|
| 1 | +// Copyright The OpenTelemetry Authors |
| 2 | +// SPDX-License-Identifier: Apache-2.0 |
| 3 | + |
| 4 | +//! This module provides adaptive array builders for Arrow arrays. |
| 5 | +//! |
| 6 | +//! Often for OTel-Arrow, we have columns that are optional on the schema. For example, the Boolean |
| 7 | +//! column may not be present in a record batch representing a list of attributes only be present |
| 8 | +//! if there are some boolean type values in the list. |
| 9 | +//! |
| 10 | +//! There are also cases where we want to dynamically use dictionary encoding with the smallest index |
| 11 | +//! the cardinality of the allows. |
| 12 | +//! |
| 13 | +//! This module contains adaptive array builders that can dynamically create either no array (for |
| 14 | +//! an all-null) column, an array that may be a dictionary, of an array or native types. It will |
| 15 | +//! handle converting between different builders dynamically based on the data which is appended. |
| 16 | +
|
| 17 | +use arrow::array::{ArrayRef, StringBuilder, StringDictionaryBuilder}; |
| 18 | +use arrow::datatypes::{DataType, UInt8Type, UInt16Type}; |
| 19 | + |
| 20 | +use crate::arrays::NullableArrayAccessor; |
| 21 | + |
| 22 | +use dictionary::{ |
| 23 | + AdaptiveDictionaryBuilder, ConvertToNativeHelper, DictionaryArrayBuilder, |
| 24 | + DictionaryBuilderError, DictionaryOptions, UpdateDictionaryIndexInto, |
| 25 | +}; |
| 26 | + |
| 27 | +pub mod dictionary; |
| 28 | +pub mod string; |
| 29 | + |
| 30 | +/// This is the base trait that array builders should implement. |
| 31 | +pub trait ArrayBuilder { |
| 32 | + type Native; |
| 33 | + |
| 34 | + fn append_value(&mut self, value: &Self::Native); |
| 35 | + |
| 36 | + fn finish(&mut self) -> ArrayWithType; |
| 37 | +} |
| 38 | + |
| 39 | +pub struct ArrayWithType { |
| 40 | + pub array: ArrayRef, |
| 41 | + pub data_type: DataType, |
| 42 | +} |
| 43 | + |
| 44 | +/// This is a helper trait that allows the adaptive builders to construct new |
| 45 | +/// instances of the builder dynamically |
| 46 | +pub trait ArrayBuilderConstructor { |
| 47 | + fn new() -> Self; |
| 48 | + |
| 49 | + // TODO, at some point we may consider optionally adding a |
| 50 | + // with_capacity function here that could be used to create |
| 51 | + // a builder with pre-allocated buffers |
| 52 | +} |
| 53 | + |
| 54 | +/// This enum is a container that abstracts array builder which is either |
| 55 | +/// dictionary or native. It converts from the dictionary builder to the |
| 56 | +/// native builder when the dictionary builder overflows. |
| 57 | +enum MaybeDictionaryBuilder< |
| 58 | + NativeBuilder: ArrayBuilder + ArrayBuilderConstructor, |
| 59 | + DictBuilderU8: DictionaryArrayBuilder<UInt8Type> + ArrayBuilderConstructor, |
| 60 | + DictBuilderU16: DictionaryArrayBuilder<UInt16Type> + ArrayBuilderConstructor, |
| 61 | +> { |
| 62 | + Native(NativeBuilder), |
| 63 | + Dictionary(AdaptiveDictionaryBuilder<DictBuilderU8, DictBuilderU16>), |
| 64 | +} |
| 65 | + |
| 66 | +impl<T, TN, TD8, TD16> ArrayBuilder for MaybeDictionaryBuilder<TN, TD8, TD16> |
| 67 | +where |
| 68 | + TN: ArrayBuilder<Native = T> + ArrayBuilderConstructor, |
| 69 | + TD8: DictionaryArrayBuilder<UInt8Type, Native = T> |
| 70 | + + ArrayBuilderConstructor |
| 71 | + + ConvertToNativeHelper, |
| 72 | + <TD8 as ConvertToNativeHelper>::Accessor: NullableArrayAccessor<Native = T> + 'static, |
| 73 | + TD16: DictionaryArrayBuilder<UInt16Type, Native = T> |
| 74 | + + ArrayBuilderConstructor |
| 75 | + + ConvertToNativeHelper, |
| 76 | + <TD16 as ConvertToNativeHelper>::Accessor: NullableArrayAccessor<Native = T> + 'static, |
| 77 | + TD8: UpdateDictionaryIndexInto<TD16>, |
| 78 | +{ |
| 79 | + type Native = T; |
| 80 | + |
| 81 | + fn append_value( |
| 82 | + &mut self, |
| 83 | + value: &<MaybeDictionaryBuilder<TN, TD8, TD16> as ArrayBuilder>::Native, |
| 84 | + ) { |
| 85 | + match self { |
| 86 | + Self::Native(array_builder) => array_builder.append_value(value), |
| 87 | + Self::Dictionary(dict_array_builder) => match dict_array_builder.append_value(value) { |
| 88 | + // we've overflowed the dictionary, so we must convert to the native builder type |
| 89 | + Err(DictionaryBuilderError::DictOverflow {}) => { |
| 90 | + let mut native = TN::new(); |
| 91 | + dict_array_builder.to_native(&mut native); |
| 92 | + native.append_value(value); |
| 93 | + *self = Self::Native(native); |
| 94 | + } |
| 95 | + _ => { |
| 96 | + // do nothing here, as the append was successful |
| 97 | + } |
| 98 | + }, |
| 99 | + } |
| 100 | + } |
| 101 | + |
| 102 | + fn finish(&mut self) -> ArrayWithType { |
| 103 | + match self { |
| 104 | + Self::Dictionary(dict_array_builder) => dict_array_builder.finish(), |
| 105 | + Self::Native(array_builder) => array_builder.finish(), |
| 106 | + } |
| 107 | + } |
| 108 | +} |
| 109 | + |
| 110 | +#[derive(Default)] |
| 111 | +pub struct ArrayOptions { |
| 112 | + pub dictionary_options: Option<DictionaryOptions>, |
| 113 | + pub nullable: bool, |
| 114 | +} |
| 115 | + |
| 116 | +pub struct AdaptiveArrayBuilder< |
| 117 | + TN: ArrayBuilder + ArrayBuilderConstructor, |
| 118 | + TD8: DictionaryArrayBuilder<UInt8Type> + ArrayBuilderConstructor, |
| 119 | + TD16: DictionaryArrayBuilder<UInt16Type> + ArrayBuilderConstructor, |
| 120 | +> { |
| 121 | + dictionary_options: Option<DictionaryOptions>, |
| 122 | + inner: Option<MaybeDictionaryBuilder<TN, TD8, TD16>>, |
| 123 | +} |
| 124 | + |
| 125 | +impl<T, TN, TD8, TD16> AdaptiveArrayBuilder<TN, TD8, TD16> |
| 126 | +where |
| 127 | + TN: ArrayBuilder<Native = T> + ArrayBuilderConstructor, |
| 128 | + TD8: DictionaryArrayBuilder<UInt8Type, Native = T> |
| 129 | + + ArrayBuilderConstructor |
| 130 | + + ConvertToNativeHelper, |
| 131 | + <TD8 as ConvertToNativeHelper>::Accessor: NullableArrayAccessor<Native = T> + 'static, |
| 132 | + TD16: DictionaryArrayBuilder<UInt16Type, Native = T> |
| 133 | + + ArrayBuilderConstructor |
| 134 | + + ConvertToNativeHelper, |
| 135 | + <TD16 as ConvertToNativeHelper>::Accessor: NullableArrayAccessor<Native = T> + 'static, |
| 136 | + TD8: UpdateDictionaryIndexInto<TD16>, |
| 137 | +{ |
| 138 | + pub fn new(options: ArrayOptions) -> Self { |
| 139 | + let inner = if options.nullable { |
| 140 | + None |
| 141 | + } else { |
| 142 | + Some(Self::initial_builder(&options.dictionary_options)) |
| 143 | + }; |
| 144 | + |
| 145 | + Self { |
| 146 | + dictionary_options: options.dictionary_options, |
| 147 | + inner, |
| 148 | + } |
| 149 | + } |
| 150 | + |
| 151 | + // Initializes the builder, which may either be a builder for the, if dictionary |
| 152 | + // options is `Some`, otherwise it will construct the native builder builder variant |
| 153 | + fn initial_builder( |
| 154 | + dictionary_options: &Option<DictionaryOptions>, |
| 155 | + ) -> MaybeDictionaryBuilder<TN, TD8, TD16> { |
| 156 | + match dictionary_options.as_ref() { |
| 157 | + Some(dictionary_options) => MaybeDictionaryBuilder::Dictionary( |
| 158 | + AdaptiveDictionaryBuilder::new(dictionary_options), |
| 159 | + ), |
| 160 | + None => MaybeDictionaryBuilder::Native(TN::new()), |
| 161 | + } |
| 162 | + } |
| 163 | + |
| 164 | + fn append_value(&mut self, value: &T) { |
| 165 | + if self.inner.is_none() { |
| 166 | + // TODO -- when we handle nulls here we need to keep track of how many |
| 167 | + // nulls have been appended before the first value, and prefix this |
| 168 | + // newly initialized array with that number of nulls |
| 169 | + // https://github.com/open-telemetry/otel-arrow/issues/534 |
| 170 | + self.inner = Some(Self::initial_builder(&self.dictionary_options)); |
| 171 | + } |
| 172 | + |
| 173 | + let inner = self |
| 174 | + .inner |
| 175 | + .as_mut() |
| 176 | + .expect("inner should now be initialized"); |
| 177 | + inner.append_value(value) |
| 178 | + } |
| 179 | + |
| 180 | + fn finish(&mut self) -> Option<ArrayWithType> { |
| 181 | + self.inner.as_mut().map(|builder| builder.finish()) |
| 182 | + } |
| 183 | +} |
| 184 | + |
| 185 | +pub type StringArrayBuilder = AdaptiveArrayBuilder< |
| 186 | + StringBuilder, |
| 187 | + StringDictionaryBuilder<UInt8Type>, |
| 188 | + StringDictionaryBuilder<UInt16Type>, |
| 189 | +>; |
| 190 | + |
| 191 | +#[cfg(test)] |
| 192 | +pub mod test { |
| 193 | + use super::*; |
| 194 | + |
| 195 | + use std::sync::Arc; |
| 196 | + |
| 197 | + use arrow::array::{StringArray, UInt8Array, UInt8DictionaryArray}; |
| 198 | + use arrow::datatypes::DataType; |
| 199 | + |
| 200 | + #[test] |
| 201 | + fn test_array_builder() { |
| 202 | + let mut builder = StringArrayBuilder::new(ArrayOptions { |
| 203 | + nullable: true, |
| 204 | + dictionary_options: Some(DictionaryOptions { |
| 205 | + min_cardinality: 4, |
| 206 | + max_cardinality: 4, |
| 207 | + }), |
| 208 | + }); |
| 209 | + |
| 210 | + // expect that for empty array, we get a None value because the builder is nullable |
| 211 | + let result = builder.finish(); |
| 212 | + assert!(result.is_none()); |
| 213 | + |
| 214 | + // expect that when we add values, we get a dictionary |
| 215 | + builder.append_value(&"a".to_string()); |
| 216 | + builder.append_value(&"a".to_string()); |
| 217 | + builder.append_value(&"b".to_string()); |
| 218 | + |
| 219 | + let result = builder.finish().unwrap(); |
| 220 | + assert_eq!( |
| 221 | + result.data_type, |
| 222 | + DataType::Dictionary(Box::new(DataType::UInt8), Box::new(DataType::Utf8)) |
| 223 | + ); |
| 224 | + |
| 225 | + let mut expected_dict_values = StringBuilder::new(); |
| 226 | + expected_dict_values.append_value("a"); |
| 227 | + expected_dict_values.append_value("b"); |
| 228 | + let expected_dict_keys = UInt8Array::from_iter_values(vec![0, 0, 1]); |
| 229 | + let expected = |
| 230 | + UInt8DictionaryArray::new(expected_dict_keys, Arc::new(expected_dict_values.finish())); |
| 231 | + |
| 232 | + assert_eq!( |
| 233 | + result |
| 234 | + .array |
| 235 | + .as_any() |
| 236 | + .downcast_ref::<UInt8DictionaryArray>() |
| 237 | + .unwrap(), |
| 238 | + &expected |
| 239 | + ); |
| 240 | + } |
| 241 | + |
| 242 | + #[test] |
| 243 | + fn test_array_builder_non_nullable_empty() { |
| 244 | + let mut builder = StringArrayBuilder::new(ArrayOptions { |
| 245 | + nullable: false, |
| 246 | + dictionary_options: Some(DictionaryOptions { |
| 247 | + min_cardinality: 4, |
| 248 | + max_cardinality: 4, |
| 249 | + }), |
| 250 | + }); |
| 251 | + |
| 252 | + // check that since the type we're building is not nullable, we get an empty array |
| 253 | + let result = builder.finish().unwrap(); |
| 254 | + assert_eq!( |
| 255 | + result.data_type, |
| 256 | + DataType::Dictionary(Box::new(DataType::UInt8), Box::new(DataType::Utf8)) |
| 257 | + ); |
| 258 | + assert_eq!(result.array.len(), 0); |
| 259 | + } |
| 260 | + |
| 261 | + #[test] |
| 262 | + fn test_array_builder_dict_overflow() { |
| 263 | + let mut builder = StringArrayBuilder::new(ArrayOptions { |
| 264 | + nullable: false, |
| 265 | + dictionary_options: Some(DictionaryOptions { |
| 266 | + min_cardinality: 4, |
| 267 | + max_cardinality: 4, |
| 268 | + }), |
| 269 | + }); |
| 270 | + |
| 271 | + // expect that when we add values, we get a dictionary |
| 272 | + builder.append_value(&"a".to_string()); |
| 273 | + builder.append_value(&"b".to_string()); |
| 274 | + builder.append_value(&"c".to_string()); |
| 275 | + builder.append_value(&"d".to_string()); |
| 276 | + builder.append_value(&"e".to_string()); |
| 277 | + |
| 278 | + let result = builder.finish().unwrap(); |
| 279 | + assert_eq!(result.data_type, DataType::Utf8); |
| 280 | + |
| 281 | + let mut expected_values = StringBuilder::new(); |
| 282 | + expected_values.append_value("a"); |
| 283 | + expected_values.append_value("b"); |
| 284 | + expected_values.append_value("c"); |
| 285 | + expected_values.append_value("d"); |
| 286 | + expected_values.append_value("e"); |
| 287 | + |
| 288 | + assert_eq!( |
| 289 | + result.array.as_any().downcast_ref::<StringArray>().unwrap(), |
| 290 | + &expected_values.finish() |
| 291 | + ) |
| 292 | + } |
| 293 | +} |
0 commit comments