learningcircuit--local-deep-research/tests/document_loaders/test_loader_registry.py

"""Tests for loader_registry module."""

from pathlib import Path

from langchain_core.document_loaders import BaseLoader

from local_deep_research.document_loaders.loader_registry import (
    LOADER_REGISTRY,
    get_loader_class_for_extension,
    get_loader_for_path,
    get_supported_extensions,
    is_extension_supported,
)


class TestGetSupportedExtensions:
    """Tests for get_supported_extensions function."""

    def test_returns_list(self) -> None:
        """Test that function returns a list."""
        result = get_supported_extensions()
        assert isinstance(result, list)

    def test_includes_common_extensions(self) -> None:
        """Test that common extensions are included."""
        extensions = get_supported_extensions()

        # Core document formats
        assert ".pdf" in extensions
        assert ".txt" in extensions
        assert ".md" in extensions
        assert ".docx" in extensions
        # .doc (legacy OLE binary) is only registered when LibreOffice is
        # installed, so it is intentionally not asserted here.

        # Spreadsheets
        assert ".csv" in extensions
        assert ".xlsx" in extensions

        # Web
        assert ".html" in extensions

        # Data formats
        assert ".json" in extensions
        assert ".yaml" in extensions
        assert ".yml" in extensions

    def test_includes_yaml_extensions(self) -> None:
        """Test that YAML extensions are included."""
        extensions = get_supported_extensions()

        assert ".yaml" in extensions
        assert ".yml" in extensions

    def test_no_duplicates(self) -> None:
        """Test that there are no duplicate extensions."""
        extensions = get_supported_extensions()
        assert len(extensions) == len(set(extensions))


class TestIsExtensionSupported:
    """Tests for is_extension_supported function."""

    def test_supported_with_dot(self) -> None:
        """Test checking supported extension with leading dot."""
        assert is_extension_supported(".pdf") is True
        assert is_extension_supported(".txt") is True
        assert is_extension_supported(".json") is True

    def test_supported_without_dot(self) -> None:
        """Test checking supported extension without leading dot."""
        assert is_extension_supported("pdf") is True
        assert is_extension_supported("txt") is True
        assert is_extension_supported("json") is True

    def test_unsupported_extension(self) -> None:
        """Test checking unsupported extension."""
        assert is_extension_supported(".xyz") is False
        assert is_extension_supported(".unknown") is False
        assert is_extension_supported("notreal") is False

    def test_case_insensitive(self) -> None:
        """Test that extension check is case insensitive."""
        assert is_extension_supported(".PDF") is True
        assert is_extension_supported(".Pdf") is True
        assert is_extension_supported("PDF") is True

    def test_yaml_extensions(self) -> None:
        """Test YAML extensions are supported."""
        assert is_extension_supported(".yaml") is True
        assert is_extension_supported(".yml") is True
        assert is_extension_supported("yaml") is True
        assert is_extension_supported("yml") is True


class TestGetLoaderForPath:
    """Tests for get_loader_for_path function."""

    def test_returns_loader_for_txt(self, tmp_path: Path) -> None:
        """Test getting loader for text file."""
        txt_file = tmp_path / "test.txt"
        txt_file.write_text("Hello world")

        loader = get_loader_for_path(txt_file)

        assert loader is not None
        assert isinstance(loader, BaseLoader)

    def test_returns_loader_for_json(self, tmp_path: Path) -> None:
        """Test getting loader for JSON file."""
        json_file = tmp_path / "test.json"
        json_file.write_text('{"key": "value"}')

        loader = get_loader_for_path(json_file)

        assert loader is not None
        assert isinstance(loader, BaseLoader)

    def test_returns_loader_for_yaml(self, tmp_path: Path) -> None:
        """Test getting loader for YAML file."""
        yaml_file = tmp_path / "test.yaml"
        yaml_file.write_text("key: value")

        loader = get_loader_for_path(yaml_file)

        assert loader is not None
        assert isinstance(loader, BaseLoader)

    def test_returns_loader_for_yml(self, tmp_path: Path) -> None:
        """Test getting loader for .yml file."""
        yml_file = tmp_path / "test.yml"
        yml_file.write_text("key: value")

        loader = get_loader_for_path(yml_file)

        assert loader is not None
        assert isinstance(loader, BaseLoader)

    def test_returns_none_for_unsupported(self, tmp_path: Path) -> None:
        """Test that None is returned for unsupported extension."""
        unknown_file = tmp_path / "test.xyz"
        unknown_file.write_text("content")

        loader = get_loader_for_path(unknown_file)

        assert loader is None

    def test_accepts_string_path(self, tmp_path: Path) -> None:
        """Test that function accepts string path."""
        txt_file = tmp_path / "test.txt"
        txt_file.write_text("content")

        loader = get_loader_for_path(str(txt_file))

        assert loader is not None

    def test_case_insensitive_extension(self, tmp_path: Path) -> None:
        """Test that extension matching is case insensitive."""
        txt_file = tmp_path / "test.TXT"
        txt_file.write_text("content")

        loader = get_loader_for_path(txt_file)

        assert loader is not None


class TestGetLoaderClassForExtension:
    """Tests for get_loader_class_for_extension function."""

    def test_returns_tuple_for_supported(self) -> None:
        """Test that function returns (class, kwargs) tuple."""
        result = get_loader_class_for_extension(".txt")

        assert result is not None
        assert isinstance(result, tuple)
        assert len(result) == 2

        loader_class, kwargs = result
        assert callable(loader_class)
        assert isinstance(kwargs, dict)

    def test_returns_none_for_unsupported(self) -> None:
        """Test that function returns None for unsupported extension."""
        result = get_loader_class_for_extension(".xyz")
        assert result is None

    def test_with_dot_prefix(self) -> None:
        """Test with extension that has dot prefix."""
        result = get_loader_class_for_extension(".pdf")
        assert result is not None

    def test_without_dot_prefix(self) -> None:
        """Test with extension without dot prefix."""
        result = get_loader_class_for_extension("pdf")
        assert result is not None

    def test_case_insensitive(self) -> None:
        """Test case insensitivity."""
        result1 = get_loader_class_for_extension(".PDF")
        result2 = get_loader_class_for_extension(".pdf")

        assert result1 is not None
        assert result2 is not None
        assert result1[0] == result2[0]

    def test_json_returns_simple_json_loader(self) -> None:
        """Test that JSON extension returns SimpleJSONLoader."""
        from local_deep_research.document_loaders.json_loader import (
            SimpleJSONLoader,
        )

        result = get_loader_class_for_extension(".json")
        assert result is not None
        assert result[0] == SimpleJSONLoader


class TestLoaderRegistry:
    """Tests for LOADER_REGISTRY constant."""

    def test_registry_is_dict(self) -> None:
        """Test that registry is a dictionary."""
        assert isinstance(LOADER_REGISTRY, dict)

    def test_registry_entries_have_loader_class(self) -> None:
        """Test that all entries have loader_class."""
        for ext, entry in LOADER_REGISTRY.items():
            assert "loader_class" in entry, f"Missing loader_class for {ext}"
            assert callable(entry["loader_class"]), (
                f"loader_class not callable for {ext}"
            )

    def test_registry_entries_have_loader_kwargs(self) -> None:
        """Test that all entries have loader_kwargs (even if empty)."""
        for ext, entry in LOADER_REGISTRY.items():
            # loader_kwargs should exist or default to empty dict
            kwargs = entry.get("loader_kwargs", {})
            assert isinstance(kwargs, dict), f"loader_kwargs not dict for {ext}"

    def test_registry_keys_start_with_dot(self) -> None:
        """Test that all registry keys start with a dot."""
        for ext in LOADER_REGISTRY.keys():
            assert ext.startswith("."), (
                f"Extension {ext} doesn't start with dot"
            )

    def test_registry_keys_are_lowercase(self) -> None:
        """Test that all registry keys are lowercase."""
        for ext in LOADER_REGISTRY.keys():
            assert ext == ext.lower(), f"Extension {ext} is not lowercase"