finoex/dingguo/parser/amazon.py

# -*- coding: utf-8 -*-

import datetime
import dateutil
import dateutil.tz
import dingguo
import email
import ioex
import re

def parse_order_confirmation_mail(mail):

    assert isinstance(mail, email.message.Message)

    msg_text = mail.get_payload()[0].get_payload(decode = True).decode('utf-8')

    if not u'Amazon.de Bestellbestätigung' in msg_text:
        raise Exception('no amazon order confirmation')

    orders = []

    for order_text in re.split(ur'={32,}', msg_text)[1:-1]:

        order_id = re.search(r'Bestellnummer #(.+)', order_text).group(1)

        order_date_formatted = re.search(ur'Aufgegeben am (.+)', order_text, re.UNICODE).group(1)
        mail_date = dateutil.parser.parse(mail['date'])
        with ioex.setlocale('de_DE.UTF-8'):
            order_date = datetime.datetime.strptime(
                    order_date_formatted.encode('utf-8'),
                    '%d. %B %Y',
                    )
            assert order_date.date() == mail_date.date()

        order = dingguo.Order(
            u'amazon.de',
            order_id,
            order_date.date(),
            )

        for articles_text in re.findall(
                    ur'Bestellte\(r\) Artikel:\s+'
                        + ur'([\W\w]+?)\s+'
                        + ur'(Lieferung \d|_{10,})',
                    order_text,
                    re.UNICODE,
                    ):
            for article_text in re.split(ur'\n\t*\n', articles_text[0]):
                article_match = re.match(
                    ur' *((?P<quantity>\d+) x )?(?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
                    )
                assert article_match is not None, repr(article_text)
                article = article_match.groupdict()
                order.items.append(dingguo.Article(
                    name = article['name'],
                    price_brutto = dingguo.Sum(
                        float(article['price_brutto'].replace(',', '.')),
                        article['price_brutto_currency']
                        ),
                    quantity = int(article['quantity']) if article['quantity'] else 1,
                    authors = article['authors'].split(',') if article['authors'] else None,
                    state = article['state'],
                    reseller = article['reseller'],
                    shipper = article['shipper'],
                    ))

        assert re.search(
                ur'Verpackung und Versand: \w+ (\d+,\d+)',
                order_text,
                ).group(1) == '0,00'
        shipping_match = re.finditer(
                ur'Zustellung:\s+(?P<t>[\W\w]+?)\s+'
                    + ur'Versandart',
                order_text,
                re.UNICODE,
                )
        for shipping_attr in [m.groupdict() for m in shipping_match]:
            with ioex.setlocale('de_AT.UTF-8'):
                s = datetime.datetime.strptime(
                        shipping_attr['t'].encode('utf8'),
                        '%A, %d. %B',
                        ).replace(
                            year = order.order_date.year,
                            tzinfo = dateutil.tz.gettz('Europe/Berlin'),
                            )
                assert s.date() > order.order_date
            e = s + datetime.timedelta(days = 1)
            order.items.append(dingguo.Shipping(
                price_brutto = dingguo.Sum(0.0, u'EUR'),
                estimated_arrival_time = ioex.datetimeex.Period(start = s, end = e),
                ))

        orders.append(order)

    return orders