|
| 1 | +// SPDX-License-Identifier: Apache-2.0 |
| 2 | +// SPDX-FileCopyrightText: Copyright the Vortex contributors |
| 3 | + |
| 4 | +use vortex_error::VortexResult; |
| 5 | +use vortex_mask::Mask; |
| 6 | + |
| 7 | +use super::slice::inherit_selection_stats; |
| 8 | +use super::slice::slice_constant_code; |
| 9 | +use crate::ArrayRef; |
| 10 | +use crate::IntoArray; |
| 11 | +use crate::array::ArrayView; |
| 12 | +use crate::arrays::Constant; |
| 13 | +use crate::arrays::Dict; |
| 14 | +use crate::arrays::DictArray; |
| 15 | +use crate::arrays::dict::DictArraySlotsExt; |
| 16 | +use crate::arrays::filter::FilterReduce; |
| 17 | + |
| 18 | +impl FilterReduce for Dict { |
| 19 | + fn filter(array: ArrayView<'_, Dict>, mask: &Mask) -> VortexResult<Option<ArrayRef>> { |
| 20 | + if let Some(code) = array.codes().as_opt::<Constant>() { |
| 21 | + return slice_constant_code(array, code.scalar(), mask.true_count()); |
| 22 | + } |
| 23 | + |
| 24 | + let codes = array.codes().filter(mask.clone())?; |
| 25 | + if let Some(code) = codes.as_opt::<Constant>() { |
| 26 | + return slice_constant_code(array, code.scalar(), codes.len()); |
| 27 | + } |
| 28 | + inherit_selection_stats(array.codes(), &codes); |
| 29 | + |
| 30 | + // SAFETY: Selection preserves the integer code type and non-null code bounds, so every code |
| 31 | + // indexes the unchanged values. `new_unchecked` resets `all_values_referenced` because |
| 32 | + // selection can remove a value's last reference. |
| 33 | + let filtered = unsafe { DictArray::new_unchecked(codes, array.values().clone()) }; |
| 34 | + |
| 35 | + Ok(Some(filtered.into_array())) |
| 36 | + } |
| 37 | +} |
| 38 | + |
| 39 | +#[cfg(test)] |
| 40 | +mod tests { |
| 41 | + use vortex_buffer::buffer; |
| 42 | + use vortex_error::VortexResult; |
| 43 | + use vortex_mask::Mask; |
| 44 | + |
| 45 | + use crate::IntoArray; |
| 46 | + use crate::VortexSessionExecute; |
| 47 | + use crate::array_session; |
| 48 | + use crate::arrays::ConstantArray; |
| 49 | + use crate::arrays::Dict; |
| 50 | + use crate::arrays::DictArray; |
| 51 | + use crate::arrays::PrimitiveArray; |
| 52 | + use crate::arrays::dict::DictArraySlotsExt; |
| 53 | + use crate::assert_arrays_eq; |
| 54 | + use crate::expr::stats::Precision; |
| 55 | + use crate::expr::stats::Stat; |
| 56 | + use crate::expr::stats::StatsProviderExt; |
| 57 | + use crate::scalar::ScalarValue; |
| 58 | + |
| 59 | + #[test] |
| 60 | + fn filter_constant_valid_code_keeps_only_referenced_value() -> VortexResult<()> { |
| 61 | + let mut ctx = array_session().create_execution_ctx(); |
| 62 | + let dict = DictArray::new( |
| 63 | + ConstantArray::new(1u8, 5).into_array(), |
| 64 | + buffer![10i32, 20, 30].into_array(), |
| 65 | + ); |
| 66 | + let filtered = dict.filter(Mask::from_iter([true, false, true, true, false]))?; |
| 67 | + |
| 68 | + let filtered_dict = filtered.as_::<Dict>(); |
| 69 | + assert_arrays_eq!( |
| 70 | + filtered_dict.values(), |
| 71 | + buffer![20i32].into_array(), |
| 72 | + &mut ctx |
| 73 | + ); |
| 74 | + assert_arrays_eq!( |
| 75 | + filtered, |
| 76 | + PrimitiveArray::from_iter([20i32, 20, 20]), |
| 77 | + &mut ctx |
| 78 | + ); |
| 79 | + Ok(()) |
| 80 | + } |
| 81 | + |
| 82 | + #[test] |
| 83 | + fn filter_inherits_true_order_stats_from_codes() -> VortexResult<()> { |
| 84 | + let codes = buffer![0u8, 0, 1, 2].into_array(); |
| 85 | + codes |
| 86 | + .statistics() |
| 87 | + .set(Stat::IsSorted, Precision::exact(ScalarValue::Bool(true))); |
| 88 | + codes |
| 89 | + .statistics() |
| 90 | + .set(Stat::IsConstant, Precision::exact(ScalarValue::Bool(false))); |
| 91 | + let dict = DictArray::new(codes, buffer![10i32, 20, 30].into_array()); |
| 92 | + |
| 93 | + let filtered = dict.filter(Mask::from_iter([true, false, true, true]))?; |
| 94 | + |
| 95 | + let filtered_codes = filtered.as_::<Dict>().codes().clone(); |
| 96 | + assert_eq!( |
| 97 | + filtered_codes.statistics().get_as::<bool>(Stat::IsSorted), |
| 98 | + Precision::exact(true) |
| 99 | + ); |
| 100 | + assert_eq!( |
| 101 | + filtered_codes.statistics().get_as::<bool>(Stat::IsConstant), |
| 102 | + Precision::Absent |
| 103 | + ); |
| 104 | + Ok(()) |
| 105 | + } |
| 106 | +} |
0 commit comments