feat(route): add 东北大学医学与生物信息工程学院 (#10349)

* feat(route): add 东北大学医学与生物信息工程学院

* Update lib/routes/universities/neu/bmie.js

Co-authored-by: Tony <TonyRL@users.noreply.github.com>

* Update lib/routes/universities/neu/bmie.js

Co-authored-by: Tony <TonyRL@users.noreply.github.com>

* fix(route): 更新BMIE 格式化日期、采用V2路由

* Apply suggestions from code review

Co-authored-by: Tony <TonyRL@users.noreply.github.com>

* fix(route):采用 pubDate 解析

* refactor: migrate to v2

Co-authored-by: Athena <qingcaomc+zen@gmail.com>
This commit is contained in:
天王州アテネ 2022-07-31 02:51:59 +08:00 committed by GitHub
parent 640c3102cd
commit 1d6a70a152
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
8 changed files with 211 additions and 67 deletions

View File

@ -891,13 +891,51 @@ xskb1 对应 <http://www.auto.uestc.edu.cn/index/xskb1.htm>
## 东北大学
### 东北大学新闻网
### 新闻网
<Route author="JeasonLau" example="/neu/news/ddyw" path="/neu/news/:type" :paramsDesc="['种类名']">
<Route author="JeasonLau" example="/neu/news/ddyw" path="/neu/news/:type" :paramsDesc="['种类名,见下表']">
| 东大要闻 | 媒体东大 | 通知公告 | 新闻纵横 | 人才培养 | 学术科研 | 英文新闻 | 招生就业 | 考研出国 | 校园文学 | 校友风采 | 时事热点 | 教育前沿 | 文化体育 | 最新科技 |
| ---- | ---- | ---- | ---- | ---: | ---- | ---- | ---- | ---- | ---- | ---- | ---- | ---- | ---- | ---- |
| ddyw | mtdd | tzgg | xwzh | rcpy | xsky | 217 | zsjy | kycg | xywx | xyfc | ssrd | jyqy | whty | zxkj |
| 种类名 | 参数 |
| ---- | ---- |
| 东大要闻 | ddyw |
| 媒体东大 | mtdd |
| 通知公告 | tzgg |
| 新闻纵横 | xwzh |
| 人才培养 | rcpy |
| 学术科研 | xsky |
| 英文新闻 | 217 |
| 招生就业 | zsjy |
| 考研出国 | kycg |
| 校园文学 | xywx |
| 校友风采 | xyfc |
| 时事热点 | ssrd |
| 教育前沿 | jyqy |
| 文化体育 | whty |
| 最新科技 | zxkj |
</Route>
### 医学与生物信息工程学院
<Route author="TennousuAthena" example="/neu/bmie/news" path="/neu/bmie/:type" :paramsDesc="['分类 id 见下表']">
| Id | 名称 |
| ---------------------- | ----- |
| news | 学院新闻 |
| academic | 学术科研 |
| talent_development | 人才培养 |
| international_exchange | 国际交流 |
| announcement | 通知公告 |
| undergraduate_dev | 本科生培养 |
| postgraduate_dev | 研究生培养 |
| undergraduate_recruit | 本科生招募 |
| postgraduate_recruit | 研究生招募 |
| CPC_build | 党的建设 |
| CPC_work | 党委工作 |
| union_work | 工会工作 |
| CYL_work | 共青团工作 |
| security_management | 安全管理 |
| alumni_style | 校友风采 |
</Route>

View File

@ -2473,7 +2473,7 @@ router.get('/haohaozhu/whole-house/:keyword?', lazyloadRouteHandler('./routes/ha
router.get('/haohaozhu/discover/:keyword?', lazyloadRouteHandler('./routes/haohaozhu/discover'));
// 东北大学
router.get('/neu/news/:type', lazyloadRouteHandler('./routes/universities/neu/news'));
// router.get('/neu/news/:type', lazyloadRouteHandler('./routes/universities/neu/news'));
// 快递100
router.get('/kuaidi100/track/:number/:id/:phone?', lazyloadRouteHandler('./routes/kuaidi100/index'));

View File

@ -1,61 +0,0 @@
const got = require('@/utils/got');
const cheerio = require('cheerio');
const baseUrl = 'https://neunews.neu.edu.cn';
module.exports = async (ctx) => {
const type = ctx.params.type;
const newsUrl = `${baseUrl}/${type}/list.htm`;
const response = await got({
method: 'get',
url: newsUrl,
});
const data = response.data;
const $ = cheerio.load(data);
const title = $('title').text();
const items = $('.column-news-list > .news_list > .news > .news_title').slice(0, 5).get();
const results = await Promise.all(
items.map(async (item) => {
const $ = cheerio.load(item);
const title = $('a').attr('title');
const url = baseUrl + $('a').attr('href');
const description = await ctx.cache.tryGet(url, async () => {
const result = await got.get(url);
const $ = cheerio.load(result.data);
// 处理部分格式
$('article')
.find('span')
.each(function () {
const temp = $(this).text();
$(this).replaceWith(temp);
});
$('article')
.find('div')
.each(function () {
const temp = $(this).html();
$(this).replaceWith(temp);
});
$('article').find('a').remove();
$('article')
.find('p')
.each(function () {
$(this).removeAttr('style');
$(this).removeAttr('class');
});
return $('article').html();
});
const result = {
title,
description,
link: url,
};
return Promise.resolve(result);
})
);
ctx.state.data = {
title: `东北大学新闻网-${title}`,
link: newsUrl,
item: results,
};
};

94
lib/v2/neu/bmie.js Normal file
View File

@ -0,0 +1,94 @@
const got = require('@/utils/got');
const cheerio = require('cheerio');
const { parseDate } = require('@/utils/parse-date');
const baseUrl = 'http://www.bmie.neu.edu.cn';
const map = {
news: 561,
academic: 562,
talent_development: 563,
international_exchange: 'gjjl3',
announcement: 564,
undergraduate_dev: 573,
postgraduate_dev: 574,
undergraduate_recruit: 'bks',
postgraduate_recruit: 574,
CPC_build: 556,
CPC_work: 576,
union_work: 577,
CYL_work: 'gqtgz',
security_management: 569,
alumni_style: 557,
};
module.exports = async (ctx) => {
let type = ctx.params.type;
if (typeof map[type] !== 'undefined') {
type = map[type];
}
const newsUrl = `${baseUrl}/${type}/list.htm`;
const response = await got(newsUrl);
const data = response.data;
const $ = cheerio.load(data);
const title = $('title').text();
const items = $('#subIndex > div.main_frame_sub > div.detail_sub > div > div > div > ul > li').slice(0, 7).get();
const results = await Promise.all(
items.map(async (item) => {
const $ = cheerio.load(item);
const title = $('a').attr('title');
const url = baseUrl + $('a').attr('href');
const data = await ctx.cache.tryGet(url, async () => {
const result = await got(url);
const $ = cheerio.load(result.data);
const info = $($('.ny_con p')[1]).text();
const s = info.search(/\d{4}-\d{2}-\d{2}/);
const date = info.substring(s, s + 10);
const au_start = info.indexOf('作者:') + 3;
const au_end = info.lastIndexOf('|');
const auhor = info.substring(au_start, au_end).trim();
$('.entry')
.find('span')
.each(function () {
const temp = $(this).text();
$(this).replaceWith(temp);
});
$('.entry')
.find('div')
.each(function () {
const temp = $(this).html();
$(this).replaceWith(temp);
});
$('.entry').find('a').remove();
$('.entry')
.find('p')
.each(function () {
$(this).removeAttr('style');
$(this).removeAttr('class');
});
$('.wp_art_adjoin').remove();
return { description: $('.entry').html(), date, author: auhor };
});
const description = data.description;
const pubDate = parseDate(data.date);
const author = data.author;
const result = {
title,
description,
link: url,
pubDate,
author,
};
return result;
})
);
ctx.state.data = {
title: `东北大学 医学与生物信息工程学院 ${title}`,
link: newsUrl,
item: results,
};
};

4
lib/v2/neu/maintainer.js Normal file
View File

@ -0,0 +1,4 @@
module.exports = {
'/bmie/:type': ['tennousuathena'],
'/news/:type': ['JeasonLau'],
};

44
lib/v2/neu/news.js Normal file
View File

@ -0,0 +1,44 @@
const got = require('@/utils/got');
const cheerio = require('cheerio');
const { parseDate } = require('@/utils/parse-date');
const baseUrl = 'https://neunews.neu.edu.cn';
module.exports = async (ctx) => {
const type = ctx.params.type;
const newsUrl = `${baseUrl}/${type}/list.htm`;
const response = await got(newsUrl);
const data = response.data;
const $ = cheerio.load(data);
const items = $('.column-news-list > .news_list > .news')
.toArray()
.map((item) => {
item = $(item);
return {
title: item.find('a').attr('title'),
link: new URL(item.find('a').attr('href'), baseUrl).href,
pubDate: parseDate(item.find('.news_meta').text(), 'YYYY-MM-DD'),
};
});
const results = await Promise.all(
items.map((item) =>
ctx.cache.tryGet(item.link, async () => {
const result = await got(item.link);
const $ = cheerio.load(result.data);
item.author = $('.arti-metas').text().split('更新日期')[0];
item.description = $('.article_content').html();
return item;
})
)
);
ctx.state.data = {
title: `东北大学新闻网-${$('head title').text()}`,
link: newsUrl,
item: results,
};
};

21
lib/v2/neu/radar.js Normal file
View File

@ -0,0 +1,21 @@
module.exports = {
'neu.edu.cn': {
_name: '东北大学',
neunews: [
{
title: '新闻网',
docs: 'https://docs.rsshub.app/university.html#dong-bei-da-xue',
source: ['/:type/list.htm'],
target: '/neu/news/:type',
},
],
'www.bmie': [
{
title: '学院新闻 - 医学与生物信息工程学院',
docs: 'https://docs.rsshub.app/university.html#dong-bei-da-xue',
source: ['/'],
target: '/neu/bmie/news',
},
],
},
};

4
lib/v2/neu/router.js Normal file
View File

@ -0,0 +1,4 @@
module.exports = (router) => {
router.get('/bmie/:type', require('./bmie'));
router.get('/news/:type', require('./news'));
};