datajoin.py

import csv
from collections import OrderedDict

def main():

    #npa
    npa = {}
    with open('npa.csv') as f:
        for line in f:
            current_line = line.split(',')
            current_line[1] = current_line[1].strip()
            if current_line[0] == "NPA":
                print "Skipping title"
            elif current_line[1] in npa: # if key is in dictionary
                print("Two or locations mapped to a NPA")
                current_line[0] = current_line[0] + ", " + npa.get(current_line[1])
                npa.update({current_line[0] : current_line[1]}) # add new NPA with location
            else:
                npa.update({current_line[0] : current_line[1]}) # add new Location and NPA to npa dictionary

    #sample_date
    sample = {}
    with open('sample_data.csv') as f:
        for line in f:
            current_line = line.split(',')
            current_line[2] = int(current_line[2].strip())
            if current_line[0][1:4] in sample: # if key is in dictionary
                current_line[2] = current_line[2] + int(sample.get(current_line[0][1:4])) # add the duration onto the previous entry
                sample.update({current_line[0][1:4] : current_line[2]}) # add new NPA with duration
            else:
                sample.update({current_line[0][1:4] : current_line[2]}) # add new NPA and duration to dictionary

    #joined_data
    joined = {}
    sortnpa = OrderedDict(sorted(npa.items(), key=lambda t: t[0]))
    sortsample = OrderedDict(sorted(sample.items(), key=lambda t: t[0]))
    for key, value in sortnpa.items():
        if value in joined:
            if type(sortsample.get(key)) is int and type(joined.get(value)) is int:
                new = sortsample.get(key) + int(joined.get(value)) # add extra minutes if in dictionary
                joined.update({value : new}) # add new entry
        else:
            joined.update({value : sortsample.get(key)}) # add new entry

    #csv
    with open('joined_data.csv', 'wb') as csv_file:
        writer = csv.writer(csv_file)
        writer.writerow(["Location", "Minutes"])
        for key, value in joined.items():
            writer.writerow([key, value]) # write location and duration in minutes

main()