gistfile1.py

import requests
from bs4 import BeautifulSoup

def PageCal(Page):
	return (Page-1)*25

UrlBase0='http://book.douban.com/subject/'
UrlBase1='/reviews'
UrlBase2='?start='


def GetReviews(BookID):
	Url=UrlBase0+BookID+UrlBase1
	#<div class="paginator">
	try:
		Page=requests.get(Url)
	except:
		pass
	else:
		PageSoup=BeautifulSoup(Page.content)
		PageNumber=PageSoup.find('div',{'class':'paginator'}).findAll('a')[-2].text
		for i in range(1,int(PageNumber)+1):
			#<div class="ctsh">
			try:
				nPage=requests.get(UrlBase0+BookID+UrlBase1+UrlBase2+str(PageCal(i)))
			except:
				pass
			else:
				nPageSoup=BeautifulSoup(nPage.content)
				Comment=nPageSoup.findAll('div',{'class':'ctsh'})
				for CommentItem in Comment:
					print CommentItem

GetReviews('26317953')