datafusion_functions_aggregate/
approx_median.rs

1// Licensed to the Apache Software Foundation (ASF) under one
2// or more contributor license agreements.  See the NOTICE file
3// distributed with this work for additional information
4// regarding copyright ownership.  The ASF licenses this file
5// to you under the Apache License, Version 2.0 (the
6// "License"); you may not use this file except in compliance
7// with the License.  You may obtain a copy of the License at
8//
9//   http://www.apache.org/licenses/LICENSE-2.0
10//
11// Unless required by applicable law or agreed to in writing,
12// software distributed under the License is distributed on an
13// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
14// KIND, either express or implied.  See the License for the
15// specific language governing permissions and limitations
16// under the License.
17
18//! Defines physical expressions for APPROX_MEDIAN that can be evaluated MEDIAN at runtime during query execution
19
20use arrow::datatypes::DataType::{Float64, UInt64};
21use arrow::datatypes::{DataType, Field, FieldRef};
22use datafusion_common::types::NativeType;
23use datafusion_functions_aggregate_common::noop_accumulator::NoopAccumulator;
24use std::any::Any;
25use std::fmt::Debug;
26use std::sync::Arc;
27
28use datafusion_common::{Result, not_impl_err};
29use datafusion_expr::function::{AccumulatorArgs, StateFieldsArgs};
30use datafusion_expr::utils::format_state_name;
31use datafusion_expr::{
32    Accumulator, AggregateUDFImpl, Coercion, Documentation, Signature, TypeSignature,
33    TypeSignatureClass, Volatility,
34};
35use datafusion_macros::user_doc;
36
37use crate::approx_percentile_cont::ApproxPercentileAccumulator;
38
39make_udaf_expr_and_func!(
40    ApproxMedian,
41    approx_median,
42    expression,
43    "Computes the approximate median of a set of numbers",
44    approx_median_udaf
45);
46
47/// APPROX_MEDIAN aggregate expression
48#[user_doc(
49    doc_section(label = "Approximate Functions"),
50    description = "Returns the approximate median (50th percentile) of input values. It is an alias of `approx_percentile_cont(0.5) WITHIN GROUP (ORDER BY x)`.",
51    syntax_example = "approx_median(expression)",
52    sql_example = r#"```sql
53> SELECT approx_median(column_name) FROM table_name;
54+-----------------------------------+
55| approx_median(column_name)        |
56+-----------------------------------+
57| 23.5                              |
58+-----------------------------------+
59```"#,
60    standard_argument(name = "expression",)
61)]
62#[derive(Debug, PartialEq, Eq, Hash)]
63pub struct ApproxMedian {
64    signature: Signature,
65}
66
67impl Default for ApproxMedian {
68    fn default() -> Self {
69        Self::new()
70    }
71}
72
73impl ApproxMedian {
74    /// Create a new APPROX_MEDIAN aggregate function
75    pub fn new() -> Self {
76        Self {
77            signature: Signature::one_of(
78                vec![
79                    TypeSignature::Coercible(vec![Coercion::new_exact(
80                        TypeSignatureClass::Integer,
81                    )]),
82                    TypeSignature::Coercible(vec![Coercion::new_implicit(
83                        TypeSignatureClass::Float,
84                        vec![TypeSignatureClass::Decimal],
85                        NativeType::Float64,
86                    )]),
87                ],
88                Volatility::Immutable,
89            ),
90        }
91    }
92}
93
94impl AggregateUDFImpl for ApproxMedian {
95    fn as_any(&self) -> &dyn Any {
96        self
97    }
98
99    fn state_fields(&self, args: StateFieldsArgs) -> Result<Vec<FieldRef>> {
100        if args.input_fields[0].data_type().is_null() {
101            Ok(vec![
102                Field::new(
103                    format_state_name(args.name, self.name()),
104                    DataType::Null,
105                    true,
106                )
107                .into(),
108            ])
109        } else {
110            Ok(vec![
111                Field::new(format_state_name(args.name, "max_size"), UInt64, false),
112                Field::new(format_state_name(args.name, "sum"), Float64, false),
113                Field::new(format_state_name(args.name, "count"), UInt64, false),
114                Field::new(format_state_name(args.name, "max"), Float64, false),
115                Field::new(format_state_name(args.name, "min"), Float64, false),
116                Field::new_list(
117                    format_state_name(args.name, "centroids"),
118                    Field::new_list_field(Float64, true),
119                    false,
120                ),
121            ]
122            .into_iter()
123            .map(Arc::new)
124            .collect())
125        }
126    }
127
128    fn name(&self) -> &str {
129        "approx_median"
130    }
131
132    fn signature(&self) -> &Signature {
133        &self.signature
134    }
135
136    fn return_type(&self, arg_types: &[DataType]) -> Result<DataType> {
137        Ok(arg_types[0].clone())
138    }
139
140    fn accumulator(&self, acc_args: AccumulatorArgs) -> Result<Box<dyn Accumulator>> {
141        if acc_args.is_distinct {
142            return not_impl_err!(
143                "APPROX_MEDIAN(DISTINCT) aggregations are not available"
144            );
145        }
146
147        if acc_args.expr_fields[0].data_type().is_null() {
148            Ok(Box::new(NoopAccumulator::default()))
149        } else {
150            Ok(Box::new(ApproxPercentileAccumulator::new(
151                0.5_f64,
152                acc_args.expr_fields[0].data_type().clone(),
153            )))
154        }
155    }
156
157    fn documentation(&self) -> Option<&Documentation> {
158        self.doc()
159    }
160}
datafusion_functions_aggregate/approx_median.rs

datafusion_functions_aggregate/
approx_median.rs
