fix(route/ruankao): Remove font presetting and other details. (#18160)

* fix: remove font presetting and other details.

* fix: add anti-crawler tag.

* fix: migrate jsdom to cheerio.

* fix: fix news-router name.
This commit is contained in:
但为君故 2025-01-23 05:38:21 +08:00 committed by GitHub
parent 6b3af0bd21
commit 90c6432838
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 26 additions and 10 deletions

View File

@ -1,11 +1,26 @@
import { DataItem, Route } from '@/types';
import type { DataItem, Route } from '@/types';
import cache from '@/utils/cache';
import got from '@/utils/got';
import { parseDate } from '@/utils/parse-date';
import { load } from 'cheerio';
const BASE_URL = 'https://www.ruankao.org.cn/index/work';
const removeFontPresetting = (html: string = ''): string => {
const $ = load(html);
$('[style]').each((_, element) => {
const style = $(element).attr('style') || '';
const cleanedStyle = style.replaceAll(/font-family:[^;]*;?/gi, '').trim();
$(element).attr('style', cleanedStyle || null);
});
$('style').each((_, styleElement) => {
const cssText = $(styleElement).html() || '';
const cleanedCssText = cssText.replaceAll(/font-family:[^;]*;?/gi, '');
$(styleElement).html(cleanedCssText);
});
return $.html();
};
const handler: Route['handler'] = async () => {
// Fetch the index page
const { data: listResponse } = await got(BASE_URL);
@ -30,8 +45,8 @@ const handler: Route['handler'] = async () => {
});
return {
title: '计算机职业技术资格考试(软考)最新动态',
description: '计算机职业技术资格考试(软考)网站最新动态和消息推送',
title: '计算机职业技术资格考试(软考)动态',
description: '计算机职业技术资格考试(软考)消息推送',
link: BASE_URL,
image: 'https://bm.ruankao.org.cn/asset/image/public/logo.png',
item: (await Promise.all(
@ -40,7 +55,7 @@ const handler: Route['handler'] = async () => {
const CONTENT_SELECTOR = '#contentTxt';
const { data: contentResponse } = await got(item.link);
const contentPage = load(contentResponse);
const content = contentPage(CONTENT_SELECTOR).html() || '';
const content = removeFontPresetting(contentPage(CONTENT_SELECTOR).html() || '');
return {
title: item.title,
pubDate: item.date,
@ -52,13 +67,13 @@ const handler: Route['handler'] = async () => {
image: 'https://bm.ruankao.org.cn/asset/image/public/logo.png',
content,
updated: item.date,
language: 'zh-cn',
language: 'zh-CN',
};
})
)
)) as DataItem[],
allowEmpty: true,
language: 'zh-cn',
language: 'zh-CN',
feedLink: 'https://rsshub.app/ruankao/news',
id: 'https://rsshub.app/ruankao/news',
};
@ -66,14 +81,15 @@ const handler: Route['handler'] = async () => {
export const route: Route = {
path: '/news',
name: '软考最新动态',
name: '软考动态',
description: '**注意:** 官方网站限制了国外网络请求,可能需要通过部署在中国大陆内的 RSSHub 实例访问。',
maintainers: ['PrinOrange'],
handler,
categories: ['study'],
features: {
requireConfig: false,
requirePuppeteer: false,
antiCrawler: false,
antiCrawler: true,
supportBT: false,
supportPodcast: false,
supportScihub: false,
@ -81,7 +97,7 @@ export const route: Route = {
},
radar: [
{
title: '软考新闻动态',
title: '计算机职业技术资格考试(软考)动态',
source: ['www.ruankao.org.cn/index/work', 'www.ruankao.org.cn'],
target: `/news`,
},