etrade-to-txf

#!/usr/bin/env python3

# https://github.com/simonmichael/hledger/blob/master/examples/TXF-spec-v042.org
# https://github.com/andreasg123/stock-gain-tax-import/blob/main/create-txf-2018.py

from collections import defaultdict
from datetime import date
from io import StringIO
from pprint import pprint
from sys import argv

symbols = defaultdict(list)
with open(argv[1]) as fh:
    # skip pass the header
    lines = fh.read().split('\n')
    for i, line in enumerate(lines):
        if line.startswith('Symbol'):
            break
    # parse the symbols
    for i in range(i+1, len(lines)):
        line = lines[i]
        if line[0] != ' ':
            # new symbol
            symbol = line.split(',', 1)[0]
            continue
        (
            _,
            quantity,
            acquired,
            cost_per_share,
            cost,
            disposed,
            price_per_share,
            proceeds,
            gain,
            deferred_loss,
            term,
            lot_selection,
            _
        ) = line.strip().split(',')
        if deferred_loss in ('--', '.00'):
            deferred_loss = None
        symbols[symbol].append({
             'quantity': quantity,
             'acquired': acquired,
             'cost_per_share': cost_per_share,
             'cost': cost,
             'disposed': disposed,
             'price_per_share': price_per_share,
             'proceeds': proceeds,
             'gain': gain,
             'deferred_loss': deferred_loss,
             'term': term,
             'lot_selection': lot_selection
        })

print('V042')
print('Aextrade-to-txf.py')
print('D', date.today().strftime('%m/%d/%y'))
print('^')


# WARNING: e*Trade doesn't provide enough information to know what was reported
# to the IRS and what wasn't in their CSVs. If you hav box B, E, or X stuff.
# You'll have to grab the pdf version and manually change the boxes for those
# :-(
codes = {
    # A
    'Short': 321,
    # D
    'Long': 323,
}
wash_codes = {
    # A
    'Short': 682,
    # D
    'Long': 323,
}

for symbol, lots in sorted(symbols.items()):
    for lot in lots:
        print('TD')
        if lot['deferred_loss']:
            code = wash_codes[lot['term']]
        else:
            code = codes[lot['term']]
        print(f'N{code}')
        print(f'P {lot["quantity"]} sh {symbol}')
        print(f'D{lot["acquired"]}')
        print(f'D{lot["disposed"]}')
        print(f'${lot["cost"]}')
        print(f'${lot["proceeds"]}')
        if lot['deferred_loss']:
            print(f'${-lot["deferred_loss"]}')
        print('^')