"""Wrapper around IBM GENAI APIs for use in Langchain"""
import logging
from pathlib import Path
from typing import Any, Iterator, Optional, Union
from pydantic import ConfigDict
from pydantic.v1 import validator
from genai import Client
from genai._types import EnumLike
from genai._utils.general import to_model_optional
from genai.extensions._common.utils import (
_prepare_chat_generation_request,
create_generation_info_from_response,
)
from genai.schema import (
AIMessage,
BaseMessage,
HumanMessage,
ModerationParameters,
SystemMessage,
TextGenerationParameters,
TrimMethod,
)
try:
from langchain_core.callbacks.manager import CallbackManagerForLLMRun
from langchain_core.language_models.chat_models import BaseChatModel
from langchain_core.messages import AIMessage as LCAIMessage
from langchain_core.messages import BaseMessage as LCBaseMessage
from langchain_core.messages import ChatMessage as LCChatMessage
from langchain_core.messages import HumanMessage as LCHumanMessage
from langchain_core.messages import SystemMessage as LCSystemMessage
from langchain_core.messages import get_buffer_string
from langchain_core.outputs import ChatGeneration, ChatResult
from genai.extensions.langchain.utils import (
CustomAIMessageChunk,
CustomChatGenerationChunk,
create_llm_output,
dump_optional_model,
load_config,
update_token_usage_stream,
)
except ImportError:
raise ImportError("Could not import langchain: Please install ibm-generative-ai[langchain] extension.") # noqa: B904
__all__ = ["LangChainChatInterface"]
logger = logging.getLogger(__name__)
Message = Union[LCBaseMessage, BaseMessage]
Messages = Union[list[LCBaseMessage], list[Message]]
def _convert_message_to_genai(message: Message) -> BaseMessage:
def convert_message_content(content: Any) -> str:
if not isinstance(content, str):
raise TypeError(
f"Cannot convert non-string message content. Got {content} of type {type(content)}, expected string."
)
return content
if isinstance(message, BaseMessage):
return message
elif isinstance(message, LCChatMessage) or isinstance(message, LCHumanMessage):
return HumanMessage(content=convert_message_content(message.content))
elif isinstance(message, LCAIMessage):
return AIMessage(content=convert_message_content(message.content))
elif isinstance(message, LCSystemMessage):
return SystemMessage(content=convert_message_content(message.content))
else:
raise ValueError(f"Got unknown message type '{message}'")
def _convert_messages_to_genai(messages: Messages) -> list[BaseMessage]:
return [_convert_message_to_genai(msg) for msg in messages]
[docs]
class LangChainChatInterface(BaseChatModel):
"""
Class representing the LangChainChatInterface for interacting with the LangChain chat API.
Example::
from genai import Client, Credentials
from genai.extensions.langchain import LangChainChatInterface
from langchain_core.messages import HumanMessage, SystemMessage
from genai.schema import TextGenerationParameters
client = Client(credentials=Credentials.from_env())
llm = LangChainChatInterface(
client=client,
model_id="meta-llama/llama-3-70b-instruct",
parameters=TextGenerationParameters(
max_new_tokens=250,
)
)
response = chat_model.generate(messages=[HumanMessage(content="Hello world!")])
print(response)
"""
model_config = ConfigDict(extra="forbid", protected_namespaces=())
client: Client
model_id: str
prompt_id: Optional[str] = None
parameters: Optional[TextGenerationParameters] = None
moderations: Optional[ModerationParameters] = None
parent_id: Optional[str] = None
prompt_template_id: Optional[str] = None
trim_method: Optional[EnumLike[TrimMethod]] = None
use_conversation_parameters: Optional[bool] = None
conversation_id: Optional[str] = None
streaming: Optional[bool] = None
[docs]
@validator("parameters", "moderations", pre=True, always=True)
@classmethod
def validate_data_models(cls, value, values, config, field):
return to_model_optional(value, Model=field.type_, copy=False)
[docs]
@classmethod
def is_lc_serializable(cls) -> bool:
return True
@property
def lc_secrets(self) -> dict[str, str]:
return {"client": "CLIENT"}
[docs]
@classmethod
def load_from_file(cls, file: Union[str, Path], *, client: Client):
config = load_config(file)
return cls(**config, client=client)
@property
def _identifying_params(self) -> dict[str, Any]:
return {
"model_id": self.model_id,
"prompt_id": self.prompt_id,
"parameters": dump_optional_model(self.parameters),
"moderations": dump_optional_model(self.moderations),
"parent_id": self.parent_id,
"prompt_template_id": self.prompt_template_id,
"trim_method": self.trim_method,
"use_conversation_parameters": self.use_conversation_parameters,
**super()._identifying_params,
}
@property
def _llm_type(self) -> str:
return "ibmgenai_chat_llm"
def _prepare_request(self, **kwargs):
updated = {k: kwargs.pop(k, v) for k, v in self._identifying_params.items()}
return _prepare_chat_generation_request(**kwargs, **updated)
def _stream(
self,
messages: Messages,
stop: Optional[list[str]] = None,
run_manager: Optional[CallbackManagerForLLMRun] = None,
**kwargs: Any,
) -> Iterator[CustomChatGenerationChunk]:
for response in self.client.text.chat.create_stream(
**self._prepare_request(messages=_convert_messages_to_genai(messages), stop=stop, **kwargs)
):
if not response:
continue
def send_chunk(*, text: str = "", generation_info: dict):
logger.info("Chunk received: {}".format(text))
chunk = CustomChatGenerationChunk(
message=CustomAIMessageChunk(content=text, generation_info=generation_info),
generation_info=generation_info,
)
yield chunk
if run_manager:
run_manager.on_llm_new_token(token=text, chunk=chunk, response=response) # noqa: B023
# Function definition does not bind loop variable `response`: linter is probably just confused here
if response.moderations:
generation_info = create_generation_info_from_response(response, result=response.moderations)
yield from send_chunk(generation_info=generation_info)
for result in response.results or []:
generation_info = create_generation_info_from_response(response, result=result)
yield from send_chunk(text=result.generated_text or "", generation_info=generation_info)
def _generate(
self,
messages: Messages,
stop: Optional[list[str]] = None,
run_manager: Optional[CallbackManagerForLLMRun] = None,
**kwargs: Any,
) -> ChatResult:
def handle_stream():
final_generation: Optional[CustomChatGenerationChunk] = None
for result in self._stream(
messages=messages,
stop=stop,
run_manager=run_manager,
**kwargs,
):
if final_generation:
token_usage = result.generation_info.pop("token_usage")
final_generation += result
update_token_usage_stream(
target=final_generation.generation_info["token_usage"],
source=token_usage,
)
else:
final_generation = result
assert final_generation and final_generation.generation_info
return {
"text": final_generation.text,
"generation_info": final_generation.generation_info.copy(),
}
def handle_non_stream():
response = self.client.text.chat.create(
**self._prepare_request(messages=_convert_messages_to_genai(messages), stop=stop, **kwargs),
)
assert response.results
result = response.results[0]
return {
"text": result.generated_text or "",
"generation_info": create_generation_info_from_response(response, result=result),
}
result = handle_stream() if self.streaming else handle_non_stream()
return ChatResult(
generations=[
ChatGeneration(
message=LCAIMessage(content=result["text"]),
generation_info=result["generation_info"].copy(),
)
],
llm_output=create_llm_output(
model=result["generation_info"].get("model_id", self.model_id or ""),
token_usages=[result["generation_info"]["token_usage"]],
),
)
[docs]
def get_num_tokens(self, text: str) -> int:
response = list(self.client.text.tokenization.create(model_id=self.model_id, input=[text]))[0]
return response.results[0].token_count
[docs]
def get_num_tokens_from_messages(self, messages: list[LCBaseMessage]) -> int:
return sum(
sum(result.token_count for result in response.results)
for response in self.client.text.tokenization.create(
model_id=self.model_id, input=[get_buffer_string([message]) for message in messages]
)
)
def _combine_llm_outputs(self, llm_outputs: list[Optional[dict]]) -> dict:
token_usages: list[Optional[dict]] = []
model = ""
for output in llm_outputs:
if output:
model = model or output.get("meta", {}).get("model_id")
token_usages.append(output.get("token_usage"))
return create_llm_output(model=model or self.model_id, token_usages=token_usages)
[docs]
def get_token_ids(self, text: str) -> list[int]:
raise NotImplementedError("API does not support returning token ids.")