depr(schema): schedule Schema.delete() and Schema.append() for removal

Author: kszucs

ibis/backends/base/sql/ddl.py

@@ -241,13 +241,13 @@ def _pieces(self):
                 part_fields = {name: self.schema[name] for name in part_schema}
                 part_schema = sch.Schema(part_fields)
 
-            to_delete = []
-            for name in self.partition:
-                if name in self.schema:
-                    to_delete.append(name)
-
-            if len(to_delete):
-                main_schema = main_schema.delete(to_delete)
+            to_delete = {name for name in self.partition if name in self.schema}
+            fields = {
+                name: dtype
+                for name, dtype in main_schema.items()
+                if name not in to_delete
+            }
+            main_schema = sch.Schema(fields)
 
             yield format_schema(main_schema)
             yield f'PARTITIONED BY {format_schema(part_schema)}'

ibis/backends/impala/tests/conftest.py

@@ -497,8 +497,12 @@ def impala_create_test_database(con, env):
 
 
 PARQUET_SCHEMAS = {
-    'functional_alltypes': TEST_TABLES["functional_alltypes"].delete(
-        ["index", "Unnamed: 0"]
+    "functional_alltypes": ibis.schema(
+        {
+            name: dtype
+            for name, dtype in TEST_TABLES["functional_alltypes"].items()
+            if name not in {"index", "Unnamed: 0"}
+        }
     ),
     "tpch_region": ibis.schema(
         [

ibis/expr/datatypes/core.py

@@ -9,7 +9,7 @@
 from public import public
 
 import ibis.expr.types as ir
-from ibis.common.annotations import optional
+from ibis.common.annotations import attribute, optional
 from ibis.common.exceptions import IbisTypeError
 from ibis.common.grounds import Concrete, Singleton
 from ibis.common.validators import (
@@ -734,12 +734,12 @@ def from_dict(
     def pairs(self) -> Mapping[str, DataType]:
         return self.fields
 
-    @property
+    @attribute.default
     def names(self) -> tuple[str, ...]:
         """Return the names of the struct's fields."""
         return tuple(self.fields.keys())
 
-    @property
+    @attribute.default
     def types(self) -> tuple[DataType, ...]:
         """Return the types of the struct's fields."""
         return tuple(self.fields.values())

ibis/expr/operations/relations.py

@@ -181,7 +181,7 @@ def __init__(self, left, right, predicates, **kwargs):
     @property
     def schema(self):
         # For joins retaining both table schemas, merge them together here
-        return self.left.schema.append(self.right.schema)
+        return self.left.schema.merge(self.right.schema)
 
 
 @public

ibis/expr/schema.py

@@ -74,17 +74,6 @@ def copy(self, fields=None):
             fields = self.fields
         return type(self)(fields)
 
-    @attribute.default
-    def _name_locs(self) -> dict[str, int]:
-        # validate unique field names
-        name_locs = {v: i for i, v in enumerate(self.names)}
-        if len(name_locs) < len(self.names):
-            duplicate_names = list(self.names)
-            for v in name_locs:
-                duplicate_names.remove(v)
-            raise IntegrityError(f'Duplicate column name(s): {duplicate_names}')
-        return name_locs
-
     def __repr__(self) -> str:
         space = 2 + max(map(len, self.names), default=0)
         return "ibis.Schema {{{}\n}}".format(
@@ -109,14 +98,18 @@ def __contains__(self, name: str) -> bool:
     def __getitem__(self, name: str) -> dt.DataType:
         return self.fields[name]
 
-    @property
+    @attribute.default
     def names(self):
         return tuple(self.fields.keys())
 
-    @property
+    @attribute.default
     def types(self):
         return tuple(self.fields.values())
 
+    @attribute.default
+    def _name_locs(self) -> dict[str, int]:
+        return {v: i for i, v in enumerate(self.names)}
+
     def equals(self, other: Schema) -> bool:
         """Return whether `other` is equal to `self`.
 
@@ -140,6 +133,11 @@ def equals(self, other: Schema) -> bool:
             )
         return self.__cached_equals__(other)
 
+    @deprecated(
+        as_of="4.1",
+        removed_in="5.0",
+        instead="construct a new Schema without the undesired names instead",
+    )
     def delete(self, names_to_delete: Iterable[str]) -> Schema:
         """Remove `names_to_delete` names from `self`.
 
@@ -247,8 +245,14 @@ def __ge__(self, other: Schema) -> bool:
         """Return whether `self` is a superset of or equal to `other`."""
         return set(self.items()) >= set(other.items())
 
+    @deprecated(as_of="4.1", removed_in="5.0", instead="use Schema.merge() instead")
     def append(self, other: Schema) -> Schema:
-        """Append `schema` to `self`.
+        return self.merge(other)
+
+    def merge(self, other: Schema) -> Schema:
+        """Merge `other` to `self`.
+
+        Raise an `IntegrityError` if there are duplicate column names.
 
         Parameters
         ----------
@@ -265,7 +269,7 @@ def append(self, other: Schema) -> Schema:
         >>> import ibis
         >>> first = ibis.Schema.from_dict({"a": "int", "b": "string"})
         >>> second = ibis.Schema.from_dict({"c": "float", "d": "int16"})
-        >>> first.append(second)
+        >>> first.merge(second)
         ibis.Schema {
           a  int64
           b  string

ibis/tests/expr/test_schema.py

@@ -217,6 +217,7 @@ def test_names_types():
 
 def test_schema_delete():
     s1 = ibis.schema({"a": "int64", "b": "string", "c": "float64", "d": "int64"})
-    s2 = s1.delete(["b", "d"])
+    with pytest.warns(FutureWarning):
+        s2 = s1.delete(["b", "d"])
 
     assert s2 == ibis.schema({"a": "int64", "c": "float64"})

ibis/tests/expr/test_table.py

@@ -1012,7 +1012,7 @@ def test_cross_join(table):
 
     joined = table.cross_join(scalar_aggs)
     agg_schema = api.Schema({'sum_a': 'int64', 'mean_b': 'double'})
-    ex_schema = table.schema().append(agg_schema)
+    ex_schema = table.schema().merge(agg_schema)
     assert_equal(joined.schema(), ex_schema)