Carlos Duarte

"""
Unit tests for redb.models.dataclasses.

These tests verify construction, field defaults, and type expectations for
every dataclass in the module (excluding Mach-O dataclasses, which are
covered by test_macho_dataclasses.py).
"""
import pytest
from dataclasses import asdict, fields

pytestmark = [pytest.mark.unit, pytest.mark.dataclass]


# ============================================================================
# Hash / Hashes
# ============================================================================

class TestHashDataclass:
    """Tests for the Hash dataclass."""

    def test_creation(self):
        from redb.models.dataclasses import Hash

        h = Hash(
            md5="d41d8cd98f00b204e9800998ecf8427e",
            sha1="da39a3ee5e6b4b0d3255bfef95601890afd80709",
            sha256="e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
        )
        assert h.md5 == "d41d8cd98f00b204e9800998ecf8427e"
        assert h.sha1 == "da39a3ee5e6b4b0d3255bfef95601890afd80709"
        assert h.sha256 == "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855"

    def test_asdict(self):
        from redb.models.dataclasses import Hash

        h = Hash(md5="a", sha1="b", sha256="c")
        d = asdict(h)
        assert d == {"md5": "a", "sha1": "b", "sha256": "c"}


class TestHashesDataclass:
    """Tests for the Hashes dataclass."""

    def test_required_fields(self):
        from redb.models.dataclasses import Hashes

        h = Hashes(
            md5="md5val",
            sha1="sha1val",
            sha256="sha256val",
            ssdeep_hash="ssdeep",
            tlsh_hash="tlsh",
        )
        assert h.md5 == "md5val"
        assert h.ssdeep_hash == "ssdeep"

    def test_optional_pe_fields_default_none(self):
        from redb.models.dataclasses import Hashes

        h = Hashes(md5="a", sha1="b", sha256="c", ssdeep_hash="d", tlsh_hash="e")
        assert h.authentihash is None
        assert h.imphash is None
        assert h.richhash is None
        assert h.richpe_hash is None
        assert h.richpv_hash is None
        assert h.richpv_hash_sorted is None
        assert h.typerefhash is None
        assert h.impfuzzy is None

    def test_optional_elf_fields_default_none(self):
        from redb.models.dataclasses import Hashes

        h = Hashes(md5="a", sha1="b", sha256="c", ssdeep_hash="d", tlsh_hash="e")
        assert h.import_hash is None
        assert h.export_hash is None
        assert h.section_hash is None
        assert h.symbol_hash is None
        assert h.dynamic_hash is None

    def test_optional_macho_fields_default_none(self):
        from redb.models.dataclasses import Hashes

        h = Hashes(md5="a", sha1="b", sha256="c", ssdeep_hash="d", tlsh_hash="e")
        assert h.macho_dylib_hash is None
        assert h.macho_import_hash is None
        assert h.macho_export_hash is None
        assert h.macho_entitlement_hash is None
        assert h.macho_symhash is None

    def test_macho_fields_settable(self):
        from redb.models.dataclasses import Hashes

        h = Hashes(
            md5="a", sha1="b", sha256="c", ssdeep_hash="d", tlsh_hash="e",
            macho_dylib_hash="dylib_h",
            macho_import_hash="import_h",
            macho_export_hash="export_h",
            macho_entitlement_hash="ent_h",
            macho_symhash="sym_h",
        )
        assert h.macho_dylib_hash == "dylib_h"
        assert h.macho_import_hash == "import_h"
        assert h.macho_export_hash == "export_h"
        assert h.macho_entitlement_hash == "ent_h"
        assert h.macho_symhash == "sym_h"


# ============================================================================
# BasicProperties
# ============================================================================

class TestBasicPropertiesDataclass:
    """Tests for the BasicProperties dataclass."""

    def test_required_fields(self):
        from redb.models.dataclasses import BasicProperties

        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,
        )
        assert bp.filename == "test.exe"
        assert bp.filesize == 1024
        assert bp.file_entropy == 6.5

    def test_optional_fields_default_none(self):
        from redb.models.dataclasses import BasicProperties

        bp = BasicProperties(
            filename="x", sample_name="x", filesize=0,
            filetype="t", filetype_mime="m", filetype_magika="k", file_entropy=0.0,
        )
        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_fat_fields(self):
        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 len(bp.child_sha256) == 2
        assert bp.child_architecture == ["x86_64", "arm64"]

    def test_asdict_roundtrip(self):
        from redb.models.dataclasses import BasicProperties

        bp = BasicProperties(
            filename="f", sample_name="s", filesize=10,
            filetype="t", filetype_mime="m", filetype_magika="k", file_entropy=1.0,
        )
        d = asdict(bp)
        assert d["filename"] == "f"
        assert d["is_fat"] is None


# ============================================================================
# DIEinfo / CAPA
# ============================================================================

class TestDIEinfoDataclass:
    def test_creation(self):
        from redb.models.dataclasses import DIEinfo

        d = DIEinfo(die_full_dump="dump_text")
        assert d.die_full_dump == "dump_text"
        assert d.die_info == {}

    def test_with_info(self):
        from redb.models.dataclasses import DIEinfo

        d = DIEinfo(die_full_dump="x", die_info={"packer": "UPX"})
        assert d.die_info["packer"] == "UPX"


class TestCAPADataclass:
    def test_creation(self):
        from redb.models.dataclasses import CAPA

        c = CAPA(
            capa_dump="dump",
            capabilities=["cap1"],
            namespaces=["ns1"],
            attack_dump="attack",
            tactics=["TA0001"],
            techniques=["T1059"],
            techniques_id=["T1059.001"],
            mbc_dump="mbc",
            mbc_objectives=["obj1"],
            mbc_behaviors=["beh1"],
            mbc_behaviors_id=["B0001"],
        )
        assert c.capa_dump == "dump"
        assert "cap1" in c.capabilities
        assert "TA0001" in c.tactics


# ============================================================================
# Malcontent
# ============================================================================

class TestMalcontentDataclass:
    def test_creation(self):
        from redb.models.dataclasses import Malcontent

        m = Malcontent(
            malcontent_dump='{"RiskScore": 2}',
            version="malcontent v1.0.0",
            risk_score=2,
            risk_level="MEDIUM"
        )
        assert m.malcontent_dump == '{"RiskScore": 2}'
        assert m.version == "malcontent v1.0.0"
        assert m.risk_score == 2
        assert m.risk_level == "MEDIUM"

    def test_with_json_content(self):
        from redb.models.dataclasses import Malcontent
        import json

        # Now stores unwrapped content (without Files/<path> wrapper)
        file_content = {
            "RiskScore": 3,
            "RiskLevel": "HIGH",
            "IsMalcontent": True,
            "Behaviors": [
                {"RuleName": "exec/shell", "RiskLevel": "HIGH", "Description": "executes shell commands"}
            ]
        }
        m = Malcontent(
            malcontent_dump=json.dumps(file_content),
            version="v0.5.0",
            risk_score=3,
            risk_level="HIGH"
        )
        parsed = json.loads(m.malcontent_dump)
        assert parsed["RiskLevel"] == "HIGH"
        assert parsed["Behaviors"][0]["RuleName"] == "exec/shell"

    def test_asdict(self):
        from redb.models.dataclasses import Malcontent
        from dataclasses import asdict

        m = Malcontent(malcontent_dump="{}", version="v1", risk_score=0, risk_level="")
        d = asdict(m)
        assert d == {"malcontent_dump": "{}", "version": "v1", "risk_score": 0, "risk_level": ""}


# ============================================================================
# String
# ============================================================================

class TestStringDataclass:
    def test_creation(self):
        from redb.models.dataclasses import String

        s = String(string="hello", string_length=5, string_entropy=2.3, string_frequency=10)
        assert s.string == "hello"
        assert s.md5 == []
        assert s.sha1 == []
        assert s.sha256 == []

    def test_with_hashes(self):
        from redb.models.dataclasses import String

        s = String(
            string="test", string_length=4, string_entropy=2.0, string_frequency=1,
            md5=["abc"], sha256=["def"],
        )
        assert s.md5 == ["abc"]
        assert s.sha256 == ["def"]


# ============================================================================
# PE Dataclasses
# ============================================================================

class TestPEDataclass:
    def test_required_fields(self):
        from redb.models.dataclasses import PE

        pe = PE(
            dos_header="dos", nt_header="nt", optional_header="opt",
            file_header="fh", magic="0x10b", entry_point="0x1000",
            major_linker_version=14, minor_linker_version=0,
            target_machine="I386", architecture="x86",
            compilation_time=1609459200, compilation_time_utc="2021-01-01T00:00:00",
            is_dotnet=False, is_signed=False, has_overlay=False,
            number_of_sections=4, number_of_imports=10,
            number_of_exports=0, number_of_resources=2,
            type="EXE",
        )
        assert pe.architecture == "x86"
        assert pe.type == "EXE"

    def test_optional_fields_default_none(self):
        from redb.models.dataclasses import PE

        pe = PE(
            dos_header="d", nt_header="n", optional_header="o",
            file_header="f", magic="m", entry_point="e",
            major_linker_version=0, minor_linker_version=0,
            target_machine="t", architecture="a",
            compilation_time=0, compilation_time_utc="",
            is_dotnet=False, is_signed=False, has_overlay=False,
            number_of_sections=0, number_of_imports=0,
            number_of_exports=0, number_of_resources=0,
            type="DLL",
        )
        assert pe.dbg_struct is None
        assert pe.tls_struct is None
        assert pe.rich_header_dump is None
        assert pe.version_info is None


class TestPEImportDataclass:
    def test_creation(self):
        from redb.models.dataclasses import PEImport

        pi = PEImport(pe_imports_total=5)
        assert pi.pe_imports_total == 5
        assert pi.pe_import_libraryName is None
        assert pi.pe_import_functions is None


class TestPESectionDataclass:
    def test_creation(self):
        from redb.models.dataclasses import PESection

        ps = PESection(
            _id="sec_id", section_entropy=6.5, section_sha256="sha",
            section_md5="md5", section_name=".text", section_name_b64=".text",
            section_pointer_to_raw_data="0x400",
            section_size=4096, section_v_addr="0x1000",
            section_v_addr_hex="0x1000", section_v_size=8192,
        )
        assert ps.section_name == ".text"
        assert ps.section_entropy == 6.5


class TestPEResourceDataclass:
    def test_creation(self):
        from redb.models.dataclasses import PEResource

        pr = PEResource(
            _id="res_id", resource_type="RT_ICON", resource_entropy=3.5,
            resource_sha256="sha", resource_filetype="image/png",
            resource_magika="png", resource_language="LANG_ENGLISH",
            resource_sub_lang="SUBLANG_DEFAULT",
            resource_rva="0x2000", resource_size=1024,
        )
        assert pr.resource_type == "RT_ICON"


class TestPEOverlayDataclass:
    def test_creation(self):
        from redb.models.dataclasses import PEOverlay

        po = PEOverlay(
            _id="ovl_id", overlay_size=512, overlay_entropy=7.9,
            overlay_offset="0x5000", overlay_mimetype="application/octet-stream",
            overlay_type="data", overlay_magika="unknown",
            overlay_sha256="sha", overlay_sha1="sha1", overlay_md5="md5",
        )
        assert po.overlay_size == 512


class TestPESignerDataclass:
    def test_creation(self):
        from redb.models.dataclasses import PESigner

        ps = PESigner(
            signer_serial_number="01", signer_subject="CN=Test",
            signer_issuer="CN=CA", signer_valid_from="2021-01-01",
            signer_valid_to="2025-01-01", signer_thumbprint="aabb",
            signer_algorithm="sha256WithRSAEncryption",
        )
        assert ps.signer_subject == "CN=Test"


class TestPECertificateDataclass:
    def test_creation(self):
        from redb.models.dataclasses import PECertificate

        pc = PECertificate(
            certificate_serial_number="02", certificate_subject="CN=Cert",
            certificate_issuer="CN=CA", certificate_valid_from="2021-01-01",
            certificate_valid_to="2025-01-01", certificate_thumbprint="ccdd",
            certificate_algorithm="sha256",
        )
        assert pc.certificate_serial_number == "02"


class TestPECodeSigningInfoDataclass:
    def test_creation(self):
        from redb.models.dataclasses import PECodeSigningInfo

        cs = PECodeSigningInfo(
            _id="cs_id", signature_verified=True, number_of_certificates=2,
        )
        assert cs.signature_verified is True
        assert cs.x509_certificates is None

    def test_with_certificates(self):
        from redb.models.dataclasses import PECodeSigningInfo

        cs = PECodeSigningInfo(
            _id="cs_id", signature_verified=False, number_of_certificates=0,
            x509_certificates=[{"cn": "test"}],
        )
        assert len(cs.x509_certificates) == 1


class TestPEExtraFindingDataclass:
    def test_creation(self):
        from redb.models.dataclasses import PEExtraFinding

        ef = PEExtraFinding(context="anomaly_detection")
        assert ef.context == "anomaly_detection"
        assert ef.finding is None


class TestPEDotNetDataclass:
    def test_creation(self):
        from redb.models.dataclasses import PEDotNet

        dn = PEDotNet()
        assert dn.dotnet == {}

    def test_with_data(self):
        from redb.models.dataclasses import PEDotNet

        dn = PEDotNet(dotnet={"runtime": "v4.0.30319"})
        assert dn.dotnet["runtime"] == "v4.0.30319"


class TestDotNetInconsistencyTestsDataclass:
    def test_creation(self):
        from redb.models.dataclasses import DotNetInconsistencyTests

        it = DotNetInconsistencyTests(
            test_dotnet_data_dir_hidden=False,
            test_dotnet_fake_types=True,
            test_dotnet_extra_data=None,
            test_dotnet_invalid_type_ref=False,
            test_dotnet_fake_datastreams=False,
            test_dotnet_extra_module_table=None,
            test_dotnet_extra_assembly_table=False,
            test_dotnet_invalid_strings_stream=False,
            test_dotnet_streams_mixed_case=True,
            test_dotnet_method_def_invalid_table=False,
            test_dotnet_max_len_exceeding_strings=False,
        )
        assert it.test_dotnet_fake_types is True
        assert it.test_dotnet_extra_data is None


class TestPEInconsistencyTestsDataclass:
    def test_creation(self):
        from redb.models.dataclasses import PEInconsistencyTests

        it = PEInconsistencyTests(
            test_rich_header_checksum=True,
            test_rich_header_duplicate=False,
            test_rich_header_linker=None,
            test_rich_header_import_count=True,
        )
        assert it.test_rich_header_checksum is True
        assert it.test_rich_header_linker is None


# ============================================================================
# YARA Dataclasses
# ============================================================================

class TestYaraRuleDataclass:
    def test_creation(self):
        from redb.models.dataclasses import YaraRule

        yr = YaraRule(
            rule_name="test_rule",
            rule_tags=["malware", "trojan"],
            rule_meta={"author": "test", "description": "A test rule"},
        )
        assert yr.rule_name == "test_rule"
        assert "malware" in yr.rule_tags


class TestYaraMatchDataclass:
    def test_creation(self):
        from redb.models.dataclasses import YaraMatch

        ym = YaraMatch(rule_name="test_rule", match_strings=["$s1", "$s2"])
        assert ym.rule_name == "test_rule"
        assert len(ym.match_strings) == 2


# ============================================================================
# ELF Dataclasses
# ============================================================================

class TestELFFeaturesDataclass:
    def test_creation(self):
        from redb.models.dataclasses import ELFFeatures

        ef = ELFFeatures(
            ei_class=2, ei_data=1, ei_version=1, ei_osabi=0, ei_abiversion=0,
            e_type=2, e_machine=62, e_version=1, e_entry=0x400000, e_flags=0,
            ei_class_str="ELFCLASS64", ei_data_str="ELFDATA2LSB",
            ei_osabi_str="ELFOSABI_SYSV", e_type_str="ET_EXEC",
            e_machine_str="EM_X86_64",
            is_64bit=1, is_stripped=0, is_pie=0, has_canary=1, has_nx=1,
            has_relro=1, has_fortify=0,
            number_of_segments=9, number_of_sections=31,
            number_of_symbols=100, number_of_dynamic_symbols=50,
            number_of_relocations=25, number_of_dependencies=5,
            gnu_hash_present=1, has_debug_info=0,
        )
        assert ef.ei_class == 2
        assert ef.e_machine_str == "EM_X86_64"
        assert ef.build_id is None

    def test_with_build_id(self):
        from redb.models.dataclasses import ELFFeatures

        ef = ELFFeatures(
            ei_class=2, ei_data=1, ei_version=1, ei_osabi=0, ei_abiversion=0,
            e_type=3, e_machine=62, e_version=1, e_entry=0, e_flags=0,
            ei_class_str="64", ei_data_str="LE", ei_osabi_str="SYSV",
            e_type_str="DYN", e_machine_str="X86_64",
            is_64bit=1, is_stripped=1, is_pie=1, has_canary=0, has_nx=1,
            has_relro=1, has_fortify=0,
            number_of_segments=8, number_of_sections=28,
            number_of_symbols=0, number_of_dynamic_symbols=30,
            number_of_relocations=10, number_of_dependencies=3,
            gnu_hash_present=1, has_debug_info=0,
            build_id="abc123def456",
        )
        assert ef.build_id == "abc123def456"


class TestELFDependencyDataclass:
    def test_creation(self):
        from redb.models.dataclasses import ELFDependency

        dep = ELFDependency(
            dependency_name="libc.so.6",
            dependency_type=1,
            dependency_type_str="NEEDED",
        )
        assert dep.dependency_name == "libc.so.6"


class TestELFImportDataclass:
    def test_creation(self):
        from redb.models.dataclasses import ELFImport

        ei = ELFImport(elf_imports_total=3)
        assert ei.elf_imports_total == 3
        assert ei.elf_import_libraries == []
        assert ei.elf_import_functions == []


class TestELFExportDataclass:
    def test_creation(self):
        from redb.models.dataclasses import ELFExport

        ee = ELFExport(elf_exports_total=5)
        assert ee.elf_exports_total == 5
        assert ee.elf_export_functions == []


class TestELFSectionDataclass:
    def test_creation(self):
        from redb.models.dataclasses import ELFSection

        es = ELFSection(
            section_name=".text", section_type=1, section_type_str="SHT_PROGBITS",
            section_flags=6, section_flags_str=["SHF_ALLOC", "SHF_EXECINSTR"],
            section_addr=0x400000, section_offset=0x1000, section_size=4096,
            section_link=0, section_info=0, section_addralign=16, section_entsize=0,
            section_entropy=6.8, section_sha256="sha", section_md5="md5",
        )
        assert es.section_name == ".text"
        assert "SHF_ALLOC" in es.section_flags_str


class TestELFSegmentDataclass:
    def test_creation(self):
        from redb.models.dataclasses import ELFSegment

        seg = ELFSegment(
            segment_type=1, segment_type_str="PT_LOAD",
            segment_flags=5, segment_flags_str=["PF_R", "PF_X"],
            segment_offset=0, segment_vaddr=0x400000, segment_paddr=0x400000,
            segment_filesz=0x1000, segment_memsz=0x1000, segment_align=0x200000,
            segment_entropy=5.5, segment_sha256="sha", segment_md5="md5",
        )
        assert seg.segment_type_str == "PT_LOAD"


class TestELFSymbolDataclass:
    def test_creation(self):
        from redb.models.dataclasses import ELFSymbol

        sym = ELFSymbol(
            symbol_name="main", symbol_value=0x401000, symbol_size=100,
            symbol_type=2, symbol_type_str="STT_FUNC",
            symbol_bind=1, symbol_bind_str="STB_GLOBAL",
            symbol_visibility=0, symbol_visibility_str="STV_DEFAULT",
            symbol_section_index=14, symbol_section_index_str="14",
            is_dynamic=0,
        )
        assert sym.symbol_name == "main"
        assert sym.symbol_type_str == "STT_FUNC"


class TestELFRelocationDataclass:
    def test_creation(self):
        from redb.models.dataclasses import ELFRelocation

        rel = ELFRelocation(
            relocation_offset=0x601000, relocation_type=7,
            relocation_type_str="R_X86_64_JUMP_SLOT",
            relocation_symbol_index=1, relocation_symbol_name="printf",
            relocation_section=".rela.plt",
        )
        assert rel.relocation_addend is None

    def test_with_addend(self):
        from redb.models.dataclasses import ELFRelocation

        rel = ELFRelocation(
            relocation_offset=0x601000, relocation_type=7,
            relocation_type_str="R_X86_64_JUMP_SLOT",
            relocation_symbol_index=1, relocation_symbol_name="printf",
            relocation_section=".rela.plt",
            relocation_addend=-4,
        )
        assert rel.relocation_addend == -4


class TestELFNoteDataclass:
    def test_creation(self):
        from redb.models.dataclasses import ELFNote

        note = ELFNote(
            note_name="GNU", note_type=3,
            note_type_str="NT_GNU_BUILD_ID",
            note_desc="abc123", note_section=".note.gnu.build-id",
        )
        assert note.note_name == "GNU"
        assert note.note_type_str == "NT_GNU_BUILD_ID"


# ============================================================================
# Decompiled
# ============================================================================

class TestDecompiledDataclass:
    def test_creation(self):
        from redb.models.dataclasses import Decompiled

        d = Decompiled(
            _id="func_hash",
            decompiled_function_name="main",
            decompiled_function_address="0x401000",
            decompiled_function="int main() { return 0; }",
        )
        assert d.decompiled_function_name == "main"