An error occurred: SQLDatabaseChain is not fully defined; you should define BaseCache, then call SQLDatabaseChain.model_rebuild().

[OpenSystemsConsulting]

SO I am doing a basic python script to connect to an existing oracle database and nothing I do will stop this error from comming up.
I think it may be some version issue with particular libraries. Python 3.12.3
Here is the code

#from langchain_community.llms import OpenAI, Cohere
#from langchain import Cohere
from langchain_community.llms import Cohere
from langchain_community.utilities import SQLDatabase
from langchain_experimental.sql import SQLDatabaseChain
from typing import Dict, Any
from cache.base_cache import BaseCache
from langchain_community.cache import InMemoryCache
from pydantic import BaseModel

Define the BaseCache class

class BaseCache(BaseModel):
cache: Dict[str, Any]

  def get(self, key: str) -> Any:
      return self.cache.get(key)

  def set(self, key: str, value: Any) -> None:
      self.cache[key] = value

Initialize the cache

cache = BaseCache(cache={})

import cx_Oracle
import os
import cohere
from langchain.chains import load_chain
#from cache.in_memory_cache import InMemoryCache
from langchain_community.cache import InMemoryCache
import os

#cache = BaseCache()

COHERE_API_KEY='WQkUQh3NrSYVJBTEWNjRHmBX3912HUVuuePk4H4N'
os.environ["COHERE_API_KEY"] = COHERE_API_KEY

lib_dir = os.path.join(os.environ.get("HOME"), "Development", "instantclient_23_6")
cx_Oracle.init_oracle_client(lib_dir=lib_dir)

hostname='192.168.160.83'
port='1521'
service_name='xxxxxx'
username='xxxxxxxx'
password='xxxxxx'

cx_Oracle.init_oracle_client(lib_dir=lib_dir)

oracle_connection_string_fmt = (
'oracle+cx_oracle://{username}:{password}@' +
cx_Oracle.makedsn('{hostname}', '{port}', service_name='{service_name}')
)
url = oracle_connection_string_fmt.format(
username=username, password=password,
hostname=hostname, port=port,
service_name=service_name,
)
from sqlalchemy import create_engine
engine=create_engine(url, echo=True)
db = SQLDatabase(engine)

#llm = Cohere(temperature=1, verbose=True)
#db_chain = SQLDatabaseChain.from_llm(llm, db,cache=cache, verbose=True)
#db_chain.model_rebuild()
#db_chain.run("Is Casey Brown in the database?")

try:
llm = Cohere(temperature=1, verbose=True)

    # Initialize SQLDatabaseChain directly with all parameters
   db_chain = SQLDatabaseChain( llm_chain=llm, database=db, cache=cache, verbose=True)

   rint("Attempting model rebuild...")
   db_chain.model_rebuild()

   print("Running query...")
   result = db_chain.run("Is Casey Brown in the database?")
   rint(result)

except Exception as e:
print(f"An error occurred: {str(e)}")

Any Ideas

[syilayavilli]

from langchain_google_genai import ChatGoogleGenerativeAI
from langchain_google_genai import GoogleGenerativeAI
from langchain_experimental.sql import SQLDatabaseChain
gooapi_key="xxxx"

llm = GoogleGenerativeAI(model="models/gemini-1.0-pro", google_api_key=gooapi_key, temperature=0.2)

db_chain = SQLDatabaseChain.from_llm(llm=llm, db=db, verbose=True)
result = db_chain.run("What is the min date?")

I get PydanticUserError: SQLDatabaseChain is not fully defined; you should define BaseCache, then call SQLDatabaseChain.model_rebuild().

For further information visit https://errors.pydantic.dev/2.10/u/class-not-fully-defined

[syilayavilli]

Re-write on a clean Jupyter notebook to avoid the error

Also look at the
#25282

experimental: Ensure MYSQL queries are generated perfectly so can be used by the next sequence in SQLDatabaseChain to 
execute them directly on database"

infra: langchain/libs/langchain/langchain/chains/sql_database/prompt.py --> modified _mysql_prompt

Description: This PR addresses an issue where the SQL queries generated by the GoogleGenerativeAI model in the langchain-
experimental package included unwanted syntax. This query when used through SQLDatabaseChain caused the error. The 
generated queries now start directly with SELECT and do not contain any additional formatting or text. The fix involved modifying 
the prompt in the prompt.py file to ensure clean SQL query generation.