quick and dirty skim over the list of CA breach notices

ca_breaches.py

#!/usr/bin/env python

from collections import Counter 
import itertools
import urllib

from BeautifulSoup import BeautifulSoup
import timestring

URL="http://oag.ca.gov/ecrime/databreach/list"

html = urllib.urlopen(URL).read()

soup = BeautifulSoup(html)
trs = soup.findAll('tr')
trs = filter(lambda x: x.attrs != [], trs)
# -1 is date reported
# -2 is date breached, if known (or n/a)
dates = [ x.findAll('td')[-2].text.split(',') for x in trs ]
dates = filter(lambda x: x != 'n/a', list(itertools.chain.from_iterable(dates)))

# counts by year
print Counter([ x.year for x in map(timestring.Date, dates) ])

# most common reporting orgs
print Counter([ x.findAll('td')[0].text for x in trs ]).most_common()


"""
breaches by year: {2014: 180, 2013: 155, 2012: 146, 2015: 32, 2011: 26, 2010: 3, 2007: 1, 2009: 1}

most common orgs breached:
(u'American Express Travel Related Services Company, Inc and /or its Affiliates (\u201cAXP\u201d)',
  61),
 (u'Discover Financial Services', 12),
 (u'Massachusetts Mutual Life Insurance Company', 5),
 (u'Yolo Federal Credit Union', 3),
 (u'California Department of Public Health', 3),
 (u'East West Bank', 3),
 (u'California Department of Corrections and Rehabilitation', 3)
"""