litellm/litellm/llms/sagemaker/chat/handler.py

import json
from collections.abc import Callable
from copy import deepcopy
from typing import Final

import httpx

from litellm.litellm_core_utils.aws_partition import get_aws_dns_suffix
from litellm.llms.bedrock.base_aws_llm import BaseAWSLLM, pop_aws_auth_params
from litellm.llms.custom_httpx.http_handler import AsyncHTTPHandler, HTTPHandler
from litellm.utils import ModelResponse, get_secret

from ..common_utils import AWSEventStreamDecoder
from .transformation import SagemakerChatConfig


class SagemakerChatHandler(BaseAWSLLM):
    def _load_credentials(
        self,
        optional_params: dict,
    ):
        try:
            from botocore.credentials import Credentials
        except ImportError:
            raise ImportError("Missing boto3 to call bedrock. Run 'pip install boto3'.")
        auth_params: Final = pop_aws_auth_params(optional_params)
        aws_region_name = optional_params.pop("aws_region_name", None)
        optional_params.pop("aws_bedrock_runtime_endpoint", None)

        ### SET REGION NAME ###
        if aws_region_name is None:
            # check env #
            litellm_aws_region_name: Final = get_secret("AWS_REGION_NAME", None)

            if litellm_aws_region_name is not None and isinstance(litellm_aws_region_name, str):
                aws_region_name = litellm_aws_region_name

            standard_aws_region_name: Final = get_secret("AWS_REGION", None)
            if standard_aws_region_name is not None and isinstance(standard_aws_region_name, str):
                aws_region_name = standard_aws_region_name

            if aws_region_name is None:
                aws_region_name = "us-west-2"

        credentials: Final[Credentials] = self.resolve_credentials(auth_params, aws_region_name)
        return credentials, aws_region_name

    def _prepare_request(
        self,
        credentials,
        model: str,
        data: dict,
        optional_params: dict,
        aws_region_name: str,
        extra_headers: dict | None = None,
    ):
        try:
            from botocore.auth import SigV4Auth
            from botocore.awsrequest import AWSRequest
        except ImportError:
            raise ImportError("Missing boto3 to call bedrock. Run 'pip install boto3'.")

        sigv4: Final = SigV4Auth(credentials, "sagemaker", aws_region_name)
        dns_suffix: Final = get_aws_dns_suffix(aws_region_name)
        if optional_params.get("stream") is True:
            api_base = f"https://runtime.sagemaker.{aws_region_name}.{dns_suffix}/endpoints/{model}/invocations-response-stream"
        else:
            api_base = f"https://runtime.sagemaker.{aws_region_name}.{dns_suffix}/endpoints/{model}/invocations"

        sagemaker_base_url: Final = optional_params.get("sagemaker_base_url", None)
        if sagemaker_base_url is not None:
            api_base = sagemaker_base_url

        encoded_data: Final = json.dumps(data).encode("utf-8")
        headers = {"Content-Type": "application/json"}
        if extra_headers is not None:
            headers = {"Content-Type": "application/json", **extra_headers}
        request: Final = AWSRequest(method="POST", url=api_base, data=encoded_data, headers=headers)
        sigv4.add_auth(request)
        if (
            extra_headers is not None and "Authorization" in extra_headers
        ):  # prevent sigv4 from overwriting the auth header
            request.headers["Authorization"] = extra_headers["Authorization"]

        prepped_request: Final = request.prepare()

        return prepped_request

    def completion(
        self,
        model: str,
        messages: list,
        model_response: ModelResponse,
        print_verbose: Callable,
        encoding,
        logging_obj,
        optional_params: dict,
        litellm_params: dict,
        timeout: float | httpx.Timeout | None = None,
        custom_prompt_dict={},
        logger_fn=None,
        acompletion: bool = False,
        headers: dict = {},
        client: HTTPHandler | AsyncHTTPHandler | None = None,
    ):
        # pop streaming if it's in the optional params as 'stream' raises an error with sagemaker
        credentials, aws_region_name = self._load_credentials(optional_params)
        inference_params: Final = deepcopy(optional_params)
        stream: Final = inference_params.pop("stream", None)

        from litellm.llms.openai_like.chat.handler import OpenAILikeChatHandler

        openai_like_chat_completions: Final = OpenAILikeChatHandler()
        inference_params["stream"] = True if stream is True else False
        _data: Final = SagemakerChatConfig().transform_request(
            model=model,
            messages=messages,
            optional_params=inference_params,
            litellm_params=litellm_params,
            headers=headers,
        )

        prepared_request: Final = self._prepare_request(
            model=model,
            data=_data,
            optional_params=optional_params,
            credentials=credentials,
            aws_region_name=aws_region_name,
        )

        custom_stream_decoder: Final = AWSEventStreamDecoder(model="", is_messages_api=True)

        return openai_like_chat_completions.completion(
            model=model,
            messages=messages,
            api_base=prepared_request.url,
            api_key=None,
            custom_prompt_dict=custom_prompt_dict,
            model_response=model_response,
            print_verbose=print_verbose,
            logging_obj=logging_obj,
            optional_params=inference_params,
            acompletion=acompletion,
            litellm_params=litellm_params,
            logger_fn=logger_fn,
            timeout=timeout,
            encoding=encoding,
            headers=prepared_request.headers,
            custom_endpoint=True,
            custom_llm_provider="sagemaker_chat",
            streaming_decoder=custom_stream_decoder,
            client=client,
        )