[BugFix]: Fix local mypy issues (#34739)

Signed-off-by: Martin Hickey <martin.hickey@ie.ibm.com>
Co-authored-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>

vllm/distributed/kv_transfer/kv_connector/v1/nixl_connector.py

@@ -935,6 +935,10 @@ class NixlConnectorWorker:
             ]
 
             if rsv_cores_for_kv:
+                if not hasattr(os, "sched_setaffinity"):
+                    raise NotImplementedError(
+                        "os.sched_setaffinity is not available on this platform"
+                    )
                 os.sched_setaffinity(0, rsv_cores_for_kv)
 
         # support for oot platform which can't register nixl memory

vllm/entrypoints/openai/chat_completion/protocol.py

@@ -5,7 +5,6 @@
 # https://github.com/lm-sys/FastChat/blob/168ccc29d3f7edc50823016105c024fe2282732a/fastchat/protocol/openai_api_protocol.py
 import json
 import time
-from dataclasses import replace
 from typing import Annotated, Any, ClassVar, Literal
 
 import torch
@@ -16,6 +15,7 @@ from openai.types.chat.chat_completion_message import Annotation as OpenAIAnnota
 from pydantic import Field, model_validator
 
 from vllm.config import ModelConfig
+from vllm.config.utils import replace
 from vllm.entrypoints.chat_utils import (
     ChatCompletionMessageParam,
     ChatTemplateContentFormatOption,

vllm/entrypoints/openai/completion/protocol.py

@@ -5,13 +5,13 @@
 # https://github.com/lm-sys/FastChat/blob/168ccc29d3f7edc50823016105c024fe2282732a/fastchat/protocol/openai_api_protocol.py
 import json
 import time
-from dataclasses import replace
 from typing import Annotated, Any, Literal
 
 import torch
 from pydantic import Field, model_validator
 
 from vllm.config import ModelConfig
+from vllm.config.utils import replace
 from vllm.entrypoints.openai.engine.protocol import (
     AnyResponseFormat,
     LegacyStructuralTagResponseFormat,

vllm/entrypoints/openai/responses/protocol.py

@@ -337,7 +337,9 @@ class ResponsesRequest(OpenAIBaseModel):
                 and response_format.schema_ is not None
             ):
                 structured_outputs = StructuredOutputsParams(
-                    json=response_format.schema_
+                    json=response_format.schema_  # type: ignore[call-arg]
+                    # --follow-imports skip hides the class definition but also hides
+                    # multiple third party conflicts, so best of both evils
                 )
 
         stop = self.stop if self.stop else []

vllm/entrypoints/openai/responses/serving.py

@@ -8,7 +8,6 @@ from collections import deque
 from collections.abc import AsyncGenerator, AsyncIterator, Callable, Sequence
 from contextlib import AsyncExitStack
 from copy import copy
-from dataclasses import replace
 from http import HTTPStatus
 from typing import Final
 
@@ -40,6 +39,7 @@ from openai_harmony import Message as OpenAIHarmonyMessage
 from pydantic import TypeAdapter
 
 from vllm import envs
+from vllm.config.utils import replace
 from vllm.engine.protocol import EngineClient
 from vllm.entrypoints.chat_utils import (
     ChatCompletionMessageParam,

vllm/sampling_params.py

@@ -3,7 +3,7 @@
 """Sampling parameters for text generation."""
 
 import copy
-import json
+import json as json_mod
 from dataclasses import field
 from enum import Enum, IntEnum
 from functools import cached_property
@@ -791,7 +791,7 @@ class SamplingParams(
                 skip_guidance = False
                 if so_params.json:
                     if isinstance(so_params.json, str):
-                        schema = json.loads(so_params.json)
+                        schema = json_mod.loads(so_params.json)
                     else:
                         schema = so_params.json
                     skip_guidance = has_guidance_unsupported_json_features(schema)