Files
datafusion/datafusion-examples/examples/builtin_functions/function_factory.rs
T
Adrian Garcia Badaracco 35ff4ab0a0 Allow logical optimizer to be run without evaluating now() & refactor SimplifyInfo (#19505)
In trying to fix #19418 I kept getting turned around about what was
needed where. The `SimplifyInfo` trait made it extra hard to understand.
I ended up realizing that the main reason for the trait to exist was
tests. Removing the trait and adding a builder style API to
`SimplifyContext` made it IMO more ergonomic for tests and other call
sites, easier to track the code (no trait opaqueness) and clearer what
simplification capabilities are available in each site. This got rid of
e.g. some places where we were calling `ExecutionProps::new()` just to
pass that into `SimplifyContext` which in turn would hand out references
to the default query time, a default `ContigOptions`, etc; or in
`datafusion/core/src/execution/session_state.rs` where we did `let
dummy_schema = DFSchema::empty()`.

This let me solve several problems:
- Can store optimized logical plans for prepared statements
- Users can optionally run an optimizer pass on logical plans without
evaluating time functions

Compared to #19426 this avoids adding a config option and is actually
less lines of code (negative diff).

Fixes #19418, closes #19426 (replaces it).

---------

Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
2026-01-08 00:38:46 +00:00

224 lines
7.7 KiB
Rust

// 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.
//! See `main.rs` for how to run it.
use arrow::datatypes::DataType;
use datafusion::common::tree_node::{Transformed, TreeNode};
use datafusion::common::{DataFusionError, exec_datafusion_err, exec_err, internal_err};
use datafusion::error::Result;
use datafusion::execution::context::{
FunctionFactory, RegisterFunction, SessionContext, SessionState,
};
use datafusion::logical_expr::simplify::{ExprSimplifyResult, SimplifyContext};
use datafusion::logical_expr::sort_properties::{ExprProperties, SortProperties};
use datafusion::logical_expr::{
ColumnarValue, CreateFunction, Expr, ScalarFunctionArgs, ScalarUDF, ScalarUDFImpl,
Signature, Volatility,
};
use std::hash::Hash;
use std::result::Result as RResult;
use std::sync::Arc;
/// This example shows how to utilize [FunctionFactory] to implement simple
/// SQL-macro like functions using a `CREATE FUNCTION` statement. The same
/// functionality can support functions defined in any language or library.
///
/// Apart from [FunctionFactory], this example covers
/// [ScalarUDFImpl::simplify()] which is often used at the same time, to replace
/// a function call with another expression at runtime.
///
/// This example is rather simple and does not cover all cases required for a
/// real implementation.
pub async fn function_factory() -> Result<()> {
// First we must configure the SessionContext with our function factory
let ctx = SessionContext::new()
// register custom function factory
.with_function_factory(Arc::new(CustomFunctionFactory::default()));
// With the function factory, we can now call `CREATE FUNCTION` SQL functions
// Let us register a function called f which takes a single argument and
// returns that value plus one
let sql = r#"
CREATE FUNCTION f1(BIGINT)
RETURNS BIGINT
RETURN $1 + 1
"#;
ctx.sql(sql).await?.show().await?;
// Now, let us register a function called f2 which takes two arguments and
// returns the first argument added to the result of calling f1 on that
// argument
let sql = r#"
CREATE FUNCTION f2(BIGINT, BIGINT)
RETURNS BIGINT
RETURN $1 + f1($2)
"#;
ctx.sql(sql).await?.show().await?;
// Invoke f2, and we expect to see 1 + (1 + 2) = 4
// Note this function works on columns as well as constants.
let sql = r#"
SELECT f2(1, 2)
"#;
ctx.sql(sql).await?.show().await?;
// Now we clean up the session by dropping the functions
ctx.sql("DROP FUNCTION f1").await?.show().await?;
ctx.sql("DROP FUNCTION f2").await?.show().await?;
Ok(())
}
/// This is our FunctionFactory that is responsible for converting `CREATE
/// FUNCTION` statements into function instances
#[derive(Debug, Default)]
struct CustomFunctionFactory {}
#[async_trait::async_trait]
impl FunctionFactory for CustomFunctionFactory {
/// This function takes the parsed `CREATE FUNCTION` statement and returns
/// the function instance.
async fn create(
&self,
_state: &SessionState,
statement: CreateFunction,
) -> Result<RegisterFunction> {
let f: ScalarFunctionWrapper = statement.try_into()?;
Ok(RegisterFunction::Scalar(Arc::new(ScalarUDF::from(f))))
}
}
/// this function represents the newly created execution engine.
#[derive(Debug, PartialEq, Eq, Hash)]
struct ScalarFunctionWrapper {
/// The text of the function body, `$1 + f1($2)` in our example
name: String,
expr: Expr,
signature: Signature,
return_type: DataType,
}
impl ScalarUDFImpl for ScalarFunctionWrapper {
fn as_any(&self) -> &dyn std::any::Any {
self
}
fn name(&self) -> &str {
&self.name
}
fn signature(&self) -> &Signature {
&self.signature
}
fn return_type(&self, _arg_types: &[DataType]) -> Result<DataType> {
Ok(self.return_type.clone())
}
fn invoke_with_args(&self, _args: ScalarFunctionArgs) -> Result<ColumnarValue> {
// Since this function is always simplified to another expression, it
// should never actually be invoked
internal_err!("This function should not get invoked!")
}
/// The simplify function is called to simply a call such as `f2(2)`. This
/// function parses the string and returns the resulting expression
fn simplify(
&self,
args: Vec<Expr>,
_info: &SimplifyContext,
) -> Result<ExprSimplifyResult> {
let replacement = Self::replacement(&self.expr, &args)?;
Ok(ExprSimplifyResult::Simplified(replacement))
}
fn output_ordering(&self, _input: &[ExprProperties]) -> Result<SortProperties> {
Ok(SortProperties::Unordered)
}
}
impl ScalarFunctionWrapper {
// replaces placeholders such as $1 with actual arguments (args[0]
fn replacement(expr: &Expr, args: &[Expr]) -> Result<Expr> {
let result = expr.clone().transform(|e| {
let r = match e {
Expr::Placeholder(placeholder) => {
let placeholder_position =
Self::parse_placeholder_identifier(&placeholder.id)?;
if placeholder_position < args.len() {
Transformed::yes(args[placeholder_position].clone())
} else {
exec_err!(
"Function argument {} not provided, argument missing!",
placeholder.id
)?
}
}
_ => Transformed::no(e),
};
Ok(r)
})?;
Ok(result.data)
}
// Finds placeholder identifier such as `$X` format where X >= 1
fn parse_placeholder_identifier(placeholder: &str) -> Result<usize> {
if let Some(value) = placeholder.strip_prefix('$') {
Ok(value.parse().map(|v: usize| v - 1).map_err(|e| {
exec_datafusion_err!("Placeholder `{placeholder}` parsing error: {e}!")
})?)
} else {
exec_err!("Placeholder should start with `$`!")
}
}
}
/// This impl block creates a scalar function from
/// a parsed `CREATE FUNCTION` statement (`CreateFunction`)
impl TryFrom<CreateFunction> for ScalarFunctionWrapper {
type Error = DataFusionError;
fn try_from(definition: CreateFunction) -> RResult<Self, Self::Error> {
Ok(Self {
name: definition.name,
expr: definition
.params
.function_body
.expect("Expression has to be defined!"),
return_type: definition
.return_type
.expect("Return type has to be defined!"),
signature: Signature::exact(
definition
.args
.unwrap_or_default()
.into_iter()
.map(|a| a.data_type)
.collect(),
definition.params.behavior.unwrap_or(Volatility::Volatile),
),
})
}
}