Parse yowsup-cli message output, see https://github.com/tgalal/yowsup/issues/450

parseYowsupCli.py

#!/usr/bin/env python3

import sys
import re
from datetime import datetime
import sqlite3

CLIDUMP = './your-cli.dump'


test_str = """[1234567890123/123456789012-1234567898@g.us(10-10-2042 11:30)]:[AC103CD12DER5]	 Your Messages goes here
Message AC103CD12DER5: Sent delivered receipt
[connected]:

[1234567890123/123456789012-1234567898@g.us(10-10-2042 11:30)]:[AC103CD12DER5]	 Another message.
Message AC103CD12DER5: Sent delivered receipt
[connected]:"""


if __name__ == "__main__":
    messageRE = re.compile('\[(?P<a>[0-9]*)\/(?P<b>[0-9]*-[0-9]*)@g.us\((?P<c>(?:[0-9]{2}\-){2}2015 [0-9]{2}:[0-9]{2})\)\]:\[(?P<d>[A-Z0-9a-z\/]{13}|.*)\]\t (?P<e>.*)\nMessage (?P=d): Sent delivered receipt\n\[connected]:')
    typeRE = re.compile('\[Media Type:image, Size:[0-9]*, URL:(?P<url>.*)\]')
    
    messages = []

    # It's your memory, deal with it ;P
    with open(CLIDUMP) as f:
        raw = f.read()
    # raw = test_str

    messagesraw = raw.split('\n\n')

    i = 0
    for m in messagesraw:
        match = re.search(messageRE, m)
        message ={'senderid': match.group('a'),
                  'groupid': match.group('b'),
                  'time': int(datetime.strptime(match.group('c'),'%d-%m-%Y %H:%M').timestamp()) + i,
                  'id': match.group('d'),
                  'body': match.group('e'),
                  'type':'text'}
        # Create unique timestamps for closely succeeding messages by adding 0-59 seconds
        i = (i + 1) % 60

        # Parse messages of type image
        typeMatch = re.search(typeRE, message['body'])
        if typeMatch:
            message.update({'body': typeMatch.group('url'),
                            'type': 'image'})

        messages.append(message)

    timestamps = list(m['time'] for m in messages)
    if len(timestamps) > len(set(timestamps)):
        sys.exit("Timestamps aren't uniue, we failed!")

    for message in messages:
        print(message)