Weichao Zhuang

"""
Integration tests for the BasicPropertiesExtractor class.
"""
import pytest
import os
from unittest.mock import Mock, patch, MagicMock

pytestmark = [pytest.mark.integration]


class TestBasicPropertiesExtractorInitialization:
    """Tests for BasicPropertiesExtractor initialization."""

    def test_initialization(self, elf_binary_path, mock_logger):
        """Test basic initialization."""
        from redb.extractors.basicproperties import BasicPropertiesExtractor

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

            assert extractor.filepath == elf_binary_path
            assert extractor.basic_properties is None
            assert extractor.is_packed is None
            assert extractor.is_fat is None
            assert extractor.parent_sha256 is None
            assert extractor.child_sha256 is None
            assert extractor.child_architecture is None
            assert extractor.child_filetype is None

    def test_initialization_with_fat_params(self, elf_binary_path, mock_logger):
        """Test initialization with FAT Mach-O parameters."""
        from redb.extractors.basicproperties import BasicPropertiesExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            child_hashes = ["abc123", "def456"]
            child_archs = ["x86_64", "arm64"]
            child_types = ["macho", "macho"]

            extractor = BasicPropertiesExtractor(
                elf_binary_path,
                mock_logger,
                is_fat=True,
                child_sha256=child_hashes,
                child_architecture=child_archs,
                child_filetype=child_types
            )

            assert extractor.is_fat is True
            assert extractor.child_sha256 == child_hashes
            assert extractor.child_architecture == child_archs
            assert extractor.child_filetype == child_types

    def test_initialization_with_parent_sha256(self, elf_binary_path, mock_logger):
        """Test initialization with parent_sha256 for slice binaries."""
        from redb.extractors.basicproperties import BasicPropertiesExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            parent_hash = "parent_sha256_hash_here"

            extractor = BasicPropertiesExtractor(
                elf_binary_path,
                mock_logger,
                parent_sha256=parent_hash
            )

            assert extractor.parent_sha256 == parent_hash

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

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

            assert "-basic_properties" in extractor.elastic_index


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

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

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

            assert extractor.tag() == Tag.BASIC_PROPERTIES.value
            assert extractor.tag() == "basic_properties"


class TestBasicPropertiesExtract:
    """Tests for extract method."""

    def test_extract_elf_binary(self, elf_binary_path, mock_logger):
        """Test extracting properties from ELF binary."""
        from redb.extractors.basicproperties import BasicPropertiesExtractor
        from redb.models.dataclasses import BasicProperties

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

            result = extractor.extract()

            assert result is not None
            assert isinstance(result, BasicProperties)
            assert result.sample_name == os.path.basename(elf_binary_path)
            assert result.filesize > 0
            assert result.file_entropy >= 0
            assert result.filetype is not None
            assert result.filetype_mime is not None
            assert result.filetype_magika is not None

    def test_extract_pe_binary(self, pe_binary_path, mock_logger):
        """Test extracting properties from PE binary."""
        from redb.extractors.basicproperties import BasicPropertiesExtractor
        from redb.models.dataclasses import BasicProperties

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

            result = extractor.extract()

            assert result is not None
            assert isinstance(result, BasicProperties)
            assert result.filesize > 0
            # PE file should be detected
            assert "PE" in result.filetype or "executable" in result.filetype.lower()

    def test_extract_text_file(self, temp_text_file, mock_logger):
        """Test extracting properties from text file."""
        from redb.extractors.basicproperties import BasicPropertiesExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = BasicPropertiesExtractor(temp_text_file, mock_logger)

            result = extractor.extract()

            assert result is not None
            assert "text" in result.filetype.lower() or "ASCII" in result.filetype
            assert result.file_entropy >= 0

    def test_extract_stores_in_basic_properties(self, elf_binary_path, mock_logger):
        """Test that extract stores result in basic_properties attribute."""
        from redb.extractors.basicproperties import BasicPropertiesExtractor

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

            assert extractor.basic_properties is None
            result = extractor.extract()
            assert extractor.basic_properties is result


class TestBasicPropertiesFiletype:
    """Tests for file type detection."""

    def test_elf_filetype(self, elf_binary_path, mock_logger):
        """Test ELF file type detection."""
        from redb.extractors.basicproperties import BasicPropertiesExtractor

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

            result = extractor.extract()

            assert result is not None
            # Should detect as ELF
            assert "ELF" in result.filetype

    def test_pe_filetype_mime(self, pe_binary_path, mock_logger):
        """Test PE file type MIME detection."""
        from redb.extractors.basicproperties import BasicPropertiesExtractor

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

            result = extractor.extract()

            assert result is not None
            # PE MIME types (varies by libmagic version)
            assert "dosexec" in result.filetype_mime or "x-msdownload" in result.filetype_mime or "portable-executable" in result.filetype_mime

    def test_text_filetype_mime(self, temp_text_file, mock_logger):
        """Test text file type MIME detection."""
        from redb.extractors.basicproperties import BasicPropertiesExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = BasicPropertiesExtractor(temp_text_file, mock_logger)

            result = extractor.extract()

            assert result is not None
            # Text MIME types
            assert "text" in result.filetype_mime


class TestBasicPropertiesEntropy:
    """Tests for entropy calculation in BasicPropertiesExtractor."""

    def test_entropy_reasonable_range(self, elf_binary_path, mock_logger):
        """Test that entropy is within reasonable range (0-8)."""
        from redb.extractors.basicproperties import BasicPropertiesExtractor

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

            result = extractor.extract()

            assert result is not None
            assert 0 <= result.file_entropy <= 8

    def test_entropy_rounded(self, elf_binary_path, mock_logger):
        """Test that entropy is rounded to 3 decimal places."""
        from redb.extractors.basicproperties import BasicPropertiesExtractor

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

            result = extractor.extract()

            assert result is not None
            # Check rounding to 3 decimal places
            entropy_str = str(result.file_entropy)
            if '.' in entropy_str:
                decimal_places = len(entropy_str.split('.')[1])
                assert decimal_places <= 3


class TestBasicPropertiesFilesize:
    """Tests for file size calculation."""

    def test_filesize_matches_actual(self, elf_binary_path, mock_logger):
        """Test that filesize matches actual file size."""
        from redb.extractors.basicproperties import BasicPropertiesExtractor

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

            result = extractor.extract()

            actual_size = os.path.getsize(elf_binary_path)
            assert result.filesize == actual_size

    def test_filesize_matches_binary_length(self, elf_binary_path, mock_logger, elf_binary_content):
        """Test that filesize matches binary content length."""
        from redb.extractors.basicproperties import BasicPropertiesExtractor

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

            result = extractor.extract()

            assert result.filesize == len(elf_binary_content)


class TestBasicPropertiesPrepareExportData:
    """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.basicproperties import BasicPropertiesExtractor

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

            result = extractor.prepare_export_data("ElasticsearchExporter")

            assert result is extractor.basic_properties

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

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = BasicPropertiesExtractor(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 len(data) == 1  # One row
            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 'filesize' in column_names
            assert 'file_entropy' in column_names
            assert 'filetype' in column_names
            assert 'filetype_mime' in column_names
            assert 'filetype_magika' in column_names
            # New FAT/child columns
            assert 'parent_sha256' in column_names
            assert 'child_sha256' in column_names
            assert 'child_architecture' in column_names
            assert 'child_filetype' in column_names
            assert 'is_fat' in column_names

    def test_prepare_export_clickhouse_data_types(self, elf_binary_path, mock_logger):
        """Test that ClickHouse export data has correct types."""
        from redb.extractors.basicproperties import BasicPropertiesExtractor

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

            data, column_names, column_type_names = extractor.prepare_export_data("ClickHouseExporter")

            row = data[0]

            # Check that sha256 is a string with correct length
            sha256_idx = column_names.index('sha256')
            assert isinstance(row[sha256_idx], str)
            assert len(row[sha256_idx]) == 64

            # Check that filesize is an integer
            filesize_idx = column_names.index('filesize')
            assert isinstance(row[filesize_idx], int)

            # Check that file_entropy is a float
            entropy_idx = column_names.index('file_entropy')
            assert isinstance(row[entropy_idx], float)


class TestBasicPropertiesClickHouseTable:
    """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.basicproperties import BasicPropertiesExtractor

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

            assert extractor.get_clickhouse_table() == "redb_basic_properties"


class TestBasicPropertiesExportData:
    """Tests for export_data method."""

    def test_export_data_with_print_exporter(self, elf_binary_path, mock_logger, mock_print_exporter, capsys):
        """Test export_data with PrintExporter."""
        from redb.extractors.basicproperties import BasicPropertiesExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = BasicPropertiesExtractor(
                elf_binary_path,
                mock_logger,
                exporters=[mock_print_exporter]
            )

            result = extractor.export_data()

            assert result is True

    def test_export_data_empty_exporters(self, elf_binary_path, mock_logger):
        """Test export_data with no exporters."""
        from redb.extractors.basicproperties import BasicPropertiesExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            extractor = BasicPropertiesExtractor(
                elf_binary_path,
                mock_logger,
                exporters=[]
            )

            result = extractor.export_data()

            # Should succeed with no exporters
            assert result is True


class TestBasicPropertiesMagika:
    """Tests for Magika file type detection."""

    def test_magika_elf_detection(self, elf_binary_path, mock_logger):
        """Test Magika detection of ELF files."""
        from redb.extractors.basicproperties import BasicPropertiesExtractor

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

            result = extractor.extract()

            assert result is not None
            # Magika should detect ELF
            assert result.filetype_magika == "elf" or "elf" in result.filetype_magika.lower()

    def test_magika_pe_detection(self, pe_binary_path, mock_logger):
        """Test Magika detection of PE files."""
        from redb.extractors.basicproperties import BasicPropertiesExtractor

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

            result = extractor.extract()

            assert result is not None
            # Magika should detect PE
            assert "pe" in result.filetype_magika.lower() or "exe" in result.filetype_magika.lower()


class TestBasicPropertiesFATSupport:
    """Tests for FAT Mach-O and child file support."""

    def test_extract_with_fat_flag(self, elf_binary_path, mock_logger):
        """Test extraction with is_fat flag set."""
        from redb.extractors.basicproperties import BasicPropertiesExtractor

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

            result = extractor.extract()

            assert result is not None
            assert result.is_fat is True

    def test_extract_with_child_hashes(self, elf_binary_path, mock_logger):
        """Test extraction with child_sha256 provided."""
        from redb.extractors.basicproperties import BasicPropertiesExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            child_hashes = ["abc123def456", "789ghi012jkl"]

            extractor = BasicPropertiesExtractor(
                elf_binary_path,
                mock_logger,
                child_sha256=child_hashes
            )

            result = extractor.extract()

            assert result is not None
            assert result.child_sha256 == child_hashes

    def test_extract_with_child_architecture(self, elf_binary_path, mock_logger):
        """Test extraction with child_architecture provided."""
        from redb.extractors.basicproperties import BasicPropertiesExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            child_archs = ["x86_64", "arm64"]

            extractor = BasicPropertiesExtractor(
                elf_binary_path,
                mock_logger,
                child_architecture=child_archs
            )

            result = extractor.extract()

            assert result is not None
            assert result.child_architecture == child_archs

    def test_extract_with_child_filetype(self, elf_binary_path, mock_logger):
        """Test extraction with child_filetype provided."""
        from redb.extractors.basicproperties import BasicPropertiesExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            child_types = ["macho", "macho"]

            extractor = BasicPropertiesExtractor(
                elf_binary_path,
                mock_logger,
                child_filetype=child_types
            )

            result = extractor.extract()

            assert result is not None
            assert result.child_filetype == child_types

    def test_clickhouse_export_with_fat_data(self, elf_binary_path, mock_logger):
        """Test ClickHouse export includes FAT-related fields correctly."""
        from redb.extractors.basicproperties import BasicPropertiesExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            child_hashes = ["abc123", "def456"]
            child_archs = ["x86_64", "arm64"]
            child_types = ["macho", "macho"]

            extractor = BasicPropertiesExtractor(
                elf_binary_path,
                mock_logger,
                is_fat=True,
                child_sha256=child_hashes,
                child_architecture=child_archs,
                child_filetype=child_types,
                parent_sha256=None  # FAT container has no parent
            )
            extractor.extract()

            data, column_names, column_type_names = extractor.prepare_export_data("ClickHouseExporter")
            row = data[0]

            # Check FAT fields
            is_fat_idx = column_names.index('is_fat')
            assert row[is_fat_idx] is True

            child_sha256_idx = column_names.index('child_sha256')
            assert row[child_sha256_idx] == child_hashes

            child_arch_idx = column_names.index('child_architecture')
            assert row[child_arch_idx] == child_archs

            child_type_idx = column_names.index('child_filetype')
            assert row[child_type_idx] == child_types

    def test_clickhouse_export_for_slice(self, elf_binary_path, mock_logger):
        """Test ClickHouse export for a slice binary with parent_sha256."""
        from redb.extractors.basicproperties import BasicPropertiesExtractor

        with patch('redb.settings.ELASTIC_BINARIES_COLLECTION', 'test'):
            parent_hash = "parent_container_sha256_hash"

            extractor = BasicPropertiesExtractor(
                elf_binary_path,
                mock_logger,
                parent_sha256=parent_hash
            )
            extractor.extract()

            data, column_names, column_type_names = extractor.prepare_export_data("ClickHouseExporter")
            row = data[0]

            parent_idx = column_names.index('parent_sha256')
            assert row[parent_idx] == parent_hash

    def test_default_fat_fields_are_none_or_empty(self, elf_binary_path, mock_logger):
        """Test that FAT-related fields default to None/empty for non-FAT binaries."""
        from redb.extractors.basicproperties import BasicPropertiesExtractor

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

            result = extractor.extract()

            assert result is not None
            assert result.is_fat is None
            assert result.child_sha256 is None
            assert result.child_architecture is None
            assert result.child_filetype is None


@pytest.mark.unit
@pytest.mark.dataclass
class TestBasicPropertiesDataclass:
    """Tests for BasicProperties dataclass structure."""

    def test_dataclass_has_all_expected_fields(self):
        """Test that BasicProperties dataclass has all expected fields."""
        from redb.models.dataclasses import BasicProperties

        # Create a sample instance
        bp = BasicProperties(
            filename="test.exe",
            sample_name="test.exe",
            filesize=1024,
            filetype="PE32",
            filetype_mime="application/x-dosexec",
            filetype_magika="pebin",
            file_entropy=6.5
        )

        # Check required fields
        assert bp.filename == "test.exe"
        assert bp.sample_name == "test.exe"
        assert bp.filesize == 1024
        assert bp.filetype == "PE32"
        assert bp.filetype_mime == "application/x-dosexec"
        assert bp.filetype_magika == "pebin"
        assert bp.file_entropy == 6.5

        # Check optional fields default to None
        assert bp.is_packed is None
        assert bp.is_fat is None
        assert bp.child_sha256 is None
        assert bp.child_architecture is None
        assert bp.child_filetype is None

    def test_dataclass_with_optional_fat_fields(self):
        """Test BasicProperties dataclass with FAT-related optional fields."""
        from redb.models.dataclasses import BasicProperties

        bp = BasicProperties(
            filename="universal.app",
            sample_name="universal.app",
            filesize=2048,
            filetype="Mach-O universal",
            filetype_mime="application/x-mach-binary",
            filetype_magika="macho",
            file_entropy=7.2,
            is_fat=True,
            child_sha256=["hash1", "hash2"],
            child_architecture=["x86_64", "arm64"],
            child_filetype=["macho", "macho"]
        )

        assert bp.is_fat is True
        assert bp.child_sha256 == ["hash1", "hash2"]
        assert bp.child_architecture == ["x86_64", "arm64"]
        assert bp.child_filetype == ["macho", "macho"]