Merge pull request #41048 from HUAHAODIA/litellm_ratchet_strict_rules

chore(lint): graduate 12 rules from the strict-gate ratchet

litellm/integrations/arize/arize_phoenix_prompt_manager.py

@@ -379,7 +379,7 @@ class ArizePhoenixPromptManager(CustomPromptManagement):
         """Reload prompts from Arize Phoenix."""
         if self.prompt_id:
             self._prompt_manager = None  # Reset to force reload
-            self.prompt_manager  # This will trigger reload
+            _ = self.prompt_manager  # access triggers lazy reload
 
     def should_run_prompt_management(
         self,

litellm/integrations/bitbucket/bitbucket_prompt_manager.py

@@ -406,7 +406,7 @@ class BitBucketPromptManager(CustomPromptManagement):
         """Reload prompts from BitBucket."""
         if self.prompt_id:
             self._prompt_manager = None  # Reset to force reload
-            self.prompt_manager  # This will trigger reload
+            _ = self.prompt_manager  # access triggers lazy reload
 
     def should_run_prompt_management(
         self,

litellm/integrations/weights_biases.py

@@ -4,18 +4,10 @@ imported_openAIResponse = True
 try:
     import io
     import logging
-    import sys
-    from typing import Any, TypeVar
+    from typing import Any, Literal, Protocol, TypeVar
 
     from wandb.sdk.data_types import trace_tree
 
-    if sys.version_info >= (3, 8):
-        from typing import Literal, Protocol
-    else:
-        from typing import Literal
-
-        from typing_extensions import Protocol
-
     logger: Final = logging.getLogger(__name__)
 
     K = TypeVar("K", bound=str)

litellm/litellm_core_utils/coroutine_checker.py

@@ -36,7 +36,7 @@ class CoroutineChecker:
         target = callback
         if not inspect.isfunction(target) and not inspect.ismethod(target):
             try:
-                call_attr: Final = getattr(target, "__call__", None)
+                call_attr: Final = getattr(target, "__call__", None)  # noqa: B004  # value unwrap so iscoroutinefunction sees through functors
                 if call_attr is not None:
                     target = call_attr
             except Exception:

litellm/litellm_core_utils/prompt_templates/factory.py

@@ -1757,7 +1757,7 @@ def convert_to_anthropic_tool_invoke(
     anthropic_tool_invoke: Final[list[AnthropicMessagesToolUseParam | dict[str, object]]] = []
 
     for tool in tool_calls:
-        if not get_attribute_or_key(tool, "type") == "function":
+        if get_attribute_or_key(tool, "type") != "function":
             continue
 
         tool_id = cast(str, get_attribute_or_key(tool, "id"))

litellm/litellm_core_utils/token_counter.py

@@ -454,7 +454,7 @@ def token_counter(
         params: Final = _MessageCountParams(model, custom_tokenizer)
         num_tokens = _count_messages(params, new_messages, use_default_image_token_count, default_token_count)
         if count_response_tokens is False:
-            includes_system_message: Final = any([message.get("role", None) == "system" for message in new_messages])
+            includes_system_message: Final = any(message.get("role", None) == "system" for message in new_messages)
             num_tokens += _count_extra(params.count_function, tools, tool_choice, includes_system_message)
 
     else:

litellm/llms/azure_ai/common_utils.py

@@ -144,10 +144,13 @@ class AzureFoundryModelInfo(BaseLLMModelInfo):
     def get_api_key(api_key: str | None = None) -> str | None:
         return api_key or litellm.api_key or get_secret_str("AZURE_AI_API_KEY")
 
+    @staticmethod
+    def get_api_version(api_version: str | None = None) -> str | None:
+        return api_version or litellm.api_version or get_secret_str("AZURE_API_VERSION")
+
     @property
-    def api_version(self, api_version: str | None = None) -> str | None:
-        api_version = api_version or litellm.api_version or get_secret_str("AZURE_API_VERSION")
-        return api_version
+    def api_version(self) -> str | None:
+        return AzureFoundryModelInfo.get_api_version()
 
     def get_token_counter(self) -> BaseTokenCounter | None:
         """

litellm/proxy/guardrails/guardrail_hooks/javelin/javelin.py

@@ -44,7 +44,7 @@ class JavelinGuardrail(CustomGuardrail):
         application: str | None = None,
         **kwargs,
     ):
-        f"""
+        """
         Initialize the JavelinGuardrail class.
 
         This calls: {api_base}/{api_version}/guardrail/{guardrail_name}/apply

litellm/proxy/guardrails/guardrail_hooks/mcp_end_user_permission/__init__.py

@@ -15,7 +15,7 @@ def initialize_guardrail(litellm_params: "LitellmParams", guardrail: "Guardrail"
     # We check the raw guardrail dict because LitellmParams normalizes None → False,
     # making it impossible to distinguish "not set" from "explicitly false" via litellm_params.
     _raw_default_on: Final = cast(dict[str, Any], guardrail).get("litellm_params", {}).get("default_on")
-    _default_on: Final = False if _raw_default_on is False else True
+    _default_on: Final = _raw_default_on is not False
 
     _callback: Final = MCPEndUserPermissionGuardrail(
         guardrail_name=guardrail.get("guardrail_name", ""),

litellm/proxy/management_helpers/team_metadata_validation.py

@@ -115,15 +115,15 @@ async def run_team_metadata_validation(
                 "error": f"custom_team_metadata_validate is an Enterprise feature. {CommonProxyErrors.not_premium_user.value}"
             },
         )
-    if not (
-        inspect.iscoroutinefunction(validator) or inspect.iscoroutinefunction(getattr(validator, "__call__", None))
-    ):
-        raise HTTPException(
-            status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
-            detail={  # mutable-ok: HTTPException.detail has no immutable form
-                "error": "custom_team_metadata_validate must be an async function"
-            },
-        )
+    if not inspect.iscoroutinefunction(validator):
+        validator_call: Final = getattr(validator, "__call__", None)  # noqa: B004  # value unwrap for the functor check
+        if not inspect.iscoroutinefunction(validator_call):
+            raise HTTPException(
+                status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+                detail={  # mutable-ok: HTTPException.detail has no immutable form
+                    "error": "custom_team_metadata_validate must be an async function"
+                },
+            )
 
     try:
         raw_result: Final = await asyncio.wait_for(validator(payload), timeout=timeout_seconds)

litellm/proxy/spend_tracking/spend_tracking_utils.py

@@ -159,7 +159,7 @@ def _get_spend_logs_metadata(
             requester_ip_address=None,
             additional_usage_values=None,
             applied_guardrails=None,
-            status=None or "success",
+            status="success",
             error_information=None,
             proxy_server_request=None,
             batch_models=None,

litellm/proxy/utils.py

@@ -4,6 +4,7 @@ import copy
 import hashlib
 import inspect
 import json
+import math
 import os
 import smtplib
 import ssl
@@ -6404,7 +6405,7 @@ class PrismaClient:
             return None
         try:
             value: Final = float(response_time_ms)
-            return value if value == value and value not in (float("inf"), float("-inf")) else None
+            return value if math.isfinite(value) else None
         except (ValueError, TypeError):
             verbose_proxy_logger.warning("Invalid response_time_ms value: %s", response_time_ms)
             return None

litellm/router.py

@@ -4105,16 +4105,16 @@ class Router:
         models: Final = [m.strip() for m in model.split(",")]
 
         async def _async_completion_no_exceptions(
-            model: str, messages: list[dict[str, str]], stream: bool, **kwargs: Any
+            model_name: str, messages: list[dict[str, str]], stream: bool, **kwargs: Any
         ) -> ModelResponse | CustomStreamWrapper | Exception:
             """
             Wrapper around self.acompletion that catches exceptions and returns them as a result
             """
             try:
-                result = await self.acompletion(model=model, messages=messages, stream=stream, **kwargs)
+                result = await self.acompletion(model=model_name, messages=messages, stream=stream, **kwargs)
                 return result
             except asyncio.CancelledError:
-                verbose_router_logger.debug("Received 'task.cancel'. Cancelling call w/ model=%s.", model)
+                verbose_router_logger.debug("Received 'task.cancel'. Cancelling call w/ model=%s.", model_name)
                 raise
             except Exception as e:
                 return e
@@ -4141,9 +4141,9 @@ class Router:
                 except KeyError:
                     pass
 
-        for model in models:
+        for model_name in models:
             task = asyncio.create_task(
-                _async_completion_no_exceptions(model=model, messages=messages, stream=stream, **kwargs)
+                _async_completion_no_exceptions(model_name=model_name, messages=messages, stream=stream, **kwargs)
             )
             pending_tasks.append(task)
 

litellm/utils.py

@@ -6260,7 +6260,7 @@ def function_to_dict(input_function) -> dict:
             "enum": param_enum,
         }
 
-        parameters[param_name] = dict([(k, v) for k, v in param_dict.items() if isinstance(v, str)])
+        parameters[param_name] = {k: v for k, v in param_dict.items() if isinstance(v, str)}
 
         # Check if the parameter has no default value (i.e., it's required)
         if param.default == param.empty:

ruff.toml

@@ -4,11 +4,12 @@ lint.ignore = ["F405", "E402", "F403"]
 # That gives editors and `ruff check --fix` the diagnostic, which the gate script cannot.
 lint.extend-select = [
     "T20", "PGH004", "RUF008", "RUF009", "RUF100",
-    "B033", "FURB136", "FURB168", "FURB188", "I001", "PERF402", "PIE790", "PIE800", "PLC0208",
-    "PLR0402", "PLR1711", "PLR1730", "PLR2044", "PLW0133", "PYI030", "PYI041", "PYI064", "RET501",
-    "RUF010", "RUF022", "RUF023", "RUF051", "S113", "SIM114", "SIM118", "TC005", "UP006", "UP007",
-    "UP008",
-    "UP012", "UP018", "UP024", "UP032", "UP034", "UP035", "UP037", "UP045",
+    "B004", "B018", "B021", "B033", "FURB136", "FURB168", "FURB188", "I001", "PERF402", "PIE790",
+    "PIE800", "PLC0208", "PLR0124", "PLR0402", "PLR0206", "PLR1704", "PLR1711", "PLR1730", "PLR2044",
+    "PLW0133", "PYI030", "PYI041", "PYI064", "RET501", "RUF010", "RUF022", "RUF023", "RUF051", "S113",
+    "SIM114", "SIM118", "SIM201", "SIM211", "SIM222", "TC005", "UP006", "UP007", "UP008",
+    "UP012", "UP018", "UP024", "UP032", "UP034", "UP035", "UP036", "UP037", "UP045",
+    "C404", "C419",
 ]
 # RUF100 (unused-noqa) only knows the rules enabled in THIS config, so it would strip
 # `# noqa` directives that protect rules enforced elsewhere. List those codes as external