deepanything 0.1.2__tar.gz → 0.1.4__tar.gz

{deepanything-0.1.2 → deepanything-0.1.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: deepanything
-Version: 0.1.2
+Version: 0.1.4
 Summary: DeepAnything is a project that provides DeepSeek R1's deep thinking capabilities for various large language models (LLMs).
 Author: Junity
 Author-email: 1727636624@qq.com
@@ -140,6 +140,8 @@ async def main():
 asyncio.run(main())
 ```
 
+More example can be find in [examples](examples).
+
 ### 2. Use as a Server
 
 ```bash

{deepanything-0.1.2 → deepanything-0.1.4}/README.md

@@ -122,6 +122,8 @@ async def main():
 asyncio.run(main())
 ```
 
+More example can be find in [examples](examples).
+
 ### 2. Use as a Server
 
 ```bash

{deepanything-0.1.2 → deepanything-0.1.4}/deepanything/DeepAnythingClient.py

@@ -45,15 +45,19 @@ def _build_message(
         role="assistant",
         content=reason_prompt.format(reason_content)
     )]
-def _proecess_reason_chunk(chunk, reasoning_contents, reason_usage,show_model, created, _id):
-    delta = chunk.choices[0].delta
-    reasoning_contents.append(delta.reasoning_content)
+def _process_reason_chunk(chunk, reasoning_contents, reason_usage, show_model, created, _id):
     new_chunk = chunk.model_copy(deep=False)
     new_chunk.model = show_model
     new_chunk.created = created
     new_chunk.id = _id
+
     if new_chunk.usage is not None:
         reason_usage = new_chunk.usage
+
+    if chunk.choices:
+        delta = chunk.choices[0].delta
+        reasoning_contents.append(delta.reasoning_content)
+
     return new_chunk, reason_usage
 
 def _process_response_chunk(chunk, reason_usage,show_model, created, _id):
@@ -116,6 +120,8 @@ def chat_completion(
 
     if max_tokens is not None:
         max_tokens -= reason_chat_completion.usage.completion_tokens
+        if max_tokens <= 0:
+            return reason_chat_completion
         response_args["max_tokens"] = max_tokens
 
     response_chat_completion: ChatCompletion = response_client.chat_completions(
@@ -166,11 +172,13 @@ def chat_completion_stream(
         reason_usage = make_usage(0, 0, 0)
 
         for chunk in reason_stream:
-            new_chunk, reason_usage = _proecess_reason_chunk(chunk, reasoning_contents, reason_usage,show_model,created,_id)
+            new_chunk, reason_usage = _process_reason_chunk(chunk, reasoning_contents, reason_usage, show_model, created, _id)
             yield new_chunk
 
         if max_tokens is not None:
             max_tokens -= reason_usage.completion_tokens
+            if max_tokens <= 0:
+                return
             response_args["max_tokens"] = max_tokens
 
         new_messages = _build_message(messages,reason_content="".join(reasoning_contents),reason_prompt=reason_prompt)
@@ -219,7 +227,8 @@ async def chat_completion_async(
             response_args=response_args,
             created=created,
             _id=_id,
-            reason_prompt=reason_prompt
+            reason_prompt=reason_prompt,
+            max_tokens=max_tokens
         )
 
     if max_tokens is not None:
@@ -233,6 +242,8 @@ async def chat_completion_async(
 
     if max_tokens is not None:
         max_tokens -= reason_chat_completion.usage.completion_tokens
+        if max_tokens <= 0:
+            return reason_chat_completion
         response_args["max_tokens"] = max_tokens
 
     response_chat_completion:ChatCompletion = await response_client.chat_completions(
@@ -283,13 +294,15 @@ async def chat_completion_stream_async(
         reason_usage = make_usage(0,0,0)
 
         async for chunk in reason_stream:
-            new_chunk,reason_usage = _proecess_reason_chunk(chunk, reasoning_contents, reason_usage,show_model,created,_id)
+            new_chunk,reason_usage = _process_reason_chunk(chunk, reasoning_contents, reason_usage, show_model, created, _id)
             yield new_chunk
 
         new_messages = _build_message(messages,reason_content="".join(reasoning_contents),reason_prompt=reason_prompt)
 
         if max_tokens is not None:
             max_tokens -= reason_usage.completion_tokens
+            if max_tokens <= 0:
+                return
             response_args["max_tokens"] = max_tokens
 
         response_stream = await response_client.chat_completions_stream(

{deepanything-0.1.2 → deepanything-0.1.4}/deepanything/ReasonClient.py

@@ -76,11 +76,14 @@ class DeepseekReasonClient(ReasonClient):
             messages=messages,
             model=model,
             stream=True,
+            stream_options = {"include_usage": True},
             **kwargs
         )
 
         def _iter():
             for chunk in stream:
+                if len(chunk.choices) == 0:
+                    yield chunk
                 if chunk.choices[0].delta.reasoning_content is not None:
                     yield chunk
                 else:
@@ -111,11 +114,15 @@ class AsyncDeepseekReasonClient(AsyncReasonClient):
             messages=messages,
             model=model,
             stream=True,
+            stream_options = {"include_usage": True},
             **kwargs
         )
 
         async def _iter():
             async for chunk in stream:
+                if len(chunk.choices) == 0:
+                    yield chunk
+                    continue
                 if chunk.choices[0].delta.reasoning_content is not None:
                     yield chunk
                 else:
@@ -130,8 +137,9 @@ class AsyncDeepseekReasonClient(AsyncReasonClient):
 def _rebuild_chunk_for_openai(
         chunk:chat_completion_chunk.ChatCompletionChunk
 ) -> chat_completion_chunk.ChatCompletionChunk:
-    chunk.choices[0].delta.reasoning_content = chunk.choices[0].delta.content
-    chunk.choices[0].delta.content = None
+    if len(chunk.choices):
+        chunk.choices[0].delta.reasoning_content = chunk.choices[0].delta.content
+        chunk.choices[0].delta.content = None
     return chunk
 
 
@@ -159,6 +167,7 @@ class OpenaiReasonClient(ReasonClient):
             messages=messages,
             model=model,
             stream=True,
+            stream_options = {"include_usage": True},
             **kwargs
         )
 
@@ -205,6 +214,7 @@ class AsyncOpenaiReasonClient(AsyncReasonClient):
             messages=messages,
             model=model,
             stream=True,
+            stream_options = {"include_usage": True},
             **kwargs
         )
 

{deepanything-0.1.2 → deepanything-0.1.4}/deepanything/ResponseClient.py

@@ -55,6 +55,7 @@ class OpenaiResponseClient(ResponseClient):
                 messages=messages,
                 model=model,
                 stream=True,
+                stream_options = {"include_usage": True},
                 **kwargs
             )
 
@@ -83,5 +84,6 @@ class AsyncOpenaiResponseClient(AsyncResponseClient):
                 messages=messages,
                 model=model,
                 stream=True,
+                stream_options = {"include_usage": True},
                 **kwargs
             )

{deepanything-0.1.2 → deepanything-0.1.4}/deepanything/Server/Server.py

@@ -1,7 +1,8 @@
 from dataclasses import dataclass
 import time
 import uvicorn
-from typing import Dict, List, Optional
+from typing import Dict, List, Optional, Any
+import json
 
 from openai.types.model import Model as OpenaiModel
 from fastapi import FastAPI,Depends, HTTPException, status,Header
@@ -21,8 +22,8 @@ class ModelInfo:
     reason_model : str
     response_client : str
     response_model : str
-    created : int
-    reason_prompt : str
+    created : int = int(time.time())
+    reason_prompt : str = "<Think>{}</Think>"
 
 class DeepAnythingServer:
     app : FastAPI = FastAPI()
@@ -35,9 +36,12 @@ class DeepAnythingServer:
     api_keys : List[str] = []
     security = HTTPBearer()
 
-    def __init__(self, host:str = None, port:int = None,config_object : Dict = None):
-        if config_object is not None:
-            self.load_config(config_object)
+    def __init__(self, host:str = None, port:int = None, config : Any or str = None):
+        if config is not None:
+            if isinstance(config,str):
+                with open(config) as f:
+                    config = json.load(f)
+            self.load_config(config)
 
         if host:
             self.host = host
@@ -50,6 +54,13 @@ class DeepAnythingServer:
     def run(self):
         uvicorn.run(self.app,host=self.host,port=self.port,log_level="trace")
 
+    @staticmethod
+    def _extract_args(query : Types.ChatCompletionQuery) -> dict:
+        args = query.dict()
+        for key in ["messages","model","stream"]:
+            args.pop(key)
+        return args
+
     def load_config(self,config_object : Dict) -> None:
         self.host = config_object.get("host","0.0.0.0")
         self.port = config_object.get("port",8000)
@@ -106,8 +117,17 @@ class DeepAnythingServer:
 
         self.api_keys = config_object.get("api_keys",[])
 
+
+    def add_reason_client(self,name:str,client:AsyncReasonClient):
+        self.reason_clients[name] = client
+
+    def add_response_client(self,name:str,client:AsyncResponseClient):
+        self.response_clients[name] = client
+
+    def add_model(self,name:str,model:ModelInfo):
+        self.models[name] = model
     def verify_authorization(self, authorization:Optional[str]):
-        if self.api_keys == []:
+        if not self.api_keys:
             return
 
         if authorization is None:
@@ -140,6 +160,13 @@ class DeepAnythingServer:
                 yield f"data: {chunk.model_dump_json(indent=None)}\n\n"
             yield "data: [DONE]"
 
+        args = DeepAnythingServer._extract_args(query)
+
+        max_tokens = None
+        if "max_tokens" in args:
+            max_tokens = args["max_tokens"]
+            args.pop("max_tokens")
+
         if query.stream:
             res = sse(
                 await chat_completion_stream_async(
@@ -150,6 +177,9 @@ class DeepAnythingServer:
                     response_model=model.response_model,
                     show_model=model.name,
                     reason_prompt=model.reason_prompt,
+                    response_args=args,
+                    reason_args=args,
+                    max_tokens=max_tokens
                 )
             )
             return StreamingResponse(
@@ -164,11 +194,14 @@ class DeepAnythingServer:
                     response_client=self.response_clients[model.response_client],
                     response_model=model.response_model,
                     show_model=model.name,
-                    reason_prompt=model.reason_prompt
+                    reason_prompt=model.reason_prompt,
+                    response_args=args,
+                    reason_args=args,
+                    max_tokens=max_tokens
             )
             return res.model_dump_json()
 
-    async def get_models(self) -> Types.ModelsListResponse:
+    def get_models(self) -> Types.ModelsListResponse:
         return Types.ModelsListResponse(
             data = [OpenaiModel(
                     id = model_info.name,

deepanything-0.1.4/deepanything/Server/Types.py

@@ -0,0 +1,22 @@
+from pydantic import BaseModel
+from openai.types.model import Model as OpenaiModel
+from typing import Dict, List, Optional
+
+
+class ModelsListResponse(BaseModel):
+    data : List[OpenaiModel]
+
+class ChatCompletionMessage(BaseModel):
+    role : str
+    content : str
+
+class ChatCompletionQuery(BaseModel):
+    model : str
+    messages : List[ChatCompletionMessage]
+    stream : bool = False
+    temperature : Optional[float] = 0.7
+    top_p : Optional[float] = 1.0
+    frequency_penalty : Optional[float] = 0.0
+    presence_penalty : Optional[float] = 0.0
+    max_tokens : Optional[int] = None
+    stop : Optional[List[str]] = None

{deepanything-0.1.2 → deepanything-0.1.4}/deepanything/Utility.py

@@ -168,7 +168,7 @@ def merge_chunk(
         model: str,
         created: int = int(time.time())
 ) -> chat_completion.ChatCompletion:
-    """同步合并 chunk 流"""
+
     reasoning_content = ""
     content = ""
     comp_tokens = prompt_tokens = total_tokens = 0

{deepanything-0.1.2 → deepanything-0.1.4}/deepanything/__main__.py

@@ -12,7 +12,7 @@ def main():
     if args.config is not None:
         with open(args.config) as f:
             config = json.load(f)
-        server = DeepAnythingServer(host=args.host,port=args.port,config_object=config)
+        server = DeepAnythingServer(host=args.host, port=args.port, config=config)
         server.run()
     else:
         print("No config file specified.")

{deepanything-0.1.2 → deepanything-0.1.4}/deepanything.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: deepanything
-Version: 0.1.2
+Version: 0.1.4
 Summary: DeepAnything is a project that provides DeepSeek R1's deep thinking capabilities for various large language models (LLMs).
 Author: Junity
 Author-email: 1727636624@qq.com
@@ -140,6 +140,8 @@ async def main():
 asyncio.run(main())
 ```
 
+More example can be find in [examples](examples).
+
 ### 2. Use as a Server
 
 ```bash

{deepanything-0.1.2 → deepanything-0.1.4}/setup.py

@@ -8,7 +8,7 @@ with open("requirements.txt") as f:
 
 setup(
     name="deepanything",
-    version="0.1.2",
+    version="0.1.4",
     author="Junity",
     author_email="1727636624@qq.com",
     description="DeepAnything is a project that provides DeepSeek R1's deep thinking capabilities for various large language models (LLMs).",

{deepanything-0.1.2 → deepanything-0.1.4}/test/server.py

@@ -4,5 +4,5 @@ import json
 with open("test/config.json") as f:
     conf = json.load(f)
 
-server = DeepAnythingServer(config_object=conf)
+server = DeepAnythingServer(config=conf)
 server.run()

deepanything-0.1.2/deepanything/Server/Types.py

@@ -1,15 +0,0 @@
-from pydantic import BaseModel
-from openai.types.model import Model as OpenaiModel
-from typing import Dict,List
-
-class ModelsListResponse(BaseModel):
-    data : List[OpenaiModel]
-
-class ChatCompletionMessage(BaseModel):
-    role : str
-    content : str
-
-class ChatCompletionQuery(BaseModel):
-    model : str
-    messages : List[ChatCompletionMessage]
-    stream : bool = False