Implementation:Eventual Inc Daft DataFrame To Arrow

Knowledge Sources	Daft Daft Docs
Domains	Data_Engineering, Interoperability
Last Updated	2026-02-08 00:00 GMT

Overview

Concrete tool for converting a Daft DataFrame to a PyArrow Table provided by the Daft library. This is a wrapper doc type for the Arrow interoperability method.

Description

The to_arrow method on DataFrame converts the current Daft DataFrame to a PyArrow Table. It first streams all partitions as Arrow RecordBatches via to_arrow_iter(results_buffer_size=None), then combines them into a single PyArrow Table using pyarrow.Table.from_batches with the DataFrame's schema converted to a PyArrow schema. This is a blocking call that triggers execution of the lazy query plan.

Usage

Call df.to_arrow() on a DataFrame instance. Requires the pyarrow package. Use when you need Arrow-native output for interoperability with other data systems.

Code Reference

Source Location

• Repository: Daft
• File: daft/dataframe/dataframe.py
• Lines: L4438-4468

Signature

def to_arrow(self) -> pyarrow.Table

Import

# Method on DataFrame, no separate import needed
# Requires: pyarrow
table = df.to_arrow()

I/O Contract

Inputs

Name	Type	Required	Description
(none)	--	--	The method takes no arguments.

Outputs

Name	Type	Description
return	pyarrow.Table	A PyArrow Table containing the materialized data from the Daft DataFrame.

External Dependencies

• pyarrow - required for output type

Usage Examples

Basic Usage

import daft

df = daft.from_pydict({"a": [1, 2, 3], "b": [4, 5, 6]})
arrow_table = df.to_arrow()
print(arrow_table)
# pyarrow.Table
# a: int64
# b: int64
# ----
# a: [[1,2,3]]
# b: [[4,5,6]]

Interop with DuckDB

import daft
import duckdb

df = daft.from_pydict({"x": [1, 2, 3], "y": [4, 5, 6]})
arrow_table = df.to_arrow()

# Query the Arrow table with DuckDB
result = duckdb.query("SELECT * FROM arrow_table WHERE x > 1")

Related Pages

Implements Principle

• Principle:Eventual_Inc_Daft_Arrow_Export

Requires Environment

• Environment:Eventual_Inc_Daft_Python_PyArrow_Core