feat(route): add Shan Dong Museum exhibition and news route (#22187)
* feat(route): add Shan Dong Museum exhibition and news route * fix: replaceAll to replace * fix: replace for fullDurationDate * fix: change xmlmode to xml * fix: oxlint warning --------- Co-authored-by: Bian Jiamin <jiamin@BiandeMac-mini.local>
This commit is contained in:
Jiamin
GitHub
parent
cdf95e67fe
commit
d3086acfe0
|
|
@ -0,0 +1,162 @@
|
|||
import { load } from 'cheerio';
|
||||
import { renderToString } from 'hono/jsx/dom/server';
|
||||
|
||||
import type { DataItem, Route } from '@/types';
|
||||
import cache from '@/utils/cache';
|
||||
import got from '@/utils/got';
|
||||
import { parseDate } from '@/utils/parse-date';
|
||||
|
||||
import { namespace } from './namespace';
|
||||
|
||||
// format the date to YYYY-MM-DD and handle missing year or month
|
||||
const extractDates = (durationStr: string) => {
|
||||
let startDate: string | undefined;
|
||||
let endDate: string | undefined;
|
||||
|
||||
if (!durationStr) {
|
||||
return { startDate, endDate };
|
||||
}
|
||||
|
||||
const parts = durationStr.split(/[-—~]+/).map((p) => p.trim());
|
||||
const startStr = parts[0];
|
||||
const endStr = parts[1];
|
||||
|
||||
let startYear: string | undefined;
|
||||
let startMonth: string | undefined;
|
||||
|
||||
const startRegex = /(\d{4})年(\d{1,2})月(\d{1,2})日/;
|
||||
const startMatch = startStr.match(startRegex);
|
||||
|
||||
if (startMatch) {
|
||||
startYear = startMatch[1];
|
||||
startMonth = startMatch[2].padStart(2, '0');
|
||||
const startDay = startMatch[3].padStart(2, '0');
|
||||
startDate = `${startYear}-${startMonth}-${startDay}`;
|
||||
}
|
||||
|
||||
if (endStr && startDate) {
|
||||
const endRegex = /(?:(\d{4})年)?(?:(\d{1,2})月)?(\d{1,2})日/;
|
||||
const endMatch = endStr.match(endRegex);
|
||||
|
||||
if (endMatch) {
|
||||
const matchYear = endMatch[1];
|
||||
const matchMonth = endMatch[2]?.padStart(2, '0');
|
||||
const matchDay = endMatch[3].padStart(2, '0');
|
||||
|
||||
const finalEndYear = matchYear || startYear;
|
||||
const finalEndMonth = matchMonth || startMonth;
|
||||
const finalEndDay = matchDay;
|
||||
endDate = `${finalEndYear}-${finalEndMonth}-${finalEndDay}`;
|
||||
}
|
||||
}
|
||||
|
||||
return { startDate, endDate };
|
||||
};
|
||||
|
||||
export const route: Route = {
|
||||
path: '/exhibitions',
|
||||
categories: ['travel'],
|
||||
example: '/sdmuseum/exhibitions',
|
||||
parameters: {},
|
||||
name: 'Temporary Exhibitions',
|
||||
maintainers: ['magazian'],
|
||||
radar: [
|
||||
{
|
||||
source: ['www.sdmuseum.com/col/col271702/index.html'],
|
||||
target: '/exhibitions',
|
||||
},
|
||||
],
|
||||
|
||||
handler: async () => {
|
||||
const baseUrl = 'https://www.sdmuseum.com';
|
||||
const apiUrl = `${baseUrl}/col/col271702/index.html`;
|
||||
const museumName = namespace.zh?.name || namespace.name;
|
||||
|
||||
const response = await got({
|
||||
method: 'get',
|
||||
url: apiUrl,
|
||||
});
|
||||
|
||||
const $ = load(response.data, {
|
||||
xml: true, // use xmlMode to preserve CDATA sections inside <script>
|
||||
});
|
||||
|
||||
const items = await Promise.all(
|
||||
$('record')
|
||||
.toArray()
|
||||
.map((el) => {
|
||||
const itemHtml = $(el).text();
|
||||
const $item = load(itemHtml);
|
||||
|
||||
const title = $item('.ltit a').attr('title');
|
||||
const link = new URL($item('.ltit a').attr('href')!, baseUrl).href;
|
||||
const imgUrlRaw = $item('.img img').attr('src') || '';
|
||||
const imgUrl = new URL(imgUrlRaw, baseUrl).href;
|
||||
|
||||
const location = $item('.item.add').text().trim();
|
||||
const fullDuration = $item('.item.time').text().trim();
|
||||
const fullDurationDate = fullDuration.replace(/开展时间\s*[:：]\s*/, '').trim(); // use regex to remove "开展时间" prefix if it exists, so replace is used here.
|
||||
|
||||
const { startDate, endDate } = extractDates(fullDurationDate);
|
||||
|
||||
// get pubDate from the detail page
|
||||
return cache.tryGet(link, async () => {
|
||||
const detailResponse = await got({
|
||||
method: 'get',
|
||||
url: link,
|
||||
});
|
||||
const $detail = load(detailResponse.data);
|
||||
|
||||
const pubDateStr = $detail('meta[name="PubDate"]').attr('content') || '';
|
||||
const pubDate = parseDate(pubDateStr, 'YYYY-MM-DD HH:mm');
|
||||
|
||||
const description = renderToString(
|
||||
<div>
|
||||
<img src={imgUrl} />
|
||||
<br />
|
||||
<p>
|
||||
<b>地点：</b>
|
||||
{location}
|
||||
</p>
|
||||
<p>
|
||||
<b>开展：</b>
|
||||
{startDate ?? '未定/常设'}
|
||||
</p>
|
||||
<p>
|
||||
<b>闭展：</b>
|
||||
{endDate ?? '未定/常设'}
|
||||
</p>
|
||||
{fullDuration && (
|
||||
<p>
|
||||
<small>原始展期：{fullDuration}</small>
|
||||
</p>
|
||||
)}
|
||||
</div>
|
||||
);
|
||||
|
||||
return {
|
||||
title,
|
||||
link,
|
||||
pubDate,
|
||||
description,
|
||||
_extra: {
|
||||
museumName,
|
||||
title,
|
||||
location,
|
||||
startDate,
|
||||
endDate,
|
||||
itemLink: link,
|
||||
},
|
||||
};
|
||||
});
|
||||
})
|
||||
);
|
||||
|
||||
return {
|
||||
title: `${museumName} - 临时展览`,
|
||||
link: apiUrl,
|
||||
language: 'zh-CN',
|
||||
item: items as DataItem[],
|
||||
};
|
||||
},
|
||||
};
|
||||
|
|
@ -0,0 +1,10 @@
|
|||
import type { Namespace } from '@/types';
|
||||
|
||||
export const namespace: Namespace = {
|
||||
name: 'Shan Dong Museum',
|
||||
url: 'www.sdmuseum.com/',
|
||||
|
||||
zh: {
|
||||
name: '山东博物馆',
|
||||
},
|
||||
};
|
||||
|
|
@ -0,0 +1,63 @@
|
|||
import { load } from 'cheerio';
|
||||
|
||||
import type { DataItem, Route } from '@/types';
|
||||
import got from '@/utils/got';
|
||||
import { parseDate } from '@/utils/parse-date';
|
||||
|
||||
import { namespace } from './namespace';
|
||||
|
||||
export const route: Route = {
|
||||
path: '/news',
|
||||
categories: ['travel'],
|
||||
example: '/sdmuseum/news',
|
||||
parameters: {},
|
||||
name: 'News',
|
||||
maintainers: ['magazian'],
|
||||
radar: [
|
||||
{
|
||||
source: ['www.sdmuseum.com/col/col270324/index.html'],
|
||||
target: '/news',
|
||||
},
|
||||
],
|
||||
|
||||
handler: async () => {
|
||||
const baseUrl = 'https://www.sdmuseum.com';
|
||||
const apiUrl = `${baseUrl}/col/col270324/index.html`;
|
||||
const museumName = namespace.zh?.name || namespace.name;
|
||||
|
||||
const response = await got({
|
||||
method: 'get',
|
||||
url: apiUrl,
|
||||
});
|
||||
const $ = load(response.data, {
|
||||
xml: true,
|
||||
});
|
||||
|
||||
const items = $('record')
|
||||
.toArray()
|
||||
.map((el) => {
|
||||
const itemHtml = $(el).text();
|
||||
const $item = load(itemHtml);
|
||||
|
||||
const $a = $item('.ltit a');
|
||||
const title = $a.attr('title');
|
||||
const link = new URL($a.attr('href')!, baseUrl).href;
|
||||
|
||||
// date format 2026.06.03
|
||||
const dateStr = $item('.data.av').text().trim();
|
||||
|
||||
return {
|
||||
title,
|
||||
link,
|
||||
pubDate: parseDate(dateStr),
|
||||
};
|
||||
});
|
||||
|
||||
return {
|
||||
title: `${museumName} - 公告`,
|
||||
link: apiUrl,
|
||||
language: 'zh-CN',
|
||||
item: items as DataItem[],
|
||||
};
|
||||
},
|
||||
};
|
||||
Loading…
Reference in New Issue