docs/source/user-guide/common-operations/spark-functions.md

<!---
  Licensed to the Apache Software Foundation (ASF) under one
  or more contributor license agreements.  See the NOTICE file
  distributed with this work for additional information
  regarding copyright ownership.  The ASF licenses this file
  to you under the Apache License, Version 2.0 (the
  "License"); you may not use this file except in compliance
  with the License.  You may obtain a copy of the License at

    http://www.apache.org/licenses/LICENSE-2.0

  Unless required by applicable law or agreed to in writing,
  software distributed under the License is distributed on an
  "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
  KIND, either express or implied.  See the License for the
  specific language governing permissions and limitations
  under the License.
-->

(spark-functions)=

# Spark-Compatible Functions

DataFusion ships Spark-compatible versions of a wide set of functions
(string, math, datetime, hash, array, aggregate) through the upstream
`datafusion-spark` crate. `datafusion-python` exposes these under
`datafusion.functions.spark` for use from the DataFrame API, and via
{py:meth}`~datafusion.SessionContext.enable_spark_functions` for use from
SQL.

## Why a Separate Namespace?

Several Spark functions share names with DataFusion built-ins but differ in
semantics. The most common divergences:

- `concat` propagates NULL. `concat('a', NULL, 'b')` returns NULL under
  Spark semantics, whereas the DataFusion default returns `'ab'`.
- `substring` is 1-indexed and supports negative positions counting from
  the end of the string.
- `round` uses HALF_UP rounding mode (`round(2.5, 0) == 3`).
- Numeric functions (`floor`, `ceil`, `mod`) follow Spark's edge-case
  handling for negative values and decimals.

Enabling Spark functions does not affect the DataFrame API: you choose which
implementation to call by which module you import from.

## DataFrame API

Import `spark` and use it like any other functions module. The Spark
functions can go anywhere you'd put a DataFusion expression — inside
`select`, `filter`, `with_column`, `aggregate`, and so on.

```python
from datafusion import SessionContext, col, lit
from datafusion.functions import spark

ctx = SessionContext()
df = ctx.from_pydict({"s": ["hello", "world"]})

# SHA-256 hash with Spark semantics
df.select(spark.sha2(col("s"), lit(256)).alias("h")).show()

# 1-indexed substring
df.select(spark.substring(col("s"), lit(1), lit(3)).alias("p")).show()
```

## SQL

To use Spark functions in SQL queries, call
{py:meth}`~datafusion.SessionContext.enable_spark_functions` on the context.
This registers every Spark UDF/UDAF/UDWF, overriding any DataFusion built-in
of the same name.

```python
from datafusion import SessionContext

ctx = SessionContext()
ctx.enable_spark_functions()

ctx.sql("SELECT sha2('hello', 256)").show()
ctx.sql("SELECT concat('a', NULL, 'b')").show()   # -> NULL, not 'ab'
```

The override applies for the lifetime of the session. To call DataFusion's
built-in versions afterwards, create a fresh `SessionContext`.

## Function Reference

The full, up-to-date list of available Spark functions — with signatures
and per-function docstrings — lives in the
{py:mod}`datafusion.functions.spark` API reference.