Dmrpp migration

conftest.py

@@ -339,6 +339,15 @@ def hdf5_empty(tmp_path: Path) -> str:
     return str(filepath)
 
 
+@pytest.fixture
+def hdf5_missing_value(tmp_path: Path) -> str:
+    """Create an empty HDF5 file."""
+    filepath = tmp_path / "compact_lowlevel.h5"
+    with h5py.File(filepath, "w") as f:
+        f.create_dataset("my_dataset", shape=(10,), dtype="int64")
+    return str(filepath)
+
+
 @pytest.fixture
 def hdf5_scalar(tmp_path: Path) -> str:
     """Create an HDF5 file with a scalar dataset."""

docs/releases.md

@@ -16,6 +16,10 @@
 
 ### Internal changes
 
+- Migrates bulk of `dmrpp parser` to `Pydap`.
+  ([#902](https://github.com/zarr-developers/VirtualiZarr/pull/902)).
+  By [Miguel Jimenez-Urias](https://github.com/Mikejmnez).
+
 ## v2.5.0 (23rd March 2026)
 
 Brings `region`-writing support in `.to_icechunk()`, a `ZarrParser` with orders of magnitude better performance, more FAQ docs, and various bugfixes.
@@ -69,6 +73,7 @@ Brings `region`-writing support in `.to_icechunk()`, a `ZarrParser` with orders
 - Fix `ZarrParser` to use public attribute instead of private one ([#916](https://github.com/zarr-developers/VirtualiZarr/pull/916)).
   By [Max Jones](https://github.com/maxrjones).
 
+
 ### Documentation
 
 - Added FAQ answer comparing the Kerchunk and Icechunk serialization formats. ([#818](https://github.com/zarr-developers/VirtualiZarr/pull/818)).

pyproject.toml

@@ -41,6 +41,7 @@ remote = [
     "requests",
     "aiohttp",
     "s3fs",
+    "pydap>=3.5.9",
 ]
 
 # non-kerchunk-based parsers
@@ -110,6 +111,7 @@ upstream = [
     'kerchunk @ git+https://github.com/fsspec/kerchunk',
     'icechunk @ git+https://github.com/earth-mover/icechunk#subdirectory=icechunk-python',
     'virtual_tiff @ git+https://github.com/virtual-zarr/virtual-tiff',
+    'pydap @ git+https://github.com/pydap/pydap.git',
 ]
 docs = [
     "mkdocs-material[imaging]>=9.6.14",
@@ -262,6 +264,7 @@ module = [
     "ujson",
     "zarr",
     "requests",
+    "pydap.*",
 ]
 ignore_missing_imports = true