Python 爬虫，爬取”百思不得姐“搞笑视频

最新推荐文章于 2024-04-07 08:10:55 发布

gf574723691

最新推荐文章于 2024-04-07 08:10:55 发布

阅读量1.4k

点赞数

分类专栏： Python

本文链接：https://blog.csdn.net/gf574723691/article/details/52756491

版权

Python 专栏收录该内容

3 篇文章 0 订阅

订阅专栏

# -*- coding:utf-8 -*-
import urllib
import re
import MySQLdb
import sys
import os
#print sys.getdefaultencoding()

urlz = 'http://www.budejie.com/video'


def video(i):
    html = urllib.urlopen(i).read()
    html = html.decode('utf-8')
    reg = r'data-text="(.*?)">'  # <a href="(.*?).mp4"'
    seg = r'<a href="(.*?)" target="_blank" download="" class="ipad-down-href" style="border:0;outline: none;">'
    result = re.findall(reg, html)
    goal = re.findall(seg, html)
    for i in range(len(goal)):
        Insert(result[i * 2 + 1], goal[i])
        path = r"D:\budejie\%s.mp4".encode('gb2312') %result[2 * i + 1]
        print path
        #os.makenod(path)
        try:
             urllib.urlretrieve(goal[i],path)
        except Exception as e:
            print e
        print result[i * 2 + 1] + '\t' + goal[i]

def Insert(name, url):
    conn = MySQLdb.connect('127.0.0.1', 'root', '123', 'game')
    cursor = conn.cursor()
    conn.set_character_set('utf8')
    cursor.execute('SET NAMES utf8;')
    cursor.execute('SET CHARACTER SET utf8;')
    cursor.execute('SET character_set_connection=utf8;')
    try:
        sql = 'insert into video_budejie values(%s,%s)'
        tmp = (name, url)
        cursor.execute(sql, tmp)
        conn.commit()
    except Exception as e:
        print e


lm = [urlz]
for i in range(1, 51):
    tmp = urlz + '/%d' % i
    lm.append(tmp)
for each in lm:
    print each
    video(each)