Remove tuple stuff

Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Signed-off-by: weichen <calvin_zhu0210@outlook.com>

vllm/v1/core/sched/policy/weighted_score_sorter.py

@@ -2,9 +2,9 @@
 # SPDX-FileCopyrightText: Copyright contributors to the vLLM project
 import time
 from functools import total_ordering
-from typing import Any
 
 from vllm.v1.core.sched.policy.normalized_scorer import TimeAndLengthScorer
+from vllm.v1.request import Request
 
 TimeAndLengthScorer_Instance = None
 
@@ -20,15 +20,11 @@ if TimeAndLengthScorer_Instance is None:
 
 @total_ordering
 class WeightedScoreSorter:
-    def __init__(
+    def __init__(self, request: Request):
-        self,
+        self.request = request
-        request_length: int,
+        assert request.prompt_token_ids is not None
-        request_arrival_time: float,
+        self.request_length = len(request.prompt_token_ids)
-        request_slo_requirement: list[Any] | None = None,
+        self.request_arrival_time = request.arrival_time
-    ):
-        self.request_length = request_length
-        self.request_arrival_time = request_arrival_time
-        self.request_slo_requirement = request_slo_requirement
         self.__update_stats()
 
     def __lt__(self, other_request_weighted_score: "WeightedScoreSorter") -> bool:

vllm/v1/core/sched/request_queue.py

@@ -228,18 +228,13 @@ class SJFRequestQueue(RequestHeap):
     """A Shortest Job First (SJF) queue where requests are ordered by weighted score.
     Requests with higher weighted scores (shorter jobs) are processed first."""
 
-    def _request_to_heap(self, request: Request) -> tuple[WeightedScoreSorter, Request]:
+    def _request_to_heap(self, request: Request) -> WeightedScoreSorter:
-        """Convert request to (weighted_score, request) tuple for heap."""
+        """Convert request to `WeightedScoreSorter` for heap."""
-        assert request.prompt_token_ids is not None
+        return WeightedScoreSorter(request)
-        return (
-            WeightedScoreSorter(len(request.prompt_token_ids), request.arrival_time),
-            request,
-        )
 
-    def _heap_to_request(self, element: tuple[WeightedScoreSorter, Request]) -> Request:
+    def _heap_to_request(self, element: WeightedScoreSorter) -> Request:
-        """Extract request from the (score, request) tuple with type checking."""
+        """Extract request from the `WeightedScoreSorter`."""
-        _, request = element
+        return element.request
-        return request
 
 
 def create_request_queue(policy: SchedulingPolicy) -> RequestQueue: