projects.mako.cc - wikipedia-api-cdsw/blob - wikipedia-raw2.py

   1 import time
   2 import simplejson as json
   3 from urllib2 import urlopen
   4
   5 url_base = 'https://en.wikipedia.org/w/api.php?action=query&prop=revisions&titles=%s&rvlimit=100&rvprop=timestamp|user&format=json'
   6
   7 pages = ["Benjamin_Mako_Hill", "Python", "Data_science"]
   8
   9 for page_title in pages:
  10
  11     wp_call = urlopen(url_base % page_title)
  12     response = json.loads(wp_call.read())
  13
  14     for page_id in response["query"]["pages"].keys():
  15         page_title = response["query"]["pages"][page_id]["title"]
  16         revisions = response["query"]["pages"][page_id]["revisions"]
  17
  18         for rev in revisions:
  19             print page_title + "\t" + rev["user"] + "\t" + rev["timestamp"]
  20
  21
  22     time.sleep(3)