Tiny move files to utils folder (#11166)

test/srt/test_patch_torch.py

@@ -6,7 +6,7 @@ from typing import Dict, List
 import torch
 import torch.multiprocessing as mp
 
-from sglang.srt.patch_torch import monkey_patch_torch_reductions
+from sglang.srt.utils.patch_torch import monkey_patch_torch_reductions
 
 
 class TestReleaseMemoryOccupation(unittest.TestCase):

test/srt/test_sagemaker_server.py

@@ -7,8 +7,8 @@ import unittest
 
 import requests
 
-from sglang.srt.hf_transformers_utils import get_tokenizer
 from sglang.srt.utils import kill_process_tree
+from sglang.srt.utils.hf_transformers_utils import get_tokenizer
 from sglang.test.test_utils import (
     DEFAULT_SMALL_MODEL_NAME_FOR_TEST,
     DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,

test/srt/test_session_control.py

@@ -13,8 +13,8 @@ import unittest
 import aiohttp
 import requests
 
-from sglang.srt.hf_transformers_utils import get_tokenizer
 from sglang.srt.utils import kill_process_tree
+from sglang.srt.utils.hf_transformers_utils import get_tokenizer
 from sglang.test.test_utils import (
     DEFAULT_SMALL_MODEL_NAME_FOR_TEST,
     DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,

test/srt/test_srt_engine.py

@@ -12,8 +12,8 @@ import torch
 
 import sglang as sgl
 from sglang.bench_offline_throughput import BenchArgs, throughput_test
-from sglang.srt.hf_transformers_utils import get_tokenizer
 from sglang.srt.server_args import ServerArgs
+from sglang.srt.utils.hf_transformers_utils import get_tokenizer
 from sglang.test.few_shot_gsm8k_engine import run_eval
 from sglang.test.test_utils import (
     DEFAULT_SMALL_EMBEDDING_MODEL_NAME_FOR_TEST,

test/srt/test_tokenizer_batch_encode.py

@@ -34,7 +34,9 @@ class TestTokenizerBatchEncode(unittest.TestCase):
 
         with patch("zmq.asyncio.Context"), patch(
             "sglang.srt.utils.get_zmq_socket"
-        ), patch("sglang.srt.hf_transformers_utils.get_tokenizer") as mock_tokenizer:
+        ), patch(
+            "sglang.srt.utils.hf_transformers_utils.get_tokenizer"
+        ) as mock_tokenizer:
 
             mock_tokenizer.return_value = Mock(vocab_size=32000)
             self.tokenizer_manager = TokenizerManager(self.server_args, self.port_args)

test/srt/test_tokenizer_manager.py

@@ -31,7 +31,9 @@ class TestInputFormatDetection(unittest.TestCase):
 
         with patch("zmq.asyncio.Context"), patch(
             "sglang.srt.utils.get_zmq_socket"
-        ), patch("sglang.srt.hf_transformers_utils.get_tokenizer") as mock_tokenizer:
+        ), patch(
+            "sglang.srt.utils.hf_transformers_utils.get_tokenizer"
+        ) as mock_tokenizer:
             mock_tokenizer.return_value = Mock(vocab_size=32000)
             self.tokenizer_manager = TokenizerManager(self.server_args, self.port_args)
 
@@ -125,7 +127,9 @@ class TestTokenizerInputPreparation(unittest.TestCase):
 
         with patch("zmq.asyncio.Context"), patch(
             "sglang.srt.utils.get_zmq_socket"
-        ), patch("sglang.srt.hf_transformers_utils.get_tokenizer") as mock_tokenizer:
+        ), patch(
+            "sglang.srt.utils.hf_transformers_utils.get_tokenizer"
+        ) as mock_tokenizer:
             mock_tokenizer.return_value = Mock(vocab_size=32000)
             self.tokenizer_manager = TokenizerManager(self.server_args, self.port_args)
 
@@ -177,7 +181,9 @@ class TestTokenizerResultExtraction(unittest.TestCase):
 
         with patch("zmq.asyncio.Context"), patch(
             "sglang.srt.utils.get_zmq_socket"
-        ), patch("sglang.srt.hf_transformers_utils.get_tokenizer") as mock_tokenizer:
+        ), patch(
+            "sglang.srt.utils.hf_transformers_utils.get_tokenizer"
+        ) as mock_tokenizer:
             mock_tokenizer.return_value = Mock(vocab_size=32000)
             self.tokenizer_manager = TokenizerManager(self.server_args, self.port_args)
 
@@ -279,7 +285,9 @@ class TestTokenizerManagerIntegration(unittest.TestCase):
 
         with patch("zmq.asyncio.Context"), patch(
             "sglang.srt.utils.get_zmq_socket"
-        ), patch("sglang.srt.hf_transformers_utils.get_tokenizer") as mock_tokenizer:
+        ), patch(
+            "sglang.srt.utils.hf_transformers_utils.get_tokenizer"
+        ) as mock_tokenizer:
             mock_tokenizer.return_value = Mock(vocab_size=32000)
             self.tokenizer_manager = TokenizerManager(self.server_args, self.port_args)