rsshub 1.0.0-master.f9ec483 → 1.0.0-master.fa04a11

package/lib/routes/joneslanglasalle/index.ts

@@ -0,0 +1,311 @@
+import path from 'node:path';
+
+import { type CheerioAPI, type Cheerio, type Element, load } from 'cheerio';
+import { type Context } from 'hono';
+
+import { type DataItem, type Route, type Data, ViewType } from '@/types';
+
+import { art } from '@/utils/render';
+import cache from '@/utils/cache';
+import { getCurrentPath } from '@/utils/helpers';
+import ofetch from '@/utils/ofetch';
+import { parseDate } from '@/utils/parse-date';
+
+const __dirname = getCurrentPath(import.meta.url);
+
+const cleanHtml = (html: string, preservedTags: string[]): string => {
+    const $ = load(html);
+
+    $('div.informationbox').remove();
+    $('div.contributors').remove();
+
+    $('*')
+        .not(preservedTags.join(', '))
+        .contents()
+        .filter((_, el) => el.type === 'text')
+        .remove();
+
+    $('*')
+        .not(preservedTags.join(', '))
+        .filter((_, el) => $(el).children().length === 0)
+        .remove();
+
+    return $.html() || '';
+};
+
+export const handler = async (ctx: Context): Promise<Data> => {
+    const { language: lang = 'zh', category = 'trends-and-insights' } = ctx.req.param();
+    // default limit is 12
+    const limit: number = Number.parseInt(ctx.req.query('limit') ?? '12', 10);
+
+    const rootUrl: string = 'https://www.joneslanglasalle.com.cn';
+    const targetUrl: string = new URL(`${lang}/${category}`, rootUrl).href;
+
+    const response = await ofetch(targetUrl);
+    const $: CheerioAPI = load(response);
+    const language: string = $('html').prop('lang') ?? 'en';
+
+    let items: DataItem[] = $('div.ti-title')
+        .slice(0, limit)
+        .toArray()
+        .map((item): DataItem => {
+            const $item: Cheerio<Element> = $(item);
+            const aEl = $item.closest('a');
+
+            const title: string = $item.text();
+            const link: string | undefined = aEl.prop('href');
+
+            const description: string = art(path.join(__dirname, 'templates/description.art'), {
+                intro: aEl.find('p.ti-teaser').text(),
+            });
+
+            const image: string | undefined = aEl.find('div.ti-image-container img').prop('src') ? new URL(aEl.find('div.ti-image-container img').prop('src') as string, rootUrl).href : undefined;
+
+            return {
+                title,
+                description,
+                pubDate: parseDate(aEl.find('span.ti-date').text(), ['MM月DD日', 'MMMM DD']),
+                link: link ? new URL(link, rootUrl).href : undefined,
+                category: [aEl.find('span.ti-type').text()].filter(Boolean),
+                content: {
+                    html: description,
+                    text: aEl.find('p.ti-teaser').text(),
+                },
+                image,
+                banner: image,
+                language,
+            };
+        });
+
+    items = (
+        await Promise.all(
+            items.map((item) => {
+                if (!item.link && typeof item.link !== 'string') {
+                    return item;
+                }
+
+                return cache.tryGet(item.link, async (): Promise<DataItem> => {
+                    try {
+                        const detailResponse = await ofetch(item.link);
+                        const $$: CheerioAPI = load(detailResponse);
+
+                        const title: string = $$('meta[property="og:title"]').prop('content');
+                        const guid: string = $$('meta[property="og:url"]').prop('content');
+                        const image: string | undefined = $$('meta[property="og:image"]').prop('content');
+
+                        const pubDate: Date = parseDate($$('div.publicationdate').text().trim(), ['YYYY 年MM 月DD 日', 'MMMM DD, YYYY']);
+
+                        const author: DataItem['author'] = $$('div.contributors ul li')
+                            .toArray()
+                            .map((el) => ({
+                                name: $$(el).text(),
+                            }));
+
+                        const media: Record<string, Record<string, string>> = {};
+
+                        $$('picture').each((_, el) => {
+                            const $$el = $$(el);
+
+                            const src = $$el.find('source').last().prop('srcset') ? new URL($$el.find('source').last().prop('srcset') as string, rootUrl).href : undefined;
+
+                            if (src) {
+                                $$el.replaceWith(
+                                    art(path.join(__dirname, 'templates/description.art'), {
+                                        images: [
+                                            {
+                                                src,
+                                            },
+                                        ],
+                                    })
+                                );
+
+                                const mediaType: string | undefined = src.split(/\./).pop();
+
+                                if (mediaType) {
+                                    media[mediaType] = { url: src };
+                                }
+                            }
+                        });
+
+                        const extraLinks = $$('div.related-content a.content-card')
+                            .toArray()
+                            .map((el) => {
+                                const $$el: Cheerio<Element> = $$(el);
+
+                                return {
+                                    url: new URL($$el.prop('href') as string, rootUrl).href,
+                                    type: 'related',
+                                    content_html: $$el.find('div.content-card__body').html(),
+                                };
+                            })
+                            .filter((link): link is { url: string; type: string; content_html: string } => true);
+
+                        const description: string = art(path.join(__dirname, 'templates/description.art'), {
+                            description: cleanHtml($$('div.page-section').eq(1).html() ?? $$('div.copy-block').html() ?? '', ['div.richtext p', 'h3', 'h4', 'h5', 'h6', 'figure', 'img', 'ul', 'li', 'span', 'b']),
+                        });
+
+                        return {
+                            title,
+                            description,
+                            pubDate,
+                            category: $$('meta[property="article:tag"]').prop('content').split(/,\s/),
+                            author,
+                            guid,
+                            id: guid,
+                            content: {
+                                html: description,
+                                text: description,
+                            },
+                            image,
+                            banner: image,
+                            language,
+                            media: Object.keys(media).length > 0 ? media : undefined,
+                            _extra: {
+                                links: extraLinks.length > 0 ? extraLinks : undefined,
+                            },
+                        };
+                    } catch {
+                        return item;
+                    }
+                });
+            })
+        )
+    ).filter((_): _ is DataItem => true);
+
+    const title = $('title').text();
+    const feedImage = $('img.logo').prop('src') ? new URL($('img.logo').prop('src') as string, rootUrl).href : undefined;
+
+    return {
+        title,
+        description: $('meta[property="og:description"]').prop('content'),
+        link: targetUrl,
+        item: items,
+        allowEmpty: true,
+        image: feedImage,
+        author: title.split(/\|/).pop(),
+        language,
+        id: $('meta[property="og:url"]').prop('content'),
+    };
+};
+
+export const route: Route = {
+    path: '/:language?/:category{.+}?',
+    name: 'Trends & Insights',
+    url: 'joneslanglasalle.com.cn',
+    maintainers: ['nczitzk', 'pseudoyu'],
+    handler,
+    example: '/joneslanglasalle/en/trends-and-insights',
+    parameters: {
+        language: 'Language, `zh` by default',
+        category: 'Category, `trends-and-insights` by default',
+    },
+    description: `::: tip
+If you subscribe to [Trends & Insights](https://www.joneslanglasalle.com.cn/en/trends-and-insights)，where the URL is \`https://www.joneslanglasalle.com.cn/en/trends-and-insights\`, extract the part \`https://joneslanglasalle.com.cn/\` to the end. Use \`zh\` and \`trends-and-insights\` as the parameters to fill in. Therefore, the route will be [\`/joneslanglasalle/en/trends-and-insights\`](https://rsshub.app/joneslanglasalle/en/trends-and-insights).
+:::
+
+| Category  | ID                            |
+| --------- | ----------------------------- |
+| Latest    | trends-and-insights           |
+| Workplace | trends-and-insights/workplace |
+| Investor  | trends-and-insights/investor  |
+| Cities    | trends-and-insights/cities    |
+| Research  | trends-and-insights/research  |
+`,
+    categories: ['new-media'],
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportRadar: true,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['joneslanglasalle.com.cn/:language/:category'],
+            target: (params) => {
+                const language = params.language;
+                const category = params.category;
+
+                return language ? `/${language}${category ? `/${category}` : ''}` : '';
+            },
+        },
+        {
+            title: 'Latest',
+            source: ['joneslanglasalle.com.cn/en/trends-and-insights'],
+            target: '/en/trends-and-insights',
+        },
+        {
+            title: 'Workplace',
+            source: ['joneslanglasalle.com.cn/en/trends-and-insights/workplace'],
+            target: '/en/trends-and-insights/workplace',
+        },
+        {
+            title: 'Investor',
+            source: ['joneslanglasalle.com.cn/en/trends-and-insights/investor'],
+            target: '/en/trends-and-insights/investor',
+        },
+        {
+            title: 'Cities',
+            source: ['joneslanglasalle.com.cn/en/trends-and-insights/cities'],
+            target: '/en/trends-and-insights/cities',
+        },
+        {
+            title: 'Research',
+            source: ['joneslanglasalle.com.cn/en/trends-and-insights/research'],
+            target: '/en/trends-and-insights/research',
+        },
+        {
+            title: '房地产趋势与洞察',
+            source: ['joneslanglasalle.com.cn/zh/trends-and-insights'],
+            target: '/zh/trends-and-insights',
+        },
+        {
+            title: '办公空间',
+            source: ['joneslanglasalle.com.cn/zh/trends-and-insights/workplace'],
+            target: '/zh/trends-and-insights/workplace',
+        },
+        {
+            title: '投资者',
+            source: ['joneslanglasalle.com.cn/zh/trends-and-insights/investor'],
+            target: '/zh/trends-and-insights/investor',
+        },
+        {
+            title: '城市',
+            source: ['joneslanglasalle.com.cn/zh/trends-and-insights/cities'],
+            target: '/zh/trends-and-insights/cities',
+        },
+        {
+            title: '研究报告',
+            source: ['joneslanglasalle.com.cn/zh/trends-and-insights/research'],
+            target: '/zh/trends-and-insights/research',
+        },
+    ],
+    view: ViewType.Articles,
+
+    zh: {
+        path: '/:language?/:category{.+}?',
+        name: '房地产趋势与洞察',
+        url: 'joneslanglasalle.com.cn',
+        maintainers: ['nczitzk'],
+        handler,
+        example: '/joneslanglasalle/zh/trends-and-insights',
+        parameters: {
+            language: '语言，默认为 `zh`，可在对应分类页 URL 中找到',
+            category: '分类，默认为 `trends-and-insights`，可在对应分类页 URL 中找到',
+        },
+        description: `::: tip
+若订阅 [房地产趋势与洞察](https://www.joneslanglasalle.com.cn/zh/trends-and-insights)，网址为 \`https://www.joneslanglasalle.com.cn/zh/trends-and-insights\`，请截取 \`https://joneslanglasalle.com.cn/\` 到末尾的部分 \`zh\` 和 \`trends-and-insights\` 作为 \`language\` 和 \`category\` 参数填入，此时目标路由为 [\`/joneslanglasalle/zh/trends-and-insights\`](https://rsshub.app/joneslanglasalle/zh/trends-and-insights)。
+:::
+
+| 分类名称   | 分类 ID                       |
+| ---------- | ----------------------------- |
+| 趋势及洞察 | trends-and-insights           |
+| 办公空间   | trends-and-insights/workplace |
+| 投资者     | trends-and-insights/investor  |
+| 城市       | trends-and-insights/cities    |
+| 研究报告   | trends-and-insights/research  |
+`,
+    },
+};

package/lib/routes/joneslanglasalle/namespace.ts

@@ -0,0 +1,13 @@
+import type { Namespace } from '@/types';
+
+export const namespace: Namespace = {
+    name: 'Jones Lang LaSalle',
+    url: 'joneslanglasalle.com.cn',
+    categories: ['new-media'],
+    description: 'JLL is a global real estate services firm in commercial property and investment management, providing services for real estate owners, occupiers and investors.',
+    lang: 'zh-CN',
+    zh: {
+        name: '仲量联行JLL',
+        description: '仲量联行JLL是全球领先的房地产专业服务和投资管理公司，为企业、房地产业主、投资者及政府提供各类资产的施工、租赁、管理、投资咨询服务。仲量联行也致力于高质量城市发展、打造理想空间、提供可持续的房地产解决方案。',
+    },
+};

package/lib/routes/joneslanglasalle/templates/description.art

@@ -0,0 +1,21 @@
+{{ if images }}
+  {{ each images image }}
+    {{ if !videos?.[0]?.src && image?.src }}
+      <figure>
+        <img
+          {{ if image.alt }}
+            alt="{{ image.alt }}"
+          {{ /if }}
+        src="{{ image.src }}">
+      </figure>
+    {{ /if }}
+  {{ /each }}
+{{ /if }}
+
+{{ if intro }}
+  <blockquote>{{ intro }}</blockquote>
+{{ /if }}
+
+{{ if description }}
+  {{@ description }}
+{{ /if }}

package/lib/routes/jrj/index.ts

@@ -0,0 +1,127 @@
+import { Route } from '@/types';
+
+import cache from '@/utils/cache';
+import { load } from 'cheerio';
+import { parseDate } from '@/utils/parse-date';
+import { ofetch } from 'ofetch';
+
+const options = {
+    '103': '财经资讯',
+    '508': '科技资讯',
+    '106': '商业资讯',
+    '632': '消费资讯',
+    '630': '医疗资讯',
+    '119': '康养资讯',
+    '004': '汽车资讯',
+    '009': '房产资讯',
+    '629': 'ESG 资讯',
+    '010': 'A股资讯',
+    '001': '港股资讯',
+    '102': '美股资讯',
+    '113': '银行资讯',
+    '115': '保险资讯',
+    '104': '基金资讯',
+    '503': '私募资讯',
+    '112': '信托资讯',
+    '007': '外汇资讯',
+    '107': '期货资讯',
+    '118': '债券资讯',
+    '603': '券商资讯',
+    '105': '观点',
+};
+
+export const route: Route = {
+    path: '/:channelNum',
+    categories: ['finance'],
+    example: '/jrj/103',
+    parameters: {
+        channelNum: {
+            description: '栏目编号',
+            options: Object.entries(options).map(([value, label]) => ({ value, label })),
+        },
+    },
+    url: 'www.jrj.com.cn',
+    name: '资讯',
+    description: `
+| column | Description |
+| ---   | ---   |
+| 103   | 财经资讯 |
+| 508   | 科技资讯 |
+| 106   | 商业资讯 |
+| 632   | 消费资讯 |
+| 630   | 医疗资讯 |
+| 119   | 康养资讯 |
+| 004   | 汽车资讯 |
+| 009   | 房产资讯 |
+| 629   | ESG 资讯 |
+| 001   | 港股资讯 |
+| 102   | 美股资讯 |
+| 113   | 银行资讯 |
+| 115   | 保险资讯 |
+| 104   | 基金资讯 |
+| 503   | 私募资讯 |
+| 112   | 信托资讯 |
+| 007   | 外汇资讯 |
+| 107   | 期货资讯 |
+| 118   | 债券资讯 |
+| 603   | 券商资讯 |
+| 105   | 观点 |
+    `,
+    maintainers: ['p3psi-boo'],
+    handler,
+};
+
+async function handler(ctx) {
+    const channelNum = ctx.req.param('channelNum');
+
+    const url = 'https://gateway.jrj.com/jrj-news/news/queryNewsList';
+
+    const response = await ofetch(url, {
+        method: 'post',
+        body: {
+            sortBy: 1,
+            pageSize: 20,
+            makeDate: '',
+            channelNum,
+            infoCls: '',
+        },
+    });
+
+    const alist = response.data.data;
+
+    const list = alist.map((item) => {
+        const link = item.pcInfoUrl;
+        const title = item.title;
+        const author = item.paperMediaSource;
+        const pubDate = parseDate(item.makeDate);
+
+        return {
+            title,
+            link,
+            author,
+            pubDate,
+        };
+    });
+
+    const items = await Promise.all(
+        list.map((item) =>
+            cache.tryGet(item.link, async () => {
+                const articleUrl = item.link;
+                const response = await ofetch(articleUrl);
+
+                const $ = load(response);
+
+                const content = $('.article_content').html();
+
+                item.description = content;
+                return item;
+            })
+        )
+    );
+
+    return {
+        title: `${options[channelNum]} - 金融界`,
+        link: 'https://jrj.com',
+        item: items,
+    };
+}

package/lib/routes/jrj/namespace.ts

@@ -0,0 +1,8 @@
+import type { Namespace } from '@/types';
+
+export const namespace: Namespace = {
+    name: '金融界',
+    url: 'www.jrj.com.cn',
+    description: '金融界是国内领先的金融信息服务平台，日均触达千万用户，年度访问量超过3亿，受众覆盖中国主流金融机构、上市公司和活跃投资理财群体',
+    lang: 'zh-CN',
+};

package/lib/routes/kcna/news.ts

@@ -7,9 +7,11 @@ import got from '@/utils/got';
 import { load } from 'cheerio';
 import asyncPool from 'tiny-async-pool';
 import { art } from '@/utils/render';
-import { parseJucheDate, fixDesc, fetchPhoto, fetchVideo } from './utils';
+import { fixDesc, fetchPhoto, fetchVideo } from './utils';
 import path from 'node:path';
 import sanitizeHtml from 'sanitize-html';
+import { parseDate } from '@/utils/parse-date';
+import timezone from '@/utils/timezone';
 
 export const route: Route = {
     path: '/:lang/:category?',
@@ -66,12 +68,12 @@ async function handler(ctx) {
         .map((_, item) => {
             item = $(item);
             const dateElem = item.find('.publish-time');
-            const dateString = dateElem.text();
+            const dateString = dateElem.text().match(/\d+\.\d+\.\d+/);
             dateElem.remove();
             return {
                 title: item.text(),
                 link: rootUrl + item.attr('href'),
-                pubDate: parseJucheDate(dateString),
+                pubDate: timezone(parseDate(dateString[0]), +9),
             };
         })
         .get();
@@ -87,9 +89,9 @@ async function handler(ctx) {
             item.title = $('article-main-title').text() || item.title;
 
             const dateElem = $('.publish-time');
-            const dateString = dateElem.text();
+            const dateString = dateElem.text().match(/\d+\.\d+\.\d+/);
             dateElem.remove();
-            item.pubDate = parseJucheDate(dateString) || item.pubDate;
+            item.pubDate = dateString ? timezone(parseDate(dateString[0]), +9) : item.pubDate;
 
             const description = fixDesc($, $('.article-content-body .content-wrapper'));
 

package/lib/routes/keep/user.ts

@@ -8,7 +8,7 @@ import path from 'node:path';
 
 export const route: Route = {
     path: '/user/:id',
-    categories: ['social-media'],
+    categories: ['social-media', 'popular'],
     example: '/keep/user/556b02c1ab59390afea671ea',
     parameters: { id: 'Keep 用户 id' },
     features: {

package/lib/routes/kemono/index.ts

@@ -36,10 +36,10 @@ export const route: Route = {
   | ----- | ------- | ------------ | ------- | ------------- | ------ | ------- | ------ |
   | posts | patreon | fanbox       | gumroad | subscribestar | dlsite | discord | fantia |
 
-  :::tip
+::: tip
   When \`posts\` is selected as the value of the parameter **source**, the parameter **id** does not take effect.
   There is an optinal parameter **limit** which controls the number of posts to fetch, default value is 25.
-  :::`,
+:::`,
 };
 
 async function handler(ctx) {

package/lib/routes/keylol/index.ts

@@ -41,10 +41,10 @@ export const route: Route = {
     ],
     maintainers: ['nczitzk', 'kennyfong19931'],
     handler,
-    description: `:::tip
+    description: `::: tip
   若订阅 [热点聚焦](https://keylol.com/f161-1)，网址为 \`https://keylol.com/f161-1\`。截取 \`https://keylol.com/\` 到末尾的部分 \`f161-1\` 作为参数，此时路由为 [\`/keylol/f161-1\`](https://rsshub.app/keylol/f161-1)。
   若订阅子分类 [试玩免费 - 热点聚焦](https://keylol.com/forum.php?mod=forumdisplay&fid=161&filter=typeid&typeid=459)，网址为 \`https://keylol.com/forum.php?mod=forumdisplay&fid=161&filter=typeid&typeid=459\`。提取\`fid\`及\`typeid\` 作为参数，此时路由为 [\`/keylol/fid=161&typeid=459\`](https://rsshub.app/keylol/fid=161&typeid=459)。注意不要包括\`filter\`，会调用[全局的内容过滤](https://docs.rsshub.app/guide/parameters#filtering)。
-  :::`,
+:::`,
 };
 
 async function handler(ctx) {

package/lib/routes/koreaherald/index.ts

@@ -0,0 +1,70 @@
+import { Route } from '@/types';
+import cache from '@/utils/cache';
+import got from '@/utils/got';
+import { load } from 'cheerio';
+import { parseDate } from '@/utils/parse-date';
+import timezone from '@/utils/timezone';
+
+export const route: Route = {
+    path: '/:category{.+}?',
+    categories: ['traditional-media'],
+    example: '/koreaherald/National',
+    parameters: {
+        category: 'Category from the path of the URL of the corresponding site, `National` by default',
+    },
+    features: {
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+        requireConfig: false,
+    },
+    name: 'News',
+    maintainers: ['quiniapiezoelectricity'],
+    handler,
+    description: `
+::: tip
+For example, the category for the page https://www.koreaherald.com/Business and https://www.koreaherald.com/Business/Market would be \`/Business\` and \`/Business/Market\` respectively. 
+:::
+`,
+    radar: [
+        {
+            source: ['www.koreaherald.com/:category'],
+            target: '/:category',
+        },
+    ],
+};
+
+async function handler(ctx) {
+    const category = ctx.req.param('category') ?? 'National';
+    const baseUrl = 'https://www.koreaherald.com/';
+
+    const response = await got(new URL(category, baseUrl).href);
+    const $ = load(response.data);
+    const title = $('ul.gnb').find('[class="on"]').length > 0 ? $('ul.gnb').find('[class="on"]').text() : $('div.nav_area > a.category').text();
+    const list = $('article.recent_news > ul.news_list > li')
+        .toArray()
+        .map((item) => new URL($(item).find('a').attr('href'), baseUrl).href);
+    const items = await Promise.all(
+        list.map((url) =>
+            cache.tryGet(url, async () => {
+                const response = await got(url);
+                const $ = load(response.data);
+                const metadata = JSON.parse($('[type="application/ld+json"]').text());
+                return {
+                    title: metadata.headline,
+                    link: url,
+                    pubDate: timezone(parseDate(metadata.datePublished), +9),
+                    author: metadata.author.name,
+                    description: $('article.article-body').html(),
+                };
+            })
+        )
+    );
+    return {
+        title: `The Korea Herald - ${title}`,
+        link: new URL(category, baseUrl).href,
+        item: items,
+    };
+}

package/lib/routes/koreaherald/namespace.ts

@@ -0,0 +1,6 @@
+import type { Namespace } from '@/types';
+
+export const namespace: Namespace = {
+    name: 'The Korea Herald',
+    url: 'koreaherald.com',
+};

package/lib/routes/kuaidi100/index.ts

@@ -19,11 +19,11 @@ export const route: Route = {
     handler,
     description: `快递公司代号如果不能确定，可通过下方快递列表获得。
 
-  :::warning
+::: warning
   1.  构造链接前请确认所有参数正确：错误\`快递公司 - 订单号\`组合将会缓存信息一小段时间防止产生无用查询
   2.  正常查询的订单在未签收状态下不会被缓存：请控制查询频率
   3.  订单完成后请尽快取消订阅，避免资源浪费
-  :::`,
+:::`,
 };
 
 async function handler(ctx) {

package/lib/routes/kuaishou/namespace.ts

@@ -0,0 +1,8 @@
+import type { Namespace } from '@/types';
+
+export const namespace: Namespace = {
+    name: '快手',
+    url: 'kuaishou.com',
+    categories: ['social-media'],
+    lang: 'zh-CN',
+};