temp

apis/python/src/tiledbsc/annotation_dataframe.py

@@ -229,7 +229,7 @@ def from_dataframe(self, dataframe: pd.DataFrame, extent: int) -> None:
             mode=mode,
         )
 
-        self.set_soma_object_type_metadata()
+        self._set_soma_object_type_metadata()
 
         if self._verbose:
             print(util.format_elapsed(s, f"{self._indent}FINISH WRITING {self.uri}"))

apis/python/src/tiledbsc/annotation_matrix.py

@@ -105,7 +105,7 @@ def from_matrix_and_dim_values(self, matrix, dim_values):
         else:
             self._numpy_ndarray_or_scipy_sparse_csr_matrix(matrix, dim_values)
 
-        self.set_soma_object_type_metadata()
+        self._set_soma_object_type_metadata()
 
         if self._verbose:
             print(util.format_elapsed(s, f"{self._indent}FINISH WRITING {self.uri}"))

apis/python/src/tiledbsc/assay_matrix.py

@@ -125,7 +125,7 @@ def from_matrix_and_dim_values(self, matrix, row_names, col_names) -> None:
         else:
             self._create_empty_array(matrix_dtype=matrix.dtype)
 
-        self.set_soma_object_type_metadata()
+        self._set_soma_object_type_metadata()
 
         self._ingest_data(matrix, row_names, col_names)
         if self._verbose:

apis/python/src/tiledbsc/tiledb_array.py

@@ -17,7 +17,8 @@ def __init__(
         self,
         uri: str,
         name: str,
-        #### XXX parent: Optional[TileDBGroup] = None,
+        # It's a circular import if we say this, but this is really:
+        # parent: Optional[TileDBGroup] = None,
         parent=None,
     ):
         """
@@ -98,7 +99,7 @@ def has_attr_name(self, attr_name: str) -> bool:
         """
         return attr_name in self.attr_names()
 
-    def set_soma_object_type_metadata(self) -> None:
+    def _set_soma_object_type_metadata(self) -> None:
         """
         This helps nested-structured traversals (especially those that start at the SOMACollection
         level) confidently navigate with a minimum of introspection on group contents.

apis/python/src/tiledbsc/tiledb_group.py

@@ -62,11 +62,45 @@ def _create(self):
         if self._verbose:
             print(f"{self._indent}Creating TileDB group {self.uri}")
         tiledb.group_create(uri=self.uri, ctx=self._ctx)
+
+        self._set_soma_object_type_metadata()
+
+    def _set_soma_object_type_metadata(self):
+        """
+        This helps nested-structured traversals (especially those that start at the SOMACollection
+        level) confidently navigate with a minimum of introspection on group contents.
+        """
         with self._open("w") as G:
             G.meta[
                 tiledbsc.util_tiledb.SOMA_OBJECT_TYPE_METADATA_KEY
             ] = self.__class__.__name__
 
+    def _set_soma_object_type_metadata_recursively(self):
+        """
+        SOMAs/SOCOs written very early on in the development of this project may not have these set.
+        Using this method we can after-populate these, without needig to re-ingest entire datasets.
+        Any SOMAs/SOCOs ingested from June 2022 onward won't need this -- this metadata will be
+        written at ingestion time.
+        """
+        self._set_soma_object_type_metadata()
+        with self._open() as G:
+            for O in G:  # This returns a tiledb.object.Object
+                # It might appear simpler to have all this code within TileDBObject class,
+                # rather than (with a little duplication) in TileDBGroup and TileDBArray.
+                # However, getting it to work with a recursive data structure and finding the
+                # required methods, it was simpler to split the logic this way.
+                object_type = tiledb.object_type(O.uri)
+                if object_type == "group":
+                    group = TileDBGroup(uri=O.uri, name=O.name, parent=self)
+                    group._set_soma_object_type_metadata_recursively()
+                elif object_type == "array":
+                    array = TileDBArray(uri=O.uri, name=O.name, parent=self)
+                    array._set_soma_object_type_metadata()
+                else:
+                    raise Exception(
+                        f"Unexpected object_type found: {object_type} at {O.uri}"
+                    )
+
     def _open(self, mode="r"):
         """
         This is just a convenience wrapper around tiledb group-open.
@@ -157,4 +191,6 @@ def show_metadata(self, recursively=True, indent=""):
                         array = TileDBArray(uri=O.uri, name=O.name, parent=self)
                         array.show_metadata(recursively, indent=child_indent)
                     else:
-                        raise Exception(f"Unexpected object_type found: {object_type}")
+                        raise Exception(
+                            f"Unexpected object_type found: {object_type} at {O.uri}"
+                        )

apis/python/src/tiledbsc/uns_group.py

@@ -114,12 +114,7 @@ def to_dict_of_matrices(self) -> Dict:
         """
         Reads the recursive group/array uns data from TileDB storage and returns them as a recursive dict of matrices.
         """
-        grp = None
-        try:  # Not all groups have uns
-            grp = tiledb.Group(self.uri, mode="r")
-        except:
-            pass
-        if grp == None:
+        if not self.exists():
             if self._verbose:
                 print(f"{self._indent}{self.uri} not found")
             return {}
@@ -128,24 +123,23 @@ def to_dict_of_matrices(self) -> Dict:
             s = util.get_start_stamp()
             print(f"{self._indent}START  read {self.uri}")
 
-        retval = {}
-        for element in grp:
-            name = os.path.basename(element.uri)  # TODO: update for tiledb cloud
+        with self._open() as G:
+            retval = {}
+            for element in G:
+                name = os.path.basename(element.uri)  # TODO: update for tiledb cloud
 
-            if element.type == tiledb.tiledb.Group:
-                child_group = UnsGroup(uri=element.uri, name=name, parent=self)
-                retval[name] = child_group.to_dict_of_matrices()
+                if element.type == tiledb.tiledb.Group:
+                    child_group = UnsGroup(uri=element.uri, name=name, parent=self)
+                    retval[name] = child_group.to_dict_of_matrices()
 
-            elif element.type == tiledb.libtiledb.Array:
-                child_array = UnsArray(uri=element.uri, name=name, parent=self)
-                retval[name] = child_array.to_matrix()
+                elif element.type == tiledb.libtiledb.Array:
+                    child_array = UnsArray(uri=element.uri, name=name, parent=self)
+                    retval[name] = child_array.to_matrix()
 
-            else:
-                raise Exception(
-                    f"Internal error: found uns group element neither group nor array: type is {str(element.type)}"
-                )
-
-        grp.close()
+                else:
+                    raise Exception(
+                        f"Internal error: found uns group element neither group nor array: type is {str(element.type)}"
+                    )
 
         if self._verbose:
             print(util.format_elapsed(s, f"{self._indent}FINISH read {self.uri}"))