rsshub 1.0.0-master.fc91df6 → 1.0.0-master.fccb918

package/lib/routes/hkej/index.js

@@ -0,0 +1,138 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const dayjs = require('dayjs');
+const utc = require('dayjs/plugin/utc');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+
+const categories = {
+    index: {
+        name: '',
+        link: '/instantnews/',
+        title: '即時香港中國 國際金融 股市經濟新聞',
+        description: '全天候即時港股、香港財經、國際金融和經濟新聞、中國經濟新聞資訊和分析',
+    },
+    stock: {
+        name: '港股直擊',
+        link: '/instantnews/stock',
+        title: '即時香港股市 股份板塊 攻略分析',
+        description: '全天候即時港股追蹤和直擊分析，股份異動、大行報告、沽空、速評',
+    },
+    hongkong: {
+        name: '香港財經',
+        link: '/instantnews/hongkong',
+        title: '即時香港經濟 中港經濟融合追蹤分析',
+        description: '香港經濟和焦點行業 中港融合和商機的分析',
+    },
+    china: {
+        name: '中國財經',
+        link: '/instantnews/china',
+        title: '即時中國經濟 國策焦點 中港融合追蹤分析',
+        description: '香港經濟和焦點行業 中港融合和商機的分析',
+    },
+    international: {
+        name: '國際財經',
+        link: '/instantnews/international',
+        title: '即時國際財經 股市匯市 央行政策',
+        description: '國際財經 金融股市 央行政策的新聞和分析',
+    },
+    property: {
+        name: '地產新聞',
+        link: '/property/news',
+        title: '地產投資',
+        description: '即時地產新聞, 新盤資訊, 樓市分析, 藍籌屋苑數據及室內設計鑑賞',
+    },
+    current: {
+        name: '時事脈搏',
+        link: '/instantnews/current',
+        title: '即時香港中國 國際金融 股市經濟新聞',
+        description: '全天候即時香港股市、金融、經濟新聞資訊和分析，致力與讀者一起剖釋香港、關注兩岸、放眼全球政經格局',
+    },
+};
+
+module.exports = async (ctx) => {
+    const category = ctx.params.category ?? 'index';
+    const cat = categories[category];
+    const baseUrl = 'https://www2.hkej.com';
+
+    dayjs.extend(utc);
+    const yesterday = dayjs().utcOffset(8).subtract(1, 'day').format('YYYY-MM-DD');
+    const today = dayjs().utcOffset(8).format('YYYY-MM-DD');
+
+    const response = await got({
+        method: 'get',
+        url: baseUrl + cat.link,
+        headers: {
+            Referer: baseUrl,
+        },
+    });
+
+    const $ = cheerio.load(response.data);
+
+    const list = $('h3.in_news_u_t a, h4.hkej_hl-news_topic_2014 a, div.hkej_toc_listingAll_news2_2014 h3 a, div.hkej_toc_cat_top_detail h3 a, div.allNews div.news h1 a, div#div_listingAll div.news2 h3 a')
+        .map((_, item) => {
+            item = $(item);
+            return {
+                title: item.text().trim(),
+                link: baseUrl + item.attr('href').substring(0, item.attr('href').lastIndexOf('/')),
+            };
+        })
+        .get();
+
+    const items = await Promise.all(
+        list &&
+            list.map((item) =>
+                ctx.cache.tryGet(item.link, async () => {
+                    const article = await got({
+                        method: 'get',
+                        url: item.link,
+                        headers: {
+                            Referer: cat.link,
+                        },
+                    });
+                    const content = cheerio.load(article.data);
+
+                    // remove unwanted elements
+                    content('#ad_popup').remove();
+                    content('[class^=ad-]').remove();
+                    content('[id^=ad-]').remove();
+                    content('[id^=div-gpt-ad-]').remove();
+                    content('.hkej_sub_ex_article_nonsubscriber_ad_2014').remove();
+
+                    // fix article image
+                    content('div.hkej_detail_thumb_2014 a').each((_, e) => {
+                        e = $(e);
+                        if (e.attr('href')) {
+                            content(e).after(`<figure><img src="${e.attr('href')}" alt="${e.attr('title')}" title="${e.attr('title')}"><figcaption>${e.attr('title')}</figcaption></figure>`);
+                            content(e).remove();
+                        }
+                    });
+                    const articleImg = content('.hkej_detail_thumb_2014 .enlargeImg').html() ?? '';
+
+                    const pubDate = content('p.info span.date')
+                        .text()
+                        .trim()
+                        .replace('日', '')
+                        .replace('今', today)
+                        .replace('昨', yesterday)
+                        .replace(/[年月]/g, '-');
+
+                    item.category = content('p.info span.cate a')
+                        .toArray()
+                        .map((e) => content(e).text().trim());
+                    item.description = articleImg + content('div#article-content').html();
+                    item.pubDate = timezone(parseDate(pubDate), +8);
+
+                    return item;
+                })
+            )
+    );
+
+    ctx.state.data = {
+        title: `信報網站 - ${cat.title} - 信報網站 hkej.com`,
+        link: baseUrl + cat.link,
+        description: `信報網站(www.hkej.com)即時新聞${cat.name}，提供${cat.description}。`,
+        item: items,
+        language: 'zh-hk',
+    };
+};

package/lib/routes/infzm/news.js

@@ -1,3 +1,4 @@
+const config = require('@/config').value;
 const got = require('@/utils/got');
 const timezone = require('@/utils/timezone');
 const cheerio = require('cheerio');
@@ -28,26 +29,17 @@ module.exports = async (ctx) => {
             if (value) {
                 description = value;
             } else {
+                const cookie = config.infzm.cookie;
                 const response = await got({
                     method: 'get',
-                    url: `http://api.infzm.com/mobile/contents/${id}`,
+                    url: `http://www.infzm.com/content/${id}`,
+                    headers: {
+                        Referer: link,
+                        Cookie: cookie || `passport_session=${Math.floor(Math.random() * 100)};`,
+                    },
                 });
-
-                if (response.data.code === 200) {
-                    description = response.data.data.content.fulltext;
-                } else {
-                    // 有时候接口会返回未发布但实际已发布
-                    const response = await got({
-                        method: 'get',
-                        url: `http://www.infzm.com/content/${id}`,
-                        headers: {
-                            Referer: link,
-                            Cookie: `PHPSESSID=${Math.floor(Math.random() * 100)};`,
-                        },
-                    });
-                    const $ = cheerio.load(response.data);
-                    description = $('#content').html();
-                }
+                const $ = cheerio.load(response.data);
+                description = $('div.nfzm-content__content').html();
 
                 ctx.cache.set(key, description);
             }

package/lib/routes/instagram/index.js

@@ -2,6 +2,7 @@ const cheerio = require('cheerio');
 const { IgLoginRequiredError } = require('instagram-private-api');
 const { ig, login } = require('./utils');
 const logger = require('@/utils/logger');
+const config = require('@/config').value;
 
 // loadContent pulls the desired user/tag/etc
 async function loadContent(category, nameOrId) {
@@ -23,7 +24,7 @@ async function loadContent(category, nameOrId) {
             }
 
             description = userInfo.biography;
-            logo = userInfo.hd_profile_pic_url_info.url;
+            logo = userInfo.profile_pic_url;
             const fullName = userInfo.full_name;
             title = `${fullName} (@${username}) - Instagram`;
             link = `https://www.instagram.com/${username}`;
@@ -66,8 +67,8 @@ module.exports = async (ctx) => {
         throw Error('Such feed is not supported.');
     }
 
-    if (process.env.IG_PROXY) {
-        ig.state.proxyUrl = process.env.IG_PROXY;
+    if (config.instagram && config.instagram.proxy) {
+        ig.state.proxyUrl = config.instagram.proxy;
     }
 
     // Unique key for a feed category

package/lib/routes/instagram/utils.js

@@ -1,26 +1,27 @@
 const { IgApiClient } = require('instagram-private-api');
 const logger = require('@/utils/logger');
+const config = require('@/config').value;
 
 const ig = new IgApiClient();
 () => login(ig); // deepscan-disable-line UNUSED_EXPR
 
 async function login(ig) {
-    if (process.env.IG_USERNAME && process.env.IG_PASSWORD) {
+    if (config.instagram && config.instagram.username && config.instagram.password) {
         try {
-            ig.state.generateDevice(process.env.IG_USERNAME);
+            ig.state.generateDevice(config.instagram.username);
             try {
                 await ig.simulate.preLoginFlow();
             } catch (error) {
                 logger.info('Instagram preLoginFlow fail: ' + error);
             }
-            await ig.account.login(process.env.IG_USERNAME, process.env.IG_PASSWORD);
+            await ig.account.login(config.instagram.username, config.instagram.password);
             process.nextTick(() => ig.simulate.postLoginFlow());
             logger.info('Instagram login success.');
         } catch (error) {
             logger.error('Instagram login fail: ' + error);
         }
     } else {
-        throw Error('Instagram username and password are required to be set in the environment.');
+        throw Error('Instagram RSS is disabled due to the lack of <a href="https://docs.rsshub.app/install/#pei-zhi-bu-fen-rss-mo-kuai-pei-zhi">relevant config</a>');
     }
 }
 

package/lib/routes/ipsw/index.js

@@ -23,7 +23,7 @@ module.exports = async (ctx) => {
     const response = await got({
         method: 'get',
         url: link,
-        header: {
+        headers: {
             Referer: host,
         },
     });
@@ -77,7 +77,7 @@ module.exports = async (ctx) => {
             const response = await got({
                 method: 'get',
                 url: itemUrl,
-                header: {
+                headers: {
                     Referer: host,
                 },
             });

package/lib/routes/jike/user.js

@@ -49,10 +49,12 @@ module.exports = async (ctx) => {
                 });
 
             let content = item.content || (item.linkInfo && item.linkInfo.title) || (item.question && item.question.title) || item.title || '';
+            content = content.replace(/\r\n|\n|\r/g, '<br>');
 
             let shortenTitle = '一条动态';
             if (content) {
                 shortenTitle = content.length > 75 ? `${content.substr(0, 75)}...` : content;
+                shortenTitle = shortenTitle.replace(/(<br>)+/g, ' ');
                 content = `${content}<br><br>`;
             }
 
@@ -65,8 +67,8 @@ module.exports = async (ctx) => {
                         repostImgTemplate += `<br><img referrerpolicy="no-referrer" src="${item.thumbnailUrl}">`;
                     });
 
-                const repostContent = `转发 ${screenNameTemplate}: ${item.target.content}${repostImgTemplate}`;
-                content = `${content}${repostContent}`.replace(/\n|\r/g, '<br>');
+                const repostContent = `转发 ${screenNameTemplate}: ${item.target.content}${repostImgTemplate}`.replace(/\r\n|\n|\r/g, '<br>');
+                content = `${content}${repostContent}`;
             }
             // 部分功能未知
             /* else if (item.type === 'ANSWER') {
@@ -101,7 +103,7 @@ module.exports = async (ctx) => {
 
             const single = {
                 title: `${typeMap[item.type]}了: ${shortenTitle}`,
-                description: `${content}${linkTemplate}${imgTemplate}`,
+                description: `${content}${linkTemplate}${imgTemplate}`.replace(/(<br>|\s)+$/, ''),
                 pubDate: new Date(item.createdAt).toUTCString(),
                 link: linkMap[item.type],
             };

package/lib/routes/lastfm/loved.js

@@ -3,7 +3,7 @@ const config = require('@/config').value;
 
 module.exports = async (ctx) => {
     if (!config.lastfm || !config.lastfm.api_key) {
-        throw 'Last.fm RSS is disabled due to the lack of <a href="https://docs.rsshub.app/install/#bu-fen-rss-mo-kuai-pei-zhi">relevant config</a>';
+        throw 'Last.fm RSS is disabled due to the lack of <a href="https://docs.rsshub.app/install/#pei-zhi-bu-fen-rss-mo-kuai-pei-zhi">relevant config</a>';
     }
 
     const user = ctx.params.user;

package/lib/routes/lastfm/recent.js

@@ -3,7 +3,7 @@ const config = require('@/config').value;
 
 module.exports = async (ctx) => {
     if (!config.lastfm || !config.lastfm.api_key) {
-        throw 'Last.fm RSS is disabled due to the lack of <a href="https://docs.rsshub.app/install/#bu-fen-rss-mo-kuai-pei-zhi">relevant config</a>';
+        throw 'Last.fm RSS is disabled due to the lack of <a href="https://docs.rsshub.app/install/#pei-zhi-bu-fen-rss-mo-kuai-pei-zhi">relevant config</a>';
     }
 
     const user = ctx.params.user;

package/lib/routes/lastfm/top.js

@@ -3,7 +3,7 @@ const config = require('@/config').value;
 
 module.exports = async (ctx) => {
     if (!config.lastfm || !config.lastfm.api_key) {
-        throw 'Last.fm RSS is disabled due to the lack of <a href="https://docs.rsshub.app/install/#bu-fen-rss-mo-kuai-pei-zhi">relevant config</a>';
+        throw 'Last.fm RSS is disabled due to the lack of <a href="https://docs.rsshub.app/install/#pei-zhi-bu-fen-rss-mo-kuai-pei-zhi">relevant config</a>';
     }
 
     const country = ctx.params.country;

package/lib/routes/lizhi/user.js

@@ -18,7 +18,7 @@ module.exports = async (ctx) => {
     const list = response.data.audios.map((item) => ({
         title: item.name,
         enclosure_url: item.url,
-        enclosure_length: item.duration,
+        itunes_duration: item.duration,
         enclosure_type: 'audio/mpeg',
         link: `https://www.lizhi.fm/${$('div.user_info div.desc span').eq(0).text().replace('FM ', '')}/${item.id}`,
         pubDate: new Date(item.create_time).toUTCString(),

package/lib/routes/lolapp/article.js

@@ -0,0 +1,53 @@
+const got = require('@/utils/got');
+const { parseRelativeDate } = require('@/utils/parse-date');
+const { parseDate } = require('@/utils/parse-date');
+const dayjs = require('dayjs');
+module.exports = async (ctx) => {
+    const uuid = ctx.params.uuid;
+    const url = 'https://mlol.qt.qq.com/go/mlol_news/author_feeds?author_uuid=' + uuid;
+    const res = await got.get(url).json();
+    const data = res.data.feedsInfo;
+    const name = data[0].feedNews.footer.source;
+    const newData = [];
+    let pubDate = [];
+    for (let i = 0; i < data.length; i++) {
+        newData[i] = data[i].feedNews.body;
+        newData[i].link = 'https://mlol.qt.qq.com/go/mlol_news/varcache_article?is_lqt=true&docid=' + newData[i].commentID;
+        pubDate[i] = getPublishedDate(newData[i].link);
+    }
+    pubDate = await Promise.all(pubDate);
+    for (let i = 0; i < data.length; i++) {
+        newData[i].pubDate = pubDate[i];
+    }
+    ctx.state.data = {
+        title: `${name}的文章`,
+        link: String(url),
+        description: `${name}的文章`,
+        item: newData.map((item) => ({
+            title: item.title,
+            description: `<img src='${item.imgUrl}'>`,
+            link: item.link,
+            pubDate: item.pubDate,
+        })),
+    };
+};
+
+async function getPublishedDate(url) {
+    const res = await got.get(url).text();
+    const start = res.indexOf('createdStr') + 15;
+    const end = res.indexOf('"', start);
+    const date = res.substring(start, end);
+    let publishedDate;
+    if (date === '刚刚') {
+        publishedDate = dayjs();
+    } else if (date.indexOf('前') > 0) {
+        // XX分钟/小时前
+        publishedDate = parseRelativeDate(date);
+    } else if (date.indexOf('年') > 0) {
+        // YYYY年MM月DD日+MM月DD日 hh:mm
+        publishedDate = parseDate(date, ['YYYY年MM月DD日', 'MM月DD日 hh:mm']);
+    } else {
+        publishedDate = dayjs();
+    }
+    return publishedDate;
+}

package/lib/routes/manhuagui/comic.js

@@ -3,23 +3,47 @@ const cheerio = require('cheerio');
 const got = require('@/utils/got');
 const LZString = require('lz-string');
 
-const getChapters = ($) =>
-    $('.chapter-list > ul')
+const getChapters = ($) => {
+    let time_mark = 100;
+    // 用于一次更新多个新章节的排序
+    let new_time_mark = 0;
+    return $('h4')
         .toArray()
-        .reverse()
-        .reduce((acc, curr) => acc.concat($(curr).children('li').toArray()), [])
         .map((ele) => {
-            const a = $(ele).children('a');
-            return {
-                link: resolve('https://www.manhuagui.com/', a.attr('href')),
-                title: a.attr('title'),
-                num: a.find('i').text(),
-            };
-        });
-
+            const categoryName = $(ele).text();
+            while (!$(ele.next).hasClass('chapter-list')) {
+                ele = ele.next;
+            }
+            ele = ele.next;
+            return $(ele)
+                .children('ul')
+                .toArray()
+                .reverse()
+                .reduce((acc, curr) => acc.concat($(curr).children('li').toArray()), [])
+                .map((ele) => {
+                    const a = $(ele).children('a');
+                    // 通过操作发布时间来对章节进行排序,如果是刚刚更新的单行本或者番外,保留最新更新时间
+                    let pDate = new Date(new Date($.pubDate) - time_mark++ * 1000);
+                    if (a.find('em').length > 0) {
+                        // 对更新的章节也进行排序
+                        pDate = new Date(new Date($.pubDate) - new_time_mark++ * 1000);
+                        $.newChapterCnt++;
+                    }
+                    return {
+                        link: resolve('https://www.manhuagui.com/', a.attr('href')),
+                        title: a.attr('title'),
+                        pub_date: pDate,
+                        num: a.find('i').text(),
+                        guid: resolve('https://www.manhuagui.com/', a.attr('href')),
+                        category: categoryName,
+                    };
+                });
+        })
+        .reduce((acc, curr) => acc.concat(curr));
+};
 module.exports = async (ctx) => {
     const { id } = ctx.params;
-
+    const chapterCnt = Number(ctx.params.chapterCnt || 0);
     const { data } = await got.get(`https://www.manhuagui.com/comic/${id}/`);
     const $ = cheerio.load(data);
 
@@ -30,24 +54,46 @@ module.exports = async (ctx) => {
             $('#__VIEWSTATE').remove();
         }
     }
-
     const bookTitle = $('.book-title > h1').text();
     const bookIntro = $('#intro-all').text();
     const coverImgSrc = $('.book-cover img').attr('src');
-    const chapters = getChapters($);
-
-    const genResult = (chapter) => ({
-        link: chapter.link,
-        title: chapter.title,
-        description: `
+    // 对最新更新的章节增加了pubDate
+    const reg = new RegExp('最近于.+更新至');
+    // 处理已下架的漫画
+    if ($('.status > span').text().indexOf('已下架') > 0) {
+        ctx.state.data = {
+            title: `看漫画 - ${bookTitle} 已下架`,
+            link: `https://www.manhuagui.com/comic/${id}/`,
+            description: bookIntro,
+            item: [{ link: `https://www.manhuagui.com/comic/${id}/`, title: bookTitle, description: '已下架' }],
+        };
+    } else {
+        const pub_date_str = $('.status > span').text().match(reg)[0].replace('最近于 [', '').replace('] 更新至', '');
+        // 为了能在闭包内访问到这个日期而不是每次需要处理这个最近更新日期
+        $.pubDate = new Date(pub_date_str).toUTCString();
+        $.newChapterCnt = 0;
+        const chapters = getChapters($);
+        const genResult = (chapter) => ({
+            link: chapter.link,
+            title: chapter.title,
+            pubDate: chapter.pub_date,
+            guid: chapter.guid,
+            category: chapter.category,
+            description: `
             <h1>${chapter.num}</h1>
-            <img src="${coverImgSrc}" />
+            <img src='${coverImgSrc}' />
         `.trim(),
-    });
-    ctx.state.data = {
-        title: `看漫画 - ${bookTitle}`,
-        link: `https://www.manhuagui.com/comic/${id}/`,
-        description: bookIntro,
-        item: chapters.map(genResult),
-    };
+        });
+        const items = chapters.map(genResult);
+        let itemsLen = items.length;
+        if (chapterCnt > 0) {
+            itemsLen = chapterCnt < $.newChapterCnt ? $.newChapterCnt : chapterCnt;
+        }
+        ctx.state.data = {
+            title: `看漫画 - ${bookTitle}`,
+            link: `https://www.manhuagui.com/comic/${id}/`,
+            description: bookIntro,
+            item: items.slice(0, itemsLen),
+        };
+    }
 };

package/lib/routes/manxiaosi/book.js

@@ -8,7 +8,7 @@ module.exports = async (ctx) => {
     const url = host + `/book/${id}`;
     const response = await got.get(url);
     const $ = cheerio.load(response.data);
-    const list = $('#detail-list-select > li > a').reverse().slice(0, 10).get();
+    const list = $('#detail-list-select > li > a').get().reverse().slice(0, 10);
 
     const items = await Promise.all(
         list.map(async (item) => {

package/lib/routes/maoyan/hotComplete.js

@@ -0,0 +1,87 @@
+const got = require('@/utils/got');
+
+function sleep(time) {
+    return new Promise((resolve) => setTimeout(resolve, time));
+}
+
+module.exports = async (ctx) => {
+    const orderby = ctx.params.orderby;
+    const ascOrDesc = ctx.params.ascOrDesc;
+    const top = ctx.params.top;
+
+    const response = await got({
+        method: 'get',
+        url: 'https://m.maoyan.com/ajax/movieOnInfoList',
+        headers: {
+            Referer: 'https://m.maoyan.com/',
+        },
+    });
+    let data = response.data.movieList;
+    const movieIds = response.data.movieIds;
+
+    const movieIdsArr = [];
+    for (let i = data.length; i < movieIds.length; i += 10) {
+        const movieIdsTemp = movieIds.slice(i, i + 10);
+        movieIdsArr.push(movieIdsTemp);
+    }
+
+    const dataArrs = await Promise.all(
+        movieIdsArr.map(async (idsTemp) => {
+            await sleep(500);
+            const responseTemp = await got({
+                method: 'get',
+                url: 'https://m.maoyan.com/ajax/moreComingList',
+                searchParams: {
+                    movieIds: idsTemp.toString(),
+                    token: '',
+                },
+                headers: {
+                    Referer: 'https://m.maoyan.com/',
+                },
+            });
+            const dataTemp = responseTemp.data.coming;
+            if (dataTemp && dataTemp.length > 0) {
+                return dataTemp;
+            } else {
+                return [];
+            }
+        })
+    );
+
+    for (let i = 0; i < dataArrs.length; i++) {
+        data.push(...dataArrs[i]);
+    }
+
+    if (orderby) {
+        data.sort((a, b) => {
+            if (orderby === 'pubDate') {
+                return ascOrDesc === 'asc' ? new Date(a.rt).getTime() - new Date(b.rt).getTime() : new Date(b.rt).getTime() - new Date(a.rt).getTime();
+            } else {
+                return ascOrDesc === 'asc' ? a.sc - b.sc : b.sc - a.sc;
+            }
+        });
+    }
+    if (top && top > 0 && top < data.length) {
+        data = data.slice(0, top - 1);
+    }
+
+    const items = await Promise.all(
+        data.map((item) => {
+            const rating = item.sc > 0 ? `评分：${item.sc}` : '';
+
+            return {
+                title: `${item.nm} ${rating}`,
+                description: `<img src="${item.img.replace('w.h', '1000.1000')}"> <br> ${rating} <br> 演员：${item.star} <br> 上映信息：${item.showInfo}`,
+                link: `https://maoyan.com/films/${item.id}`,
+                pubDate: new Date(item.rt).toUTCString(),
+            };
+        })
+    );
+
+    ctx.state.data = {
+        title: `猫眼电影 - 正在热映 - 完整版`,
+        link: `https://maoyan.com/films`,
+        description: `猫眼电影 - 正在热映`,
+        item: items,
+    };
+};