Files
pysim/contrib/saip-tool.py
Harald Welte b6532b56d2 saip-tool: Add 'extract-apps' to dump all applications from eSIM profile
This new action can be used to dump all java applications as either raw
IJC file or converted to CAP format (the usual format generated by
JavaCard toolchains).

Change-Id: I51cffa5ba3ddbea491341d678ec9249d7cf470a5
2024-06-11 08:45:27 +02:00

223 lines
9.3 KiB
Python
Executable File

#!/usr/bin/env python3
# (C) 2024 by Harald Welte <laforge@osmocom.org>
#
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU Affero General Public License as published by
# the Free Software Foundation, either version 3 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 Affero General Public License for more details.
#
# You should have received a copy of the GNU Affero General Public License
# along with this program. If not, see <http://www.gnu.org/licenses/>.
import os
import sys
import argparse
import logging
import zipfile
from pathlib import Path
from typing import List
from pySim.esim.saip import *
from pySim.esim.saip.validation import CheckBasicStructure
from pySim import javacard
from pySim.utils import h2b, b2h, swap_nibbles
from pySim.pprint import HexBytesPrettyPrinter
pp = HexBytesPrettyPrinter(indent=4,width=500)
logging.basicConfig(level=logging.DEBUG)
parser = argparse.ArgumentParser(description="""
Utility program to work with eSIM SAIP (SimAlliance Interoperable Profile) files.""")
parser.add_argument('INPUT_UPP', help='Unprotected Profile Package Input file')
subparsers = parser.add_subparsers(dest='command', help="The command to perform", required=True)
parser_split = subparsers.add_parser('split', help='Split PE-Sequence into individual PEs')
parser_split.add_argument('--output-prefix', default='.', help='Prefix path/filename for output files')
parser_dump = subparsers.add_parser('dump', help='Dump information on PE-Sequence')
parser_dump.add_argument('mode', choices=['all_pe', 'all_pe_by_type', 'all_pe_by_naa'])
parser_dump.add_argument('--dump-decoded', action='store_true', help='Dump decoded PEs')
parser_check = subparsers.add_parser('check', help='Run constraint checkers on PE-Sequence')
parser_rpe = subparsers.add_parser('remove-pe', help='Remove specified PEs from PE-Sequence')
parser_rpe.add_argument('--output-file', required=True, help='Output file name')
parser_rpe.add_argument('--identification', type=int, action='append', help='Remove PEs matching specified identification')
parser_rn = subparsers.add_parser('remove-naa', help='Remove speciifed NAAs from PE-Sequence')
parser_rn.add_argument('--output-file', required=True, help='Output file name')
parser_rn.add_argument('--naa-type', required=True, choices=NAAs.keys(), help='Network Access Application type to remove')
# TODO: add an --naa-index or the like, so only one given instance can be removed
parser_info = subparsers.add_parser('info', help='Display information about the profile')
parser_eapp = subparsers.add_parser('extract-apps', help='Extract applications as loadblock file')
parser_eapp.add_argument('--output-dir', default='.', help='Output directory (where to store files)')
parser_eapp.add_argument('--format', default='cap', choices=['ijc', 'cap'], help='Data format of output files')
def do_split(pes: ProfileElementSequence, opts):
i = 0
for pe in pes.pe_list:
basename = Path(opts.INPUT_UPP).stem
if not pe.identification:
fname = '%s-%02u-%s.der' % (basename, i, pe.type)
else:
fname = '%s-%02u-%05u-%s.der' % (basename, i, pe.identification, pe.type)
print("writing single PE to file '%s'" % fname)
with open(os.path.join(opts.output_prefix, fname), 'wb') as outf:
outf.write(pe.to_der())
i += 1
def do_dump(pes: ProfileElementSequence, opts):
def print_all_pe(pes: ProfileElementSequence, dump_decoded:bool = False):
# iterate over each pe in the pes (using its __iter__ method)
for pe in pes:
print("="*70 + " " + pe.type)
if dump_decoded:
pp.pprint(pe.decoded)
def print_all_pe_by_type(pes: ProfileElementSequence, dump_decoded:bool = False):
# sort by PE type and show all PE within that type
for pe_type in pes.pe_by_type.keys():
print("="*70 + " " + pe_type)
for pe in pes.pe_by_type[pe_type]:
pp.pprint(pe)
if dump_decoded:
pp.pprint(pe.decoded)
def print_all_pe_by_naa(pes: ProfileElementSequence, dump_decoded:bool = False):
for naa in pes.pes_by_naa:
i = 0
for naa_instance in pes.pes_by_naa[naa]:
print("="*70 + " " + naa + str(i))
i += 1
for pe in naa_instance:
pp.pprint(pe.type)
if dump_decoded:
for d in pe.decoded:
print(" %s" % d)
if opts.mode == 'all_pe':
print_all_pe(pes, opts.dump_decoded)
elif opts.mode == 'all_pe_by_type':
print_all_pe_by_type(pes, opts.dump_decoded)
elif opts.mode == 'all_pe_by_naa':
print_all_pe_by_naa(pes, opts.dump_decoded)
def do_check(pes: ProfileElementSequence, opts):
print("Checking PE-Sequence structure...")
checker = CheckBasicStructure()
checker.check(pes)
print("All good!")
def do_remove_pe(pes: ProfileElementSequence, opts):
new_pe_list = []
for pe in pes.pe_list:
identification = pe.identification
if identification:
if identification in opts.identification:
print("Removing PE %s (id=%u) from Sequence..." % (pe, identification))
continue
new_pe_list.append(pe)
pes.pe_list = new_pe_list
pes._process_pelist()
print("Writing %u PEs to file '%s'..." % (len(pes.pe_list), opts.output_file))
with open(opts.output_file, 'wb') as f:
f.write(pes.to_der())
def do_remove_naa(pes: ProfileElementSequence, opts):
if not opts.naa_type in NAAs:
raise ValueError('unsupported NAA type %s' % opts.naa_type)
naa = NAAs[opts.naa_type]
print("Removing NAAs of type '%s' from Sequence..." % opts.naa_type)
pes.remove_naas_of_type(naa)
print("Writing %u PEs to file '%s'..." % (len(pes.pe_list), opts.output_file))
with open(opts.output_file, 'wb') as f:
f.write(pes.to_der())
def do_info(pes: ProfileElementSequence, opts):
def get_naa_count(pes: ProfileElementSequence) -> dict:
"""return a dict with naa-type (usim, isim) as key and the count of NAA instances as value."""
ret = {}
for naa_type in pes.pes_by_naa:
ret[naa_type] = len(pes.pes_by_naa[naa_type])
return ret
pe_hdr_dec = pes.pe_by_type['header'][0].decoded
print()
print("SAIP Profile Version: %u.%u" % (pe_hdr_dec['major-version'], pe_hdr_dec['minor-version']))
print("Profile Type: '%s'" % pe_hdr_dec['profileType'])
print("ICCID: %s" % b2h(pe_hdr_dec['iccid']))
print("Mandatory Services: %s" % ', '.join(pe_hdr_dec['eUICC-Mandatory-services'].keys()))
print()
naa_strs = ["%s[%u]" % (k, v) for k, v in get_naa_count(pes).items()]
print("NAAs: %s" % ', '.join(naa_strs))
for naa_type in pes.pes_by_naa:
for naa_inst in pes.pes_by_naa[naa_type]:
first_pe = naa_inst[0]
adf_name = ''
if hasattr(first_pe, 'adf_name'):
adf_name = '(' + first_pe.adf_name + ')'
print("NAA %s %s" % (first_pe.type, adf_name))
if hasattr(first_pe, 'imsi'):
print("\tIMSI: %s" % first_pe.imsi)
# applications
print()
apps = pes.pe_by_type.get('application', [])
print("Number of applications: %u" % len(apps))
for app_pe in apps:
print("App Load Package AID: %s" % b2h(app_pe.decoded['loadBlock']['loadPackageAID']))
print("\tMandated: %s" % ('mandated' in app_pe.decoded['app-Header']))
print("\tLoad Block Size: %s" % len(app_pe.decoded['loadBlock']['loadBlockObject']))
for inst in app_pe.decoded.get('instanceList', []):
print("\tInstance AID: %s" % b2h(inst['instanceAID']))
pass
def do_extract_apps(pes:ProfileElementSequence, opts):
apps = pes.pe_by_type.get('application', [])
for app_pe in apps:
package_aid = b2h(app_pe.decoded['loadBlock']['loadPackageAID'])
fname = os.path.join(opts.output_dir, '%s-%s.%s' % (pes.iccid, package_aid, opts.format))
load_block_obj = app_pe.decoded['loadBlock']['loadBlockObject']
print("Writing Load Package AID: %s to file %s" % (package_aid, fname))
if opts.format == 'ijc':
with open(fname, 'wb') as f:
f.write(load_block_obj)
else:
with io.BytesIO(load_block_obj) as f, zipfile.ZipFile(fname, 'w') as z:
javacard.ijc_to_cap(f, z, package_aid)
if __name__ == '__main__':
opts = parser.parse_args()
with open(opts.INPUT_UPP, 'rb') as f:
pes = ProfileElementSequence.from_der(f.read())
print("Read %u PEs from file '%s'" % (len(pes.pe_list), opts.INPUT_UPP))
if opts.command == 'split':
do_split(pes, opts)
elif opts.command == 'dump':
do_dump(pes, opts)
elif opts.command == 'check':
do_check(pes, opts)
elif opts.command == 'remove-pe':
do_remove_pe(pes, opts)
elif opts.command == 'remove-naa':
do_remove_naa(pes, opts)
elif opts.command == 'info':
do_info(pes, opts)
elif opts.command == 'extract-apps':
do_extract_apps(pes, opts)