Merge pull request #6643 from ikvarxt/master

feat: add 辽宁工程技术大学教务公告全文抓取
This commit is contained in:
NeverBehave
2021-01-10 19:36:26 -05:00
committed by GitHub
+33 -13
View File
@@ -1,26 +1,46 @@
const got = require('@/utils/got');
const cheerio = require('cheerio');
const url = require('url');
module.exports = async (ctx) => {
const link = 'http://jwzx.lntu.edu.cn/index/jwgg.htm';
const response = await got.get(link);
const $ = cheerio.load(response.data);
const list = $('.tr-ri li').slice(0, 10);
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 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: link,
description: '辽宁工程技术大学教务公告',
item:
list &&
list
.map((index, item) => {
item = $(item);
return {
title: item.find('li a').text(),
link: item.find('li a').attr('href'),
};
})
.get(),
item: res,
};
};