handle pydantic v1 gracefully (#195)

* check pydantic version

* forward ref the ZepModel type

* mod init

* lint

src/zep_cloud/external_clients/memory.py

@@ -1,16 +1,23 @@
 import datetime
 import json
 import typing
+from packaging import version
+
+import pydantic
 
 from zep_cloud.core.client_wrapper import AsyncClientWrapper, SyncClientWrapper
-from zep_cloud.extractor.models import ZepModel
 from zep_cloud.memory.client import (
     AsyncMemoryClient as AsyncBaseMemoryClient,
 )
 from zep_cloud.memory.client import (
     MemoryClient as BaseMemoryClient,
 )
 
+if typing.TYPE_CHECKING:
+    from zep_cloud.extractor.models import ZepModel
+
+MIN_PYDANTIC_VERSION = "2.0"
+
 
 class MemoryClient(BaseMemoryClient):
     def __init__(self, *, client_wrapper: SyncClientWrapper):
@@ -19,7 +26,7 @@ def __init__(self, *, client_wrapper: SyncClientWrapper):
     def extract(
         self,
         session_id: str,
-        model: ZepModel,
+        model: "ZepModel",
         current_date_time: typing.Optional[datetime.datetime] = None,
         last_n: int = 4,
         validate: bool = False,
@@ -65,16 +72,22 @@ class CustomerInfo(ZepModel):
 
         print(customer_data.name)  # Access extracted and validated customer name
         """
+
+        if version.parse(pydantic.VERSION) < version.parse(MIN_PYDANTIC_VERSION):
+            raise RuntimeError(
+                f"Pydantic version {MIN_PYDANTIC_VERSION} or greater is required."
+            )
+
         model_schema = json.dumps(model.model_json_schema())
 
         result = self.extract_data(
             session_id=session_id,
             model_schema=model_schema,
             validate=validate,
             last_n=last_n,
-            current_date_time=current_date_time.isoformat()
-            if current_date_time
-            else None,
+            current_date_time=(
+                current_date_time.isoformat() if current_date_time else None
+            ),
         )
 
         return model.model_validate(result)
@@ -87,62 +100,68 @@ def __init__(self, *, client_wrapper: AsyncClientWrapper):
     async def extract(
         self,
         session_id: str,
-        model: ZepModel,
+        model: "ZepModel",
         current_date_time: typing.Optional[datetime.datetime] = None,
         last_n: int = 4,
         validate: bool = False,
     ):
         """Extracts structured data from a session based on a ZepModel schema.
-       This method retrieves data based on a given model and session details.
-       It then returns the extracted and validated data as an instance of the given ZepModel.
-
-       Parameters
-       ----------
-       session_id: str
-           Session ID.
-       model: ZepModel
-           An instance of a ZepModel subclass defining the expected data structure and field types.
-       current_date_time: typing.Optional[datetime.datetime]
-           Your current date and time in ISO 8601 format including timezone.
-           This is used for determining relative dates.
-       last_n: typing.Optional[int]
-           The number of messages in the chat history from which to extract data.
-       validate: typing.Optional[bool]
-           Validate that the extracted data is present in the dialog and correct per the field description.
-           Mitigates hallucination, but is slower and may result in false negatives.
-
-       Returns
-       -------
-       ZepModel: An instance of the provided ZepModel subclass populated with the
-           extracted and validated data.
-
-       Examples
-       --------
-       class CustomerInfo(ZepModel):
-           name: Optional[ZepText] = Field(description="Customer name", default=None)
-           name: Optional[ZepEmail] = Field(description="Customer email", default=None)
-           signup_date: Optional[ZepDate] = Field(description="Customer Sign up date", default=None)
-
-       client = AsyncMemoryClient(...)
-
-       customer_data = await client.memory.extract(
-           session_id="session123",
-           model=CustomerInfo(),
-           current_date_time=datetime.datetime.now(),  # Filter data up to now
-       )
-
-       print(customer_data.name)  # Access extracted and validated customer name
-       """
+        This method retrieves data based on a given model and session details.
+        It then returns the extracted and validated data as an instance of the given ZepModel.
+
+        Parameters
+        ----------
+        session_id: str
+            Session ID.
+        model: ZepModel
+            An instance of a ZepModel subclass defining the expected data structure and field types.
+        current_date_time: typing.Optional[datetime.datetime]
+            Your current date and time in ISO 8601 format including timezone.
+            This is used for determining relative dates.
+        last_n: typing.Optional[int]
+            The number of messages in the chat history from which to extract data.
+        validate: typing.Optional[bool]
+            Validate that the extracted data is present in the dialog and correct per the field description.
+            Mitigates hallucination, but is slower and may result in false negatives.
+
+        Returns
+        -------
+        ZepModel: An instance of the provided ZepModel subclass populated with the
+            extracted and validated data.
+
+        Examples
+        --------
+        class CustomerInfo(ZepModel):
+            name: Optional[ZepText] = Field(description="Customer name", default=None)
+            name: Optional[ZepEmail] = Field(description="Customer email", default=None)
+            signup_date: Optional[ZepDate] = Field(description="Customer Sign up date", default=None)
+
+        client = AsyncMemoryClient(...)
+
+        customer_data = await client.memory.extract(
+            session_id="session123",
+            model=CustomerInfo(),
+            current_date_time=datetime.datetime.now(),  # Filter data up to now
+        )
+
+        print(customer_data.name)  # Access extracted and validated customer name
+        """
+
+        if version.parse(pydantic.VERSION) < version.parse(MIN_PYDANTIC_VERSION):
+            raise RuntimeError(
+                f"Pydantic version {MIN_PYDANTIC_VERSION} or greater is required."
+            )
+
         model_schema = json.dumps(model.model_json_schema())
 
         result = await self.extract_data(
             session_id=session_id,
             model_schema=model_schema,
             validate=validate,
             last_n=last_n,
-            current_date_time=current_date_time.isoformat()
-            if current_date_time
-            else None,
+            current_date_time=(
+                current_date_time.isoformat() if current_date_time else None
+            ),
         )
 
         return model.model_validate(result)

src/zep_cloud/extractor/__init__.py

@@ -1,15 +1,40 @@
-from zep_cloud.extractor.models import (
-    ZepModel,
-    ZepText,
-    ZepNumber,
-    ZepFloat,
-    ZepRegex,
-    ZepZipCode,
-    ZepDate,
-    ZepDateTime,
-    ZepEmail,
-    ZepPhoneNumber,
-)
+# mypy: disable-error-code=no-redef
+
+from typing import Type
+
+# Zep Extraction requires Pydantic v2. If v2 is not installed, catch the error
+# and set the variables to PydanticV2Required
+
+
+class PydanticV2Required:
+    def __init__(self, *args, **kwargs):
+        raise RuntimeError("Pydantic v2 is required to use this class.")
+
+
+try:
+    from zep_cloud.extractor.models import (
+        ZepModel,
+        ZepText,
+        ZepNumber,
+        ZepFloat,
+        ZepRegex,
+        ZepZipCode,
+        ZepDate,
+        ZepDateTime,
+        ZepEmail,
+        ZepPhoneNumber,
+    )
+except ImportError:
+    ZepModel: Type = PydanticV2Required
+    ZepText: Type = PydanticV2Required
+    ZepNumber: Type = PydanticV2Required
+    ZepFloat: Type = PydanticV2Required
+    ZepRegex: Type = PydanticV2Required
+    ZepZipCode: Type = PydanticV2Required
+    ZepDate: Type = PydanticV2Required
+    ZepDateTime: Type = PydanticV2Required
+    ZepEmail: Type = PydanticV2Required
+    ZepPhoneNumber: Type = PydanticV2Required
 
 __all__ = [
     "ZepModel",