extract_plist_xml_data.py

#!/usr/bin/env python3

# extract base64 <data> in XML plist
# use `plistutil` or `plist -convert xml1` to convert from binary

import xml.sax
from sys import argv
from base64 import b64decode
import re


class MyHandler(xml.sax.ContentHandler):
    def startDocument(self):
        self.in_data = None
        self.in_key = None
        self.last_key = "unknown"
        self.count = 0

    def startElement(self, name, attrs):
        if name == "data":
            self.in_data = ""
        if name == "key":
            self.in_key = ""

    def endElement(self, name):
        if self.in_data is not None:
            self.count += 1
            date_name = f"data{self.count}_{self.last_key}.bin"
            with open(date_name, "wb") as f:
                data = re.sub(r"\s", "", self.in_data, re.DOTALL)
                f.write(b64decode(data))
                print(f"write  {date_name}, {len(data)} bytes")
            self.in_data = None
        if self.in_key:
            self.last_key = self.in_key.strip()
            self.in_key = None

    def characters(self, content):
        if self.in_data is not None:
            self.in_data += content
        elif self.in_key is not None:
            self.in_key += content


parser = xml.sax.make_parser()
parser.setFeature(xml.sax.handler.feature_namespaces, 0)
Handler = MyHandler()
parser.setContentHandler(Handler)
parser.parse(argv[1])