EECS293 CSV translation project

modernizer.py

#!/usr/bin/env python
import fileinput
import re
import sys


def _prepare_line_list(line):
    line_list = re.split("\s+", line)
    # Use of re.split adds an empty element to the list
    del line_list[len(line_list) - 1]
    _check_field_quantity(line_list)
    _check_field_values(line_list)
    del line_list[5]
    return line_list


def _check_field_quantity(line_list):
    # Check that we have the proper number of fields
    if len(line_list) not in [6, 7]:
        raise ValueError("Incorrect number of fields on line %d. Bad file?" %
                         fileinput.filelineno())
    # Add empty field if we don't have a middle name
    if len(line_list) is 6:
        line_list.insert(3, "")


def _check_field_values(line_list):

    if not line_list[0].isdigit():
        raise ValueError(
            "First field on line %d is not all digits.  Bad file?" %
            fileinput.filelineno())

    if not line_list[5] == "F13":
        raise ValueError('Fifth Field on line %d is not "F13".  Bad file?' %
                         fileinput.filelineno())

    if not line_list[6].startswith("S"):
        raise ValueError(
            'Sixth field on line %d does not start with "S". Bad file?' %
            fileinput.filelineno())


def _capitalize_at_index(string, index):
    if index < len(string):
        return string[:index] + string[index].upper() + string[index + 1:]
    else:
        return string


def _capitalize(line_list):
    # This is an indexed loop instead of list iteration because we only
    # want to operate on indices [1,4], but slicing creates a new list.
    for i in xrange(1, 4):
        line_list[i] = line_list[i].title()
∏ if (line_list[i].startswith("Mc", 0, 2) or
            line_list[i].startswith("O'", 0, 2)):
            line_list[i] = _capitalize_at_index(line_list[i], 2)
        elif line_list[i].startswith("Mac", 0, 2):
            line_list[i] = _capitalize_at_index(line_list[i], 4)

    return line_list


def main():
    try:
        for line in fileinput.input():
            line_list = _prepare_line_list(line)
            _capitalize(line_list)
            print ",".join(line_list)
    finally:
        fileinput.close()

if __name__ == '__main__':
    main()