vortex_fsst/compute/
cast.rs1use vortex_array::ArrayRef;
5use vortex_array::ArrayView;
6use vortex_array::ExecutionCtx;
7use vortex_array::IntoArray;
8use vortex_array::arrays::VarBinArray;
9use vortex_array::arrays::varbin::VarBinArraySlotsExt;
10use vortex_array::dtype::DType;
11use vortex_array::scalar_fn::fns::cast::CastKernel;
12use vortex_array::scalar_fn::fns::cast::CastReduce;
13use vortex_array::validity::Validity;
14use vortex_error::VortexResult;
15
16use crate::FSST;
17use crate::FSSTArrayExt;
18use crate::FSSTArraySlotsExt;
19
20fn build_with_codes_validity(
21 array: ArrayView<'_, FSST>,
22 dtype: &DType,
23 new_codes_validity: Validity,
24) -> VortexResult<ArrayRef> {
25 let codes = array.codes();
26 let new_codes = unsafe {
29 VarBinArray::new_unchecked_from_handle(
30 codes.offsets().clone(),
31 codes.bytes_handle().clone(),
32 codes.dtype().with_nullability(dtype.nullability()),
33 new_codes_validity,
34 )
35 };
36
37 Ok(unsafe {
38 FSST::new_unchecked_with_symbol_table(
39 dtype.clone(),
40 array.symbol_table(),
41 new_codes,
42 array.uncompressed_lengths().clone(),
43 )
44 }
45 .into_array())
46}
47
48impl CastReduce for FSST {
49 fn cast(array: ArrayView<'_, Self>, dtype: &DType) -> VortexResult<Option<ArrayRef>> {
50 if !array.dtype().eq_ignore_nullability(dtype) {
51 return Ok(None);
52 }
53
54 let codes = array.codes();
55 let Some(new_codes_validity) = codes
56 .validity()?
57 .trivially_cast_nullability(dtype.nullability(), codes.len())?
58 else {
59 return Ok(None);
60 };
61
62 Ok(Some(build_with_codes_validity(
63 array,
64 dtype,
65 new_codes_validity,
66 )?))
67 }
68}
69
70impl CastKernel for FSST {
71 fn cast(
72 array: ArrayView<'_, Self>,
73 dtype: &DType,
74 ctx: &mut ExecutionCtx,
75 ) -> VortexResult<Option<ArrayRef>> {
76 if !array.dtype().eq_ignore_nullability(dtype) {
77 return Ok(None);
78 }
79
80 let codes = array.codes();
81 let new_codes_validity =
82 codes
83 .validity()?
84 .cast_nullability(dtype.nullability(), codes.len(), ctx)?;
85
86 Ok(Some(build_with_codes_validity(
87 array,
88 dtype,
89 new_codes_validity,
90 )?))
91 }
92}
93
94#[cfg(test)]
95mod tests {
96 use std::sync::LazyLock;
97
98 use rstest::rstest;
99 use vortex_array::IntoArray;
100 use vortex_array::VortexSessionExecute;
101 use vortex_array::arrays::VarBinArray;
102 use vortex_array::compute::conformance::cast::test_cast_conformance;
103 use vortex_array::dtype::DType;
104 use vortex_array::dtype::Nullability;
105 use vortex_error::VortexResult;
106 use vortex_session::VortexSession;
107
108 use crate::fsst_compress;
109 use crate::fsst_train_compressor;
110 use crate::initialize;
111
112 static SESSION: LazyLock<VortexSession> = LazyLock::new(|| {
113 let session = vortex_array::array_session();
114 initialize(&session);
115 session
116 });
117
118 #[rstest]
119 #[case(VarBinArray::from_iter(
120 vec![Some("hello"), Some("world"), Some("hello world")],
121 DType::Utf8(Nullability::NonNullable)
122 ))]
123 #[case(VarBinArray::from_iter(
124 vec![Some("foo"), None, Some("bar"), Some("foobar")],
125 DType::Utf8(Nullability::Nullable)
126 ))]
127 #[case(VarBinArray::from_iter(
128 vec![Some("test")],
129 DType::Utf8(Nullability::NonNullable)
130 ))]
131 fn test_cast_fsst_conformance(#[case] array: VarBinArray) -> VortexResult<()> {
132 let mut ctx = SESSION.create_execution_ctx();
133 let array = array.into_array();
134 let compressor = fsst_train_compressor(&array, &mut ctx)?;
135 let fsst = fsst_compress(&array, &compressor, &mut ctx)?;
136 test_cast_conformance(&fsst.into_array(), &mut ctx);
137 Ok(())
138 }
139}