简介
朋友问我能不能做一个下载他在豆瓣读书上的短评的工具,于是就做了这个“豆瓣用户读书短评下载工具”。
GitHub链接:https://github.com/xiaff/dbc-downloader。
这个小工具使用Python3.4编写,其工作流程为:
- 用户输入其豆瓣ID;
- 抓取用户评论列表网页
- 对网页进行解析;
- 存储评论相关信息;
- 将Markdown格式文件转换为Html。
用到的库主要有:
- urllib.request
- BeautifulSoup4
- markdown
抓取网页
所需要抓取的信息在这样的网页中:http://book.douban.com/people/ahbei/collect?sort=time&start=0&filter=all&mode=grid&tags_sort=count,URL中包含了用户ID(people/
之后)、评论序号(start=
)等信息。
url_1=‘http://book.douban.com/people/‘
url_2=‘/collect?sort=time&start=‘
url_3=‘&filter=all&mode=grid&tags_sort=count‘
url=url_1+uId+url_2+index+url_3
,其中 UID 为豆瓣用户ID,index 为评论序号。评论序号从0开始编号,每页显示15条,因为每个url中的序号依次为0、15、30……15*i
。 i的最大值即为 网页页数-1,在解析第一张网页的时候可以获取页数。
在抓取网页的时候可以选择使用代理服务器,因此使用urllib.request设置代理:
proxyInfo=input(‘Please type in your HTTP Proxy: ‘)
proxySupport=urllib.request.ProxyHandler({‘http‘:proxyInfo})
opener=urllib.request.build_opener(proxySupport)
urllib.request.install_opener(opener)
不过,如果只设置了代理就访问豆瓣的用户读书评论列表,豆瓣会返回403 Forbidden
。
解决办法就是添加请求标头(Request Headers)来模拟浏览器访问。 标头信息可以在浏览器中打开网页时按F12进入控制台,在Network选项卡中找到 请求标头(Request Headers) 。
比如,这是我在Edge浏览器中访问豆瓣的请求标头。
head= { ‘Accept‘:‘text/html, application/xhtml+xml, image/jxr, */*‘, ‘Accept-Language‘: ‘zh-Hans-CN, zh-Hans; q=0.5‘, ‘Connection‘:‘Keep-Alive‘, ‘Cookie‘:‘bid=lkpO8Id/Kbs; __utma=30149280.1824146216.1438612767.1440248573.1440319237.13; __utmz=30149280.1438612767.1.1.utmcsr=(direct)|utmccn=(direct)|utmcmd=(none); as=http://book.douban.com/people/133476248/; ll=108288; viewed=26274009_1051580; ap=1; ps=y; ct=y; __utmb=30149280.23.10.1440319237; __utmc=30149280; __utmt_douban=1; _pk_id.100001.3ac3=b288f385b4d73e38.1438657126.3.1440319394.1440248628.; __utma=81379588.142106303.1438657126.1440248573.1440319240.3; __utmz=81379588.1440319240.3.2.utmcsr=movie.douban.com|utmccn=(referral)|utmcmd=referral|utmcct=/; _pk_ses.100001.3ac3=*; __utmb=81379588.23.10.1440319240; __utmt=1; __utmc=81379588; _pk_ref.100001.3ac3=%5B%22%22%2C%22%22%2C1440319240%2C%22http%3A%2F%2Fmovie.douban.com%2F%22%5D‘, ‘Host‘:‘book.douban.com‘, ‘User-Agent‘:‘Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/42.0.2311.135 Safari/537.36 Edge/12.10240‘}
然后在访问网页的时候加上header信息:
full_url=urllib.request.Request(url,headers=head) response=urllib.request.urlopen(full_url) html=response.read()
这样就可以正确抓取到网页内容了。
解析网页
在我之前一片文章[Python]从豆瓣电影批量获取看过某部电影的用户列表讲过了使用 BeautifulSoup 解析网页了,其实只要看看官方文档就很容易上手了。 这里就不再赘述了。
Markdown转Html
最后一步是将以Markdown格式保存的文件转换成Html文件,这样可以让不熟悉Markdown的人在浏览器中直接查看或者另存为PDF文件。markdown
包可以做到这一点:
md = markdown.markdown(contents)
html = ‘<html><meta charset="UTF-8">‘
html+=‘<title>‘+title+‘</title>‘
html += "<body>" + md + "</body></html>"
md = markdown.markdown(contents)
转换出来的md是不包含<html>
标签的,因此需要自己加上这些标签后再保存。
源代码
#coding=utf-8 #Python 3.4 ##从豆瓣网页中得到用户的所有读书短评 ##网页地址类型:http://book.douban.com/people/1000001/collect?sort=time&start=0&filter=all&mode=grid&tags_sort=count ## http://book.douban.com/people/1000001/collect?sort=time&start=15&filter=all&mode=grid&tags_sort=count from bs4 import BeautifulSoup import time import urllib.request,urllib.parse from urllib.error import URLError,HTTPError import os import markdown #换行符 lineSep=‘\n‘ #设置HTTP代理 ans=input(‘Do you want to use a HTTP Proxy (N/y)? ‘) ans=ans.lower() if ans==‘y‘ or ans==‘yes‘: print(‘HTTP Proxy formart: IP:PORT \nExample: 127.0.0.1:80‘) print(‘Do NOT contain any unnecessary characters.‘) proxyInfo=input(‘Please type in your HTTP Proxy: ‘) proxySupport=urllib.request.ProxyHandler({‘http‘:proxyInfo}) opener=urllib.request.build_opener(proxySupport) urllib.request.install_opener(opener) else: pass #头信息 head= { ‘Accept‘:‘text/html, application/xhtml+xml, image/jxr, */*‘, ‘Accept-Language‘: ‘zh-Hans-CN, zh-Hans; q=0.5‘, ‘Connection‘:‘Keep-Alive‘, ‘Cookie‘:‘bid=lkpO8Id/Kbs; __utma=30149280.1824146216.1438612767.1440248573.1440319237.13; __utmz=30149280.1438612767.1.1.utmcsr=(direct)|utmccn=(direct)|utmcmd=(none); as=http://book.douban.com/people/133476248/; ll=108288; viewed=26274009_1051580; ap=1; ps=y; ct=y; __utmb=30149280.23.10.1440319237; __utmc=30149280; __utmt_douban=1; _pk_id.100001.3ac3=b288f385b4d73e38.1438657126.3.1440319394.1440248628.; __utma=81379588.142106303.1438657126.1440248573.1440319240.3; __utmz=81379588.1440319240.3.2.utmcsr=movie.douban.com|utmccn=(referral)|utmcmd=referral|utmcct=/; _pk_ses.100001.3ac3=*; __utmb=81379588.23.10.1440319240; __utmt=1; __utmc=81379588; _pk_ref.100001.3ac3=%5B%22%22%2C%22%22%2C1440319240%2C%22http%3A%2F%2Fmovie.douban.com%2F%22%5D‘, ‘Host‘:‘book.douban.com‘, ‘User-Agent‘:‘Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/42.0.2311.135 Safari/537.36 Edge/12.10240‘} #url=url_1+uId+url_2+Index+url_3 url_1=‘http://book.douban.com/people/‘ url_2=‘/collect?sort=time&start=‘ url_3=‘&filter=all&mode=grid&tags_sort=count‘ def is_chinese(uchar): """判断一个unicode是否是汉字 """ if uchar >= u‘\u4e00‘ and uchar<=u‘\u9fa5‘: return True else: return False def isChineseBook(title): """判断书名是否为中文书名 """ for c in title: if(is_chinese(c)): return True return False def getHtml(url): """返回指定的网页内容 """ print(‘Loading: ‘+url+‘......‘) full_url=urllib.request.Request(url,headers=head) TRY_TIMES=3 response=None while TRY_TIMES>0 and response==None : TRY_TIMES-=1 try: response=urllib.request.urlopen(full_url) #open=urlopen except HTTPError as e: print(‘HTTP Error:‘,e.code) except URLError as e: print(‘URL Error: ‘,e.reason) if response==None: print(‘Error!‘) os.system("pause") exit() html=response.read() return html def getBookComment(html): """解析网页并返回5个列表: 书名,出版信息,标记日期,标签,评论 """ titleList=[] #书名 pubList=[] #出版信息 dateList=[] #标记日期 tagsList=[] #标签 commentList=[] #评论 soup=BeautifulSoup(html,‘html.parser‘) lis=soup.findAll(‘li‘,‘subject-item‘) for li in lis: infoDiv=li.find(‘div‘,‘info‘) commentP=infoDiv.find(‘p‘,‘comment‘) if commentP!=None: a=infoDiv.a #书名 title1=a.get(‘title‘).strip() title2Span=a.span if title2Span!=None: title2=a.span.text.strip() else: title2=‘‘ title=title1+title2 c1=title[0] c2=title[-1] #如果是中文书名,则加上书名号 if isChineseBook(title): title=u‘《‘+title+u‘》‘ else: #英文书加斜体 title=‘*‘+title+‘*‘ titleList.append(title) #出版信息 pubDiv=infoDiv.find(‘div‘,‘pub‘) pub=pubDiv.text.strip() pubList.append(pub) #标记日期 dataSpan=infoDiv.find(‘span‘,‘date‘) words=dataSpan.text.split(‘\n‘) date=words[0]+words[1] dateList.append(date) #标签 tagsSpan=infoDiv.find(‘span‘,‘tags‘) if tagsSpan!=None: tags=tagsSpan.text.strip() else: tags=‘‘ tagsList.append(tags) #评论 comment=commentP.text.strip() commentList.append(comment) return (titleList,pubList,dateList,tagsList,commentList) def getHtmlTitle(html): """ 获取网页标题 """ soup=BeautifulSoup(html,‘html.parser‘) title=soup.head.title.text return title def clearOldFile(uId): """ 清除之前已保存的文件 """ fileName=‘booksComments_‘+uId+‘.md‘ temp=open(fileName,‘w‘,encoding=‘utf-8‘) temp.close() def saveBookComment(titleList,pubList,dateList,tagsList,commentList,uId): """保存书评至文件 """ fileName=‘booksComments_‘+uId+‘.md‘ wf=open(fileName,mode=‘a‘,encoding=‘utf-8‘) size=len(titleList) for i in range(size): title=titleList[i] pub=pubList[i] date=dateList[i] tags=tagsList[i] comment=commentList[i] wf.write(‘## ‘+title+lineSep) wf.write(pub+‘ ‘+lineSep) wf.write(date+‘ ‘+lineSep) wf.write(tags+lineSep+lineSep) wf.write(comment+lineSep+lineSep) wf.close() return fileName def getPageNum(html): """解析第一页网页,返回该用户的书评页数 """ soup=BeautifulSoup(html,‘html.parser‘) paginator=soup.find(‘div‘,‘paginator‘) pas=paginator.findAll(‘a‘) num=int(pas[-2].text) return num def convertMd2Html(mdName,title): """ 将Markdown文件转换为Html格式文件 """ htmlName=mdName.replace(‘.md‘,‘.html‘) mdFile=open(mdName,‘r‘,encoding=‘utf-8‘) contents=mdFile.read() mdFile.close() md = markdown.markdown(contents) html = ‘<html><meta charset="UTF-8">‘ html+=‘<title>‘+title+‘</title>‘ html += "<body>" + md + "</body></html>" htmlFile=open(htmlName,‘w‘,encoding=‘utf-8‘) htmlFile.write(html) htmlFile.close() return htmlName #输入User-Id print(‘\nYou can find User-Id in the url.‘) print(‘E.g. Someone\‘s homepage\‘url is http://book.douban.com/people/1000001/ , the User-Id should be 1000001 .‘) uId=input(‘User-Id: ‘) while(uId==‘‘): uId=input(‘User-Id: ‘) #计数器 count=0 #读取第一页 index=0 url=url=url_1+uId+url_2+str(index)+url_3 html=getHtml(url) (titleList,pubList,dateList,tagsList,commentList)=getBookComment(html) htmlTitle=getHtmlTitle(html) clearOldFile(uId); fileName=saveBookComment(titleList,pubList,dateList,tagsList,commentList,uId) count+=len(titleList) try: pageNum=getPageNum(html) #用户读过的书的网页页数 except: pageNum=1 index+=1 #读取后续页 for i in range(index*15,15*pageNum,15): print(‘Sleep for 5 seconds.‘) time.sleep(5) print(‘%d/%d‘ %(i/15+1,pageNum)) url=url=url_1+uId+url_2+str(i)+url_3 html=getHtml(url) (titleList,pubList,dateList,tagsList,commentList)=getBookComment(html) count+=len(titleList) saveBookComment(titleList,pubList,dateList,tagsList,commentList,uId) print(‘\nMission accomplished!‘) print(‘%d comments have been saved to %s.‘ %(count,fileName)) ans=input(‘\nDo you want to convert Markdown file to html file(Y/n)?‘) ans=ans.lower() if ans!=‘n‘: htmlName=convertMd2Html(fileName,htmlTitle) print(‘Convert success: %s‘ %htmlName) os.system("pause")