Added partial support for compact datasets.

pyfive/h5d.py

@@ -98,7 +98,7 @@ def __init__(self, dataobject, pseudo_chunking_size_MB=4):
         # throws a flake8 wobbly for Python<3.10; match is Py3.10+ syntax
         match self.layout_class:  # noqa
             case 0:  #compact storage
-                raise NotImplementedError("Compact Storage")
+                self._data = self._get_compact_data(dataobject)
             case 1:  # contiguous storage
                 self.data_offset, = struct.unpack_from('<Q', dataobject.msg_data, self.property_offset)
             case 2:  # chunked storage
@@ -154,14 +154,14 @@ def read_direct_chunk(self, chunk_position, **kwargs):
             raise OSError("Chunk coordinates must lie on chunk boundaries")
         storeinfo = self._index[chunk_position]
         return storeinfo.filter_mask, self._get_raw_chunk(storeinfo)
-    
+
     def get_data(self, args, fillvalue):
         """ Called by the dataset getitem method """
         dtype = self._dtype
         # throws a flake8 wobbly for Python<3.10; match is Py3.10+ syntax
         match self.layout_class:  # noqa
             case 0:  #compact storage
-                raise NotImplementedError("Compact Storage")
+                return self._read_compact_data(args, fillvalue)
             case 1:  # contiguous storage
                 if self.data_offset == UNDEFINED_ADDRESS:
                     # no storage is backing array, return an array of
@@ -373,6 +373,40 @@ def _get_contiguous_data(self, args):
             except UnsupportedOperation:
                 return self._get_direct_from_contiguous(args)
 
+    def _get_compact_data(self, dataobject):
+        data = None
+        layout = None
+        for msg in dataobject.msgs:
+            if msg["type"] == 8:
+                layout = msg
+                break
+        if layout is None:
+            raise ValueError("No layout message in compact dataset?")
+        byts = dataobject.msg_data[msg["offset_to_message"]:msg["offset_to_message"]+msg["size"]]
+        layout_version = byts[0]
+        if layout_version == 1 or layout_version == 2:
+            raise NotImplementedError("Compact layout v1 and v2.")
+        elif layout_version == 3 or layout_version == 4:
+            size = int.from_bytes(byts[2:4], "little")
+            data = byts[4:4+size]
+        else:
+            raise ValueError("Unknown layout version.")
+        return data
+
+    def _read_compact_data(self, args, fillvalue):
+        if self._data is None:
+            if isinstance(self._dtype, tuple):
+                dtype = np.array(fillvalue).dtype
+            return np.full(self.shape, fillvalue, dtype=dtype)[args]
+        else:
+            view = np.frombuffer(
+                self._data,
+                dtype=self._dtype,
+            ).reshape(self.shape)
+            # Create the sub-array
+            result = view[args]
+            return result
+
 
     def _get_direct_from_contiguous(self, args=None):
         """

tests/test_compact.py

@@ -0,0 +1,36 @@
+import os
+
+import numpy as np
+import pytest
+from numpy.testing import assert_array_equal
+
+import pyfive
+import h5py
+
+
+def test_compact_dataset_hdf5(name, data):
+    with pyfive.File(name) as hfile:
+        # check data
+        dset1 = hfile['compact']
+        assert_array_equal(dset1[...], data)
+
+
+@pytest.fixture(scope='module')
+def data():
+    return np.array([1, 2, 3, 4], dtype=np.int32)
+
+
+@pytest.fixture(scope='module')
+def name(data):
+    name = os.path.join(os.path.dirname(__file__), 'compact.hdf5')
+
+    f = h5py.File(name, 'w', libver='earliest')
+    dtype = h5py.h5t.NATIVE_INT32
+    space = h5py.h5s.create_simple(data.shape)
+    dcpl = h5py.h5p.create(h5py.h5p.DATASET_CREATE)
+    dcpl.set_layout(h5py.h5d.COMPACT)
+    dset_id = h5py.h5d.create(f.id, b"compact", dtype, space, dcpl=dcpl)
+    dset_id.write(h5py.h5s.ALL, h5py.h5s.ALL, data)
+    f.close()
+
+    return name