Find always empty columns in .csv

data.csv

gistfile1.py

#!/usr/bin/env python

# Prints any always empty columns in a CSV input file.
# usage: cat data.csv | ./mr --map | sort -k1n | ./mr --reduce

import sys

def map_main():
    import csv

    for row in csv.reader(sys.stdin):
        i = 0
        for field in row:
            if field:
                field = '.'
            print str(i) + "\t" + field
            i = i + 1

def reduce_main():
    def mr_reduce(key, values):
        empty_values = True
        for value in values:
            if value:
                empty_values = False
                break
        if empty_values:
            yield (key, None)

    prev_key = None
    values = []

    while True:
        line = sys.stdin.readline(1024)
        if not line:
            break

        line = line.rstrip('\n')

        i = 0
        key, value = line.split("\t", 1)
        if prev_key == None:
            prev_key = key
            values.append(value)
            continue

        if key == prev_key:
            values.append(value)
            continue

        for k, v in mr_reduce(prev_key, values):
            print "column " + str(k) + " is always empty"

        prev_key = key
        values = [value]

    for k, v in mr_reduce(prev_key, values):
        print "column " + str(k) + " is always empty"

assert len(sys.argv) == 2 and sys.argv[1] in ['--map', '--reduce'], "call with either --map or --reduce"

if sys.argv[1] == "--map":
    map_main()
else:
    reduce_main()