A quick hack to tally OpenStack Member affiliations

openstack_foundation_affiliations.py

from pprint import pprint
from urllib2 import urlopen

from bs4 import BeautifulSoup


soup = BeautifulSoup(urlopen("http://www.openstack.org/community/members/").read())
bullets = soup.find_all("li")
affiliations = [x.text.split("(")[1].split(")")[0].lower().strip()
                for x in bullets if "(" in x.text]
counts = {}
for aff in affiliations:
  if "rackspace" in aff or aff == "rax" or "racksapce" in aff:
    aff = u"rackspace"
  elif (aff is "none" or aff is "non" or aff is "na" or aff is "n/a"
        or aff is "" or aff is "0" or aff is "none."):
    aff = u""
  elif "canonical" in aff:
    aff = u"canonical"
  elif "alcatel" in aff and "lucent" in aff:
    aff = u"alcatel lucent"
  elif "wikimedia" in aff:
    aff = u"wikimedia"
  elif "inktank" in aff:
    aff = u"inktank"
  elif "yahoo" in aff:
    aff = u"yahoo"
  elif "suse" in aff:
    aff = u"suse"
  elif "tipit" in aff:
    aff = u"tipit"
  elif "cloudscaling" in aff:
    aff = u"cloudscaling"
  elif "comcast" in aff:
    aff = u"comcast"
  elif aff.startswith("aptira"):
    aff = u"aptira"
  elif aff.startswith("ntt"):
    aff = u"ntt"
  elif aff.startswith("target"):
    aff = u"target"
  elif aff.startswith("switfstack"):
    aff = u"switftstack"
  elif aff.startswith("redhat") or aff.startswith("red hat"):
    aff = u"red hat"
  elif aff.startswith("piston cloud") or aff.startswith("pistoncloud"):
    aff = u"piston cloud"
  elif aff.startswith("opscode"):
    aff = u"opscode"
  elif aff.startswith("citrix"):
    aff = u"citrix"
  elif aff.startswith("dey storage"):
    aff = u"dey storage"
  elif aff.startswith("dell"):
    aff = u"dell"
  elif aff.startswith("zhaw"):
    aff = u"zhaw"
  elif aff.startswith("nicira"):
    aff = u"nicira"
  elif aff.startswith("nexenta"):
    aff = u"nexenta"
  elif aff.startswith("nebula"):
    aff = u"nebula"
  elif aff.startswith("morphlabs"):
    aff = u"morphlabs"
  elif aff.startswith("maldivica"):
    aff = u"maldivica"
  elif aff.startswith("cisco"):
    aff = u"cisco"
  elif aff.startswith("emc"):
    aff = u"emc"
  elif (aff.startswith("hp")
        or " hp" in aff
        or "hewllet" in aff
        or "hewlett" in aff):
    aff = u"hp"
  ign = counts.setdefault(aff.lower(), 0)
  counts[aff.lower()] += 1


# display the count data, number of members per organization
pprint(counts)
# list all the affiliations, sorted by highest representation
pprint(sorted([(v,k) for k, v in counts.items()], reverse=True))
# give the total number of affiliations recorded
print(len(counts))
# display the top 20
pprint(sorted([(v,k) for k, v in counts.items()], reverse=True)[:20])