finished with api

Dockerfile

@@ -0,0 +1,13 @@
+FROM python:3.12.7
+
+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", "main:app", "--host", "0.0.0.0", "--port", "7860"]

app/api/chat/__init__.py

@@ -1,7 +1,7 @@
 from fastapi.routing import APIRouter
 
 chat_router = APIRouter(
-    prefix="/chat", tags=["chat"]
+    prefix="/api/chat", tags=["chat"]
 )
 
 from . import views

app/api/chat/dto.py

@@ -0,0 +1,8 @@
+from enum import Enum
+
+
+class ModelType(Enum):
+    gpt_4o = "gpt-4o"
+    gpt_4o_mini = "gpt-4o-mini"
+    o1_mini = "o1-mini"
+    o1_preview = "o1-preview"

app/api/chat/model.py

@@ -3,7 +3,7 @@ from datetime import datetime
 from pydantic import Field
 
 from app.api.account.model import AccountModel
-from app.api.common.dto import ModelType
+from app.api.chat.dto import ModelType
 from app.core.database import MongoBaseModel
 
 

app/api/chat/schemas.py

@@ -1,7 +1,8 @@
 from pydantic import BaseModel
 
+from app.api.chat.dto import ModelType
 from app.api.chat.model import ChatModel
-from app.api.common.dto import Paging, ModelType
+from app.api.common.dto import Paging
 from app.core.wrappers import HectoolResponseWrapper
 
 

app/api/common/dto.py

@@ -6,11 +6,4 @@ from pydantic import BaseModel
 class Paging(BaseModel):
     pageSize: int
     pageIndex: int
-    totalCount: int
-
-
-class ModelType(Enum):
-    gpt_4o = "gpt-4o"
-    gpt_4o_mini = "gpt-4o-mini"
-    o1_mini = "o1-mini"
-    o1_preview = "o1-preview"
+    totalCount: int

app/api/message/__init__.py

@@ -1,7 +1,7 @@
 from fastapi.routing import APIRouter
 
 message_router = APIRouter(
-    prefix="/message", tags=["message"]
+    prefix="/api/message", tags=["message"]
 )
 
 from . import views

app/api/message/ai/openai_request.py

@@ -0,0 +1,33 @@
+from app.api.chat.model import ChatModel
+from app.api.message.dto import Author
+from app.api.message.model import MessageModel
+from app.api.message.schemas import MessageWrapper
+from app.core.config import settings
+
+
+async def response_generator(chat: ChatModel, messages: list):
+    completion = await settings.OPENAI_CLIENT.chat.completions.create(
+        model=chat.model.value,
+        messages=messages,
+        n=1,
+        temperature=0.4,
+        stream=True
+    )
+
+    message_obj, full_response = None, ''
+
+    async for chunk in completion:
+        chunk_text = chunk.choices[0].delta.content
+        if chunk_text:
+            full_response += chunk_text
+            message_obj = MessageModel(
+                chatId=chat.id,
+                author=Author.Assistant,
+                text=chunk_text,
+            )
+            yield MessageWrapper(data=message_obj).model_dump_json()
+
+    message_obj.text = full_response
+    await settings.DB_CLIENT.messages.insert_one(message_obj.to_mongo())
+
+

app/api/message/ai/prompts.py

@@ -0,0 +1,2 @@
+class Prompts:
+    generate_response = """You must always Return 'Hello world!'"""

app/api/message/ai/utils.py

@@ -0,0 +1,30 @@
+from app.api.message.ai.prompts import Prompts
+from app.api.message.model import MessageModel
+
+
+def transform_messages_to_openai(messages: list[MessageModel]) -> list[dict]:
+    openai_messages = [{"role": "system", "content": Prompts.generate_response}]
+    for message in messages:
+
+        if message.file:
+            content = [
+                {
+                    "type": "text", "text": message.text
+                },
+                {
+                    "type": "image_url",
+                    "image_url": {
+                        "url": f"data:image/jpeg;base64,{message.file.base64String}",
+                        "detail": "low"
+                    }
+                },
+            ]
+        else:
+            content = message.text
+
+        openai_messages.append({
+            "role": message.author.value,
+            "content": content
+        })
+
+    return openai_messages

app/api/message/db_requests.py

@@ -4,6 +4,7 @@ from fastapi import HTTPException
 
 from app.api.account.model import AccountModel
 from app.api.chat.model import ChatModel
+from app.api.message.dto import Author
 from app.api.message.model import MessageModel
 from app.api.message.schemas import CreateMessageRequest
 from app.core.config import settings
@@ -29,13 +30,8 @@ async def get_all_chat_messages_obj(
 
 async def create_message_obj(
         chat_id: str, message_data: CreateMessageRequest, account: AccountModel
-) -> MessageModel:
-    message = MessageModel(**message_data.model_dump(), chatId=chat_id)
-    _, chat = await asyncio.gather(
-        settings.DB_CLIENT.insert_one(message.to_mongo()),
-        settings.DB_CLIENT.chats.find_one({"id": chat_id})
-    )
-
+) -> tuple[MessageModel, ChatModel]:
+    chat = await settings.DB_CLIENT.chats.find_one({"id": chat_id})
     if not chat:
         raise HTTPException(status_code=404, detail="Chat not found")
 
@@ -43,4 +39,7 @@ async def create_message_obj(
     if account and chat.account != account:
         raise HTTPException(status_code=403, detail="Chat account not match")
 
-    return message
+    message = MessageModel(**message_data.model_dump(), chatId=chat_id, author=Author.User)
+    await settings.DB_CLIENT.messages.insert_one(message.to_mongo())
+
+    return message, chat

app/api/message/dto.py

@@ -1,6 +1,6 @@
 from enum import Enum
 
-from openai import BaseModel
+from pydantic import BaseModel
 
 
 class Author(Enum):

app/api/message/schemas.py

@@ -1,24 +1,23 @@
 from pydantic import BaseModel
 
-from app.api.chat.model import ChatModel
 from app.api.common.dto import Paging
 from app.api.message.dto import Author, File
+from app.api.message.model import MessageModel
 from app.core.wrappers import HectoolResponseWrapper
 
 
 class CreateMessageRequest(BaseModel):
-    author: Author
     text: str
     file: File | None = None
 
 
-class MessageWrapper(HectoolResponseWrapper[ChatModel]):
+class MessageWrapper(HectoolResponseWrapper[MessageModel]):
     pass
 
 
 class AllMessageResponse(BaseModel):
     paging: Paging
-    data: list[ChatModel]
+    data: list[MessageModel]
 
 
 class AllMessageWrapper(HectoolResponseWrapper[AllMessageResponse]):

app/api/message/views.py

@@ -1,10 +1,13 @@
 from fastapi.params import Depends
+from starlette.responses import StreamingResponse
 
 from app.api.account.model import AccountModel
 from app.api.common.dto import Paging
 from app.api.message import message_router
 from app.api.message.db_requests import get_all_chat_messages_obj, create_message_obj
-from app.api.message.schemas import AllMessageWrapper, MessageWrapper, AllMessageResponse, CreateMessageRequest
+from app.api.message.ai.openai_request import response_generator
+from app.api.message.schemas import AllMessageWrapper, AllMessageResponse, CreateMessageRequest
+from app.api.message.ai.utils import transform_messages_to_openai
 from app.core.security import PermissionDependency
 
 
@@ -25,6 +28,8 @@ async def create_message(
         chatId: str,
         message_data: CreateMessageRequest,
         account: AccountModel = Depends(PermissionDependency(is_public=True))
-) -> MessageWrapper:
-    message = await create_message_obj(chatId, message_data, account)
-    return MessageWrapper(data=message)
+) -> StreamingResponse:
+    _, chat = await create_message_obj(chatId, message_data, account)
+    messages = await get_all_chat_messages_obj(chatId, account)
+    openai_messages = transform_messages_to_openai(messages)
+    return StreamingResponse(response_generator(chat, openai_messages), media_type='application/json')

app/core/config.py

@@ -11,13 +11,7 @@ load_dotenv()
 class BaseConfig:
     BASE_DIR: pathlib.Path = pathlib.Path(__file__).parent.parent
     SECRET_KEY = os.getenv('SECRET')
-    DB_CLIENT = motor.motor_asyncio.AsyncIOMotorClient(
-        'mongodb://'
-        f'{os.getenv("DB_USER")}:'
-        f'{os.getenv("DB_PASSWORD")}@'
-        f'{os.getenv("DB_HOST")}:'
-        f'{os.getenv("DB_PORT")}?authSource=hectool'
-    ).hectool
+    DB_CLIENT = motor.motor_asyncio.AsyncIOMotorClient(os.getenv("MONGO_DB_URL")).hectool
     OPENAI_CLIENT = AsyncClient(api_key=os.getenv('OPENAI_API_KEY'))
 
 

app/core/wrappers.py

@@ -5,7 +5,6 @@ from fastapi import HTTPException
 from pydantic import BaseModel
 from starlette.responses import JSONResponse
 
-from app.api.common.dto import ModelType
 from app.core.config import settings
 
 T = TypeVar('T')
@@ -43,28 +42,3 @@ def exception_wrapper(http_error: int, error_message: str):
         return wrapper
 
     return decorator
-
-def openai_wrapper(
-        temperature: int | float = 0,
-        model: ModelType = ModelType.gpt_4o_mini,
-        is_json: bool = False
-):
-    def decorator(func):
-        @wraps(func)
-        async def wrapper(*args, **kwargs) -> str:
-            messages = await func(*args, **kwargs)
-            completion = await settings.OPENAI_CLIENT.chat.completions.create(
-                messages=messages,
-                temperature=temperature,
-                n=1,
-                model=model.value,
-                response_format={"type": "json_object"} if is_json else {"type": "text"}
-            )
-            response = completion.choices[0].message.content
-            if is_json:
-                response = json.loads(response)
-            return response
-
-        return wrapper
-
-    return decorator