Fragmenter 1

class Signature_Element(object):
    """docstring for sub_seq_factory"""
    def __init__(self):
        self.byte_seq_reference = None
        self.subseq_min_frag_len = None
        self.subseq_position = None
        self.subseq_max_offset= None
        self.subseq_min_offset = None
        self.sequence = None
        self.default_shift = None


    def do_grep_grammer_convert(self,sequence):
        """converts the sigfile notation for masks e.g. ... to DROID e.g {3}"""
        grep_mask_length = sequence.count(".")
        sequence = re.sub ("\.+"," {} ",sequence)
        grep_insert = "{%s}" % (str(grep_mask_length))
        sequence = sequence.replace("{}", grep_insert)
        return sequence

    def do_fragmenter(self,sequence):
        string = 1
        token = 0
        change_next_loop = 0
        reserve_start_list = ["(","[","{","*","?"]
        reserve_stop_list = [")","]","}","*","?"]
        seq_word, token_word = [],[]
        seq_dict = {}
        sequence = sequence.replace(" ","")
        last_item = len(sequence)
        sequence_fragment_counter = 1
        seq_written_marker = 1
        for i, nibble in enumerate(sequence):
            print sequence_fragment_counter, nibble #token_word
            print string, token, change_next_loop
            ## stream switcher ##
            if seq_written_marker != sequence_fragment_counter:
                if string == 1:
                    label = "string::"
                else:
                    label = "token::"
                seq_dict[sequence_fragment_counter] = label+token_word
                seq_written_marker += 1

            if nibble in reserve_start_list and change_next_loop == 1:
               sequence_fragment_counter += 1
               token_word = "".join(token_word)
               seq_dict[sequence_fragment_counter] = "token::"+token_word
               token_word=[]

            if i == (last_item-1):
                if string == 1:
                    seq_word.append(nibble)
                else:
                    token_word.append(nibble)
                token_word = "".join(token_word)
                if token_word != "":
                    seq_dict[sequence_fragment_counter] = "token::"+token_word
                seq_word = "".join(seq_word)
                if seq_word != "":
                    seq_dict[sequence_fragment_counter] = "string::"+seq_word
            else:
                if change_next_loop == 1 and nibble not in reserve_start_list:
                    string = 1
                    token = 0
                    change_next_loop = 0
                elif nibble in reserve_start_list:
                    string = 0
                    token = 1
                    change_next_loop = 0
                if nibble in reserve_stop_list:
                    change_next_loop = 1

            ## split stream maker ##
                if string == 1:
                    seq_word.append(nibble)
                    if token_word != []:
                        token_word = "".join(token_word)
                        seq_dict[sequence_fragment_counter] = "token::"+token_word
                        sequence_fragment_counter += 1
                        token_word = []
                else:
                    token_word.append(nibble)
                    if seq_word != []:
                        seq_word = "".join(seq_word)
                        seq_dict[sequence_fragment_counter] = "string::"+seq_word
                        sequence_fragment_counter += 1
                        seq_word = []

        print seq_dict


def main():
    sig= Signature_Element()
    sig.sequence  = "524946(46|58){4}434452367672736E"
    sig.offset = "0"
    sig.sequence = sig.do_grep_grammer_convert(sig.sequence)
    sig.sequence = sig.do_fragmenter(sig.sequence)

if __name__ == '__main__':
    main()