CSV merge-subtract-doo-dah

duper.py

#!/usr/bin/python
"""
Imports two exported CSV files containing tabular data, merges them and
outputs the difference between one list and the merged copy.

"""

import csv

# Import
def loadEmails(fileName, email_index):
    emails = []
    reader = csv.reader(open(fileName, "r"), delimiter=',', quotechar='"')
    i = 0
    for row in reader:
        if i > 0:
            emails.append(row[email_index])
        i += 1
    return emails

oldList = loadEmails("old.csv", 14)
newList = loadEmails("new.csv", 4)

"""
A = old
B = new
(UNIQUE(A + B)) - B
"""

# A + B without duplicates
merged = list(oldList)
for b in newList:
    if b not in merged:
        merged.append(b)

print "Old: %d" % len(oldList)
print "New: %d" % len(newList)
print "Merged: %d" % len(merged)

# U-B
for b in newList:
    if b in merged:
        merged.remove(b)

print "Non-cont: %d" % len(merged)

for o in oldList:
    if o not in merged:
        print o

with open("output.txt", "wb") as out:
    for m in merged:
        out.write(m + "\n")