phoenix/litellm-mirror
0
0
Fork 1
mirror of https://github.com/BerriAI/litellm.git synced 2025-04-26 03:04:13 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

fix(proxy/utils.py): add stronger typing for litellm params in failure call logging

Browse source
This commit is contained in:
Krrish Dholakia 2024-07-22 21:31:21 -07:00
parent 34575293a5
commit 1a83935aa4
3 changed files with 30 additions and 4 deletions
Download patch file Download diff file Expand all files Collapse all files

14
litellm/proxy/utils.py
View file

@ -25,7 +25,7 @@ from typing_extensions import overload
import litellm import litellm
import litellm.litellm_core_utils import litellm.litellm_core_utils
import litellm.litellm_core_utils.litellm_logging import litellm.litellm_core_utils.litellm_logging
from litellm import EmbeddingResponse, ImageResponse, ModelResponse from litellm import EmbeddingResponse, ImageResponse, ModelResponse, get_litellm_params
from litellm._logging import verbose_proxy_logger from litellm._logging import verbose_proxy_logger
from litellm._service_logger import ServiceLogging, ServiceTypes from litellm._service_logger import ServiceLogging, ServiceTypes
from litellm.caching import DualCache, RedisCache from litellm.caching import DualCache, RedisCache
@ -50,7 +50,7 @@ from litellm.proxy.hooks.max_budget_limiter import _PROXY_MaxBudgetLimiter
from litellm.proxy.hooks.parallel_request_limiter import ( from litellm.proxy.hooks.parallel_request_limiter import (
_PROXY_MaxParallelRequestsHandler, _PROXY_MaxParallelRequestsHandler,
) )
from litellm.types.utils import CallTypes from litellm.types.utils import CallTypes, LoggedLiteLLMParams
if TYPE_CHECKING: if TYPE_CHECKING:
from opentelemetry.trace import Span as _Span from opentelemetry.trace import Span as _Span
@ -602,14 +602,20 @@ class ProxyLogging:
if litellm_logging_obj is not None: if litellm_logging_obj is not None:
## UPDATE LOGGING INPUT ## UPDATE LOGGING INPUT
_optional_params = {} _optional_params = {}
_litellm_params = {}
litellm_param_keys = LoggedLiteLLMParams.__annotations__.keys()
for k, v in request_data.items(): for k, v in request_data.items():
if k != "model" and k != "user" and k != "litellm_params": if k in litellm_param_keys:
_litellm_params[k] = v
elif k != "model" and k != "user":
_optional_params[k] = v _optional_params[k] = v
litellm_logging_obj.update_environment_variables( litellm_logging_obj.update_environment_variables(
model=request_data.get("model", ""), model=request_data.get("model", ""),
user=request_data.get("user", ""), user=request_data.get("user", ""),
optional_params=_optional_params, optional_params=_optional_params,
litellm_params=request_data.get("litellm_params", {}), litellm_params=_litellm_params,
) )
input: Union[list, str, dict] = "" input: Union[list, str, dict] = ""

1
litellm/tests/test_custom_callback_input.py
View file

@ -234,6 +234,7 @@ class CompletionCustomHandler(
) )
assert isinstance(kwargs["optional_params"], dict) assert isinstance(kwargs["optional_params"], dict)
assert isinstance(kwargs["litellm_params"], dict) assert isinstance(kwargs["litellm_params"], dict)
assert isinstance(kwargs["litellm_params"]["metadata"], Optional[dict])
assert isinstance(kwargs["start_time"], (datetime, type(None))) assert isinstance(kwargs["start_time"], (datetime, type(None)))
assert isinstance(kwargs["stream"], bool) assert isinstance(kwargs["stream"], bool)
assert isinstance(kwargs["user"], (str, type(None))) assert isinstance(kwargs["user"], (str, type(None)))

19
litellm/types/utils.py
View file

@ -1029,3 +1029,22 @@ class GenericImageParsingChunk(TypedDict):
class ResponseFormatChunk(TypedDict, total=False): class ResponseFormatChunk(TypedDict, total=False):
type: Required[Literal["json_object", "text"]] type: Required[Literal["json_object", "text"]]
response_schema: dict response_schema: dict
class LoggedLiteLLMParams(TypedDict, total=False):
force_timeout: Optional[float]
custom_llm_provider: Optional[str]
api_base: Optional[str]
litellm_call_id: Optional[str]
model_alias_map: Optional[dict]
metadata: Optional[dict]
model_info: Optional[dict]
proxy_server_request: Optional[dict]
acompletion: Optional[bool]
preset_cache_key: Optional[str]
no_log: Optional[bool]
input_cost_per_second: Optional[float]
input_cost_per_token: Optional[float]
output_cost_per_token: Optional[float]
output_cost_per_second: Optional[float]
cooldown_time: Optional[float]