projects.mako.cc - wikipedia-api-cdsw/blob - wikipedia2-2.py

   1 import requests
   2
   3 # base url:
   4 # https://en.wikipedia.org/w/api.php?action=query&prop=revisions&titles=%s&rvlimit=100&rvprop=timestamp|user&format=json
   5
   6 pages = ["Benjamin Mako Hill", "University of Washington", "Data science"]
   7
   8 parameters = {'action' : 'query',
   9               'prop' : 'revisions',
  10               'rvlimit' : 100,
  11               'rvprop' : 'timestamp|user',
  12               'format' : 'json',
  13               'continue' : ''}
  14
  15 for page_title in pages:
  16     parameters['titles'] = page_title
  17
  18     while True:
  19         wp_call = requests.get('https://en.wikipedia.org/w/api.php', params=parameters)
  20         response = wp_call.json()
  21
  22         for page_id in response["query"]["pages"].keys():
  23             page_title = response["query"]["pages"][page_id]["title"]
  24             revisions = response["query"]["pages"][page_id]["revisions"]
  25
  26             for rev in revisions:
  27                 print(page_title + "\t" + rev["user"] + "\t" + rev["timestamp"])
  28
  29         if 'continue' in response:
  30             parameters.update(response['continue'])
  31         else:
  32             break