Merge branch 'litellm_internal_staging' into litellm_remove_types_ruff_exclusion

Import-block conflicts against #35962's unused-import removals resolve
to this branch's side: its modernization already removed every typing
import staging trimmed, plus the usages

litellm/__init__.py

@@ -2150,9 +2150,9 @@ def __getattr__(name: str) -> Any:
 
     # Lazy load encoding from main.py to avoid heavy tiktoken import
     if name == "encoding":
-        from ._lazy_imports import _get_litellm_globals
+        from ._lazy_imports import get_litellm_globals
 
-        _globals = _get_litellm_globals()
+        _globals = get_litellm_globals()
         # Check if already cached
         if "encoding" not in _globals:
             from .main import encoding as _encoding
@@ -2162,9 +2162,9 @@ def __getattr__(name: str) -> Any:
 
     # Lazy load bedrock_tool_name_mappings instance
     if name == "bedrock_tool_name_mappings":
-        from ._lazy_imports import _get_litellm_globals
+        from ._lazy_imports import get_litellm_globals
 
-        _globals = _get_litellm_globals()
+        _globals = get_litellm_globals()
         # Check if already cached
         if "bedrock_tool_name_mappings" not in _globals:
             from .llms.bedrock.chat.invoke_handler import (
@@ -2176,9 +2176,9 @@ def __getattr__(name: str) -> Any:
 
     # Lazy load AzureOpenAIError exception class
     if name == "AzureOpenAIError":
-        from ._lazy_imports import _get_litellm_globals
+        from ._lazy_imports import get_litellm_globals
 
-        _globals = _get_litellm_globals()
+        _globals = get_litellm_globals()
         # Check if already cached
         if "AzureOpenAIError" not in _globals:
             from .llms.azure.common_utils import AzureOpenAIError as _AzureOpenAIError
@@ -2188,9 +2188,9 @@ def __getattr__(name: str) -> Any:
 
     # Lazy load openaiOSeriesConfig instance
     if name == "openaiOSeriesConfig":
-        from ._lazy_imports import _get_litellm_globals
+        from ._lazy_imports import get_litellm_globals
 
-        _globals = _get_litellm_globals()
+        _globals = get_litellm_globals()
         if "openaiOSeriesConfig" not in _globals:
             # Import the config class and instantiate it
             config_class = __getattr__("OpenAIOSeriesConfig")
@@ -2206,9 +2206,9 @@ def __getattr__(name: str) -> Any:
         "nvidiaNimEmbeddingConfig": "NvidiaNimEmbeddingConfig",
     }
     if name in _config_instances:
-        from ._lazy_imports import _get_litellm_globals
+        from ._lazy_imports import get_litellm_globals
 
-        _globals = _get_litellm_globals()
+        _globals = get_litellm_globals()
         if name not in _globals:
             # Import the config class and instantiate it
             config_class = __getattr__(_config_instances[name])
@@ -2221,9 +2221,9 @@ def __getattr__(name: str) -> Any:
 
     # Lazy load provider_list
     if name == "provider_list":
-        from ._lazy_imports import _get_litellm_globals
+        from ._lazy_imports import get_litellm_globals
 
-        _globals = _get_litellm_globals()
+        _globals = get_litellm_globals()
         # Check if already cached
         if "provider_list" not in _globals:
             # LlmProviders is eagerly imported above, so we can import it directly
@@ -2234,9 +2234,9 @@ def __getattr__(name: str) -> Any:
 
     # Lazy load priority_reservation_settings instance
     if name == "priority_reservation_settings":
-        from ._lazy_imports import _get_litellm_globals
+        from ._lazy_imports import get_litellm_globals
 
-        _globals = _get_litellm_globals()
+        _globals = get_litellm_globals()
         # Check if already cached
         if "priority_reservation_settings" not in _globals:
             # Import the class and instantiate it
@@ -2246,9 +2246,9 @@ def __getattr__(name: str) -> Any:
 
     # Lazy load logging_callback_manager instance
     if name == "logging_callback_manager":
-        from ._lazy_imports import _get_litellm_globals
+        from ._lazy_imports import get_litellm_globals
 
-        _globals = _get_litellm_globals()
+        _globals = get_litellm_globals()
         # Check if already cached
         if "logging_callback_manager" not in _globals:
             # Import the class and instantiate it
@@ -2258,9 +2258,9 @@ def __getattr__(name: str) -> Any:
 
     # Lazy load _service_logger module
     if name == "_service_logger":
-        from ._lazy_imports import _get_litellm_globals
+        from ._lazy_imports import get_litellm_globals
 
-        _globals = _get_litellm_globals()
+        _globals = get_litellm_globals()
         # Check if already cached
         if "_service_logger" not in _globals:
             # Import the module lazily

litellm/_lazy_imports.py

@@ -54,7 +54,7 @@ from ._lazy_imports_registry import (
 )
 
 
-def _get_litellm_globals() -> dict:
+def get_litellm_globals() -> dict:
     """
     Get the globals dictionary of the litellm module.
 
@@ -233,7 +233,7 @@ def _generic_lazy_import(name: str, import_map: dict[str, tuple[str, str]], cate
         raise AttributeError(f"{category} lazy import: unknown attribute {name!r}")
 
     # Step 2: Get the cache (where we store imported things)
-    _globals: Final = _get_litellm_globals()
+    _globals: Final = get_litellm_globals()
 
     # Step 3: If we've already imported it, just return the cached version
     if name in _globals:
@@ -332,7 +332,7 @@ def _lazy_import_utils_module(name: str) -> Any:
     Handler for utils module lazy imports.
 
     This uses a custom implementation because utils module needs to use
-    _get_utils_globals() instead of _get_litellm_globals() for caching.
+    _get_utils_globals() instead of get_litellm_globals() for caching.
     """
     # Check if this attribute exists in our map
     if name not in _UTILS_MODULE_IMPORT_MAP:
@@ -379,7 +379,7 @@ def _lazy_import_llm_client_cache(name: str) -> Any:
     - "in_memory_llm_clients_cache" is a singleton instance of that class
     So we need custom logic to handle both cases.
     """
-    _globals: Final = _get_litellm_globals()
+    _globals: Final = get_litellm_globals()
 
     # If already cached, return it
     if name in _globals:
@@ -412,7 +412,7 @@ def _lazy_import_http_handlers(name: str) -> Any:
     - They need configuration (timeout, etc.) from the module globals
     - They use factory functions instead of direct instantiation
     """
-    _globals: Final = _get_litellm_globals()
+    _globals: Final = get_litellm_globals()
 
     if name == "module_level_aclient":
         # Create an async HTTP client using the factory function

litellm/litellm_core_utils/audio_utils/utils.py

@@ -180,6 +180,7 @@ def get_audio_file_content_hash(file_obj: FileTypes) -> str:
     if isinstance(file_obj, tuple):
         if len(file_obj) < 2:
             fallback_filename = str(file_obj[0]) if len(file_obj) > 0 else None
+            file_content_obj = None
         else:
             fallback_filename = str(file_obj[0]) if file_obj[0] is not None else None
             file_content_obj = file_obj[1]
@@ -206,7 +207,7 @@ def get_audio_file_content_hash(file_obj: FileTypes) -> str:
             except OSError:
                 fallback_filename = str(file_content_obj)
                 file_content = None
-        elif hasattr(file_content_obj, "read"):
+        elif file_content_obj is not None and hasattr(file_content_obj, "read"):
             try:
                 current_position: Final = file_content_obj.tell() if hasattr(file_content_obj, "tell") else None
                 if hasattr(file_content_obj, "seek"):

litellm/litellm_core_utils/prompt_templates/factory.py

@@ -3684,7 +3684,7 @@ def _convert_to_bedrock_tool_call_invoke(
                             # cache_control applies to the whole original
                             # tool call; attach after the last split block.
                             if tool.get("cache_control", None) is not None:
-                                _cache_point_block = litellm.AmazonConverseConfig()._get_cache_point_block(
+                                _cache_point_block = litellm.AmazonConverseConfig().get_cache_point_block(
                                     {"cache_control": tool["cache_control"]},
                                     block_type="content_block",
                                     model=model,
@@ -3701,7 +3701,7 @@ def _convert_to_bedrock_tool_call_invoke(
 
                 # Check for cache_control and add a separate cachePoint block
                 if tool.get("cache_control", None) is not None:
-                    cache_point_block = litellm.AmazonConverseConfig()._get_cache_point_block(
+                    cache_point_block = litellm.AmazonConverseConfig().get_cache_point_block(
                         {"cache_control": tool["cache_control"]},
                         block_type="content_block",
                         model=model,
@@ -4360,7 +4360,7 @@ class BedrockConverseMessagesProcessor:
                             elif element["type"] == "document":
                                 _part = BedrockConverseMessagesProcessor._process_document_message(element)
                                 _parts.append(_part)
-                            _cache_point_block = litellm.AmazonConverseConfig()._get_cache_point_block(
+                            _cache_point_block = litellm.AmazonConverseConfig().get_cache_point_block(
                                 message_block=cast(OpenAIMessageContentListBlock, element),
                                 block_type="content_block",
                                 model=model,
@@ -4370,7 +4370,7 @@ class BedrockConverseMessagesProcessor:
                     user_content.extend(_parts)
                 elif message_block["content"] and isinstance(message_block["content"], str):
                     _part = BedrockContentBlock(text=messages[msg_i]["content"])
-                    _cache_point_block = litellm.AmazonConverseConfig()._get_cache_point_block(
+                    _cache_point_block = litellm.AmazonConverseConfig().get_cache_point_block(
                         message_block, block_type="content_block", model=model
                     )
                     user_content.append(_part)
@@ -4417,7 +4417,7 @@ class BedrockConverseMessagesProcessor:
 
                 # Add a separate cachePoint block if cache_control is present
                 if tool_msg_cache_control is not None:
-                    cache_point_block = litellm.AmazonConverseConfig()._get_cache_point_block(
+                    cache_point_block = litellm.AmazonConverseConfig().get_cache_point_block(
                         {"cache_control": tool_msg_cache_control},
                         block_type="content_block",
                         model=model,
@@ -4496,7 +4496,7 @@ class BedrockConverseMessagesProcessor:
                                 assistants_part = await BedrockImageProcessor.process_image_async(image_url=image_url)
                                 assistants_parts.append(assistants_part)
                                 # Add cache point block for assistant content elements
-                        _cache_point_block = litellm.AmazonConverseConfig()._get_cache_point_block(
+                        _cache_point_block = litellm.AmazonConverseConfig().get_cache_point_block(
                             message_block=cast(OpenAIMessageContentListBlock, element),
                             block_type="content_block",
                             model=model,
@@ -4510,7 +4510,7 @@ class BedrockConverseMessagesProcessor:
                         assistant_content.append(BedrockContentBlock(text=_assistant_content))
                     # If content is empty/whitespace, skip it (don't add a placeholder)
                     # Add cache point block for assistant string content
-                    _cache_point_block = litellm.AmazonConverseConfig()._get_cache_point_block(
+                    _cache_point_block = litellm.AmazonConverseConfig().get_cache_point_block(
                         assistant_message_block, block_type="content_block", model=model
                     )
                     if _cache_point_block is not None:
@@ -4733,7 +4733,7 @@ def _bedrock_converse_messages_pt(
                         elif element["type"] == "document":
                             _part = BedrockConverseMessagesProcessor._process_document_message(element)
                             _parts.append(_part)
-                        _cache_point_block = litellm.AmazonConverseConfig()._get_cache_point_block(
+                        _cache_point_block = litellm.AmazonConverseConfig().get_cache_point_block(
                             message_block=cast(OpenAIMessageContentListBlock, element),
                             block_type="content_block",
                             model=model,
@@ -4743,7 +4743,7 @@ def _bedrock_converse_messages_pt(
                 user_content.extend(_parts)
             elif message_block["content"] and isinstance(message_block["content"], str):
                 _part = BedrockContentBlock(text=messages[msg_i]["content"])
-                _cache_point_block = litellm.AmazonConverseConfig()._get_cache_point_block(
+                _cache_point_block = litellm.AmazonConverseConfig().get_cache_point_block(
                     message_block, block_type="content_block", model=model
                 )
                 user_content.append(_part)
@@ -4792,7 +4792,7 @@ def _bedrock_converse_messages_pt(
 
             # Add a separate cachePoint block if cache_control is present
             if tool_msg_cache_control is not None:
-                cache_point_block = litellm.AmazonConverseConfig()._get_cache_point_block(
+                cache_point_block = litellm.AmazonConverseConfig().get_cache_point_block(
                     {"cache_control": tool_msg_cache_control},
                     block_type="content_block",
                     model=model,
@@ -4874,7 +4874,7 @@ def _bedrock_converse_messages_pt(
                             assistants_part = BedrockImageProcessor.process_image_sync(image_url=image_url)
                             assistants_parts.append(assistants_part)
                         # Add cache point block for assistant content elements
-                        _cache_point_block = litellm.AmazonConverseConfig()._get_cache_point_block(
+                        _cache_point_block = litellm.AmazonConverseConfig().get_cache_point_block(
                             message_block=cast(OpenAIMessageContentListBlock, element),
                             block_type="content_block",
                             model=model,
@@ -4887,7 +4887,7 @@ def _bedrock_converse_messages_pt(
                 if _assistant_content.strip():
                     assistant_content.append(BedrockContentBlock(text=_assistant_content))
                 # Add cache point block for assistant string content
-                _cache_point_block = litellm.AmazonConverseConfig()._get_cache_point_block(
+                _cache_point_block = litellm.AmazonConverseConfig().get_cache_point_block(
                     assistant_message_block, block_type="content_block", model=model
                 )
                 if _cache_point_block is not None:

litellm/llms/bedrock/chat/converse_transformation.py

@@ -1081,7 +1081,7 @@ class AmazonConverseConfig(BaseConfig):
                 optional_params["maxTokens"] = thinking_token_budget + DEFAULT_MAX_TOKENS
 
     @overload
-    def _get_cache_point_block(
+    def get_cache_point_block(
         self,
         message_block: OpenAIMessageContentListBlock
         | ChatCompletionUserMessage
@@ -1093,7 +1093,7 @@ class AmazonConverseConfig(BaseConfig):
         pass
 
     @overload
-    def _get_cache_point_block(
+    def get_cache_point_block(
         self,
         message_block: OpenAIMessageContentListBlock
         | ChatCompletionUserMessage
@@ -1104,7 +1104,7 @@ class AmazonConverseConfig(BaseConfig):
     ) -> ContentBlock | None:
         pass
 
-    def _get_cache_point_block(
+    def get_cache_point_block(
         self,
         message_block: OpenAIMessageContentListBlock
         | ChatCompletionUserMessage
@@ -1149,14 +1149,14 @@ class AmazonConverseConfig(BaseConfig):
                 system_prompt_indices.append(idx)
                 if isinstance(message["content"], str) and message["content"]:
                     system_content_blocks.append(SystemContentBlock(text=message["content"]))
-                    cache_block = self._get_cache_point_block(message, block_type="system", model=model)
+                    cache_block = self.get_cache_point_block(message, block_type="system", model=model)
                     if cache_block:
                         system_content_blocks.append(cache_block)
                 elif isinstance(message["content"], list):
                     for m in message["content"]:
                         if m.get("type") == "text" and m.get("text"):
                             system_content_blocks.append(SystemContentBlock(text=m["text"]))
-                            cache_block = self._get_cache_point_block(m, block_type="system", model=model)
+                            cache_block = self.get_cache_point_block(m, block_type="system", model=model)
                             if cache_block:
                                 system_content_blocks.append(cache_block)
         if len(system_prompt_indices) > 0:

litellm/llms/xai/oauth.py

@@ -40,7 +40,7 @@ class XAIOAuthLoginRequiredError(XAIOAuthError):
 
 
 class _CallbackHandler(BaseHTTPRequestHandler):
-    server: "_CallbackServer"
+    server: "_CallbackServer"  # pyright: ignore[reportIncompatibleVariableOverride]  # stdlib stubs type server as BaseServer; _CallbackServer is the only server this handler is registered on
 
     def do_GET(self) -> None:
         parsed: Final = urlparse(self.path)

litellm/proxy/_experimental/mcp_server/elicitation_handler.py

@@ -9,10 +9,19 @@ MCP Spec Reference:
     https://modelcontextprotocol.io/specification/2025-11-25/client/elicitation
 """
 
-from typing import Any, Final, Union
+from typing import TYPE_CHECKING, Any, Final, Union
 
 from litellm._logging import verbose_logger
 
+if TYPE_CHECKING:
+    from mcp.types import (
+        ElicitRequestFormParams,
+        ElicitRequestParams,
+        ElicitRequestURLParams,
+        ElicitResult,
+        ErrorData,
+    )
+
 # Guard imports that require the mcp package
 try:
     from mcp.types import (

litellm/proxy/_experimental/mcp_server/sampling_handler.py

@@ -18,7 +18,15 @@ if typing.TYPE_CHECKING:
     from fastapi import Request
     from mcp.client.session import ClientSession
     from mcp.shared.context import RequestContext
-    from mcp.types import ContentBlock, SamplingMessageContentBlock
+    from mcp.types import (
+        ContentBlock,
+        CreateMessageResult,
+        CreateMessageResultWithTools,
+        ErrorData,
+        SamplingMessageContentBlock,
+        TextContent,
+        ToolUseContent,
+    )
 
     from litellm.proxy._types import UserAPIKeyAuth
     from litellm.proxy.utils import ProxyLogging

litellm/proxy/_experimental/mcp_server/server.py

@@ -79,6 +79,8 @@ from litellm.types.utils import CallTypes, StandardLoggingMCPToolCall
 from litellm.utils import Rules, client, function_setup
 
 if TYPE_CHECKING:
+    from mcp.server.session import ServerSession as _McpServerSession
+
     from litellm.proxy._experimental.mcp_server.db import OAuthCredentialPayload
 
 # Short-lived in-memory cache for BYOK credentials.
@@ -144,10 +146,6 @@ try:
 
     # Robust auth lookup keyed by session_object.
     _session_obj_auth_storage: "weakref.WeakKeyDictionary[Any, MCPAuthenticatedUser]" = weakref.WeakKeyDictionary()
-
-    active_mcp_session_var: Final[contextvars.ContextVar[_McpServerSession | None]] = contextvars.ContextVar(
-        "active_mcp_session", default=None
-    )
 except ImportError as e:
     verbose_logger.debug("MCP module not found: %s", e)
     MCP_AVAILABLE = False
@@ -163,6 +161,10 @@ except ImportError as e:
     Server = None
     TextResourceContents = None
 
+active_mcp_session_var: Final[contextvars.ContextVar["_McpServerSession | None"]] = contextvars.ContextVar(
+    "active_mcp_session", default=None
+)
+
 
 # Global variables to track initialization
 _SESSION_MANAGERS_INITIALIZED = False

litellm/proxy/example_config_yaml/custom_guardrail.py

@@ -1,11 +1,10 @@
-from typing import Any, Dict, Final, List, Literal, Optional, Union
+from typing import Dict, Final, Optional, Union
 
 import litellm
 from litellm._logging import verbose_proxy_logger
 from litellm.caching.caching import DualCache
 from litellm.integrations.custom_guardrail import CustomGuardrail
 from litellm.proxy._types import UserAPIKeyAuth
-from litellm.proxy.guardrails.guardrail_helpers import should_proceed_based_on_metadata
 from litellm.types.utils import CallTypesLiteral
 
 # Global counter for tracking which guardrail was called (for load balancing tests)

litellm/proxy/example_config_yaml/custom_handler.py

@@ -1,9 +1,7 @@
-import time
-from typing import Any, Final, Optional
+from typing import Final
 
 import litellm
-from litellm import CustomLLM, ImageObject, ImageResponse, completion, get_llm_provider
-from litellm.llms.custom_httpx.http_handler import AsyncHTTPHandler
+from litellm import CustomLLM
 from litellm.types.utils import ModelResponse
 
 

ui/litellm-dashboard/eslint-budgets.json

@@ -1,8 +1,8 @@
 {
   "@typescript-eslint/no-explicit-any": { "max": 2040, "target": 1500 },
   "no-console": { "max": 12, "target": 0 },
-  "complexity": { "max": 121, "target": 80 },
-  "max-depth": { "max": 55, "target": 30 },
-  "local/no-large-inline-object-arg": { "max": 469, "target": 300 },
-  "local/no-long-condition-chain": { "max": 217, "target": 120 }
+  "complexity": { "max": 140, "target": 80 },
+  "max-depth": { "max": 70, "target": 30 },
+  "local/no-large-inline-object-arg": { "max": 560, "target": 300 },
+  "local/no-long-condition-chain": { "max": 265, "target": 120 }
 }