From d503966ae80beae6228776cf64d2ced3eb427f9d Mon Sep 17 00:00:00 2001
From: Maarten Pronk <git@evetion.nl>
Date: Tue, 26 Aug 2025 17:32:22 +0200
Subject: [PATCH] Add extension on HDF5.

---
 Project.toml   |   6 ++-
 ext/HDF5Ext.jl | 105 +++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 109 insertions(+), 2 deletions(-)
 create mode 100644 ext/HDF5Ext.jl

diff --git a/Project.toml b/Project.toml
index fa7fb8d..7f61db1 100644
--- a/Project.toml
+++ b/Project.toml
@@ -1,14 +1,13 @@
 name = "YAXArrayBase"
 uuid = "90b8fcef-0c2d-428d-9c56-5f86629e9d14"
 authors = ["Fabian Gans <fgans@bgc-jena.mpg.de>"]
-version = "0.7.6"
+version = "0.7.7"
 
 [deps]
 DataStructures = "864edb3b-99cc-5e75-8d2d-829cb0a9cfe8"
 Dates = "ade2ca70-3891-5945-98fb-dc099432e06a"
 Downloads = "f43a241f-c20a-4ad4-852c-f6b1247861c6"
 
-
 [compat]
 DataStructures = "0.17,0.18"
 julia = "1.9"
@@ -16,6 +15,7 @@ ArchGDAL = "0.10"
 AxisArrays = "0.4"
 AxisKeys = "0.2"
 DimensionalData = "0.27, 0.28, 0.29"
+HDF5 = "0.17"
 NetCDF = "0.11, 0.12"
 Zarr = "0.8, 0.9"
 
@@ -24,6 +24,7 @@ ArchGDALExt = "ArchGDAL"
 AxisArraysExt = "AxisArrays"
 AxisKeysExt = "AxisKeys"
 DimensionalDataExt = "DimensionalData"
+HDF5Ext = "HDF5"
 NamedDimsExt = "NamedDims"
 NetCDFExt = "NetCDF"
 ZarrExt = "Zarr"
@@ -33,6 +34,7 @@ ArchGDAL = "c9ce4bd3-c3d5-55b8-8973-c0e20141b8c3"
 AxisArrays = "39de3d68-74b9-583c-8d2d-e117c070f3a9"
 AxisKeys = "94b1ba4f-4ee9-5380-92f1-94cde586c3c5"
 DimensionalData = "0703355e-b756-11e9-17c0-8b28908087d0"
+HDF5 = "f67ccb44-e63f-5c2f-98bd-6dc0ccc4ba2f"
 NamedDims = "356022a1-0364-5f58-8944-0da4b18d706f"
 NetCDF = "30363a11-5582-574a-97bb-aa9a979735b9"
 Zarr = "0a941bbe-ad1d-11e8-39d9-ab76183a1d99"
diff --git a/ext/HDF5Ext.jl b/ext/HDF5Ext.jl
new file mode 100644
index 0000000..55aa828
--- /dev/null
+++ b/ext/HDF5Ext.jl
@@ -0,0 +1,105 @@
+module HDF5Ext
+import YAXArrayBase: YAXArrayBase as YAB
+using HDF5
+
+"""
+    HDF5Dataset
+
+Dataset backend to read HDF5 files using HDF5.jl
+"""
+struct HDF5Dataset
+    filename::String
+    mode::String
+    handle::Base.RefValue{Union{Nothing,HDF5.File}}
+end
+function HDF5Dataset(filename; mode="r")
+    HDF5Dataset(filename, mode, Ref{Union{Nothing,HDF5.File}}(nothing))
+end
+function dsopen(f, ds::HDF5Dataset)
+    if ds.handle[] === nothing || !Base.isopen(ds.handle[])
+        HDF5.h5open(f, ds.filename, ds.mode)
+    else
+        f(ds.handle[])
+    end
+end
+function YAB.open_dataset_handle(f, ds::HDF5Dataset)
+    if ds.handle[] === nothing || !Base.isopen(ds.handle[])
+        try
+            ds.handle[] = HDF5.h5open(ds.filename, ds.mode)
+            f(ds)
+        finally
+            ds.handle[] = nothing
+        end
+    else
+        f(ds)
+    end
+end
+
+function __init__()
+    @debug "new driver key :HDF5, updating backendlist."
+    YAB.backendlist[:HDF5] = HDF5Dataset
+    push!(YAB.backendregex, r".h5$" => HDF5Dataset)
+end
+
+function get_all_paths(file, prefix="")
+    paths = String[]
+
+    for key in keys(file)
+        full_path = isempty(prefix) ? key : "$prefix/$key"
+        obj = file[key]
+
+        if isa(obj, HDF5.Dataset)
+            push!(paths, full_path)
+        elseif isa(obj, HDF5.Group)
+            append!(paths, get_all_paths(obj, full_path))
+        end
+    end
+
+    return paths
+end
+
+function get_dims(f, var)
+    dims = String[]
+    ds = f[var]
+    if haskey(ds, "DIMENSION_LIST")
+        dimension_list = read_attribute(ds, "DIMENSION_LIST")
+        for dimensions in dimension_list
+            for dim_ref in dimensions
+                push!(dims, HDF5.name(f[dim_ref]))
+            end
+        end
+    end
+    return dims
+end
+
+"Return a list of variable names"
+YAB.get_varnames(ds::HDF5Dataset) = dsopen(get_all_paths, ds)
+
+"Return a list of dimension names for a given variable"
+YAB.get_var_dims(ds::HDF5Dataset, name) = dsopen(x -> get_dims(x, name), ds)
+
+function get_var_attrs(file, name)
+    attributes = Dict(attrs(file[name]))
+    pop!(attributes, "DIMENSION_LIST", nothing)  # Remove DIMENSION_LIST if present
+    return attributes
+end
+
+"Return a dict with the attributes for a given variable"
+YAB.get_var_attrs(ds::HDF5Dataset, name) = dsopen(v -> get_var_attrs(v, name), ds)
+
+"Return a dict with global attributes for the dataset"
+YAB.get_global_attrs(ds::HDF5Dataset) = dsopen(h5 -> Dict(attrs(h5)), ds)
+
+"Return a DiskArray handle to a dataset"
+function YAB.get_var_handle(ds::HDF5Dataset, i; persist=true)
+    if persist || ds.handle[] === nothing
+        s, et = NetCDF.open(j -> (size(j), eltype(j)), ds.filename, i)
+        NetCDFVariable{et,length(s)}(ds.filename, i, s)
+    else
+        ds.handle[][i]
+    end
+end
+Base.haskey(ds::HDF5Dataset, k) = dsopen(h5 -> haskey(h5, k), ds)
+
+
+end