BerriAI/litellm · error · NotImplementedError

Export format '{self.export_format}' not supported. Use 'par

Error message

Export format '{self.export_format}' not supported. Use 'parquet' or 'csv'.

What it means

FocusExportEngine._init_serializer raises NotImplementedError when export_format is neither 'csv' nor 'parquet'. The engine maps the format string to a serializer class (FocusCsvSerializer / FocusParquetSerializer); anything else has no implementation.

Source

Thrown at litellm/integrations/focus/export_engine.py:45

    ) -> None:
        self.provider = provider
        self.export_format = export_format
        self.prefix = prefix
        self._destination = FocusDestinationFactory.create(
            provider=self.provider,
            prefix=self.prefix,
            config=destination_config,
        )
        self._serializer = self._init_serializer()
        self._transformer = FocusTransformer()
        self._database = FocusLiteLLMDatabase()

    def _init_serializer(self) -> FocusSerializer:
        if self.export_format == "csv":
            return FocusCsvSerializer()
        if self.export_format == "parquet":
            return FocusParquetSerializer()
        raise NotImplementedError(f"Export format '{self.export_format}' not supported. Use 'parquet' or 'csv'.")

    async def dry_run_export_usage_data(self, limit: int | None) -> dict[str, Any]:
        data: Final = await self._database.get_usage_data(limit=limit)
        normalized: Final = self._transformer.transform(data)

        usage_sample: Final = data.head(min(50, len(data))).to_dicts()
        normalized_sample: Final = normalized.head(min(50, len(normalized))).to_dicts()

        summary: Final = {
            "total_records": len(normalized),
            "total_spend": self._sum_column(data, "spend"),
            "total_tokens": self._sum_column(data, "total_tokens"),
            "unique_teams": self._count_unique(data, "team_id"),
            "unique_models": self._count_unique(data, "model"),
        }

        return {
            "usage_data": usage_sample,

View on GitHub (pinned to 6c2dcb801b)

Solutions

1. Set export_format to exactly 'parquet' or 'csv' (lowercase).
2. Validate/normalize the value (strip + lower) in your config layer before constructing the engine.
3. If you need another format, subclass FocusSerializer and register it — do not expect built-in support.

Example fix

# before
FocusExportEngine(export_format="JSON", ...)

# after
FocusExportEngine(export_format="parquet", ...)

fmt = (export_format or "").strip().lower()
if fmt not in {"parquet", "csv"}:
    raise ValueError(f"export_format must be 'parquet' or 'csv', got {export_format!r}")
engine = FocusExportEngine(export_format=fmt, ...)

from typing import Literal
FocusFormat = Literal["parquet", "csv"]

def is_focus_format(v: str) -> bool:
    return v in {"parquet", "csv"}

When it happens

Trigger: Initializing FocusExportEngine with export_format='json', 'JSON', 'Parquet' (case-sensitive), or None via constructor or settings/env-driven config.

Common situations: Case mismatch ('CSV' vs 'csv'); copy-pasting a format from a different tool's docs; leaving an empty string default from templated config.

Related errors

• bucket_name must be provided for S3 destination
• Unsupported frequency: {self.frequency}
• Event hook {hook} is not in the supported event hooks {suppo
• Event hook {event_hook} is not in the supported event hooks
• Invalid environment: {environment}. Please use one of the fo

AI-assisted analysis of BerriAI/litellm@6c2dcb801b (2026-08-15). Data as JSON: /api/errors/c7e9af4be5d2eab0. Report an issue: GitHub.