debug [skip-ci]

apis/python/src/tiledbsoma/_arrow_types.py

@@ -263,17 +263,63 @@ def df_to_arrow(df: pd.DataFrame) -> pa.Table:
     #   will later be appending, or updating, with null data.
     # * Note that Arrow has a per-field nullable flag in its schema metadata
     #   -- and so do TileDB array schemas.
+    print()
+    print()
+    print("DF")
+    print(df.dtypes)
+    print()
+    print()
+
+    print()
+    print()
+    print("NF1")
+    print(nullable_fields)
+    print()
+    print()
+
     for key in df:
-        if df[key].dtype.name == "object":
+        print(">>>>", key, df[key].dtype.name)
+        if df[key].dtype.name in ["object", "string"]:
+            print("]]]]", key, "YES")
             nullable_fields.add(key)
+        else:
+            print("]]]]", key, "NO")
+
+    print()
+    print()
+    print("DF")
+    print(df)
+    print()
+    print()
+
+    print()
+    print()
+    print("NF2")
+    print(nullable_fields)
+    print()
+    print()
 
     arrow_table = pa.Table.from_pandas(df)
 
+    print()
+    print()
+    print("AT1")
+    print(arrow_table.schema)
+    print()
+    print()
+
     if nullable_fields:
         md = arrow_table.schema.metadata
         md.update(dict.fromkeys(nullable_fields, "nullable"))
         arrow_table = arrow_table.replace_schema_metadata(md)
 
+    print()
+    print()
+    print("AT2")
+    print(arrow_table.schema)
+    print()
+    print()
+
     # For tiledbsoma.io (for which this method exists) _any_ dataset can be appended to
     # later on. This means that on fresh ingest we must use a larger bit-width than
     # the bare minimum necessary.
@@ -297,6 +343,13 @@ def df_to_arrow(df: pd.DataFrame) -> pa.Table:
 
     arrow_table = pa.Table.from_pandas(df, schema=new_schema)
 
+    print()
+    print()
+    print("AT3")
+    print(arrow_table.schema)
+    print()
+    print()
+
     return arrow_table
 
 

apis/python/src/tiledbsoma/_dataframe.py

@@ -445,6 +445,13 @@ def write(
 
         clib_dataframe = self._handle._handle
 
+        print()
+        print()
+        print("VSDF SELF.SCHEMA")
+        print(self.schema)
+        print()
+        print()
+
         values = _util.cast_values_to_target_schema(clib_dataframe, values, self.schema)
 
         for batch in values.to_batches():

apis/python/src/tiledbsoma/_util.py

@@ -378,8 +378,30 @@ def cast_values_to_target_schema(
             target_schema.append(target_field.with_type(pa.uint8()))
         else:
             target_schema.append(target_field)
+
+    print()
+    print()
+    print("CCCC1")
+    print(target_schema)
+    print()
+    print()
+
     new_schema = pa.schema(target_schema, values.schema.metadata)
 
+    print()
+    print()
+    print("CCCC2a")
+    print(values)
+    print()
+    print()
+
+    print()
+    print()
+    print("CCCC2b")
+    print(new_schema)
+    print()
+    print()
+
     return values.cast(new_schema)
 
 

apis/python/tests/test_update_dataframes.py

@@ -252,3 +252,37 @@ def test_change_counts(
         verify_updates(experiment_path, new_obs2, new_var2, exc=ValueError)
         verify_obs_and_var_eq(old_anndata, new_anndata)
         verify_schemas(experiment_path, obs_schema, var_schema)
+
+
+# @pytest.mark.parametrize("multiple_fixtures_with_readback", [False, True], indirect=True)
+@pytest.mark.parametrize("multiple_fixtures_with_readback", [False], indirect=True)
+# @pytest.mark.parametrize("multiple_fixtures_with_readback", [True], indirect=True)
+def test_update_non_null_to_null(exp_path, new_obs, new_var, o1, v1):
+    print()
+    print()
+    print("NEW_OBS1")
+    print(new_obs)
+    print()
+    print()
+
+    new_obs["batch_id"] = "testing"
+
+    print()
+    print()
+    print("NEW_OBS2")
+    print(new_obs)
+    print()
+    print()
+
+    verify_updates(exp_path, new_obs, new_var)
+
+    new_obs["batch_id"] = pd.NA
+
+    print()
+    print()
+    print("NEW_OBS3")
+    print(new_obs)
+    print()
+    print()
+
+    verify_updates(exp_path, new_obs, new_var)