The outgoing calls information for the "Megafon" mobile operator

megafon_calls_info.py

# coding=utf-8

"""
The outgoing calls information for the "Megafon" mobile operator

"""

import xml.etree.ElementTree as ElementTree
import re

import numpy as np


ALL_CALLS = re.compile(r'Исходящие (местные|междугородные|международные) вызовы')
LOCAL_CALLS = re.compile(r'Исходящие местные вызовы')
LONG_DIST_CALLS = re.compile(r'Исходящие междугородные вызовы')
INTERNATIONAL_CALLS = re.compile(r'Исходящие международные вызовы')


def get_calls_info(file_name, call_type=ALL_CALLS):
    tree = ElementTree.parse(file_name)
    root = tree.getroot()

    prefix = '{urn:schemas-microsoft-com:office:spreadsheet}'

    worksheet = root.find(prefix + 'Worksheet')
    table = worksheet.find(prefix + 'Table')

    calls_info = {
        'date': [],
        'time': [],
        'number': [],
        'type': [],
        'minutes': [],
        'cost': [],
    }

    for row in table.findall(prefix + 'Row'):
        cells = list(row.findall(prefix + 'Cell'))

        for cell in cells:
            if cell.attrib[prefix + 'Index'] == '11':
                data = cell.find(prefix + 'Data')
                if data is None:
                    continue

                if call_type.search(data.text):
                    calls_info['type'].append(data.text)

                    for cell_ in cells:
                        if cell_.attrib[prefix + 'Index'] == '1':
                            date = cell_.find(prefix + 'Data').text
                            calls_info['date'].append(date)

                        if cell_.attrib[prefix + 'Index'] == '5':
                            time = cell_.find(prefix + 'Data').text
                            calls_info['time'].append(time)

                        if cell_.attrib[prefix + 'Index'] == '7':
                            number = cell_.find(prefix + 'Data').text
                            calls_info['number'].append(number)

                        if cell_.attrib[prefix + 'Index'] == '8':
                            minutes = cell_.find(prefix + 'Data').text
                            calls_info['minutes'].append(int(minutes))

                        if cell_.attrib[prefix + 'Index'] == '14':
                            cost = cell_.find(prefix + 'Data').text
                            calls_info['cost'].append(float(cost))

    return calls_info


def filter_by(calls_info, key='number', value=None):
    filtered_calls_info = {}

    def _filter(info):
        if not key or not value:
            return info

        filtered = list(filter(lambda x: calls_info[key][x[0]] == value,
                               enumerate(info)))
        if not filtered:
            return []

        return list(zip(*filtered))[1]

    for k, v in calls_info.items():
        filtered_calls_info[k] = _filter(v)

    return filtered_calls_info


def print_report(calls_info):
    cumsum_minutes = np.cumsum(calls_info['minutes']).tolist()
    cumsum_cost = np.cumsum(calls_info['cost']).tolist()

    for date, time, call_type, number, minutes, cost, cum_minutes, cum_cost in zip(
        calls_info['date'], calls_info['time'], calls_info['type'],
            calls_info['number'], calls_info['minutes'], calls_info['cost'],
            cumsum_minutes, cumsum_cost):

        print('{}, {}, {}, {}, {} ({} минут, {} рублей)'.format(
            date, time, number, minutes, cost, cum_minutes, round(cum_cost, 2)))


if __name__ == '__main__':
    calls_info = get_calls_info('megafon_report.xls', ALL_CALLS)
    print_report(filter_by(calls_info, 'number', '79212223344'))