landmarkLeech.py

   1 #!/usr/bin/python
   2 # vi: set softtabstop=4 shiftwidth=4 tabstop=8 expandtab:
   3
   4 """Downloads a list of landmarks from the server."""
   5
   6 import re
   7 import sys
   8 import urllib
   9
  10 import Planner
  11
  12 class LandmarkLeech(Planner.TravelPlannerClient):
  13
  14     def leech(self, fp):
  15         html = self._sendRequest("FromLandmarkSetup.oci", None)
  16         for match in _category_rx.finditer(html):
  17             self.grabCategory(fp, match.group("code"), match.group("desc"))
  18
  19     def grabCategory(self, fp, code, desc):
  20         # the description is just cosmetic - it gets displayed at the
  21         # top of the FromLandmarkType page.
  22         params = (("landmarkType", code), ("landmarkDesc", desc))
  23         html = self._sendRequest("FromLandmarkType.oci", params)
  24         for landmark in _landmark_rx.findall(html):
  25             print >> fp, "|".join((code, urllib.unquote(landmark)))
  26
  27 _category_re = '(?i)<option\s+value="(?P<code>.*?)">(?P<desc>.*?)</option>'
  28 _category_rx = re.compile(_category_re)
  29
  30 _landmark_re = ('(?i)<a href="FromLandmarkMultipleMatch.oci\?'
  31                 'landmarkAddress=(.*)">')
  32 _landmark_rx = re.compile(_landmark_re)
  33
  34 def main(argv=None):
  35     if argv is None:
  36         argv = sys.argv
  37
  38     l = LandmarkLeech()
  39     l.leech(sys.stdout)
  40
  41     return 0
  42
  43 if __name__ == '__main__':
  44     sys.exit(main())