A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 2, 03:30 UTC
[CI] Add missing vllm/parser/ CI trigger and fix test_parse.py (#44352)
Signed-off-by: sfeng33 <4florafeng@gmail.com>
Flora Feng committed
Jun 3, 2026 at 00:05 UTC
e67063826b331ddd10e70f42eab9739ed8abf643
2 files changed
+7
-4
.buildkite/test_areas/misc.yaml
+1
@@ -281,6 +281,7 @@ steps:
281
- vllm/model_executor/layers/quantization/quark/
282
- vllm/multimodal/
283
- vllm/outputs.py
284
+ - vllm/parser/
285
- vllm/platforms/
286
- vllm/pooling_params.py
287
- vllm/ray/
tests/parser/test_parse.py
+6
-4
@@ -6,7 +6,7 @@ import json
6
import pytest
7
8
from vllm.entrypoints.openai.chat_completion.protocol import ChatCompletionRequest
9
-from vllm.parser.abstract_parser import _WrappedParser
9
+from vllm.parser.abstract_parser import DelegatingParser
10
from vllm.reasoning.basic_parsers import BaseThinkingReasoningParser
11
from vllm.tool_parsers.hermes_tool_parser import Hermes2ProToolParser
12
@@ -65,9 +65,11 @@ TOOLS = [
65
66
67
def make_parser(tokenizer, reasoning=False, tool=False):
68
- _WrappedParser.reasoning_parser_cls = ThinkReasoningParser if reasoning else None
69
- _WrappedParser.tool_parser_cls = Hermes2ProToolParser if tool else None
70
- return _WrappedParser(tokenizer)
68
+ class TestParser(DelegatingParser):
69
+ reasoning_parser_cls = ThinkReasoningParser if reasoning else None
70
+ tool_parser_cls = Hermes2ProToolParser if tool else None
71
+
72
+ return TestParser(tokenizer)
73
74
75
@pytest.mark.parametrize(