Move parsers under a single folder (#9912)

docs/advanced_features/separate_reasoning.ipynb

@@ -313,7 +313,7 @@
    "outputs": [],
    "source": [
     "import sglang as sgl\n",
-    "from sglang.srt.reasoning_parser import ReasoningParser\n",
+    "from sglang.srt.parser.reasoning_parser import ReasoningParser\n",
     "from sglang.utils import print_highlight\n",
     "\n",
     "llm = sgl.Engine(model_path=\"deepseek-ai/DeepSeek-R1-Distill-Qwen-7B\")\n",

docs/advanced_features/vlm_query.ipynb

@@ -44,7 +44,7 @@
     "import requests\n",
     "from PIL import Image\n",
     "\n",
-    "from sglang.srt.conversation import chat_templates\n",
+    "from sglang.srt.parser.conversation import chat_templates\n",
     "\n",
     "image = Image.open(\n",
     "    BytesIO(\n",
@@ -182,7 +182,7 @@
     "import requests\n",
     "from PIL import Image\n",
     "\n",
-    "from sglang.srt.conversation import chat_templates\n",
+    "from sglang.srt.parser.conversation import chat_templates\n",
     "\n",
     "image = Image.open(\n",
     "    BytesIO(\n",

examples/runtime/engine/offline_batch_inference_vlm.py

@@ -7,7 +7,7 @@ import argparse
 import dataclasses
 
 import sglang as sgl
-from sglang.srt.conversation import chat_templates
+from sglang.srt.parser.conversation import chat_templates
 from sglang.srt.server_args import ServerArgs
 
 

python/sglang/lang/interpreter.py

@@ -740,7 +740,7 @@ class StreamExecutor:
             # Execute the stored lazy generation calls
             self.backend.role_end_generate(self)
 
-        from sglang.srt.reasoning_parser import ReasoningParser
+        from sglang.srt.parser.reasoning_parser import ReasoningParser
 
         reasoning_parser = ReasoningParser(expr.model_type)
         other = expr.expr

python/sglang/srt/entrypoints/http_server.py

@@ -102,7 +102,7 @@ from sglang.srt.managers.multi_tokenizer_mixin import (
 from sglang.srt.managers.template_manager import TemplateManager
 from sglang.srt.managers.tokenizer_manager import ServerStatus, TokenizerManager
 from sglang.srt.metrics.func_timer import enable_func_timer
-from sglang.srt.reasoning_parser import ReasoningParser
+from sglang.srt.parser.reasoning_parser import ReasoningParser
 from sglang.srt.server_args import PortArgs, ServerArgs
 from sglang.srt.utils import (
     add_api_key_middleware,

python/sglang/srt/entrypoints/openai/serving_chat.py

@@ -8,7 +8,6 @@ from typing import Any, AsyncGenerator, Dict, List, Optional, Union
 from fastapi import Request
 from fastapi.responses import ORJSONResponse, StreamingResponse
 
-from sglang.srt.conversation import generate_chat_conv
 from sglang.srt.entrypoints.openai.protocol import (
     ChatCompletionRequest,
     ChatCompletionResponse,
@@ -33,11 +32,12 @@ from sglang.srt.entrypoints.openai.utils import (
     to_openai_style_logprobs,
 )
 from sglang.srt.function_call.function_call_parser import FunctionCallParser
-from sglang.srt.jinja_template_utils import process_content_for_template_format
 from sglang.srt.managers.io_struct import GenerateReqInput
 from sglang.srt.managers.template_manager import TemplateManager
 from sglang.srt.managers.tokenizer_manager import TokenizerManager
-from sglang.srt.reasoning_parser import ReasoningParser
+from sglang.srt.parser.conversation import generate_chat_conv
+from sglang.srt.parser.jinja_template_utils import process_content_for_template_format
+from sglang.srt.parser.reasoning_parser import ReasoningParser
 from sglang.utils import convert_json_schema_to_str
 
 logger = logging.getLogger(__name__)

python/sglang/srt/entrypoints/openai/serving_completions.py

@@ -5,7 +5,6 @@ from typing import Any, AsyncGenerator, Dict, List, Optional, Union
 from fastapi import Request
 from fastapi.responses import ORJSONResponse, StreamingResponse
 
-from sglang.srt.code_completion_parser import generate_completion_prompt_from_request
 from sglang.srt.entrypoints.openai.protocol import (
     CompletionRequest,
     CompletionResponse,
@@ -23,6 +22,9 @@ from sglang.srt.entrypoints.openai.utils import (
 from sglang.srt.managers.io_struct import GenerateReqInput
 from sglang.srt.managers.template_manager import TemplateManager
 from sglang.srt.managers.tokenizer_manager import TokenizerManager
+from sglang.srt.parser.code_completion_parser import (
+    generate_completion_prompt_from_request,
+)
 from sglang.utils import convert_json_schema_to_str
 
 logger = logging.getLogger(__name__)

python/sglang/srt/entrypoints/openai/serving_embedding.py

@@ -3,7 +3,6 @@ from typing import Any, Dict, List, Optional, Union
 from fastapi import Request
 from fastapi.responses import ORJSONResponse
 
-from sglang.srt.conversation import generate_embedding_convs
 from sglang.srt.entrypoints.openai.protocol import (
     EmbeddingObject,
     EmbeddingRequest,
@@ -16,6 +15,7 @@ from sglang.srt.entrypoints.openai.serving_base import OpenAIServingBase
 from sglang.srt.managers.io_struct import EmbeddingReqInput
 from sglang.srt.managers.template_manager import TemplateManager
 from sglang.srt.managers.tokenizer_manager import TokenizerManager
+from sglang.srt.parser.conversation import generate_embedding_convs
 
 
 class OpenAIServingEmbedding(OpenAIServingBase):

python/sglang/srt/entrypoints/openai/serving_responses.py

@@ -56,7 +56,7 @@ from sglang.srt.entrypoints.openai.tool_server import MCPToolServer, ToolServer
 from sglang.srt.managers.io_struct import GenerateReqInput
 from sglang.srt.managers.template_manager import TemplateManager
 from sglang.srt.managers.tokenizer_manager import TokenizerManager
-from sglang.srt.reasoning_parser import ReasoningParser
+from sglang.srt.parser.reasoning_parser import ReasoningParser
 from sglang.srt.utils import random_uuid
 
 logger = logging.getLogger(__name__)

python/sglang/srt/function_call/gpt_oss_detector.py

@@ -10,7 +10,7 @@ from sglang.srt.function_call.core_types import (
     ToolCallItem,
     _GetInfoFunc,
 )
-from sglang.srt.harmony_parser import HarmonyParser
+from sglang.srt.parser.harmony_parser import HarmonyParser
 
 logger = logging.getLogger(__name__)
 

python/sglang/srt/layers/moe/utils.py

@@ -162,7 +162,6 @@ def get_deepep_config() -> str:
 def is_tbo_enabled() -> bool:
     global IS_TBO_ENABLED
     if IS_TBO_ENABLED is None:
-        logger.warning("IS_TBO_ENABLED is not initialized, using False")
         IS_TBO_ENABLED = False
     return IS_TBO_ENABLED
 

python/sglang/srt/managers/scheduler.py

@@ -141,7 +141,7 @@ from sglang.srt.mem_cache.lora_radix_cache import LoRARadixCache
 from sglang.srt.mem_cache.radix_cache import RadixCache
 from sglang.srt.mem_cache.swa_radix_cache import SWARadixCache
 from sglang.srt.model_executor.forward_batch_info import ForwardMode, PPProxyTensors
-from sglang.srt.reasoning_parser import ReasoningParser
+from sglang.srt.parser.reasoning_parser import ReasoningParser
 from sglang.srt.server_args import PortArgs, ServerArgs
 from sglang.srt.speculative.spec_info import SpeculativeAlgorithm
 from sglang.srt.torch_memory_saver_adapter import TorchMemorySaverAdapter

python/sglang/srt/managers/template_manager.py

@@ -24,20 +24,20 @@ import os
 import re
 from typing import Optional
 
-from sglang.srt.code_completion_parser import (
+from sglang.srt.parser.code_completion_parser import (
     CompletionTemplate,
     FimPosition,
     completion_template_exists,
     register_completion_template,
 )
-from sglang.srt.conversation import (
+from sglang.srt.parser.conversation import (
     Conversation,
     SeparatorStyle,
     chat_template_exists,
     get_conv_template_by_model_path,
     register_conv_template,
 )
-from sglang.srt.jinja_template_utils import detect_jinja_template_content_format
+from sglang.srt.parser.jinja_template_utils import detect_jinja_template_content_format
 
 logger = logging.getLogger(__name__)
 

python/sglang/srt/model_executor/model_runner.py

@@ -1655,7 +1655,7 @@ class ModelRunner:
 
     def apply_torch_tp(self):
         logger.info(f"Enabling torch tensor parallelism on {self.tp_size} devices.")
-        from sglang.srt.model_parallel import tensor_parallel
+        from sglang.srt.layers.model_parallel import tensor_parallel
 
         device_mesh = torch.distributed.init_device_mesh(self.device, (self.tp_size,))
         tensor_parallel(self.model, device_mesh)

python/sglang/srt/models/torch_native_llama.py

@@ -22,7 +22,7 @@ Reference: https://pytorch.org/docs/stable/distributed.tensor.parallel.html
 
 Here is a quick example to enable TP:
 ```python
-from sglang.srt.model_parallel import tensor_parallel
+from sglang.srt.layers.model_parallel import tensor_parallel
 
 device_mesh = torch.distributed.init_device_mesh("cuda", (tp_size,))
 tensor_parallel(model, device_mesh)