Counts various statistics of claims, such as total claims, number of independent claims, dependent claims

ClaimCounts.py

# ClaimCounts()
# Counts various statistics of claims, such as total claims,
# number of independent claims, dependent claims
# files to create database can be found at: 
# https://github.com/aniemerg/Patent-Tools
# written on May 30, 2012 for a blog post

import MySQLdb as mdb
import sys
import datetime
import re
from pprint import pprint 
import math

def ClaimCounts():
  print "Started Counting Claims at Time:"
  print str(datetime.datetime.now())
  #mdb.connect('localhost', 'username', 'password', 'database');
   
  con = mdb.connect('localhost', 'root', 'password', 'database');
    
  with con:
    cur = con.cursor()
    cur.execute("SELECT USPatents.claims from USPatents LIMIT 10000")
    rows = cur.fetchall()
    #dictionary for the number of claims 
    totalclaims = {}
    #dictionary for the number of independent claims
    indclaims = {}
    #dictionary for the number of dependent claims 
    depclaims = {}
    #Process the claims
    for x in rows:
        claims = []
        claims = re.split("\d\.\ ", x[0])
        claims = claims[1:]
        totalclaims[len(claims)] = totalclaims.get(len(claims),0) + 1
        indclaimsc = 0
        for claim in claims:
          if re.findall('claim', claim) == []:
            indclaimsc += 1
        indclaims[indclaimsc] = indclaims.get(indclaimsc, 0) +1
        if len(claims) < indclaimsc:
           print "There is a serious error: More claims than independent claims were found!"
        depclaims[len(claims) - indclaimsc] = depclaims.get(len(claims) - indclaimsc, 0) + 1

          #Now write this stuff out to a file
  
          #Save Total Claims to File
    outfile = open('Total_Claim_Counts.csv', 'w') 
    results = totalclaims.items()
    results.sort()
    for result in results:
      line = "[\'%s\', %s],\n" % (result[0], result[1]) 
      outfile.write(line)      
    outfile.close()

    #Save Independent Claims to File
    outfile = open('Independent_Claim_Counts.csv', 'w') 
    results = indclaims.items()
    results.sort()
    for result in results:
      line = "[\'%s\', %s],\n" % (result[0], result[1]) 
      outfile.write(line)      
    outfile.close()

    #Save Independent Claims to File
    outfile = open('Dependent_Claim_Counts.csv', 'w') 
    results = depclaims.items()
    results.sort()
    for result in results:
      line = "[\'%s\', %s],\n" % (result[0], result[1]) 
      outfile.write(line)      
    outfile.close()
        
        
if __name__ == '__main__':  
    ClaimCounts()