finoex/dingguo/parser/amazon_cn.py

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

import BeautifulSoup
import datetime
import dateutil.parser
import dingguo
import email.message
import ioex
import pytz
import re

def parse_order_confirmation_mail(mail):

    assert isinstance(mail, email.message.Message)

    html = mail.get_payload()[1].get_payload(decode = True).decode('utf-8')

    if not u'订单详情' in html:
        raise Exception()

    doc = BeautifulSoup.BeautifulSoup(html)

    for br in doc.findAll('br'):
        br.replaceWith('\n')

    def get_text(tag):
        return u''.join([c if type(c) is BeautifulSoup.NavigableString else get_text(c) for c in tag.contents])

    order = dingguo.Order(
        platform = u'amazon.cn',
        order_id = doc.find(text = re.compile(u'订单号')).findNext().text,
        order_date = dateutil.parser.parse(mail['date']),
        )

    article_table = doc.find(text = re.compile(u'订购日期')).findNext('table')
    for a in article_table.find('tbody').findChildren(recursive = False):
        image_col_tag, details_tag, price_tag = a.findAll('td', recursive = False)
        name_tag = details_tag.find('a')
        order.items.append(dingguo.Article(
            name = name_tag.text.strip(),
            price_brutto = dingguo.Sum.parse_text(price_tag.text),
            quantity = 1,
            reseller = name_tag.findNext(text = re.compile(u'由')).findNext().text,
            ))

    with ioex.setlocale('zh_CN.utf-8'):
        estimated_arrival_date_start = pytz.timezone('Asia/Shanghai').localize(
            datetime.datetime.strptime(
                doc.find(text = re.compile(u'预计送达日期:')).findNext('b').text.encode('utf-8'),
                '%A, %m/%d',
                ).replace(year = order.order_date.year)
            )
        assert estimated_arrival_date_start >= order.order_date
    order.items.append(dingguo.Shipping(
        name = doc.find(text = re.compile(u'送货方式:')).findNext('b').text,
        price_brutto = dingguo.Sum.parse_text(
            doc.find(text = re.compile(u'配送费:')).findNext().text
            ),
        estimated_arrival_time = ioex.datetimeex.Period(
            start = estimated_arrival_date_start,
            end = (estimated_arrival_date_start + datetime.timedelta(days = 1)),
            ),
        destination_point = '\n'.join([l.strip() for l in
            get_text(
                doc.find(text = re.compile(u'您的订单发送至:')).findNext('b')
                ).split('\n')
            ]),
        ))

    return [order]