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

package/lib/v2/hizu/index.js

@@ -0,0 +1,76 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+
+const titles = {
+    '5dd92265e4b0bf88dd8c1175': '热点',
+    '5dd921a7e4b0bf88dd8c116f': '订阅',
+    '604f1cbbe4b0cf5c2234d470': '学党史',
+    '5dd92242e4b0bf88dd8c1174': '政经',
+    '61259fd6e4b0d294f7f9786d': '合作区',
+    '61dfe511e4b0248b60d1c568': '名记名播',
+    '5dd9222ce4b0bf88dd8c1173': '大湾区',
+    '617805e4e4b037abacfd4820': '网评',
+    '5dd9220de4b0bf88dd8c1172': 'TV新闻',
+    '5e6edd50e4b02ebde0ab061e': '音频',
+    '600e8ad4e4b02c3a6af6aaa8': '澳门',
+    '600f760fe4b0e33cf6f8e68e': '政务',
+    '5ff7c0fde4b0e2f210d05e20': '教育',
+    '5fc88615e4b0e3055e693e0a': '深圳',
+    '600e8a93e4b02c3a6af6aa80': '中山',
+    '5dd921ece4b0bf88dd8c1170': '民生',
+    '61148184e4b08d3215364396': '社区',
+    '5dd9215fe4b0bf88dd8c116b': '专题',
+    '5e2e5107e4b0c14b5d0e3d04': '战疫',
+    '5f88eaf2e4b0a27cd404e09e': '横琴',
+    '5f86a3f5e4b09d75f99dde7d': '香洲',
+    '5e8c42b4e4b0347c7e5836e0': '金湾',
+    '5ee70534e4b07b8a779a1ad6': '斗门',
+    '607d37ade4b05c59ac2f3d40': '高新',
+};
+
+module.exports = async (ctx) => {
+    const column = ctx.params.column ?? '5dd92265e4b0bf88dd8c1175';
+
+    const rootUrl = 'https://www.hizh.cn';
+    const currentUrl = `${rootUrl}/channels/zjyapp/columns/${column}/stories.json`;
+
+    const response = await got({
+        method: 'get',
+        url: currentUrl,
+    });
+
+    let items = response.data.stories.slice(0, ctx.query.limit ? parseInt(ctx.query.limit) : 50).map((item) => ({
+        title: item.title,
+        author: item.author,
+        link: item.jsonUrl,
+        category: item.columnName,
+        pubDate: parseDate(item.published),
+    }));
+
+    items = await Promise.all(
+        items.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const detailResponse = await got({
+                    method: 'get',
+                    url: item.link,
+                });
+
+                const content = cheerio.load(detailResponse.data.content);
+
+                content('figure').last().remove();
+
+                item.description = content.html();
+                item.link = `${rootUrl}/content.html?jsonUrl=${item.link}`;
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: `${titles[column]} - 珠海网`,
+        link: rootUrl,
+        item: items,
+    };
+};

package/lib/v2/hizu/maintainer.js

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

package/lib/v2/hizu/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'hizh.cn': {
+        _name: '珠海网',
+        '.': [
+            {
+                title: '栏目',
+                docs: 'https://docs.rsshub.app/new-media.html#zhu-hai-wang-lan-mu',
+                source: '/',
+                target: '/hizu/:column?',
+            },
+        ],
+    },
+};

package/lib/v2/hizu/router.js

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

package/lib/v2/hkej/index.js

@@ -0,0 +1,154 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate, parseRelativeDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+const { art } = require('@/utils/render');
+const path = require('path');
+const { CookieJar } = require('tough-cookie');
+
+const cookieJar = new CookieJar();
+
+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';
+
+    const response = await got({
+        method: 'get',
+        url: baseUrl + cat.link,
+        headers: {
+            Referer: baseUrl,
+        },
+        cookieJar,
+    });
+
+    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 renderArticleImg = (pics) =>
+        art(path.join(__dirname, 'templates/articleImg.art'), {
+            pics,
+        });
+
+    const renderDesc = (pics, desc) =>
+        art(path.join(__dirname, 'templates/description.art'), {
+            pics: renderArticleImg(pics),
+            desc,
+        });
+
+    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,
+                        },
+                        cookieJar,
+                    });
+                    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
+                    const articleImg = (content('div.hkej_detail_thumb_2014 td a').length ? content('div.hkej_detail_thumb_2014 td a') : content('div.thumb td a'))
+                        .map((_, e) => {
+                            e = $(e);
+                            return {
+                                href: e.attr('href'),
+                                title: e.attr('title'),
+                            };
+                        })
+                        .get();
+
+                    const pubDate = content('p.info span.date').text().trim();
+
+                    item.category = content('p.info span.cate a')
+                        .toArray()
+                        .map((e) => content(e).text().trim());
+                    item.description = renderDesc(articleImg, content('div#article-content').html());
+                    item.pubDate = timezone(/(今|昨)/.test(pubDate) ? parseRelativeDate(pubDate) : parseDate(pubDate, 'YYYY M D'), +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',
+    };
+
+    ctx.state.json = {
+        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',
+        cookieJar,
+    };
+};

package/lib/v2/hkej/maintainer.js

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

package/lib/v2/hkej/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'hkej.com': {
+        _name: '信报财经新闻',
+        '.': [
+            {
+                title: '即时新闻',
+                docs: 'https://docs.rsshub.app/traditional-media.html##xin-bao-cai-jing-xin-wen',
+                source: ['/'],
+                target: '/hkej/:category?',
+            },
+        ],
+    },
+};

package/lib/v2/hkej/router.js

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

package/lib/v2/hkej/templates/articleImg.art

@@ -0,0 +1,3 @@
+{{ each pics }}
+    <figure><img src="{{ $value.href }}" alt="{{ $value.title }}"><figcaption>{{ $value.title }}</figcaption></figure>
+{{ /each }}

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

@@ -0,0 +1,2 @@
+{{@ pics }}
+{{@ desc }}

package/lib/v2/hkepc/index.js

@@ -0,0 +1,206 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+const path = require('path');
+const { art } = require('@/utils/render');
+
+module.exports = async (ctx) => {
+    const category = ctx.params.category ?? '';
+    const baseUrl = `https://www.hkepc.com`;
+
+    let response;
+    if (category === 'price' || category === 'review' || category === 'coverStory' || category === 'news' || category === 'press' || category === 'member') {
+        response = await got({
+            url: `${baseUrl}/news`,
+            headers: {
+                Referer: baseUrl,
+            },
+        });
+    } else if (category === 'digital' || category === 'entertainment' || category === 'latest' || category === '') {
+        response = await got(baseUrl);
+    } else {
+        // category === ocLab
+        response = await got({
+            url: `${baseUrl}/${category}`,
+            headers: {
+                Referer: baseUrl,
+            },
+        });
+    }
+
+    const $ = cheerio.load(response.data);
+
+    let feedTitle = '電腦領域 HKEPC';
+    let items;
+    switch (category) {
+        case 'price':
+            feedTitle += ' - 腦場新聞';
+            items = $('#sidebar > div:nth-child(1) > div.content > ul > li')
+                .find('a')
+                .toArray()
+                .map((item) => ({
+                    title: $(item).text(),
+                    link: baseUrl + $(item).attr('href'),
+                }));
+            break;
+        case 'review':
+            feedTitle += ' - 新品快遞';
+            items = $('#sidebar > div:nth-child(2) > div.content > ul > li')
+                .find('a')
+                .toArray()
+                .map((item) => ({
+                    title: $(item).text(),
+                    link: baseUrl + $(item).attr('href'),
+                }));
+            break;
+        case 'coverStory':
+            feedTitle += ' - 專題報導';
+            items = $('#sidebar > div:nth-child(3) > div.content > ul > li')
+                .find('a')
+                .toArray()
+                .map((item) => ({
+                    title: $(item).text(),
+                    link: baseUrl + $(item).attr('href'),
+                }));
+            break;
+        case 'news':
+            feedTitle += ' - 新聞中心';
+            items = $('#sidebar > div:nth-child(4) > div.content > ul > li')
+                .find('a')
+                .toArray()
+                .map((item) => ({
+                    title: $(item).text(),
+                    link: baseUrl + $(item).attr('href'),
+                }));
+            break;
+        case 'press':
+            feedTitle += ' - 業界資訊';
+            items = $('#sidebar > div:nth-child(5) > div.content > ul > li')
+                .find('a')
+                .toArray()
+                .map((item) => ({
+                    title: $(item).text(),
+                    link: baseUrl + $(item).attr('href'),
+                }));
+            break;
+        case 'member':
+            feedTitle += ' - 會員消息';
+            items = $('#sidebar > div:nth-child(6) > div.content > ul > li')
+                .find('a')
+                .toArray()
+                .map((item) => ({
+                    title: $(item).text(),
+                    link: baseUrl + $(item).attr('href'),
+                }));
+            break;
+        case 'digital':
+            feedTitle += ' - 流動數碼';
+            items = $('#contentR5 > div.left > div.article > div.heading')
+                .find('a')
+                .toArray()
+                .map((item) => ({
+                    title: $(item).text(),
+                    link: baseUrl + $(item).attr('href'),
+                }));
+            break;
+        case 'entertainment':
+            feedTitle += ' - 生活娛樂';
+            items = $('#contentR5 > div.right > div.article > div.heading')
+                .find('a')
+                .toArray()
+                .map((item) => ({
+                    title: $(item).text(),
+                    link: baseUrl + $(item).attr('href'),
+                }));
+            break;
+        case 'latest':
+        case '':
+            feedTitle += ' - 最新消息';
+            items = $('div .item')
+                .find('a')
+                .toArray()
+                .map((item) => ({
+                    title: $(item).text(),
+                    link: baseUrl + $(item).attr('href'),
+                }));
+            break;
+        case 'ocLab':
+            feedTitle += ' - 超頻領域';
+            items = $('.heading')
+                .toArray()
+                .map((item) => ({
+                    title: $(item).text(),
+                    link: baseUrl + $(item).attr('href'),
+                }));
+            break;
+        default:
+            break;
+    }
+
+    items = await Promise.all(
+        items.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const detailResponse = await got({
+                    url: item.link,
+                    headers: {
+                        Referer: baseUrl,
+                    },
+                });
+
+                const content = cheerio.load(detailResponse.data);
+
+                // remove unwanted elements
+                content('#view > div.advertisement').remove();
+                content('div#comments').remove();
+                content('div#share_btn').remove();
+
+                // Non-breaking space U+00A0, ` ` in html
+                // Taken from /caixin/blog.js
+                content('#view > p')
+                    .filter((_, e) => e.children[0].data === String.fromCharCode(160))
+                    .remove();
+
+                // fix lazyload image
+                content('#view > p > img').each((_, e) => {
+                    e = content(e);
+                    e.after(
+                        art(path.join(__dirname, 'templates/image.art'), {
+                            src: e.attr('rel') ?? e.attr('src'),
+                            alt: e.attr('alt'),
+                        })
+                    );
+                    e.remove();
+                });
+
+                item.author = content('.newsAuthor').text().trim() ?? content('#articleHead div.author').text().trim();
+                item.category = content('div#relatedArticles div.tags a')
+                    .toArray()
+                    .map((e) => content(e).text().trim());
+                item.description = art(path.join(__dirname, 'templates/description.art'), {
+                    desc: content('#view').html(),
+                });
+                item.pubDate = timezone(parseDate(content('.publishDate').text()), +8);
+                item.guid = item.link.substring(0, item.link.lastIndexOf('/'));
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: feedTitle,
+        link: `https://www.hkepc.com/${category}`,
+        description: '電腦領域 HKEPC Hardware - 全港 No.1 PC網站',
+        language: 'zh-hk',
+        item: items,
+    };
+
+    ctx.state.json = {
+        title: feedTitle,
+        link: `https://www.hkepc.com/${category}`,
+        description: '電腦領域 HKEPC Hardware - 全港 No.1 PC網站',
+        language: 'zh-hk',
+        item: items,
+    };
+};

package/lib/v2/hkepc/maintainer.js

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

package/lib/v2/hkepc/radar.js

@@ -0,0 +1,19 @@
+module.exports = {
+    'hkepc.com': {
+        _name: 'HKEPC',
+        '.': [
+            {
+                title: 'HKEPC 电脑领域',
+                docs: 'https://docs.rsshub.app/new-media.html#hkepc',
+                source: ['/:category?'],
+                target: '/hkepc/:category?',
+            },
+            {
+                title: 'HKEPC 电脑领域',
+                docs: 'https://docs.rsshub.app/new-media.html#hkepc',
+                source: ['/'],
+                target: '/hkepc',
+            },
+        ],
+    },
+};

package/lib/v2/hkepc/router.js

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

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

@@ -0,0 +1 @@
+{{@ desc }}

package/lib/v2/hkepc/templates/image.art

@@ -0,0 +1 @@
+<img src="{{ src }}" alt="{{ alt }}">

package/lib/v2/hket/index.js

@@ -0,0 +1,122 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+const path = require('path');
+const { art } = require('@/utils/render');
+
+const urlMap = {
+    srac: {
+        baseUrl: 'https://china.hket.com',
+    },
+    sran: {
+        baseUrl: 'https://inews.hket.com',
+    },
+    srat: {
+        baseUrl: 'https://topick.hket.com',
+    },
+    sraw: {
+        baseUrl: 'https://wealth.hket.com',
+    },
+};
+
+module.exports = async (ctx) => {
+    const category = ctx.params.category ?? 'sran001';
+    const baseUrl = urlMap[category.substring(0, 4)].baseUrl;
+
+    const response = await got({
+        method: 'get',
+        url: baseUrl + '/' + category,
+        headers: {
+            Referer: 'https://www.hket.com',
+        },
+    });
+
+    const $ = cheerio.load(response.data);
+
+    const list = $('div.listing-title > a')
+        .map((_, item) => {
+            item = $(item);
+            return {
+                title: item.text().trim(),
+                link: item.attr('href').startsWith('/')
+                    ? // remove tracking parameters
+                      baseUrl + item.attr('href').split('?')[0].substring(0, item.attr('href').lastIndexOf('/'))
+                    : item.attr('href').split('?')[0].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: baseUrl + '/' + category,
+                        },
+                    });
+                    const content = cheerio.load(article.data);
+
+                    // extract categories
+                    const categories = content('.contentTags-container > .hotkey-container-wrapper > .hotkey-container > a')
+                        .toArray()
+                        .map((e) => content(e).text().trim());
+
+                    // remove unwanted elements
+                    content('p.adunit').remove();
+                    content('p.article-detail_caption').remove();
+                    content('div.template-default.hket-row.detail-widget.show-xs-img.relatedContents-container').remove();
+                    content('div.template-default.hket-row.no-padding.detail-widget').remove();
+                    content('div.gallery-related-container').remove();
+                    content('div.contentTags-container').remove();
+                    content('div.article-details-center-sharing-btn').remove();
+                    content('source').remove();
+                    content('span.click-to-enlarge').remove();
+                    content('span').each((_, e) => {
+                        if (content(e).text().startsWith('+')) {
+                            content(e).remove();
+                        }
+                    });
+
+                    // fix lazyload image and caption
+                    content('img').each((_, e) => {
+                        e = content(e);
+                        e.after(
+                            art(path.join(__dirname, 'templates/image.art'), {
+                                alt: e.attr('data-alt'),
+                                src: e.attr('data-src') ?? e.attr('src'),
+                            })
+                        );
+                        e.remove();
+                    });
+
+                    item.category = categories;
+                    item.description = art(path.join(__dirname, 'templates/description.art'), {
+                        desc: content('div.article-detail-body-container').html(),
+                    });
+                    item.pubDate = timezone(parseDate(content('.article-details-info-container_date').text().trim()), +8);
+
+                    return item;
+                })
+            )
+    );
+
+    ctx.state.data = {
+        title: $('head meta[name=title]').attr('content').trim(),
+        link: baseUrl + '/' + category,
+        description: $('head meta[name=description]').attr('content').trim(),
+        item: items,
+        language: 'zh-hk',
+    };
+
+    ctx.state.json = {
+        title: $('head meta[name=title]').attr('content').trim(),
+        link: baseUrl + '/' + category,
+        description: $('head meta[name=description]').attr('content').trim(),
+        item: items,
+        language: 'zh-hk',
+    };
+};

package/lib/v2/hket/maintainer.js

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