Oleksiy Mazhelis

52 papers B 4C 7Misc 2Journal 13Unranked 25
YearRankTypeTitle / Venue / Authors
2018 conf
ICSOB
Eetu Luoma, Gabriella Laatikainen, Oleksiy Mazhelis
2017 C conf
WEBIST
Aleksei Romanov, Alexander Semenov, Oleksiy Mazhelis, Jari Veijalainen
2016 J jnl
Comput. Commun.
Julien Mineraud, Oleksiy Mazhelis, Xiang Su, Sasu Tarkoma
2016 J jnl
Int. J. Innov. Digit. Econ.
Tapio Levä, Mahya Ilaghi, Vilen Looga, Miika Komu, Nicklas Beijar, Oleksiy Mazhelis
2016 J jnl
J. Syst. Softw.
Gabriella Laatikainen, Oleksiy Mazhelis, Pasi Tyrväinen
2016 conf
ISC2
Oleksiy Mazhelis, Antti Hämäläinen, Tomi Asp, Pasi Tyrväinen
2015 J jnl
CoRR
Julien Mineraud, Oleksiy Mazhelis, Xiang Su, Sasu Tarkoma
2015 J jnl
CoRR
Julien Mineraud, Oleksiy Mazhelis, Xiang Su, Sasu Tarkoma
2015 C conf
ICIS
Gabriella Laatikainen, Oleksiy Mazhelis, Pasi Tyrväinen
2015 conf
ICSOB
Teemu Toivanen, Oleksiy Mazhelis, Eetu Luoma
2014 conf
WF-IoT
Oleksiy Mazhelis, Pasi Tyrväinen
2014 J jnl
Decis. Support Syst.
Tapio Levä, Oleksiy Mazhelis, Henna Suomi
2014 J jnl
Decis. Support Syst.
Gabriella Laatikainen, Oleksiy Mazhelis, Pasi Tyrväinen
2013 conf
ISNN (1)
Fengyu Cong, Hannu Hautakangas, Jukka Nieminen, Oleksiy Mazhelis, Mikko Perttunen, Jukka Riekki, Tapani Ristaniemi
2013 conf
ICSOB
Gabriella Laatikainen, Arto Ojala, Oleksiy Mazhelis
2013 conf
AIMS
Oleksiy Mazhelis, Martin Waldburger, Guilherme Sperb Machado, Burkhard Stiller, Pasi Tyrväinen
2013 J jnl
Inf. Softw. Technol.
Oleksiy Mazhelis, Pasi Tyrväinen, Lauri Frank
2012 conf
ICSOB
Oleksiy Mazhelis, Eetu Luoma, Arto Ojala
2012 conf
ICSOB
Oleksiy Mazhelis
2012 conf
NEW2AN
Oleksiy Mazhelis, Eetu Luoma, Henna Warma
2012 J jnl
Inf. Syst. Frontiers
Oleksiy Mazhelis, Pasi Tyrväinen
2012 C conf
CLOSER
Yrjö Raivio, Oleksiy Mazhelis, Koushik Annapureddy, Ramasivakarthik Mallavarapu, Pasi Tyrväinen
2012 conf
IEEE CLOUD
Oleksiy Mazhelis, Gabriella Fazekas, Pasi Tyrväinen
2012 ch.
Telecommunication Economics
Lauri Frank, Eetu Luoma, Oleksiy Mazhelis, Mirja Pulkkinen, Pasi Tyrväinen
2011 B conf
Discovery Science
Oleksiy Mazhelis, Indre Zliobaite, Mykola Pechenizkiy
2011 C conf
CLOSER
Oleksiy Mazhelis, Pasi Tyrväinen, Tan Kuan Eeik, Jari Hiltunen
2011 C conf
UIC
Mikko Perttunen, Oleksiy Mazhelis, Fengyu Cong, Mikko Kauppila, Teemu Leppänen, Jouni Kantola, Jussi Collin, Susanna Pirttikangas, Janne Haverinen, Tapani Ristaniemi, Jukka Riekki
2011 B conf
Intelligent Vehicles Symposium
Oleksiy Mazhelis
2011 conf
EUROMICRO-SEAA
Oleksiy Mazhelis, Pasi Tyrväinen
2010 conf
INC
Oleksiy Mazhelis, Hannakaisa Isomäki
2010 conf
ICSOB
Oskari Miettinen, Oleksiy Mazhelis, Eetu Luoma
2010 conf
ICSOB
Eetu Luoma, Oleksiy Mazhelis, Pertti Paakkolanvaara
2010 conf
ITSC
Oleksiy Mazhelis
2008 conf
ICSOFT (SE/MUSE/GSDCA)
Oleksiy Mazhelis, Pasi Tyrväinen, Jarmo Matilainen
2008 conf
INC
Oleksiy Mazhelis, Hannakaisa Isomäki
2007 J jnl
Comput. Secur.
Oleksiy Mazhelis, Seppo Puuronen
2007 B conf
ARES
Oleksiy Mazhelis, Seppo Puuronen
2007 J jnl
ARIMA J.
Oleksiy Mazhelis
2007 conf
WEBIST (3)
Mirja Pulkkinen, Oleksiy Mazhelis, Pentti Marttiin, Jouni Meriluoto
2006 conf
HICSS
Oleksiy Mazhelis, Jari A. Lehto, Jouni Markkula, Mirja Pulkkinen
2006 C conf
ICICS
Oleksiy Mazhelis, Seppo Puuronen, Mika Raento
2006 Misc conf
SEC
Oleksiy Mazhelis, Seppo Puuronen, Mika Raento
2006 J jnl
South Afr. Comput. J.
Oleksiy Mazhelis
2006 conf
WOSIS
Oleksiy Mazhelis, Anton Naumenko
2005 J jnl
Inf. Manag. Comput. Security
Oleksiy Mazhelis, Jouni Markkula, Jari Veijalainen
2005 B conf
PROFES
Oleksiy Mazhelis, Jouni Markkula, Markus Jakobsson
2004 conf
IICIS
Oleksiy Mazhelis, Seppo Puuronen
2004 conf
ICEIS (4)
Oleksiy Mazhelis, Seppo Puuronen
2004 C conf
ICICS
Oleksiy Mazhelis, Seppo Puuronen, Jari Veijalainen
2002 conf
Certification and Security in E-Services
Oleksiy Mazhelis, Alexandr Seleznyov, Seppo Puuronen
2002 Misc conf
SAC
Alexandr Seleznyov, Oleksiy Mazhelis
2001 conf
MMM-ACNS
Alexandr Seleznyov, Oleksiy Mazhelis, Seppo Puuronen
tests/integration/test_hashes.py
← Index tests/integration/test_hashes.py python
"""
Integration tests for the HashExtractor class.
"""
import pytest
import hashlib
from unittest.mock import Mock, patch, MagicMock

pytestmark = [pytest.mark.integration]


class TestHashExtractorInitialization:
    """Tests for HashExtractor initialization."""

    def test_initialization_elf(self, elf_binary_path, mock_logger):
        """Test initialization with ELF file."""
        from redb.extractors.hashes import HashExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(elf_binary_path, mock_logger)

            assert extractor.filepath == elf_binary_path
            assert extractor.hashes is None
            assert extractor.filetype == "elf"
            assert extractor.macho is None

    def test_initialization_with_macho_object(self, elf_binary_path, mock_logger):
        """Test initialization with pre-parsed macho object."""
        from redb.extractors.hashes import HashExtractor

        mock_macho = MagicMock()
        mock_macho.get_architectures.return_value = ["x86_64"]

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(elf_binary_path, mock_logger, macho=mock_macho)

            assert extractor.macho is mock_macho

    def test_initialization_pe(self, pe_binary_path, mock_logger):
        """Test initialization with PE file."""
        from redb.extractors.hashes import HashExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(pe_binary_path, mock_logger)

            assert extractor.filepath == pe_binary_path
            assert extractor.filetype == "pebin"
            # PE object should be created
            assert extractor.pe is not None

    def test_elastic_index_suffix(self, elf_binary_path, mock_logger):
        """Test that elastic_index has correct suffix."""
        from redb.extractors.hashes import HashExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(elf_binary_path, mock_logger)

            assert "-hashes" in extractor.elastic_index


class TestHashExtractorTag:
    """Tests for tag method."""

    def test_tag_value(self, elf_binary_path, mock_logger):
        """Test that tag returns correct value."""
        from redb.extractors.hashes import HashExtractor
        from redb.extractors.enum import Tag

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(elf_binary_path, mock_logger)

            assert extractor.tag() == Tag.HASH.value
            assert extractor.tag() == "hash"


class TestHashExtractorBasicHashes:
    """Tests for basic hash extraction."""

    def test_extract_md5(self, elf_binary_path, mock_logger, known_hashes):
        """Test MD5 hash extraction."""
        from redb.extractors.hashes import HashExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(elf_binary_path, mock_logger)

            result = extractor.extract()

            assert result is not None
            assert result.md5 == known_hashes['md5']

    def test_extract_sha1(self, elf_binary_path, mock_logger, known_hashes):
        """Test SHA1 hash extraction."""
        from redb.extractors.hashes import HashExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(elf_binary_path, mock_logger)

            result = extractor.extract()

            assert result is not None
            assert result.sha1 == known_hashes['sha1']

    def test_extract_sha256(self, elf_binary_path, mock_logger, known_hashes):
        """Test SHA256 hash extraction."""
        from redb.extractors.hashes import HashExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(elf_binary_path, mock_logger)

            result = extractor.extract()

            assert result is not None
            assert result.sha256 == known_hashes['sha256']

    def test_extract_ssdeep(self, elf_binary_path, mock_logger):
        """Test ssdeep hash extraction."""
        from redb.extractors.hashes import HashExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(elf_binary_path, mock_logger)

            result = extractor.extract()

            assert result is not None
            # ssdeep should be a string
            assert isinstance(result.ssdeep_hash, str)

    def test_extract_tlsh(self, elf_binary_path, mock_logger):
        """Test TLSH hash extraction."""
        from redb.extractors.hashes import HashExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(elf_binary_path, mock_logger)

            result = extractor.extract()

            assert result is not None
            # TLSH should be a string
            assert isinstance(result.tlsh_hash, str)


class TestHashExtractorPEHashes:
    """Tests for PE-specific hash extraction."""

    def test_extract_imphash(self, pe_binary_path, mock_logger):
        """Test imphash extraction for PE file."""
        from redb.extractors.hashes import HashExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(pe_binary_path, mock_logger)

            result = extractor.extract()

            assert result is not None
            # Imphash should be set for PE files
            if extractor.pe is not None:
                assert result.imphash is not None or result.imphash is None  # May or may not have imports

    def test_extract_authentihash(self, pe_binary_path, mock_logger):
        """Test authentihash extraction for PE file."""
        from redb.extractors.hashes import HashExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(pe_binary_path, mock_logger)

            result = extractor.extract()

            assert result is not None
            # Authentihash should be set for PE files
            if extractor.pe is not None:
                assert result.authentihash is not None

    def test_pe_hashes_not_set_for_elf(self, elf_binary_path, mock_logger):
        """Test that PE-specific hashes are not set for ELF files."""
        from redb.extractors.hashes import HashExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(elf_binary_path, mock_logger)

            result = extractor.extract()

            assert result is not None
            # PE-specific hashes should be None for ELF
            assert result.imphash is None
            assert result.authentihash is None
            assert result.richhash is None


class TestHashExtractorRichHeaderHashes:
    """Tests for Rich header hash extraction."""

    def test_compute_richhash(self, pe_binary_path, mock_logger):
        """Test Rich header hash computation."""
        from redb.extractors.hashes import HashExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(pe_binary_path, mock_logger)

            result = extractor.extract()

            assert result is not None
            # Rich hash may or may not be present depending on the PE
            if result.richhash is not None:
                assert isinstance(result.richhash, str)
                assert len(result.richhash) == 32  # MD5 hex length

    def test_compute_richpe_hash(self, pe_binary_path, mock_logger):
        """Test RichPE hash computation."""
        from redb.extractors.hashes import HashExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(pe_binary_path, mock_logger)

            result = extractor.extract()

            assert result is not None
            # RichPE hash may or may not be present
            if result.richpe_hash is not None:
                assert isinstance(result.richpe_hash, str)
                assert len(result.richpe_hash) == 32  # MD5 hex length

    def test_compute_richpv_hash(self, pe_binary_path, mock_logger):
        """Test RichPV hash computation."""
        from redb.extractors.hashes import HashExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(pe_binary_path, mock_logger)

            result = extractor.extract()

            assert result is not None
            # RichPV hashes may or may not be present
            if result.richpv_hash is not None:
                assert isinstance(result.richpv_hash, str)
            if result.richpv_hash_sorted is not None:
                assert isinstance(result.richpv_hash_sorted, str)


class TestHashExtractorELFHashes:
    """Tests for ELF-specific hash extraction."""

    def test_extract_elf_import_hash(self, elf_binary_path, mock_logger):
        """Test ELF import hash extraction."""
        from redb.extractors.hashes import HashExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(elf_binary_path, mock_logger)

            result = extractor.extract()

            assert result is not None
            # Import hash may or may not be present
            if result.import_hash is not None:
                assert isinstance(result.import_hash, str)
                assert len(result.import_hash) == 32

    def test_extract_elf_export_hash(self, elf_binary_path, mock_logger):
        """Test ELF export hash extraction."""
        from redb.extractors.hashes import HashExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(elf_binary_path, mock_logger)

            result = extractor.extract()

            assert result is not None
            # Export hash may or may not be present
            if result.export_hash is not None:
                assert isinstance(result.export_hash, str)

    def test_extract_elf_section_hash(self, elf_binary_path, mock_logger):
        """Test ELF section hash extraction."""
        from redb.extractors.hashes import HashExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(elf_binary_path, mock_logger)

            result = extractor.extract()

            assert result is not None
            # Section hash should be present for valid ELF
            if result.section_hash is not None:
                assert isinstance(result.section_hash, str)

    def test_extract_elf_symbol_hash(self, elf_binary_path, mock_logger):
        """Test ELF symbol hash extraction."""
        from redb.extractors.hashes import HashExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(elf_binary_path, mock_logger)

            result = extractor.extract()

            assert result is not None
            # Symbol hash (symhash) may or may not be present
            if result.symbol_hash is not None:
                assert isinstance(result.symbol_hash, str)

    def test_extract_elf_dynamic_hash(self, elf_binary_path, mock_logger):
        """Test ELF dynamic hash extraction."""
        from redb.extractors.hashes import HashExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(elf_binary_path, mock_logger)

            result = extractor.extract()

            assert result is not None
            # Dynamic hash may or may not be present
            if result.dynamic_hash is not None:
                assert isinstance(result.dynamic_hash, str)


class TestHashExtractorPrepareExportData:
    """Tests for prepare_export_data method."""

    def test_prepare_export_elasticsearch(self, elf_binary_path, mock_logger):
        """Test prepare_export_data for Elasticsearch."""
        from redb.extractors.hashes import HashExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(elf_binary_path, mock_logger)
            extractor.extract()

            result = extractor.prepare_export_data("ElasticsearchExporter")

            assert result is extractor.hashes

    def test_prepare_export_clickhouse(self, elf_binary_path, mock_logger):
        """Test prepare_export_data for ClickHouse."""
        from redb.extractors.hashes import HashExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(elf_binary_path, mock_logger)
            extractor.extract()

            result = extractor.prepare_export_data("ClickHouseExporter")

            # Should return tuple of (data, column_names, column_type_names)
            assert isinstance(result, tuple)
            assert len(result) == 3

            data, column_names, column_type_names = result

            assert isinstance(data, list)
            assert isinstance(column_names, list)
            assert isinstance(column_type_names, list)

            # Check expected columns
            assert 'sha256' in column_names
            assert 'md5' in column_names
            assert 'sha1' in column_names
            assert 'ssdeep_hash' in column_names
            assert 'tlsh_hash' in column_names

    def test_prepare_export_clickhouse_pe(self, pe_binary_path, mock_logger):
        """Test prepare_export_data for ClickHouse with PE file."""
        from redb.extractors.hashes import HashExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(pe_binary_path, mock_logger)
            extractor.extract()

            result = extractor.prepare_export_data("ClickHouseExporter")

            data, column_names, column_type_names = result

            # PE-specific columns should be present
            assert 'authentihash' in column_names
            assert 'imphash' in column_names
            assert 'richhash' in column_names

    def test_prepare_export_clickhouse_has_macho_columns(self, elf_binary_path, mock_logger):
        """Test prepare_export_data for ClickHouse includes Mach-O hash columns."""
        from redb.extractors.hashes import HashExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(elf_binary_path, mock_logger)
            extractor.extract()

            result = extractor.prepare_export_data("ClickHouseExporter")

            data, column_names, column_type_names = result

            # Mach-O hash columns should be present
            assert 'macho_dylib_hash' in column_names
            assert 'macho_import_hash' in column_names
            assert 'macho_export_hash' in column_names
            assert 'macho_entitlement_hash' in column_names
            assert 'macho_symhash' in column_names


class TestHashExtractorClickHouseTable:
    """Tests for get_clickhouse_table method."""

    def test_clickhouse_table_name(self, elf_binary_path, mock_logger):
        """Test correct ClickHouse table name."""
        from redb.extractors.hashes import HashExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(elf_binary_path, mock_logger)

            assert extractor.get_clickhouse_table() == "redb_hashes"


@pytest.mark.unit
@pytest.mark.dataclass
class TestHashExtractorHashesDataclass:
    """Tests for Hashes dataclass structure."""

    def test_hashes_dataclass_structure(self, elf_binary_path, mock_logger):
        """Test that Hashes dataclass has expected structure."""
        from redb.extractors.hashes import HashExtractor
        from redb.models.dataclasses import Hashes

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(elf_binary_path, mock_logger)

            result = extractor.extract()

            assert isinstance(result, Hashes)
            # Required fields
            assert hasattr(result, 'md5')
            assert hasattr(result, 'sha1')
            assert hasattr(result, 'sha256')
            assert hasattr(result, 'ssdeep_hash')
            assert hasattr(result, 'tlsh_hash')
            # Optional fields
            assert hasattr(result, 'authentihash')
            assert hasattr(result, 'imphash')
            assert hasattr(result, 'richhash')

    def test_hashes_dataclass_has_macho_fields(self):
        """Test that Hashes dataclass has Mach-O hash fields."""
        from redb.models.dataclasses import Hashes

        # Create instance with default values
        hashes = Hashes(
            md5="d41d8cd98f00b204e9800998ecf8427e",
            sha1="da39a3ee5e6b4b0d3255bfef95601890afd80709",
            sha256="e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
            ssdeep_hash="",
            tlsh_hash=""
        )

        # Verify Mach-O hash fields exist and default to None
        assert hasattr(hashes, 'macho_dylib_hash')
        assert hasattr(hashes, 'macho_import_hash')
        assert hasattr(hashes, 'macho_export_hash')
        assert hasattr(hashes, 'macho_entitlement_hash')
        assert hasattr(hashes, 'macho_symhash')

        assert hashes.macho_dylib_hash is None
        assert hashes.macho_import_hash is None
        assert hashes.macho_export_hash is None
        assert hashes.macho_entitlement_hash is None
        assert hashes.macho_symhash is None

    def test_hashes_dataclass_with_macho_values(self):
        """Test that Hashes dataclass can be created with Mach-O hash values."""
        from redb.models.dataclasses import Hashes

        hashes = Hashes(
            md5="d41d8cd98f00b204e9800998ecf8427e",
            sha1="da39a3ee5e6b4b0d3255bfef95601890afd80709",
            sha256="e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
            ssdeep_hash="",
            tlsh_hash="",
            macho_dylib_hash="abc123def456789012345678901234567",
            macho_import_hash="def456abc123789012345678901234567",
            macho_export_hash="789012abc123def456345678901234567",
            macho_entitlement_hash="345678abc123def456789012901234567",
            macho_symhash="901234abc123def456789012345678567"
        )

        assert hashes.macho_dylib_hash == "abc123def456789012345678901234567"
        assert hashes.macho_import_hash == "def456abc123789012345678901234567"
        assert hashes.macho_export_hash == "789012abc123def456345678901234567"
        assert hashes.macho_entitlement_hash == "345678abc123def456789012901234567"
        assert hashes.macho_symhash == "901234abc123def456789012345678567"


class TestHashExtractorErrorHandling:
    """Tests for error handling in HashExtractor."""

    def test_extract_handles_hash_errors(self, elf_binary_path, mock_logger):
        """Test that extract handles hash computation errors gracefully."""
        from redb.extractors.hashes import HashExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(elf_binary_path, mock_logger)

            # Even if some hashes fail, we should get a result
            result = extractor.extract()

            assert result is not None

    def test_extract_with_invalid_pe(self, elf_binary_path, mock_logger):
        """Test extraction when PE parsing fails."""
        from redb.extractors.hashes import HashExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            # Force filetype to be 'pebin' but use ELF file
            with patch.object(HashExtractor, '__init__', lambda self, *args, **kwargs: None):
                extractor = HashExtractor.__new__(HashExtractor)
                extractor.filepath = elf_binary_path
                extractor.log = mock_logger
                extractor.hashes = None
                extractor.pe = None
                extractor.filetype = 'pebin'

                with open(elf_binary_path, 'rb') as f:
                    extractor.binary = f.read()

                # Should not crash
                result = extractor._extract_hashes()
                assert result is not None


class TestHashExtractorMachoHashes:
    """Tests for Mach-O specific hash extraction."""

    def test_macho_hashes_not_set_for_elf(self, elf_binary_path, mock_logger):
        """Test that Mach-O hashes are not set for ELF files."""
        from redb.extractors.hashes import HashExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(elf_binary_path, mock_logger)

            result = extractor.extract()

            assert result is not None
            assert result.macho_dylib_hash is None
            assert result.macho_import_hash is None
            assert result.macho_export_hash is None
            assert result.macho_entitlement_hash is None
            assert result.macho_symhash is None

    def test_macho_hashes_not_set_for_pe(self, pe_binary_path, mock_logger):
        """Test that Mach-O hashes are not set for PE files."""
        from redb.extractors.hashes import HashExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(pe_binary_path, mock_logger)

            result = extractor.extract()

            assert result is not None
            assert result.macho_dylib_hash is None
            assert result.macho_import_hash is None
            assert result.macho_export_hash is None
            assert result.macho_entitlement_hash is None
            assert result.macho_symhash is None

    def test_extract_macho_hashes_with_mock(self, elf_binary_path, mock_logger):
        """Test Mach-O hash extraction with mocked macho object."""
        from redb.extractors.hashes import HashExtractor

        # Create mock macho object
        mock_macho = MagicMock()
        mock_macho.get_architectures.return_value = ["x86_64"]
        mock_macho.get_similarity_hashes.return_value = {
            'dylib_hash': 'abc123def456789012345678901234ab',
            'import_hash': 'def456abc789012345678901234567cd',
            'export_hash': '789012abc456def345678901234567ef',
            'entitlement_hash': '345678abc123def789012901234567gh',
            'symhash': '901234abc123def456789012345678ij'
        }

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(elf_binary_path, mock_logger, macho=mock_macho)
            # Override filetype to trigger macho hash extraction
            extractor.filetype = 'macho'

            result = extractor.extract()

            assert result is not None
            assert result.macho_dylib_hash == 'abc123def456789012345678901234ab'
            assert result.macho_import_hash == 'def456abc789012345678901234567cd'
            assert result.macho_export_hash == '789012abc456def345678901234567ef'
            assert result.macho_entitlement_hash == '345678abc123def789012901234567gh'
            assert result.macho_symhash == '901234abc123def456789012345678ij'

    def test_extract_macho_hashes_fat_binary_mock(self, elf_binary_path, mock_logger):
        """Test Mach-O hash extraction for FAT binary with mocked macho object."""
        from redb.extractors.hashes import HashExtractor

        # Create mock macho object for FAT binary
        mock_macho = MagicMock()
        mock_macho.get_architectures.return_value = ["x86_64", "arm64"]  # Multiple archs = FAT
        mock_macho.get_similarity_hashes.return_value = {
            'fat': {
                'dylib_hash': 'fat_dylib_hash_12345678901234567',
                'import_hash': 'fat_import_hash_12345678901234567',
            },
            'combined': {
                'dylib_hash': 'combined_dylib_hash_1234567890123',
            }
        }

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(elf_binary_path, mock_logger, macho=mock_macho)
            extractor.filetype = 'macho'

            result = extractor.extract()

            assert result is not None
            # Should use 'fat' key for FAT binaries
            assert result.macho_dylib_hash == 'fat_dylib_hash_12345678901234567'
            assert result.macho_import_hash == 'fat_import_hash_12345678901234567'

    def test_extract_macho_hashes_handles_empty(self, elf_binary_path, mock_logger):
        """Test Mach-O hash extraction handles empty hash dict."""
        from redb.extractors.hashes import HashExtractor

        mock_macho = MagicMock()
        mock_macho.get_architectures.return_value = ["x86_64"]
        mock_macho.get_similarity_hashes.return_value = {}

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(elf_binary_path, mock_logger, macho=mock_macho)
            extractor.filetype = 'macho'

            result = extractor.extract()

            # Should not crash, hashes should be None
            assert result is not None
            assert result.macho_dylib_hash is None

    def test_extract_macho_hashes_handles_none(self, elf_binary_path, mock_logger):
        """Test Mach-O hash extraction handles None return."""
        from redb.extractors.hashes import HashExtractor

        mock_macho = MagicMock()
        mock_macho.get_architectures.return_value = ["x86_64"]
        mock_macho.get_similarity_hashes.return_value = None

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(elf_binary_path, mock_logger, macho=mock_macho)
            extractor.filetype = 'macho'

            result = extractor.extract()

            # Should not crash
            assert result is not None


class TestHashExtractorConsistency:
    """Tests for hash consistency."""

    def test_hash_consistency_across_extractions(self, elf_binary_path, mock_logger):
        """Test that hashes are consistent across multiple extractions."""
        from redb.extractors.hashes import HashExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor1 = HashExtractor(elf_binary_path, mock_logger)
            result1 = extractor1.extract()

            extractor2 = HashExtractor(elf_binary_path, mock_logger)
            result2 = extractor2.extract()

            assert result1.md5 == result2.md5
            assert result1.sha1 == result2.sha1
            assert result1.sha256 == result2.sha256

    def test_hash_values_are_lowercase(self, elf_binary_path, mock_logger):
        """Test that hash values are lowercase hex strings."""
        from redb.extractors.hashes import HashExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(elf_binary_path, mock_logger)

            result = extractor.extract()

            assert result.md5 == result.md5.lower()
            assert result.sha1 == result.sha1.lower()
            assert result.sha256 == result.sha256.lower()

    def test_hash_lengths_correct(self, elf_binary_path, mock_logger):
        """Test that hash lengths are correct."""
        from redb.extractors.hashes import HashExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = HashExtractor(elf_binary_path, mock_logger)

            result = extractor.extract()

            assert len(result.md5) == 32  # MD5 = 128 bits = 32 hex chars
            assert len(result.sha1) == 40  # SHA1 = 160 bits = 40 hex chars
            assert len(result.sha256) == 64  # SHA256 = 256 bits = 64 hex chars