GEOPY-2756: H5 flags to optimize performances

geoh5py/workspace/workspace.py

@@ -245,6 +245,9 @@ def close(self):
 
             self.repack = False
 
+        if isinstance(self._h5file, BytesIO):
+            self._h5file.seek(0)
+
     @property
     def contributors(self) -> np.ndarray:
         """
@@ -1296,17 +1299,13 @@ def _create_h5(self) -> h5py.File:
         """
         Generate a new geoh5 file with core structure.
         """
-        if isinstance(self.h5file, BytesIO):
-            self._geoh5 = h5py.File(self.h5file, "a")
-
-        elif isinstance(self.h5file, Path):
-            self._geoh5 = h5py.File(
-                self.h5file,
-                "x",
-                fs_strategy="page",
-                page_buf_size=DEFAULT_PAGE_BUF_SIZE,
-            )
-
+        self._geoh5 = h5py.File(
+            self.h5file,
+            "x",
+            fs_strategy="page",
+            page_buf_size=DEFAULT_PAGE_BUF_SIZE,
+            libver=("v110", "v114"),
+        )
         H5Writer.init_geoh5(self._geoh5, self)
 
         return self._geoh5

tests/add_filename_data_test.py

@@ -33,8 +33,8 @@
 
 
 def test_add_file(tmp_path: Path):
-    workspace = Workspace(tmp_path / "test.geoh5")
-    workspace_copy = Workspace()
+    workspace = Workspace.create(tmp_path / f"{__name__}.geoh5")
+    workspace_copy = Workspace.create(tmp_path / f"{__name__}_copy.geoh5")
     with pytest.warns(UserWarning, match="No 'vertices' provided."):
         curve = Curve.create(workspace)
     group = ContainerGroup.create(workspace)

tests/point_data_test.py

@@ -35,10 +35,11 @@ def test_create_point_data(tmp_path):
     # Generate a random cloud of points
     values = np.random.randn(12)
     h5file_path = tmp_path / r"testPoints.geoh5"
-    workspace = Workspace.create(h5file_path)
+    workspace = Workspace()
     points = Points.create(workspace, vertices=np.random.randn(12, 3), allow_move=False)
     data = points.add_data({"DataValues": {"association": "VERTEX", "values": values}})
 
+    workspace.save_as(h5file_path)
     with pytest.raises(ValueError, match="Association flag should be one of"):
         points.add_data({"test": {"association": "ABC", "values": values}})
 

tests/workspace_test.py

@@ -61,15 +61,15 @@ def test_workspace_from_kwargs(tmp_path: Path):
                 f"Error changing value for attribute {key}."
             )
 
-    assert workspace.geoh5.libver == ("earliest", "v114")
+    assert workspace.geoh5.libver == ("v110", "v114")
     workspace.close()
 
 
 def test_empty_workspace(tmp_path):
-    Workspace.create(
+    with Workspace.create(
         tmp_path / r"test.geoh5",
-    ).close()
-
+    ) as workspace:
+        assert workspace.geoh5.libver == ("v110", "v114")
     with File(tmp_path / r"test.geoh5", "r+") as file:
         del file["GEOSCIENCE"]["Groups"]
         del file["GEOSCIENCE"]["Data"]