feat: crawl 36kr newsflashes (#2984)

* crawl 36kr newsflashes

* xueqiu hot statuses
This commit is contained in:
hillerliao
2019-09-03 12:05:08 +08:00
committed by DIYgod
parent ac80164099
commit b4183a2510
7 changed files with 95 additions and 6 deletions

View File

@@ -6,6 +6,10 @@ pageClass: routes
## 36kr
### 快讯
<Route author="hillerliao" example="/36kr/newsflashes" path="/36kr/newsflashes" />
### 搜索文章
<Route author="xyqfer kt286" example="/36kr/search/article/ofo" path="/36kr/search/article/:keyword" :paramsDesc="['关键字']" />

View File

@@ -751,6 +751,10 @@ pageClass: routes
</Route>
### 热帖
<Route author="hillerliao" example="/xueqiu/hots" path="/xueqiu/hots"/>
## 一亩三分地
### 主题帖

View File

@@ -379,6 +379,7 @@ router.get('/xueqiu/user_stock/:id', require('./routes/xueqiu/user_stock'));
router.get('/xueqiu/fund/:id', require('./routes/xueqiu/fund'));
router.get('/xueqiu/stock_info/:id/:type?', require('./routes/xueqiu/stock_info'));
router.get('/xueqiu/snb/:id', require('./routes/xueqiu/snb'));
router.get('/xueqiu/hots', require('./routes/xueqiu/hots'));
// Greasy Fork
router.get('/greasyfork/:language/:domain?', require('./routes/greasyfork/scripts'));
@@ -973,6 +974,7 @@ router.get('/tingdiantz/nanjing', require('./routes/tingdiantz/nanjing'));
// 36kr
router.get('/36kr/search/article/:keyword', require('./routes/36kr/search/article'));
router.get('/36kr/newsflashes', require('./routes/36kr/newsflashes'));
// icourse163
router.get('/icourse163/newest', require('./routes/icourse163/newest'));

View File

@@ -0,0 +1,40 @@
const got = require('@/utils/got');
module.exports = async (ctx) => {
const response = await got({
method: 'get',
url: `https://36kr.com/api/newsflash?per_page=30`,
headers: {
'Content-Type': 'application/json; charset=utf-8',
},
});
const newsflashes = response.data.data.items;
let newsflashesList = [];
for (let i = 0; i < newsflashes.length; i++) {
newsflashesList = newsflashesList.concat(newsflashes[i]);
}
const out = newsflashesList.map((item) => {
const date = item.published_at;
const link = item.news_url;
const title = item.title;
const description = item.description;
const single = {
title,
link,
pubDate: new Date(date).toUTCString(),
description,
};
return single;
});
ctx.state.data = {
title: `快讯 - 36氪`,
link: `https://36kr.com/newsflashes`,
item: out,
};
};

View File

@@ -30,7 +30,7 @@ module.exports = async (ctx) => {
let name = '';
const out = announcementsList.map((item) => {
const date = item.announcementTime;
const link = `http://www.cninfo.com.cn/new/disclosure/detail?announcementId=${item.announcementId}`;
const link = `http://static.cninfo.com.cn/${item.adjunctUrl}`;
name = item.secName;
const title = name + ': ' + item.announcementTitle;

42
lib/routes/xueqiu/hots.js Normal file
View File

@@ -0,0 +1,42 @@
const got = require('@/utils/got');
module.exports = async (ctx) => {
const res1 = await got({
method: 'get',
url: 'https://xueqiu.com/',
});
const token = res1.headers['set-cookie'].find((s) => s.startsWith('xq_a_token=')).split(';')[0];
const res2 = await got({
method: 'get',
url: 'https://xueqiu.com/statuses/hots.json',
params: {
a: '1',
count: '10',
page: '1',
scope: 'day',
type: 'status',
meigu: '0',
},
headers: {
Cookie: token,
Referer: `https://xueqiu.com/`,
},
});
const data = res2.data;
ctx.state.data = {
title: `热帖 - 雪球`,
link: `https://xueqiu.com/`,
description: `雪球热门帖子`,
item: data.map((item) => {
const description = item.text;
return {
title: item.title ? item.title : description,
description: item.text,
pubDate: new Date(item.created_at).toUTCString(),
link: `https://xueqiu.com${item.target}`,
};
}),
};
};

View File

@@ -11,16 +11,13 @@ module.exports = async (ctx) => {
const res2 = await got({
method: 'get',
url: 'https://stock.xueqiu.com/v5/stock/portfolio/stock/list.json?category=1&size=1000',
params: {
uid: id,
},
url: `https://stock.xueqiu.com/v5/stock/portfolio/stock/list.json?category=1&size=1000&uid=${id}`,
headers: {
Cookie: token,
Referer: `https://xueqiu.com/u/${id}`,
},
});
const data = res2.data.stocks;
const data = res2.data.data.stocks;
const res3 = await got({
method: 'get',