fix: Model and Parser protocols added

dataclasses_avroschema/dacite_config.py

@@ -53,16 +53,16 @@ def generate_dacite_config(model: typing.Type["AvroModel"]) -> Config:
     """
     Get the default config for dacite and always include the self reference
     """
-    # We need to make sure that the `avro schemas` has been generated, otherwise cls._klass is empty
+    # We need to make sure that the `avro schemas` has been generated, otherwise cls._dataclass is empty
     # It won't affect the performance because the rendered schema will be store in model._rendered_schema
     model.generate_schema()
-    dacite_user_config = model._metadata.dacite_config  # type: ignore
+    dacite_user_config = model._parser.metadata.dacite_config  # type: ignore
 
     dacite_config = {
         "check_types": False,
         "cast": [],
         "forward_references": {
-            model._klass.__name__: model._klass,  # type: ignore
+            model._parser.dataclass.__name__: model._parser.dataclass,  # type: ignore
         },
         "type_hooks": {
             datetime: parse_datetime,

dataclasses_avroschema/faust/main.py

@@ -14,9 +14,6 @@
     raise Exception("faust-streaming must be installed in order to use AvroRecord") from ex  # pragma: no cover
 
 
-CT = typing.TypeVar("CT", bound="AvroRecord")
-
-
 class AvroRecord(Record, AvroModel):  # type: ignore
     def validate_avro(self) -> bool:
         """
@@ -56,5 +53,5 @@ def to_dict(self) -> JsonDict:
         return self.standardize_type(include_type=False)
 
     @classmethod
-    def _generate_parser(cls: typing.Type[CT]) -> FaustParser:
-        return FaustParser(type=cls._klass, metadata=cls.get_metadata(), parent=cls._parent or cls)
+    def _generate_parser(cls: typing.Type["AvroRecord"]) -> FaustParser:
+        return FaustParser(type=cls, parent=cls._parent or cls)

dataclasses_avroschema/faust/parser.py

@@ -5,8 +5,20 @@
 from dataclasses_avroschema.fields.fields import AvroField
 from dataclasses_avroschema.parser import Parser
 
+if typing.TYPE_CHECKING:
+    from .main import AvroRecord  # pragma: no cover
+
 
 class FaustParser(Parser):
+    def __init__(
+        self,
+        type,
+        parent,
+    ):
+        super().__init__(type, parent)
+        self.type: typing.Type["AvroRecord"]
+        self.parent: typing.Type["AvroRecord"]
+
     def parse_fields(self, exclude: typing.List) -> typing.List[Field]:
         schema_fields = []
 

dataclasses_avroschema/fields/fields.py

@@ -831,7 +831,10 @@ def get_avro_type(self) -> typing.Union[str, typing.List, typing.Dict]:
         meta = getattr(self.type, "Meta", type)
         metadata = utils.SchemaMetadata.create(meta)
 
-        alias = self.parent._metadata.get_alias_nested_items(self.name) or metadata.get_alias_nested_items(self.name)  # type: ignore  # noqa E501
+        if self.parent is not None and self.parent._parser is not None:
+            alias = self.parent._parser.metadata.get_alias_nested_items(self.name)
+        else:
+            alias = metadata.get_alias_nested_items(self.name)  # type: ignore  # noqa E501
 
         # The priority for the schema name
         # 1. Check if exists an alias_nested_items in parent llass or Meta class of own model

dataclasses_avroschema/main.py

@@ -2,7 +2,7 @@
 import inspect
 import json
 from collections import OrderedDict
-from typing import Any, Dict, List, Optional, Set, Type, TypeVar, Union
+from typing import Any, Dict, List, Optional, Set, Type, Union
 
 from dacite import Config, from_dict
 from fastavro.validation import validate
@@ -12,41 +12,20 @@
 from .fields.base import Field
 from .parser import Parser
 from .types import JsonDict
-from .utils import SchemaMetadata, UserDefinedType, standardize_custom_type
-
-CT = TypeVar("CT", bound="AvroModel")
-
+from .utils import UserDefinedType, standardize_custom_type
 
 _schemas_cache: Dict["Type[AvroModel]", dict] = {}
 _dacite_config_cache: Dict["Type[AvroModel]", Config] = {}
 
 
 class AvroModel:
     _parser: Optional[Parser] = None
-    _klass: Optional[Type] = None
-    _metadata: Optional[SchemaMetadata] = None
-    _parent: Any = None
+    _parent: Optional[Type["AvroModel"]] = None
     _user_defined_types: Set[UserDefinedType] = set()
     _rendered_schema: OrderedDict = dataclasses.field(default_factory=OrderedDict)
 
     @classmethod
-    def generate_dataclass(cls: "Type[CT]") -> "Type[CT]":
-        if cls is AvroModel:
-            raise AttributeError("Schema generation must be called on a subclass of AvroModel, not AvroModel itself.")
-
-        if dataclasses.is_dataclass(cls):
-            return cls  # type: ignore
-        return dataclasses.dataclass(cls)
-
-    @classmethod
-    def get_metadata(cls: "Type[CT]") -> SchemaMetadata:
-        if cls._metadata is None:
-            meta = getattr(cls._klass, "Meta", type)
-            cls._metadata = SchemaMetadata.create(meta)
-        return cls._metadata
-
-    @classmethod
-    def get_fullname(cls) -> str:
+    def get_fullname(cls: Type["AvroModel"]) -> str:
         """
         Fullname is composed of two parts: a name and a namespace
         separated by a dot. A namespace is a dot-separated sequence of such names.
@@ -56,26 +35,25 @@ def get_fullname(cls) -> str:
         """
         # we need to make sure that the schema has been generated
         cls.generate_schema()
-        metadata = cls.get_metadata()
+        assert cls._parser
+        metadata = cls._parser.metadata
 
         if metadata.namespace:
             # if the current record has a namespace we use it
             return f"{metadata.namespace}.{cls.__name__}"
         elif cls._parent is not None:
             # if the record has a parent then we try to use the parent namespace
-            parent_metadata = cls._parent.get_metadata()
+            assert cls._parent._parser
+            parent_metadata = cls._parent._parser.metadata
             if parent_metadata.namespace:
                 return f"{parent_metadata.namespace}.{cls.__name__}"
         return cls.__name__
 
     @classmethod
     def generate_schema(
-        cls: "Type[CT]", schema_type: serialization.SerializationType = "avro"
+        cls: Type["AvroModel"], schema_type: serialization.SerializationType = "avro"
     ) -> Optional[OrderedDict]:
         if cls._parser is None:
-            # Generate dataclass and metadata
-            cls._klass = cls.generate_dataclass()
-
             # let's live open the possibility to define different
             # schema definitions like json
             if schema_type == "avro":
@@ -98,17 +76,17 @@ def _get_serialization_context(cls) -> JsonDict:
         return {user_type.model.__name__: user_type.model for user_type in cls._user_defined_types}
 
     @classmethod
-    def _generate_parser(cls: "Type[CT]") -> Parser:
-        return Parser(type=cls._klass, metadata=cls.get_metadata(), parent=cls._parent or cls)
+    def _generate_parser(cls: Type["AvroModel"]) -> Parser:
+        return Parser(type=cls, parent=cls._parent or cls)
 
     @classmethod
-    def avro_schema(cls: "Type[CT]", case_type: Optional[str] = None, **kwargs) -> str:
+    def avro_schema(cls: Type["AvroModel"], case_type: Optional[str] = None, **kwargs) -> str:
         return json.dumps(cls.avro_schema_to_python(case_type=case_type), **kwargs)
 
     @classmethod
     def avro_schema_to_python(
-        cls: "Type[CT]",
-        parent: Optional["CT"] = None,
+        cls: Type["AvroModel"],
+        parent: Optional[Type["AvroModel"]] = None,
         case_type: Optional[str] = None,
     ) -> Dict[str, Any]:
         if parent is not None:
@@ -135,13 +113,13 @@ def avro_schema_to_python(
         return json.loads(json.dumps(avro_schema))
 
     @classmethod
-    def get_fields(cls: "Type[CT]") -> List[Field]:
+    def get_fields(cls: Type["AvroModel"]) -> List[Field]:
         if cls._parser is None:
             cls.generate_schema()
         return cls._parser.fields  # type: ignore
 
     @classmethod
-    def _reset_parser(cls: "Type[CT]") -> None:
+    def _reset_parser(cls: Type["AvroModel"]) -> None:
         """
         Reset all the values to original state.
         """
@@ -151,12 +129,12 @@ def _reset_parser(cls: "Type[CT]") -> None:
 
     @classmethod
     def deserialize(
-        cls: "Type[CT]",
+        cls: Type["AvroModel"],
         data: bytes,
         serialization_type: serialization.SerializationType = "avro",
         create_instance: bool = True,
-        writer_schema: Optional[Union[JsonDict, "Type[CT]"]] = None,
-    ) -> Union[JsonDict, CT]:
+        writer_schema: Optional[Union[JsonDict, Type["AvroModel"]]] = None,
+    ) -> Union[JsonDict, "AvroModel"]:
         payload = cls.deserialize_to_python(data, serialization_type, writer_schema)
         obj = cls.parse_obj(payload)
 
@@ -166,10 +144,10 @@ def deserialize(
 
     @classmethod
     def deserialize_to_python(  # This can be used straight with a pydantic dataclass to bypass dacite
-        cls: "Type[CT]",
+        cls: Type["AvroModel"],
         data: bytes,
         serialization_type: serialization.SerializationType = "avro",
-        writer_schema: Union[JsonDict, "Type[CT]", None] = None,
+        writer_schema: Union[JsonDict, Type["AvroModel"], None] = None,
     ) -> dict:
         if inspect.isclass(writer_schema) and issubclass(writer_schema, AvroModel):
             # mypy does not understand redefinitions
@@ -188,15 +166,15 @@ def deserialize_to_python(  # This can be used straight with a pydantic dataclas
         )
 
     @classmethod
-    def parse_obj(cls: "Type[CT]", data: Dict) -> CT:
+    def parse_obj(cls: Type["AvroModel"], data: Dict) -> "AvroModel":
         config = _dacite_config_cache.get(cls)
         if config is None:
             config = generate_dacite_config(cls)
             _dacite_config_cache[cls] = config
         return from_dict(data_class=cls, data=data, config=config)
 
     @classmethod
-    def fake(cls: "Type[CT]", **data: Any) -> CT:
+    def fake(cls: Type["AvroModel"], **data: Any) -> "AvroModel":
         """
         Creates a fake instance of the model.
 
@@ -215,7 +193,7 @@ def asdict(self) -> JsonDict:
             field.name: standardize_custom_type(
                 field_name=field.name, value=getattr(self, field.name), model=self, base_class=AvroModel
             )
-            for field in dataclasses.fields(self)  # type: ignore
+            for field in dataclasses.fields(self)  # type: ignore[arg-type]
         }
 
     def serialize(self, serialization_type: serialization.SerializationType = "avro") -> bytes:

dataclasses_avroschema/parser.py

@@ -3,14 +3,14 @@
 import typing
 from collections import OrderedDict
 
-from . import utils
 from .fields.base import Field
-from .version import PY_VERSION
+from .utils import SchemaMetadata
 
-KWARGS = {"slots": True} if PY_VERSION >= (3, 10) else {}
+if typing.TYPE_CHECKING:
+    from .main import AvroModel
+    # from .protocol import ModelProtocol
 
 
-@dataclasses.dataclass(**KWARGS)  # type: ignore
 class Parser:
     """
     Parse python dataclasses to represent it as an avro schema.
@@ -19,18 +19,34 @@ class Parser:
     be represented as an avro type.
     """
 
-    type: typing.Any
-    parent: typing.Any
-    metadata: utils.SchemaMetadata
-    fields: typing.List[Field] = dataclasses.field(default_factory=list)
-    # mapping of field_name: Field
-    fields_map: typing.Dict[str, Field] = dataclasses.field(default_factory=dict)
+    def __init__(
+        self,
+        type: typing.Type["AvroModel"],
+        parent: typing.Type["AvroModel"],
+    ):
+        self.type = type
+        self.parent = parent
 
-    def __post_init__(self) -> None:
+        # generate the dataclass for thr given type
+        self.dataclass = self.generate_dataclass()
+
+        meta = getattr(type, "Meta", type)
+        self.metadata = SchemaMetadata.create(meta)
         exclude = self.metadata.exclude
+
         self.fields = self.parse_fields(exclude=exclude)
         self.fields_map = {field.name: field for field in self.fields}
 
+    def generate_dataclass(self) -> typing.Type:
+        from .main import AvroModel
+
+        if self.type is AvroModel:
+            raise AttributeError("Schema generation must be called on a subclass of AvroModel, not AvroModel itself.")
+
+        if dataclasses.is_dataclass(self.type):
+            return self.type
+        return dataclasses.dataclass(self.type)
+
     def parse_fields(self, exclude: typing.List) -> typing.List[Field]:
         from .fields.fields import AvroField
 
@@ -44,19 +60,19 @@ def parse_fields(self, exclude: typing.List) -> typing.List[Field]:
                 model_metadata=self.metadata,
                 parent=self.parent,
             )
-            for dataclass_field in dataclasses.fields(self.type)
+            for dataclass_field in dataclasses.fields(self.dataclass)
             if dataclass_field.name not in exclude
         ]
 
     def get_fields_map(self) -> typing.Dict[str, Field]:
         return self.fields_map
 
     def get_schema_name(self) -> str:
-        return self.type._metadata.schema_name or self.type.__name__
+        return self.metadata.schema_name or self.type.__name__
 
     def generate_documentation(self) -> typing.Optional[str]:
         if isinstance(self.metadata.schema_doc, str):
-            doc = self.metadata.schema_doc
+            doc: typing.Optional[str] = self.metadata.schema_doc
         else:
             doc = self.type.__doc__
             # dataclasses create a (in avro context) useless docstring by default,