(
string_array: &GenericStringArray<T>,
args: &[ArrayRef],
)
| 344 | |
| 345 | #[expect(clippy::needless_range_loop)] |
| 346 | fn generic_string_substr<T: OffsetSizeTrait>( |
| 347 | string_array: &GenericStringArray<T>, |
| 348 | args: &[ArrayRef], |
| 349 | ) -> Result<ArrayRef> { |
| 350 | // We'd like to return a StringViewArray that points into the input string |
| 351 | // array's values buffer. Since the Arrow spec defines StringView offsets |
| 352 | // as i32, we can't use this approach when the values buffer is >2GB, so |
| 353 | // fallback to copying. |
| 354 | if !values_fit_in_i32(string_array) { |
| 355 | return generic_string_substr_copy(string_array, args); |
| 356 | } |
| 357 | |
| 358 | let start_array = as_int64_array(&args[0])?; |
| 359 | let count_array_opt = args.get(1).map(|a| as_int64_array(a)).transpose()?; |
| 360 | |
| 361 | let is_ascii = enable_ascii_fast_path(&string_array, start_array, count_array_opt); |
| 362 | let offsets = string_array.value_offsets(); |
| 363 | let mut views_buf = Vec::with_capacity(string_array.len()); |
| 364 | let mut has_out_of_line = false; |
| 365 | |
| 366 | // Combine null bitmaps from all inputs in bulk. |
| 367 | let nulls = NullBuffer::union_many([ |
| 368 | string_array.nulls(), |
| 369 | start_array.nulls(), |
| 370 | count_array_opt.and_then(|a| a.nulls()), |
| 371 | ]); |
| 372 | |
| 373 | for i in 0..string_array.len() { |
| 374 | if nulls.as_ref().is_some_and(|n| n.is_null(i)) { |
| 375 | views_buf.push(0); |
| 376 | continue; |
| 377 | } |
| 378 | |
| 379 | let string = string_array.value(i); |
| 380 | let source_offset = offsets[i].as_usize(); |
| 381 | let start = start_array.value(i); |
| 382 | let count = count_array_opt.map(|a| a.value(i)); |
| 383 | |
| 384 | let (byte_start, byte_end) = get_true_start_end(string, start, count, is_ascii)?; |
| 385 | has_out_of_line |= append_view_from_buffer( |
| 386 | &mut views_buf, |
| 387 | &string[byte_start..byte_end], |
| 388 | source_offset + byte_start, |
| 389 | ); |
| 390 | } |
| 391 | |
| 392 | let views_buf = ScalarBuffer::from(views_buf); |
| 393 | |
| 394 | // If all result strings are stored inline, we don't need to retain the |
| 395 | // input string array. |
| 396 | let data_buffers = if has_out_of_line { |
| 397 | vec![string_array.values().clone()] |
| 398 | } else { |
| 399 | vec![] |
| 400 | }; |
| 401 | |
| 402 | // Safety: |
| 403 | // (1) The blocks of the given views are all provided |
no test coverage detected
searching dependent graphs…