pySim/utils.py

# -*- coding: utf-8 -*-

""" pySim: various utilities
"""

import json
import abc
import string
from io import BytesIO
from typing import Optional, List, Dict, Any, Tuple, NewType

# Copyright (C) 2009-2010  Sylvain Munaut <tnt@246tNt.com>
# Copyright (C) 2021 Harald Welte <laforge@osmocom.org>
#
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 2 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program.  If not, see <http://www.gnu.org/licenses/>.
#

# just to differentiate strings of hex nibbles from everything else
Hexstr = NewType('Hexstr', str)
SwHexstr = NewType('SwHexstr', str)
SwMatchstr = NewType('SwMatchstr', str)
ResTuple = Tuple[Hexstr, SwHexstr]

def h2b(s: Hexstr) -> bytearray:
    """convert from a string of hex nibbles to a sequence of bytes"""
    return bytearray.fromhex(s)


def b2h(b: bytearray) -> Hexstr:
    """convert from a sequence of bytes to a string of hex nibbles"""
    return ''.join(['%02x' % (x) for x in b])


def h2i(s: Hexstr) -> List[int]:
    """convert from a string of hex nibbles to a list of integers"""
    return [(int(x, 16) << 4)+int(y, 16) for x, y in zip(s[0::2], s[1::2])]


def i2h(s: List[int]) -> Hexstr:
    """convert from a list of integers to a string of hex nibbles"""
    return ''.join(['%02x' % (x) for x in s])


def h2s(s: Hexstr) -> str:
    """convert from a string of hex nibbles to an ASCII string"""
    return ''.join([chr((int(x, 16) << 4)+int(y, 16)) for x, y in zip(s[0::2], s[1::2])
                    if int(x + y, 16) != 0xff])


def s2h(s: str) -> Hexstr:
    """convert from an ASCII string to a string of hex nibbles"""
    b = bytearray()
    b.extend(map(ord, s))
    return b2h(b)


def i2s(s: List[int]) -> str:
    """convert from a list of integers to an ASCII string"""
    return ''.join([chr(x) for x in s])


def swap_nibbles(s: Hexstr) -> Hexstr:
    """swap the nibbles in a hex string"""
    return ''.join([x+y for x, y in zip(s[1::2], s[0::2])])


def rpad(s: str, l: int, c='f') -> str:
    """pad string on the right side.
    Args:
            s : string to pad
            l : total length to pad to
            c : padding character
    Returns:
            String 's' padded with as many 'c' as needed to reach total length of 'l'
    """
    return s + c * (l - len(s))


def lpad(s: str, l: int, c='f') -> str:
    """pad string on the left side.
    Args:
            s : string to pad
            l : total length to pad to
            c : padding character
    Returns:
            String 's' padded with as many 'c' as needed to reach total length of 'l'
    """
    return c * (l - len(s)) + s


def half_round_up(n: int) -> int:
    return (n + 1)//2


def str_sanitize(s: str) -> str:
    """replace all non printable chars, line breaks and whitespaces, with ' ', make sure that
    there are no whitespaces at the end and at the beginning of the string.

    Args:
            s : string to sanitize
    Returns:
            filtered result of string 's'
    """

    chars_to_keep = string.digits + string.ascii_letters + string.punctuation
    res = ''.join([c if c in chars_to_keep else ' ' for c in s])
    return res.strip()

#########################################################################
# poor man's COMPREHENSION-TLV decoder.
#########################################################################


def comprehensiontlv_parse_tag_raw(binary: bytes) -> Tuple[int, bytes]:
    """Parse a single Tag according to ETSI TS 101 220 Section 7.1.1"""
    if binary[0] in [0x00, 0x80, 0xff]:
        raise ValueError("Found illegal value 0x%02x in %s" %
                         (binary[0], binary))
    if binary[0] == 0x7f:
        # three-byte tag
        tag = binary[0] << 16 | binary[1] << 8 | binary[2]
        return (tag, binary[3:])
    elif binary[0] == 0xff:
        return None, binary
    else:
        # single byte tag
        tag = binary[0]
        return (tag, binary[1:])


def comprehensiontlv_parse_tag(binary: bytes) -> Tuple[dict, bytes]:
    """Parse a single Tag according to ETSI TS 101 220 Section 7.1.1"""
    if binary[0] in [0x00, 0x80, 0xff]:
        raise ValueError("Found illegal value 0x%02x in %s" %
                         (binary[0], binary))
    if binary[0] == 0x7f:
        # three-byte tag
        tag = (binary[1] & 0x7f) << 8
        tag |= binary[2]
        compr = True if binary[1] & 0x80 else False
        return ({'comprehension': compr, 'tag': tag}, binary[3:])
    else:
        # single byte tag
        tag = binary[0] & 0x7f
        compr = True if binary[0] & 0x80 else False
        return ({'comprehension': compr, 'tag': tag}, binary[1:])


def comprehensiontlv_encode_tag(tag) -> bytes:
    """Encode a single Tag according to ETSI TS 101 220 Section 7.1.1"""
    # permit caller to specify tag also as integer value
    if isinstance(tag, int):
        compr = True if tag < 0xff and tag & 0x80 else False
        tag = {'tag': tag, 'comprehension': compr}
    compr = tag.get('comprehension', False)
    if tag['tag'] in [0x00, 0x80, 0xff] or tag['tag'] > 0xff:
        # 3-byte format
        byte3 = tag['tag'] & 0xff
        byte2 = (tag['tag'] >> 8) & 0x7f
        if compr:
            byte2 |= 0x80
        return b'\x7f' + byte2.to_bytes(1, 'big') + byte3.to_bytes(1, 'big')
    else:
        # 1-byte format
        ret = tag['tag']
        if compr:
            ret |= 0x80
        return ret.to_bytes(1, 'big')

# length value coding is equal to BER-TLV


def comprehensiontlv_parse_one(binary: bytes) -> Tuple[dict, int, bytes, bytes]:
    """Parse a single TLV IE at the start of the given binary data.
    Args:
            binary : binary input data of BER-TLV length field
    Returns:
            Tuple of (tag:dict, len:int, remainder:bytes)
    """
    (tagdict, remainder) = comprehensiontlv_parse_tag(binary)
    (length, remainder) = bertlv_parse_len(remainder)
    value = remainder[:length]
    remainder = remainder[length:]
    return (tagdict, length, value, remainder)


#########################################################################
# poor man's BER-TLV decoder. To be a more sophisticated OO library later
#########################################################################

def bertlv_parse_tag_raw(binary: bytes) -> Tuple[int, bytes]:
    """Get a single raw Tag from start of input according to ITU-T X.690 8.1.2
    Args:
            binary : binary input data of BER-TLV length field
    Returns:
    Tuple of (tag:int, remainder:bytes)
    """
    # check for FF padding at the end, as customary in SIM card files
    if binary[0] == 0xff and len(binary) == 1 or binary[0] == 0xff and binary[1] == 0xff:
        return None, binary
    tag = binary[0] & 0x1f
    if tag <= 30:
        return binary[0], binary[1:]
    else:  # multi-byte tag
        tag = binary[0]
        i = 1
        last = False
        while not last:
            last = False if binary[i] & 0x80 else True
            tag <<= 8
            tag |= binary[i]
            i += 1
        return tag, binary[i:]


def bertlv_parse_tag(binary: bytes) -> Tuple[dict, bytes]:
    """Parse a single Tag value according to ITU-T X.690 8.1.2
    Args:
            binary : binary input data of BER-TLV length field
    Returns:
            Tuple of ({class:int, constructed:bool, tag:int}, remainder:bytes)
    """
    cls = binary[0] >> 6
    constructed = True if binary[0] & 0x20 else False
    tag = binary[0] & 0x1f
    if tag <= 30:
        return ({'class': cls, 'constructed': constructed, 'tag': tag}, binary[1:])
    else:  # multi-byte tag
        tag = 0
        i = 1
        last = False
        while not last:
            last = False if binary[i] & 0x80 else True
            tag <<= 7
            tag |= binary[i] & 0x7f
            i += 1
        return ({'class': cls, 'constructed': constructed, 'tag': tag}, binary[i:])


def bertlv_encode_tag(t) -> bytes:
    """Encode a single Tag value according to ITU-T X.690 8.1.2
    """
    def get_top7_bits(inp: int) -> Tuple[int, int]:
        """Get top 7 bits of integer. Returns those 7 bits as integer and the remaining LSBs."""
        remain_bits = inp.bit_length()
        if remain_bits >= 7:
            bitcnt = 7
        else:
            bitcnt = remain_bits
        outp = inp >> (remain_bits - bitcnt)
        remainder = inp & ~ (inp << (remain_bits - bitcnt))
        return outp, remainder

    if isinstance(t, int):
        # FIXME: multiple byte tags
        tag = t & 0x1f
        constructed = True if t & 0x20 else False
        cls = t >> 6
    else:
        tag = t['tag']
        constructed = t['constructed']
        cls = t['class']
    if tag <= 30:
        t = tag & 0x1f
        if constructed:
            t |= 0x20
        t |= (cls & 3) << 6
        return bytes([t])
    else:  # multi-byte tag
        t = 0x1f
        if constructed:
            t |= 0x20
        t |= (cls & 3) << 6
        tag_bytes = bytes([t])
        remain = tag
        while True:
            t, remain = get_top7_bits(remain)
            if remain:
                t |= 0x80
            tag_bytes += bytes([t])
            if not remain:
                break
        return tag_bytes


def bertlv_parse_len(binary: bytes) -> Tuple[int, bytes]:
    """Parse a single Length value according to ITU-T X.690 8.1.3;
    only the definite form is supported here.
    Args:
            binary : binary input data of BER-TLV length field
    Returns:
            Tuple of (length, remainder)
    """
    if binary[0] < 0x80:
        return (binary[0], binary[1:])
    else:
        num_len_oct = binary[0] & 0x7f
        length = 0
        for i in range(1, 1+num_len_oct):
            length <<= 8
            length |= binary[i]
        return (length, binary[1+num_len_oct:])


def bertlv_encode_len(length: int) -> bytes:
    """Encode a single Length value according to ITU-T X.690 8.1.3;
    only the definite form is supported here.
    Args:
            length : length value to be encoded
    Returns:
            binary output data of BER-TLV length field
    """
    if length < 0x80:
        return length.to_bytes(1, 'big')
    elif length <= 0xff:
        return b'\x81' + length.to_bytes(1, 'big')
    elif length <= 0xffff:
        return b'\x82' + length.to_bytes(2, 'big')
    elif length <= 0xffffff:
        return b'\x83' + length.to_bytes(3, 'big')
    elif length <= 0xffffffff:
        return b'\x84' + length.to_bytes(4, 'big')
    else:
        raise ValueError("Length > 32bits not supported")


def bertlv_parse_one(binary: bytes) -> Tuple[dict, int, bytes, bytes]:
    """Parse a single TLV IE at the start of the given binary data.
    Args:
            binary : binary input data of BER-TLV length field
    Returns:
            Tuple of (tag:dict, len:int, remainder:bytes)
    """
    (tagdict, remainder) = bertlv_parse_tag(binary)
    (length, remainder) = bertlv_parse_len(remainder)
    value = remainder[:length]
    remainder = remainder[length:]
    return (tagdict, length, value, remainder)


# IMSI encoded format:
# For IMSI 0123456789ABCDE:
#
# |     byte 1      | 2 upper | 2 lower  | 3 upper | 3 lower | ... | 9 upper | 9 lower |
# | length in bytes |    0    | odd/even |    2    |    1    | ... |    E    |    D    |
#
# If the IMSI is less than 15 characters, it should be padded with 'f' from the end.
#
# The length is the total number of bytes used to encoded the IMSI. This includes the odd/even
# parity bit. E.g. an IMSI of length 14 is 8 bytes long, not 7, as it uses bytes 2 to 9 to
# encode itself.
#
# Because of this, an odd length IMSI fits exactly into len(imsi) + 1 // 2 bytes, whereas an
# even length IMSI only uses half of the last byte.

def enc_imsi(imsi: str):
    """Converts a string IMSI into the encoded value of the EF"""
    l = half_round_up(
        len(imsi) + 1)  # Required bytes - include space for odd/even indicator
    oe = len(imsi) & 1			# Odd (1) / Even (0)
    ei = '%02x' % l + swap_nibbles('%01x%s' % ((oe << 3) | 1, rpad(imsi, 15)))
    return ei


def dec_imsi(ef: Hexstr) -> Optional[str]:
    """Converts an EF value to the IMSI string representation"""
    if len(ef) < 4:
        return None
    l = int(ef[0:2], 16) * 2		# Length of the IMSI string
    l = l - 1						# Encoded length byte includes oe nibble
    swapped = swap_nibbles(ef[2:]).rstrip('f')
    if len(swapped) < 1:
        return None
    oe = (int(swapped[0]) >> 3) & 1  # Odd (1) / Even (0)
    if not oe:
        # if even, only half of last byte was used
        l = l-1
    if l != len(swapped) - 1:
        return None
    imsi = swapped[1:]
    return imsi


def dec_iccid(ef: Hexstr) -> str:
    return swap_nibbles(ef).strip('f')


def enc_iccid(iccid: str) -> Hexstr:
    return swap_nibbles(rpad(iccid, 20))


def enc_plmn(mcc: Hexstr, mnc: Hexstr) -> Hexstr:
    """Converts integer MCC/MNC into 3 bytes for EF"""

    # Make sure there are no excess whitespaces in the input
    # parameters
    mcc = mcc.strip()
    mnc = mnc.strip()

    # Make sure that MCC/MNC are correctly padded with leading
    # zeros or 'F', depending on the length.
    if len(mnc) == 0:
        mnc = "FFF"
    elif len(mnc) == 1:
        mnc = "F0" + mnc
    elif len(mnc) == 2:
        mnc += "F"

    if len(mcc) == 0:
        mcc = "FFF"
    elif len(mcc) == 1:
        mcc = "00" + mcc
    elif len(mcc) == 2:
        mcc = "0" + mcc

    return (mcc[1] + mcc[0]) + (mnc[2] + mcc[2]) + (mnc[1] + mnc[0])


def dec_plmn(threehexbytes: Hexstr) -> dict:
    res = {'mcc': "0", 'mnc': "0"}
    dec_mcc_from_plmn_str(threehexbytes)
    res['mcc'] = dec_mcc_from_plmn_str(threehexbytes)
    res['mnc'] = dec_mnc_from_plmn_str(threehexbytes)
    return res


def dec_spn(ef):
    """Obsolete, kept for API compatibility"""
    from ts_51_011 import EF_SPN
    abstract_data = EF_SPN().decode_hex(ef)
    show_in_hplmn = abstract_data['show_in_hplmn']
    hide_in_oplmn = abstract_data['hide_in_oplmn']
    name = abstract_data['spn']
    return (name, show_in_hplmn, hide_in_oplmn)


def enc_spn(name: str, show_in_hplmn=False, hide_in_oplmn=False):
    """Obsolete, kept for API compatibility"""
    from ts_51_011 import EF_SPN
    abstract_data = {
        'hide_in_oplmn': hide_in_oplmn,
        'show_in_hplmn': show_in_hplmn,
        'spn': name,
    }
    return EF_SPN().encode_hex(abstract_data)


def hexstr_to_Nbytearr(s, nbytes):
    return [s[i:i+(nbytes*2)] for i in range(0, len(s), (nbytes*2))]

# Accepts hex string representing three bytes


def dec_mcc_from_plmn(plmn: Hexstr) -> int:
    ia = h2i(plmn)
    digit1 = ia[0] & 0x0F		# 1st byte, LSB
    digit2 = (ia[0] & 0xF0) >> 4  # 1st byte, MSB
    digit3 = ia[1] & 0x0F		# 2nd byte, LSB
    if digit3 == 0xF and digit2 == 0xF and digit1 == 0xF:
        return 0xFFF  # 4095
    return derive_mcc(digit1, digit2, digit3)


def dec_mcc_from_plmn_str(plmn: Hexstr) -> str:
    digit1 = plmn[1]  # 1st byte, LSB
    digit2 = plmn[0]  # 1st byte, MSB
    digit3 = plmn[3]  # 2nd byte, LSB
    res = digit1 + digit2 + digit3
    return res.upper().strip("F")


def dec_mnc_from_plmn(plmn: Hexstr) -> int:
    ia = h2i(plmn)
    digit1 = ia[2] & 0x0F		# 3rd byte, LSB
    digit2 = (ia[2] & 0xF0) >> 4  # 3rd byte, MSB
    digit3 = (ia[1] & 0xF0) >> 4  # 2nd byte, MSB
    if digit3 == 0xF and digit2 == 0xF and digit1 == 0xF:
        return 0xFFF  # 4095
    return derive_mnc(digit1, digit2, digit3)


def dec_mnc_from_plmn_str(plmn: Hexstr) -> str:
    digit1 = plmn[5]  # 3rd byte, LSB
    digit2 = plmn[4]  # 3rd byte, MSB
    digit3 = plmn[2]  # 2nd byte, MSB
    res = digit1 + digit2 + digit3
    return res.upper().strip("F")


def dec_act(twohexbytes: Hexstr) -> List[str]:
    act_list = [
        {'bit': 15, 'name': "UTRAN"},
        {'bit': 14, 'name': "E-UTRAN"},
        {'bit': 11, 'name': "NG-RAN"},
        {'bit':  7, 'name': "GSM"},
        {'bit':  6, 'name': "GSM COMPACT"},
        {'bit':  5, 'name': "cdma2000 HRPD"},
        {'bit':  4, 'name': "cdma2000 1xRTT"},
    ]
    ia = h2i(twohexbytes)
    u16t = (ia[0] << 8) | ia[1]
    sel = []
    for a in act_list:
        if u16t & (1 << a['bit']):
            if a['name'] == "E-UTRAN":
                # The Access technology identifier of E-UTRAN
                # allows a more detailed specification:
                if u16t & (1 << 13) and u16t & (1 << 12):
                    sel.append("E-UTRAN WB-S1")
                    sel.append("E-UTRAN NB-S1")
                elif u16t & (1 << 13):
                    sel.append("E-UTRAN WB-S1")
                elif u16t & (1 << 12):
                    sel.append("E-UTRAN NB-S1")
                else:
                    sel.append("E-UTRAN")
            else:
                sel.append(a['name'])
    return sel


def dec_xplmn_w_act(fivehexbytes: Hexstr) -> Dict[str, Any]:
    res = {'mcc': "0", 'mnc': "0", 'act': []}
    plmn_chars = 6
    act_chars = 4
    # first three bytes (six ascii hex chars)
    plmn_str = fivehexbytes[:plmn_chars]
    # two bytes after first three bytes
    act_str = fivehexbytes[plmn_chars:plmn_chars + act_chars]
    res['mcc'] = dec_mcc_from_plmn_str(plmn_str)
    res['mnc'] = dec_mnc_from_plmn_str(plmn_str)
    res['act'] = dec_act(act_str)
    return res


def format_xplmn_w_act(hexstr):
    s = ""
    for rec_data in hexstr_to_Nbytearr(hexstr, 5):
        rec_info = dec_xplmn_w_act(rec_data)
        if rec_info['mcc'] == "" and rec_info['mnc'] == "":
            rec_str = "unused"
        else:
            rec_str = "MCC: %s MNC: %s AcT: %s" % (
                rec_info['mcc'], rec_info['mnc'], ", ".join(rec_info['act']))
        s += "\t%s # %s\n" % (rec_data, rec_str)
    return s


def dec_loci(hexstr):
    res = {'tmsi': '',  'mcc': 0, 'mnc': 0, 'lac': '', 'status': 0}
    res['tmsi'] = hexstr[:8]
    res['mcc'] = dec_mcc_from_plmn(hexstr[8:14])
    res['mnc'] = dec_mnc_from_plmn(hexstr[8:14])
    res['lac'] = hexstr[14:18]
    res['status'] = h2i(hexstr[20:22])
    return res


def dec_psloci(hexstr):
    res = {'p-tmsi': '', 'p-tmsi-sig': '', 'mcc': 0,
           'mnc': 0, 'lac': '', 'rac': '', 'status': 0}
    res['p-tmsi'] = hexstr[:8]
    res['p-tmsi-sig'] = hexstr[8:14]
    res['mcc'] = dec_mcc_from_plmn(hexstr[14:20])
    res['mnc'] = dec_mnc_from_plmn(hexstr[14:20])
    res['lac'] = hexstr[20:24]
    res['rac'] = hexstr[24:26]
    res['status'] = h2i(hexstr[26:28])
    return res


def dec_epsloci(hexstr):
    res = {'guti': '', 'mcc': 0, 'mnc': 0, 'tac': '', 'status': 0}
    res['guti'] = hexstr[:24]
    res['tai'] = hexstr[24:34]
    res['mcc'] = dec_mcc_from_plmn(hexstr[24:30])
    res['mnc'] = dec_mnc_from_plmn(hexstr[24:30])
    res['tac'] = hexstr[30:34]
    res['status'] = h2i(hexstr[34:36])
    return res


def dec_xplmn(threehexbytes: Hexstr) -> dict:
    res = {'mcc': 0, 'mnc': 0, 'act': []}
    plmn_chars = 6
    # first three bytes (six ascii hex chars)
    plmn_str = threehexbytes[:plmn_chars]
    res['mcc'] = dec_mcc_from_plmn_str(plmn_str)
    res['mnc'] = dec_mnc_from_plmn_str(plmn_str)
    return res


def format_xplmn(hexstr: Hexstr) -> str:
    s = ""
    for rec_data in hexstr_to_Nbytearr(hexstr, 3):
        rec_info = dec_xplmn(rec_data)
        if not rec_info['mcc'] and not rec_info['mnc']:
            rec_str = "unused"
        else:
            rec_str = "MCC: %s MNC: %s" % (rec_info['mcc'], rec_info['mnc'])
        s += "\t%s # %s\n" % (rec_data, rec_str)
    return s


def derive_milenage_opc(ki_hex: Hexstr, op_hex: Hexstr) -> Hexstr:
    """
    Run the milenage algorithm to calculate OPC from Ki and OP
    """
    from Cryptodome.Cipher import AES
    # pylint: disable=no-name-in-module
    from Cryptodome.Util.strxor import strxor
    from pySim.utils import b2h

    # We pass in hex string and now need to work on bytes
    ki_bytes = bytes(h2b(ki_hex))
    op_bytes = bytes(h2b(op_hex))
    aes = AES.new(ki_bytes, AES.MODE_ECB)
    opc_bytes = aes.encrypt(op_bytes)
    return b2h(strxor(opc_bytes, op_bytes))


def calculate_luhn(cc) -> int:
    """
    Calculate Luhn checksum used in e.g. ICCID and IMEI
    """
    num = list(map(int, str(cc)))
    check_digit = 10 - sum(num[-2::-2] + [sum(divmod(d * 2, 10))
                           for d in num[::-2]]) % 10
    return 0 if check_digit == 10 else check_digit


def mcc_from_imsi(imsi: str) -> Optional[str]:
    """
    Derive the MCC (Mobile Country Code) from the first three digits of an IMSI
    """
    if imsi == None:
        return None

    if len(imsi) > 3:
        return imsi[:3]
    else:
        return None


def mnc_from_imsi(imsi: str, long: bool = False) -> Optional[str]:
    """
    Derive the MNC (Mobile Country Code) from the 4th to 6th digit of an IMSI
    """
    if imsi == None:
        return None

    if len(imsi) > 3:
        if long:
            return imsi[3:6]
        else:
            return imsi[3:5]
    else:
        return None


def derive_mcc(digit1: int, digit2: int, digit3: int) -> int:
    """
    Derive decimal representation of the MCC (Mobile Country Code)
    from three given digits.
    """

    mcc = 0

    if digit1 != 0x0f:
        mcc += digit1 * 100
    if digit2 != 0x0f:
        mcc += digit2 * 10
    if digit3 != 0x0f:
        mcc += digit3

    return mcc


def derive_mnc(digit1: int, digit2: int, digit3: int = 0x0f) -> int:
    """
    Derive decimal representation of the MNC (Mobile Network Code)
    from two or (optionally) three given digits.
    """

    mnc = 0

    # 3-rd digit is optional for the MNC. If present
    # the algorythm is the same as for the MCC.
    if digit3 != 0x0f:
        return derive_mcc(digit1, digit2, digit3)

    if digit1 != 0x0f:
        mnc += digit1 * 10
    if digit2 != 0x0f:
        mnc += digit2

    return mnc


def dec_msisdn(ef_msisdn: Hexstr) -> Optional[Tuple[int, int, Optional[str]]]:
    """
    Decode MSISDN from EF.MSISDN or EF.ADN (same structure).
    See 3GPP TS 31.102, section 4.2.26 and 4.4.2.3.
    """

    # Convert from str to (kind of) 'bytes'
    ef_msisdn = h2b(ef_msisdn)

    # Make sure mandatory fields are present
    if len(ef_msisdn) < 14:
        raise ValueError("EF.MSISDN is too short")

    # Skip optional Alpha Identifier
    xlen = len(ef_msisdn) - 14
    msisdn_lhv = ef_msisdn[xlen:]

    # Parse the length (in bytes) of the BCD encoded number
    bcd_len = msisdn_lhv[0]
    # BCD length = length of dial num (max. 10 bytes) + 1 byte ToN and NPI
    if bcd_len == 0xff:
        return None
    elif bcd_len > 11 or bcd_len < 1:
        raise ValueError(
            "Length of MSISDN (%d bytes) is out of range" % bcd_len)

    # Parse ToN / NPI
    ton = (msisdn_lhv[1] >> 4) & 0x07
    npi = msisdn_lhv[1] & 0x0f
    bcd_len -= 1

    # No MSISDN?
    if not bcd_len:
        return (npi, ton, None)

    msisdn = swap_nibbles(b2h(msisdn_lhv[2:][:bcd_len])).rstrip('f')
    # International number 10.5.118/3GPP TS 24.008
    if ton == 0x01:
        msisdn = '+' + msisdn

    return (npi, ton, msisdn)


def enc_msisdn(msisdn: str, npi: int = 0x01, ton: int = 0x03) -> Hexstr:
    """
    Encode MSISDN as LHV so it can be stored to EF.MSISDN.
    See 3GPP TS 31.102, section 4.2.26 and 4.4.2.3. (The result
    will not contain the optional Alpha Identifier at the beginning.)

    Default NPI / ToN values:
      - NPI: ISDN / telephony numbering plan (E.164 / E.163),
      - ToN: network specific or international number (if starts with '+').
    """

    # If no MSISDN is supplied then encode the file contents as all "ff"
    if msisdn == "" or msisdn == "+":
        return "ff" * 14

    # Leading '+' indicates International Number
    if msisdn[0] == '+':
        msisdn = msisdn[1:]
        ton = 0x01

    # An MSISDN must not exceed 20 digits
    if len(msisdn) > 20:
        raise ValueError("msisdn must not be longer than 20 digits")

    # Append 'f' padding if number of digits is odd
    if len(msisdn) % 2 > 0:
        msisdn += 'f'

    # BCD length also includes NPI/ToN header
    bcd_len = len(msisdn) // 2 + 1
    npi_ton = (npi & 0x0f) | ((ton & 0x07) << 4) | 0x80
    bcd = rpad(swap_nibbles(msisdn), 10 * 2)  # pad to 10 octets

    return ('%02x' % bcd_len) + ('%02x' % npi_ton) + bcd + ("ff" * 2)


def dec_st(st, table="sim") -> str:
    """
    Parses the EF S/U/IST and prints the list of available services in EF S/U/IST
    """

    if table == "isim":
        from pySim.ts_31_103 import EF_IST_map
        lookup_map = EF_IST_map
    elif table == "usim":
        from pySim.ts_31_102 import EF_UST_map
        lookup_map = EF_UST_map
    else:
        from pySim.ts_51_011 import EF_SST_map
        lookup_map = EF_SST_map

    st_bytes = [st[i:i+2] for i in range(0, len(st), 2)]

    avail_st = ""
    # Get each byte and check for available services
    for i in range(0, len(st_bytes)):
        # Byte i contains info about Services num (8i+1) to num (8i+8)
        byte = int(st_bytes[i], 16)
        # Services in each byte are in order MSB to LSB
        # MSB - Service (8i+8)
        # LSB - Service (8i+1)
        for j in range(1, 9):
            if byte & 0x01 == 0x01 and ((8*i) + j in lookup_map):
                # Byte X contains info about Services num (8X-7) to num (8X)
                # bit = 1: service available
                # bit = 0: service not available
                avail_st += '\tService %d - %s\n' % (
                    (8*i) + j, lookup_map[(8*i) + j])
            byte = byte >> 1
    return avail_st


def first_TLV_parser(bytelist):
    '''
    first_TLV_parser([0xAA, 0x02, 0xAB, 0xCD, 0xFF, 0x00]) -> (170, 2, [171, 205])

    parses first TLV format record in a list of bytelist
    returns a 3-Tuple: Tag, Length, Value
    Value is a list of bytes
    parsing of length is ETSI'style 101.220
    '''
    Tag = bytelist[0]
    if bytelist[1] == 0xFF:
        Len = bytelist[2]*256 + bytelist[3]
        Val = bytelist[4:4+Len]
    else:
        Len = bytelist[1]
        Val = bytelist[2:2+Len]
    return (Tag, Len, Val)


def TLV_parser(bytelist):
    '''
    TLV_parser([0xAA, ..., 0xFF]) -> [(T, L, [V]), (T, L, [V]), ...]

    loops on the input list of bytes with the "first_TLV_parser()" function
    returns a list of 3-Tuples
    '''
    ret = []
    while len(bytelist) > 0:
        T, L, V = first_TLV_parser(bytelist)
        if T == 0xFF:
            # padding bytes
            break
        ret.append((T, L, V))
        # need to manage length of L
        if L > 0xFE:
            bytelist = bytelist[L+4:]
        else:
            bytelist = bytelist[L+2:]
    return ret


def enc_st(st, service, state=1):
    """
    Encodes the EF S/U/IST/EST and returns the updated Service Table

    Parameters:
            st - Current value of SIM/USIM/ISIM Service Table
            service - Service Number to encode as activated/de-activated
            state - 1 mean activate, 0 means de-activate

    Returns:
            s - Modified value of SIM/USIM/ISIM Service Table

    Default values:
            - state: 1 - Sets the particular Service bit to 1
    """
    st_bytes = [st[i:i+2] for i in range(0, len(st), 2)]

    s = ""
    # Check whether the requested service is present in each byte
    for i in range(0, len(st_bytes)):
        # Byte i contains info about Services num (8i+1) to num (8i+8)
        if service in range((8*i) + 1, (8*i) + 9):
            byte = int(st_bytes[i], 16)
            # Services in each byte are in order MSB to LSB
            # MSB - Service (8i+8)
            # LSB - Service (8i+1)
            mod_byte = 0x00
            # Copy bit by bit contents of byte to mod_byte with modified bit
            # for requested service
            for j in range(1, 9):
                mod_byte = mod_byte >> 1
                if service == (8*i) + j:
                    mod_byte = state == 1 and mod_byte | 0x80 or mod_byte & 0x7f
                else:
                    mod_byte = byte & 0x01 == 0x01 and mod_byte | 0x80 or mod_byte & 0x7f
                byte = byte >> 1

            s += ('%02x' % (mod_byte))
        else:
            s += st_bytes[i]

    return s


def dec_addr_tlv(hexstr):
    """
    Decode hex string to get EF.P-CSCF Address or EF.ePDGId or EF.ePDGIdEm.
    See 3GPP TS 31.102 version 13.4.0 Release 13, section 4.2.8, 4.2.102 and 4.2.104.
    """

    # Convert from hex str to int bytes list
    addr_tlv_bytes = h2i(hexstr)

    # Get list of tuples containing parsed TLVs
    tlvs = TLV_parser(addr_tlv_bytes)

    for tlv in tlvs:
        # tlv = (T, L, [V])
        # T = Tag
        # L = Length
        # [V] = List of value

        # Invalid Tag value scenario
        if tlv[0] != 0x80:
            continue

        # Empty field - Zero length
        if tlv[1] == 0:
            continue

        # First byte in the value has the address type
        addr_type = tlv[2][0]
        # TODO: Support parsing of IPv6
        # Address Type: 0x00 (FQDN), 0x01 (IPv4), 0x02 (IPv6), other (Reserved)
        if addr_type == 0x00:  # FQDN
            # Skip address tye byte i.e. first byte in value list
            content = tlv[2][1:]
            return (i2s(content), '00')

        elif addr_type == 0x01:  # IPv4
            # Skip address tye byte i.e. first byte in value list
            # Skip the unused byte in Octect 4 after address type byte as per 3GPP TS 31.102
            ipv4 = tlv[2][2:]
            content = '.'.join(str(x) for x in ipv4)
            return (content, '01')
        else:
            raise ValueError("Invalid address type")

    return (None, None)


def enc_addr_tlv(addr, addr_type='00'):
    """
    Encode address TLV object used in EF.P-CSCF Address, EF.ePDGId and EF.ePDGIdEm.
    See 3GPP TS 31.102 version 13.4.0 Release 13, section 4.2.8, 4.2.102 and 4.2.104.

    Default values:
      - addr_type: 00 - FQDN format of Address
    """

    s = ""

    # TODO: Encoding of IPv6 address
    if addr_type == '00':  # FQDN
        hex_str = s2h(addr)
        s += '80' + ('%02x' % ((len(hex_str)//2)+1)) + '00' + hex_str
    elif addr_type == '01':  # IPv4
        ipv4_list = addr.split('.')
        ipv4_str = ""
        for i in ipv4_list:
            ipv4_str += ('%02x' % (int(i)))

        # Unused bytes shall be set to 'ff'. i.e 4th Octet after Address Type is not used
        # IPv4 Address is in octet 5 to octet 8 of the TLV data object
        s += '80' + ('%02x' % ((len(ipv4_str)//2)+2)) + '01' + 'ff' + ipv4_str

    return s


def is_hex(string: str, minlen: int = 2, maxlen: Optional[int] = None) -> bool:
    """
    Check if a string is a valid hexstring
    """

    # Filter obviously bad strings
    if not string:
        return False
    if len(string) < minlen or minlen < 2:
        return False
    if len(string) % 2:
        return False
    if maxlen and len(string) > maxlen:
        return False

    # Try actual encoding to be sure
    try:
        try_encode = h2b(string)
        return True
    except:
        return False


def sanitize_pin_adm(pin_adm, pin_adm_hex=None) -> Hexstr:
    """
    The ADM pin can be supplied either in its hexadecimal form or as
    ascii string. This function checks the supplied opts parameter and
    returns the pin_adm as hex encoded string, regardless in which form
    it was originally supplied by the user
    """

    if pin_adm is not None:
        if len(pin_adm) <= 8:
            pin_adm = ''.join(['%02x' % (ord(x)) for x in pin_adm])
            pin_adm = rpad(pin_adm, 16)

        else:
            raise ValueError("PIN-ADM needs to be <=8 digits (ascii)")

    if pin_adm_hex is not None:
        if len(pin_adm_hex) == 16:
            pin_adm = pin_adm_hex
            # Ensure that it's hex-encoded
            try:
                try_encode = h2b(pin_adm)
            except ValueError:
                raise ValueError(
                    "PIN-ADM needs to be hex encoded using this option")
        else:
            raise ValueError(
                "PIN-ADM needs to be exactly 16 digits (hex encoded)")

    return pin_adm


def enc_ePDGSelection(hexstr, mcc, mnc, epdg_priority='0001', epdg_fqdn_format='00'):
    """
    Encode ePDGSelection so it can be stored at EF.ePDGSelection or EF.ePDGSelectionEm.
    See 3GPP TS 31.102 version 15.2.0 Release 15, section 4.2.104 and 4.2.106.

    Default values:
            - epdg_priority: '0001' - 1st Priority
            - epdg_fqdn_format: '00' - Operator Identifier FQDN
    """

    plmn1 = enc_plmn(mcc, mnc) + epdg_priority + epdg_fqdn_format
    # TODO: Handle encoding of Length field for length more than 127 Bytes
    content = '80' + ('%02x' % (len(plmn1)//2)) + plmn1
    content = rpad(content, len(hexstr))
    return content


def dec_ePDGSelection(sixhexbytes):
    """
    Decode ePDGSelection to get EF.ePDGSelection or EF.ePDGSelectionEm.
    See 3GPP TS 31.102 version 15.2.0 Release 15, section 4.2.104 and 4.2.106.
    """

    res = {'mcc': 0, 'mnc': 0, 'epdg_priority': 0, 'epdg_fqdn_format': ''}
    plmn_chars = 6
    epdg_priority_chars = 4
    epdg_fqdn_format_chars = 2
    # first three bytes (six ascii hex chars)
    plmn_str = sixhexbytes[:plmn_chars]
    # two bytes after first three bytes
    epdg_priority_str = sixhexbytes[plmn_chars:plmn_chars +
                                    epdg_priority_chars]
    # one byte after first five bytes
    epdg_fqdn_format_str = sixhexbytes[plmn_chars +
                                       epdg_priority_chars:plmn_chars + epdg_priority_chars + epdg_fqdn_format_chars]
    res['mcc'] = dec_mcc_from_plmn(plmn_str)
    res['mnc'] = dec_mnc_from_plmn(plmn_str)
    res['epdg_priority'] = epdg_priority_str
    res['epdg_fqdn_format'] = epdg_fqdn_format_str == '00' and 'Operator Identifier FQDN' or 'Location based FQDN'
    return res


def format_ePDGSelection(hexstr):
    ePDGSelection_info_tag_chars = 2
    ePDGSelection_info_tag_str = hexstr[:2]
    s = ""
    # Minimum length
    len_chars = 2
    # TODO: Need to determine length properly - definite length support only
    # Inconsistency in spec: 3GPP TS 31.102 version 15.2.0 Release 15, 4.2.104
    # As per spec, length is 5n, n - number of PLMNs
    # But, each PLMN entry is made of PLMN (3 Bytes) + ePDG Priority (2 Bytes) + ePDG FQDN format (1 Byte)
    # Totalling to 6 Bytes, maybe length should be 6n
    len_str = hexstr[ePDGSelection_info_tag_chars:ePDGSelection_info_tag_chars+len_chars]

    # Not programmed scenario
    if int(len_str, 16) == 255 or int(ePDGSelection_info_tag_str, 16) == 255:
        len_chars = 0
        ePDGSelection_info_tag_chars = 0
    if len_str[0] == '8':
        # The bits 7 to 1 denotes the number of length octets if length > 127
        if int(len_str[1]) > 0:
            # Update number of length octets
            len_chars = len_chars * int(len_str[1])
            len_str = hexstr[ePDGSelection_info_tag_chars:len_chars]

    content_str = hexstr[ePDGSelection_info_tag_chars+len_chars:]
    # Right pad to prevent index out of range - multiple of 6 bytes
    content_str = rpad(content_str, len(content_str) +
                       (12 - (len(content_str) % 12)))
    for rec_data in hexstr_to_Nbytearr(content_str, 6):
        rec_info = dec_ePDGSelection(rec_data)
        if rec_info['mcc'] == 0xFFF and rec_info['mnc'] == 0xFFF:
            rec_str = "unused"
        else:
            rec_str = "MCC: %03d MNC: %03d ePDG Priority: %s ePDG FQDN format: %s" % \
                (rec_info['mcc'], rec_info['mnc'],
                 rec_info['epdg_priority'], rec_info['epdg_fqdn_format'])
        s += "\t%s # %s\n" % (rec_data, rec_str)
    return s


def get_addr_type(addr):
    """
    Validates the given address and returns it's type (FQDN or IPv4 or IPv6)
    Return: 0x00 (FQDN), 0x01 (IPv4), 0x02 (IPv6), None (Bad address argument given)

    TODO: Handle IPv6
    """

    # Empty address string
    if not len(addr):
        return None

    addr_list = addr.split('.')

    # Check for IPv4/IPv6
    try:
        import ipaddress
        # Throws ValueError if addr is not correct
        ipa = ipaddress.ip_address(addr)

        if ipa.version == 4:
            return 0x01
        elif ipa.version == 6:
            return 0x02
    except Exception as e:
        invalid_ipv4 = True
        for i in addr_list:
            # Invalid IPv4 may qualify for a valid FQDN, so make check here
            # e.g. 172.24.15.300
            import re
            if not re.match('^[0-9_]+$', i):
                invalid_ipv4 = False
                break

        if invalid_ipv4:
            return None

    fqdn_flag = True
    for i in addr_list:
        # Only Alpha-numeric characters and hyphen - RFC 1035
        import re
        if not re.match("^[a-zA-Z0-9]+(?:-[a-zA-Z0-9]+)?$", i):
            fqdn_flag = False
            break

    # FQDN
    if fqdn_flag:
        return 0x00

    return None


def sw_match(sw: str, pattern: str) -> bool:
    """Match given SW against given pattern."""
    # Create a masked version of the returned status word
    sw_lower = sw.lower()
    sw_masked = ""
    for i in range(0, 4):
        if pattern[i] == '?':
            sw_masked = sw_masked + '?'
        elif pattern[i] == 'x':
            sw_masked = sw_masked + 'x'
        else:
            sw_masked = sw_masked + sw_lower[i]
    # Compare the masked version against the pattern
    return sw_masked == pattern


def tabulate_str_list(str_list, width: int = 79, hspace: int = 2, lspace: int = 1,
                      align_left: bool = True) -> str:
    """Pretty print a list of strings into a tabulated form.

    Args:
            width : total width in characters per line
            space : horizontal space between cells
            lspace : number of spaces before row
            align_lef : Align text to the left side
    Returns:
            multi-line string containing formatted table
    """
    if str_list == None:
        return ""
    if len(str_list) <= 0:
        return ""
    longest_str = max(str_list, key=len)
    cellwith = len(longest_str) + hspace
    cols = width // cellwith
    rows = (len(str_list) - 1) // cols + 1
    table = []
    for i in iter(range(rows)):
        str_list_row = str_list[i::rows]
        if (align_left):
            format_str_cell = '%%-%ds'
        else:
            format_str_cell = '%%%ds'
        format_str_row = (format_str_cell % cellwith) * len(str_list_row)
        format_str_row = (" " * lspace) + format_str_row
        table.append(format_str_row % tuple(str_list_row))
    return '\n'.join(table)


def auto_int(x):
    """Helper function for argparse to accept hexadecimal integers."""
    return int(x, 0)


def expand_hex(hexstring, length):
    """Expand a given hexstring to a specified length by replacing "." or ".."
       with a filler that is derived from the neighboring nibbles respective
       bytes. Usually this will be the nibble respective byte before "." or
       "..", execpt when the string begins with "." or "..", then the nibble
       respective byte after "." or ".." is used.". In case the string cannot
       be expanded for some reason, the input string is returned unmodified.

    Args:
            hexstring : hexstring to expand
            length : desired length of the resulting hexstring.
    Returns:
            expanded hexstring
    """

    # expand digit aligned
    if hexstring.count(".") == 1:
        pos = hexstring.index(".")
        if pos > 0:
            filler = hexstring[pos - 1]
        else:
            filler = hexstring[pos + 1]

        missing = length * 2 - (len(hexstring) - 1)
        if missing <= 0:
            return hexstring

        return hexstring.replace(".", filler * missing)

    # expand byte aligned
    elif hexstring.count("..") == 1:
        if len(hexstring) % 2:
            return hexstring

        pos = hexstring.index("..")

        if pos % 2:
            return hexstring

        if pos > 1:
            filler = hexstring[pos - 2:pos]
        else:
            filler = hexstring[pos + 2:pos+4]

        missing = length * 2 - (len(hexstring) - 2)
        if missing <= 0:
            return hexstring

        return hexstring.replace("..", filler * (missing // 2))

    # no change
    return hexstring


class JsonEncoder(json.JSONEncoder):
    """Extend the standard library JSONEncoder with support for more types."""

    def default(self, o):
        if isinstance(o, BytesIO) or isinstance(o, bytes) or isinstance(o, bytearray):
            return b2h(o)
        return json.JSONEncoder.default(self, o)


def boxed_heading_str(heading, width=80):
    """Generate a string that contains a boxed heading."""
    # Auto-enlarge box if heading exceeds length
    if len(heading) > width - 4:
        width = len(heading) + 4

    res = "#" * width
    fstr = "\n# %-" + str(width - 4) + "s #\n"
    res += fstr % (heading)
    res += "#" * width
    return res


class DataObject(abc.ABC):
    """A DataObject (DO) in the sense of ISO 7816-4.  Contrary to 'normal' TLVs where one
    simply has any number of different TLVs that may occur in any order at any point, ISO 7816
    has the habit of specifying TLV data but with very spcific ordering, or specific choices of
    tags at specific points in a stream.  This class tries to represent this."""

    def __init__(self, name: str, desc: Optional[str] = None, tag: Optional[int] = None):
        """
        Args:
            name: A brief, all-lowercase, underscore separated string identifier
            desc: A human-readable description of what this DO represents
            tag : The tag associated with this DO
        """
        self.name = name
        self.desc = desc
        self.tag = tag
        self.decoded = None
        self.encoded = None

    def __str__(self):
        return self.name

    def __repr__(self) -> str:
        return '%s(%s)' % (self.__class__, self.name)

    def __or__(self, other) -> 'DataObjectChoice':
        """OR-ing DataObjects together renders a DataObjectChoice."""
        if isinstance(other, DataObject):
            # DataObject | DataObject = DataObjectChoice
            return DataObjectChoice(None, members=[self, other])
        else:
            raise TypeError

    def __add__(self, other) -> 'DataObjectCollection':
        """ADD-ing DataObjects together renders a DataObjectCollection."""
        if isinstance(other, DataObject):
            # DataObject + DataObject = DataObjectCollectin
            return DataObjectCollection(None, members=[self, other])
        else:
            raise TypeError

    def _compute_tag(self) -> int:
        """Compute the tag (sometimes the tag encodes part of the value)."""
        return self.tag

    def to_dict(self) -> dict:
        """Return a dict in form "name: decoded_value" """
        return {self.name: self.decoded}

    @abc.abstractmethod
    def from_bytes(self, do: bytes):
        """Parse the value part of the DO into the internal state of this instance.
        Args:
            do : binary encoded bytes
        """

    @abc.abstractmethod
    def to_bytes(self) -> bytes:
        """Encode the internal state of this instance into the TLV value part.
        Returns:
            binary bytes encoding the internal state
        """

    def from_tlv(self, do: bytes) -> bytes:
        """Parse binary TLV representation into internal state.  The resulting decoded
        representation is _not_ returned, but just internalized in the object instance!
        Args:
            do : input bytes containing TLV-encoded representation
        Returns:
            bytes remaining at end of 'do' after parsing one TLV/DO.
        """
        if do[0] != self.tag:
            raise ValueError('%s: Can only decode tag 0x%02x' %
                             (self, self.tag))
        length = do[1]
        val = do[2:2+length]
        self.from_bytes(val)
        # return remaining bytes
        return do[2+length:]

    def to_tlv(self) -> bytes:
        """Encode internal representation to binary TLV.
        Returns:
            bytes encoded in TLV format.
        """
        val = self.to_bytes()
        return bertlv_encode_tag(self._compute_tag()) + bertlv_encode_len(len(val)) + val

    # 'codec' interface
    def decode(self, binary: bytes) -> Tuple[dict, bytes]:
        """Decode a single DOs from the input data.
        Args:
            binary : binary bytes of encoded data
        Returns:
            tuple of (decoded_result, binary_remainder)
        """
        tag = binary[0]
        if tag != self.tag:
            raise ValueError('%s: Unknown Tag 0x%02x in %s; expected 0x%02x' %
                             (self, tag, binary, self.tag))
        remainder = self.from_tlv(binary)
        return (self.to_dict(), remainder)

    # 'codec' interface
    def encode(self) -> bytes:
        return self.to_tlv()


class TL0_DataObject(DataObject):
    """Data Object that has Tag, Len=0 and no Value part."""

    def __init__(self, name: str, desc: str, tag: int, val=None):
        super().__init__(name, desc, tag)
        self.val = val

    def from_bytes(self, binary: bytes):
        if len(binary) != 0:
            raise ValueError
        self.decoded = self.val

    def to_bytes(self) -> bytes:
        return b''


class DataObjectCollection:
    """A DataObjectCollection consits of multiple Data Objects identified by their tags.
    A given encoded DO may contain any of them in any order, and may contain multiple instances
    of each DO."""

    def __init__(self, name: str, desc: Optional[str] = None, members=None):
        self.name = name
        self.desc = desc
        self.members = members or []
        self.members_by_tag = {}
        self.members_by_name = {}
        self.members_by_tag = {m.tag: m for m in members}
        self.members_by_name = {m.name: m for m in members}

    def __str__(self) -> str:
        member_strs = [str(x) for x in self.members]
        return '%s(%s)' % (self.name, ','.join(member_strs))

    def __repr__(self) -> str:
        member_strs = [repr(x) for x in self.members]
        return '%s(%s)' % (self.__class__, ','.join(member_strs))

    def __add__(self, other) -> 'DataObjectCollection':
        """Extending DataCollections with other DataCollections or DataObjects."""
        if isinstance(other, DataObjectCollection):
            # adding one collection to another
            members = self.members + other.members
            return DataObjectCollection(self.name, self.desc, members)
        elif isinstance(other, DataObject):
            # adding a member to a collection
            return DataObjectCollection(self.name, self.desc, self.members + [other])
        else:
            raise TypeError

    # 'codec' interface
    def decode(self, binary: bytes) -> Tuple[List, bytes]:
        """Decode any number of DOs from the collection until the end of the input data,
        or uninitialized memory (0xFF) is found.
        Args:
            binary : binary bytes of encoded data
        Returns:
            tuple of (decoded_result, binary_remainder)
        """
        res = []
        remainder = binary
        # iterate until no binary trailer is left
        while len(remainder):
            tag = remainder[0]
            if tag == 0xff:  # uninitialized memory at the end?
                return (res, remainder)
            if not tag in self.members_by_tag:
                raise ValueError('%s: Unknown Tag 0x%02x in %s; expected %s' %
                                 (self, tag, remainder, self.members_by_tag.keys()))
            obj = self.members_by_tag[tag]
            # DO from_tlv returns remainder of binary
            remainder = obj.from_tlv(remainder)
            # collect our results
            res.append(obj.to_dict())
        return (res, remainder)

    # 'codec' interface
    def encode(self, decoded) -> bytes:
        res = bytearray()
        for i in decoded:
            obj = self.members_by_name(i[0])
            res.append(obj.to_tlv())
        return res


class DataObjectChoice(DataObjectCollection):
    """One Data Object from within a choice, identified by its tag.
    This means that exactly one member of the choice must occur, and which one occurs depends
    on the tag."""

    def __add__(self, other):
        """We overload the add operator here to avoid inheriting it from DataObjecCollection."""
        raise TypeError

    def __or__(self, other) -> 'DataObjectChoice':
        """OR-ing a Choice to another choice extends the choice, as does OR-ing a DataObject."""
        if isinstance(other, DataObjectChoice):
            # adding one collection to another
            members = self.members + other.members
            return DataObjectChoice(self.name, self.desc, members)
        elif isinstance(other, DataObject):
            # adding a member to a collection
            return DataObjectChoice(self.name, self.desc, self.members + [other])
        else:
            raise TypeError

    # 'codec' interface
    def decode(self, binary: bytes) -> Tuple[dict, bytes]:
        """Decode a single DOs from the choice based on the tag.
        Args:
            binary : binary bytes of encoded data
        Returns:
            tuple of (decoded_result, binary_remainder)
        """
        tag = binary[0]
        if tag == 0xff:
            return (None, binary)
        if not tag in self.members_by_tag:
            raise ValueError('%s: Unknown Tag 0x%02x in %s; expected %s' %
                             (self, tag, binary, self.members_by_tag.keys()))
        obj = self.members_by_tag[tag]
        remainder = obj.from_tlv(binary)
        return (obj.to_dict(), remainder)

    # 'codec' interface
    def encode(self, decoded) -> bytes:
        obj = self.members_by_name[list(decoded)[0]]
        obj.decoded = list(decoded.values())[0]
        return obj.to_tlv()


class DataObjectSequence:
    """A sequence of DataObjects or DataObjectChoices. This allows us to express a certain
       ordered sequence of DOs or choices of DOs that have to appear as per the specification.
       By wrapping them into this formal DataObjectSequence, we can offer convenience methods
       for encoding or decoding an entire sequence."""

    def __init__(self, name: str, desc: Optional[str] = None, sequence=None):
        self.sequence = sequence or []
        self.name = name
        self.desc = desc

    def __str__(self) -> str:
        member_strs = [str(x) for x in self.sequence]
        return '%s(%s)' % (self.name, ','.join(member_strs))

    def __repr__(self) -> str:
        member_strs = [repr(x) for x in self.sequence]
        return '%s(%s)' % (self.__class__, ','.join(member_strs))

    def __add__(self, other) -> 'DataObjectSequence':
        """Add (append) a DataObject or DataObjectChoice to the sequence."""
        if isinstance(other, 'DataObject'):
            return DataObjectSequence(self.name, self.desc, self.sequence + [other])
        elif isinstance(other, 'DataObjectChoice'):
            return DataObjectSequence(self.name, self.desc, self.sequence + [other])
        elif isinstance(other, 'DataObjectSequence'):
            return DataObjectSequence(self.name, self.desc, self.sequence + other.sequence)

    # 'codec' interface
    def decode(self, binary: bytes) -> Tuple[list, bytes]:
        """Decode a sequence by calling the decoder of each element in the sequence.
        Args:
            binary : binary bytes of encoded data
        Returns:
            tuple of (decoded_result, binary_remainder)
        """
        remainder = binary
        res = []
        for e in self.sequence:
            (r, remainder) = e.decode(remainder)
            if r:
                res.append(r)
        return (res, remainder)

    # 'codec' interface
    def decode_multi(self, do: bytes) -> Tuple[list, bytes]:
        """Decode multiple occurrences of the sequence from the binary input data.
        Args:
            do : binary input data to be decoded
        Returns:
            list of results of the decoder of this sequences
        """
        remainder = do
        res = []
        while len(remainder):
            (r, remainder2) = self.decode(remainder)
            if r:
                res.append(r)
            if len(remainder2) < len(remainder):
                remainder = remainder2
            else:
                remainder = remainder2
                break
        return (res, remainder)

    # 'codec' interface
    def encode(self, decoded) -> bytes:
        """Encode a sequence by calling the encoder of each element in the sequence."""
        encoded = bytearray()
        i = 0
        for e in self.sequence:
            encoded += e.encode(decoded[i])
            i += 1
        return encoded

    def encode_multi(self, decoded) -> bytes:
        """Encode multiple occurrences of the sequence from the decoded input data.
        Args:
            decoded : list of json-serializable input data; one sequence per list item
        Returns:
            binary encoded output data
        """
        encoded = bytearray()
        for d in decoded:
            encoded += self.encode(d)
        return encoded


class CardCommand:
    """A single card command / instruction."""

    def __init__(self, name, ins, cla_list=None, desc=None):
        self.name = name
        self.ins = ins
        self.cla_list = cla_list or []
        self.cla_list = [x.lower() for x in self.cla_list]
        self.desc = desc

    def __str__(self):
        return self.name

    def __repr__(self):
        return '%s(INS=%02x,CLA=%s)' % (self.name, self.ins, self.cla_list)

    def match_cla(self, cla):
        """Does the given CLA match the CLA list of the command?."""
        if not isinstance(cla, str):
            cla = '%02u' % cla
        cla = cla.lower()
        for cla_match in self.cla_list:
            cla_masked = ""
            for i in range(0, 2):
                if cla_match[i] == 'x':
                    cla_masked += 'x'
                else:
                    cla_masked += cla[i]
            if cla_masked == cla_match:
                return True
        return False


class CardCommandSet:
    """A set of card instructions, typically specified within one spec."""

    def __init__(self, name, cmds=[]):
        self.name = name
        self.cmds = {c.ins: c for c in cmds}

    def __str__(self):
        return self.name

    def __getitem__(self, idx):
        return self.cmds[idx]

    def __add__(self, other):
        if isinstance(other, CardCommand):
            if other.ins in self.cmds:
                raise ValueError('%s: INS 0x%02x already defined: %s' %
                                 (self, other.ins, self.cmds[other.ins]))
            self.cmds[other.ins] = other
        elif isinstance(other, CardCommandSet):
            for c in other.cmds.keys():
                self.cmds[c] = other.cmds[c]
        else:
            raise ValueError(
                '%s: Unsupported type to add operator: %s' % (self, other))

    def lookup(self, ins, cla=None):
        """look-up the command within the CommandSet."""
        ins = int(ins)
        if not ins in self.cmds:
            return None
        cmd = self.cmds[ins]
        if cla and not cmd.match_cla(cla):
            return None
        return cmd


def all_subclasses(cls) -> set:
    """Recursively get all subclasses of a specified class"""
    return set(cls.__subclasses__()).union([s for c in cls.__subclasses__() for s in all_subclasses(c)])