vortex_array/builders/
extension.rs1use std::any::Any;
5
6use vortex_error::VortexResult;
7use vortex_error::vortex_ensure;
8use vortex_mask::Mask;
9
10use crate::ArrayRef;
11use crate::ExecutionCtx;
12use crate::IntoArray;
13use crate::arrays::ExtensionArray;
14use crate::arrays::extension::ExtensionArrayExt;
15use crate::builders::ArrayBuilder;
16use crate::builders::DEFAULT_BUILDER_CAPACITY;
17use crate::builders::builder_with_capacity;
18use crate::canonical::Canonical;
19use crate::dtype::DType;
20use crate::dtype::extension::ExtDTypeRef;
21use crate::scalar::ExtScalar;
22use crate::scalar::Scalar;
23
24pub struct ExtensionBuilder {
26 dtype: DType,
27 storage: Box<dyn ArrayBuilder>,
28}
29
30impl ExtensionBuilder {
31 pub fn new(ext_dtype: ExtDTypeRef) -> Self {
33 Self::with_capacity(ext_dtype, DEFAULT_BUILDER_CAPACITY)
34 }
35
36 pub fn with_capacity(ext_dtype: ExtDTypeRef, capacity: usize) -> Self {
38 Self {
39 storage: builder_with_capacity(ext_dtype.storage_dtype(), capacity),
40 dtype: DType::Extension(ext_dtype),
41 }
42 }
43
44 pub fn append_value(&mut self, value: ExtScalar) -> VortexResult<()> {
46 self.storage.append_scalar(&value.to_storage_scalar())
47 }
48
49 pub(crate) fn append_extension_array(
52 &mut self,
53 array: &ExtensionArray,
54 ctx: &mut ExecutionCtx,
55 ) -> VortexResult<()> {
56 array
57 .storage_array()
58 .append_to_builder(self.storage.as_mut(), ctx)
59 }
60
61 pub fn finish_into_extension(&mut self) -> ExtensionArray {
63 let storage = self.storage.finish();
64 ExtensionArray::new(self.ext_dtype(), storage)
65 }
66
67 fn ext_dtype(&self) -> ExtDTypeRef {
71 if let DType::Extension(ext_dtype) = &self.dtype {
72 ext_dtype.clone()
73 } else {
74 unreachable!()
75 }
76 }
77}
78
79impl ArrayBuilder for ExtensionBuilder {
80 fn as_any(&self) -> &dyn Any {
81 self
82 }
83
84 fn as_any_mut(&mut self) -> &mut dyn Any {
85 self
86 }
87
88 fn dtype(&self) -> &DType {
89 &self.dtype
90 }
91
92 fn len(&self) -> usize {
93 self.storage.len()
94 }
95
96 fn append_zeros(&mut self, n: usize) {
97 self.storage.append_zeros(n)
98 }
99
100 unsafe fn append_nulls_unchecked(&mut self, n: usize) {
101 self.storage.append_nulls(n)
102 }
103
104 fn append_scalar(&mut self, scalar: &Scalar) -> VortexResult<()> {
105 vortex_ensure!(
106 scalar.dtype() == self.dtype(),
107 "ExtensionBuilder expected scalar with dtype {}, got {}",
108 self.dtype(),
109 scalar.dtype()
110 );
111
112 self.append_value(scalar.as_extension())
113 }
114
115 fn reserve_exact(&mut self, capacity: usize) {
116 self.storage.reserve_exact(capacity)
117 }
118
119 unsafe fn set_validity_unchecked(&mut self, validity: Mask) {
120 unsafe { self.storage.set_validity_unchecked(validity) };
121 }
122
123 fn finish(&mut self) -> ArrayRef {
124 self.finish_into_extension().into_array()
125 }
126
127 fn finish_into_canonical(&mut self, _ctx: &mut ExecutionCtx) -> Canonical {
128 Canonical::Extension(self.finish_into_extension())
129 }
130}
131
132#[cfg(test)]
133mod tests {
134 use super::*;
135 use crate::VortexSessionExecute;
136 use crate::array_session;
137 use crate::arrays::PrimitiveArray;
138 use crate::assert_arrays_eq;
139 use crate::builders::ArrayBuilder;
140 use crate::dtype::Nullability;
141 use crate::extension::datetime::Date;
142 use crate::extension::datetime::TimeUnit;
143 use crate::scalar::Scalar;
144
145 #[test]
146 fn test_append_scalar() {
147 let mut ctx = array_session().create_execution_ctx();
148 let ext_dtype = Date::new(TimeUnit::Days, Nullability::Nullable).erased();
149
150 let mut builder = ExtensionBuilder::new(ext_dtype.clone());
151
152 let storage1 = Scalar::from(Some(42i32));
154 let ext_scalar1 = Scalar::extension::<Date>(TimeUnit::Days, storage1);
155 builder.append_scalar(&ext_scalar1).unwrap();
156
157 let storage2 = Scalar::from(Some(84i32));
159 let ext_scalar2 = Scalar::extension::<Date>(TimeUnit::Days, storage2);
160 builder.append_scalar(&ext_scalar2).unwrap();
161
162 let null_storage = Scalar::null(DType::Primitive(
164 crate::dtype::PType::I32,
165 Nullability::Nullable,
166 ));
167 let null_scalar = Scalar::extension::<Date>(TimeUnit::Days, null_storage);
168 builder.append_scalar(&null_scalar).unwrap();
169
170 let array = builder.finish_into_extension();
171 let expected = ExtensionArray::new(
172 ext_dtype.clone(),
173 PrimitiveArray::from_option_iter([Some(42i32), Some(84), None]).into_array(),
174 );
175
176 assert_arrays_eq!(&array, &expected, &mut ctx);
177 assert_eq!(array.len(), 3);
178
179 let mut builder = ExtensionBuilder::new(ext_dtype);
181 let wrong_scalar = Scalar::from(true);
182 assert!(builder.append_scalar(&wrong_scalar).is_err());
183 }
184}