#coding=utf-8#__author__ = 'Abel'
importurllibimporturllib2importre#处理页面标签类
classTool:#去除img标签,7位长空格
removeImg = re.compile('| {7}|')#删除超链接标签
removeAddr = re.compile('|')#把换行的标签换为
replaceLine = re.compile('
replaceTD= re.compile('
')#把段落开头换为 加空两格replacePara = re.compile('
')#将换行符或双换行符替换为replaceBR = re.compile('
|
')#将其余标签剔除
removeExtraTag = re.compile('<.*?>')defreplace(self,x):
x= re.sub(self.removeImg,"",x)
x= re.sub(self.removeAddr,"",x)
x= re.sub(self.replaceLine," ",x)
x= re.sub(self.replaceTD," ",x)
x= re.sub(self.replacePara," ",x)
x= re.sub(self.replaceBR," ",x)
x= re.sub(self.removeExtraTag,"",x)#strip()将前后多余内容删除
returnx.strip()#百度贴吧爬虫类
classBDTB:#初始化,传入基地址,是否只看楼主的参数
def __init__(self,baseUrl,seeLZ,floorTag):#base链接地址
self.baseURL =baseUrl#是否只看楼主
self.seeLZ = '?see_lz='+str(seeLZ)#HTML标签剔除工具类对象
self.tool =Tool()#全局file变量,文件写入操作对象
self.file =None#楼层标号,初始为1
self.floor = 1
#默认的标题,如果没有成功获取到标题的话则会用这个标题
self.defaultTitle = u"百度贴吧"
#是否写入楼分隔符的标记
self.floorTag =floorTag#传入页码,获取该页帖子的代码
defgetPage(self,pageNum):try:#构建URL
url = self.baseURL+ self.seeLZ + '&pn=' +str(pageNum)
request=urllib2.Request(url)
response=urllib2.urlopen(request)#返回UTF-8格式编码内容
return response.read().decode('utf-8')#无法连接,报错
excepturllib2.URLError, e:if hasattr(e,"reason"):print u"连接百度贴吧失败,错误原因",e.reasonreturnNone#获取帖子标题
defgetTitle(self,page):#得到标题的正则表达式
pattern = re.compile('
(.*?)
',re.S)result=re.search(pattern,page)ifresult:#如果存在,则返回标题
return result.group(1).strip()else:returnNone#获取帖子一共有多少页
defgetPageNum(self,page):#获取帖子页数的正则表达式
pattern = re.compile('
.*?(.*?)',re.S)result=re.search(pattern,page)ifresult:return result.group(1).strip()else:returnNone#获取每一层楼的内容,传入页面内容
defgetContent(self,page):#匹配所有楼层的内容
pattern = re.compile('
items=re.findall(pattern,page)
contents=[]for item initems:#将文本进行去除标签处理,同时在前后加入换行符
content = " "+self.tool.replace(item)+" "contents.append(content.encode('utf-8'))returncontentsdefsetFileTitle(self,title):#如果标题不是为None,即成功获取到标题
if title is notNone:
self.file= open(title + ".txt","w+")else:
self.file= open(self.defaultTitle + ".txt","w+")defwriteData(self,contents):#向文件写入每一楼的信息
for item incontents:if self.floorTag == '1':#楼之间的分隔符
floorLine = " " + str(self.floor) + u"----------------------------------------------------------------------------------------- "self.file.write(floorLine)
self.file.write(item)
self.floor+= 1
defstart(self):
indexPage= self.getPage(1)
pageNum=self.getPageNum(indexPage)
title=self.getTitle(indexPage)
self.setFileTitle(title)if pageNum ==None:print "URL已失效,请重试"
return
try:print "该帖子共有" + str(pageNum) + "页"
for i in range(1,int(pageNum)+1):print "正在写入第" + str(i) + "页数据"page=self.getPage(i)
contents=self.getContent(page)
self.writeData(contents)#出现写入异常
exceptIOError,e:print "写入异常,原因" +e.messagefinally:print "写入任务完成"
print u"请输入帖子代号"baseURL= 'http://tieba.baidu.com/p/3560761301' + str(raw_input(u'http://tieba.baidu.com/p/3560761301'))
seeLZ= raw_input("是否只获取楼主发言,是输入1,否输入0 ")
floorTag= raw_input("是否写入楼层信息,是输入1,否输入0 ")
bdtb=BDTB(baseURL,seeLZ,floorTag)
bdtb.start()