Script to parse Clipper activity logs and calculate sums

parse_clipper_statement.py

#!/usr/bin/env python3

"""
@author Demitri Morgan <demitri.morgan@gmail.com>

Takes the ugly, poorly-formatted text output from Clipper's web portal...
https://www.clippercard.com/ClipperCard/dashboard.jsf

...and prints it to stdout in three columns: time, credit and debit.

Totals printed to stderr (in case you want to pipe to awk and do other things).

The way it's expected to work: 

1. Copy and paste the table of Clipper Card activity into a text file
2. Pass the file as the sole argument to this script.

I wrote this as a stopgap for record keeping and metrics until Clipper improves
their website with some kind of CSV export, or if they're really nice, an API.
"""


import argparse
import re
import sys

datestamp = re.compile(
    r'^(\d{2}\/\d{2}\/\d{4})',
    flags=re.MULTILINE
)
currency_amount = re.compile(
    r'^\d+\.\d{2}$'
)

def main():
    ap = argparse.ArgumentParser("Reads screwy input copied from Clipper's tra"\
        "sactions dashboard and translates it to cleanly-formatted data.")
    ap.add_argument("in_file", type=argparse.FileType('r'))
    args = ap.parse_args()
    fh = args.in_file

    file_contents = fh.read()
    fh.close()

    credit = 0.
    debit  = 0.

    with_linebreaks = datestamp.sub(r'\n\1', file_contents)
    for l in with_linebreaks.split('\n\n'):
        entries = l.strip().split('\n')
        datetime = entries[0]
        if len(entries) < 5:
            continue # Malformed entry
        balance = entries.pop()
        if currency_amount.match(entries[-1]) is None:
            continue # A no-fare entrance (charged fare upon exit)
        if 're-load' in entries[1]:
            transaction = (datetime, entries[-1], '0.00')
        else:
            transaction = (datetime, '0.00', entries[-1])

        credit += float(transaction[1])
        debit  += float(transaction[2])
        print('{0}\t{1}\t{2}'.format(*transaction))

    print("      Total fare: %.2f"%debit, file=sys.stderr) 
    print("  Total deposits: %.2f"%credit, file=sys.stderr) 
        

if __name__=='__main__':
    main()