finoex/order-confirmation-mail-parser

#!/usr/bin/env python
# -*- coding: utf-8 -*-
# PYTHON_ARGCOMPLETE_OK

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

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

def parse_amazon(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>.*)\n'
                + ur'( *von (?P<authors>.*)\n)?'
                + ur' *(?P<price_brutto_currency>[A-Z]+) (?P<price_brutto>\d+,\d+)\n'
                + ur'( *Zustand: (?P<state>.*)\n)?'
                + ur' *Verkauft von: (?P<reseller>.*)'
                + ur'(\n *Versand durch (?P<shipper>.*))?',
            article_text,
            re.MULTILINE | re.UNICODE
            )
        if article_match is None:
            sys.stderr.write(repr(article_text) + '\n')
            raise Exception('could not match article')
        article = article_match.groupdict()
        if article['authors']:
            article['authors'] = article['authors'].split(',')
        else:
            del article['authors']
        article['price_brutto'] = float(article['price_brutto'].replace(',', '.'))
        order['articles'].append(article)

    return order

def parse_oebb(msg):

    msg_text = re.sub(
        r'<[^>]+>',
        '',
        HTMLParser.HTMLParser().unescape(msg.get_payload(decode = True).decode('utf8'))
        )

    order_match = re.search(
        ur'Booking code:\s+(?P<order_id>[\d ]+)\s+'
            + ur'Customer number:\s+(?P<customer_id>PV\d+)\s+'
            + ur'Booking date:\s+(?P<order_date>.* \d{4})\s',
        msg_text,
        re.MULTILINE | re.UNICODE
        )
    order = order_match.groupdict()
    order['platform'] = 'oebb.at'
    locale.setlocale(locale.LC_ALL, 'en_US.UTF-8')
    order['order_date'] = datetime.datetime.strptime(
        order['order_date'],
        '%b %d, %Y'
        ).strftime('%Y-%m-%d')

    article_match = re.search(
        ur'(?P<price_brutto_currency>.)(?P<price_brutto>\d+\.\d+)'
            + ur'[\W\w]+'
            + ur'Your Booking\s+'
            + ur'(?P<departure_point>.*)\s+>\s+(?P<destination_point>.*)',
        msg_text,
        re.MULTILINE | re.UNICODE
        )
    article = article_match.groupdict()
    article['name'] = 'Train Ticket'
    article['price_brutto'] = float(article['price_brutto'])
    if article['price_brutto_currency'] == u'€':
        article['price_brutto_currency'] = 'EUR'
    else:
        raise Exception('currency %s is not supported' % article['price_brutto_currency'])
    order['articles'] = [article]

    return order

def parse(msg):

    tracebacks = {}

    try:
        return parse_amazon(msg)
    except:
        tracebacks['amazon'] = traceback.format_exc()

    try:
        return parse_oebb(msg)
    except:
        tracebacks['oebb'] = traceback.format_exc()

    for parser_name in tracebacks:
        print('%s parser: \n%s' % (parser_name, tracebacks[parser_name]))

    print('failed')
    # raise Exception('failed to parse')

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)
    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:]))