星标日志 - 运维进行时 - 互联网运维与架构

分页： 30/37

25 26 27 28 29 30 31 32 33 34

[ 显示模式：摘要 | 列表 ]

写一个爬虫程序[原创]

刘天斯 , 2010/03/08 10:33 , Python , 评论(3) , 阅读(35242) , Via 本站原创

写爬虫是一项复杂、枯噪、反复的工作，考虑的问题包括采集效率、链路异常处理、数据质量(与站点编码规范关系很大)等。整理自己写一个爬虫程序，单台服务器可以启用1~8个实例同时采集，然后将数据入库。

#-*- coding:utf-8 -*-
#!/usr/local/bin/python
import sys, time, os,string
import mechanize
import urlparse
from BeautifulSoup import BeautifulSoup
import re
import MySQLdb
import logging
import cgi
from optparse import OptionParser
#----------------------------------------------------------------------------#
# Name:        TySpider.py                                                            #
# Purpose:     WebSite  Spider Module                                         #
# Author:      刘天斯                                                                     #
# Email:       liutiansi@gamil.com                                                  #
# Created:     2010/02/16                                                            #
# Copyright:   (c) 2010                                                                #
#----------------------------------------------------------------------------#

"""
|--------------------------------------------------------------------------
| 定义 loging class;
|--------------------------------------------------------------------------
|
| 功能：记录系统相关日志信息。
| 
|
"""
class Pubclilog():
    def __init__(self):
        self.logfile = 'website_log.txt'

def iniLog(self):
        logger = logging.getLogger()
        filehandler = logging.FileHandler(self.logfile)
        streamhandler = logging.StreamHandler()
        fmt = logging.Formatter('%(asctime)s, %(funcName)s, %(message)s')
        logger.setLevel(logging.DEBUG) 
        logger.addHandler(filehandler) 
        logger.addHandler(streamhandler)
        return [logger,filehandler]

"""
|--------------------------------------------------------------------------
| 定义 tySpider class;
|--------------------------------------------------------------------------
|
| 功能：抓取分类、标题等信息
| 
|
"""
class BaseTySpider:

#初始化相关成员方法
    def __init__(self,X,log_switch):

#数据库连接
        self.conn = MySQLdb.connect(db='dbname',host='192.168.0.10', user='dbuser',passwd='SDFlkj934y5jsdgfjh435',charset='utf8')

#分类及标题页面Community
        self.CLASS_URL = 'http://test.abc.com/aa/CommTopicsPage?'

#发表回复页
        self.Content_URL = 'http://test.bac.com/aa/CommMsgsPage?'

#开始comm值
        self.X=X

#当前comm id取模，方面平均到表
        self.mod=self.X%5

#Community文件下载页
        self.body=""

#self.bodySoup对象
        self.soup=None

#发表回复页下载内容变量
        self.Contentbody=""

#发表回复页内容self.ContentbodySoup对象
        self.Contentsoup=None

#日志开关
        self.log_switch=log_switch

#======================获取名称及分类方法==========================
    def _SpiderClass(self,nextpage=None):
        if nextpage==None:
            FIXED_QUERY = 'cmm='+str(self.X)
        else:
            FIXED_QUERY = nextpage[1:]

try:
            rd = mechanize.Browser()
            rd.addheaders = [("User-agent", "Tianya/2010 (compatible; MSIE 6.0;Windows NT 5.1)")]
            rd.open(self.CLASS_URL + FIXED_QUERY)
            self.body=rd.response().read()
            #rd=mechanize.Request(self.CLASS_URL + FIXED_QUERY)
            #response = mechanize.urlopen(rd)
            #self.body=response.read()

except Exception,e:
            if self.log_switch=="on":
                logapp=Pubclilog()
                logger,hdlr = logapp.iniLog()
                logger.info(self.CLASS_URL + FIXED_QUERY+str(e))
                hdlr.flush()
                logger.removeHandler(hdlr)
                return
        self.soup = BeautifulSoup(self.body)
        NextPageObj= self.soup("a", {'class' : re.compile("fs-paging-item fs-paging-next")})
        self.cursor = self.conn.cursor()
        if nextpage==None:
            try:
                Ttag=str(self.soup.table)
                #print Ttag

"""
                ------------------分析结构体-----------------
                <table cellspacing="0" cellpadding="0">
                    <tr>
                        <td>
                            <h1 title="Dunhill">Dunhill</h1>
                        </td>
                       <td valign="middle">
                            <div class="fs-comm-cat">
                                <span class="fs-icons fs-icon-cat"> </span> <a href="TopByCategoryPage?cid=211&ref=commnav-cat">中国</a> » <a href="TopByCategoryPage?cid=211&subcid=273&ref=commnav-cat">人民</a>
                            </div>
                        </td>
                    </tr>
                </table>
                """

soupTable=BeautifulSoup(Ttag)

#定位到第一个h1标签
                tableh1 = soupTable("h1")
                #print self.X
                #print "Name:"+tableh1[0].string.strip().encode('utf-8')

#处理无类型的
                try:
                    #定位到表格中符合规则“^TopByCategory”A链接块，tablea[0]为第一个符合条件的连接文字，tablea[1]...
                    tablea = soupTable("a", {'href' : re.compile("^TopByCategory")})
                    if tablea[0].string.strip()=="":
                        pass
                    #print "BigCLass:"+tablea[0].string.strip().encode('utf-8')
                    #print "SubClass:"+tablea[1].string.strip().encode('utf-8')
                except Exception,e:
                    if self.log_switch=="on":
                        logapp=Pubclilog()
                        logger,hdlr = logapp.iniLog()
                        logger.info("[noClassInfo]"+str(self.X)+str(e))
                        hdlr.flush()
                        logger.removeHandler(hdlr)
                    self.cursor.execute("insert into baname"+str(self.mod)+" values('%d','%d','%s')" %(self.X,-1,tableh1[0].string.strip().encode('utf-8')))
                    self.conn.commit() 
                    self._SpiderTitle()
                    if NextPageObj:
                        NextPageURL=NextPageObj[0]['href']
                        self._SpiderClass(NextPageURL)
                        return
                    else:
                        return

#获取链接二对象的href值
                classlink=tablea[1]['href']
                par_dict=cgi.parse_qs(urlparse.urlparse(classlink).query)
                #print "CID:"+par_dict["cid"][0]
                #print "SubCID:"+par_dict["subcid"][0]
                #print "---------------------------------------"
                
                #插入数据库
                self.cursor.execute("insert into class values('%d','%s')" %(int(par_dict["cid"][0]),tablea[0].string.strip().encode('utf-8')))
                self.cursor.execute("insert into subclass values('%d','%d','%s')" %(int(par_dict["subcid"][0]),int(par_dict["cid"][0]),tablea[1].string.strip().encode('utf-8'))) 
                self.cursor.execute("insert into baname"+str(self.mod)+" values('%d','%d','%s')" %(self.X,int(par_dict["subcid"][0]),tableh1[0].string.strip().encode('utf-8')))
                self.conn.commit() 
                
                self._SpiderTitle()
                if NextPageObj:
                    NextPageURL=NextPageObj[0]['href']
                    self._SpiderClass(NextPageURL)

self.body=None
                self.soup=None
                Ttag=None
                soupTable=None
                table=None
                table1=None
                classlink=None
                par_dict=None
            except Exception,e:
                if self.log_switch=="on":
                    logapp=Pubclilog()
                    logger,hdlr = logapp.iniLog()
                    logger.info("[ClassInfo]"+str(self.X)+str(e))
                    hdlr.flush()
                    logger.removeHandler(hdlr)
        else:
            self._SpiderTitle()
            if NextPageObj:
                NextPageURL=NextPageObj[0]['href']
                self._SpiderClass(NextPageURL)

#====================获取标题方法=========================
    def _SpiderTitle(self):
        #查找标题表格对象(table)
        soupTitleTable=self.soup("table", {'class' : "fs-topic-list"})
        
        #查找标题行对象(tr)
        TitleTr = soupTitleTable[0]("tr", {'onmouseover' : re.compile("^this\.className='fs-row-hover'")})
        """
        -----------分析结构体--------------
        <tr class="fs-alt-row" onmouseover="this.className='fs-row-hover'" onmouseout="this.className='fs-alt-row'">
            <td valign="middle" class="fs-hot-topic-dots-ctn">
                <div class="fs-hot-topic-dots" style="background-position:0 -0px" title="点击量：12"></div>
            </td>
            <td valign="middle" class="fs-topic-name">
                <a href="CommMsgsPage?cmm=16081&tid=2718969307756232842&ref=regulartopics" id="a53" title="【新人报到】欢迎美国人民加入" target="_blank">【新人报到】欢迎美国人民加入</a>
                <span class="fs-meta">
                <span class="fs-icons fs-icon-mini-reply"> </span>0
                  /
                  <span class="fs-icons fs-icon-pageview"> </span>12</span>
            </td>
            <td valign="middle">
                <a class="fs-tiny-user-avatar umhook " href="ProfilePage?uid=8765915421039908242" title="中国人"><img src="http://img1.sohu.com.cn/aa/images/138/0/P/1/s.jpg" /></a>
            </td>
            <td valign="middle" style="padding-left:4px">
                <a href="Profile?uid=8765915421039908242" id="b53" title="中国人" class="umhook">中国人</a>
            </td>
            <td valign="middle" class="fs-topic-last-mdfy fs-meta">2-14</td>
        </tr>
        """
        for CurrTr in TitleTr:
            try:
                #初始化置顶及精华状态
                Title_starred='N'
                Title_sticky='N'

#获取当前记录的BeautifulSoup对象
                soupCurrTr=BeautifulSoup(str(CurrTr))

#BeautifulSoup分析HTML有误，只能通过span的标志数来获取贴子状态，会存在一定误差
                #如只有精华时也会当成置顶来处理。
                TitleStatus=soupCurrTr("span", {'title' : ""})
                TitlePhotoViewer=soupCurrTr("a", {'href' : re.compile("^PhotoViewer")})
                if TitlePhotoViewer.__len__()==1:
                    TitlePhotoViewerBool=0
                else:
                    TitlePhotoViewerBool=1
                if TitleStatus.__len__()==3-TitlePhotoViewerBool:
                    Title_starred='Y'
                    Title_sticky='Y'
                elif TitleStatus.__len__()==2-TitlePhotoViewerBool:
                    Title_sticky='Y'

#获取贴子标题
                Title=soupCurrTr.a.next.strip()

#获取贴子ID
                par_dict=cgi.parse_qs(urlparse.urlparse(soupCurrTr.a['href']).query)

#获取回复数及浏览器
                TitleNum=soupCurrTr("td", {'class' : "fs-topic-name"})
                TitleArray=string.split(str(TitleNum[0]),'\n')
                Title_ReplyNum=string.split(TitleArray[len(TitleArray)-4],'>')[2]
                Title_ViewNum=string.split(TitleArray[len(TitleArray)-2],'>')[2][:-6]

#获取贴子作者
                TitleAuthorObj=soupCurrTr("td", {'style' : "padding-left:4px"})
                Title_Author=TitleAuthorObj[0].next.next.next.string.strip().encode('utf-8')

#获取回复时间
                TitleTime=soupCurrTr("td", {'class' : re.compile("^fs-topic-last-mdfy fs-meta")})

"""
                print "X:"+str(self.X)
                print "Title_starred:"+Title_starred
                print "Title_sticky:"+Title_sticky
                print "Title:"+Title

#获取贴子内容连接URL
                print "Title_link:"+soupCurrTr.a['href']
                print "CID:"+par_dict["tid"][0]
                print "Title_ReplyNum:"+Title_ReplyNum
                print "Title_ViewNum:"+Title_ViewNum
                print "Title_Author:"+Title_Author
                print "TitleTime:"+TitleTime[0].string.strip().encode('utf-8')
                """

#入库
                self.cursor.execute("insert into Title"+str(self.mod)+" values('%s','%d','%s','%d','%d','%s','%s','%s','%s')" %(par_dict["tid"][0], \
                                                                         self.X,Title,int(Title_ReplyNum),int(Title_ViewNum),Title_starred,Title_sticky, \
                                                                         Title_Author.decode('utf-8'),TitleTime[0].string.strip().encode('utf-8')))
                self.conn.commit()
                self._SpiderContent(par_dict["tid"][0])
            except Exception,e:
                if self.log_switch=="on":
                    logapp=Pubclilog()
                    logger,hdlr = logapp.iniLog()
                    logger.info("[Title]"+str(self.X)+'-'+par_dict["tid"][0]+'-'+str(e))
                    hdlr.flush()
                    logger.removeHandler(hdlr)

#======================获取发表及回复方法=======================
    def _SpiderContent(self,ID,nextpage=None):
        if nextpage==None:
            FIXED_QUERY = 'cmm='+str(self.X)+'&tid='+ID+'&ref=regulartopics'
        else:
            FIXED_QUERY = nextpage[9:]

rd = mechanize.Browser()
        rd.addheaders = [("User-agent", "Tianya/2010 (compatible; MSIE 6.0;Windows NT 5.1)")]
        rd.open(self.Content_URL + FIXED_QUERY)
        self.Contentbody=rd.response().read()

#rd=mechanize.Request(self.Content_URL + FIXED_QUERY)
        #response = mechanize.urlopen(rd)
        #self.Contentbody=response.read()

self.Contentsoup = BeautifulSoup(self.Contentbody)
        NextPageObj= self.Contentsoup("a", {'class' : re.compile("fs-paging-item fs-paging-next")})
        try:
            Tdiv=self.Contentsoup("div", {'class' : "fs-user-action"})
            i=0
            for Currdiv in Tdiv:
                if i==0:
                    Ctype='Y'
                else:
                    Ctype='N'
                #发表时间
                soupCurrdiv=BeautifulSoup(str(Currdiv))
                PosttimeObj=soupCurrdiv("span", {'class' : "fs-meta"})
                Posttime=PosttimeObj[0].next[1:]
                Posttime=Posttime[0:-3]
                #IP地址
                IPObj=soupCurrdiv("a", {'href' : re.compile("CommMsgAddress")})
                if IPObj:
                    IP=IPObj[0].next.strip()
                else:
                    IP=''

#发表／回复内容
                ContentObj=soupCurrdiv("div", {'class' :"fs-user-action-body"})
                Content=ContentObj[0].renderContents().strip()
                """
                print "ID:"+str(self.X)
                print "ID:"+ID
                print "Ctype:"+Ctype
                print "POSTTIME:"+Posttime
                print "IP:"+IP
                print "Content:"+Content
                """

self.cursor.execute("insert into Content"+str(self.mod)+" values('%s','%d','%s','%s','%s','%s')" %(ID,self.X,Ctype,Posttime,IP,Content.decode('utf-8')))
                self.conn.commit() 
                i+=1

except Exception,e:
            if self.log_switch=="on":
                logapp=Pubclilog()
                logger,hdlr = logapp.iniLog()
                logger.info("[Content]"+str(self.X)+'-'+ID+'-'+str(e))
                hdlr.flush()
                logger.removeHandler(hdlr)

#如“下一页”有链接刚继续遍历
        if NextPageObj:
            NextPageURL=NextPageObj[0]['href']
            self._SpiderContent(ID,NextPageURL)

def __del__(self):
        try:
            self.cursor.close()
            self.conn.close()
        except Exception,e:
            pass

#遍历comm范围
def initapp(StartValue,EndValue,log_switch):
    for x in range(StartValue,EndValue): 
        app=BaseTySpider(x,log_switch)
        app._SpiderClass()
        app=None

if __name__ == "__main__":
    
    #定义命令行参数
    MSG_USAGE = "TySpider.py [ -s StartNumber EndNumber ] -l [on|off] [-v][-h]"
    parser = OptionParser(MSG_USAGE)
    parser.add_option("-s", "--set", nargs=2,action="store", 
            dest="comm_value",
            type="int",
            default=False, 
            help="配置名称ID值范围。".decode('utf-8'))
    parser.add_option("-l", "--log", action="store", 
            dest="log_switch",
            type="string",
            default="on", 
            help="错误日志开关".decode('utf-8'))
    parser.add_option("-v","--version", action="store_true", dest="verbose",
            help="显示版本信息".decode('utf-8'))
    opts, args = parser.parse_args()

if opts.comm_value:
        if opts.comm_value[0]>opts.comm_value[1]:
            print "终止值比起始值还小？"
            exit();
        if opts.log_switch=="on":
            log_switch="on"
        else:
            log_switch="off"
        initapp(opts.comm_value[0],opts.comm_value[1],log_switch)
        exit();

if opts.verbose:
        print "WebSite Scider V1.0 beta."
        exit;

如大家有什么疑问或感兴趣的话题可以通过weibo与我交流：http://t.qq.com/yorkoliu

Tags: 爬虫 , python

排查Centos与ubuntu系统awk运行效率差异[原创]

刘天斯 , 2010/03/05 18:02 , Linux , 评论(4) , 阅读(31147) , Via 本站原创

同一个分析日志的awk脚本，在Centos、ubuntu操作系统运行效率存在巨大差异。即在ubuntu中只需1分钟，而在Centos中则需要20分钟。以下为我的排查步骤：
1、检查服务器内核版本

引用

拿一台升级过最新Linux内核(2.6.32.3)的CentOS5.4服务器来测试，结果还是没有改善。

2、检查内核ulimit参数

引用

在Centos服务器调整所有ulimit参数与ubuntu系统一致，结果还是一样。

3、优化awk脚本

引用

由于mawk与gawk部分语法上存在差异，如将转义符‘\’换成'\\'，双引号换成单引号，依然没有效果。

4、检查gawk版本

引用

      检查两个系统的gawk版本，发现所有Centos版本默认自带的gawk都低于ubuntu系统自带的3.1.6，尝试在Centos服务器下载、安装源码gawk3.1.6，结果速度提升了20倍。测试结果如下：
       time /usr/local/bin/gawk -f test.awk access.log >"temp.txt"
       real    0m48.739s
       user    0m42.904s
       sys     0m5.389s

分页： 30/37

25 26 27 28 29 30 31 32 33 34

[ 显示模式：摘要 | 列表 ]

运维进行时

写一个爬虫程序[原创]

排查Centos与ubuntu系统awk运行效率差异[原创]

个人简介

分类

日历

最新日志

最新评论

链接

归档

其他

< 2024 > < 12 >
日	一	二	三	四	五	六
1	2	3	4	5	6	7
8	9	10	11	12	13	14
15	16	17	18	19	20	21
22	23	24	25	26	27	28
29	30	31