libpurple-to-markdown/libpurple_to_markdown/synctech_sms.py

"""Backend for SyncTech Backup & Restore.

[SyncTech Backup & Restore](https://www.synctech.com.au/sms-backup-restore/)
for Android is a free app for backing up your SMS and MMS messages. It uses an
XML format as backup format, which this backend reads and converts to the
standardized Message format.
"""

import datetime
import logging
from collections.abc import Iterator
from pathlib import Path

import bs4

from .data import MYSELF, Message

logger = logging.getLogger(__name__)

def normalize_phone_number(num: str) -> str:
    num = num.replace(' ', '')
    if num.startswith('00'):
        num = '+' + num.removeprefix('00')
    if len(num) == 8:
        num = '+45' + num
    elif len(num) >= 10 and num[0] != '+':
        num = '+' + num
    return num


def sms_soup_to_message(soup: bs4.BeautifulSoup) -> Message:
    # TODO: Require myself
    sent_at = datetime.datetime.fromtimestamp(int(soup['date']) / 1000)

    phone_num = normalize_phone_number(soup['address'])
    contact_name = soup.get('contact_name') or phone_num
    if soup['type'] == '2':
        sender = MYSELF
    else:
        sender = contact_name

    text = soup['body']
    chat_id = f'SMS {contact_name} {phone_num}'
    return Message(sent_at, sender, text, chat_id=chat_id)


def parse_messages_in_backup_xml_file(path: Path) -> Iterator[Message]:
    logger.info('Parsing %s', path)

    with open(path) as f:
        soup = bs4.BeautifulSoup(f, 'lxml-xml')

    for sms in soup.find_all('sms'):
        yield sms_soup_to_message(sms)
        del sms