1.[代碼][Python]代碼

#-*- coding: UTF-8 -*-

#Python UTF-8 抓取百度關(guān)鍵詞V1.0

#key.txt是抓取文件配置

#author PHPer.yang@gmail.com

import cgi,urllib #URL讀取

import re #正則匹配

import MySQLdb #MySQL

import datetime #時(shí)間

#import time,thread #多線程

"""

MySQL表結(jié)構(gòu)

CREATE TABLE `baidu` (

`id` int(10) unsigned NOT NULL auto_increment,

`url` varchar(200) NOT NULL,

`title` varchar(600) NOT NULL,

`keys` varchar(100) NOT NULL,

`bdurl` varchar(200) NOT NULL,

`date` date NOT NULL,

PRIMARY KEY (`id`)

) ENGINE=MyISAM DEFAULT CHARSET=utf8 AUTO_INCREMENT=1 ;

"""

def Yang_Config ():

fp = open('key.txt','r')

for line in fp.read().split('@'):

word = line.split(',') #word 是字典

#for item in word :

#print item.encode("UTF-8")

#print '------'

if len(word) > 1:

yang_u = word[0]

yang_k = word[1]

Yang_Spider(yang_u,yang_k)

#抓取頁(yè)面開(kāi)始

def Yang_Spider(yang_u,yang_k):

url = 'http://www.baidu.com/s?wd=%s+site:%s&&rn=100'% (yang_k,yang_u)

print url

fp = urllib.urlopen(url).read()

#print fp re.search

m = re.findall(r"

()?(.*?)\s*?.*?.*? ((\d{4}\-\d{1,2}\-\d{1,2})|(\d+小時(shí)前)|(\d+分鐘前)) .*?.*? ",fp) if m: #print m # for s in m:#數(shù)組抓取過(guò)來(lái)是gbk 轉(zhuǎn)碼成utf8.encode("UTF-8") 是漢字decode('gbk') ASNII轉(zhuǎn)UTF8 入數(shù)據(jù)庫(kù)操作print str(s[3]) # print '~~~'.join(s) #切割數(shù)組 Yang_MySQL (yang_k,yang_u,s) #入庫(kù) #for i, s in enumerate(m.group(3)): #print i,s else: print 'not search' def Yang_MySQL (k,u,s): global cursor,d cursor.execute("set names utf8") key_unicode = s[3].decode('gb2312') #gb2312 key_utf8 = key_unicode.encode('utf-8') SQL = " INSERT INTO `baidukey`.`baidu` (`url` ,`title` ,`keys` ,`bdurl` ,`date`) VALUES ('%s', '%s', '%s','%s','%s'); " % (s[2],key_utf8,k,u,d) insert = cursor.execute(SQL) #print SQL #運(yùn)行抓取函數(shù) conn = MySQLdb.connect(host="localhost",user="phper",passwd="123456",db="baidukey") cursor = conn.cursor() t = datetime.datetime.now() d = t.strftime('%Y-%m-%d')#%H:%M:%S Del = " DELETE FROM `baidukey`.`baidu` WHERE date = '%s'" % (d) cursor.execute(Del) Yang_Config()

總結(jié)

以上是生活随笔為你收集整理的python关键词大全_Python 批量获取Baidu关键词的排名并入库的全部?jī)?nèi)容，希望文章能夠幫你解決所遇到的問(wèn)題。

如果覺(jué)得生活随笔網(wǎng)站內(nèi)容還不錯(cuò)，歡迎將生活随笔推薦給好友。