""" LDIF parsing module """

import base64

def parseLdif(input, field_separator=':', dn_name='dn', key_process=lambda s: s, value_process=str.lstrip, logger=lambda bad_line_number, bad_data: None):
    """ accepts iterable input with lines printed by ldapsearch
    returns dict of dicts with parsed data """
    temp_dict = {}
    line_count = 0
    line, key, value = '', '', ''
    comment, b64 = False, False
    
    do_work = True
    state = 0
    lines = iter(input)
    while do_work:
        if state == 0: # read the line
            try:
                line_ = lines.next()
                line_count += 1
                line = line_.rstrip()
                if line:
                    state = 1
                else:
                    state = 5
            except StopIteration:
                line = ''
                state = 6
        elif state == 1: # evaluate the line
            char = line[0]
            if char == '#':
                state = 2
            elif char == ' ':
                state = 3
            elif key and value:
                state = 5
            else:
                state = 4
        elif state == 2: # it was the comment
            comment = True
            state = 0
        elif state == 3: # line continuation
            if (not comment) and key:
                value += line[1:]
            state = 0
        elif state == 4: # line seems to be legit - parse it
            if line:
                (key_, sep_, value_) = line.partition(field_separator)
                if sep_: # the line had something good
                    b64 = len(value_) > 2 and value_[0] == ':'
                    key = key_
                    value = value_[1:].strip() if b64 else value_
                elif b64 and key:
                    value += line
                else:
                    b64 = False
                    logger(line_count, line_)
            comment = False
            line = ''
            state = 0
        elif state == 5 or state == 6: # add to temp_dict
            if key and value:
                if b64:
                    value = base64.b64decode(value)
                key = key_process(key)
                value = value_process(value)
                if key not in temp_dict:
                    temp_dict[key] = []
                temp_dict[key].append(value)
            key, value = '', ''
            b64 = False
            if line: # return to line parsing
                state = 4
            else:
                if state == 6: # caught exception - no more lines
                    state = 8
                else:
                    state = 7
        elif state == 7 or state == 8: # yield results
            if dn_name in temp_dict:
                yield temp_dict
            else: # no dn in temp_dict
                logger(line_count, temp_dict)
            temp_dict = {}
            do_work = state != 8
            if do_work:
                state = 0