#!/usr/bin/python

#split vcf files

import re

working_dir = '/home/umair/Documents/Contacts/'
input_file = 'contacts starred 26-06-2014.vcf'
output_seed = 'contacts-part-'
vcards_per_file = 1

with open(working_dir + input_file,'r') as f:
    count = 0
    output_count = 1
    results = []
    for line in f:
        if ("BEGIN:VCARD" in line):
            count += 1
        if (count <= vcards_per_file):
            if line.startswith('N:'):
                # LAST;FIRST -> FIRST;LAST Names
                line = re.sub(r'N:([^;]*);([^;]+)(;+)', r'N:\2;\1\3', line)
            results.append(line)
        else:
            #output file with stored values
            with open(working_dir + output_seed + str(output_count) + '.vcf','w') as oFile:
                for item in results:
                    oFile.write(item)

            #increment outputfile count
            output_count += 1

            #clear results list and append last read line
            del results[:]
            results.append(line)

            #set counter back to 1
            count = 1


    #write the last set of results to a file
    with open(working_dir + output_seed + str(output_count) + '.vcf','w') as oFile:
        for item in results:
            oFile.write(item)