# Copyright (c) "Neo4j"
# Neo4j Sweden AB [https://neo4j.com]
# #
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
# #
# https://www.apache.org/licenses/LICENSE-2.0
# #
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from __future__ import annotations
import os
from typing import Any, Iterable, List, Optional, Type, Union, cast, overload
from pydantic import BaseModel, ValidationError
from neo4j_graphrag.exceptions import LLMGenerationError
from neo4j_graphrag.llm.base import LLMInterface, LLMInterfaceV2
from neo4j_graphrag.llm.types import (
BaseMessage,
LLMResponse,
MessageList,
SystemMessage,
UserMessage,
)
from neo4j_graphrag.message_history import MessageHistory
from neo4j_graphrag.types import LLMMessage
from neo4j_graphrag.utils.rate_limit import (
RateLimitHandler,
)
from neo4j_graphrag.utils.rate_limit import (
async_rate_limit_handler as async_rate_limit_handler_decorator,
)
from neo4j_graphrag.utils.rate_limit import (
rate_limit_handler as rate_limit_handler_decorator,
)
try:
from mistralai import (
AssistantMessage,
Messages,
Mistral,
)
from mistralai import (
SystemMessage as MistralSystemMessage,
)
from mistralai import (
UserMessage as MistralUserMessage,
)
from mistralai.models.sdkerror import SDKError
except ImportError:
Mistral = None # type: ignore
SDKError = None # type: ignore
# pylint: disable=redefined-builtin, arguments-differ, raise-missing-from, no-else-return
[docs]
class MistralAILLM(LLMInterface, LLMInterfaceV2):
def __init__(
self,
model_name: str,
model_params: Optional[dict[str, Any]] = None,
rate_limit_handler: Optional[RateLimitHandler] = None,
**kwargs: Any,
):
"""
Args:
model_name (str):
model_params (str): Parameters for LLMInterface(V1) like temperature and such that will be
passed to the chat completions endpoint
rate_limit_handler (Optional[RateLimitHandler]): Handler for rate limiting for LLMInterface(V1). Defaults to retry with exponential backoff.
kwargs: All other parameters will be passed to the Mistral client.
"""
if Mistral is None:
raise ImportError(
"""Could not import Mistral Python client.
Please install it with `pip install "neo4j-graphrag[mistralai]"`."""
)
LLMInterfaceV2.__init__(
self,
model_name=model_name,
model_params=model_params or {},
rate_limit_handler=rate_limit_handler,
**kwargs,
)
api_key = kwargs.pop("api_key", None)
if api_key is None:
api_key = os.getenv("MISTRAL_API_KEY", "")
self.client = Mistral(api_key=api_key, **kwargs)
# overloads for LLMInterface and LLMInterfaceV2 methods
@overload # type: ignore[no-overload-impl]
def invoke(
self,
input: str,
message_history: Optional[Union[List[LLMMessage], MessageHistory]] = None,
system_instruction: Optional[str] = None,
) -> LLMResponse: ...
@overload
def invoke(
self,
input: List[LLMMessage],
response_format: Optional[Union[Type[BaseModel], dict[str, Any]]] = None,
**kwargs: Any,
) -> LLMResponse: ...
@overload # type: ignore[no-overload-impl]
async def ainvoke(
self,
input: str,
message_history: Optional[Union[List[LLMMessage], MessageHistory]] = None,
system_instruction: Optional[str] = None,
) -> LLMResponse: ...
@overload
async def ainvoke(
self,
input: List[LLMMessage],
response_format: Optional[Union[Type[BaseModel], dict[str, Any]]] = None,
**kwargs: Any,
) -> LLMResponse: ...
# switching logics to LLMInterface or LLMInterfaceV2
[docs]
def invoke( # type: ignore[no-redef]
self,
input: Union[str, List[LLMMessage]],
message_history: Optional[Union[List[LLMMessage], MessageHistory]] = None,
system_instruction: Optional[str] = None,
response_format: Optional[Union[Type[BaseModel], dict[str, Any]]] = None,
**kwargs: Any,
) -> LLMResponse:
if isinstance(input, str):
return self.__invoke_v1(input, message_history, system_instruction)
elif isinstance(input, list):
return self.__invoke_v2(input, response_format=response_format, **kwargs)
else:
raise ValueError(f"Invalid input type for invoke method - {type(input)}")
[docs]
async def ainvoke( # type: ignore[no-redef]
self,
input: Union[str, List[LLMMessage]],
message_history: Optional[Union[List[LLMMessage], MessageHistory]] = None,
system_instruction: Optional[str] = None,
response_format: Optional[Union[Type[BaseModel], dict[str, Any]]] = None,
**kwargs: Any,
) -> LLMResponse:
if isinstance(input, str):
return await self.__ainvoke_v1(input, message_history, system_instruction)
elif isinstance(input, list):
return await self.__ainvoke_v2(
input, response_format=response_format, **kwargs
)
else:
raise ValueError(f"Invalid input type for ainvoke method - {type(input)}")
# implementations
@rate_limit_handler_decorator
def __invoke_v1(
self,
input: str,
message_history: Optional[Union[List[LLMMessage], MessageHistory]] = None,
system_instruction: Optional[str] = None,
) -> LLMResponse:
"""Sends a text input to the Mistral chat completion model
and returns the response's content.
Args:
input (str): Text sent to the LLM.
message_history (Optional[Union[List[LLMMessage], MessageHistory]]): A collection previous messages, with each message having a specific role assigned.
system_instruction (Optional[str]): An option to override the llm system message for this invocation.
Returns:
LLMResponse: The response from MistralAI.
Raises:
LLMGenerationError: If anything goes wrong.
"""
try:
if isinstance(message_history, MessageHistory):
message_history = message_history.messages
messages = self.get_messages(input, message_history, system_instruction)
response = self.client.chat.complete(
model=self.model_name,
messages=messages,
**self.model_params,
)
content: str = ""
if response and response.choices:
possible_content = response.choices[0].message.content
if isinstance(possible_content, str):
content = possible_content
return LLMResponse(content=content)
except SDKError as e:
raise LLMGenerationError(e)
def __invoke_v2(
self,
input: List[LLMMessage],
response_format: Optional[Union[Type[BaseModel], dict[str, Any]]] = None,
**kwargs: Any,
) -> LLMResponse:
"""Sends a text input to the Mistral chat completion model
and returns the response's content.
Args:
input (List[LLMMessage]): Messages sent to the LLM.
response_format: Not supported by MistralAILLM.
Returns:
LLMResponse: The response from MistralAI.
Raises:
LLMGenerationError: If anything goes wrong.
"""
if response_format is not None:
raise NotImplementedError(
"MistralAILLM does not currently support structured output"
)
try:
messages = self.get_messages_v2(input)
response = self.client.chat.complete(
model=self.model_name, messages=messages, **self.model_params, **kwargs
)
content: str = ""
if response and response.choices:
possible_content = response.choices[0].message.content
if isinstance(possible_content, str):
content = possible_content
return LLMResponse(content=content)
except SDKError as e:
raise LLMGenerationError(e)
@async_rate_limit_handler_decorator
async def __ainvoke_v1(
self,
input: str,
message_history: Optional[Union[List[LLMMessage], MessageHistory]] = None,
system_instruction: Optional[str] = None,
) -> LLMResponse:
"""Asynchronously sends a text input to the MistralAI chat
completion model and returns the response's content.
Args:
input (str): Text sent to the LLM.
message_history (Optional[Union[List[LLMMessage], MessageHistory]]): A collection previous messages,
with each message having a specific role assigned.
system_instruction (Optional[str]): An option to override the llm system message for this invocation.
Returns:
LLMResponse: The response from MistralAI.
Raises:
LLMGenerationError: If anything goes wrong.
"""
try:
if isinstance(message_history, MessageHistory):
message_history = message_history.messages
messages = self.get_messages(input, message_history, system_instruction)
response = await self.client.chat.complete_async(
model=self.model_name,
messages=messages,
**self.model_params,
)
content: str = ""
if response and response.choices:
possible_content = response.choices[0].message.content
if isinstance(possible_content, str):
content = possible_content
return LLMResponse(content=content)
except SDKError as e:
raise LLMGenerationError(e)
async def __ainvoke_v2(
self,
input: List[LLMMessage],
response_format: Optional[Union[Type[BaseModel], dict[str, Any]]] = None,
**kwargs: Any,
) -> LLMResponse:
"""Asynchronously sends a text input to the MistralAI chat
completion model and returns the response's content.
Args:
input (List[LLMMessage]): Messages sent to the LLM.
response_format: Not supported by MistralAILLM.
Returns:
LLMResponse: The response from MistralAI.
Raises:
LLMGenerationError: If anything goes wrong.
"""
if response_format is not None:
raise NotImplementedError(
"MistralAILLM does not currently support structured output"
)
try:
messages = self.get_messages_v2(input)
response = await self.client.chat.complete_async(
model=self.model_name,
messages=messages,
**self.model_params,
**kwargs,
)
content: str = ""
if response and response.choices:
possible_content = response.choices[0].message.content
if isinstance(possible_content, str):
content = possible_content
return LLMResponse(content=content)
except SDKError as e:
raise LLMGenerationError(e)
# subsidiary methods
[docs]
def get_messages(
self,
input: str,
message_history: Optional[Union[List[LLMMessage], MessageHistory]] = None,
system_instruction: Optional[str] = None,
) -> list[Messages]:
"""Constructs the message list for the Mistral chat completion model."""
messages = []
if system_instruction:
messages.append(SystemMessage(content=system_instruction).model_dump())
if message_history:
if isinstance(message_history, MessageHistory):
message_history = message_history.messages
try:
MessageList(messages=cast(list[BaseMessage], message_history))
except ValidationError as e:
raise LLMGenerationError(e.errors()) from e
messages.extend(cast(Iterable[dict[str, Any]], message_history))
messages.append(UserMessage(content=input).model_dump())
return cast(list[Messages], messages)
[docs]
def get_messages_v2(
self,
input: list[LLMMessage],
) -> list[Messages]:
"""Constructs the message list for the Mistral chat completion model."""
messages: list[Messages] = []
for m in input:
if m["role"] == "system":
messages.append(MistralSystemMessage(content=m["content"]))
continue
if m["role"] == "user":
messages.append(MistralUserMessage(content=m["content"]))
continue
if m["role"] == "assistant":
messages.append(AssistantMessage(content=m["content"]))
continue
raise ValueError(f"Unknown role: {m['role']}")
return messages