(
value_type: DictValueType,
cardinality: Cardinality,
null_percent: NullPercent,
size: usize,
)
| 257 | ]; |
| 258 | |
| 259 | pub fn generate_dict( |
| 260 | value_type: DictValueType, |
| 261 | cardinality: Cardinality, |
| 262 | null_percent: NullPercent, |
| 263 | size: usize, |
| 264 | ) -> ArrayRef { |
| 265 | let num_distinct = cardinality.num_distinct(size); |
| 266 | |
| 267 | let dict_values: ArrayRef = match value_type { |
| 268 | DictValueType::Utf8 => { |
| 269 | let strings: StringArray = (0..num_distinct) |
| 270 | .map(|i| format!("value_{i}")) |
| 271 | .collect::<Vec<_>>() |
| 272 | .into(); |
| 273 | Arc::new(strings) |
| 274 | } |
| 275 | DictValueType::ListUtf8 => { |
| 276 | let flat: StringArray = (0..num_distinct) |
| 277 | .flat_map(|i| (0..ITEMS_PER_VALUE).map(move |j| format!("value_{i}_{j}"))) |
| 278 | .collect::<Vec<_>>() |
| 279 | .into(); |
| 280 | |
| 281 | let offsets: Vec<i32> = (0..=(num_distinct * ITEMS_PER_VALUE)) |
| 282 | .step_by(ITEMS_PER_VALUE) |
| 283 | .map(|o| o as i32) |
| 284 | .collect(); |
| 285 | |
| 286 | Arc::new(ListArray::new( |
| 287 | Arc::new(Field::new("item", DataType::Utf8, false)), |
| 288 | OffsetBuffer::new(offsets.into()), |
| 289 | Arc::new(flat), |
| 290 | None, |
| 291 | )) |
| 292 | } |
| 293 | }; |
| 294 | |
| 295 | let null_every = null_percent.null_every(); |
| 296 | |
| 297 | let mut key_builder = Int32Array::builder(size); |
| 298 | for i in 0..size { |
| 299 | if null_every.is_some_and(|n| i % n == 0) { |
| 300 | key_builder.append_null(); |
| 301 | } else { |
| 302 | key_builder.append_value((i % num_distinct) as i32); |
| 303 | } |
| 304 | } |
| 305 | let keys = key_builder.finish(); |
| 306 | |
| 307 | Arc::new( |
| 308 | DictionaryArray::<Int32Type>::try_new(keys, dict_values) |
| 309 | .expect("valid dictionary array"), |
| 310 | ) |
| 311 | } |
| 312 | impl RunOpt { |
| 313 | pub async fn run(self) -> Result<()> { |
| 314 | println!( |
no test coverage detected
searching dependent graphs…