mirror of
https://github.com/DIYgod/RSSHub.git
synced 2025-12-16 11:37:44 +08:00
fix: fetch content with tryGet
This commit is contained in:
@@ -1,67 +1,44 @@
|
||||
// 以下代码参考 bupt/yz.js & bupt/utils.js 修改完成,在此感谢。
|
||||
|
||||
const got = require('@/utils/got');
|
||||
const cheerio = require('cheerio');
|
||||
const url = require('url');
|
||||
|
||||
module.exports = async(ctx) => {
|
||||
const meta = {
|
||||
selector: {
|
||||
list: '.tr-ri ul',
|
||||
item: 'li',
|
||||
content: '.v_news_content',
|
||||
},
|
||||
url: 'http://jwzx.lntu.edu.cn/index/jwgg.htm',
|
||||
}
|
||||
const response = await got({
|
||||
method: 'get',
|
||||
url: `${meta.url}`,
|
||||
});
|
||||
const data = response.data;
|
||||
const link = 'http://jwzx.lntu.edu.cn/index/jwgg.htm';
|
||||
const response = await got.get(link);
|
||||
|
||||
const $ = cheerio.load(data);
|
||||
const list = $(meta.selector.list).find(meta.selector.item).get();
|
||||
const $ = cheerio.load(response.data);
|
||||
const list = $('.tr-ri ul').find('li').get();
|
||||
|
||||
const res = await Promise.all(
|
||||
list.map(async(item) => {
|
||||
let $ = cheerio.load(item);
|
||||
// 通过解析后的子项地址
|
||||
const item_link = url.resolve(link, $('a').attr('href'));
|
||||
|
||||
const res = await ProcessFeed(list, meta);
|
||||
const description = await ctx.cache.tryGet(item_link, async() => {
|
||||
const result = await got.get(item_link);
|
||||
|
||||
$ = cheerio.load(result.data);
|
||||
// remove style
|
||||
$('img, div, span, p, table, td, tr').removeAttr('style');
|
||||
$('style, script').remove();
|
||||
|
||||
return $('.v_news_content').html().replace(/(<span[^>]*>|<\/span>)/g, "");
|
||||
});
|
||||
|
||||
const rssitem = {
|
||||
title: $('title').text().split('-')[0],
|
||||
description: description,
|
||||
link: item_link,
|
||||
author: '辽宁工程技术大学教务处',
|
||||
};
|
||||
return Promise.resolve(rssitem);
|
||||
})
|
||||
);
|
||||
|
||||
ctx.state.data = {
|
||||
title: '辽宁工程技术大学教务公告',
|
||||
link: `${meta.url}`,
|
||||
link: link,
|
||||
item: res,
|
||||
};
|
||||
};
|
||||
|
||||
const ProcessFeed = async(list, meta) =>
|
||||
await Promise.all(
|
||||
list.map(async(item) => {
|
||||
let $ = cheerio.load(item);
|
||||
// 通过解析过后的子项地址
|
||||
const $url = url.resolve(meta.url, $('a').attr('href'));
|
||||
|
||||
// 加载内容页面
|
||||
const response = await got({
|
||||
method: 'get',
|
||||
url: $url,
|
||||
});
|
||||
|
||||
const data = response.data;
|
||||
$ = cheerio.load(data); // 使用 cheerio 加载返回的 HTML
|
||||
|
||||
// 去除样式,好文明
|
||||
$('img, div, span, p, table, td, tr').removeAttr('style');
|
||||
$('style, script').remove();
|
||||
|
||||
// 截取掉「……教务处」
|
||||
let title = $('title').text().split('-')[0];
|
||||
|
||||
// 列表上提取到的信息
|
||||
return {
|
||||
title: title,
|
||||
description: $(meta.selector.content).html().replace(/<span*>(.*?)<\/span>/g, '$1').replace(/<span[^>]*>/g, '').replace(/<\/span[^>]*>/g, ''),
|
||||
link: $url,
|
||||
author: '辽宁工程技术大学教务处',
|
||||
guid: $url, // 文章唯一标识
|
||||
};
|
||||
})
|
||||
);
|
||||
};
|
||||
Reference in New Issue
Block a user