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])

    orders = []

    order_id_label_tags = doc.findAll(text = re.compile(u'订单号'))
    assert len(order_id_label_tags) % 2 == 0
    order_id_label_tags = order_id_label_tags[len(order_id_label_tags)/2:]
    for order_id_label_tag in order_id_label_tags:

        order = dingguo.Order(
            platform = u'amazon.cn',
            order_id = order_id_label_tag.findNext().text,
            order_date = dateutil.parser.parse(mail['date']),
            )

        order_detail_tag = (order_id_label_tag.findNext().findNext('table')
            if len(order_id_label_tags) > 1
            else order_id_label_tag.findParent('table').findPrevious('table'))
        article_table = order_detail_tag.findNext('table')
        if len(order_id_label_tags) == 1:
            article_table = article_table.findNext('table')
        for article_tag in article_table.find('tbody').findChildren(recursive = False):
            image_col_tag, details_tag, price_tag = article_tag.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,
                ))

        del_date_str = order_detail_tag.find(text = re.compile(u'预计送达日期:')).findNext().text
        def parse_date(s):
            with ioex.setlocale('zh_CN.utf-8'):
                return pytz.timezone('Asia/Shanghai').localize(
                    datetime.datetime.strptime(
                        s.encode('utf-8'),
                        '%A, %m/%d',
                        ).replace(year = order.order_date.year)
                    )
        del_date_range = [parse_date(s.strip()) for s in del_date_str.split('-')]
        if len(del_date_range) == 1:
            del_date_range.append(del_date_range[0])
        order.items.append(dingguo.Shipping(
           name = order_detail_tag.find(text = re.compile(u'送货方式:')).findNext('b').text,
           price_brutto = dingguo.Sum.parse_text(
               order_detail_tag.findNext(text = re.compile(u'配送费:')).findNext().text
               ),
            estimated_arrival_time = ioex.datetimeex.Period(
               start = del_date_range[0],
               end = (del_date_range[1] + datetime.timedelta(days = 1)),
               ),
           destination_point = '\n'.join([l.strip() for l in
               get_text(
                   order_detail_tag.find(text = re.compile(u'您的订单发送至:')).findNext('b')
                   ).split('\n')
               ]),
            ))

        orders.append(order)

    return orders