rsshub 1.0.0-master.f6ab7b6 → 1.0.0-master.f6b1d4c

package/lib/v2/huxiu/utils.js

@@ -1,69 +1,154 @@
 const cheerio = require('cheerio');
-const date = require('@/utils/date');
+const { parseDate } = require('@/utils/parse-date');
 const got = require('@/utils/got');
+const { art } = require('@/utils/render');
+const path = require('path');
+const CryptoJS = require('crypto-js');
 
-const ProcessFeed = async (list, cache) => {
-    const host = 'https://www.huxiu.com';
+const accountApi = 'https://account-api.huxiu.com';
+const articleApi = 'https://api-article.huxiu.com';
+const baseUrl = 'https://www.huxiu.com';
+const briefApi = 'https://api-brief.huxiu.com';
+const momentApi = 'https://moment-api.huxiu.com';
+const searchApi = 'https://search-api.huxiu.com';
 
-    const items = await Promise.all(
-        list.map(async (e) => {
-            const link = new URL(e, host).href;
+const ProcessFeed = (list, cache) =>
+    Promise.all(
+        list.map((item) =>
+            cache.tryGet(item.link, async () => {
+                const response = await got(item.link, {
+                    https: {
+                        rejectUnauthorized: false,
+                    },
+                });
+                const $ = cheerio.load(response.data);
+                const initialState = parseInitialState($);
 
-            const single = await cache.tryGet(link, async () => {
-                const response = await got(link);
+                if (item.link.startsWith(`${baseUrl}/article/`)) {
+                    fetchArticle(initialState, item);
+                } else if (item.link.startsWith(`${baseUrl}/brief/`)) {
+                    fetchBrief(initialState, item);
+                }
 
-                const $ = cheerio.load(response.data);
+                return item;
+            })
+        )
+    );
 
-                // const __INITIAL_STATE__ = Function(
-                //     'let window={};' +
-                //         $('*')
-                //             .html()
-                //             .match(/window\.__INITIAL_STATE__.*};/gm)[0] +
-                //         'return window.__INITIAL_STATE__'
-                // )();
-
-                $('.article__content')
-                    .find('.lazyImg')
-                    .each((i, e) => {
-                        $(e).attr('src', $(e).attr('_src'));
-                    });
-
-                $('.article__content')
-                    .find('.text-big-title')
-                    .each((i, e) => (e.tagName = 'h3'));
-
-                $('p').each(function () {
-                    if ($(this).find('img').length === 0 && $(this).text().match(/^\s*$/)) {
-                        $(this).remove();
-                    }
-                });
-                $('img.dialog_add_wxxy_qrcode_icon').remove();
+const fetchArticle = (initialState, item) => {
+    const { articleDetail } = initialState.articleDetail;
+    const topImg = art(path.join(__dirname, 'templates/img.art'), {
+        img: articleDetail.pic_path.split('?')[0],
+    });
+    let video = '';
+    if (articleDetail.video_info) {
+        video = art(path.join(__dirname, 'templates/video.art'), {
+            video_info: articleDetail.video_info,
+        });
+    }
 
-                let description = '';
+    const $ = cheerio.load(articleDetail.content, null, false);
 
-                // if ('video_info' in __INITIAL_STATE__.articleDetail.articleDetail) {
-                //     description += `<video height="100%" poster="${__INITIAL_STATE__.articleDetail.articleDetail.video_info.cover}" controls src="${__INITIAL_STATE__.articleDetail.articleDetail.video_info.hd_link}"></video>`;
-                // }
-                if ($('.top-img').html() !== null) {
-                    description += $('.top-img').html();
-                }
+    $('.text-big-title').each((_, e) => (e.tagName = 'h3'));
+    fixImg($);
+    removeExtraLinebreak($);
+    // $('img.dialog_add_wxxy_qrcode_icon').remove();
+
+    item.description = video + topImg + $.html();
+    item.author = articleDetail.author;
+    item.category = articleDetail.tags_info.map((tag) => tag.name);
+};
+
+const fetchBrief = (initialState, item) => {
+    const { brief_detail } = initialState.briefStoreModule;
+    const { brief_column } = brief_detail;
+
+    const description = art(path.join(__dirname, 'templates/brief.art'), {
+        brief: brief_detail.brief,
+    });
 
-                return {
-                    title: $('.article__title').text().trim(),
-                    description: description + $('.article__content').html(),
-                    pubDate: date($('.article__time').text(), 8),
-                    author: $('.author-info__username').text(),
-                    link,
-                };
-            });
-
-            return single;
-        })
+    const $ = cheerio.load(description, null, false);
+
+    $('button.black-dot').remove();
+    fixImg($);
+    removeExtraLinebreak($);
+
+    item.description = $.html();
+    item.author = brief_detail.brief.publisher_list.map((item) => item.username).join(', ');
+    item.pubDate = parseDate(brief_column.update_time, 'X');
+};
+
+const parseInitialState = ($) =>
+    JSON.parse(
+        $('script')
+            .text()
+            .match(/window\.__INITIAL_STATE__=(\{.*?\});\(function\(\)/)[1]
     );
 
-    return items;
+const fixImg = ($) => {
+    $('img.lazyImg, img.js-preview').each((_, e) => {
+        if (e.attribs._src) {
+            e.attribs.src = e.attribs._src.split('?')[0];
+            delete e.attribs._src;
+        }
+        if (e.attribs.src.includes('?')) {
+            e.attribs.src = e.attribs.src.split('?')[0];
+        }
+        if (e.attribs['data-w']) {
+            e.attribs.width = e.attribs['data-w'];
+            delete e.attribs['data-w'];
+        }
+        if (e.attribs['data-h']) {
+            e.attribs.height = e.attribs['data-h'];
+            delete e.attribs['data-h'];
+        }
+    });
+};
+
+const removeExtraLinebreak = ($) => {
+    $('p').each((_, e) => {
+        e = $(e);
+        if (e.find('img').length === 0 && e.text().match(/^\s*$/)) {
+            e.remove();
+        }
+    });
+};
+
+const generateNonce = () => {
+    let nonce = '';
+    const e = 'abcdefghijklmnopqrstuvwxyz1234567890ABCDEFGHIJKLMNOPQRSTUVWXYZ';
+    const t = 16;
+    for (let i = 0; i < t; i++) {
+        nonce += e.charAt(Math.floor(Math.random() * e.length));
+    }
+    return nonce;
+};
+
+const generateSignature = () => {
+    const timestamp = Math.round(new Date().getTime() / 1000).toString();
+    // const appSecret = '4nHzGgGt7WX4zFTsTKocpxg4dzU6-wUi';
+    const appSecret = 'hUzaVKtNfDE-6UiyaJdfsmjW-8dwoyVc'; // appid: 'hx_search'
+    const nonce = generateNonce();
+    const r = [appSecret, timestamp, nonce].sort();
+    return {
+        nonce,
+        timestamp,
+        signature: CryptoJS.SHA1(r[0] + r[1] + r[2]).toString(),
+    };
 };
 
 module.exports = {
+    accountApi,
+    articleApi,
+    baseUrl,
+    briefApi,
+    momentApi,
+    searchApi,
     ProcessFeed,
+    fetchArticle,
+    fetchBrief,
+    parseInitialState,
+    fixImg,
+    removeExtraLinebreak,
+    generateSignature,
 };

package/lib/v2/i-cable/category.js

@@ -0,0 +1,74 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const { art } = require('@/utils/render');
+const { join } = require('path');
+
+module.exports = async (ctx) => {
+    const { category } = ctx.params;
+    const { limit = '100' } = ctx.query;
+    const baseUrl = 'https://www.i-cable.com';
+
+    let link;
+    let id;
+    if (category) {
+        const categoryDetail = await ctx.cache.tryGet(`i-cable:category:${category}`, async () => {
+            const { data: res } = await got(`${baseUrl}/wp-json/wp/v2/categories`, {
+                searchParams: {
+                    search: category,
+                },
+            });
+            const data = res[0];
+            if (!data) {
+                throw Error('Invalid category');
+            }
+            return {
+                id: data.id,
+                link: data.link,
+            };
+        });
+        id = categoryDetail.id;
+        link = categoryDetail.link;
+    }
+
+    const { data: response } = await got(`${baseUrl}/wp-json/wp/v2/posts`, {
+        searchParams: {
+            per_page: limit,
+            categories: id,
+            _embed: 1,
+        },
+    });
+
+    const items = response.map((item) => {
+        const $ = cheerio.load(item.content.rendered, null, false);
+        $('img').each((_, img) => {
+            img = $(img);
+            const src = img.attr('src');
+            if (src.match(/-\d+x\d+.(jpg|png)$/)) {
+                img.attr('src', src.replace(/-\d+x\d+.(jpg|png)$/, '.$1'));
+            }
+            img.removeAttr('srcset');
+        });
+        return {
+            title: item.title.rendered,
+            description: art(join(__dirname, 'templates/desc.art'), {
+                content: $.html(),
+                imgs: item.yoast_head_json.og_image.map((img) => img.url).filter((i) => !i.endsWith('Open-Graph-2.png')),
+            }),
+            pubDate: parseDate(item.date_gmt), // 2023-05-11T15:07:51
+            updated: parseDate(item.modified_gmt),
+            guid: item.guid.rendered,
+            link: item.link,
+            author: item._embedded.author.map((author) => author.name).join(', '),
+            category: item._embedded['wp:term'][0].map((category) => category.name),
+        };
+    });
+
+    ctx.state.data = {
+        title: `${category || '主頁'} | 有線寬頻 i-CABLE`,
+        description: '24小時無間斷緊貼香港本地及國際新聞，即時報道新聞事件的最新發展。內容包括時事、財經、體育、天氣消息等多元化資訊。',
+        image: 'https://icable-prod.s3.ap-southeast-1.amazonaws.com/wp-content/uploads/2022/05/19010704/Open-Graph-1.png',
+        link: link || baseUrl,
+        item: items,
+    };
+};

package/lib/v2/i-cable/maintainer.js

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

package/lib/v2/i-cable/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'i-cable.com': {
+        _name: '有線寬頻 i-CABLE',
+        '.': [
+            {
+                title: '有線新聞 | Cable News',
+                docs: 'https://docs.rsshub.app/traditional-media.html#you-xian-kuan-pin-i-cable',
+                source: ['/category/*path', '/'],
+                target: (params) => `/i-cable${params.path ? decodeURIComponent(params.path.slice(params.path.lastIndexOf('/'))) : ''}`,
+            },
+        ],
+    },
+};

package/lib/v2/i-cable/router.js

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

package/lib/v2/i-cable/templates/desc.art

@@ -0,0 +1,8 @@
+{{ if imgs }}
+    {{ each imgs img }}
+        <img src="{{ img }}"><br>
+    {{ /each }}
+{{ /if }}
+{{ if content }}
+    {{@ content }}
+{{ /if }}

package/lib/v2/ibc/maintainer.js

@@ -0,0 +1,4 @@
+module.exports = {
+    '/maitta': ['fengkx'],
+    '/radio/:id?': ['nczitzk'],
+};

package/lib/v2/ibc/maitta.js

@@ -0,0 +1,32 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const url = 'https://www.ibc.co.jp/radio/maitta/audio/';
+const { parseDate } = require('@/utils/parse-date');
+
+module.exports = async (ctx) => {
+    const res = await got(url);
+    const $ = cheerio.load(res.data);
+    const items = $('.up').toArray();
+    ctx.state.data = {
+        title: 'IBCラジオ　イヤーマイッタマイッタ｜IBC岩手放送',
+        link: 'http://www.ibc.co.jp/radio/maitta/audio',
+        description: $('meta[name=description]').attr('content'),
+        itunes_author: '水越アナと大塚アナ',
+        image: 'https://cdn.ibc.co.jp/radio/maitta/audio/images/header.png',
+        language: 'ja',
+        item: items.map((item) => {
+            item = $(item);
+            const link = new URL(item.find('a').attr('href'), url).href;
+            const uid = link.match(/\/(\d+)\/$/)[1];
+            return {
+                title: item.find('h3').text(),
+                description: item.find('.audio_intro').text().trim(),
+                link,
+                pubDate: parseDate(item.find('.onair_date').text().split('日')[0].replace(/年|月/g, '-')),
+                itunes_item_image: 'https://assets.blubrry.com/coverart/300/357155.png',
+                enclosure_url: `https://media-data.cdn.ibc.co.jp/out/sound/28_${uid}/28_${uid}.mp3`, // 音频链接
+                enclosure_type: 'audio/mpeg',
+            };
+        }),
+    };
+};

package/lib/v2/ibc/radar.js

@@ -0,0 +1,19 @@
+module.exports = {
+    'ibc.co.jp': {
+        _name: 'IBC 岩手放送',
+        '.': [
+            {
+                title: 'イヤーマイッタマイッタ',
+                docs: 'https://docs.rsshub.app/multimedia.html#ibc-yan-shou-fang-song',
+                source: ['/radio/maitta/audio', '/'],
+                target: '/ibc/maitta',
+            },
+            {
+                title: 'ラジオ',
+                docs: 'https://docs.rsshub.app/multimedia.html#ibc-yan-shou-fang-song',
+                source: ['/radio/:id/audio', '/'],
+                target: '/ibc/radio/:id?',
+            },
+        ],
+    },
+};

package/lib/v2/ibc/radio.js

@@ -0,0 +1,73 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const timezone = require('@/utils/timezone');
+const { parseDate } = require('@/utils/parse-date');
+const { art } = require('@/utils/render');
+const path = require('path');
+
+module.exports = async (ctx) => {
+    const id = ctx.params.id ?? 'maitta';
+    const limit = ctx.query.limit ? parseInt(ctx.query.limit) : 150;
+
+    const rootUrl = 'https://www.ibc.co.jp';
+    const currentUrl = `${rootUrl}/radio/${id}/audio/`;
+
+    let response = await got({
+        method: 'get',
+        url: currentUrl,
+    });
+
+    let $ = cheerio.load(response.data);
+
+    const image = $('meta[property="og:image"]').attr('content');
+
+    let radioId = $('link[title="RSS"]').attr('href')?.split('/rss/').pop().replace(/\/$/, '') ?? '';
+
+    let items = $('.audio_list section').toArray();
+
+    if (!radioId) {
+        const item = $(items[0]);
+
+        response = await got({
+            method: 'get',
+            url: new URL(item.find('a').attr('href'), rootUrl).href,
+        });
+
+        $ = cheerio.load(response.data);
+
+        radioId = $('audio')
+            .attr('src')
+            .match(/\/sound\/(\d+)_/)[1];
+    }
+
+    items = items.slice(0, limit).map((item) => {
+        item = $(item);
+
+        const link = new URL(item.find('a').attr('href'), rootUrl).href;
+        const audioId = link.match(/\/aitem\/(\d+)/)[1];
+        const audio = `https://media-data.cdn.ibc.co.jp/out/sound/${radioId}_${audioId}/${radioId}_${audioId}.mp3`;
+
+        return {
+            link,
+            title: item.find('h3').text(),
+            pubDate: timezone(parseDate(item.find('.update_date').text(), 'YYYY年MM月DD日'), +9),
+            enclosure_url: audio,
+            enclosure_type: 'audio/mpeg',
+            itunes_item_image: image,
+            description: art(path.join(__dirname, 'templates/description.art'), {
+                description: item.find('.audio_intro').text(),
+                audio,
+            }),
+        };
+    });
+
+    ctx.state.data = {
+        title: $('title').text() || `IBC岩手放送｜${$('.programinfo h3').text()}`,
+        link: currentUrl,
+        item: items,
+        description: $('meta[property="og:description"]').attr('content'),
+        itunes_author: 'IBC岩手放送',
+        image,
+        allowEmpty: true,
+    };
+};

package/lib/v2/ibc/router.js

@@ -0,0 +1,4 @@
+module.exports = (router) => {
+    router.get('/maitta', require('./maitta'));
+    router.get('/radio/:id?', require('./radio'));
+};

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

@@ -0,0 +1,6 @@
+{{ if description }}
+<p>{{ description }}</p>
+{{ /if }}
+{{ if audio }}
+<audio controls src="{{ audio }}"></audio>
+{{ /if }}

package/lib/v2/icac/radar.js

@@ -1,6 +1,6 @@
 module.exports = {
-    'https://www.icac.org.hk': {
-        _name: '廉政公署',
+    'icac.org.hk': {
+        _name: '香港廉政公署',
         '.': [
             {
                 title: '新闻公布',

package/lib/v2/idolypride/maintainer.js

@@ -0,0 +1,3 @@
+module.exports = {
+    '/news': ['Mingxia1'],
+};

package/lib/v2/idolypride/news.js

@@ -0,0 +1,29 @@
+const got = require('@/utils/got');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+
+module.exports = async (ctx) => {
+    const response = await got({
+        method: 'get',
+        url: 'https://idolypride.jp/wp-json/wp/v2/news',
+    });
+    const list = response.data;
+
+    ctx.state.data = {
+        title: '偶像荣耀-新闻',
+        link: 'https://idolypride.jp/news',
+        item: list.map((item) => {
+            const title = item.title.rendered;
+            const link = item.link;
+            const pubDate = timezone(parseDate(item.date_gmt), 0);
+            const rendered = item.content.rendered;
+
+            return {
+                title,
+                link,
+                pubDate,
+                description: rendered,
+            };
+        }),
+    };
+};

package/lib/v2/idolypride/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'idolypride.jp': {
+        _name: '偶像荣耀',
+        '.': [
+            {
+                title: '新闻',
+                docs: 'https://docs.rsshub.app/anime.html#ou-xiang-rong-yao',
+                source: '/news',
+                target: '/idolypride/news',
+            },
+        ],
+    },
+};

package/lib/v2/idolypride/router.js

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

package/lib/v2/ieee/earlyaccess.js

@@ -0,0 +1,77 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const path = require('path');
+const { art } = require('@/utils/render');
+
+const { CookieJar } = require('tough-cookie');
+const cookieJar = new CookieJar();
+
+module.exports = async (ctx) => {
+    const isnumber = ctx.params.journal;
+    const sortType = ctx.params.sortType ?? 'vol-only-seq';
+    const host = 'https://ieeexplore.ieee.org';
+    const jrnlUrl = `${host}/xpl/tocresult.jsp?isnumber=${isnumber}`;
+
+    const response = await got(`${host}/rest/publication/home/metadata?issueid=${isnumber}`, {
+        cookieJar,
+    }).json();
+    const punumber = response.publicationNumber;
+    const volume = response.currentIssue.volume;
+    const jrnlName = response.displayTitle;
+
+    const response2 = await got
+        .post(`${host}/rest/search/pub/${punumber}/issue/${isnumber}/toc`, {
+            cookieJar,
+            json: {
+                punumber,
+                isnumber,
+                sortType,
+                rowsPerPage: '100',
+            },
+        })
+        .json();
+    let list = response2.records.map((item) => {
+        const $2 = cheerio.load(item.articleTitle);
+        const title = $2.text();
+        const link = item.htmlLink;
+        const doi = item.doi;
+        let authors = 'Do not have author';
+        if (item.hasOwnProperty('authors')) {
+            authors = item.authors.map((itemAuth) => itemAuth.preferredName).join('; ');
+        }
+        const abstract = item.hasOwnProperty('abstract') ? item.abstract : '';
+        return {
+            title,
+            link,
+            authors,
+            doi,
+            volume,
+            abstract,
+        };
+    });
+
+    const renderDesc = (item) =>
+        art(path.join(__dirname, 'templates/description.art'), {
+            item,
+        });
+    list = await Promise.all(
+        list.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                if (item.abstract !== '') {
+                    const response3 = await got(`${host}${item.link}`);
+                    const { abstract } = JSON.parse(response3.body.match(/metadata=(.*);/)[1]);
+                    const $3 = cheerio.load(abstract);
+                    item.abstract = $3.text();
+                    item.description = renderDesc(item);
+                }
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: jrnlName,
+        link: jrnlUrl,
+        item: list,
+    };
+};

package/lib/v2/ieee/maintainer.js

@@ -1,4 +1,5 @@
 module.exports = {
     '/journal/:journal/:sortType?': ['Derekmini', 'auto-bot-ty'],
     '/journal/:journal/recent/:sortType?': ['Derekmini', 'auto-bot-ty'],
+    '/journal/:journal/earlyaccess/:sortType?': ['5upernova-heng'],
 };

package/lib/v2/ieee/radar.js

@@ -14,6 +14,12 @@ module.exports = {
                 source: '/*',
                 target: (params, url) => `/ieee/journal/${new URL(url).searchParams.get('punumber')}/recent`,
             },
+            {
+                title: 'Early Access Journal',
+                docs: 'https://docs.rsshub.app/journal.html#ieee-xplore',
+                source: '/*',
+                target: (params, url) => `/ieee/journal/${new URL(url).searchParams.get('isnumber')}/earlyaccess`,
+            },
         ],
     },
 };

package/lib/v2/ieee/router.js

@@ -1,6 +1,7 @@
 module.exports = function (router) {
     router.get('/:journal/latest/vol/:sortType?', require('./journal')); // deprecated
     router.get('/:journal/latest/date/:sortType?', require('./recent')); // deprecated
-    router.get('/journal/:journal/:sortType?', require('./journal'));
+    router.get('/journal/:journal/earlyaccess/:sortType?', require('./earlyaccess'));
     router.get('/journal/:journal/recent/:sortType?', require('./recent'));
+    router.get('/journal/:journal/:sortType?', require('./journal'));
 };

package/lib/v2/ieee-security/maintainer.js

@@ -0,0 +1,3 @@
+module.exports = {
+    '/security-privacy': ['ZeddYu'],
+};

package/lib/v2/ieee-security/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'ieee-security.org': {
+        _name: 'IEEE Computer Society',
+        '.': [
+            {
+                title: 'IEEE Symposium on Security and Privacy',
+                docs: 'https://docs.rsshub.app/journal.html#ieee-computer-society',
+                source: ['/TC/SP-Index.html', '/'],
+                target: '/ieee-security/security-privacy',
+            },
+        ],
+    },
+};

package/lib/v2/ieee-security/router.js

@@ -0,0 +1,3 @@
+module.exports = (router) => {
+    router.get('/security-privacy', require('./sp'));
+};