Merge pull request #130 from legend-exp/chunks

Allow passing lists to set HDF5 chunking

Author: gipert

src/lgdo/lh5/_serializers/write/array.py

@@ -64,6 +64,11 @@ def _h5_write_array(
         if "hdf5_settings" in obj.attrs:
             h5py_kwargs |= obj.attrs["hdf5_settings"]
 
+        # HACK: a tuple is strictly requested for the "chunks" setting, but
+        # we'd like to pass a list too in some situations
+        if "chunks" in h5py_kwargs and isinstance(h5py_kwargs["chunks"], list):
+            h5py_kwargs["chunks"] = tuple(h5py_kwargs["chunks"])
+
         # create HDF5 dataset
         ds = group.create_dataset(name, data=nda, **h5py_kwargs)
 

tests/lh5/test_lh5_write.py

@@ -44,6 +44,19 @@ def test_write_with_hdf5_compression(lgnd_file, tmptestdir):
         assert h5f["/geds/raw/waveform/values"].compression is None
         assert h5f["/geds/raw/waveform/values"].shuffle is False
 
+    store.write(
+        wft.values,
+        "/geds/raw/waveform/values",
+        f"{tmptestdir}/tmp-pygama-hdf5-compressed-wfs.lh5",
+        wo_mode="overwrite_file",
+        chunks=[1, 10],
+        compression=None,
+        shuffle=False,
+    )
+    with h5py.File(f"{tmptestdir}/tmp-pygama-hdf5-compressed-wfs.lh5") as h5f:
+        assert h5f["/geds/raw/waveform/values"].compression is None
+        assert h5f["/geds/raw/waveform/values"].shuffle is False
+
 
 def test_write_empty_vov(tmptestdir):
     vov = types.VectorOfVectors(flattened_data=[], cumulative_length=[])