csv_align.py

#!/usr/bin/env python
""" Read a tab-separated CSV file and print its contents, padding values so
they align into pretty columns. """

import csv
from collections import defaultdict
from optparse import OptionParser

option_parser = OptionParser()
option_parser.add_option("-e", default="utf-8", dest="encoding")
option_parser.add_option("-s", default="\t", dest="separator")

def pretty_print(datafile, encoding, separator):
    normalize = lambda line: unicode(line, encoding).replace('\t', ' '*4)

    lengths = defaultdict(int)
    for line in csv.reader(datafile, delimiter=separator):
        # first pass: decide on column widths
        for n, value in enumerate(line):
            value = normalize(value)
            lengths[n] = max(lengths[n], len(value))
        if n > 10000: # if the file is too big, bail out
            break

    datafile.seek(0)
    for line in csv.reader(datafile, delimiter=separator):
        # second pass: print the actual data
        for n, value in enumerate(line):
            value = normalize(value)
            print ((u'%%-%ds ' % lengths[n]) % value).encode(encoding),
        print

def main():
    options, args = option_parser.parse_args()
    if len(args) != 1:
        option_parser.error("Please specify one file name")

    with open(args[0], 'rt') as datafile:
        pretty_print(datafile, options.encoding, options.separator)

if __name__ == '__main__':
    main()