vortex_array/arrays/varbin/
builder.rs1use num_traits::AsPrimitive;
5use vortex_buffer::{BitBufferMut, BufferMut};
6use vortex_dtype::{DType, IntegerPType};
7use vortex_error::vortex_panic;
8
9use crate::IntoArray;
10use crate::arrays::primitive::PrimitiveArray;
11use crate::arrays::varbin::VarBinArray;
12use crate::validity::Validity;
13
14pub struct VarBinBuilder<O: IntegerPType> {
15 offsets: BufferMut<O>,
16 data: BufferMut<u8>,
17 validity: BitBufferMut,
18}
19
20impl<O: IntegerPType> Default for VarBinBuilder<O> {
21 fn default() -> Self {
22 Self::new()
23 }
24}
25
26impl<O: IntegerPType> VarBinBuilder<O> {
27 pub fn new() -> Self {
28 Self::with_capacity(0)
29 }
30
31 pub fn with_capacity(len: usize) -> Self {
32 let mut offsets = BufferMut::with_capacity(len + 1);
33 offsets.push(O::zero());
34 Self {
35 offsets,
36 data: BufferMut::empty(),
37 validity: BitBufferMut::with_capacity(len),
38 }
39 }
40
41 #[inline]
42 pub fn append(&mut self, value: Option<&[u8]>) {
43 match value {
44 Some(v) => self.append_value(v),
45 None => self.append_null(),
46 }
47 }
48
49 #[inline]
50 pub fn append_value(&mut self, value: impl AsRef<[u8]>) {
51 let slice = value.as_ref();
52 self.offsets
53 .push(O::from(self.data.len() + slice.len()).unwrap_or_else(|| {
54 vortex_panic!(
55 "Failed to convert sum of {} and {} to offset of type {}",
56 self.data.len(),
57 slice.len(),
58 std::any::type_name::<O>()
59 )
60 }));
61 self.data.extend_from_slice(slice);
62 self.validity.append_true();
63 }
64
65 #[inline]
66 pub fn append_null(&mut self) {
67 self.offsets.push(self.offsets[self.offsets.len() - 1]);
68 self.validity.append_false();
69 }
70
71 #[inline]
72 pub fn append_n_nulls(&mut self, n: usize) {
73 self.offsets.push_n(self.offsets[self.offsets.len() - 1], n);
74 self.validity.append_n(false, n);
75 }
76
77 #[inline]
78 pub fn append_values(&mut self, values: &[u8], end_offsets: impl Iterator<Item = O>, num: usize)
79 where
80 O: 'static,
81 usize: AsPrimitive<O>,
82 {
83 self.offsets
84 .extend(end_offsets.map(|offset| offset + self.data.len().as_()));
85 self.data.extend_from_slice(values);
86 self.validity.append_n(true, num);
87 }
88
89 pub fn finish(self, dtype: DType) -> VarBinArray {
90 let offsets = PrimitiveArray::new(self.offsets.freeze(), Validity::NonNullable);
91 let nulls = self.validity.freeze();
92
93 let validity = Validity::from_bit_buffer(nulls, dtype.nullability());
94
95 unsafe {
101 VarBinArray::new_unchecked(offsets.into_array(), self.data.freeze(), dtype, validity)
102 }
103 }
104}
105
106#[cfg(test)]
107mod tests {
108 use vortex_dtype::DType;
109 use vortex_dtype::Nullability::Nullable;
110 use vortex_scalar::Scalar;
111
112 use crate::arrays::varbin::builder::VarBinBuilder;
113
114 #[test]
115 fn test_builder() {
116 let mut builder = VarBinBuilder::<i32>::with_capacity(0);
117 builder.append(Some(b"hello"));
118 builder.append(None);
119 builder.append(Some(b"world"));
120 let array = builder.finish(DType::Utf8(Nullable));
121
122 assert_eq!(array.len(), 3);
123 assert_eq!(array.dtype().nullability(), Nullable);
124 assert_eq!(
125 array.scalar_at(0),
126 Scalar::utf8("hello".to_string(), Nullable)
127 );
128 assert!(array.scalar_at(1).is_null());
129 }
130}