grocery-manager/test/rest/test_hash_util.py

#
# Copyright (c) Daniel Sheffield 2023
# All rights reserved
#
# THIS SOFTWARE IS PROVIDED AS IS WITHOUT WARRANTY
from pytest import fixture, mark
import app.rest.hash_util as hash_util
import base32_lib as b32
from app.rest.hash_util import(
    hash_to_hex,
    hash_to_bytes,
    bytes_to_base32,
    bytes_to_base64,
    bytes_to_base85,
    bytes_to_hash,
    bytes_to_hex,
    normalize_bytes,
    normalize_hash,
    normalize_hex,
    base85_to_hash,
    base85_to_bytes,
    base85_to_hex,
    base32_to_bytes,
    base32_to_hash,
    base32_to_hex,
    base64_to_bytes,
    base64_to_hash,
    base64_to_hex,
    hash_to_base32,
    hash_to_base85,
    hash_to_base64,
    hex_to_base32,
    hex_to_base85,
    hex_to_base64,
    hex_to_bytes,
    hex_to_hash,
)

def to_unsigned(_hash: int) -> int:
    return _hash & hash_util.DIGEST_SIZE_BITMASK

def digest_size():
    for g in ( two_bytes, three_bytes, four_bytes):
        yield g

def two_bytes():
    hash_util.DIGEST_SIZE_BYTES = 2
    hash_util.DIGEST_SIZE_NIBBLES = 4
    hash_util.DIGEST_SIZE_BITMASK = 0xffff
    hash_util.DIGEST_SIZE_SIGNED_TO_UNSIGNED_BIT = 0x10000
    hash_util.DIGEST_SIZE_SIGNED_TO_UNSIGNED_BITMASK = 0x1ffff
    return hash_util.DIGEST_SIZE_BYTES

def three_bytes():
    hash_util.DIGEST_SIZE_BYTES = 3
    hash_util.DIGEST_SIZE_NIBBLES = 6
    hash_util.DIGEST_SIZE_BITMASK = 0xffffff
    hash_util.DIGEST_SIZE_SIGNED_TO_UNSIGNED_BIT = 0x1000000
    hash_util.DIGEST_SIZE_SIGNED_TO_UNSIGNED_BITMASK = 0x1ffffff
    return hash_util.DIGEST_SIZE_BYTES

def four_bytes():
    hash_util.DIGEST_SIZE_BYTES = 4
    hash_util.DIGEST_SIZE_NIBBLES = 8
    hash_util.DIGEST_SIZE_BITMASK = 0xffffffff
    hash_util.DIGEST_SIZE_SIGNED_TO_UNSIGNED_BIT = 0x100000000
    hash_util.DIGEST_SIZE_SIGNED_TO_UNSIGNED_BITMASK = 0x1ffffffff
    return hash_util.DIGEST_SIZE_BYTES

@fixture
def two_byte_bitmask():
    return 0xffff

def hash_codes():
    for _hash in [
        -7583489610679606711,
        1421958803217889556,
        10,
    ]:
        _bytes = b'\00' * 8 + _hash.to_bytes(8, byteorder='big', signed=True)[-8:]
        _hex = _bytes.hex()
        print(_hash, _hex, _bytes)
        yield _hash, _hex, _bytes

# PYTHONHASHSEED=0

def digest_size_bitmask(digest_size_bits):
    mask = 0x0
    for _ in range(digest_size_bits):
        mask = mask << 1 | 0x1
    return mask

def digest_size_force_unsigned_bitmask(digest_size):
    mask = 0x1
    for _ in range(8*digest_size):
        mask = mask << 1 | 0x1
    return mask

def test_three_byte_assertions():
    digest_size = three_bytes()

    assert hash_to_base64(-7583489610679606711) == 'k6JJ'
    assert bytes_to_base64(b'\x96\xc2\x08`\xcd\x93\xa2I') == 'k6JJ'
    assert hex_to_base64('96c20860cd93a249') == 'k6JJ'
    assert base64_to_hex('k6JJ') == '93a249'
    assert base64_to_bytes('aa') == b'\x00\x00i'
    assert base64_to_hex('aa') == '000069'
    assert base64_to_hash('aa') == 105


@mark.parametrize('_hash, _hex', [
    (_hash, _hex) for _hash, _hex, *_ in hash_codes()
])
@mark.parametrize('digest_size_gen', digest_size())
def test_hash_to_hex(_hash, _hex, digest_size_gen):
    digest_size = digest_size_gen()
    assert hash_to_hex(_hash) == _hex[-digest_size*2:]
    for i in range(1, digest_size*2+1):
        # test hex is zero padded
        assert hash_to_hex(_hash & digest_size_bitmask(i*4)) == _hex[-i:].zfill(digest_size*2)

@mark.parametrize('_hash, _bytes', [
    (_hash, _bytes) for _hash, _, _bytes, *_ in hash_codes()
])
@mark.parametrize('digest_size_gen', digest_size())
def test_hash_to_base32(_hash, _bytes, digest_size_gen):
    digest_size = digest_size_gen()
    norm = hash_to_base32(_hash)
    assert norm == b32.encode(
        to_unsigned(int.from_bytes(
            _bytes[-digest_size:], byteorder='big', signed=False)
        )
    ).upper().zfill(digest_size*8//5+1)
    assert len(norm) == digest_size*8//5 +1
    for i in range(1, digest_size+1):
        norm = hash_to_base32(_hash & digest_size_bitmask(i*8))
        # test base32 is zero padded
        assert norm == b32.encode(
            to_unsigned(int.from_bytes(
                (b'\x00'*digest_size + _bytes[-i:])[-digest_size:], byteorder='big', signed=False)
            )
        ).upper().zfill(digest_size*8//5+1)
        assert len(norm) == digest_size*8//5 +1

@mark.parametrize('_hash, _hex', [
    (_hash, _hex) for _hash, _hex, *_ in hash_codes()
])
@mark.parametrize('digest_size_gen', digest_size())
def test_hex_to_hash(_hash, _hex, digest_size_gen):
    digest_size = digest_size_gen()
    assert hex_to_hash(_hex) == _hash & digest_size_bitmask(digest_size*8)
    for i in range(1, digest_size*2+1):
        # test non-zero padded hex
        _input = _hex[-i:]
        assert digest_size*2 == i or len(_input) < digest_size*2
        assert hex_to_hash(_input) == _hash & digest_size_bitmask(i*4)

        # test zero padded hex
        _input = _hex[-i:].zfill(digest_size*2)
        assert len(_input) == digest_size*2
        assert hex_to_hash(_input) == _hash & digest_size_bitmask(i*4)

@mark.parametrize('_hex, _bytes', [
    (_hex, _bytes) for _, _hex, _bytes, *_ in hash_codes()
])
@mark.parametrize('digest_size_gen', digest_size())
def test_hex_to_bytes(_hex, _bytes, digest_size_gen):
    digest_size = digest_size_gen()
    assert hex_to_bytes(_hex) == _bytes[-digest_size:]
    for i in range(1, digest_size*2+1):
        # test non-zero padded hex
        _input = _hex[-i:]
        assert digest_size*2 == i or len(_input) < digest_size*2
        if i % 2:
            if i > 2:
                assert hex_to_bytes(_input) == (b'\x00'*digest_size + bytes([
                    _bytes[-i//2] & 0x0f, *_bytes[-i//2+1:]
                ]))[-digest_size:]
            else:
                assert hex_to_bytes(_input) == (b'\x00'*digest_size + bytes([
                    _bytes[-i//2] & 0x0f
                ]))[-digest_size:]
        else:
            assert hex_to_bytes(_input) == (b'\x00'*digest_size + _bytes[-i//2:])[-digest_size:]

        # test zero padded hex
        _input = _hex[-i:].zfill(digest_size*2)
        assert len(_input) == digest_size*2
        if i % 2:
            if i > 2:
                assert hex_to_bytes(_input) == (b'\x00'*digest_size + bytes([
                    _bytes[-i//2] & 0x0f, *_bytes[-i//2+1:]
                ]))[-digest_size:]
            else:
                assert hex_to_bytes(_input) == (b'\x00'*digest_size + bytes([
                    _bytes[-i//2] & 0x0f
                ]))[-digest_size:]
        else:
            assert hex_to_bytes(_input) == (b'\x00'*digest_size + _bytes[-i//2:])[-digest_size:]

@mark.parametrize('_hash, _bytes', [
    (_hash, _bytes) for _hash, _, _bytes, *_ in hash_codes()
])
@mark.parametrize('digest_size_gen', digest_size())
def test_bytes_to_hash(_hash, _bytes, digest_size_gen):
    digest_size = digest_size_gen()
    assert bytes_to_hash(_bytes) == _hash & digest_size_bitmask(digest_size*8)
    for i in range(1, digest_size+1):
        # test non-zero padded bytes
        _input = _bytes[-i:]
        assert digest_size == i or len(_input) < digest_size
        assert bytes_to_hash(_input) == _hash & digest_size_bitmask(i*8)

        # test zero padded bytes
        _input = (b'\x00'*digest_size + _bytes[-i:])[-digest_size:]
        assert len(_input) == digest_size
        assert bytes_to_hash(_input) == _hash & digest_size_bitmask(i*8)

@mark.parametrize('_hash, _bytes', [
    (_hash, _bytes) for _hash, _, _bytes, *_ in hash_codes()
])
@mark.parametrize('digest_size_gen', digest_size())
def test_base32_to_hash(_hash, _bytes, digest_size_gen):
    digest_size = digest_size_gen()
    _base32 = b32.encode(
        to_unsigned(int.from_bytes(
            _bytes[-digest_size:], byteorder='big', signed=False)
        )
    ).upper().zfill(digest_size*8//5+1)
    norm = base32_to_hash(_base32)
    assert norm == _hash & digest_size_bitmask(digest_size*8)

    for i in range(1, digest_size+1):
        # test non-zero padded base32
        _base32 = b32.encode(
            to_unsigned(int.from_bytes(
                (b'\x00'*digest_size + _bytes[-i:])[-digest_size:], byteorder='big', signed=False)
            )
        ).upper()
        norm = base32_to_hash(_base32)
        assert norm == _hash & digest_size_bitmask(i*8)

        # test zero padded base32
        _base32 = b32.encode(
            to_unsigned(int.from_bytes(
                (b'\x00'*digest_size + _bytes[-i:])[-digest_size:], byteorder='big', signed=False)
            )
        ).upper().zfill(digest_size*8//5+1)
        norm = base32_to_hash(_base32)
        assert norm == _hash & digest_size_bitmask(i*8)

@mark.parametrize('_hex', [
    _hex for _, _hex, *_ in hash_codes()
])
@mark.parametrize('digest_size_gen', digest_size())
def test_normalize_hex(_hex, digest_size_gen):
    digest_size = digest_size_gen()
    assert normalize_hex(_hex) == _hex[-digest_size*2:]
    assert len(normalize_hex(_hex)) == digest_size*2

    # test commutivity
    for i in range(1, digest_size*2+1):
        assert i == digest_size*2 or _hex[-i:] != _hex[-i:].zfill(digest_size*2)
        assert normalize_hex(_hex[-i:]) == _hex[-i:].zfill(digest_size*2)
        assert len(normalize_hex(_hex[-i:])) == digest_size*2

@mark.parametrize('_bytes', [
    _bytes for _, _, _bytes, *_ in hash_codes()
])
@mark.parametrize('digest_size_gen', digest_size())
def test_normalize_bytes(_bytes, digest_size_gen):
    digest_size = digest_size_gen()
    assert normalize_bytes(_bytes) == _bytes[-digest_size:]
    assert len(normalize_bytes(_bytes)) == digest_size
    
    # test commutivity
    for i in range(1, digest_size+1):
        assert i == digest_size or _bytes[-i:] != (b'\x00'*digest_size + _bytes[-i:])[-digest_size:]
        assert normalize_bytes(_bytes[-i:]) == (b'\x00'*digest_size + _bytes[-i:])[-digest_size:]
        assert len(normalize_bytes(_bytes[-i:])) == digest_size

@mark.parametrize('_hash', [
    _hash for _hash, *_ in hash_codes()
])
@mark.parametrize('digest_size_gen', digest_size())
def test_normalize_hash(_hash, digest_size_gen):
    digest_size = digest_size_gen()
    norm = normalize_hash(_hash)
    assert norm == _hash & digest_size_bitmask(digest_size*8)
    
    # test commutivity
    for i in range(1, digest_size+1):
        norm = normalize_hash(_hash & digest_size_bitmask(i*8))
        assert  norm == _hash & digest_size_bitmask(i*8)
        assert norm == norm & digest_size_bitmask(i*8)