EngineX/xc-llm-ascend
3
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

[CI] Add pre-commit check for patch logger (#7446)

Browse Source 
### What this PR does / why we need it?
See https://github.com/vllm-project/vllm-ascend/pull/7402, pre-commit
hook will forbid init_logger(__name__) in vllm_ascend patch modules

- vLLM version: v0.17.0
- vLLM main:
8a680463fa

---------

Signed-off-by: wangli <wangli858794774@gmail.com>
This commit is contained in:
Li Wang
2026-03-19 16:53:20 +08:00
committed by GitHub
parent 38e637eef5
commit 83a4065b4b
14 changed files with 99 additions and 49 deletions
Download Patch File Download Diff File Expand all files Collapse all files

34
tests/ut/quantization/test_quant_utils.py
View File

@@ -1,5 +1,4 @@
import json
import logging
import os
import tempfile
from pathlib import Path
@@ -135,16 +134,16 @@ class TestMaybeAutoDetectQuantization(TestBase):
vllm_config = self._make_vllm_config(
model_path="/fake/quant_model", quantization=None)
with self.assertLogs("vllm_ascend.quantization.utils",
level=logging.INFO) as cm:
with patch("vllm_ascend.quantization.utils.logger") as mock_logger:
maybe_auto_detect_quantization(vllm_config)
self.assertEqual(vllm_config.model_config.quantization,
ASCEND_QUANTIZATION_METHOD)
log_output = "\n".join(cm.output)
self.assertIn("Auto-detected quantization method", log_output)
self.assertIn(ASCEND_QUANTIZATION_METHOD, log_output)
self.assertIn("/fake/quant_model", log_output)
mock_logger.info.assert_called_once()
call_args = mock_logger.info.call_args[0]
self.assertIn("Auto-detected quantization method", call_args[0])
self.assertIn(ASCEND_QUANTIZATION_METHOD, call_args)
self.assertIn("/fake/quant_model", call_args)
@patch("vllm_ascend.quantization.utils.detect_quantization_method",
return_value=ASCEND_QUANTIZATION_METHOD)
@@ -155,29 +154,28 @@ class TestMaybeAutoDetectQuantization(TestBase):
model_path="/fake/quant_model",
quantization=COMPRESSED_TENSORS_METHOD)
with self.assertLogs("vllm_ascend.quantization.utils",
level=logging.WARNING) as cm:
with patch("vllm_ascend.quantization.utils.logger") as mock_logger:
maybe_auto_detect_quantization(vllm_config)
self.assertEqual(vllm_config.model_config.quantization,
COMPRESSED_TENSORS_METHOD)
log_output = "\n".join(cm.output)
self.assertIn("Auto-detected quantization method", log_output)
self.assertIn(ASCEND_QUANTIZATION_METHOD, log_output)
self.assertIn(COMPRESSED_TENSORS_METHOD, log_output)
mock_logger.warning.assert_called_once()
call_args = mock_logger.warning.call_args[0]
self.assertIn("Auto-detected quantization method", call_args[0])
self.assertIn(ASCEND_QUANTIZATION_METHOD, call_args)
self.assertIn(COMPRESSED_TENSORS_METHOD, call_args)
@patch("vllm_ascend.quantization.utils.detect_quantization_method",
return_value=None)
def test_no_detection_emits_no_log(self, mock_detect):
"""When no quantization is detected, no log should be emitted."""
vllm_config = self._make_vllm_config(quantization=None)
logger_name = "vllm_ascend.quantization.utils"
with self.assertRaises(AssertionError):
# assertLogs raises AssertionError when no logs are emitted
with self.assertLogs(logger_name, level=logging.DEBUG):
maybe_auto_detect_quantization(vllm_config)
with patch("vllm_ascend.quantization.utils.logger") as mock_logger:
maybe_auto_detect_quantization(vllm_config)
mock_logger.info.assert_not_called()
mock_logger.warning.assert_not_called()
self.assertIsNone(vllm_config.model_config.quantization)
@patch("vllm.config.VllmConfig._get_quantization_config",