RCAL-861: by default compress all arrays (#440)

changes/440.feature.rst

@@ -0,0 +1 @@
+Change default compression to lz4. The previous default was no compression.

pyproject.toml

@@ -14,6 +14,7 @@ classifiers = [
 ]
 dependencies = [
     "asdf >=3.3.0",
+    "lz4 >= 4.3.0",
     "asdf-astropy >=0.5.0",
     "gwcs >=0.19.0",
     "numpy >=1.24",

src/roman_datamodels/datamodels/_core.py

@@ -230,10 +230,11 @@ def open_asdf(self, init=None, **kwargs):
             return asdf.AsdfFile(init, **kwargs)
 
     def to_asdf(self, init, *args, **kwargs):
+        all_array_compression = kwargs.pop("all_array_compression", "lz4")
         with validate.nuke_validation(), _temporary_update_filename(self, Path(init).name):
             asdf_file = self.open_asdf(**kwargs)
             asdf_file["roman"] = self._instance
-            asdf_file.write_to(init, *args, **kwargs)
+            asdf_file.write_to(init, *args, all_array_compression=all_array_compression, **kwargs)
 
     def get_primary_array_name(self):
         """

src/roman_datamodels/maker_utils/_base.py

@@ -29,6 +29,6 @@ def save_node(node, filepath=None):
 
         af = asdf.AsdfFile()
         af.tree = {"roman": node}
-        af.write_to(filepath)
+        af.write_to(filepath, all_array_compression="lz4")
 
     return node

tests/test_models.py

@@ -1109,3 +1109,28 @@ def test_model_assignment_access_types(model_class):
     assert type(model["meta"]["filename"]) == type(model2.meta["filename"])  # noqa: E721
     assert type(model["meta"]["filename"]) == type(model2.meta.filename)  # noqa: E721
     assert type(model.meta.filename) == type(model2.meta["filename"])  # noqa: E721
+
+
+def test_default_array_compression(tmp_path):
+    """
+    Test that saving a model results in compressed arrays
+    for default options.
+    """
+    fn = tmp_path / "foo.asdf"
+    model = utils.mk_datamodel(datamodels.ImageModel)
+    model.save(fn)
+    with asdf.open(fn) as af:
+        assert af.get_array_compression(af["roman"]["data"]) == "lz4"
+
+
+@pytest.mark.parametrize("compression", [None, "bzp2"])
+def test_array_compression_override(tmp_path, compression):
+    """
+    Test that providing a compression argument changes the
+    array compression.
+    """
+    fn = tmp_path / "foo.asdf"
+    model = utils.mk_datamodel(datamodels.ImageModel)
+    model.save(fn, all_array_compression=compression)
+    with asdf.open(fn) as af:
+        assert af.get_array_compression(af["roman"]["data"]) == compression