feat(route): add Guandong Museum Exhibitions and news route (#22729)

* feat:add Guandong Museum Exhibitions and news

* fix: update selector

lib/routes/gdmuseum/exhibition.tsx

@@ -0,0 +1,126 @@
+import { load } from 'cheerio';
+import dayjs from 'dayjs';
+import type { Context } from 'hono';
+import { renderToString } from 'hono/jsx/dom/server';
+
+import type { DataItem, Route } from '@/types';
+import got from '@/utils/got';
+import { parseDate } from '@/utils/parse-date';
+
+import { namespace } from './namespace';
+
+// convert exhibition date string to YYYY-MM-DD format, e.g. "2023年5月1日" or "2023/5/1" => "2023-05-01"
+const formatExhibitionDate = (d?: string) => {
+    if (!d) {
+        return;
+    }
+    return dayjs(d.replaceAll(/[年月/.]/g, '-').replaceAll('日', '')).format('YYYY-MM-DD');
+};
+
+const parseExhibitionDuration = (duration?: string) => {
+    const allDates = duration?.match(/\d{4}[./年-]\d{1,2}[./月-]\d{1,2}日?/g) || [];
+    return {
+        startDate: formatExhibitionDate(allDates[0]),
+        endDate: formatExhibitionDate(allDates[1]),
+    };
+};
+
+export const route: Route = {
+    path: '/exhibition/:type?',
+    categories: ['travel'],
+    example: '/gdmuseum/exhibition/temp',
+    parameters: {
+        type: 'Exhibition type, supported values: temp(临时展览), default: All exhibitions.',
+    },
+    name: 'Current Exhibitions',
+    maintainers: ['magazian'],
+    radar: [
+        {
+            source: ['www.gdmuseum.org.cn/col9/list'],
+            target: '/exhibition',
+        },
+    ],
+    handler: async (ctx: Context) => {
+        const typeParam = ctx.req.param('type') || 'all';
+
+        const museumName = namespace.zh?.name || namespace.name;
+        const baseUrl = 'https://www.gdmuseum.org.cn';
+        const url = `${baseUrl}/col9/list`;
+
+        const response = await got(url);
+        const $ = load(response.data);
+
+        const list = $('.ULLIST li a[href^="/cn/col"]')
+            .toArray()
+            .map((el) => {
+                const $item = $(el);
+                const title = $item.find('.divtt.qui-dot').text();
+
+                if (typeParam === 'temp') {
+                    if (title.includes('常设展览')) {
+                        return null;
+                    }
+                } else if (typeParam !== 'all') {
+                    return null;
+                }
+
+                const rawLink = $item.attr('href');
+                const itemLink = rawLink ? new URL(rawLink, baseUrl).href : '';
+
+                const rawSrc = $item.find('img').attr('src');
+                const imgUrl = rawSrc ? new URL(rawSrc, baseUrl).href : undefined;
+
+                const textDurationAndLocation = $item.find('.quitxt.qui-dot').text().trim();
+                const [fullDuration = '', location = ''] = textDurationAndLocation.split('|').map((p) => p.trim());
+
+                const { startDate, endDate } = parseExhibitionDuration(fullDuration);
+                const pubDate = startDate ? parseDate(startDate) : undefined;
+
+                const description = renderToString(
+                    <div>
+                        {imgUrl && <img src={imgUrl} />}
+                        <br />
+                        <p>
+                            <b>地点：</b>
+                            {location || '参考详情'}
+                        </p>
+                        <p>
+                            <b>开展：</b>
+                            {startDate || '未定/常设'}
+                        </p>
+                        <p>
+                            <b>闭展：</b>
+                            {endDate || '未定/常设'}
+                        </p>
+                        {fullDuration && (
+                            <p>
+                                <small>原始展期：{fullDuration}</small>
+                            </p>
+                        )}
+                    </div>
+                );
+
+                return {
+                    title,
+                    link: itemLink,
+                    pubDate,
+                    description,
+                    // For further .ics file processing
+                    _extra: {
+                        museumName,
+                        location,
+                        startDate,
+                        endDate,
+                    },
+                } as DataItem;
+            })
+            .filter((i): i is DataItem => i !== null);
+
+        return {
+            title: `${museumName} - 正在热展${typeParam === 'temp' ? ' - 临时展览' : ''}`,
+            link: url,
+            language: 'zh-CN',
+            item: list,
+        };
+    },
+};

lib/routes/gdmuseum/namespace.ts

@@ -0,0 +1,10 @@
+import type { Namespace } from '@/types';
+
+export const namespace: Namespace = {
+    name: 'Guangdong Museum',
+    url: 'gdmuseum.org.cn',
+
+    zh: {
+        name: '广东省博物馆',
+    },
+};

lib/routes/gdmuseum/news.ts

@@ -0,0 +1,58 @@
+import { load } from 'cheerio';
+
+import type { DataItem, Route } from '@/types';
+import got from '@/utils/got';
+import { parseDate } from '@/utils/parse-date';
+import timezone from '@/utils/timezone';
+
+import { namespace } from './namespace';
+
+export const route: Route = {
+    path: '/information',
+    categories: ['travel'],
+    example: '/gdmuseum/information',
+    name: 'Information',
+    maintainers: ['magazian'],
+    radar: [
+        {
+            source: ['www.gdmuseum.org.cn/cn/col51/list'],
+            target: '/information',
+        },
+    ],
+    handler: async () => {
+        const baseUrl = 'https://www.gdmuseum.org.cn';
+        const apiUrl = `${baseUrl}/cn/col51/list`;
+        const museumName = namespace.zh?.name || namespace.name;
+
+        const response = await got(apiUrl);
+        const $ = load(response.data);
+
+        const list = $('.ULLIST li a[href^="/cn/col"]')
+            .toArray()
+            .map((el) => {
+                const $item = $(el);
+
+                const rawLink = $item.attr('href') || '';
+                const itemLink = new URL(rawLink, baseUrl).href;
+
+                const title = $item.find('h3.h3.qui-dot').text();
+                const day = $item.find('time b').text().trim();
+                const yearMonth = $item.find('time i').text().trim();
+
+                const pubDate = timezone(parseDate(`${yearMonth}-${day}`), 8);
+
+                return {
+                    title,
+                    link: itemLink,
+                    pubDate,
+                } as DataItem;
+            });
+
+        return {
+            title: `${museumName} - 公告`,
+            link: apiUrl,
+            language: 'zh-CN',
+            item: list,
+        };
+    },
+};