cluster-fu-csv-generator.py

import csv
import random

def randTime(size=500):
    timings = [700, 730, 800, 830, 900]
    return (random.choices(timings, k=size))

def generate(contents, size=500):
    r = [contents[(random.randint(0, len(contents)))]
            for i in range(size * 2)]
    origins = random.choices()
    r = list(map(lambda x: x[0], r))  # postal code is in first column of csv dump
    randomTimes = randTime(size)
    rows = []
    rows.append("Index,Origin,Destination,Time")
    for i in range(size):
        rows.append("{},{},{},{}".format(i, r[i], r[size+i], randomTimes[i]))
    return rows
    
if __name__ == "__main__":
    infile = "./postal_code_to_address.csv"
    with open(infile) as csvfile:
        csvreader = csv.reader(csvfile, delimiter=',', quotechar='"')
        next(csvreader)
        contents = list(csvreader)
        csvfile.close()

    with open("cf-file.csv", 'w') as outfile:
        outfile.write("\n".join(generate(contents, 500)))
    outfile.close()