datafusion_expr/

udaf.rs

// 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.

//! [`AggregateUDF`]: User Defined Aggregate Functions

use std::any::Any;
use std::cmp::Ordering;
use std::fmt::{self, Debug, Formatter, Write};
use std::hash::{DefaultHasher, Hash, Hasher};
use std::sync::Arc;
use std::vec;

use arrow::datatypes::{DataType, Field};

use datafusion_common::{exec_err, not_impl_err, Result, ScalarValue, Statistics};
use datafusion_physical_expr_common::physical_expr::PhysicalExpr;

use crate::expr::{
    schema_name_from_exprs, schema_name_from_exprs_comma_separated_without_space,
    schema_name_from_sorts, AggregateFunction, AggregateFunctionParams,
    WindowFunctionParams,
};
use crate::function::{
    AccumulatorArgs, AggregateFunctionSimplification, StateFieldsArgs,
};
use crate::groups_accumulator::GroupsAccumulator;
use crate::utils::format_state_name;
use crate::utils::AggregateOrderSensitivity;
use crate::{expr_vec_fmt, Accumulator, Expr};
use crate::{Documentation, Signature};

/// Logical representation of a user-defined [aggregate function] (UDAF).
///
/// An aggregate function combines the values from multiple input rows
/// into a single output "aggregate" (summary) row. It is different
/// from a scalar function because it is stateful across batches. User
/// defined aggregate functions can be used as normal SQL aggregate
/// functions (`GROUP BY` clause) as well as window functions (`OVER`
/// clause).
///
/// `AggregateUDF` provides DataFusion the information needed to plan and call
/// aggregate functions, including name, type information, and a factory
/// function to create an [`Accumulator`] instance, to perform the actual
/// aggregation.
///
/// For more information, please see [the examples]:
///
/// 1. For simple use cases, use [`create_udaf`] (examples in [`simple_udaf.rs`]).
///
/// 2. For advanced use cases, use [`AggregateUDFImpl`] which provides full API
///    access (examples in [`advanced_udaf.rs`]).
///
/// # API Note
/// This is a separate struct from `AggregateUDFImpl` to maintain backwards
/// compatibility with the older API.
///
/// [the examples]: https://github.com/apache/datafusion/tree/main/datafusion-examples#single-process
/// [aggregate function]: https://en.wikipedia.org/wiki/Aggregate_function
/// [`Accumulator`]: crate::Accumulator
/// [`create_udaf`]: crate::expr_fn::create_udaf
/// [`simple_udaf.rs`]: https://github.com/apache/datafusion/blob/main/datafusion-examples/examples/simple_udaf.rs
/// [`advanced_udaf.rs`]: https://github.com/apache/datafusion/blob/main/datafusion-examples/examples/advanced_udaf.rs
#[derive(Debug, Clone, PartialOrd)]
pub struct AggregateUDF {
    inner: Arc<dyn AggregateUDFImpl>,
}

impl PartialEq for AggregateUDF {
    fn eq(&self, other: &Self) -> bool {
        self.inner.equals(other.inner.as_ref())
    }
}

impl Eq for AggregateUDF {}

impl Hash for AggregateUDF {
    fn hash<H: Hasher>(&self, state: &mut H) {
        self.inner.hash_value().hash(state)
    }
}

impl fmt::Display for AggregateUDF {
    fn fmt(&self, f: &mut Formatter) -> fmt::Result {
        write!(f, "{}", self.name())
    }
}

/// Arguments passed to [`AggregateUDFImpl::value_from_stats`]
#[derive(Debug)]
pub struct StatisticsArgs<'a> {
    /// The statistics of the aggregate input
    pub statistics: &'a Statistics,
    /// The resolved return type of the aggregate function
    pub return_type: &'a DataType,
    /// Whether the aggregate function is distinct.
    ///
    /// ```sql
    /// SELECT COUNT(DISTINCT column1) FROM t;
    /// ```
    pub is_distinct: bool,
    /// The physical expression of arguments the aggregate function takes.
    pub exprs: &'a [Arc<dyn PhysicalExpr>],
}

impl AggregateUDF {
    /// Create a new `AggregateUDF` from a `[AggregateUDFImpl]` trait object
    ///
    /// Note this is the same as using the `From` impl (`AggregateUDF::from`)
    pub fn new_from_impl<F>(fun: F) -> AggregateUDF
    where
        F: AggregateUDFImpl + 'static,
    {
        Self::new_from_shared_impl(Arc::new(fun))
    }

    /// Create a new `AggregateUDF` from a `[AggregateUDFImpl]` trait object
    pub fn new_from_shared_impl(fun: Arc<dyn AggregateUDFImpl>) -> AggregateUDF {
        Self { inner: fun }
    }

    /// Return the underlying [`AggregateUDFImpl`] trait object for this function
    pub fn inner(&self) -> &Arc<dyn AggregateUDFImpl> {
        &self.inner
    }

    /// Adds additional names that can be used to invoke this function, in
    /// addition to `name`
    ///
    /// If you implement [`AggregateUDFImpl`] directly you should return aliases directly.
    pub fn with_aliases(self, aliases: impl IntoIterator<Item = &'static str>) -> Self {
        Self::new_from_impl(AliasedAggregateUDFImpl::new(
            Arc::clone(&self.inner),
            aliases,
        ))
    }

    /// Creates an [`Expr`] that calls the aggregate function.
    ///
    /// This utility allows using the UDAF without requiring access to
    /// the registry, such as with the DataFrame API.
    pub fn call(&self, args: Vec<Expr>) -> Expr {
        Expr::AggregateFunction(AggregateFunction::new_udf(
            Arc::new(self.clone()),
            args,
            false,
            None,
            None,
            None,
        ))
    }

    /// Returns this function's name
    ///
    /// See [`AggregateUDFImpl::name`] for more details.
    pub fn name(&self) -> &str {
        self.inner.name()
    }

    /// See [`AggregateUDFImpl::schema_name`] for more details.
    pub fn schema_name(&self, params: &AggregateFunctionParams) -> Result<String> {
        self.inner.schema_name(params)
    }

    pub fn window_function_schema_name(
        &self,
        params: &WindowFunctionParams,
    ) -> Result<String> {
        self.inner.window_function_schema_name(params)
    }

    /// See [`AggregateUDFImpl::display_name`] for more details.
    pub fn display_name(&self, params: &AggregateFunctionParams) -> Result<String> {
        self.inner.display_name(params)
    }

    pub fn window_function_display_name(
        &self,
        params: &WindowFunctionParams,
    ) -> Result<String> {
        self.inner.window_function_display_name(params)
    }

    pub fn is_nullable(&self) -> bool {
        self.inner.is_nullable()
    }

    /// Returns the aliases for this function.
    pub fn aliases(&self) -> &[String] {
        self.inner.aliases()
    }

    /// Returns this function's signature (what input types are accepted)
    ///
    /// See [`AggregateUDFImpl::signature`] for more details.
    pub fn signature(&self) -> &Signature {
        self.inner.signature()
    }

    /// Return the type of the function given its input types
    ///
    /// See [`AggregateUDFImpl::return_type`] for more details.
    pub fn return_type(&self, args: &[DataType]) -> Result<DataType> {
        self.inner.return_type(args)
    }

    /// Return an accumulator the given aggregate, given its return datatype
    pub fn accumulator(&self, acc_args: AccumulatorArgs) -> Result<Box<dyn Accumulator>> {
        self.inner.accumulator(acc_args)
    }

    /// Return the fields used to store the intermediate state for this aggregator, given
    /// the name of the aggregate, value type and ordering fields. See [`AggregateUDFImpl::state_fields`]
    /// for more details.
    ///
    /// This is used to support multi-phase aggregations
    pub fn state_fields(&self, args: StateFieldsArgs) -> Result<Vec<Field>> {
        self.inner.state_fields(args)
    }

    /// See [`AggregateUDFImpl::groups_accumulator_supported`] for more details.
    pub fn groups_accumulator_supported(&self, args: AccumulatorArgs) -> bool {
        self.inner.groups_accumulator_supported(args)
    }

    /// See [`AggregateUDFImpl::create_groups_accumulator`] for more details.
    pub fn create_groups_accumulator(
        &self,
        args: AccumulatorArgs,
    ) -> Result<Box<dyn GroupsAccumulator>> {
        self.inner.create_groups_accumulator(args)
    }

    pub fn create_sliding_accumulator(
        &self,
        args: AccumulatorArgs,
    ) -> Result<Box<dyn Accumulator>> {
        self.inner.create_sliding_accumulator(args)
    }

    pub fn coerce_types(&self, arg_types: &[DataType]) -> Result<Vec<DataType>> {
        self.inner.coerce_types(arg_types)
    }

    /// See [`AggregateUDFImpl::with_beneficial_ordering`] for more details.
    pub fn with_beneficial_ordering(
        self,
        beneficial_ordering: bool,
    ) -> Result<Option<AggregateUDF>> {
        self.inner
            .with_beneficial_ordering(beneficial_ordering)
            .map(|updated_udf| updated_udf.map(|udf| Self { inner: udf }))
    }

    /// Gets the order sensitivity of the UDF. See [`AggregateOrderSensitivity`]
    /// for possible options.
    pub fn order_sensitivity(&self) -> AggregateOrderSensitivity {
        self.inner.order_sensitivity()
    }

    /// Reserves the `AggregateUDF` (e.g. returns the `AggregateUDF` that will
    /// generate same result with this `AggregateUDF` when iterated in reverse
    /// order, and `None` if there is no such `AggregateUDF`).
    pub fn reverse_udf(&self) -> ReversedUDAF {
        self.inner.reverse_expr()
    }

    /// Do the function rewrite
    ///
    /// See [`AggregateUDFImpl::simplify`] for more details.
    pub fn simplify(&self) -> Option<AggregateFunctionSimplification> {
        self.inner.simplify()
    }

    /// Returns true if the function is max, false if the function is min
    /// None in all other cases, used in certain optimizations for
    /// or aggregate
    pub fn is_descending(&self) -> Option<bool> {
        self.inner.is_descending()
    }

    /// Return the value of this aggregate function if it can be determined
    /// entirely from statistics and arguments.
    ///
    /// See [`AggregateUDFImpl::value_from_stats`] for more details.
    pub fn value_from_stats(
        &self,
        statistics_args: &StatisticsArgs,
    ) -> Option<ScalarValue> {
        self.inner.value_from_stats(statistics_args)
    }

    /// See [`AggregateUDFImpl::default_value`] for more details.
    pub fn default_value(&self, data_type: &DataType) -> Result<ScalarValue> {
        self.inner.default_value(data_type)
    }

    /// Returns the documentation for this Aggregate UDF.
    ///
    /// Documentation can be accessed programmatically as well as
    /// generating publicly facing documentation.
    pub fn documentation(&self) -> Option<&Documentation> {
        self.inner.documentation()
    }
}

impl<F> From<F> for AggregateUDF
where
    F: AggregateUDFImpl + Send + Sync + 'static,
{
    fn from(fun: F) -> Self {
        Self::new_from_impl(fun)
    }
}

/// Trait for implementing [`AggregateUDF`].
///
/// This trait exposes the full API for implementing user defined aggregate functions and
/// can be used to implement any function.
///
/// See [`advanced_udaf.rs`] for a full example with complete implementation and
/// [`AggregateUDF`] for other available options.
///
/// [`advanced_udaf.rs`]: https://github.com/apache/datafusion/blob/main/datafusion-examples/examples/advanced_udaf.rs
///
/// # Basic Example
/// ```
/// # use std::any::Any;
/// # use std::sync::LazyLock;
/// # use arrow::datatypes::DataType;
/// # use datafusion_common::{DataFusionError, plan_err, Result};
/// # use datafusion_expr::{col, ColumnarValue, Signature, Volatility, Expr, Documentation};
/// # use datafusion_expr::{AggregateUDFImpl, AggregateUDF, Accumulator, function::{AccumulatorArgs, StateFieldsArgs}};
/// # use datafusion_expr::window_doc_sections::DOC_SECTION_AGGREGATE;
/// # use arrow::datatypes::Schema;
/// # use arrow::datatypes::Field;
///
/// #[derive(Debug, Clone)]
/// struct GeoMeanUdf {
///   signature: Signature,
/// }
///
/// impl GeoMeanUdf {
///   fn new() -> Self {
///     Self {
///       signature: Signature::uniform(1, vec![DataType::Float64], Volatility::Immutable),
///      }
///   }
/// }
///
/// static DOCUMENTATION: LazyLock<Documentation> = LazyLock::new(|| {
///         Documentation::builder(DOC_SECTION_AGGREGATE, "calculates a geometric mean", "geo_mean(2.0)")
///             .with_argument("arg1", "The Float64 number for the geometric mean")
///             .build()
///     });
///
/// fn get_doc() -> &'static Documentation {
///     &DOCUMENTATION
/// }
///    
/// /// Implement the AggregateUDFImpl trait for GeoMeanUdf
/// impl AggregateUDFImpl for GeoMeanUdf {
///    fn as_any(&self) -> &dyn Any { self }
///    fn name(&self) -> &str { "geo_mean" }
///    fn signature(&self) -> &Signature { &self.signature }
///    fn return_type(&self, args: &[DataType]) -> Result<DataType> {
///      if !matches!(args.get(0), Some(&DataType::Float64)) {
///        return plan_err!("geo_mean only accepts Float64 arguments");
///      }
///      Ok(DataType::Float64)
///    }
///    // This is the accumulator factory; DataFusion uses it to create new accumulators.
///    fn accumulator(&self, _acc_args: AccumulatorArgs) -> Result<Box<dyn Accumulator>> { unimplemented!() }
///    fn state_fields(&self, args: StateFieldsArgs) -> Result<Vec<Field>> {
///        Ok(vec![
///             Field::new("value", args.return_type.clone(), true),
///             Field::new("ordering", DataType::UInt32, true)
///        ])
///    }
///    fn documentation(&self) -> Option<&Documentation> {
///        Some(get_doc())  
///    }
/// }
///
/// // Create a new AggregateUDF from the implementation
/// let geometric_mean = AggregateUDF::from(GeoMeanUdf::new());
///
/// // Call the function `geo_mean(col)`
/// let expr = geometric_mean.call(vec![col("a")]);
/// ```
pub trait AggregateUDFImpl: Debug + Send + Sync {
    // Note: When adding any methods (with default implementations), remember to add them also
    // into the AliasedAggregateUDFImpl below!

    /// Returns this object as an [`Any`] trait object
    fn as_any(&self) -> &dyn Any;

    /// Returns this function's name
    fn name(&self) -> &str;

    /// Returns the name of the column this expression would create
    ///
    /// See [`Expr::schema_name`] for details
    ///
    /// Example of schema_name: count(DISTINCT column1) FILTER (WHERE column2 > 10) ORDER BY [..]
    fn schema_name(&self, params: &AggregateFunctionParams) -> Result<String> {
        let AggregateFunctionParams {
            args,
            distinct,
            filter,
            order_by,
            null_treatment,
        } = params;

        let mut schema_name = String::new();

        schema_name.write_fmt(format_args!(
            "{}({}{})",
            self.name(),
            if *distinct { "DISTINCT " } else { "" },
            schema_name_from_exprs_comma_separated_without_space(args)?
        ))?;

        if let Some(null_treatment) = null_treatment {
            schema_name.write_fmt(format_args!(" {}", null_treatment))?;
        }

        if let Some(filter) = filter {
            schema_name.write_fmt(format_args!(" FILTER (WHERE {filter})"))?;
        };

        if let Some(order_by) = order_by {
            schema_name.write_fmt(format_args!(
                " ORDER BY [{}]",
                schema_name_from_sorts(order_by)?
            ))?;
        };

        Ok(schema_name)
    }

    /// Returns the name of the column this expression would create
    ///
    /// See [`Expr::schema_name`] for details
    ///
    /// Different from `schema_name` in that it is used for window aggregate function
    ///
    /// Example of schema_name: count(DISTINCT column1) FILTER (WHERE column2 > 10) [PARTITION BY [..]] [ORDER BY [..]]
    fn window_function_schema_name(
        &self,
        params: &WindowFunctionParams,
    ) -> Result<String> {
        let WindowFunctionParams {
            args,
            partition_by,
            order_by,
            window_frame,
            null_treatment,
        } = params;

        let mut schema_name = String::new();
        schema_name.write_fmt(format_args!(
            "{}({})",
            self.name(),
            schema_name_from_exprs(args)?
        ))?;

        if let Some(null_treatment) = null_treatment {
            schema_name.write_fmt(format_args!(" {}", null_treatment))?;
        }

        if !partition_by.is_empty() {
            schema_name.write_fmt(format_args!(
                " PARTITION BY [{}]",
                schema_name_from_exprs(partition_by)?
            ))?;
        }

        if !order_by.is_empty() {
            schema_name.write_fmt(format_args!(
                " ORDER BY [{}]",
                schema_name_from_sorts(order_by)?
            ))?;
        };

        schema_name.write_fmt(format_args!(" {window_frame}"))?;

        Ok(schema_name)
    }

    /// Returns the user-defined display name of function, given the arguments
    ///
    /// This can be used to customize the output column name generated by this
    /// function.
    ///
    /// Defaults to `function_name([DISTINCT] column1, column2, ..) [null_treatment] [filter] [order_by [..]]`
    fn display_name(&self, params: &AggregateFunctionParams) -> Result<String> {
        let AggregateFunctionParams {
            args,
            distinct,
            filter,
            order_by,
            null_treatment,
        } = params;

        let mut display_name = String::new();

        display_name.write_fmt(format_args!(
            "{}({}{})",
            self.name(),
            if *distinct { "DISTINCT " } else { "" },
            expr_vec_fmt!(args)
        ))?;

        if let Some(nt) = null_treatment {
            display_name.write_fmt(format_args!(" {}", nt))?;
        }
        if let Some(fe) = filter {
            display_name.write_fmt(format_args!(" FILTER (WHERE {fe})"))?;
        }
        if let Some(ob) = order_by {
            display_name.write_fmt(format_args!(
                " ORDER BY [{}]",
                ob.iter()
                    .map(|o| format!("{o}"))
                    .collect::<Vec<String>>()
                    .join(", ")
            ))?;
        }

        Ok(display_name)
    }

    /// Returns the user-defined display name of function, given the arguments
    ///
    /// This can be used to customize the output column name generated by this
    /// function.
    ///
    /// Different from `display_name` in that it is used for window aggregate function
    ///
    /// Defaults to `function_name([DISTINCT] column1, column2, ..) [null_treatment] [partition by [..]] [order_by [..]]`
    fn window_function_display_name(
        &self,
        params: &WindowFunctionParams,
    ) -> Result<String> {
        let WindowFunctionParams {
            args,
            partition_by,
            order_by,
            window_frame,
            null_treatment,
        } = params;

        let mut display_name = String::new();

        display_name.write_fmt(format_args!(
            "{}({})",
            self.name(),
            expr_vec_fmt!(args)
        ))?;

        if let Some(null_treatment) = null_treatment {
            display_name.write_fmt(format_args!(" {}", null_treatment))?;
        }

        if !partition_by.is_empty() {
            display_name.write_fmt(format_args!(
                " PARTITION BY [{}]",
                expr_vec_fmt!(partition_by)
            ))?;
        }

        if !order_by.is_empty() {
            display_name
                .write_fmt(format_args!(" ORDER BY [{}]", expr_vec_fmt!(order_by)))?;
        };

        display_name.write_fmt(format_args!(
            " {} BETWEEN {} AND {}",
            window_frame.units, window_frame.start_bound, window_frame.end_bound
        ))?;

        Ok(display_name)
    }

    /// Returns the function's [`Signature`] for information about what input
    /// types are accepted and the function's Volatility.
    fn signature(&self) -> &Signature;

    /// What [`DataType`] will be returned by this function, given the types of
    /// the arguments
    fn return_type(&self, arg_types: &[DataType]) -> Result<DataType>;

    /// Whether the aggregate function is nullable.
    ///
    /// Nullable means that the function could return `null` for any inputs.
    /// For example, aggregate functions like `COUNT` always return a non null value
    /// but others like `MIN` will return `NULL` if there is nullable input.
    /// Note that if the function is declared as *not* nullable, make sure the [`AggregateUDFImpl::default_value`] is `non-null`
    fn is_nullable(&self) -> bool {
        true
    }

    /// Return a new [`Accumulator`] that aggregates values for a specific
    /// group during query execution.
    ///
    /// acc_args: [`AccumulatorArgs`] contains information about how the
    /// aggregate function was called.
    fn accumulator(&self, acc_args: AccumulatorArgs) -> Result<Box<dyn Accumulator>>;

    /// Return the fields used to store the intermediate state of this accumulator.
    ///
    /// See [`Accumulator::state`] for background information.
    ///
    /// args:  [`StateFieldsArgs`] contains arguments passed to the
    /// aggregate function's accumulator.
    ///
    /// # Notes:
    ///
    /// The default implementation returns a single state field named `name`
    /// with the same type as `value_type`. This is suitable for aggregates such
    /// as `SUM` or `MIN` where partial state can be combined by applying the
    /// same aggregate.
    ///
    /// For aggregates such as `AVG` where the partial state is more complex
    /// (e.g. a COUNT and a SUM), this method is used to define the additional
    /// fields.
    ///
    /// The name of the fields must be unique within the query and thus should
    /// be derived from `name`. See [`format_state_name`] for a utility function
    /// to generate a unique name.
    fn state_fields(&self, args: StateFieldsArgs) -> Result<Vec<Field>> {
        let fields = vec![Field::new(
            format_state_name(args.name, "value"),
            args.return_type.clone(),
            true,
        )];

        Ok(fields
            .into_iter()
            .chain(args.ordering_fields.to_vec())
            .collect())
    }

    /// If the aggregate expression has a specialized
    /// [`GroupsAccumulator`] implementation. If this returns true,
    /// `[Self::create_groups_accumulator]` will be called.
    ///
    /// # Notes
    ///
    /// Even if this function returns true, DataFusion will still use
    /// [`Self::accumulator`] for certain queries, such as when this aggregate is
    /// used as a window function or when there no GROUP BY columns in the
    /// query.
    fn groups_accumulator_supported(&self, _args: AccumulatorArgs) -> bool {
        false
    }

    /// Return a specialized [`GroupsAccumulator`] that manages state
    /// for all groups.
    ///
    /// For maximum performance, a [`GroupsAccumulator`] should be
    /// implemented in addition to [`Accumulator`].
    fn create_groups_accumulator(
        &self,
        _args: AccumulatorArgs,
    ) -> Result<Box<dyn GroupsAccumulator>> {
        not_impl_err!("GroupsAccumulator hasn't been implemented for {self:?} yet")
    }

    /// Returns any aliases (alternate names) for this function.
    ///
    /// Note: `aliases` should only include names other than [`Self::name`].
    /// Defaults to `[]` (no aliases)
    fn aliases(&self) -> &[String] {
        &[]
    }

    /// Sliding accumulator is an alternative accumulator that can be used for
    /// window functions. It has retract method to revert the previous update.
    ///
    /// See [retract_batch] for more details.
    ///
    /// [retract_batch]: datafusion_expr_common::accumulator::Accumulator::retract_batch
    fn create_sliding_accumulator(
        &self,
        args: AccumulatorArgs,
    ) -> Result<Box<dyn Accumulator>> {
        self.accumulator(args)
    }

    /// Sets the indicator whether ordering requirements of the AggregateUDFImpl is
    /// satisfied by its input. If this is not the case, UDFs with order
    /// sensitivity `AggregateOrderSensitivity::Beneficial` can still produce
    /// the correct result with possibly more work internally.
    ///
    /// # Returns
    ///
    /// Returns `Ok(Some(updated_udf))` if the process completes successfully.
    /// If the expression can benefit from existing input ordering, but does
    /// not implement the method, returns an error. Order insensitive and hard
    /// requirement aggregators return `Ok(None)`.
    fn with_beneficial_ordering(
        self: Arc<Self>,
        _beneficial_ordering: bool,
    ) -> Result<Option<Arc<dyn AggregateUDFImpl>>> {
        if self.order_sensitivity().is_beneficial() {
            return exec_err!(
                "Should implement with satisfied for aggregator :{:?}",
                self.name()
            );
        }
        Ok(None)
    }

    /// Gets the order sensitivity of the UDF. See [`AggregateOrderSensitivity`]
    /// for possible options.
    fn order_sensitivity(&self) -> AggregateOrderSensitivity {
        // We have hard ordering requirements by default, meaning that order
        // sensitive UDFs need their input orderings to satisfy their ordering
        // requirements to generate correct results.
        AggregateOrderSensitivity::HardRequirement
    }

    /// Optionally apply per-UDaF simplification / rewrite rules.
    ///
    /// This can be used to apply function specific simplification rules during
    /// optimization (e.g. `arrow_cast` --> `Expr::Cast`). The default
    /// implementation does nothing.
    ///
    /// Note that DataFusion handles simplifying arguments and  "constant
    /// folding" (replacing a function call with constant arguments such as
    /// `my_add(1,2) --> 3` ). Thus, there is no need to implement such
    /// optimizations manually for specific UDFs.
    ///
    /// # Returns
    ///
    /// [None] if simplify is not defined or,
    ///
    /// Or, a closure with two arguments:
    /// * 'aggregate_function': [crate::expr::AggregateFunction] for which simplified has been invoked
    /// * 'info': [crate::simplify::SimplifyInfo]
    ///
    /// closure returns simplified [Expr] or an error.
    ///
    fn simplify(&self) -> Option<AggregateFunctionSimplification> {
        None
    }

    /// Returns the reverse expression of the aggregate function.
    fn reverse_expr(&self) -> ReversedUDAF {
        ReversedUDAF::NotSupported
    }

    /// Coerce arguments of a function call to types that the function can evaluate.
    ///
    /// This function is only called if [`AggregateUDFImpl::signature`] returns [`crate::TypeSignature::UserDefined`]. Most
    /// UDAFs should return one of the other variants of `TypeSignature` which handle common
    /// cases
    ///
    /// See the [type coercion module](crate::type_coercion)
    /// documentation for more details on type coercion
    ///
    /// For example, if your function requires a floating point arguments, but the user calls
    /// it like `my_func(1::int)` (aka with `1` as an integer), coerce_types could return `[DataType::Float64]`
    /// to ensure the argument was cast to `1::double`
    ///
    /// # Parameters
    /// * `arg_types`: The argument types of the arguments  this function with
    ///
    /// # Return value
    /// A Vec the same length as `arg_types`. DataFusion will `CAST` the function call
    /// arguments to these specific types.
    fn coerce_types(&self, _arg_types: &[DataType]) -> Result<Vec<DataType>> {
        not_impl_err!("Function {} does not implement coerce_types", self.name())
    }

    /// Return true if this aggregate UDF is equal to the other.
    ///
    /// Allows customizing the equality of aggregate UDFs.
    /// Must be consistent with [`Self::hash_value`] and follow the same rules as [`Eq`]:
    ///
    /// - reflexive: `a.equals(a)`;
    /// - symmetric: `a.equals(b)` implies `b.equals(a)`;
    /// - transitive: `a.equals(b)` and `b.equals(c)` implies `a.equals(c)`.
    ///
    /// By default, compares [`Self::name`] and [`Self::signature`].
    fn equals(&self, other: &dyn AggregateUDFImpl) -> bool {
        self.name() == other.name() && self.signature() == other.signature()
    }

    /// Returns a hash value for this aggregate UDF.
    ///
    /// Allows customizing the hash code of aggregate UDFs. Similarly to [`Hash`] and [`Eq`],
    /// if [`Self::equals`] returns true for two UDFs, their `hash_value`s must be the same.
    ///
    /// By default, hashes [`Self::name`] and [`Self::signature`].
    fn hash_value(&self) -> u64 {
        let hasher = &mut DefaultHasher::new();
        self.name().hash(hasher);
        self.signature().hash(hasher);
        hasher.finish()
    }

    /// If this function is max, return true
    /// If the function is min, return false
    /// Otherwise return None (the default)
    ///
    ///
    /// Note: this is used to use special aggregate implementations in certain conditions
    fn is_descending(&self) -> Option<bool> {
        None
    }

    /// Return the value of this aggregate function if it can be determined
    /// entirely from statistics and arguments.
    ///
    /// Using a [`ScalarValue`] rather than a runtime computation can significantly
    /// improving query performance.
    ///
    /// For example, if the minimum value of column `x` is known to be `42` from
    /// statistics, then the aggregate `MIN(x)` should return `Some(ScalarValue(42))`
    fn value_from_stats(&self, _statistics_args: &StatisticsArgs) -> Option<ScalarValue> {
        None
    }

    /// Returns default value of the function given the input is all `null`.
    ///
    /// Most of the aggregate function return Null if input is Null,
    /// while `count` returns 0 if input is Null
    fn default_value(&self, data_type: &DataType) -> Result<ScalarValue> {
        ScalarValue::try_from(data_type)
    }

    /// Returns the documentation for this Aggregate UDF.
    ///
    /// Documentation can be accessed programmatically as well as
    /// generating publicly facing documentation.
    fn documentation(&self) -> Option<&Documentation> {
        None
    }

    /// Indicates whether the aggregation function is monotonic as a set
    /// function. See [`SetMonotonicity`] for details.
    fn set_monotonicity(&self, _data_type: &DataType) -> SetMonotonicity {
        SetMonotonicity::NotMonotonic
    }
}

impl PartialEq for dyn AggregateUDFImpl {
    fn eq(&self, other: &Self) -> bool {
        self.equals(other)
    }
}

// Manual implementation of `PartialOrd`
// There might be some wackiness with it, but this is based on the impl of eq for AggregateUDFImpl
// https://users.rust-lang.org/t/how-to-compare-two-trait-objects-for-equality/88063/5
impl PartialOrd for dyn AggregateUDFImpl {
    fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
        match self.name().partial_cmp(other.name()) {
            Some(Ordering::Equal) => self.signature().partial_cmp(other.signature()),
            cmp => cmp,
        }
    }
}

pub enum ReversedUDAF {
    /// The expression is the same as the original expression, like SUM, COUNT
    Identical,
    /// The expression does not support reverse calculation
    NotSupported,
    /// The expression is different from the original expression
    Reversed(Arc<AggregateUDF>),
}

/// AggregateUDF that adds an alias to the underlying function. It is better to
/// implement [`AggregateUDFImpl`], which supports aliases, directly if possible.
#[derive(Debug)]
struct AliasedAggregateUDFImpl {
    inner: Arc<dyn AggregateUDFImpl>,
    aliases: Vec<String>,
}

impl AliasedAggregateUDFImpl {
    pub fn new(
        inner: Arc<dyn AggregateUDFImpl>,
        new_aliases: impl IntoIterator<Item = &'static str>,
    ) -> Self {
        let mut aliases = inner.aliases().to_vec();
        aliases.extend(new_aliases.into_iter().map(|s| s.to_string()));

        Self { inner, aliases }
    }
}

impl AggregateUDFImpl for AliasedAggregateUDFImpl {
    fn as_any(&self) -> &dyn Any {
        self
    }

    fn name(&self) -> &str {
        self.inner.name()
    }

    fn signature(&self) -> &Signature {
        self.inner.signature()
    }

    fn return_type(&self, arg_types: &[DataType]) -> Result<DataType> {
        self.inner.return_type(arg_types)
    }

    fn accumulator(&self, acc_args: AccumulatorArgs) -> Result<Box<dyn Accumulator>> {
        self.inner.accumulator(acc_args)
    }

    fn aliases(&self) -> &[String] {
        &self.aliases
    }

    fn state_fields(&self, args: StateFieldsArgs) -> Result<Vec<Field>> {
        self.inner.state_fields(args)
    }

    fn groups_accumulator_supported(&self, args: AccumulatorArgs) -> bool {
        self.inner.groups_accumulator_supported(args)
    }

    fn create_groups_accumulator(
        &self,
        args: AccumulatorArgs,
    ) -> Result<Box<dyn GroupsAccumulator>> {
        self.inner.create_groups_accumulator(args)
    }

    fn create_sliding_accumulator(
        &self,
        args: AccumulatorArgs,
    ) -> Result<Box<dyn Accumulator>> {
        self.inner.accumulator(args)
    }

    fn with_beneficial_ordering(
        self: Arc<Self>,
        beneficial_ordering: bool,
    ) -> Result<Option<Arc<dyn AggregateUDFImpl>>> {
        Arc::clone(&self.inner)
            .with_beneficial_ordering(beneficial_ordering)
            .map(|udf| {
                udf.map(|udf| {
                    Arc::new(AliasedAggregateUDFImpl {
                        inner: udf,
                        aliases: self.aliases.clone(),
                    }) as Arc<dyn AggregateUDFImpl>
                })
            })
    }

    fn order_sensitivity(&self) -> AggregateOrderSensitivity {
        self.inner.order_sensitivity()
    }

    fn simplify(&self) -> Option<AggregateFunctionSimplification> {
        self.inner.simplify()
    }

    fn reverse_expr(&self) -> ReversedUDAF {
        self.inner.reverse_expr()
    }

    fn coerce_types(&self, arg_types: &[DataType]) -> Result<Vec<DataType>> {
        self.inner.coerce_types(arg_types)
    }

    fn equals(&self, other: &dyn AggregateUDFImpl) -> bool {
        if let Some(other) = other.as_any().downcast_ref::<AliasedAggregateUDFImpl>() {
            self.inner.equals(other.inner.as_ref()) && self.aliases == other.aliases
        } else {
            false
        }
    }

    fn hash_value(&self) -> u64 {
        let hasher = &mut DefaultHasher::new();
        self.inner.hash_value().hash(hasher);
        self.aliases.hash(hasher);
        hasher.finish()
    }

    fn is_descending(&self) -> Option<bool> {
        self.inner.is_descending()
    }

    fn documentation(&self) -> Option<&Documentation> {
        self.inner.documentation()
    }
}

// Aggregate UDF doc sections for use in public documentation
pub mod aggregate_doc_sections {
    use crate::DocSection;

    pub fn doc_sections() -> Vec<DocSection> {
        vec![
            DOC_SECTION_GENERAL,
            DOC_SECTION_STATISTICAL,
            DOC_SECTION_APPROXIMATE,
        ]
    }

    pub const DOC_SECTION_GENERAL: DocSection = DocSection {
        include: true,
        label: "General Functions",
        description: None,
    };

    pub const DOC_SECTION_STATISTICAL: DocSection = DocSection {
        include: true,
        label: "Statistical Functions",
        description: None,
    };

    pub const DOC_SECTION_APPROXIMATE: DocSection = DocSection {
        include: true,
        label: "Approximate Functions",
        description: None,
    };
}

/// Indicates whether an aggregation function is monotonic as a set
/// function. A set function is monotonically increasing if its value
/// increases as its argument grows (as a set). Formally, `f` is a
/// monotonically increasing set function if `f(S) >= f(T)` whenever `S`
/// is a superset of `T`.
///
/// For example `COUNT` and `MAX` are monotonically increasing as their
/// values always increase (or stay the same) as new values are seen. On
/// the other hand, `MIN` is monotonically decreasing as its value always
/// decreases or stays the same as new values are seen.
#[derive(Debug, Clone, PartialEq)]
pub enum SetMonotonicity {
    /// Aggregate value increases or stays the same as the input set grows.
    Increasing,
    /// Aggregate value decreases or stays the same as the input set grows.
    Decreasing,
    /// Aggregate value may increase, decrease, or stay the same as the input
    /// set grows.
    NotMonotonic,
}

#[cfg(test)]
mod test {
    use crate::{AggregateUDF, AggregateUDFImpl};
    use arrow::datatypes::{DataType, Field};
    use datafusion_common::Result;
    use datafusion_expr_common::accumulator::Accumulator;
    use datafusion_expr_common::signature::{Signature, Volatility};
    use datafusion_functions_aggregate_common::accumulator::{
        AccumulatorArgs, StateFieldsArgs,
    };
    use std::any::Any;
    use std::cmp::Ordering;

    #[derive(Debug, Clone)]
    struct AMeanUdf {
        signature: Signature,
    }

    impl AMeanUdf {
        fn new() -> Self {
            Self {
                signature: Signature::uniform(
                    1,
                    vec![DataType::Float64],
                    Volatility::Immutable,
                ),
            }
        }
    }

    impl AggregateUDFImpl for AMeanUdf {
        fn as_any(&self) -> &dyn Any {
            self
        }
        fn name(&self) -> &str {
            "a"
        }
        fn signature(&self) -> &Signature {
            &self.signature
        }
        fn return_type(&self, _args: &[DataType]) -> Result<DataType> {
            unimplemented!()
        }
        fn accumulator(
            &self,
            _acc_args: AccumulatorArgs,
        ) -> Result<Box<dyn Accumulator>> {
            unimplemented!()
        }
        fn state_fields(&self, _args: StateFieldsArgs) -> Result<Vec<Field>> {
            unimplemented!()
        }
    }

    #[derive(Debug, Clone)]
    struct BMeanUdf {
        signature: Signature,
    }
    impl BMeanUdf {
        fn new() -> Self {
            Self {
                signature: Signature::uniform(
                    1,
                    vec![DataType::Float64],
                    Volatility::Immutable,
                ),
            }
        }
    }

    impl AggregateUDFImpl for BMeanUdf {
        fn as_any(&self) -> &dyn Any {
            self
        }
        fn name(&self) -> &str {
            "b"
        }
        fn signature(&self) -> &Signature {
            &self.signature
        }
        fn return_type(&self, _args: &[DataType]) -> Result<DataType> {
            unimplemented!()
        }
        fn accumulator(
            &self,
            _acc_args: AccumulatorArgs,
        ) -> Result<Box<dyn Accumulator>> {
            unimplemented!()
        }
        fn state_fields(&self, _args: StateFieldsArgs) -> Result<Vec<Field>> {
            unimplemented!()
        }
    }

    #[test]
    fn test_partial_ord() {
        // Test validates that partial ord is defined for AggregateUDF using the name and signature,
        // not intended to exhaustively test all possibilities
        let a1 = AggregateUDF::from(AMeanUdf::new());
        let a2 = AggregateUDF::from(AMeanUdf::new());
        assert_eq!(a1.partial_cmp(&a2), Some(Ordering::Equal));

        let b1 = AggregateUDF::from(BMeanUdf::new());
        assert!(a1 < b1);
        assert!(!(a1 == b1));
    }
}