111

Dockerfile

@@ -0,0 +1,16 @@
+# Read the doc: https://huggingface.co/docs/hub/spaces-sdks-docker
+# you will also find guides on how best to write your Dockerfile
+
+FROM python:3.11
+
+RUN useradd -m -u 1000 user
+USER user
+ENV PATH="/home/user/.local/bin:$PATH"
+
+WORKDIR /app
+
+COPY --chown=user ./requirements.txt requirements.txt
+RUN pip install --no-cache-dir --upgrade -r requirements.txt
+
+COPY --chown=user . /app
+CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]

app.py

@@ -0,0 +1,14 @@
+# uvicorn app:app --host 0.0.0.0 --port 7860
+
+from fastapi import FastAPI
+from routers import v1, v2, openai
+
+app = FastAPI()
+
+app.include_router(v1.router, prefix="/v1")
+app.include_router(v2.router, prefix="/v2")
+app.include_router(openai.router, prefix="/openai")
+
+@app.get("/")
+def greet_json():
+    return {"Hello": "World!"}

requirements.txt

@@ -0,0 +1,3 @@
+fastapi
+uvicorn[standard]
+requests

routers/__init__.py

@@ -0,0 +1 @@
+from . import v1, v2, openai

routers/openai.py

@@ -0,0 +1,273 @@
+from fastapi import APIRouter, Request, HTTPException
+from pydantic import BaseModel
+from typing import List, Optional
+
+router = APIRouter()
+
+# 定义请求和响应的数据模型
+
+class ChatCompletionMessage(BaseModel):
+    """
+    聊天完成消息的数据模型
+    role: 消息的角色，例如 "user" 或 "assistant"
+    content: 消息的内容
+    """
+    role: str
+    content: str
+
+class ChatCompletionRequest(BaseModel):
+    """
+    聊天完成请求的数据模型
+    model: 使用的模型名称
+    messages: 消息列表，包含用户的输入和助手的回复
+    max_tokens: (可选) 生成的最大 token 数
+    temperature: (可选) 控制生成文本的随机性
+    """
+    model: str
+    messages: List[ChatCompletionMessage]
+    max_tokens: Optional[int] = None
+    temperature: Optional[float] = None
+
+class ChatCompletionChoice(BaseModel):
+    """
+    聊天完成选项的数据模型
+    index: 选项的索引
+    message: 聊天完成消息
+    finish_reason: 完成的原因，例如 "stop"
+    """
+    index: int
+    message: ChatCompletionMessage
+    finish_reason: str
+
+class ChatCompletionUsage(BaseModel):
+    """
+    聊天完成用量的数据模型
+    prompt_tokens: prompt 使用的 token 数
+    completion_tokens: 完成生成的 token 数
+    total_tokens: 总共使用的 token 数
+    """
+    prompt_tokens: int
+    completion_tokens: int
+    total_tokens: int
+
+class ChatCompletionResponse(BaseModel):
+    """
+    聊天完成响应的数据模型
+    id: 响应的 ID
+    object: 响应的对象类型，例如 "chat.completion"
+    created: 响应的创建时间戳
+    model: 使用的模型名称
+    choices: 选项列表
+    usage: 用量信息
+    """
+    id: str
+    object: str
+    created: int
+    model: str
+    choices: List[ChatCompletionChoice]
+    usage: ChatCompletionUsage
+
+class CompletionRequest(BaseModel):
+    """
+    文本生成请求的数据模型
+    model: 使用的模型名称
+    prompt: 输入的 prompt
+    max_tokens: (可选) 生成的最大 token 数
+    temperature: (可选) 控制生成文本的随机性
+    """
+    model: str
+    prompt: str
+    max_tokens: Optional[int] = None
+    temperature: Optional[float] = None
+
+class CompletionChoice(BaseModel):
+    """
+    文本生成选项的数据模型
+    index: 选项的索引
+    text: 生成的文本
+    finish_reason: 完成的原因，例如 "stop"
+    """
+    index: int
+    text: str
+    finish_reason: str
+
+class CompletionUsage(BaseModel):
+    """
+    文本生成用量的数据模型
+    prompt_tokens: prompt 使用的 token 数
+    completion_tokens: 完成生成的 token 数
+    total_tokens: 总共使用的 token 数
+    """
+    prompt_tokens: int
+    completion_tokens: int
+    total_tokens: int
+
+class CompletionResponse(BaseModel):
+    """
+    文本生成响应的数据模型
+    id: 响应的 ID
+    object: 响应的对象类型，例如 "text_completion"
+    created: 响应的创建时间戳
+    model: 使用的模型名称
+    choices: 选项列表
+    usage: 用量信息
+    """
+    id: str
+    object: str
+    created: int
+    model: str
+    choices: List[CompletionChoice]
+    usage: CompletionUsage
+
+class EmbeddingRequest(BaseModel):
+    """
+    Embedding 请求的数据模型
+    model: 使用的模型名称
+    input: 输入的文本
+    """
+    model: str
+    input: str
+
+class EmbeddingData(BaseModel):
+    """
+    Embedding 数据的数据模型
+    index: 数据的索引
+    embedding: Embedding 向量
+    """
+    index: int
+    embedding: List[float]
+
+class EmbeddingResponse(BaseModel):
+    """
+    Embedding 响应的数据模型
+    object: 响应的对象类型，例如 "list"
+    data: Embedding 数据列表
+    model: 使用的模型名称
+    usage: 用量信息，例如 {"prompt_tokens": 8, "total_tokens": 8}
+    """
+    object: str
+    data: List[EmbeddingData]
+    model: str
+    usage: dict  # Example: {"prompt_tokens": 8, "total_tokens": 8}
+
+class RerankRequest(BaseModel):
+    """
+    Rerank 请求的数据模型
+    model: 使用的模型名称
+    query: 查询文本
+    documents: 文档列表
+    max_chunks: (可选) 最大 chunk 数
+    """
+    model: str
+    query: str
+    documents: List[str]
+    max_chunks: Optional[int] = None
+
+class RerankResponse(BaseModel):
+    """
+    Rerank 响应的数据模型
+    model: 使用的模型名称
+    results: Rerank 结果列表，例如 [{"index": 0, "relevance_score": 0.9}]
+    """
+    model: str
+    results: List[dict] # Example: [{"index": 0, "relevance_score": 0.9}]
+
+
+# 实现 endpoint
+
+@router.post("/chat/completions")
+async def chat_completions(request: ChatCompletionRequest):
+    """
+    聊天完成 endpoint
+    接收 ChatCompletionRequest，返回 ChatCompletionResponse
+    """
+    # Implement chat completion logic here
+    # This is a placeholder, replace with actual implementation
+    print(f"Received chat completion request: {request}")
+    if not isinstance(request, ChatCompletionRequest):
+        raise HTTPException(status_code=400, detail="Invalid request format")
+
+    response = ChatCompletionResponse(
+        id="chatcmpl-xxxxxxxxxxxxxxxxxxxxxxxx",
+        object="chat.completion",
+        created=1629750000,
+        model=request.model,
+        choices=[
+            ChatCompletionChoice(
+                index=0,
+                message=ChatCompletionMessage(
+                    role="assistant",
+                    content="This is a dummy response for chat completion.",
+                ),
+                finish_reason="stop",
+            )
+        ],
+        usage=ChatCompletionUsage(
+            prompt_tokens=10, completion_tokens=20, total_tokens=30
+        ),
+    )
+    return response
+
+@router.post("/completions")
+async def completions(request: CompletionRequest):
+    """
+    文本生成 endpoint
+    接收 CompletionRequest，返回 CompletionResponse
+    """
+    # Implement text generation logic here
+    # This is a placeholder, replace with actual implementation
+    print(f"Received completion request: {request}")
+    response = CompletionResponse(
+        id="cmpl-xxxxxxxxxxxxxxxxxxxxxxxx",
+        object="text_completion",
+        created=1629750000,
+        model=request.model,
+        choices=[
+            CompletionChoice(
+                index=0,
+                text="This is a dummy response for text completion.",
+                finish_reason="stop",
+            )
+        ],
+        usage=CompletionUsage(
+            prompt_tokens=10, completion_tokens=20, total_tokens=30
+        ),
+    )
+    return response
+
+@router.post("/embeddings")
+async def embeddings(request: EmbeddingRequest):
+    """
+    Embedding endpoint
+    接收 EmbeddingRequest，返回 EmbeddingResponse
+    """
+    # Implement embedding logic here
+    # This is a placeholder, replace with actual implementation
+    print(f"Received embedding request: {request}")
+    response = EmbeddingResponse(
+        object="list",
+        data=[
+            EmbeddingData(
+                index=0,
+                embedding=[0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0],
+            )
+        ],
+        model=request.model,
+        usage={"prompt_tokens": 8, "total_tokens": 8},
+    )
+    return response
+
+@router.post("/rerank")
+async def rerank(request: RerankRequest):
+    """
+    Rerank endpoint
+    接收 RerankRequest，返回 RerankResponse
+    """
+    # Implement rerank logic here
+    # This is a placeholder, replace with actual implementation
+    print(f"Received rerank request: {request}")
+    response = RerankResponse(
+        model=request.model,
+        results=[{"index": 0, "relevance_score": 0.9}],
+    )
+    return response

routers/v1.py

@@ -0,0 +1,94 @@
+import requests
+from fastapi import APIRouter, HTTPException, Request
+from pydantic import BaseModel
+from typing import List, Optional
+import random
+import string
+
+router = APIRouter()
+
+def generate_chatcmpl_id():
+    characters = string.ascii_letters + string.digits
+    return "chatcmpl-" + ''.join(random.choice(characters) for i in range(29))
+
+# Define data models for request and response
+
+class ChatCompletionMessage(BaseModel):
+    role: str
+    content: str
+
+class ChatCompletionRequest(BaseModel):
+    model: str
+    messages: List[ChatCompletionMessage]
+    max_tokens: Optional[int] = None
+    temperature: Optional[float] = None
+
+class ChatCompletionChoice(BaseModel):
+    index: int
+    message: ChatCompletionMessage
+    finish_reason: str
+
+class ChatCompletionUsage(BaseModel):
+    prompt_tokens: int
+    completion_tokens: int
+    total_tokens: int
+
+class ChatCompletionResponse(BaseModel):
+    id: str
+    object: str
+    created: int
+    model: str
+    choices: List[ChatCompletionChoice]
+    usage: ChatCompletionUsage
+
+@router.post("/chat/completions")
+async def chat_completions(chat_request: ChatCompletionRequest, request: Request):
+    """
+    聊天完成 endpoint，调用 Gemini API
+    """
+    headers = dict(request.headers)
+    print(headers)
+    return "hh"
+    gemini_api_url = "https://generativelanguage.googleapis.com/v1beta/openai/chat/completions"  # 需要验证 URL 的正确性
+    auth_header = request.headers.get("Authorization")
+    headers = {"Content-Type": "application/json"}
+    if auth_header:
+        headers["Authorization"] = auth_header
+
+    try:
+        response = requests.post(gemini_api_url, headers=headers, json=chat_request.model_dump())
+        response.raise_for_status()  # 检查请求是否成功
+
+        # 将 Gemini API 的响应转换为 ChatCompletionResponse
+        gemini_response = response.json()
+        print(f"Gemini API response: {gemini_response}")  # 打印 Gemini API 的响应
+
+        chat_completion_response = ChatCompletionResponse(
+            id=generate_chatcmpl_id(),
+            object="chat.completion",
+            created=gemini_response.get("created", 1629750000),
+            model=chat_request.model,
+            choices=[
+                ChatCompletionChoice(
+                    index=i,
+                    message=ChatCompletionMessage(
+                        role=choice.get("message", {}).get("role", "assistant"),
+                        content=choice.get("message", {}).get("content", "This is a dummy response from Gemini."),
+                    ),
+                    finish_reason=choice.get("finish_reason", "stop"),
+                )
+                for i, choice in enumerate(gemini_response.get("choices", []))
+            ],
+            usage=ChatCompletionUsage(
+                prompt_tokens=gemini_response.get("usage", {}).get("prompt_tokens", 0),
+                completion_tokens=gemini_response.get("usage", {}).get("completion_tokens", 0),
+                total_tokens=gemini_response.get("usage", {}).get("total_tokens", 0),
+            ),
+        )
+        return chat_completion_response
+    except requests.exceptions.RequestException as e:
+        raise HTTPException(status_code=500, detail=str(e))
+
+@router.get("/")
+def greet_v1():
+    return {"version": "v1", "message": "Hello, World!"}

routers/v2.py

@@ -0,0 +1,7 @@
+from fastapi import APIRouter
+
+router = APIRouter()
+
+@router.get("/")
+def greet_v2():
+    return {"version": "v2", "message": "Hello, World! This is version 2."}

程序开发笔记.md

@@ -0,0 +1,4 @@
+# 程序开发笔记
+
+1. 主文件为 app.py
+2. 以版本作为路由路径，将版本对应的路由放到 routers 文件夹下，routers/v1.py、routers/v2.py