biglumber-sort - sort keys from a biglumber page by mean shortest distance

biglumber-sort.py

#!/usr/bin/python -tt
# coding: utf-8

# Manuel Pégourié-Gonnard, 2012. WTFPL v2.

"""
biglumber-sort - sort keys from a biglumber page by mean shortest distance

Example:
  biglumber-sort 'http://biglumber.com/x/web?sc=Paris'

Background:
  - biglumber.com is a site that lists people willing to sign GPG keys
  - the mean shortest distance of a key is defined here:
  http://pgp.cs.uu.nl/plot/
"""

import re
import sys
import urllib

class Key(object):
  """A GPG key"""
  def __init__(self, fingerprint):
    self._fingerprint = fingerprint
    self._msd = None

  def fingerprint(self):
    return self._fingerprint

  def short_id(self):
    return self._fingerprint[-8:]

  def __str__(self):
    return self.short_id()

  def biglumber_url(self):
    """Biglumber key page with owner details"""
    return 'http://biglumber.com/x/web?sf=' + self._fingerprint

  def wotsap_url(self):
    """Plain-text statistics"""
    return 'http://webware.lysator.liu.se/jc/wotsap/wots/latest/keystatistics/0x' + self.short_id() + '.txt'

  def kpas_url(self):
    """More human-friendly statistics"""
    return 'http://pgp.cs.uu.nl/stats/' + self.short_id() + '.html'

  def msd(self):
    """Mean shortest distance as given by wotsap statistics.
    +infinity if not listed by wotsap.
    Cached to avoid repeated network access"""
    if not self._msd:
      handle = urllib.urlopen(self.wotsap_url())
      data = handle.read()
      handle.close()
      m = re.search(r'Mean shortest distance:\s+([0-9.]+)', data)
      self._msd = float(m.group(1) if m else 'inf')
    return self._msd

  def __cmp__(self, other):
    return cmp(self.msd(), other.msd())


def get_keys(biglumber_url):
  """Get a list of keys from biglumber city/country url"""
  fh = urllib.urlopen(biglumber_url)
  text = fh.read()
  fh.close()

  fingerprints = re.findall(
      r'>Fingerprint: <a href=".*?=(.*?)">', text)
  return [ Key(fp) for fp in fingerprints ]


def main():
  werr = sys.stderr.write
  if len(sys.argv) != 2:
    werr('Error: invalid number of arguments.\n' + __doc__)
    sys.exit(1)
  source_url = sys.argv[1]

  werr("Acquiring key list... ")
  keys = get_keys(source_url)
  werr(str(len(keys)) + " keys found.\n")

  werr("Acquiring msd data... ")
  for k in keys:
    k.msd()
    werr('.')
  werr(' done.\n')

  for key in sorted(keys):
    print str(key.msd()).ljust(8) + key.biglumber_url()
  return


# Standard boilerplate to call the main() function.
if __name__ == '__main__':
  main()