Add multiple --input-model support with inheritance preservation

Author: koxudaxi

src/datamodel_code_generator/__main__.py

@@ -385,6 +385,16 @@ def validate_all_exports_collision_strategy(self: Self) -> Self:  # pyright: ign
                 raise Error(self.__validate_all_exports_collision_strategy_err)
             return self
 
+        from pydantic import field_validator as _field_validator  # noqa: PLC0415
+
+        @_field_validator("input_model", mode="before")
+        @classmethod
+        def coerce_input_model_to_list(cls, v: str | list[str] | None) -> list[str] | None:  # pyright: ignore[reportRedeclaration]
+            """Convert string input_model to list for backwards compatibility."""
+            if isinstance(v, str):
+                return [v]
+            return v
+
     else:
 
         @model_validator()  # pyright: ignore[reportArgumentType]
@@ -443,8 +453,16 @@ def validate_all_exports_collision_strategy(cls, values: dict[str, Any]) -> dict
                 raise Error(cls.__validate_all_exports_collision_strategy_err)
             return values
 
+        @field_validator("input_model", mode="before")
+        @classmethod
+        def coerce_input_model_to_list(cls, v: str | list[str] | None) -> list[str] | None:
+            """Convert string input_model to list for backwards compatibility."""
+            if isinstance(v, str):
+                return [v]
+            return v
+
     input: Optional[Union[Path, str]] = None  # noqa: UP007, UP045
-    input_model: Optional[str] = None  # noqa: UP045
+    input_model: Optional[list[str]] = None  # noqa: UP045
     input_model_ref_strategy: Optional[InputModelRefStrategy] = None  # noqa: UP045
     input_file_type: InputFileType = InputFileType.Auto
     output_model_type: DataModelType = DataModelType.PydanticBaseModel
@@ -1172,6 +1190,231 @@ def _try_rebuild_model(obj: type) -> None:
         obj.model_rebuild()
 
 
+def _get_base_model_parents(model_class: type) -> list[type]:
+    """Get parent classes that are BaseModel subclasses (excluding BaseModel itself)."""
+    return [
+        p
+        for p in model_class.__bases__
+        if isinstance(p, type) and issubclass(p, BaseModel) and p is not BaseModel
+    ]
+
+
+def _transform_single_model_to_inheritance(  # noqa: PLR0912
+    schema: dict[str, object],
+    model_class: type,
+    schema_generator: type,
+    processed_parents: dict[str, dict[str, object]] | None = None,
+) -> dict[str, object]:
+    """Transform a single model's schema to use allOf inheritance structure.
+
+    Args:
+        schema: The JSON schema generated by Pydantic
+        model_class: The Pydantic model class
+        schema_generator: The schema generator class
+        processed_parents: Cache of already processed parent schemas
+
+    Returns:
+        Transformed schema with allOf structure for inheritance
+    """
+    if processed_parents is None:
+        processed_parents = {}
+
+    direct_parents = _get_base_model_parents(model_class)
+
+    if not direct_parents:
+        return schema
+
+    parent = direct_parents[0]
+    parent_name = parent.__name__
+    parent_fields = set(parent.model_fields.keys())
+
+    defs = dict(cast("dict[str, object]", schema.get("$defs", {})))
+
+    if parent_name in processed_parents:
+        parent_schema = processed_parents[parent_name]
+    else:
+        if hasattr(parent, "model_rebuild"):
+            _try_rebuild_model(parent)
+        parent_schema = parent.model_json_schema(schema_generator=schema_generator)
+        parent_schema = _add_python_type_for_unserializable(parent_schema, parent)
+        parent_schema = _add_python_type_info(parent_schema, parent)
+        parent_schema = _transform_single_model_to_inheritance(
+            parent_schema, parent, schema_generator, processed_parents
+        )
+        processed_parents[parent_name] = parent_schema
+
+    if "$defs" in parent_schema:
+        parent_defs = cast("dict[str, object]", parent_schema["$defs"])
+        for k, v in parent_defs.items():
+            if k not in defs:
+                defs[k] = v
+
+    parent_def = {k: v for k, v in parent_schema.items() if k != "$defs"}
+    defs[parent_name] = parent_def
+
+    original_props = cast("dict[str, object]", schema.get("properties", {}))
+    child_props = {k: v for k, v in original_props.items() if k not in parent_fields}
+
+    new_schema: dict[str, object] = {}
+    if defs:
+        new_schema["$defs"] = defs
+    new_schema["allOf"] = [{"$ref": f"#/$defs/{parent_name}"}]
+    if child_props:
+        new_schema["properties"] = child_props
+    original_required = cast("list[str]", schema.get("required", []))
+    child_required = [r for r in original_required if r not in parent_fields]
+    if child_required:
+        new_schema["required"] = child_required
+    new_schema["title"] = schema.get("title")
+    new_schema["type"] = "object"
+
+    for key in schema:
+        if key not in {"$defs", "properties", "required", "title", "type", "allOf"}:
+            new_schema[key] = schema[key]
+
+    return new_schema
+
+
+def _load_multiple_model_schemas(  # noqa: PLR0912, PLR0914, PLR0915
+    input_models: list[str],
+    input_file_type: InputFileType,
+    ref_strategy: InputModelRefStrategy | None = None,
+    output_model_type: DataModelType = DataModelType.PydanticBaseModel,
+) -> dict[str, object]:
+    """Load and merge schemas from multiple Python import paths with inheritance support.
+
+    Args:
+        input_models: List of import paths in 'module.path:ObjectName' format
+        input_file_type: Current input file type setting for validation
+        ref_strategy: Strategy for handling referenced types
+        output_model_type: Target output model type for reuse-foreign strategy
+
+    Returns:
+        Merged schema dict with anyOf referencing all root models
+    """
+    import importlib.util  # noqa: PLC0415
+    import sys  # noqa: PLC0415
+
+    if len(input_models) == 1:
+        return _load_model_schema(
+            input_models[0], input_file_type, ref_strategy, output_model_type
+        )
+
+    cwd = str(Path.cwd())
+    if cwd not in sys.path:
+        sys.path.insert(0, cwd)
+
+    model_classes: list[type] = []
+    loaded_modules: dict[str, object] = {}
+
+    for input_model in input_models:
+        modname, sep, qualname = input_model.rpartition(":")
+        if not sep or not modname:
+            msg = f"Invalid --input-model format: {input_model!r}. Expected 'module:Object' or 'path/to/file.py:Object'."
+            raise Error(msg)
+
+        if modname not in loaded_modules:
+            is_path = "/" in modname or "\\" in modname
+            if not is_path and modname.endswith(".py"):
+                is_path = Path(modname).exists()
+
+            if is_path:
+                file_path = Path(modname).resolve()
+                if not file_path.exists():
+                    msg = f"File not found: {modname!r}"
+                    raise Error(msg)
+                module_name = file_path.stem
+                spec = importlib.util.spec_from_file_location(module_name, file_path)
+                if spec is None or spec.loader is None:
+                    msg = f"Cannot load module from {modname!r}"
+                    raise Error(msg)
+                module = importlib.util.module_from_spec(spec)
+                sys.modules[module_name] = module
+                spec.loader.exec_module(module)
+            else:
+                try:
+                    found_spec = importlib.util.find_spec(modname)
+                    if found_spec is None:
+                        msg = f"Cannot find module {modname!r}"
+                        raise Error(msg)
+                    module = importlib.import_module(modname)
+                except ImportError as e:
+                    msg = f"Cannot import module {modname!r}: {e}"
+                    raise Error(msg) from e
+            loaded_modules[modname] = module
+        else:
+            module = loaded_modules[modname]
+
+        try:
+            obj = getattr(module, qualname)
+        except AttributeError as e:
+            msg = f"Module {modname!r} has no attribute {qualname!r}"
+            raise Error(msg) from e
+
+        if not (isinstance(obj, type) and issubclass(obj, BaseModel)):
+            msg = f"Multiple --input-model only supports Pydantic v2 BaseModel classes, got {type(obj).__name__}"
+            raise Error(msg)
+
+        if not hasattr(obj, "model_json_schema"):
+            msg = "Multiple --input-model with Pydantic model requires Pydantic v2 runtime. Please upgrade Pydantic to v2."
+            raise Error(msg)
+
+        model_classes.append(obj)
+
+    if input_file_type not in {InputFileType.Auto, InputFileType.JsonSchema}:
+        msg = (
+            f"--input-file-type must be 'jsonschema' (or omitted) "
+            f"when --input-model points to Pydantic models, "
+            f"got '{input_file_type.value}'"
+        )
+        raise Error(msg)
+
+    schema_generator = _get_input_model_json_schema_class()
+    merged_defs: dict[str, object] = {}
+    root_refs: list[dict[str, str]] = []
+    processed_parents: dict[str, dict[str, object]] = {}
+
+    for model_class in model_classes:
+        model_name = model_class.__name__
+        if hasattr(model_class, "model_rebuild"):
+            _try_rebuild_model(model_class)
+
+        schema = model_class.model_json_schema(schema_generator=schema_generator)
+        schema = _add_python_type_for_unserializable(schema, model_class)
+        schema = _add_python_type_info(schema, model_class)
+
+        schema = _transform_single_model_to_inheritance(
+            schema, model_class, schema_generator, processed_parents
+        )
+
+        if "$defs" in schema:
+            schema_defs = cast("dict[str, object]", schema["$defs"])
+            for k, v in schema_defs.items():
+                if k not in merged_defs:
+                    merged_defs[k] = v
+
+        model_def = {k: v for k, v in schema.items() if k != "$defs"}
+        merged_defs[model_name] = model_def
+
+        root_refs.append({"$ref": f"#/$defs/{model_name}"})
+
+    final_schema: dict[str, object] = {"$defs": merged_defs}
+    if len(root_refs) == 1:
+        final_schema.update(root_refs[0])
+    else:
+        final_schema["anyOf"] = root_refs
+
+    if ref_strategy and ref_strategy != InputModelRefStrategy.RegenerateAll:
+        all_nested_models: dict[str, type] = {}
+        for model_class in model_classes:
+            all_nested_models.update(_collect_nested_models(model_class))
+        final_schema = _filter_defs_by_strategy(
+            final_schema, all_nested_models, output_model_type, ref_strategy
+        )
+
+    return final_schema
+
+
 def _load_model_schema(  # noqa: PLR0912, PLR0914, PLR0915
     input_model: str,
     input_file_type: InputFileType,
@@ -1262,6 +1505,9 @@ def _load_model_schema(  # noqa: PLR0912, PLR0914, PLR0915
         schema = _add_python_type_for_unserializable(schema, obj)
         schema = _add_python_type_info(schema, obj)
 
+        # Transform to inheritance structure if the model has BaseModel parents
+        schema = _transform_single_model_to_inheritance(schema, obj, schema_generator)
+
         if ref_strategy and ref_strategy != InputModelRefStrategy.RegenerateAll:
             nested_models = _collect_nested_models(obj)
             model_name = getattr(obj, "__name__", None)
@@ -1890,7 +2136,7 @@ def main(args: Sequence[str] | None = None) -> Exit:  # noqa: PLR0911, PLR0912,
     try:
         input_: Path | str | ParseResult
         if config.input_model:
-            schema = _load_model_schema(
+            schema = _load_multiple_model_schemas(
                 config.input_model,
                 config.input_file_type,
                 config.input_model_ref_strategy,

src/datamodel_code_generator/_types/graphql_parser_config_dict.py

@@ -30,7 +30,7 @@
     from datamodel_code_generator.types import DataTypeManager
 
 
-class GraphQLParserConfigDict(TypedDict):
+class ParserConfig(TypedDict):
     data_model_type: NotRequired[type[DataModel]]
     data_model_root_type: NotRequired[type[DataModel]]
     data_type_manager_type: NotRequired[type[DataTypeManager]]
@@ -142,5 +142,8 @@ class GraphQLParserConfigDict(TypedDict):
     read_only_write_only_model_type: NotRequired[ReadOnlyWriteOnlyModelType | None]
     field_type_collision_strategy: NotRequired[FieldTypeCollisionStrategy | None]
     target_pydantic_version: NotRequired[TargetPydanticVersion | None]
+
+
+class GraphQLParserConfigDict(ParserConfig):
     data_model_scalar_type: NotRequired[type[DataModel]]
     data_model_union_type: NotRequired[type[DataModel]]

src/datamodel_code_generator/_types/jsonschema_parser_config_dict.py

@@ -30,7 +30,7 @@
     from datamodel_code_generator.types import DataTypeManager
 
 
-class JSONSchemaParserConfigDict(TypedDict):
+class ParserConfig(TypedDict):
     data_model_type: NotRequired[type[DataModel]]
     data_model_root_type: NotRequired[type[DataModel]]
     data_type_manager_type: NotRequired[type[DataTypeManager]]
@@ -142,3 +142,7 @@ class JSONSchemaParserConfigDict(TypedDict):
     read_only_write_only_model_type: NotRequired[ReadOnlyWriteOnlyModelType | None]
     field_type_collision_strategy: NotRequired[FieldTypeCollisionStrategy | None]
     target_pydantic_version: NotRequired[TargetPydanticVersion | None]
+
+
+class JSONSchemaParserConfigDict(ParserConfig):
+    pass

src/datamodel_code_generator/_types/openapi_parser_config_dict.py

@@ -31,7 +31,7 @@
     from datamodel_code_generator.types import DataTypeManager
 
 
-class OpenAPIParserConfigDict(TypedDict):
+class ParserConfig(TypedDict):
     data_model_type: NotRequired[type[DataModel]]
     data_model_root_type: NotRequired[type[DataModel]]
     data_type_manager_type: NotRequired[type[DataTypeManager]]
@@ -143,6 +143,13 @@ class OpenAPIParserConfigDict(TypedDict):
     read_only_write_only_model_type: NotRequired[ReadOnlyWriteOnlyModelType | None]
     field_type_collision_strategy: NotRequired[FieldTypeCollisionStrategy | None]
     target_pydantic_version: NotRequired[TargetPydanticVersion | None]
+
+
+class JSONSchemaParserConfig(ParserConfig):
+    pass
+
+
+class OpenAPIParserConfigDict(JSONSchemaParserConfig):
     openapi_scopes: NotRequired[list[OpenAPIScope] | None]
     include_path_parameters: NotRequired[bool]
     use_status_code_in_response_name: NotRequired[bool]

src/datamodel_code_generator/arguments.py

@@ -159,8 +159,10 @@ def start_section(self, heading: str | None) -> None:
 )
 base_options.add_argument(
     "--input-model",
+    action="append",
     help="Python import path to a Pydantic v2 model or schema dict "
     "(e.g., 'mypackage.module:ClassName' or 'mypackage.schemas:SCHEMA_DICT'). "
+    "Can be specified multiple times for related models with inheritance. "
     "For dict input, --input-file-type is required. "
     "Cannot be used with --input or --url.",
     metavar="MODULE:NAME",

tests/data/expected/main/input_model/forked_inheritance.py

@@ -0,0 +1,26 @@
+# generated by datamodel-codegen:
+#   filename:  <stdin>
+#   timestamp: 1985-10-26T08:21:00+00:00
+
+from __future__ import annotations
+
+from typing import TypeAlias, TypedDict
+
+
+class GrandParent(TypedDict):
+    grand_field: str
+
+
+class Parent(GrandParent):
+    parent_field: int
+
+
+class ChildA(Parent):
+    child_a_field: float
+
+
+class ChildB(Parent):
+    child_b_field: bool
+
+
+Model: TypeAlias = ChildA | ChildB