remove more custom serpyco logic

Author: aaronsteers

airbyte_cdk/models/airbyte_protocol.py

@@ -7,87 +7,3 @@
 from typing import Annotated, Any, Dict, List, Mapping, Optional, Union
 
 from airbyte_protocol_dataclasses.models import *  # noqa: F403  # Allow '*'
-
-if sys.platform == 'emscripten':
-    from serpyco.metadata import Alias
-else:
-    from serpyco_rs.metadata import Alias
-
-# ruff: noqa: F405  # ignore fuzzy import issues with 'import *'
-
-
-@dataclass
-class AirbyteStateBlob:
-    """
-    A dataclass that dynamically sets attributes based on provided keyword arguments and positional arguments.
-    Used to "mimic" pydantic Basemodel with ConfigDict(extra='allow') option.
-
-    The `AirbyteStateBlob` class allows for flexible instantiation by accepting any number of keyword arguments
-    and positional arguments. These are used to dynamically update the instance's attributes. This class is useful
-    in scenarios where the attributes of an object are not known until runtime and need to be set dynamically.
-
-    Attributes:
-        kwargs (InitVar[Mapping[str, Any]]): A dictionary of keyword arguments used to set attributes dynamically.
-
-    Methods:
-        __init__(*args: Any, **kwargs: Any) -> None:
-            Initializes the `AirbyteStateBlob` by setting attributes from the provided arguments.
-
-        __eq__(other: object) -> bool:
-            Checks equality between two `AirbyteStateBlob` instances based on their internal dictionaries.
-            Returns `False` if the other object is not an instance of `AirbyteStateBlob`.
-    """
-
-    kwargs: InitVar[Mapping[str, Any]]
-
-    def __init__(self, *args: Any, **kwargs: Any) -> None:
-        # Set any attribute passed in through kwargs
-        for arg in args:
-            self.__dict__.update(arg)
-        for key, value in kwargs.items():
-            setattr(self, key, value)
-
-    def __eq__(self, other: object) -> bool:
-        return (
-            False
-            if not isinstance(other, AirbyteStateBlob)
-            else bool(self.__dict__ == other.__dict__)
-        )
-
-
-# The following dataclasses have been redeclared to include the new version of AirbyteStateBlob
-@dataclass
-class AirbyteStreamState:
-    stream_descriptor: StreamDescriptor  # type: ignore [name-defined]
-    stream_state: Optional[AirbyteStateBlob] = None
-
-
-@dataclass
-class AirbyteGlobalState:
-    stream_states: List[AirbyteStreamState]
-    shared_state: Optional[AirbyteStateBlob] = None
-
-
-@dataclass
-class AirbyteStateMessage:
-    type: Optional[AirbyteStateType] = None  # type: ignore [name-defined]
-    stream: Optional[AirbyteStreamState] = None
-    global_: Annotated[AirbyteGlobalState | None, Alias("global")] = (
-        None  # "global" is a reserved keyword in python ⇒ Alias is used for (de-)serialization
-    )
-    data: Optional[Dict[str, Any]] = None
-    sourceStats: Optional[AirbyteStateStats] = None  # type: ignore [name-defined]
-    destinationStats: Optional[AirbyteStateStats] = None  # type: ignore [name-defined]
-
-
-@dataclass
-class AirbyteMessage:
-    type: Type  # type: ignore [name-defined]
-    log: Optional[AirbyteLogMessage] = None  # type: ignore [name-defined]
-    spec: Optional[ConnectorSpecification] = None  # type: ignore [name-defined]
-    connectionStatus: Optional[AirbyteConnectionStatus] = None  # type: ignore [name-defined]
-    catalog: Optional[AirbyteCatalog] = None  # type: ignore [name-defined]
-    record: Optional[AirbyteRecordMessage] = None  # type: ignore [name-defined]
-    state: Optional[AirbyteStateMessage] = None
-    trace: Optional[AirbyteTraceMessage] = None  # type: ignore [name-defined]
-    control: Optional[AirbyteControlMessage] = None  # type: ignore [name-defined]

airbyte_cdk/test/entrypoint_wrapper.py

@@ -30,11 +30,6 @@
 import orjson
 from pydantic import ValidationError as V2ValidationError
 
-if sys.platform == 'emscripten':
-    from serpyco import SchemaValidationError
-else:
-    from serpyco_rs import SchemaValidationError
-
 from airbyte_cdk.entrypoint import AirbyteEntrypoint
 from airbyte_cdk.exception_handler import assemble_uncaught_exception
 from airbyte_cdk.logger import AirbyteLogFormatter
@@ -55,6 +50,13 @@
 from airbyte_cdk.sources import Source
 from airbyte_cdk.test.models.scenario import ExpectedOutcome
 
+JsonValidationErrors: tuple[type[Exception], ...] = (orjson.JSONDecodeError,)
+# Conditionally import and create a union type for exception handling
+if sys.platform != "emscripten":
+    from serpyco_rs import SchemaValidationError
+
+    JsonValidationErrors = (orjson.JSONDecodeError, SchemaValidationError)
+
 
 class AirbyteEntrypointException(Exception):
     """Exception raised for errors in the AirbyteEntrypoint execution.
@@ -123,7 +125,7 @@ def __init__(
     def _parse_message(message: str) -> AirbyteMessage:
         try:
             return AirbyteMessageSerializer.load(orjson.loads(message))
-        except (orjson.JSONDecodeError, SchemaValidationError):
+        except JsonValidationErrors:
             # The platform assumes that logs that are not of AirbyteMessage format are log messages
             return AirbyteMessage(
                 type=Type.LOG, log=AirbyteLogMessage(level=Level.INFO, message=message)

unit_tests/sources/test_source.py

@@ -11,11 +11,7 @@
 
 import orjson
 import pytest
-
-if sys.platform == 'emscripten':
-    from serpyco import SchemaValidationError
-else:
-    from serpyco_rs import SchemaValidationError
+from serpyco_rs import SchemaValidationError
 
 from airbyte_cdk.models import (
     AirbyteGlobalState,