finoex/amazon-order-confirmation-mail-parser

#!/usr/bin/env python
# PYTHON_ARGCOMPLETE_OK

import re
import os
import sys
import yaml
import email
import pprint
import random
import locale
import argparse
import datetime
import argcomplete

# strptime
locale.setlocale(locale.LC_ALL, 'de_DE.UTF-8')

def parse(msg):

    order = {
        'platform': 'amazon.de',
        }

    msg_text = msg.get_payload(decode = True)

    order['order_id'] = re.search(r'Bestellnummer #(.+)', msg_text).group(1)

    order_date = datetime.datetime.strptime(
        re.search(r'Aufgegeben am (.+)', msg_text).group(1),
        '%d. %B %Y'
        )
    order['order_date'] = order_date.strftime('%Y-%m-%d')

    order['articles'] = []
    articles_text = msg_text.split('Bestellte(r) Artikel:')[1].split('_' * 10)[0].strip()
    for article_text in articles_text.split('\n\n'):
        article_match = re.match(
            ur' *(?P<name>.*)\s'
                + ur' *(?P<price_brutto_currency>[A-Z]+) (?P<price_brutto>\d+,\d+)\s'
                + ur'( *Zustand: (?P<state>.*)\s)?'
                + ur' *Verkauft von: (?P<reseller>.*)',
            article_text,
            re.MULTILINE | re.UNICODE
            )
        if article_match is None:
            sys.stderr.write(repr(article_text))
            raise Exception('could not match article')
        article = article_match.groupdict()
        article['price_brutto'] = float(article['price_brutto'].replace(',', '.'))
        order['articles'].append(article)

    return order

def compute():

    msg = email.message_from_string(sys.stdin.read())

    orders = []

    if msg.is_multipart():
        for part in msg.get_payload():
            orders.append(parse(part))
    else:
        orders.append(parse(msg))

    print(yaml.safe_dump(orders, default_flow_style = False))

def _init_argparser():

    argparser = argparse.ArgumentParser(description = None)
    # argparser.add_argument('a', nargs = '*')
    # argparser.add_argument('--b')
    # argparser.add_argument('--flag', action='store_true')
    # argparser.add_argument('file', type = argparse.FileType('r'))
    # exclusive_group = argparser.add_mutually_exclusive_group(required = False)
    # exclusive_group.add_argument('--exclusive-1', action='store_true')
    # exclusive_group.add_argument('--exclusive-2', action='store_true')
    # subparsers = argparser.add_subparsers(help = None, dest = 'command')
    return argparser

def main(argv):

    argparser = _init_argparser()
    argcomplete.autocomplete(argparser)
    args = argparser.parse_args(argv)

    compute(**vars(args))

    return 0

if __name__ == "__main__":
    sys.exit(main(sys.argv[1:]))