938 Improve create_pydantic_model for multidimensional arrays (#939)

* handle multidimensional arrays in pydantic

* make sure the inner type is also correct

for example, if it's `Array(Array(Varchar()))`, the inner `Varchar` should be `constr`.

* ignore type warning

piccolo/utils/pydantic.py

@@ -74,6 +74,43 @@ def validate_columns(
     )
 
 
+def get_array_value_type(
+    column: Array, inner: t.Optional[t.Type] = None
+) -> t.Type:
+    """
+    Gets the correct type for an ``Array`` column (which might be
+    multidimensional).
+    """
+    if isinstance(column.base_column, Array):
+        inner_type = get_array_value_type(column.base_column, inner=inner)
+    else:
+        inner_type = get_pydantic_value_type(column.base_column)
+
+    return t.List[inner_type]  # type: ignore
+
+
+def get_pydantic_value_type(column: Column) -> t.Type:
+    """
+    Map the Piccolo ``Column`` to a Pydantic type.
+    """
+    value_type: t.Type
+
+    if isinstance(column, (Decimal, Numeric)):
+        value_type = pydantic.condecimal(
+            max_digits=column.precision, decimal_places=column.scale
+        )
+    elif isinstance(column, Email):
+        value_type = pydantic.EmailStr  # type: ignore
+    elif isinstance(column, Varchar):
+        value_type = pydantic.constr(max_length=column.length)
+    elif isinstance(column, Array):
+        value_type = get_array_value_type(column=column)
+    else:
+        value_type = column.value_type
+
+    return value_type
+
+
 def create_pydantic_model(
     table: t.Type[Table],
     nested: t.Union[bool, t.Tuple[ForeignKey, ...]] = False,
@@ -211,17 +248,7 @@ def create_pydantic_model(
         #######################################################################
         # Work out the column type
 
-        if isinstance(column, (Decimal, Numeric)):
-            value_type: t.Type = pydantic.condecimal(
-                max_digits=column.precision, decimal_places=column.scale
-            )
-        elif isinstance(column, Email):
-            value_type = pydantic.EmailStr
-        elif isinstance(column, Varchar):
-            value_type = pydantic.constr(max_length=column.length)
-        elif isinstance(column, Array):
-            value_type = t.List[column.base_column.value_type]  # type: ignore
-        elif isinstance(column, (JSON, JSONB)):
+        if isinstance(column, (JSON, JSONB)):
             if deserialize_json:
                 value_type = pydantic.Json
             else:
@@ -235,7 +262,7 @@ def create_pydantic_model(
                     validator  # type: ignore
                 )
         else:
-            value_type = column.value_type
+            value_type = get_pydantic_value_type(column=column)
 
         _type = t.Optional[value_type] if is_optional else value_type
 

tests/utils/test_pydantic.py

@@ -126,6 +126,32 @@ class Band(Table):
             "string",
         )
 
+    def test_multidimensional_array(self):
+        """
+        Make sure that multidimensional arrays have the correct type.
+        """
+
+        class Band(Table):
+            members = Array(Array(Varchar(length=255)), required=True)
+
+        pydantic_model = create_pydantic_model(table=Band)
+
+        self.assertEqual(
+            pydantic_model.model_fields["members"].annotation,
+            t.List[t.List[pydantic.constr(max_length=255)]],
+        )
+
+        # Should not raise a validation error:
+        pydantic_model(
+            members=[
+                ["Alice", "Bob", "Francis"],
+                ["Alan", "Georgia", "Sue"],
+            ]
+        )
+
+        with self.assertRaises(ValueError):
+            pydantic_model(members=["Bob"])
+
 
 class TestForeignKeyColumn(TestCase):
     def test_target_column(self):