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 std::any::Any;
21use std::fmt::Debug;
22
23use arrow::datatypes::DataType::{Float64, UInt64};
24use arrow::datatypes::{DataType, Field};
25
26use datafusion_common::{not_impl_err, plan_err, Result};
27use datafusion_expr::function::{AccumulatorArgs, StateFieldsArgs};
28use datafusion_expr::type_coercion::aggregates::NUMERICS;
29use datafusion_expr::utils::format_state_name;
30use datafusion_expr::{
31    Accumulator, AggregateUDFImpl, Documentation, Signature, Volatility,
32};
33use datafusion_macros::user_doc;
34
35use crate::approx_percentile_cont::ApproxPercentileAccumulator;
36
37make_udaf_expr_and_func!(
38    ApproxMedian,
39    approx_median,
40    expression,
41    "Computes the approximate median of a set of numbers",
42    approx_median_udaf
43);
44
45/// APPROX_MEDIAN aggregate expression
46#[user_doc(
47    doc_section(label = "Approximate Functions"),
48    description = "Returns the approximate median (50th percentile) of input values. It is an alias of `approx_percentile_cont(x, 0.5)`.",
49    syntax_example = "approx_median(expression)",
50    sql_example = r#"```sql
51> SELECT approx_median(column_name) FROM table_name;
52+-----------------------------------+
53| approx_median(column_name)        |
54+-----------------------------------+
55| 23.5                              |
56+-----------------------------------+
57```"#,
58    standard_argument(name = "expression",)
59)]
60pub struct ApproxMedian {
61    signature: Signature,
62}
63
64impl Debug for ApproxMedian {
65    fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
66        f.debug_struct("ApproxMedian")
67            .field("name", &self.name())
68            .field("signature", &self.signature)
69            .finish()
70    }
71}
72
73impl Default for ApproxMedian {
74    fn default() -> Self {
75        Self::new()
76    }
77}
78
79impl ApproxMedian {
80    /// Create a new APPROX_MEDIAN aggregate function
81    pub fn new() -> Self {
82        Self {
83            signature: Signature::uniform(1, NUMERICS.to_vec(), Volatility::Immutable),
84        }
85    }
86}
87
88impl AggregateUDFImpl for ApproxMedian {
89    /// Return a reference to Any that can be used for downcasting
90    fn as_any(&self) -> &dyn Any {
91        self
92    }
93
94    fn state_fields(&self, args: StateFieldsArgs) -> Result<Vec<Field>> {
95        Ok(vec![
96            Field::new(format_state_name(args.name, "max_size"), UInt64, false),
97            Field::new(format_state_name(args.name, "sum"), Float64, false),
98            Field::new(format_state_name(args.name, "count"), UInt64, false),
99            Field::new(format_state_name(args.name, "max"), Float64, false),
100            Field::new(format_state_name(args.name, "min"), Float64, false),
101            Field::new_list(
102                format_state_name(args.name, "centroids"),
103                Field::new_list_field(Float64, true),
104                false,
105            ),
106        ])
107    }
108
109    fn name(&self) -> &str {
110        "approx_median"
111    }
112
113    fn signature(&self) -> &Signature {
114        &self.signature
115    }
116
117    fn return_type(&self, arg_types: &[DataType]) -> Result<DataType> {
118        if !arg_types[0].is_numeric() {
119            return plan_err!("ApproxMedian requires numeric input types");
120        }
121        Ok(arg_types[0].clone())
122    }
123
124    fn accumulator(&self, acc_args: AccumulatorArgs) -> Result<Box<dyn Accumulator>> {
125        if acc_args.is_distinct {
126            return not_impl_err!(
127                "APPROX_MEDIAN(DISTINCT) aggregations are not available"
128            );
129        }
130
131        Ok(Box::new(ApproxPercentileAccumulator::new(
132            0.5_f64,
133            acc_args.exprs[0].data_type(acc_args.schema)?,
134        )))
135    }
136
137    fn documentation(&self) -> Option<&Documentation> {
138        self.doc()
139    }
140}
datafusion_functions_aggregate/approx_median.rs

datafusion_functions_aggregate/
approx_median.rs
