|
| 1 | +// SPDX-License-Identifier: Apache-2.0 |
| 2 | +// SPDX-FileCopyrightText: Copyright the Vortex contributors |
| 3 | + |
| 4 | +use vortex::ArrayRef; |
| 5 | +use vortex::arrays::PrimitiveArray; |
| 6 | +use vortex::buffer::Buffer; |
| 7 | +use vortex::dtype::{DType, NativePType, Nullability}; |
| 8 | +use vortex::encodings::sequence::SequenceArray; |
| 9 | +use vortex::error::{VortexExpect, VortexResult, vortex_bail}; |
| 10 | +use vortex::scalar::PValue; |
| 11 | +use vortex::validity::Validity; |
| 12 | + |
| 13 | +pub fn sequence_array_from_range<T: NativePType + TryFrom<isize> + Into<PValue>>( |
| 14 | + start: isize, |
| 15 | + stop: isize, |
| 16 | + step: isize, |
| 17 | + dtype: DType, |
| 18 | +) -> VortexResult<ArrayRef> { |
| 19 | + if step == 0 { |
| 20 | + vortex_bail!("Step must not be zero"); |
| 21 | + } |
| 22 | + |
| 23 | + let Some(len) = range_len(start, stop, step) else { |
| 24 | + let validity = match dtype.nullability() { |
| 25 | + Nullability::NonNullable => Validity::NonNullable, |
| 26 | + Nullability::Nullable => Validity::AllValid, |
| 27 | + }; |
| 28 | + return Ok(PrimitiveArray::new::<T>(Buffer::empty(), validity).to_array()); |
| 29 | + }; |
| 30 | + let Ok(start) = T::try_from(start) else { |
| 31 | + vortex_bail!( |
| 32 | + "Start, {}, does not fit in requested dtype: {}", |
| 33 | + start, |
| 34 | + dtype |
| 35 | + ); |
| 36 | + }; |
| 37 | + let Ok(step) = T::try_from(step) else { |
| 38 | + vortex_bail!("Step, {}, does not fit in requested dtype: {}", step, dtype); |
| 39 | + }; |
| 40 | + |
| 41 | + Ok(SequenceArray::typed_new::<T>(start, step, dtype.nullability(), len)?.to_array()) |
| 42 | +} |
| 43 | + |
| 44 | +fn range_len(start: isize, stop: isize, step: isize) -> Option<usize> { |
| 45 | + if step > 0 { |
| 46 | + if start > stop { |
| 47 | + return None; |
| 48 | + } |
| 49 | + |
| 50 | + let len = (stop - start + step - 1) / step; |
| 51 | + let len = |
| 52 | + usize::try_from(len).vortex_expect("stop >= start, step > 0, so len is non-negative"); |
| 53 | + Some(len) |
| 54 | + } else { |
| 55 | + assert!(step != 0); |
| 56 | + |
| 57 | + if stop > start { |
| 58 | + return None; |
| 59 | + } |
| 60 | + |
| 61 | + let len = (start - stop + -step - 1) / -step; |
| 62 | + let len = |
| 63 | + usize::try_from(len).vortex_expect("start >= stop, step < 0, so len is non-negative"); |
| 64 | + Some(len) |
| 65 | + } |
| 66 | +} |
| 67 | + |
| 68 | +#[cfg(test)] |
| 69 | +mod test { |
| 70 | + use vortex::IntoArray as _; |
| 71 | + use vortex::arrow::IntoArrowArray; |
| 72 | + use vortex::buffer::buffer; |
| 73 | + use vortex::dtype::{DType, Nullability, PType}; |
| 74 | + |
| 75 | + use crate::arrays::range_to_sequence::{range_len, sequence_array_from_range}; |
| 76 | + |
| 77 | + #[test] |
| 78 | + fn test_range_len() { |
| 79 | + assert_eq!(range_len(0, 10, 1).unwrap(), 10); |
| 80 | + assert_eq!(range_len(0, 10, 5).unwrap(), 2); |
| 81 | + assert_eq!(range_len(0, 10, 10).unwrap(), 1); |
| 82 | + assert_eq!(range_len(0, 10, 100).unwrap(), 1); |
| 83 | + assert_eq!(range_len(-5, -5, 1).unwrap(), 0); |
| 84 | + assert_eq!(range_len(-5, 5, 3).unwrap(), 4); |
| 85 | + assert_eq!(range_len(-7, -5, 1).unwrap(), 2); |
| 86 | + assert_eq!(range_len(3, -3, -1).unwrap(), 6); |
| 87 | + assert_eq!(range_len(10, 3, 1), None); |
| 88 | + assert_eq!(range_len(0, 10, -1), None); |
| 89 | + } |
| 90 | + |
| 91 | + #[test] |
| 92 | + fn test_sequence_array_from_len() { |
| 93 | + let dtype = DType::Primitive(PType::U16, Nullability::NonNullable); |
| 94 | + let arr = sequence_array_from_range::<u16>(0, 10, 1, dtype.clone()).unwrap(); |
| 95 | + assert_eq!(arr.dtype(), &dtype); |
| 96 | + assert_eq!( |
| 97 | + &arr.into_arrow_preferred().unwrap(), |
| 98 | + &buffer![0u16, 1, 2, 3, 4, 5, 6, 7, 8, 9] |
| 99 | + .into_array() |
| 100 | + .into_arrow_preferred() |
| 101 | + .unwrap() |
| 102 | + ); |
| 103 | + |
| 104 | + let dtype = DType::Primitive(PType::I32, Nullability::NonNullable); |
| 105 | + let arr = sequence_array_from_range::<i32>(0, 10, 5, dtype.clone()).unwrap(); |
| 106 | + assert_eq!(arr.dtype(), &dtype); |
| 107 | + assert_eq!( |
| 108 | + &arr.into_arrow_preferred().unwrap(), |
| 109 | + &buffer![0i32, 5] |
| 110 | + .into_array() |
| 111 | + .into_arrow_preferred() |
| 112 | + .unwrap() |
| 113 | + ); |
| 114 | + |
| 115 | + let dtype = DType::Primitive(PType::I8, Nullability::NonNullable); |
| 116 | + let arr = sequence_array_from_range::<i8>(-5, 5, 3, dtype.clone()).unwrap(); |
| 117 | + assert_eq!(arr.dtype(), &dtype); |
| 118 | + assert_eq!( |
| 119 | + &arr.into_arrow_preferred().unwrap(), |
| 120 | + &buffer![-5i8, -2, 1, 4] |
| 121 | + .into_array() |
| 122 | + .into_arrow_preferred() |
| 123 | + .unwrap() |
| 124 | + ); |
| 125 | + |
| 126 | + let dtype = DType::Primitive(PType::I8, Nullability::NonNullable); |
| 127 | + let arr = sequence_array_from_range::<i8>(3, -3, -1, dtype.clone()).unwrap(); |
| 128 | + assert_eq!(arr.dtype(), &dtype); |
| 129 | + assert_eq!( |
| 130 | + &arr.into_arrow_preferred().unwrap(), |
| 131 | + &buffer![3i8, 2, 1, 0, -1, -2] |
| 132 | + .into_array() |
| 133 | + .into_arrow_preferred() |
| 134 | + .unwrap() |
| 135 | + ); |
| 136 | + |
| 137 | + let dtype = DType::Primitive(PType::U32, Nullability::NonNullable); |
| 138 | + let result = sequence_array_from_range::<u32>(1_000_000, 10, -500_000, dtype); |
| 139 | + assert!( |
| 140 | + result.is_err_and(|err| err.to_string().contains("does not fit in requested dtype")) |
| 141 | + ); |
| 142 | + |
| 143 | + let dtype = DType::Primitive(PType::I32, Nullability::NonNullable); |
| 144 | + let arr = sequence_array_from_range::<i32>(1_000_000, 10, -500_000, dtype.clone()).unwrap(); |
| 145 | + assert_eq!(arr.dtype(), &dtype); |
| 146 | + assert_eq!( |
| 147 | + &arr.into_arrow_preferred().unwrap(), |
| 148 | + &buffer![1_000_000i32, 500_000] |
| 149 | + .into_array() |
| 150 | + .into_arrow_preferred() |
| 151 | + .unwrap() |
| 152 | + ); |
| 153 | + } |
| 154 | +} |
0 commit comments