11from urllib .request import urlopen
2+ from urllib .request import HTTPError
23from bs4 import BeautifulSoup
34import datetime
45import json
910def getLinks (articleUrl ):
1011 html = urlopen ("http://en.wikipedia.org" + articleUrl )
1112 bsObj = BeautifulSoup (html )
12- return bsObj .find ("div" , {"id" :"bodyContent" }).findAll ("a" ,
13- href = re .compile ("^(/wiki/)((?!:).)*$" ))
13+ return bsObj .find ("div" , {"id" :"bodyContent" }).findAll ("a" , href = re .compile ("^(/wiki/)((?!:).)*$" ))
1414
1515def getHistoryIPs (pageUrl ):
1616 #Format of revision history pages is:
@@ -31,8 +31,7 @@ def getHistoryIPs(pageUrl):
3131
3232def getCountry (ipAddress ):
3333 try :
34- response = urlopen ("http://freegeoip.net/json/"
35- + ipAddress ).read ().decode ('utf-8' )
34+ response = urlopen ("http://freegeoip.net/json/" + ipAddress ).read ().decode ('utf-8' )
3635 except HTTPError :
3736 return None
3837 responseJson = json .loads (response )
@@ -51,4 +50,4 @@ def getCountry(ipAddress):
5150 print (historyIP + " is from " + country )
5251
5352 newLink = links [random .randint (0 , len (links )- 1 )].attrs ["href" ]
54- links = getLinks (newLink )
53+ links = getLinks (newLink )
0 commit comments