Add .gen_paths to relative path for auto generated relative paths

src/dataregistry/registrar/dataset.py

@@ -20,6 +20,7 @@
     _read_configuration_file,
     get_directory_info,
     _relpath_from_name,
+    get_first_directory,
 )
 from .dataset_util import set_dataset_status, get_dataset_status
 
@@ -83,6 +84,18 @@ def _validate_register_inputs(
                     "External datasets require either a url or contact_email"
                 )
 
+        # Make sure the user passed `relative_path` is legal
+        # Only needed for `register` function, `replace` has no `relative_path`
+        # argument as this cannot be changed from the original `register`
+        if "relative_path" in kwargs_dict.keys():
+            if kwargs_dict["relative_path"] is not None:
+                first_dir = get_first_directory(kwargs_dict["relative_path"])
+
+                if first_dir is not None:
+                    if first_dir == ".gen_paths":
+                        raise ValueError("Can't start relative path with .gen_paths, "
+                            "this is reserved for auto-generated `relative_paths")
+
         # Assign the `owner_type`
         if kwargs_dict["owner_type"] is None:
             if self._owner_type is not None:

src/dataregistry/registrar/registrar_util.py

@@ -332,6 +332,11 @@ def _relpath_from_name(name, version):
     Construct a relative path from the name and version of a dataset.
     We use this when the `relative_path` is not explicitly defined.
 
+    Every automatically generated `relative_path` is prefixed with
+    `.gen_paths/`, meaning that all automatically generated `relative_paths` go
+    into this top level folder. This is to prevent clashes with user specified
+    `relative_path`'s.
+
     Parameters
     ----------
     name : str
@@ -345,4 +350,31 @@ def _relpath_from_name(name, version):
         Automatically generated `relative_path`
     """
 
-    return f"{name}_{version}"
+    return os.path.join(".gen_paths", f"{name}_{version}")
+
+def get_first_directory(path):
+    """
+    Get back the root folder of a path
+
+    Parameters
+    ----------
+    path : str
+        Absolute or relative path string
+
+    Returns
+    -------
+    - : str
+        Root (or 1st) folder name
+        Returns None if path had no directory separators 
+    """
+
+    # Normalize the path and split it into components
+    parts = os.path.normpath(path).split(os.sep)
+
+    # Filter out any empty strings (in case of absolute paths starting with '/')
+    parts = [part for part in parts if part]
+
+    # Return the first part, which should be the first directory
+    if parts:
+        return parts[0]
+    return None

tests/end_to_end_tests/test_register_dataset_dummy.py

@@ -49,7 +49,7 @@ def test_register_dataset_defaults(dummy_file):
     assert results["owner"][0] == os.getenv("USER")
     assert results["owner_type"][0] == "user"
     assert results["description"][0] == None
-    assert results["relative_path"][0] == f"{_NAME}_0.0.1"
+    assert results["relative_path"][0] == f".gen_paths/{_NAME}_0.0.1"
     assert results["data_org"][0] == "dummy"
     assert results["execution_id"][0] >= 0
     assert results["dataset_id"][0] >= 0
@@ -192,7 +192,7 @@ def test_dataset_bumping(dummy_file, v_type, ans):
     # Check the result
     assert results["dataset.name"][0] == _NAME
     assert results["dataset.version_string"][0] == ans
-    assert results["dataset.relative_path"][0] == f"{_NAME}_{ans}"
+    assert results["dataset.relative_path"][0] == f".gen_paths/{_NAME}_{ans}"
 
 
 @pytest.mark.parametrize("owner_type", ["user", "group", "project"])

tests/end_to_end_tests/test_register_dataset_real_data.py

@@ -146,6 +146,28 @@ def test_registering_bad_relative_path(dummy_file, link):
             relative_path=f"test/register/bad/relpath/{link}",
         )
 
+@pytest.mark.parametrize("link", ["file1.txt", "directory1"])
+def test_registering_bad_relative_path_2(dummy_file, link):
+    """
+    Make sure we cannot register a datataset to a relative path that is using
+    the auto generated .gen_paths directry.
+    """
+
+    # Establish connection to database
+    tmp_src_dir, tmp_root_dir = dummy_file
+    datareg = DataRegistry(root_dir=str(tmp_root_dir), schema=DEFAULT_SCHEMA_WORKING)
+
+    data_path = str(tmp_src_dir / link)
+
+    with pytest.raises(ValueError, match="Can't start relative path with .gen_paths"):
+        d_id = _insert_dataset_entry(
+            datareg,
+            f"DESC:datasets:test_registering_bad_relative_path_3_{link}",
+            "0.0.1",
+            old_location=data_path,
+            location_type="dataregistry",
+            relative_path=f".gen_paths/test/register/bad/relpath/{link}",
+        )
 
 @pytest.mark.parametrize("link", ["file1.txt", "directory1"])
 def test_registering_deleted_relative_path(dummy_file, link):

tests/unit_tests/test_registrar_util.py

@@ -8,6 +8,7 @@
     _read_configuration_file,
     get_directory_info,
     _relpath_from_name,
+    get_first_directory,
 )
 
 
@@ -159,7 +160,7 @@ def test_read_file(tmpdir, nchars, max_config_length, ans):
 @pytest.mark.parametrize(
     "name,version_string,ans",
     [
-        ("mydataset", "1.1.1", "mydataset_1.1.1"),
+        ("mydataset", "1.1.1", ".gen_paths/mydataset_1.1.1"),
     ],
 )
 def test_relpath_from_name(name, version_string, ans):
@@ -185,3 +186,20 @@ def test_name_from_relpath(rel_path,ans):
     """Make sure names are extracted from paths correctly"""
 
     assert _name_from_relpath(rel_path) == ans
+
+@pytest.mark.parametrize(
+    "path,ans",
+    [
+        ("/testing/test", "testing"),
+        ("./testing/test", "testing"),
+        ("/testing/test/", "testing"),
+        ("test", "test"),
+        (".gen_paths/test", ".gen_paths"),
+        ("test/testing/tested", "test"),
+        ("/test/testing/tested", "test"),
+    ],
+)
+def test_get_first_directory(path,ans):
+    """Make sure the first directory is pulled out correct"""
+
+    assert get_first_directory(path) == ans