vortex_array/builders/
mod.rs

1//! Builders for Vortex arrays.
2//!
3//! Every logical type in Vortex has a canonical (uncompressed) in-memory encoding. This module
4//! provides pre-allocated builders to construct new canonical arrays.
5//!
6//! ## Example:
7//!
8//! ```
9//! use vortex_array::builders::{builder_with_capacity, ArrayBuilderExt};
10//! use vortex_array::compute::scalar_at;
11//! use vortex_dtype::{DType, Nullability};
12//!
13//! // Create a new builder for string data.
14//! let mut builder = builder_with_capacity(&DType::Utf8(Nullability::NonNullable), 4);
15//!
16//! builder.append_scalar(&"a".into()).unwrap();
17//! builder.append_scalar(&"b".into()).unwrap();
18//! builder.append_scalar(&"c".into()).unwrap();
19//! builder.append_scalar(&"d".into()).unwrap();
20//!
21//! let strings = builder.finish();
22//!
23//! assert_eq!(scalar_at(&strings, 0).unwrap(), "a".into());
24//! assert_eq!(scalar_at(&strings, 1).unwrap(), "b".into());
25//! assert_eq!(scalar_at(&strings, 2).unwrap(), "c".into());
26//! assert_eq!(scalar_at(&strings, 3).unwrap(), "d".into());
27//! ```
28
29mod bool;
30mod decimal;
31mod extension;
32mod lazy_validity_builder;
33mod list;
34mod null;
35mod primitive;
36mod struct_;
37mod varbinview;
38
39use std::any::Any;
40
41pub use bool::*;
42pub use decimal::*;
43pub use extension::*;
44pub use list::*;
45pub use null::*;
46pub use primitive::*;
47pub use varbinview::*;
48use vortex_dtype::{DType, match_each_native_ptype};
49use vortex_error::{VortexResult, vortex_bail, vortex_err};
50use vortex_mask::Mask;
51use vortex_scalar::{
52    BinaryScalar, BoolScalar, ExtScalar, ListScalar, PrimitiveScalar, Scalar, ScalarValue,
53    StructScalar, Utf8Scalar,
54};
55
56use crate::builders::struct_::StructBuilder;
57use crate::{Array, ArrayRef};
58
59pub trait ArrayBuilder: Send {
60    fn as_any(&self) -> &dyn Any;
61
62    fn as_any_mut(&mut self) -> &mut dyn Any;
63
64    fn dtype(&self) -> &DType;
65
66    fn len(&self) -> usize;
67
68    fn is_empty(&self) -> bool {
69        self.len() == 0
70    }
71
72    /// Append a "zero" value to the array.
73    fn append_zero(&mut self) {
74        self.append_zeros(1)
75    }
76
77    /// Appends n "zero" values to the array.
78    fn append_zeros(&mut self, n: usize);
79
80    /// Append a "null" value to the array.
81    fn append_null(&mut self) {
82        self.append_nulls(1)
83    }
84
85    /// Appends n "null" values to the array.
86    fn append_nulls(&mut self, n: usize);
87
88    /// Extends the array with the provided array, canonicalizing if necessary.
89    fn extend_from_array(&mut self, array: &dyn Array) -> VortexResult<()>;
90
91    /// Ensure that the builder can hold at least `capacity` number of items
92    fn ensure_capacity(&mut self, capacity: usize);
93
94    /// Override builders validity with the one provided
95    fn set_validity(&mut self, validity: Mask);
96
97    /// Constructs an Array from the builder components.
98    ///
99    /// # Panics
100    ///
101    /// This function may panic if the builder's methods are called with invalid arguments. If only
102    /// the methods on this interface are used, the builder should not panic. However, specific
103    /// builders have interfaces that may be misued. For example, if the number of values in a
104    /// [PrimitiveBuilder]'s [vortex_buffer::BufferMut] does not match the number of validity bits,
105    /// the PrimitiveBuilder's [Self::finish] will panic.
106    fn finish(&mut self) -> ArrayRef;
107}
108
109/// Construct a new canonical builder for the given [`DType`].
110///
111///
112/// # Example
113///
114/// ```
115/// use vortex_array::builders::{builder_with_capacity, ArrayBuilderExt};
116/// use vortex_array::compute::scalar_at;
117/// use vortex_dtype::{DType, Nullability};
118///
119/// // Create a new builder for string data.
120/// let mut builder = builder_with_capacity(&DType::Utf8(Nullability::NonNullable), 4);
121///
122/// builder.append_scalar(&"a".into()).unwrap();
123/// builder.append_scalar(&"b".into()).unwrap();
124/// builder.append_scalar(&"c".into()).unwrap();
125/// builder.append_scalar(&"d".into()).unwrap();
126///
127/// let strings = builder.finish();
128///
129/// assert_eq!(scalar_at(&strings, 0).unwrap(), "a".into());
130/// assert_eq!(scalar_at(&strings, 1).unwrap(), "b".into());
131/// assert_eq!(scalar_at(&strings, 2).unwrap(), "c".into());
132/// assert_eq!(scalar_at(&strings, 3).unwrap(), "d".into());
133/// ```
134pub fn builder_with_capacity(dtype: &DType, capacity: usize) -> Box<dyn ArrayBuilder> {
135    match dtype {
136        DType::Null => Box::new(NullBuilder::new()),
137        DType::Bool(n) => Box::new(BoolBuilder::with_capacity(*n, capacity)),
138        DType::Primitive(ptype, n) => {
139            match_each_native_ptype!(ptype, |$P| {
140                Box::new(PrimitiveBuilder::<$P>::with_capacity(*n, capacity))
141            })
142        }
143        DType::Decimal(..) => {
144            todo!("(aduffy): implement Decimal builder")
145        }
146        DType::Utf8(n) => Box::new(VarBinViewBuilder::with_capacity(DType::Utf8(*n), capacity)),
147        DType::Binary(n) => Box::new(VarBinViewBuilder::with_capacity(
148            DType::Binary(*n),
149            capacity,
150        )),
151        DType::Struct(struct_dtype, n) => Box::new(StructBuilder::with_capacity(
152            struct_dtype.clone(),
153            *n,
154            capacity,
155        )),
156        DType::List(dtype, n) => Box::new(ListBuilder::<u64>::with_capacity(
157            dtype.clone(),
158            *n,
159            capacity,
160        )),
161        DType::Extension(ext_dtype) => {
162            Box::new(ExtensionBuilder::with_capacity(ext_dtype.clone(), capacity))
163        }
164    }
165}
166
167pub trait ArrayBuilderExt: ArrayBuilder {
168    /// A generic function to append a scalar value to the builder.
169    fn append_scalar_value(&mut self, value: ScalarValue) -> VortexResult<()> {
170        if value.is_null() {
171            self.append_null();
172            Ok(())
173        } else {
174            self.append_scalar(&Scalar::new(self.dtype().clone(), value))
175        }
176    }
177
178    /// A generic function to append a scalar to the builder.
179    fn append_scalar(&mut self, scalar: &Scalar) -> VortexResult<()> {
180        if scalar.dtype() != self.dtype() {
181            vortex_bail!(
182                "Builder has dtype {:?}, scalar has {:?}",
183                self.dtype(),
184                scalar.dtype()
185            )
186        }
187        match scalar.dtype() {
188            DType::Null => self
189                .as_any_mut()
190                .downcast_mut::<NullBuilder>()
191                .ok_or_else(|| vortex_err!("Cannot append null scalar to non-null builder"))?
192                .append_null(),
193            DType::Bool(_) => self
194                .as_any_mut()
195                .downcast_mut::<BoolBuilder>()
196                .ok_or_else(|| vortex_err!("Cannot append bool scalar to non-bool builder"))?
197                .append_option(BoolScalar::try_from(scalar)?.value()),
198            DType::Primitive(ptype, ..) => {
199                match_each_native_ptype!(ptype, |$P| {
200                    self
201                    .as_any_mut()
202                    .downcast_mut::<PrimitiveBuilder<$P>>()
203                    .ok_or_else(|| {
204                        vortex_err!("Cannot append primitive scalar to non-primitive builder")
205                    })?
206                    .append_option(PrimitiveScalar::try_from(scalar)?.typed_value::<$P>())
207                })
208            }
209            DType::Decimal(..) => todo!("(aduffy): implement Decimal builder"),
210            DType::Utf8(_) => self
211                .as_any_mut()
212                .downcast_mut::<VarBinViewBuilder>()
213                .ok_or_else(|| vortex_err!("Cannot append utf8 scalar to non-utf8 builder"))?
214                .append_option(Utf8Scalar::try_from(scalar)?.value()),
215            DType::Binary(_) => self
216                .as_any_mut()
217                .downcast_mut::<VarBinViewBuilder>()
218                .ok_or_else(|| vortex_err!("Cannot append binary scalar to non-binary builder"))?
219                .append_option(BinaryScalar::try_from(scalar)?.value()),
220            DType::Struct(..) => self
221                .as_any_mut()
222                .downcast_mut::<StructBuilder>()
223                .ok_or_else(|| vortex_err!("Cannot append struct scalar to non-struct builder"))?
224                .append_value(StructScalar::try_from(scalar)?)?,
225            DType::List(..) => self
226                .as_any_mut()
227                .downcast_mut::<ListBuilder<u64>>()
228                .ok_or_else(|| vortex_err!("Cannot append list scalar to non-list builder"))?
229                .append_value(ListScalar::try_from(scalar)?)?,
230            DType::Extension(..) => self
231                .as_any_mut()
232                .downcast_mut::<ExtensionBuilder>()
233                .ok_or_else(|| {
234                    vortex_err!("Cannot append extension scalar to non-extension builder")
235                })?
236                .append_value(ExtScalar::try_from(scalar)?)?,
237        }
238        Ok(())
239    }
240}
241
242impl<T: ?Sized + ArrayBuilder> ArrayBuilderExt for T {}
vortex_array/builders/mod.rs

vortex_array/builders/
mod.rs
