vortex_sampling_compressor/compressors/
alp.rs1use vortex_alp::{
2 alp_encode_components, match_each_alp_float_ptype, ALPArray, ALPEncoding, ALPRDEncoding,
3};
4use vortex_array::aliases::hash_set::HashSet;
5use vortex_array::array::PrimitiveArray;
6use vortex_array::variants::PrimitiveArrayTrait;
7use vortex_array::{Array, Encoding, EncodingId, IntoArray, IntoArrayVariant};
8use vortex_dtype::PType;
9use vortex_error::VortexResult;
10use vortex_fastlanes::BitPackedEncoding;
11
12use super::alp_rd::ALPRDCompressor;
13use crate::compressors::{CompressedArray, CompressionTree, EncodingCompressor};
14use crate::{constants, SamplingCompressor};
15
16#[derive(Debug)]
17pub struct ALPCompressor;
18
19impl EncodingCompressor for ALPCompressor {
20 fn id(&self) -> &str {
21 ALPEncoding::ID.as_ref()
22 }
23
24 fn cost(&self) -> u8 {
25 constants::ALP_COST
26 }
27
28 fn can_compress(&self, array: &Array) -> Option<&dyn EncodingCompressor> {
29 let parray = PrimitiveArray::maybe_from(array)?;
31
32 if !matches!(parray.ptype(), PType::F32 | PType::F64) {
34 return None;
35 }
36
37 Some(self)
38 }
39
40 fn compress<'a>(
41 &'a self,
42 array: &Array,
43 like: Option<CompressionTree<'a>>,
44 ctx: SamplingCompressor<'a>,
45 ) -> VortexResult<CompressedArray<'a>> {
46 let parray = array.clone().into_primitive()?;
47
48 let (exponents, encoded, patches) = match_each_alp_float_ptype!(
49 parray.ptype(), |$T| {
50 alp_encode_components::<$T>(&parray, None)
51 });
52
53 let compressed_encoded = ctx
54 .named("packed")
55 .excluding(self)
56 .compress(&encoded, like.as_ref().and_then(|l| l.child(0)))?;
57
58 let compressed_patches = patches
60 .map(|p| {
61 ctx.auxiliary("patches")
62 .excluding(self)
63 .including(&ALPRDCompressor)
64 .compress_patches(p)
65 })
66 .transpose()?;
67
68 Ok(CompressedArray::compressed(
69 ALPArray::try_new(compressed_encoded.array, exponents, compressed_patches)?
70 .into_array(),
71 Some(CompressionTree::new(self, vec![compressed_encoded.path])),
72 array,
73 ))
74 }
75
76 fn used_encodings(&self) -> HashSet<EncodingId> {
77 HashSet::from([
78 ALPEncoding::ID,
79 ALPRDEncoding::ID,
81 BitPackedEncoding::ID,
82 ])
83 }
84}