copy: add skip option

GooseHDF5/__init__.py

@@ -564,6 +564,7 @@ def copy(
     dest_datasets: list[str] = None,
     root: str = None,
     recursive: bool = True,
+    skip: bool = False,
 ):
     """
     Copy groups/datasets from one HDF5-archive ``source`` to another HDF5-archive ``dest``.
@@ -577,6 +578,7 @@ def copy(
     :param dest_datasets: List of dataset-paths in ``dest``, defaults to ``source_datasets``.
     :param root: Path prefix for all ``dest_datasets``.
     :param recursive: If the source is a group, copy all objects within that group recursively.
+    :param skip: Skip datasets that are not present in source.
     """
 
     if len(source_datasets) == 0:
@@ -589,6 +591,11 @@ def copy(
 
     dest_datasets = np.array(dest_datasets)
 
+    if skip:
+        keep = [i in source for i in source_datasets]
+        source_datasets = source_datasets[keep]
+        dest_datasets = dest_datasets[keep]
+
     if root:
         dest_datasets = np.array([join(root, path, root=True) for path in dest_datasets])
 
@@ -838,7 +845,6 @@ def compare(
     b: Union[str, h5py.File],
     paths_a: list[str] = None,
     paths_b: list[str] = None,
-    root: str = None,
     attrs: bool = True,
     matching_dtype: bool = False,
 ):

tests/copying.py

@@ -8,20 +8,30 @@
 import GooseHDF5 as g5
 
 
+dirname = os.path.join(os.path.abspath(os.path.dirname(__file__)), "mytest")
+
+
 class Test_itereator(unittest.TestCase):
-    def test_copy_plain(self):
+    """ """
 
-        dirname = "mytest"
-        sourcepath = os.path.join(dirname, "foo_1.h5")
-        destpath = os.path.join(dirname, "bar_1.h5")
+    @classmethod
+    def setUpClass(self):
 
         if not os.path.isdir(dirname):
             os.makedirs(dirname)
 
+    @classmethod
+    def tearDownClass(self):
+
+        shutil.rmtree(dirname)
+
+    def test_copy_plain(self):
+
+        sourcepath = os.path.join(dirname, "foo_1.h5")
+        destpath = os.path.join(dirname, "bar_1.h5")
         datasets = ["/a", "/b/foo", "/c/d/foo"]
 
         with h5py.File(sourcepath, "w") as source:
-
             with h5py.File(destpath, "w") as dest:
 
                 for d in datasets:
@@ -32,21 +42,30 @@ def test_copy_plain(self):
                 for path in datasets:
                     self.assertTrue(g5.equal(source, dest, path))
 
-        shutil.rmtree(dirname)
-
-    def test_copy_nonrecursive(self):
+    def test_copy_skip(self):
 
-        dirname = "mytest"
         sourcepath = os.path.join(dirname, "foo_1.h5")
         destpath = os.path.join(dirname, "bar_1.h5")
+        datasets = ["/a", "/b/foo", "/c/d/foo"]
 
-        if not os.path.isdir(dirname):
-            os.makedirs(dirname)
+        with h5py.File(sourcepath, "w") as source:
+            with h5py.File(destpath, "w") as dest:
+
+                for d in datasets:
+                    source[d] = np.random.rand(10)
+
+                g5.copy(source, dest, datasets + ["/nonexisting"], skip=True)
+
+                for path in datasets:
+                    self.assertTrue(g5.equal(source, dest, path))
 
+    def test_copy_nonrecursive(self):
+
+        sourcepath = os.path.join(dirname, "foo_1.h5")
+        destpath = os.path.join(dirname, "bar_1.h5")
         datasets = ["/a", "/b/foo", "/b/bar", "/c/d/foo"]
 
         with h5py.File(sourcepath, "w") as source:
-
             with h5py.File(destpath, "w") as dest:
 
                 for d in datasets:
@@ -61,21 +80,13 @@ def test_copy_nonrecursive(self):
                 self.assertFalse(g5.exists(dest, "/b/foo"))
                 self.assertFalse(g5.exists(dest, "/b/bar"))
 
-        shutil.rmtree(dirname)
-
     def test_copy_recursive(self):
 
-        dirname = "mytest"
         sourcepath = os.path.join(dirname, "foo_1.h5")
         destpath = os.path.join(dirname, "bar_1.h5")
-
-        if not os.path.isdir(dirname):
-            os.makedirs(dirname)
-
         datasets = ["/a", "/b/foo", "/b/bar", "/c/d/foo"]
 
         with h5py.File(sourcepath, "w") as source:
-
             with h5py.File(destpath, "w") as dest:
 
                 for d in datasets:
@@ -86,21 +97,13 @@ def test_copy_recursive(self):
                 for path in datasets:
                     self.assertTrue(g5.equal(source, dest, path))
 
-        shutil.rmtree(dirname)
-
     def test_copy_attrs(self):
 
-        dirname = "mytest"
         sourcepath = os.path.join(dirname, "foo_2.h5")
         destpath = os.path.join(dirname, "bar_2.h5")
-
-        if not os.path.isdir(dirname):
-            os.makedirs(dirname)
-
         datasets = ["/a", "/b/foo", "/c/d/foo"]
 
         with h5py.File(sourcepath, "w") as source:
-
             with h5py.File(destpath, "w") as dest:
 
                 for d in datasets:
@@ -115,21 +118,13 @@ def test_copy_attrs(self):
                 for path in datasets:
                     self.assertTrue(g5.equal(source, dest, path))
 
-        shutil.rmtree(dirname)
-
     def test_copy_groupattrs(self):
 
-        dirname = "mytest"
         sourcepath = os.path.join(dirname, "foo_3.h5")
         destpath = os.path.join(dirname, "bar_3.h5")
-
-        if not os.path.isdir(dirname):
-            os.makedirs(dirname)
-
         datasets = ["/a", "/b/foo", "/c/d/foo"]
 
         with h5py.File(sourcepath, "w") as source:
-
             with h5py.File(destpath, "w") as dest:
 
                 for d in datasets:
@@ -143,8 +138,6 @@ def test_copy_groupattrs(self):
                 for path in datasets:
                     self.assertTrue(g5.equal(source, dest, path))
 
-        shutil.rmtree(dirname)
-
 
 if __name__ == "__main__":