more updates to user None

Author: tranade

lib/sycamore/sycamore/data/document.py

@@ -224,15 +224,11 @@ def field_to_value(self, field: str) -> Any:
             Returns None if field does not exist in document.
         """
         fields = field.split(".")
-        if hasattr(self, fields[0]):
-            value = getattr(self, fields[0])
-        else:
-            return None
+        value = getattr(self, fields[0], None)
         if len(fields) > 1:
-            assert fields[0] == "properties"
             for f in fields[1:]:
-                if f in value:
-                    value = value[f]
+                if isinstance(value, dict):
+                    value = value.get(f, None)
                 else:
                     return None
         return value

lib/sycamore/sycamore/query/execution/operations.py

@@ -363,7 +363,7 @@ def make_filter_fn_join(field: str, join_set: set) -> Callable[[Document], bool]
     """
 
     def filter_fn_join(doc: Document) -> bool:
-        value = str(doc.field_to_value(field))
+        value = doc.field_to_value(field)
         return value in join_set
 
     return filter_fn_join
@@ -391,7 +391,7 @@ def join_operation(docset1: DocSet, docset2: DocSet, field1: str, field2: str) -
         doc = Document.from_row(row)
         if isinstance(doc, MetadataDocument):
             continue
-        value = str(doc.field_to_value(field1))
+        value = doc.field_to_value(field1)
         unique_vals.add(value)
 
     # filters docset2 based on matches of field2 with unique values
@@ -548,6 +548,8 @@ def ray_callable(input_dict: dict[str, Any]) -> dict[str, Any]:
 
         if unique_field is not None:
             val = str(doc.field_to_value(unique_field))
+            if val is None:
+                return {"doc": None, "key": None, "unique": None}
             # updates row to include new col
             new_doc["unique"] = val
 

lib/sycamore/sycamore/tests/unit/query/test_operations.py

@@ -348,31 +348,6 @@ def test_semantic_cluster(self, number_docset):
             elif doc.text_representation == "3" or doc.text_representation == "three":
                 assert doc.properties["_autogen_ClusterAssignment"] == "group3"
 
-    # Helpers
-    # def test_field_to_value(self):
-    #     doc = Document(
-    #         text_representation="hello",
-    #         doc_id=1,
-    #         properties={"letter": "A", "animal": "panda", "math": {"pi": 3.14, "e": 2.72, "tanx": "sinx/cosx"}},
-    #     )
-
-    #     assert field_to_value(doc, "text_representation") == "hello"
-    #     assert field_to_value(doc, "doc_id") == 1
-    #     assert field_to_value(doc, "properties.letter") == "A"
-    #     assert field_to_value(doc, "properties.animal") == "panda"
-    #     assert field_to_value(doc, "properties.math.pi") == 3.14
-    #     assert field_to_value(doc, "properties.math.e") == 2.72
-    #     assert field_to_value(doc, "properties.math.tanx") == "sinx/cosx"
-
-    #     with pytest.raises(KeyError):
-    #         field_to_value(doc, "properties.math.log")
-
-    #     with pytest.raises(Exception):
-    #         field_to_value(doc, "document_id")
-
-    #     with pytest.raises(AssertionError):
-    #         field_to_value(doc, "text_representation.text")
-
     def test_convert_string_to_date(self):
         date_string = "2024-07-21"
         expected_date = datetime(2024, 7, 21)