Module ZarrSource

zarr Provides an Ocaml implementation of the Zarr version 3 storage format specification. It supports creation of arrays and groups as well as chunking arrays along any dimension. One can store a Zarr hierarchy in memory or on disk. Zarr also supports reading zarr hierarchies created using other implementations, as long as they are spec-compliant.

Consult the examples and limitations for more info.

References

• The Zarr Version 3 specification.
• Zarr community site.

Node

Metadata

Storage

Codecs

Indexing

Utils

Ndarray

Examples

Create, read & write array.

Here we show how the library's asynchronous API using Lwt's concurrency monad can be used.

open Zarr.Metadata
open Zarr.Node
open Zarr.Codecs
open Zarr_lwt.Storage
open FilesystemStore.Deferred.Syntax

let _ =
  Lwt_main.run begin
    let store = FilesystemStore.create "testdata.zarr" in
    let group_node = GroupNode.of_path "/some/group" in
    let* () = FilesystemStore.create_group store group_node in
    let array_node = ArrayNode.(group_node / "name") in
    let* () = FilesystemStore.create_array
      ~codecs:[`Bytes BE] ~shape:[|100; 100; 50|] ~chunks:[|10; 15; 20|]
      Bigarray.Float32 Float.neg_infinity array_node store in
    let slice = Owl_types.[|R [0; 20]; I 10; R []|] in
    let* x = FilesystemStore.read_array store array_node slice Bigarray.Float32 in
    let x' = Owl.Dense.Ndarray.Generic.map (fun _ -> Owl_stats_dist.uniform_rvs 0. 10.) x
    in FilesystemStore.write_array store array_node slice x'
  end

Extension Points

This library also provides custom extensions not defined in the version 3 specification. These are tabulated below:

Limitations

Although this implementation tries to be spec compliant, it does come with a few limitations:

• Ocaml does not have support for unsigned integers as array data types and thus this library cannot support reading values of datatypes uint32, uint64 and complex128.