vortex_array/arrays/varbin/
builder.rs1use num_traits::AsPrimitive;
5use vortex_buffer::BitBufferMut;
6use vortex_buffer::BufferMut;
7use vortex_dtype::DType;
8use vortex_dtype::IntegerPType;
9use vortex_error::vortex_panic;
10
11use crate::IntoArray;
12use crate::arrays::primitive::PrimitiveArray;
13use crate::arrays::varbin::VarBinArray;
14use crate::validity::Validity;
15
16pub struct VarBinBuilder<O: IntegerPType> {
17 offsets: BufferMut<O>,
18 data: BufferMut<u8>,
19 validity: BitBufferMut,
20}
21
22impl<O: IntegerPType> Default for VarBinBuilder<O> {
23 fn default() -> Self {
24 Self::new()
25 }
26}
27
28impl<O: IntegerPType> VarBinBuilder<O> {
29 pub fn new() -> Self {
30 Self::with_capacity(0)
31 }
32
33 pub fn with_capacity(len: usize) -> Self {
34 let mut offsets = BufferMut::with_capacity(len + 1);
35 offsets.push(O::zero());
36 Self {
37 offsets,
38 data: BufferMut::empty(),
39 validity: BitBufferMut::with_capacity(len),
40 }
41 }
42
43 #[inline]
44 pub fn append(&mut self, value: Option<&[u8]>) {
45 match value {
46 Some(v) => self.append_value(v),
47 None => self.append_null(),
48 }
49 }
50
51 #[inline]
52 pub fn append_value(&mut self, value: impl AsRef<[u8]>) {
53 let slice = value.as_ref();
54 self.offsets
55 .push(O::from(self.data.len() + slice.len()).unwrap_or_else(|| {
56 vortex_panic!(
57 "Failed to convert sum of {} and {} to offset of type {}",
58 self.data.len(),
59 slice.len(),
60 std::any::type_name::<O>()
61 )
62 }));
63 self.data.extend_from_slice(slice);
64 self.validity.append_true();
65 }
66
67 #[inline]
68 pub fn append_null(&mut self) {
69 self.offsets.push(self.offsets[self.offsets.len() - 1]);
70 self.validity.append_false();
71 }
72
73 #[inline]
74 pub fn append_n_nulls(&mut self, n: usize) {
75 self.offsets.push_n(self.offsets[self.offsets.len() - 1], n);
76 self.validity.append_n(false, n);
77 }
78
79 #[inline]
80 pub fn append_values(&mut self, values: &[u8], end_offsets: impl Iterator<Item = O>, num: usize)
81 where
82 O: 'static,
83 usize: AsPrimitive<O>,
84 {
85 self.offsets
86 .extend(end_offsets.map(|offset| offset + self.data.len().as_()));
87 self.data.extend_from_slice(values);
88 self.validity.append_n(true, num);
89 }
90
91 pub fn finish(self, dtype: DType) -> VarBinArray {
92 let offsets = PrimitiveArray::new(self.offsets.freeze(), Validity::NonNullable);
93 let nulls = self.validity.freeze();
94
95 let validity = Validity::from_bit_buffer(nulls, dtype.nullability());
96
97 unsafe {
103 VarBinArray::new_unchecked(offsets.into_array(), self.data.freeze(), dtype, validity)
104 }
105 }
106}
107
108#[cfg(test)]
109mod tests {
110 use vortex_dtype::DType;
111 use vortex_dtype::Nullability::Nullable;
112 use vortex_scalar::Scalar;
113
114 use crate::arrays::varbin::builder::VarBinBuilder;
115
116 #[test]
117 fn test_builder() {
118 let mut builder = VarBinBuilder::<i32>::with_capacity(0);
119 builder.append(Some(b"hello"));
120 builder.append(None);
121 builder.append(Some(b"world"));
122 let array = builder.finish(DType::Utf8(Nullable));
123
124 assert_eq!(array.len(), 3);
125 assert_eq!(array.dtype().nullability(), Nullable);
126 assert_eq!(
127 array.scalar_at(0),
128 Scalar::utf8("hello".to_string(), Nullable)
129 );
130 assert!(array.scalar_at(1).is_null());
131 }
132}