zylon-ai/private-gpt · error · ImportError

Tabular data dependencies are not installed. Install with `u

Error message

Tabular data dependencies are not installed. Install with `uv sync --inexact --extra tool-tabular`.

What it means

ImportError raised by _load_pandas_ai_service when private_gpt.components.tabular.pandasai_service cannot be imported. The tabular data analysis tool depends on pandas-ai and friends, shipped as the optional 'tool-tabular' extra; the loader converts the raw ImportError into an actionable install message.

Source

Thrown at private_gpt/components/tools/builders/tabular_data_builder.py:62

from private_gpt.utils.dependencies import format_missing_dependency_message

if TYPE_CHECKING:
    from collections.abc import Coroutine

    from private_gpt.components.tabular.pandasai_service import PandasAIService
    from private_gpt.components.workflows.tabular.tabular_data import (
        TabularDataAnalysisInputEvent,
        TabularDataAnalysisWorkflow,
    )

config = settings()


def _load_pandas_ai_service() -> type["PandasAIService"]:
    try:
        from private_gpt.components.tabular.pandasai_service import PandasAIService
    except ImportError as e:
        raise ImportError(
            format_missing_dependency_message(
                "Tabular data",
                extras="tool-tabular",
            )
        ) from e

    return PandasAIService


def _load_tabular_workflow_dependencies() -> tuple[
    type["TabularDataAnalysisInputEvent"],
    type["TabularDataAnalysisWorkflow"],
]:
    try:
        from private_gpt.components.workflows.tabular.tabular_data import (
            TabularDataAnalysisInputEvent,
            TabularDataAnalysisWorkflow,
        )

View on GitHub (pinned to 4a030776a3)

Solutions

Run uv sync --inexact --extra tool-tabular and retry.
If the import still fails after install, check for version conflicts: python -c "import private_gpt.components.tabular.pandasai_service" and read the chained ImportError for the actual missing module.
Rebuild Docker/CI images with the extra included.

Example fix

# before
uv sync --inexact

# after
uv sync --inexact --extra tool-tabular

Defensive patterns

Strategy: type-guard

Validate before calling

def pandasai_available() -> bool:
    try:
        import private_gpt.components.tabular.pandasai_service  # noqa: F401
        return True
    except ImportError:
        return False

Try / catch

try:
    svc = _load_pandas_ai_service()
except ImportError as e:
    logger.warning("tabular tool disabled: %s", e)
    return []  # skip registering the tool

Prevention

Install uv sync --inexact --extra tool-tabular in every environment that enables the tabular tool.
Probe availability before registering the tool so missing deps do not crash at runtime.
Pin compatible versions of pandasai in the lockfile to avoid broken transitive imports.

When it happens

Trigger: Building or invoking the tabular data tool in an environment without the tool-tabular extra installed; the import of PandasAIService fails and is chained.

Common situations: Base installation used for a deployment that later enables tabular analysis; CI or Docker image built with a minimal extra set; dependency conflict causing pandasai import failure even when nominally installed.

Related errors

AI-assisted analysis of zylon-ai/private-gpt@4a030776a3 (2026-08-15). Data as JSON: /api/errors/8c5dbc94c1e00941. Report an issue: GitHub.