标签：比特币 .python3.6 行业资讯
```#!/usr/bin/env python
# -*- coding:utf-8 -*-
#Author: Mr Gee
import requests
import urllib
from urllib import request
import time
from lxml import etree
import re
import pymysql
from gevent import monkey;monkey.patch_all()
import gevent
class BithCoin:
def __init__(self):
self.user_agent = ‘Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/62.0.3202.89 Safari/537.36‘
self.header = {‘User-Agent‘: self.user_agent}
def get_html(self,url):
html = request.Request(url,headers=self.header)
respon = request.urlopen(html)
page = respon.read().decode(‘utf-8‘)
return page
def get_title(self,text):
# selector = etree.HTML(text)
title = re.compile(‘<h1 class="subject">(.*?)</h1>‘)
title_text = re.findall(title,text)
self.titles = ‘‘.join(title_text)
return self.titles
def get_content(self,text):
content = re.compile(‘<div class="article-content">(.*?)</div>‘,re.S)
content_text = re.findall(content,text)
self.contents= ‘‘.join(content_text).replace(" ","")
return self.contents
def get_author(self,text):
author = re.compile(‘<div class="mark">.*?<span class="author">(.*?)</span>‘,re.S)
author_text = re.findall(author,text)
self.author = ‘‘.join(author_text)
return self.author
# return self.‘‘.join(author_text)
def get_date(self,text):
date = re.compile(‘ <span class="date">(.*?)</span>‘,re.S)
date_text = re.findall(date,text)
self.dates = ‘‘.join(date_text)
# print(type(self.dates))
return self.dates
def my_db(self,end_url):
conn = pymysql.connect(host="localhost",user="root",passwd="123456",db="ljj",charset="utf8")
print(‘链接上了‘)
db = conn.cursor()
insert_data = ("INSERT INTO bitcoin(url,post_author,post_date,post_content,post_title)" "VALUES(%s,%s,%s,%s,%s)")
data_value = (end_url,self.author,self.dates,self.contents,self.titles)
# data_value = (end_url,‘ ‘,‘ 2017-03-17‘,‘ ‘,‘ ‘)
try:
db.execute(insert_data,data_value)
conn.commit()
except:
conn.rollback()
print (‘已经结束‘)
# data_colum=("INSERT INTO ljj(id,url)" "VALUES(%s,%s)")
# data = (‘https://www.5186.net/article-show-id-1‘,end_url)
# db.execute(data_colum,data)
db.close()
conn.close()
if __name__ == ‘__main__‘:
Coin = BithCoin()
for i in range(1,1000,1):
base_url = ‘https://www.5186.net/article-show-id-‘ + ‘%s‘%(i)
print(base_url)
# print(base_url)
end_html = Coin.get_html(base_url)
task = [gevent.spawn(Coin.get_author,end_html),gevent.spawn(Coin.get_date,end_html),gevent.spawn(Coin.get_content,end_html),gevent.spawn(Coin.get_title,end_html),gevent.spawn(Coin.my_db,base_url)]
gevent.joinall(task)
标签：比特币 .python3.6 行业资讯
原文地址：http://blog.51cto.com/9694110/2084287
