fix: add nullability check for NamedStruct in builder (#104)

ensures that the nullability of a NamedStruct is NULLABILITY_REQUIRED

---------

Signed-off-by: MBWhite <whitemat@uk.ibm.com>

Author: mbwhite

Makefile

@@ -14,5 +14,8 @@ codegen-extensions:
 lint:
 	uvx ruff@0.11.11 check
 
+lint_fix:
+	uvx ruff@0.11.11 check --fix
+
 format:
 	uvx ruff@0.11.11 format

src/substrait/builders/plan.py

@@ -34,6 +34,11 @@ def _merge_extensions(*objs):
 def read_named_table(
     names: Union[str, Iterable[str]], named_struct: stt.NamedStruct
 ) -> UnboundPlan:
+    if named_struct.struct.nullability is stt.Type.NULLABILITY_NULLABLE:
+        raise Exception("NamedStruct must not contain a nullable struct")
+    elif named_struct.struct.nullability is stt.Type.NULLABILITY_UNSPECIFIED:
+        named_struct.struct.nullability = stt.Type.NULLABILITY_REQUIRED
+
     def resolve(registry: ExtensionRegistry) -> stp.Plan:
         _names = [names] if isinstance(names, str) else names
 

src/substrait/builders/type.py

@@ -257,4 +257,9 @@ def map(key: stt.Type, value: stt.Type, nullable=True) -> stt.Type:
 
 
 def named_struct(names: Iterable[str], struct: stt.Type) -> stt.NamedStruct:
+    if struct.struct.nullability is stt.Type.NULLABILITY_NULLABLE:
+        raise Exception("NamedStruct must not contain a nullable struct")
+    elif struct.struct.nullability is stt.Type.NULLABILITY_UNSPECIFIED:
+        struct.struct.nullability = stt.Type.NULLABILITY_REQUIRED
+
     return stt.NamedStruct(names=names, struct=struct.struct)

tests/builders/extended_expression/test_aggregate_function.py

@@ -12,7 +12,8 @@
         stt.Type(i64=stt.Type.I64(nullability=stt.Type.NULLABILITY_REQUIRED)),
         stt.Type(string=stt.Type.String(nullability=stt.Type.NULLABILITY_NULLABLE)),
         stt.Type(fp32=stt.Type.FP32(nullability=stt.Type.NULLABILITY_NULLABLE)),
-    ]
+    ],
+    nullability=stt.Type.Nullability.NULLABILITY_REQUIRED,
 )
 
 named_struct = stt.NamedStruct(

tests/builders/extended_expression/test_scalar_function.py

@@ -68,7 +68,7 @@ def test_sclar_add():
         extensions=[
             ste.SimpleExtensionDeclaration(
                 extension_function=ste.SimpleExtensionDeclaration.ExtensionFunction(
-                    extension_uri_reference=1, function_anchor=1, name="test_func:i8"   
+                    extension_uri_reference=1, function_anchor=1, name="test_func:i8"
                 )
             )
         ],

tests/builders/plan/test_aggregate.py

@@ -28,7 +28,9 @@
 registry = ExtensionRegistry(load_default_extensions=False)
 registry.register_extension_dict(yaml.safe_load(content), uri="test_uri")
 
-struct = stt.Type.Struct(types=[i64(nullable=False), boolean()])
+struct = stt.Type.Struct(
+    types=[i64(nullable=False), boolean()], nullability=stt.Type.NULLABILITY_REQUIRED
+)
 
 named_struct = stt.NamedStruct(names=["id", "is_applicable"], struct=struct)
 

tests/builders/plan/test_cross.py

@@ -7,13 +7,17 @@
 
 registry = ExtensionRegistry(load_default_extensions=False)
 
-struct = stt.Type.Struct(types=[i64(nullable=False), boolean()])
+struct = stt.Type.Struct(
+    types=[i64(nullable=False), boolean()], nullability=stt.Type.NULLABILITY_REQUIRED
+)
 
 named_struct = stt.NamedStruct(names=["id", "is_applicable"], struct=struct)
 
 named_struct_2 = stt.NamedStruct(
     names=["fk_id", "name"],
-    struct=stt.Type.Struct(types=[i64(nullable=False), string()]),
+    struct=stt.Type.Struct(
+        types=[i64(nullable=False), string()], nullability=stt.Type.NULLABILITY_REQUIRED
+    ),
 )
 
 

tests/builders/plan/test_fetch.py

@@ -8,7 +8,9 @@
 
 registry = ExtensionRegistry(load_default_extensions=False)
 
-struct = stt.Type.Struct(types=[i64(nullable=False), boolean()])
+struct = stt.Type.Struct(
+    types=[i64(nullable=False), boolean()], nullability=stt.Type.NULLABILITY_REQUIRED
+)
 
 named_struct = stt.NamedStruct(names=["id", "is_applicable"], struct=struct)
 

tests/builders/plan/test_filter.py

@@ -8,7 +8,9 @@
 
 registry = ExtensionRegistry(load_default_extensions=False)
 
-struct = stt.Type.Struct(types=[i64(nullable=False), boolean()])
+struct = stt.Type.Struct(
+    types=[i64(nullable=False), boolean()], nullability=stt.Type.NULLABILITY_REQUIRED
+)
 
 named_struct = stt.NamedStruct(names=["id", "is_applicable"], struct=struct)
 

tests/builders/plan/test_join.py

@@ -8,13 +8,17 @@
 
 registry = ExtensionRegistry(load_default_extensions=False)
 
-struct = stt.Type.Struct(types=[i64(nullable=False), boolean()])
+struct = stt.Type.Struct(
+    types=[i64(nullable=False), boolean()], nullability=stt.Type.NULLABILITY_REQUIRED
+)
 
 named_struct = stt.NamedStruct(names=["id", "is_applicable"], struct=struct)
 
 named_struct_2 = stt.NamedStruct(
     names=["fk_id", "name"],
-    struct=stt.Type.Struct(types=[i64(nullable=False), string()]),
+    struct=stt.Type.Struct(
+        types=[i64(nullable=False), string()], nullability=stt.Type.NULLABILITY_REQUIRED
+    ),
 )