[Lint] Add `python/sglang` to ruff F401 checks and remove unused imports in files (#11685)

python/sglang/srt/models/qwen3_vl.py

@@ -20,18 +20,13 @@ from typing import Callable, Iterable, List, Optional, Tuple, Union
 import numpy as np
 import torch
 import torch.nn as nn
-import torch.nn.functional as F
 from einops import rearrange
 from transformers.activations import ACT2FN
 from transformers.models.qwen2_5_vl.modeling_qwen2_5_vl import (
     Qwen2_5_VisionRotaryEmbedding,
 )
 
-from sglang.srt.configs.qwen3_vl import (
-    Qwen3VLConfig,
-    Qwen3VLTextConfig,
-    Qwen3VLVisionConfig,
-)
+from sglang.srt.configs.qwen3_vl import Qwen3VLConfig, Qwen3VLVisionConfig
 from sglang.srt.layers.attention.vision import VisionAttention
 from sglang.srt.layers.linear import ColumnParallelLinear, RowParallelLinear
 from sglang.srt.layers.logits_processor import LogitsProcessor
@@ -47,11 +42,7 @@ from sglang.srt.managers.schedule_batch import (
     MultimodalDataItem,
     MultimodalInputs,
 )
-from sglang.srt.model_executor.forward_batch_info import (
-    ForwardBatch,
-    ForwardMode,
-    PPProxyTensors,
-)
+from sglang.srt.model_executor.forward_batch_info import ForwardBatch, PPProxyTensors
 from sglang.srt.model_loader.weight_utils import default_weight_loader
 from sglang.srt.models.qwen3 import Qwen3Model
 from sglang.srt.utils import add_prefix

python/sglang/srt/models/qwen3_vl_moe.py

@@ -25,12 +25,8 @@ from sglang.srt.distributed import (
     get_moe_expert_parallel_world_size,
     get_tensor_model_parallel_rank,
 )
-from sglang.srt.layers.logits_processor import LogitsProcessor
 from sglang.srt.layers.moe.fused_moe_triton.layer import FusedMoE
 from sglang.srt.layers.quantization.base_config import QuantizationConfig
-from sglang.srt.layers.vocab_parallel_embedding import ParallelLMHead
-from sglang.srt.managers.mm_utils import general_mm_embed_routine
-from sglang.srt.managers.schedule_batch import MultimodalDataItem
 from sglang.srt.model_executor.forward_batch_info import ForwardBatch, PPProxyTensors
 from sglang.srt.model_loader.weight_utils import default_weight_loader
 from sglang.srt.models.qwen3_moe import Qwen3MoeModel

python/sglang/srt/models/roberta.py

 # SPDX-License-Identifier: Apache-2.0
 
-import itertools
 from typing import Iterable, Optional, Tuple
 
 import torch

python/sglang/srt/models/sarashina2_vision.py

@@ -17,7 +17,6 @@ import logging
 from typing import Iterable, List, Optional, Tuple
 
 import torch
-import torch.nn.functional as F
 from torch import nn
 from transformers import LlamaConfig
 

python/sglang/srt/models/step3_vl.py

 import logging
 import math
-from collections.abc import Iterable
 from math import sqrt
-from typing import Any, Dict, Iterable, List, Literal, Optional, Tuple, TypedDict, Union
+from typing import Any, Dict, Iterable, List, Optional, Tuple
 
 import torch
 from torch import nn

python/sglang/srt/multimodal/processors/deepseek_vl_v2.py

@@ -18,9 +18,6 @@
 # CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 from typing import List, Union
 
-import torch
-
-from sglang.srt.managers.schedule_batch import Modality, MultimodalDataItem
 from sglang.srt.models.deepseek_vl2 import DeepseekVL2ForCausalLM
 from sglang.srt.multimodal.processors.base_processor import (
     BaseMultimodalProcessor,

python/sglang/srt/multimodal/processors/dots_vlm.py

 import asyncio
-import math
 import re
 from typing import Dict, List, Union
 

python/sglang/srt/multimodal/processors/glm4v.py

-import re
 from typing import List, Union
 
 from decord import VideoReader
@@ -9,10 +8,7 @@ from sglang.srt.models.glm4v_moe import Glm4vMoeForConditionalGeneration
 from sglang.srt.multimodal.processors.base_processor import (
     BaseMultimodalProcessor as SGLangBaseProcessor,
 )
-from sglang.srt.multimodal.processors.base_processor import (
-    BaseMultiModalProcessorOutput,
-    MultimodalSpecialTokens,
-)
+from sglang.srt.multimodal.processors.base_processor import MultimodalSpecialTokens
 
 
 class Glm4vImageProcessor(SGLangBaseProcessor):

python/sglang/srt/multimodal/processors/internvl.py

@@ -4,10 +4,8 @@ from functools import lru_cache
 
 import numpy as np
 import torch
-import torchvision.transforms as T
 from decord import VideoReader, cpu, gpu
 from PIL import Image
-from torchvision.transforms import InterpolationMode
 
 from sglang.srt.managers.schedule_batch import Modality, MultimodalDataItem
 from sglang.srt.models.interns1 import InternS1ForConditionalGeneration

python/sglang/srt/multimodal/processors/janus_pro.py

 from typing import List, Union
 
-from sglang.srt.managers.schedule_batch import Modality, MultimodalDataItem
 from sglang.srt.models.deepseek_janus_pro import MultiModalityCausalLM
 from sglang.srt.multimodal.processors.base_processor import (
     BaseMultimodalProcessor,

python/sglang/srt/multimodal/processors/mllama4.py

 from typing import List, Union
 
-import torch
-from transformers.image_utils import SizeDict
-from transformers.models.llama4.image_processing_llama4_fast import (
-    find_supported_resolutions,
-    get_best_fit,
-)
-
-from sglang.srt.managers.schedule_batch import Modality, MultimodalDataItem
 from sglang.srt.models.mllama4 import Llama4ForConditionalGeneration
 from sglang.srt.multimodal.processors.base_processor import (
     BaseMultimodalProcessor,

python/sglang/srt/multimodal/processors/phi4mm.py

@@ -3,7 +3,6 @@ from typing import List, Union
 
 from transformers.processing_utils import ProcessorMixin
 
-from sglang.srt.managers.schedule_batch import Modality, MultimodalDataItem
 from sglang.srt.models.phi4mm import Phi4MMForCausalLM
 from sglang.srt.multimodal.processors.base_processor import (
     BaseMultimodalProcessor,

python/sglang/srt/multimodal/processors/step3_vl.py

 import math
 import re
 from itertools import product
-from typing import List, Literal, Optional, TypedDict, Union
+from typing import List, Optional, Union
 
 import numpy as np
 import torch

python/sglang/srt/parser/reasoning_parser.py

-import re
 from typing import Dict, Optional, Tuple, Type
 
 from sglang.srt.parser.harmony_parser import HarmonyParser

python/sglang/srt/server_args_config_parser.py

@@ -5,7 +5,7 @@ Handles merging of YAML configuration files with command-line arguments.
 
 import logging
 from pathlib import Path
-from typing import Any, Dict, List, Union
+from typing import Any, Dict, List
 
 import yaml
 

python/sglang/srt/speculative/eagle_worker.py

@@ -56,7 +56,7 @@ from sglang.srt.utils import (
 )
 
 if is_cuda():
-    from sgl_kernel import segment_packbits
+    from sgl_kernel import segment_packbits  # noqa: F401
 
 logger = logging.getLogger(__name__)
 SGLANG_RETURN_ORIGINAL_LOGPROB = get_bool_env_var("SGLANG_RETURN_ORIGINAL_LOGPROB")

python/sglang/srt/speculative/spec_utils.py

@@ -22,8 +22,6 @@ from sglang.srt.managers.schedule_batch import Req
 from sglang.srt.utils import is_cuda, is_hip
 
 if TYPE_CHECKING:
-    from sglang.srt.mem_cache.allocator import TokenToKVPoolAllocator
-    from sglang.srt.mem_cache.memory_pool import ReqToTokenPool
     from sglang.srt.speculative.eagle_info import EagleVerifyInput
 
 

python/sglang/srt/speculative/standalone_worker.py

@@ -11,7 +11,7 @@ from sglang.srt.speculative.spec_utils import draft_tp_context, load_token_map
 from sglang.srt.utils import empty_context, get_bool_env_var, is_cuda
 
 if is_cuda():
-    from sgl_kernel import segment_packbits
+    from sgl_kernel import segment_packbits  # noqa: F401
 
 logger = logging.getLogger(__name__)
 SGLANG_RETURN_ORIGINAL_LOGPROB = get_bool_env_var("SGLANG_RETURN_ORIGINAL_LOGPROB")

python/sglang/srt/utils/common.py

@@ -228,7 +228,7 @@ def support_triton(backend: str) -> bool:
 
 
 try:
-    import sgl_kernel
+    import sgl_kernel  # noqa: F401
 
     is_intel_amx_backend_available = hasattr(
         torch.ops.sgl_kernel, "convert_weight_packed"
@@ -1556,7 +1556,7 @@ def get_hpu_memory_capacity():
 
 def get_npu_memory_capacity():
     try:
-        import torch_npu
+        import torch_npu  # noqa: F401
 
         return torch.npu.mem_get_info()[1] // 1024 // 1024  # unit: MB
     except ImportError as e:
@@ -1743,7 +1743,7 @@ def get_device(device_id: Optional[int] = None) -> str:
 
     if is_habana_available():
         try:
-            import habana_frameworks.torch.hpu
+            import habana_frameworks.torch.hpu  # noqa: F401
 
             if torch.hpu.is_available():
                 if device_id == None:
@@ -1773,7 +1773,7 @@ def get_device_count() -> int:
 
     if is_habana_available():
         try:
-            import habana_frameworks.torch.hpu
+            import habana_frameworks.torch.hpu  # noqa: F401
 
             if torch.hpu.is_available():
                 return torch.hpu.device_count()

python/sglang/srt/utils/host_shared_memory.py

 import logging
-import os
 from dataclasses import dataclass
 from multiprocessing import shared_memory
 from pathlib import Path