|
1 | | -use std::hash::Hash; |
2 | | -use arrow::array::{ArrayRef, BooleanArray, Int64Array}; |
3 | | -use arrow::datatypes::{ArrowPrimitiveType, DataType}; |
| 1 | +use arrow::array::{ |
| 2 | + Array, ArrayRef, AsArray, BooleanArray, Int64Array, ListBuilder, PrimitiveBuilder, |
| 3 | +}; |
| 4 | +use arrow::datatypes::ArrowPrimitiveType; |
| 5 | +use datafusion_common::HashSet; |
| 6 | +use datafusion_common::hash_utils::RandomState; |
4 | 7 | use datafusion_expr_common::groups_accumulator::{EmitTo, GroupsAccumulator}; |
| 8 | +use std::hash::Hash; |
| 9 | +use std::sync::Arc; |
5 | 10 |
|
6 | 11 | pub struct PrimitiveDistinctCountGroupsAccumulator<T: ArrowPrimitiveType> |
7 | | - where |
| 12 | +where |
8 | 13 | T::Native: Eq + Hash, |
9 | | - { |
| 14 | +{ |
10 | 15 | /// Count distinct per group. |
11 | | - values: Vec<T>, |
12 | | - data_type : DataType |
| 16 | + values: Vec<HashSet<T::Native, RandomState>>, |
| 17 | +} |
| 18 | + |
| 19 | +impl<T: ArrowPrimitiveType> Default for PrimitiveDistinctCountGroupsAccumulator<T> |
| 20 | +where |
| 21 | + T::Native: Eq + Hash, |
| 22 | +{ |
| 23 | + fn default() -> Self { |
| 24 | + Self::new() |
| 25 | + } |
13 | 26 | } |
14 | 27 |
|
15 | 28 | impl<T: ArrowPrimitiveType> PrimitiveDistinctCountGroupsAccumulator<T> |
16 | 29 | where |
17 | 30 | T::Native: Eq + Hash, |
18 | 31 | { |
19 | | - pub fn new(data_type: DataType) -> Self { |
20 | | - Self { |
21 | | - values: Vec::new(), |
22 | | - data_type, |
23 | | - } |
| 32 | + pub fn new() -> Self { |
| 33 | + Self { values: Vec::new() } |
24 | 34 | } |
25 | 35 | } |
26 | 36 |
|
27 | 37 | impl<T: ArrowPrimitiveType + Send + std::fmt::Debug> GroupsAccumulator |
28 | | -for PrimitiveDistinctCountGroupsAccumulator<T> |
| 38 | + for PrimitiveDistinctCountGroupsAccumulator<T> |
29 | 39 | where |
30 | 40 | T::Native: Eq + Hash, |
31 | | - { |
32 | | - fn update_batch(&mut self, values: &[ArrayRef], group_indices: &[usize], opt_filter: Option<&BooleanArray>, total_num_groups: usize) -> datafusion_common::Result<()> { |
33 | | - todo!() |
| 41 | +{ |
| 42 | + fn update_batch( |
| 43 | + &mut self, |
| 44 | + values: &[ArrayRef], |
| 45 | + group_indices: &[usize], |
| 46 | + opt_filter: Option<&BooleanArray>, |
| 47 | + total_num_groups: usize, |
| 48 | + ) -> datafusion_common::Result<()> { |
| 49 | + self.values.resize_with(total_num_groups, HashSet::default); |
| 50 | + debug_assert_eq!(values.len(), 1, "multiple arguments are not supported"); |
| 51 | + |
| 52 | + let arr = values[0].as_primitive::<T>(); |
| 53 | + |
| 54 | + for (idx, group_idx) in group_indices.iter().enumerate() { |
| 55 | + if let Some(filter) = opt_filter |
| 56 | + && !filter.value(idx) |
| 57 | + { |
| 58 | + continue; |
| 59 | + } |
| 60 | + if arr.is_valid(idx) { |
| 61 | + let value = arr.value(idx); |
| 62 | + self.values[*group_idx].insert(value); |
| 63 | + } |
| 64 | + } |
| 65 | + |
| 66 | + Ok(()) |
34 | 67 | } |
35 | 68 |
|
36 | 69 | fn evaluate(&mut self, emit_to: EmitTo) -> datafusion_common::Result<ArrayRef> { |
37 | | - todo!() |
| 70 | + let counts: Vec<i64> = emit_to |
| 71 | + .take_needed(&mut self.values) |
| 72 | + .iter() |
| 73 | + .map(|groups| groups.len() as i64) |
| 74 | + .collect(); |
| 75 | + |
| 76 | + Ok(Arc::new(Int64Array::from(counts))) |
38 | 77 | } |
39 | 78 |
|
40 | 79 | fn state(&mut self, emit_to: EmitTo) -> datafusion_common::Result<Vec<ArrayRef>> { |
41 | | - todo!() |
| 80 | + let hash_sets = emit_to.take_needed(&mut self.values); |
| 81 | + let mut builder = ListBuilder::new(PrimitiveBuilder::<T>::new()); |
| 82 | + |
| 83 | + for set in hash_sets { |
| 84 | + for value in set { |
| 85 | + builder.values().append_value(value); |
| 86 | + } |
| 87 | + builder.append(true); |
| 88 | + } |
| 89 | + Ok(vec![Arc::new(builder.finish())]) |
42 | 90 | } |
43 | 91 |
|
44 | | - fn merge_batch(&mut self, values: &[ArrayRef], group_indices: &[usize], opt_filter: Option<&BooleanArray>, total_num_groups: usize) -> datafusion_common::Result<()> { |
45 | | - todo!() |
| 92 | + fn merge_batch( |
| 93 | + &mut self, |
| 94 | + values: &[ArrayRef], |
| 95 | + group_indices: &[usize], |
| 96 | + _opt_filter: Option<&BooleanArray>, |
| 97 | + total_num_groups: usize, |
| 98 | + ) -> datafusion_common::Result<()> { |
| 99 | + self.values.resize_with(total_num_groups, HashSet::default); |
| 100 | + let list_array = values[0].as_list::<i32>(); |
| 101 | + |
| 102 | + for (row_idx, group_idx) in group_indices.iter().enumerate() { |
| 103 | + let inner = list_array.value(row_idx); |
| 104 | + let inner_set = inner.as_primitive::<T>(); |
| 105 | + for i in 0..inner.len() { |
| 106 | + self.values[*group_idx].insert(inner_set.value(i)); |
| 107 | + } |
| 108 | + } |
| 109 | + Ok(()) |
46 | 110 | } |
47 | 111 |
|
48 | 112 | fn size(&self) -> usize { |
49 | | - todo!() |
| 113 | + size_of::<Self>() |
| 114 | + + self.values.capacity() * size_of::<HashSet<T::Native, RandomState>>() |
| 115 | + + self |
| 116 | + .values |
| 117 | + .iter() |
| 118 | + .map(|s| s.capacity() * size_of::<T::Native>()) |
| 119 | + .sum::<usize>() |
50 | 120 | } |
51 | 121 | } |
0 commit comments