|
| 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 | +use arrow::{ |
| 19 | + array::{ArrayRef, ArrowNumericType}, |
| 20 | + datatypes::{i256, Decimal128Type, Decimal256Type, DecimalType}, |
| 21 | +}; |
| 22 | +use datafusion_common::{Result, ScalarValue}; |
| 23 | +use datafusion_expr_common::accumulator::Accumulator; |
| 24 | +use std::fmt::Debug; |
| 25 | +use std::mem::size_of_val; |
| 26 | + |
| 27 | +use crate::aggregate::sum_distinct::DistinctSumAccumulator; |
| 28 | +use crate::utils::DecimalAverager; |
| 29 | + |
| 30 | +/// Generic implementation of `AVG DISTINCT` for Decimal types. |
| 31 | +/// Handles both Decimal128Type and Decimal256Type. |
| 32 | +#[derive(Debug)] |
| 33 | +pub struct DecimalDistinctAvgAccumulator<T: DecimalType + Debug> { |
| 34 | + sum_accumulator: DistinctSumAccumulator<T>, |
| 35 | + sum_scale: i8, |
| 36 | + target_precision: u8, |
| 37 | + target_scale: i8, |
| 38 | +} |
| 39 | + |
| 40 | +impl<T: DecimalType + Debug> DecimalDistinctAvgAccumulator<T> { |
| 41 | + pub fn with_decimal_params( |
| 42 | + sum_scale: i8, |
| 43 | + target_precision: u8, |
| 44 | + target_scale: i8, |
| 45 | + ) -> Self { |
| 46 | + let data_type = T::TYPE_CONSTRUCTOR(T::MAX_PRECISION, sum_scale); |
| 47 | + |
| 48 | + Self { |
| 49 | + sum_accumulator: DistinctSumAccumulator::new(&data_type), |
| 50 | + sum_scale, |
| 51 | + target_precision, |
| 52 | + target_scale, |
| 53 | + } |
| 54 | + } |
| 55 | +} |
| 56 | + |
| 57 | +impl<T: DecimalType + ArrowNumericType + Debug> Accumulator |
| 58 | + for DecimalDistinctAvgAccumulator<T> |
| 59 | +{ |
| 60 | + fn state(&mut self) -> Result<Vec<ScalarValue>> { |
| 61 | + self.sum_accumulator.state() |
| 62 | + } |
| 63 | + |
| 64 | + fn update_batch(&mut self, values: &[ArrayRef]) -> Result<()> { |
| 65 | + self.sum_accumulator.update_batch(values) |
| 66 | + } |
| 67 | + |
| 68 | + fn merge_batch(&mut self, states: &[ArrayRef]) -> Result<()> { |
| 69 | + self.sum_accumulator.merge_batch(states) |
| 70 | + } |
| 71 | + |
| 72 | + fn evaluate(&mut self) -> Result<ScalarValue> { |
| 73 | + if self.sum_accumulator.distinct_count() == 0 { |
| 74 | + return ScalarValue::new_primitive::<T>( |
| 75 | + None, |
| 76 | + &T::TYPE_CONSTRUCTOR(self.target_precision, self.target_scale), |
| 77 | + ); |
| 78 | + } |
| 79 | + |
| 80 | + let sum_scalar = self.sum_accumulator.evaluate()?; |
| 81 | + |
| 82 | + match sum_scalar { |
| 83 | + ScalarValue::Decimal128(Some(sum), _, _) => { |
| 84 | + let decimal_averager = DecimalAverager::<Decimal128Type>::try_new( |
| 85 | + self.sum_scale, |
| 86 | + self.target_precision, |
| 87 | + self.target_scale, |
| 88 | + )?; |
| 89 | + let avg = decimal_averager |
| 90 | + .avg(sum, self.sum_accumulator.distinct_count() as i128)?; |
| 91 | + Ok(ScalarValue::Decimal128( |
| 92 | + Some(avg), |
| 93 | + self.target_precision, |
| 94 | + self.target_scale, |
| 95 | + )) |
| 96 | + } |
| 97 | + ScalarValue::Decimal256(Some(sum), _, _) => { |
| 98 | + let decimal_averager = DecimalAverager::<Decimal256Type>::try_new( |
| 99 | + self.sum_scale, |
| 100 | + self.target_precision, |
| 101 | + self.target_scale, |
| 102 | + )?; |
| 103 | + // `distinct_count` returns `u64`, but `avg` expects `i256` |
| 104 | + // first convert `u64` to `i128`, then convert `i128` to `i256` to avoid overflow |
| 105 | + let distinct_cnt: i128 = self.sum_accumulator.distinct_count() as i128; |
| 106 | + let count: i256 = i256::from_i128(distinct_cnt); |
| 107 | + let avg = decimal_averager.avg(sum, count)?; |
| 108 | + Ok(ScalarValue::Decimal256( |
| 109 | + Some(avg), |
| 110 | + self.target_precision, |
| 111 | + self.target_scale, |
| 112 | + )) |
| 113 | + } |
| 114 | + |
| 115 | + _ => unreachable!("Unsupported decimal type: {:?}", sum_scalar), |
| 116 | + } |
| 117 | + } |
| 118 | + |
| 119 | + fn size(&self) -> usize { |
| 120 | + let fixed_size = size_of_val(self); |
| 121 | + |
| 122 | + // Account for the size of the sum_accumulator with its contained values |
| 123 | + fixed_size + self.sum_accumulator.size() |
| 124 | + } |
| 125 | +} |
| 126 | + |
| 127 | +#[cfg(test)] |
| 128 | +mod tests { |
| 129 | + use super::*; |
| 130 | + use arrow::array::{Decimal128Array, Decimal256Array}; |
| 131 | + use std::sync::Arc; |
| 132 | + |
| 133 | + #[test] |
| 134 | + fn test_decimal128_distinct_avg_accumulator() -> Result<()> { |
| 135 | + let precision = 10_u8; |
| 136 | + let scale = 4_i8; |
| 137 | + let array = Decimal128Array::from(vec![ |
| 138 | + Some(100_0000), |
| 139 | + Some(125_0000), |
| 140 | + Some(175_0000), |
| 141 | + Some(200_0000), |
| 142 | + Some(200_0000), |
| 143 | + Some(300_0000), |
| 144 | + None, |
| 145 | + None, |
| 146 | + ]) |
| 147 | + .with_precision_and_scale(precision, scale)?; |
| 148 | + |
| 149 | + let mut accumulator = |
| 150 | + DecimalDistinctAvgAccumulator::<Decimal128Type>::with_decimal_params( |
| 151 | + scale, 14, 8, |
| 152 | + ); |
| 153 | + accumulator.update_batch(&[Arc::new(array)])?; |
| 154 | + |
| 155 | + let result = accumulator.evaluate()?; |
| 156 | + let expected_result = ScalarValue::Decimal128(Some(180_00000000), 14, 8); |
| 157 | + assert_eq!(result, expected_result); |
| 158 | + |
| 159 | + Ok(()) |
| 160 | + } |
| 161 | + |
| 162 | + #[test] |
| 163 | + fn test_decimal256_distinct_avg_accumulator() -> Result<()> { |
| 164 | + let precision = 50_u8; |
| 165 | + let scale = 2_i8; |
| 166 | + |
| 167 | + let array = Decimal256Array::from(vec![ |
| 168 | + Some(i256::from_i128(10_000)), |
| 169 | + Some(i256::from_i128(12_500)), |
| 170 | + Some(i256::from_i128(17_500)), |
| 171 | + Some(i256::from_i128(20_000)), |
| 172 | + Some(i256::from_i128(20_000)), |
| 173 | + Some(i256::from_i128(30_000)), |
| 174 | + None, |
| 175 | + None, |
| 176 | + ]) |
| 177 | + .with_precision_and_scale(precision, scale)?; |
| 178 | + |
| 179 | + let mut accumulator = |
| 180 | + DecimalDistinctAvgAccumulator::<Decimal256Type>::with_decimal_params( |
| 181 | + scale, 54, 6, |
| 182 | + ); |
| 183 | + accumulator.update_batch(&[Arc::new(array)])?; |
| 184 | + |
| 185 | + let result = accumulator.evaluate()?; |
| 186 | + let expected_result = |
| 187 | + ScalarValue::Decimal256(Some(i256::from_i128(180_000000)), 54, 6); |
| 188 | + assert_eq!(result, expected_result); |
| 189 | + |
| 190 | + Ok(()) |
| 191 | + } |
| 192 | +} |
0 commit comments