process.py

#!/usr/bin/env python
#
# Converts standard LDIF to a list of dicts
#

import sys
from base64 import b64decode
from pprint import pprint

SOURCE = "source.ldf"

def main():
    objs = ldf_to_dict(file(SOURCE))
    pprint(objs)
    return 0

def ldf_to_dict(fh):
    f = fh.read()
    fh.close()

    # sanitize file; join multiple lines and strip \r
    f = f.replace("\r", "").replace("\n ", "")

    item_objs = []
    items = f.split("\n\n")
    for item in items:
        item_obj = {}
        for line in item.split("\n"):
            if not line:
                continue

            key, value = line.split(": ", 1)
            if key[-1] == ":":
                # double colon indicates base64
                value = b64decode(value)
                key = key[:-1]

            if key not in item_obj:
                item_obj[key] = value
            elif type(item_obj[key]) == type([]):
                item_obj[key].append(value)
            else:
                item_obj[key] = [item_obj[key], value]

        item_objs.append(item_obj)

    return item_objs

if __name__ == "__main__":
    sys.exit(main())