OAS 3.1 tmulti-type cast/deserialize support

Author: p1c2u

openapi_core/casting/schemas/__init__.py

@@ -4,6 +4,7 @@
 from openapi_core.casting.schemas.casters import ArrayCaster
 from openapi_core.casting.schemas.casters import BooleanCaster
 from openapi_core.casting.schemas.casters import IntegerCaster
+from openapi_core.casting.schemas.casters import MultiTypeCaster
 from openapi_core.casting.schemas.casters import NumberCaster
 from openapi_core.casting.schemas.casters import ObjectCaster
 from openapi_core.casting.schemas.casters import PrimitiveCaster
@@ -46,10 +47,14 @@
     oas30_casters_dict,
     AnyCaster,
 )
+# OAS 3.1/3.2: ``type`` may be a list. ``multi=MultiTypeCaster`` enables the
+# real coercion path. ``multi`` is intentionally left ``None`` for OAS 3.0 so
+# any ``type: [..]`` in a 3.0 spec still raises
+# ``TypeError("caster does not accept multiple types")`` at dispatch time.
 oas31_types_caster = TypesCaster(
     oas31_casters_dict,
     AnyCaster,
-    multi=PrimitiveCaster,
+    multi=MultiTypeCaster,
 )
 oas32_types_caster = oas31_types_caster
 

openapi_core/casting/schemas/casters.py

@@ -186,6 +186,35 @@ def _cast_proparties(
         return value
 
 
+class MultiTypeCaster(PrimitiveCaster):
+    """Cast a value against a multi-type schema (OAS 3.1/3.2 ``type: [..]``).
+
+    Tries each declared type in order and returns the first cast that
+    succeeds. ``"null"`` entries are skipped — null values are short-circuited
+    upstream by ``SchemaCaster.cast`` before this caster is dispatched.
+
+    Raises ``CastError`` with the full type list when no candidate succeeds,
+    so callers see one failure with the complete declared set rather than a
+    storm of per-candidate errors.
+    """
+
+    def cast(self, value: Any) -> Any:
+        schema_types = (self.schema / "type").read_str_or_list([])
+        if isinstance(schema_types, str):
+            schema_types = [schema_types]
+        for candidate in schema_types:
+            if candidate == "null":
+                continue
+            try:
+                candidate_caster = self.schema_caster.get_type_caster(
+                    candidate
+                )
+                return candidate_caster(value)
+            except (CastError, ValueError, TypeError):
+                continue
+        raise CastError(value, list(schema_types))
+
+
 class TypesCaster:
     casters: Mapping[str, Type[PrimitiveCaster]] = {}
     multi: Optional[Type[PrimitiveCaster]] = None

openapi_core/deserializing/media_types/deserializers.py

@@ -23,6 +23,7 @@
 from openapi_core.schema.protocols import SuportsGetAll
 from openapi_core.schema.protocols import SuportsGetList
 from openapi_core.schema.schemas import get_properties
+from openapi_core.schema.types import pick_style_type
 from openapi_core.validation.schemas.validators import SchemaValidator
 
 if TYPE_CHECKING:
@@ -240,7 +241,12 @@ def decode_property_content_type(
             prop_schema,
             mimetype=prop_content_type,
         )
-        prop_schema_type = (prop_schema / "type").read_str("")
+        # Use ``read_str_or_list`` so OAS 3.1/3.2 multi-type properties
+        # (e.g. ``type: ["array", "null"]``) still trigger the multipart
+        # fan-out branch.
+        prop_schema_type = pick_style_type(
+            (prop_schema / "type").read_str_or_list("")
+        )
         if (
             self.mimetype.startswith("multipart")
             and prop_schema_type == "array"

openapi_core/deserializing/styles/casters.py

@@ -6,6 +6,7 @@
 
 from openapi_core.schema.protocols import SuportsGetAll
 from openapi_core.schema.protocols import SuportsGetList
+from openapi_core.schema.types import pick_style_type
 
 
 def split(value: str, separator: str = ",", step: int = 1) -> List[str]:
@@ -31,7 +32,7 @@ def delimited_loads(
 ) -> Any:
     value = location[name]
 
-    explode_type = (explode, schema_type)
+    explode_type = (explode, pick_style_type(schema_type))
     if explode_type == (False, "array"):
         return split(value, separator=delimiter)
     if explode_type == (False, "object"):
@@ -51,25 +52,26 @@ def matrix_loads(
     schema_type: str | list[str],
     location: Mapping[str, Any],
 ) -> Any:
+    structural_type = pick_style_type(schema_type)
     if explode == False:
         m = re.match(rf"^;{name}=(.*)$", location[f";{name}"])
         if m is None:
             raise KeyError(name)
         value = m.group(1)
         # ;color=blue,black,brown
-        if schema_type == "array":
+        if structural_type == "array":
             return split(value)
         # ;color=R,100,G,200,B,150
-        if schema_type == "object":
+        if structural_type == "object":
             return dict(map(split, split(value, step=2)))
         # .;color=blue
         return value
     else:
         # ;color=blue;color=black;color=brown
-        if schema_type == "array":
+        if structural_type == "array":
             return re.findall(rf";{name}=([^;]*)", location[f";{name}*"])
         # ;R=100;G=200;B=150
-        if schema_type == "object":
+        if structural_type == "object":
             value = location[f";{name}*"]
             return dict(
                 map(
@@ -91,23 +93,24 @@ def label_loads(
     schema_type: str | list[str],
     location: Mapping[str, Any],
 ) -> Any:
+    structural_type = pick_style_type(schema_type)
     if explode == False:
         value = location[f".{name}"]
         # .blue,black,brown
-        if schema_type == "array":
+        if structural_type == "array":
             return split(value[1:])
         # .R,100,G,200,B,150
-        if schema_type == "object":
+        if structural_type == "object":
             return dict(map(split, split(value[1:], separator=",", step=2)))
         # .blue
         return value[1:]
     else:
         value = location[f".{name}*"]
         # .blue.black.brown
-        if schema_type == "array":
+        if structural_type == "array":
             return split(value[1:], separator=".")
         # .R=100.G=200.B=150
-        if schema_type == "object":
+        if structural_type == "object":
             return dict(
                 map(
                     partial(split, separator="="),
@@ -124,7 +127,7 @@ def form_loads(
     schema_type: str | list[str],
     location: Mapping[str, Any],
 ) -> Any:
-    explode_type = (explode, schema_type)
+    explode_type = (explode, pick_style_type(schema_type))
     # color=blue,black,brown
     if explode_type == (False, "array"):
         return split(location[name], separator=",")
@@ -159,12 +162,13 @@ def simple_loads(
     location: Mapping[str, Any],
 ) -> Any:
     value = location[name]
+    structural_type = pick_style_type(schema_type)
 
     # blue,black,brown
-    if schema_type == "array":
+    if structural_type == "array":
         return split(value, separator=",")
 
-    explode_type = (explode, schema_type)
+    explode_type = (explode, structural_type)
     # R,100,G,200,B,150
     if explode_type == (False, "object"):
         return dict(map(split, split(value, separator=",", step=2)))
@@ -204,7 +208,7 @@ def deep_object_loads(
     schema_type: str | list[str],
     location: Mapping[str, Any],
 ) -> Any:
-    explode_type = (explode, schema_type)
+    explode_type = (explode, pick_style_type(schema_type))
 
     if explode_type != (True, "object"):
         raise ValueError("not available")

openapi_core/deserializing/styles/util.py

@@ -0,0 +1,39 @@
+"""Helpers for OpenAPI ``type`` (string or list of strings).
+
+OAS 3.1 and 3.2 allow the ``type`` keyword to be either a single string or
+a list of strings (e.g. ``type: ["integer", "null"]``). OAS 3.0 only allows
+a single string. Several places in the code need to make decisions based on
+the *structural* type implied by the schema (array vs. object vs. primitive)
+without caring which side of the version split they are on; this module
+centralises that mapping.
+"""
+
+from typing import Iterable
+from typing import Optional
+from typing import Union
+
+
+def pick_style_type(
+    schema_type: Optional[Union[str, Iterable[str]]],
+) -> str:
+    """Pick the structural type used by style/multipart deserializers.
+
+    Style loaders need to know whether the wire form should be parsed as an
+    array, an object, or a single scalar. They do not need to know which
+    primitive type the leaf will eventually become — that is the schema
+    caster's job.
+
+    For multi-type schemas the priority is ``array`` > ``object`` >
+    primitive. Primitive (or unknown) is represented by an empty string to
+    match the historical default returned by ``read_str_or_list("")``.
+    """
+    if schema_type is None:
+        return ""
+    if isinstance(schema_type, str):
+        return schema_type
+    types = list(schema_type)
+    if "array" in types:
+        return "array"
+    if "object" in types:
+        return "object"
+    return ""