fix (issue-1079): allow update_column to set doc as ''

Author: Tiansu Yu

pyiceberg/table/__init__.py

@@ -2492,21 +2492,22 @@ def update_column(
                 except ResolveError as e:
                     raise ValidationError(f"Cannot change column type: {full_name}: {field.field_type} -> {field_type}") from e
 
+        # if other updates for the same field exist in one transaction:
         if updated := self._updates.get(field.field_id):
             self._updates[field.field_id] = NestedField(
                 field_id=updated.field_id,
                 name=updated.name,
                 field_type=field_type or updated.field_type,
-                doc=doc or updated.doc,
-                required=updated.required,
+                doc=doc if doc is not None else updated.doc,
+                required=required or updated.required,
             )
         else:
             self._updates[field.field_id] = NestedField(
                 field_id=field.field_id,
                 name=field.name,
                 field_type=field_type or field.field_type,
-                doc=doc or field.doc,
-                required=field.required,
+                doc=doc if doc is not None else field.doc,
+                required=required or field.required,
             )
 
         if required is not None:
@@ -2878,7 +2879,7 @@ def _update_column(self, field: NestedField, existing_field: NestedField) -> Non
         if field.field_type.is_primitive and field.field_type != existing_field.field_type:
             self.update_schema.update_column(full_name, field_type=field.field_type)
 
-        if field.doc is not None and not field.doc != existing_field.doc:
+        if field.doc is not None and field.doc != existing_field.doc:
             self.update_schema.update_column(full_name, doc=field.doc)
 
     def _find_field_type(self, field_id: int) -> IcebergType:

pyproject.toml

@@ -16,7 +16,7 @@
 # under the License.
 [tool.poetry]
 name = "pyiceberg"
-version = "0.7.1"
+version = "0.7.2"
 readme = "README.md"
 homepage = "https://py.iceberg.apache.org/"
 repository = "https://github.com/apache/iceberg-python"

tests/table/test_init.py

@@ -512,6 +512,29 @@ def test_add_column(table_v2: Table) -> None:
     assert apply_schema.highest_field_id == 4
 
 
+def test_update_column_doc(table_v1: Table, table_v2: Table) -> None:
+    COMMENT2 = "comment2"
+    for table in [table_v1, table_v2]:
+        original_schema = table.schema()
+        # update existing doc to a new doc
+        assert original_schema.find_field("y").doc == "comment"
+        new_schema = table.transaction().update_schema().update_column("y", doc=COMMENT2)._apply()
+        assert new_schema.find_field("y").doc == COMMENT2, "failed to update existing field doc"
+
+        # update existing doc to an emtpy string
+        assert new_schema.find_field("y").doc == COMMENT2
+        new_schema2 = table.transaction().update_schema().update_column("y", doc="")._apply()
+        assert new_schema2.find_field("y").doc == "", "failed to remove existing field doc"
+
+        # assert the above two updates also works with union_by_name
+        assert (
+            table.update_schema().union_by_name(new_schema)._apply() == new_schema
+        ), "failed to update existing field doc with union_by_name"
+        assert (
+            table.update_schema().union_by_name(new_schema2)._apply() == new_schema2
+        ), "failed to remove existing field doc with union_by_name"
+
+
 def test_add_primitive_type_column(table_v2: Table) -> None:
     primitive_type: Dict[str, PrimitiveType] = {
         "boolean": BooleanType(),