: 'get',
-        url: apiUrl,
-    });
-
-    let items;
-
-    switch (caty) {
-        case 'books':
-            items = response.data.data.books.map((item) => ({
-                title: item.title,
-                link: `${rootUrl}/guanshutang/summary.html?id=${item.id}`,
-                description: `<img src="${item.cover}"><p>[${item.audio_time}] ${item.desc_short}</p>`,
-                pubDate: new Date(parseInt(item.cover.split('/').pop().substr(0, 10)) * 1000).toUTCString(),
-            }));
-            break;
-
-        case 'courses':
-            items = response.data.data.courses.data.map((item) => {
-                let description = '',
-                    pubDate = new Date(0);
-
-                for (const i of item.items) {
-                    const newPubDate = new Date(i.publish_time);
-                    pubDate = pubDate > newPubDate ? pubDate : newPubDate;
-                    description += `<a href="${rootUrl}/zaixianke/content.html?id=${i.id}">${i.title}</a><br>`;
-                }
-
-                return {
-                    title: item.name,
-                    link: `${rootUrl}/zaixianke/summary.html?id=${item.id}`,
-                    author: item.author_name,
-                    description: `<img src="${item.cover}"><p>${item.desc_short}</p><br>${description}`,
-                    pubDate: pubDate.toUTCString(),
-                };
-            });
-            break;
-
-        default:
-            items = response.data.data[caty].map((item) => ({
-                title: item.title,
-                link: item.jump_url,
-                author: item.author_name,
-                description: `<img src="${item.big_pic}"><p>${item.summary}</p>`,
-                enclosure_url: item.media_url,
-                enclosure_length: item.media_size,
-                enclosure_type: 'audio/mpeg',
-                pubDate: new Date(item.created_at * 1000).toUTCString(),
-            }));
-    }
-
-    ctx.state.data = {
-        title: `观学院 - ${title[caty]}`,
-        link: `${rootUrl}/index.html`,
-        item: items,
-    };
-};

package/lib/routes/guancha/topic.js

@@ -1,55 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-const date = require('@/utils/date');
-
-module.exports = async (ctx) => {
-    const id = ctx.params.id || '0';
-    const order = ctx.params.order || '1';
-
-    const rootUrl = 'https://user.guancha.cn';
-    const currentUrl = `${rootUrl}/${id === '0' ? 'main/index-list.json?' : 'topic/post-list?topic_id=' + id}&page=1&order=${order}`;
-
-    const response = await got({
-        method: 'get',
-        url: currentUrl,
-    });
-
-    const $ = cheerio.load(response.data);
-
-    const list = $('.list-item h4 a, ul.home li .list-item h4 a')
-        .slice(0, 10)
-        .map((_, item) => {
-            item = $(item);
-            return {
-                title: item.text(),
-                link: `${rootUrl}${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.pubDate = date(content('.time1').text(), 8);
-                item.author = content('.user-main h4 a').eq(0).text();
-                item.description = content('.article-txt-content').html();
-
-                return item;
-            })
-        )
-    );
-
-    $('h1.title span').remove();
-
-    ctx.state.data = {
-        title: `观察者网 - ${id === '0' ? '风闻' : $('h1.title').text()}`,
-        link: currentUrl,
-        item: items,
-    };
-};

package/lib/routes/guokr/calendar.js

@@ -1,47 +0,0 @@
-const got = require('@/utils/got');
-
-async function loadFullPage(ctx, id) {
-    const link = `https://apis.guokr.com/minisite/article/${id}.json`;
-    const content = await ctx.cache.tryGet(link, async () => {
-        const res = await got.get(link);
-        return res.data.result.content;
-    });
-    return content;
-}
-
-const channelMap = {
-    calendar: 'pac',
-    institute: 'predator',
-    foodlab: 'predator',
-    pretty: 'beauty',
-};
-
-module.exports = async (ctx) => {
-    const channel = channelMap[ctx.params.channel] ? channelMap[ctx.params.channel] : ctx.params.channel;
-
-    const response = await got.get(`https://www.guokr.com/apis/minisite/article.json?retrieve_type=by_wx&channel_key=${channel}&offset=0&limit=10`);
-    const items = response.data.result;
-
-    if (items.length === 0) {
-        throw 'Unknown channel';
-    }
-
-    const channel_name = items[0].channels[0].name;
-    const channel_url = items[0].channels[0].url;
-
-    const result = await Promise.all(
-        items.map(async (item) => ({
-            title: item.title,
-            description: await loadFullPage(ctx, item.id), // Mercury 无法正确解析全文，故这里手动加载
-            pubDate: item.date_published,
-            link: item.url,
-            author: item.author.nickname,
-        }))
-    );
-
-    ctx.state.data = {
-        title: `果壳网 ${channel_name}`,
-        link: channel_url,
-        item: result,
-    };
-};

package/lib/routes/guokr/scientific.js

@@ -1,30 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-
-module.exports = async (ctx) => {
-    const response = await got.get('https://www.guokr.com/apis/minisite/article.json?retrieve_type=by_subject&limit=20&offset=0');
-
-    const result = response.data.result;
-
-    ctx.state.data = {
-        title: '果壳网 科学人',
-        link: 'https://www.guokr.com/scientific',
-        description: '果壳网 科学人',
-        item: await Promise.all(
-            result.map((item) =>
-                ctx.cache.tryGet(item.url, async () => {
-                    const res = await got.get(item.url);
-                    const $ = cheerio.load(res.data);
-                    item.description = $('.eflYNZ #js_content').css('visibility', 'visible').html() || $('.bPfFQI').html();
-                    return {
-                        title: item.title,
-                        description: item.description,
-                        pubDate: item.date_published,
-                        link: item.url,
-                        author: item.author.nickname,
-                    };
-                })
-            )
-        ),
-    };
-};

package/lib/routes/hk01/channel.js

@@ -1,22 +0,0 @@
-const got = require('@/utils/got');
-
-module.exports = async (ctx) => {
-    const id = ctx.params.id;
-
-    const response = await got(`https://web-data.api.hk01.com/v2/page/category/${id}`);
-    const data = response.data;
-    const list = data.sections[0].items;
-
-    ctx.state.data = {
-        title: `香港01 - ${data.category.publishName}`,
-        description: data.meta.metaDesc,
-        link: data.category.publishUrl,
-        item: list.map((item) => ({
-            title: item.data.title,
-            author: item.data.authors && item.data.authors.map((e) => e.publishName).join(', '),
-            description: `<p>${item.data.description}…</p><img style="width: 100%" src="${item.data.mainImage.cdnUrl}" />`,
-            pubDate: new Date(item.data.lastModifyTime * 1000),
-            link: item.data.canonicalUrl,
-        })),
-    };
-};

package/lib/routes/hk01/hot.js

@@ -1,20 +0,0 @@
-const got = require('@/utils/got');
-
-module.exports = async (ctx) => {
-    const response = await got('https://web-data.api.hk01.com/v2/page/hot/');
-    const data = response.data;
-    const list = data.items;
-
-    ctx.state.data = {
-        title: '香港01 - 熱門',
-        description: data.meta.metaDesc,
-        link: data.meta.canonicalUrl,
-        item: list.map((item) => ({
-            title: item.data.title,
-            author: item.data.authors && item.data.authors.map((e) => e.publishName).join(', '),
-            description: `<p>${item.data.description}</p><img style="width: 100%" src="${item.data.mainImage.cdnUrl}" />`,
-            pubDate: new Date(item.data.lastModifyTime * 1000),
-            link: item.data.canonicalUrl,
-        })),
-    };
-};

package/lib/routes/hk01/issue.js

@@ -1,22 +0,0 @@
-const got = require('@/utils/got');
-
-module.exports = async (ctx) => {
-    const id = ctx.params.id;
-
-    const response = await got(`https://web-data.api.hk01.com/v2/page/issue/${id}`);
-    const data = response.data;
-    const list = data.issue.blocks[0].articles;
-
-    ctx.state.data = {
-        title: `香港01 - ${data.issue.title}`,
-        description: data.meta.metaDesc,
-        link: data.issue.publishUrl,
-        item: list.map((item) => ({
-            title: item.data.title,
-            author: item.data.authors && item.data.authors.map((e) => e.publishName).join(', '),
-            description: `<p>${item.data.description}…</p><img style="width: 100%" src="${item.data.mainImage.cdnUrl}" />`,
-            pubDate: new Date(item.data.lastModifyTime * 1000),
-            link: item.data.canonicalUrl,
-        })),
-    };
-};

package/lib/routes/hk01/tag.js

@@ -1,22 +0,0 @@
-const got = require('@/utils/got');
-
-module.exports = async (ctx) => {
-    const id = ctx.params.id;
-
-    const response = await got(`https://web-data.api.hk01.com/v2/page/tag/${id}`);
-    const data = response.data;
-    const list = data.articles;
-
-    ctx.state.data = {
-        title: `香港01 - ${data.tag.tagName}`,
-        description: data.meta.metaDesc,
-        link: data.tag.publishUrl,
-        item: list.map((item) => ({
-            title: item.data.title,
-            author: item.data.authors && item.data.authors.map((e) => e.publishName).join(', '),
-            description: `<p>${item.data.description}</p><img style="width: 100%" src="${item.data.mainImage.cdnUrl}" />`,
-            pubDate: new Date(item.data.lastModifyTime * 1000),
-            link: item.data.canonicalUrl,
-        })),
-    };
-};

package/lib/routes/hk01/zone.js

@@ -1,41 +0,0 @@
-const got = require('@/utils/got');
-
-module.exports = async (ctx) => {
-    const id = ctx.params.id;
-
-    const response = await got(`https://web-data.api.hk01.com/v2/page/zone/${id}`);
-    const data = response.data;
-    const list = data.sections[0].items;
-
-    ctx.state.data = {
-        title: `香港01 - ${data.zone.publishName}`,
-        description: data.meta.metaDesc,
-        link: data.zone.publishUrl,
-        item: list.map((item) => {
-            let author;
-            let description;
-            let pubDate;
-            switch (item.type) {
-                case 1:
-                    author = item.data.authors && item.data.authors.map((e) => e.publishName).join(', ');
-                    description = `<p>${item.data.description}</p><img style="width: 100%" src="${item.data.mainImage.cdnUrl}" />`;
-                    pubDate = new Date(item.data.lastModifyTime * 1000);
-                    break;
-                case 2:
-                    author = item.data.zonePublishName;
-                    description = `${item.data.teaser.map((e) => '<p>' + e + '</p>').join('')}<img style="width: 100%" src="${item.data.mainImage.cdnUrl}" />`;
-                    pubDate = new Date();
-                    break;
-                default:
-                    break;
-            }
-            return {
-                title: item.data.title,
-                author,
-                description,
-                pubDate,
-                link: item.data.canonicalUrl,
-            };
-        }),
-    };
-};

package/lib/routes/hupu/all.js

@@ -1,53 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-
-module.exports = async (ctx) => {
-    const rootUrl = 'https://bbs.hupu.com';
-    const currentUrl = `${rootUrl}/all-${ctx.params.caty}`;
-
-    const response = await got({
-        method: 'get',
-        url: currentUrl,
-    });
-    const $ = cheerio.load(response.data);
-
-    const list = $('div.list ul li')
-        .slice(0, 20)
-        .map((_, item) => {
-            item = $(item);
-            const a = item.find('span.textSpan a[title]');
-            return {
-                title: a.attr('title'),
-                link: `${rootUrl}${a.attr('href')}`,
-            };
-        })
-        .get();
-
-    const items = await Promise.all(
-        list.map((item) =>
-            ctx.cache.tryGet(item.link, async () => {
-                try {
-                    const detailResponse = await got({
-                        method: 'get',
-                        url: item.link,
-                    });
-                    const content = cheerio.load(detailResponse.data);
-
-                    item.description = content('div.quote-content').html();
-                    item.author = content('div.author div.left a.u').eq(0).text();
-                    item.pubDate = new Date(content('span.stime').eq(0).text() + ' GMT+8').toUTCString();
-
-                    return item;
-                } catch (error) {
-                    return Promise.resolve('');
-                }
-            })
-        )
-    );
-
-    ctx.state.data = {
-        title: $('title').text(),
-        link: currentUrl,
-        item: items,
-    };
-};