vortex_array/arrays/varbin/compute/
compare.rs1use std::cmp::Ordering;
5
6use vortex_buffer::BitBuffer;
7use vortex_error::VortexExpect as _;
8use vortex_error::VortexResult;
9use vortex_error::vortex_bail;
10
11use crate::ArrayRef;
12use crate::ExecutionCtx;
13use crate::IntoArray;
14use crate::array::ArrayView;
15use crate::arrays::BoolArray;
16use crate::arrays::PrimitiveArray;
17use crate::arrays::VarBin;
18use crate::arrays::varbin::VarBinArraySlotsExt;
19use crate::dtype::DType;
20use crate::dtype::IntegerPType;
21use crate::match_each_integer_ptype;
22use crate::scalar_fn::fns::binary::CompareKernel;
23use crate::scalar_fn::fns::operators::CompareOperator;
24
25impl CompareKernel for VarBin {
28 fn compare(
29 lhs: ArrayView<'_, VarBin>,
30 rhs: &ArrayRef,
31 operator: CompareOperator,
32 ctx: &mut ExecutionCtx,
33 ) -> VortexResult<Option<ArrayRef>> {
34 let Some(rhs_const) = rhs.as_constant() else {
35 return Ok(None);
36 };
37
38 let len = lhs.len();
39
40 let rhs_bytes: &[u8] = match rhs_const.dtype() {
43 DType::Binary(_) => rhs_const
44 .as_binary()
45 .value()
46 .vortex_expect("RHS should not be null")
47 .as_slice(),
48 DType::Utf8(_) => rhs_const
49 .as_utf8()
50 .value()
51 .vortex_expect("RHS should not be null")
52 .as_str()
53 .as_bytes(),
54 _ => vortex_bail!("VarBinArray can only have type of Binary or Utf8"),
55 };
56
57 let buffer = if rhs_bytes.is_empty() {
58 match operator {
60 CompareOperator::Gte => BitBuffer::new_set(len), CompareOperator::Lt => BitBuffer::new_unset(len), CompareOperator::Eq | CompareOperator::Lte => {
63 let lhs_offsets = lhs.offsets().clone().execute::<PrimitiveArray>(ctx)?;
64 match_each_integer_ptype!(lhs_offsets.ptype(), |P| {
65 compare_offsets_to_empty::<P>(lhs_offsets, true)
66 })
67 }
68 CompareOperator::NotEq | CompareOperator::Gt => {
69 let lhs_offsets = lhs.offsets().clone().execute::<PrimitiveArray>(ctx)?;
70 match_each_integer_ptype!(lhs_offsets.ptype(), |P| {
71 compare_offsets_to_empty::<P>(lhs_offsets, false)
72 })
73 }
74 }
75 } else {
76 let lhs_offsets = lhs.offsets().clone().execute::<PrimitiveArray>(ctx)?;
77 match_each_integer_ptype!(lhs_offsets.ptype(), |P| {
78 compare_bytes_to_constant(
79 lhs_offsets.as_slice::<P>(),
80 lhs.bytes().as_slice(),
81 rhs_bytes,
82 operator,
83 )
84 })
85 };
86
87 Ok(Some(
88 BoolArray::new(
89 buffer,
90 lhs.validity()?.union_nullability(rhs.dtype().nullability()),
91 )
92 .into_array(),
93 ))
94 }
95}
96
97fn compare_offsets_to_empty<P: IntegerPType>(offsets: PrimitiveArray, eq: bool) -> BitBuffer {
98 let fn_ = if eq { P::eq } else { P::ne };
99 let offsets = offsets.as_slice::<P>();
100 BitBuffer::collect_bool(offsets.len() - 1, |idx| {
101 let left = unsafe { offsets.get_unchecked(idx) };
102 let right = unsafe { offsets.get_unchecked(idx + 1) };
103 fn_(left, right)
104 })
105}
106
107fn compare_bytes_to_constant<P: IntegerPType>(
111 offsets: &[P],
112 bytes: &[u8],
113 constant: &[u8],
114 operator: CompareOperator,
115) -> BitBuffer {
116 match operator {
117 CompareOperator::Eq => {
118 collect_lane_bits(offsets, |start, end| value_eq(bytes, start, end, constant))
119 }
120 CompareOperator::NotEq => {
121 collect_lane_bits(offsets, |start, end| !value_eq(bytes, start, end, constant))
122 }
123 CompareOperator::Gt => collect_lane_bits(offsets, |start, end| {
124 value_cmp(bytes, start, end, constant).is_gt()
125 }),
126 CompareOperator::Gte => collect_lane_bits(offsets, |start, end| {
127 value_cmp(bytes, start, end, constant).is_ge()
128 }),
129 CompareOperator::Lt => collect_lane_bits(offsets, |start, end| {
130 value_cmp(bytes, start, end, constant).is_lt()
131 }),
132 CompareOperator::Lte => collect_lane_bits(offsets, |start, end| {
133 value_cmp(bytes, start, end, constant).is_le()
134 }),
135 }
136}
137
138fn collect_lane_bits<P: IntegerPType>(
140 offsets: &[P],
141 predicate: impl Fn(usize, usize) -> bool,
142) -> BitBuffer {
143 BitBuffer::collect_bool(offsets.len() - 1, |idx| {
144 let start = unsafe { offsets.get_unchecked(idx) }.as_();
146 let end = unsafe { offsets.get_unchecked(idx + 1) }.as_();
147 predicate(start, end)
148 })
149}
150
151#[allow(clippy::inline_always)]
157#[inline(always)]
158fn value_eq(bytes: &[u8], start: usize, end: usize, constant: &[u8]) -> bool {
159 end.wrapping_sub(start) == constant.len()
163 && bytes.get(start..end).is_some_and(|value| value == constant)
164}
165
166#[allow(clippy::inline_always)]
169#[inline(always)]
170fn value_cmp(bytes: &[u8], start: usize, end: usize, constant: &[u8]) -> Ordering {
171 bytes.get(start..end).unwrap_or_default().cmp(constant)
172}
173
174#[cfg(test)]
175mod test {
176 use vortex_buffer::ByteBuffer;
177
178 use crate::IntoArray;
179 use crate::VortexSessionExecute;
180 use crate::array_session;
181 use crate::arrays::BoolArray;
182 use crate::arrays::ConstantArray;
183 use crate::arrays::VarBinArray;
184 use crate::arrays::VarBinViewArray;
185 use crate::assert_arrays_eq;
186 use crate::builtins::ArrayBuiltins;
187 use crate::dtype::DType;
188 use crate::dtype::Nullability;
189 use crate::scalar::Scalar;
190 use crate::scalar_fn::fns::operators::Operator;
191
192 #[test]
193 fn test_binary_compare() {
194 let mut ctx = array_session().create_execution_ctx();
195 let array = VarBinArray::from_iter(
196 [Some(b"abc".to_vec()), None, Some(b"def".to_vec())],
197 DType::Binary(Nullability::Nullable),
198 );
199 let result = array
200 .into_array()
201 .binary(
202 ConstantArray::new(
203 Scalar::binary(ByteBuffer::copy_from(b"abc"), Nullability::Nullable),
204 3,
205 )
206 .into_array(),
207 Operator::Eq,
208 )
209 .unwrap();
210
211 assert_arrays_eq!(
212 result,
213 BoolArray::from_iter([Some(true), None, Some(false)]),
214 &mut ctx
215 );
216 }
217
218 #[test]
219 fn varbinview_compare() {
220 let mut ctx = array_session().create_execution_ctx();
221 let array = VarBinArray::from_iter(
222 [Some(b"abc".to_vec()), None, Some(b"def".to_vec())],
223 DType::Binary(Nullability::Nullable),
224 );
225 let vbv = VarBinViewArray::from_iter(
226 [None, None, Some(b"def".to_vec())],
227 DType::Binary(Nullability::Nullable),
228 );
229 let result = array
230 .into_array()
231 .binary(vbv.into_array(), Operator::Eq)
232 .unwrap();
233
234 assert_arrays_eq!(
235 result,
236 BoolArray::from_iter([None, None, Some(true)]),
237 &mut ctx
238 );
239 }
240}
241
242#[cfg(test)]
243mod tests {
244 use vortex_buffer::ByteBuffer;
245
246 use crate::IntoArray;
247 use crate::VortexSessionExecute;
248 use crate::array_session;
249 use crate::arrays::BoolArray;
250 use crate::arrays::ConstantArray;
251 use crate::arrays::VarBinArray;
252 use crate::arrays::varbin::builder::VarBinBuilder;
253 use crate::assert_arrays_eq;
254 use crate::builtins::ArrayBuiltins;
255 use crate::dtype::DType;
256 use crate::dtype::Nullability;
257 use crate::scalar::Scalar;
258 use crate::scalar_fn::fns::operators::Operator;
259
260 #[test]
261 fn test_null_compare() {
262 let arr = VarBinArray::from_iter([Some("h")], DType::Utf8(Nullability::NonNullable));
263
264 let const_ = ConstantArray::new(Scalar::utf8("", Nullability::Nullable), 1);
265
266 let result = arr
267 .into_array()
268 .binary(const_.into_array(), Operator::Eq)
269 .unwrap();
270 assert_arrays_eq!(
271 result,
272 BoolArray::from_iter([Some(false)]),
273 &mut array_session().create_execution_ctx()
274 );
275 }
276
277 #[test]
283 fn varbin_i64_offsets_compare_constant() {
284 let mut ctx = array_session().create_execution_ctx();
285 let mut builder = VarBinBuilder::<i64>::with_capacity_in(
286 DType::Utf8(Nullability::NonNullable),
287 3,
288 vortex_buffer::BufferAllocatorRef::static_ref(),
289 );
290 builder.append_value(b"abc");
291 builder.append_value(b"xyz");
292 builder.append_value(b"abc");
293 let array = builder.finish_into_varbin();
294
295 let result = array
296 .into_array()
297 .binary(
298 ConstantArray::new(Scalar::utf8("abc", Nullability::NonNullable), 3).into_array(),
299 Operator::Eq,
300 )
301 .unwrap();
302
303 let expected = BoolArray::from_iter([true, false, true]);
304 assert_arrays_eq!(result, expected, &mut ctx);
305 }
306
307 #[test]
308 fn varbin_i64_offsets_compare_constant_binary() {
309 let mut ctx = array_session().create_execution_ctx();
310 let mut builder = VarBinBuilder::<i64>::with_capacity_in(
311 DType::Binary(Nullability::NonNullable),
312 3,
313 vortex_buffer::BufferAllocatorRef::static_ref(),
314 );
315 builder.append_value(b"abc");
316 builder.append_value(b"xyz");
317 builder.append_value(b"abc");
318 let array = builder.finish_into_varbin();
319
320 let result = array
321 .into_array()
322 .binary(
323 ConstantArray::new(
324 Scalar::binary(ByteBuffer::copy_from(b"abc"), Nullability::NonNullable),
325 3,
326 )
327 .into_array(),
328 Operator::Eq,
329 )
330 .unwrap();
331
332 let expected = BoolArray::from_iter([true, false, true]);
333 assert_arrays_eq!(result, expected, &mut ctx);
334 }
335}