Method write

pandas/io/parquet.py:174–238

(
        self,
        df: DataFrame,
        path: FilePath | WriteBuffer[bytes],
        compression: ParquetCompressionOptions = "snappy",
        index: bool | None = None,
        storage_options: StorageOptions | None = None,
        partition_cols: list[str] | None = None,
        filesystem=None,
        **kwargs,
    )

        self.api = pyarrow

    def write(
        self,
        df: DataFrame,
        path: FilePath | WriteBuffer[bytes],
        compression: ParquetCompressionOptions = "snappy",
        index: bool | None = None,
        storage_options: StorageOptions | None = None,
        partition_cols: list[str] | None = None,
        filesystem=None,
        **kwargs,
    ) -> None:
        self.validate_dataframe(df)

        from_pandas_kwargs: dict[str, Any] = {"schema": kwargs.pop("schema", None)}
        if index is not None:
            from_pandas_kwargs["preserve_index"] = index

        table = self.api.Table.from_pandas(df, **from_pandas_kwargs)

        if df.attrs:
            df_metadata = {"PANDAS_ATTRS": json.dumps(df.attrs)}
            existing_metadata = table.schema.metadata
            merged_metadata = {**existing_metadata, **df_metadata}
            table = table.replace_schema_metadata(merged_metadata)

        path_or_handle, handles, filesystem = _get_path_or_handle(
            path,
            filesystem,
            storage_options=storage_options,
            mode="wb",
            is_dir=partition_cols is not None,
        )
        if (
            isinstance(path_or_handle, io.BufferedWriter)
            and hasattr(path_or_handle, "name")
            and isinstance(path_or_handle.name, (str, bytes))
        ):
            if isinstance(path_or_handle.name, bytes):
                path_or_handle = path_or_handle.name.decode()
            else:
                path_or_handle = path_or_handle.name

        try:
            if partition_cols is not None:
                # writes to multiple files under the given path
                self.api.parquet.write_to_dataset(
                    table,
                    path_or_handle,
                    compression=compression,
                    partition_cols=partition_cols,
                    filesystem=filesystem,
                    **kwargs,
                )
            else:
                # write to single output file
                self.api.parquet.write_table(
                    table,
                    path_or_handle,