fix: support for Zarr V2 zstd encoded data created with numcodecs < 0.13 (#121)

Author: LDeakin

CHANGELOG.md

@@ -13,6 +13,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Add `ArrayShardedExt::is_exclusively_sharded`
 - Add `ArrayShardedReadableExtCache::array_is_exclusively_sharded`
 - Add `Vlen{Array,Bytes,Utf8}Codec`, replacing `VlenV2Codec`
+- Add `ZstdCodecConfigurationNumCodecs`
+  - Adds support for Zarr V2 `zstd` encoded data created with `numcodecs` < 0.13
 
 ### Changed
 - **Breaking**: Seal `Array` extension traits: `ArraySharded[Readable]Ext` and `ArrayChunkCacheExt`

zarrs/tests/data/v3/array_zstd.zarr/zarr.json

@@ -32,8 +32,8 @@
     {
       "name": "zstd",
       "configuration": {
-        "checksum": false,
-        "level": 5
+        "level": 5,
+        "checksum": false
       }
     }
   ],

zarrs_metadata/src/v2/array/codec/zstd.rs

@@ -1 +1,40 @@
+use derive_more::derive::{Display, From};
+use serde::{Deserialize, Serialize};
+
 pub use crate::v3::array::codec::zstd::ZstdCodecConfigurationV1;
+
+use crate::v3::array::codec::zstd::{ZstdCodecConfiguration, ZstdCompressionLevel};
+
+type ZstdCodecConfigurationNumCodecs0_13 = ZstdCodecConfigurationV1;
+
+/// A wrapper to handle various versions of `zstd` codec configuration parameters.
+#[derive(Serialize, Deserialize, Clone, Eq, PartialEq, Debug, Display, From)]
+#[serde(untagged)]
+pub enum ZstdCodecConfigurationNumCodecs {
+    /// `numcodecs` version 0.13.
+    V0_13(ZstdCodecConfigurationNumCodecs0_13),
+    /// `numcodecs` version 0.1.
+    V0_1(ZstdCodecConfigurationNumCodecs0_1),
+}
+
+/// Configuration parameters for the `zstd` codec (`numcodecs` version 0.1).
+#[derive(Serialize, Deserialize, Clone, Eq, PartialEq, Debug, Display)]
+#[serde(deny_unknown_fields)]
+#[display("{}", serde_json::to_string(self).unwrap_or_default())]
+pub struct ZstdCodecConfigurationNumCodecs0_1 {
+    /// The compression level.
+    pub level: ZstdCompressionLevel,
+}
+
+/// Convert [`ZstdCodecConfigurationNumCodecs`] to [`ZstdCodecConfiguration`].
+#[must_use]
+pub fn codec_zstd_v2_numcodecs_to_v3(
+    zstd: &ZstdCodecConfigurationNumCodecs,
+) -> ZstdCodecConfiguration {
+    match zstd {
+        ZstdCodecConfigurationNumCodecs::V0_13(zstd) => ZstdCodecConfiguration::V1(zstd.clone()),
+        ZstdCodecConfigurationNumCodecs::V0_1(zstd) => {
+            ZstdCodecConfiguration::V1(ZstdCodecConfigurationV1::new(zstd.level.clone(), false))
+        }
+    }
+}

zarrs_metadata/src/v2_to_v3.rs

@@ -6,6 +6,7 @@ use crate::{
             codec::{
                 blosc::{codec_blosc_v2_numcodecs_to_v3, BloscCodecConfigurationNumcodecs},
                 zfpy::{codec_zfpy_v2_numcodecs_to_v3, ZfpyCodecConfigurationNumcodecs},
+                zstd::{codec_zstd_v2_numcodecs_to_v3, ZstdCodecConfigurationNumCodecs},
             },
             data_type_metadata_v2_to_endianness, ArrayMetadataV2Order, DataTypeMetadataV2,
             DataTypeMetadataV2InvalidEndiannessError, FillValueMetadataV2,
@@ -214,6 +215,16 @@ pub fn array_metadata_v2_to_v3(
                     &configuration,
                 )?);
             }
+            crate::v3::array::codec::zstd::IDENTIFIER => {
+                let zstd = serde_json::from_value::<ZstdCodecConfigurationNumCodecs>(
+                    serde_json::to_value(compressor.configuration())?,
+                )?;
+                let configuration = codec_zstd_v2_numcodecs_to_v3(&zstd);
+                codecs.push(MetadataV3::new_with_serializable_configuration(
+                    crate::v3::array::codec::zstd::IDENTIFIER,
+                    &configuration,
+                )?);
+            }
             _ => codecs.push(MetadataV3::new_with_configuration(
                 compressor.id(),
                 compressor.configuration().clone(),