방구석 브리핑_Backend

BIS_Backend_Overview

Implement a functionality using FastAPI to periodically provide clients with briefings on the results of an AI Inference Pipeline. Enable interaction between the client and the server through request and response mechanisms.

Environment Installation

Please refer to the attached "requirements.txt" for the versions of packages and libraries needed for environment installation.

fastapi

pip install fastapi
pip install "uvicorn[standard]"

websocket

pip install websockets

apscheduler

pip install apscheduler

mongodb

pip install pymongo

Install MongoDB following the instructions on the

official MongoDB website

.env.production

pip install python-dotenv

Create a separate file named .env.production and add the following content:

DATABASE_ID="your_id"
DATABASE_PASSWORD="your_pw"
DATABASE_URI=mongodb+srv://<DATABASE_ID>:<DATABASE_PASSWORD>@"your_dbname".keizkxa.mongodb.net

Ensure that you keep this file secure and do not share sensitive information.

This file contains configuration details for your production environment, including database credentials. Keep in mind the importance of securing this information and not exposing it publicly.

API List-up

Index	Method	URI	Description
1	POST	/login	로그인 API
2	WebSocket	/ws	Web socket (Db 변화 감지)
3	WebSocket	/reloadBriefing	Web socket (진행 중 briefing 확인)
4	POST	/endBriefing	Briefing 종료
5	POST	/getAllRequest	모든 request 조회
6	POST	/getBriefing	모든 briefing 조회
7	POST	/aiSummary	브리핑 요약
8	POST	/getCustom	Custom 획득
9	POST	/updateCustom	Custom 갱신

---

API Specification

• login

  • description

    • Utilize check_user in models.py to verify new and existing users.

      • In the case of an existing user, return {"message": "signup", "user_id": <user_id>}.
      • For a new user, return {"message": "login", "user_id": <user_id>}.

Endpoint	Method
POST /login	POST

	Request
Header
"No special headers required"
Body
Parameter	Type
username	String

Response
1) Success Response
Status Code: 200 OK
Body:
json { "message" : "signup", "user_id" : "user_id_value"}
or
json { "message" : "login", "user_id" : "user_id_value"}
2) Error Response
Status Code: 500 Internal Server Error
Body:
json { "detail": "Login Error: error_message"}

• websocket_endpoint

  • Description:
    • Handle WebSocket connections for continuous updates.
    • Establish WebSocket connection.
    • Store new request in the database and register with the scheduler.
    • Continuously send updates to the frontend through the WebSocket.

Endpoint	Method
websocket/ws	WebSocket

Parameters

Name	Type	Description
websocket	WebSocket	The WebSocket connection object.
username	String	User's username.
interval	String	Requested interval for updates.
endtime	String	Requested end time for updates.

• websocket_reloadBriefing

  • Description
    • Display the ongoing briefing.
    • Establish WebSocket connection.
    • Send reloaded briefing data to the frontend.
    • Get the latest request ID and continuously watch the database for updates.

Endpoint	Method
websocket/reloadBriefing	WebSocket

Parameters

Name	Type	Description
websocket	WebSocket	The WebSocket connection object.
username	String	User's username.

• endBriefing

  • description

    • Terminate the ongoing briefing.

    • The function endBriefing uses the endBriefing_worker from utils.py and end_scheduler from scheduler.py.

      • Upon success, it prints a success message and returns a success response with the message "Briefing successfully terminated".
      • In case of failure (e.g., an exception), it catches the "HTTPException", logs the error message, and returns an error response with the corresponding status code and details.

Endpoint	Method
POST /endBriefing	POST

	Request
Header
"No special headers required"
Body
Parameter	Type
username	String

Response
1) Success Response
Status Code: 200 OK
Body:
"message": "브리핑 성공적으로 종료"
2)Error Response
Status Code: 500 Internal Server Error
Body:
Internal Server Error

• getAllRequest

  • description
    • Implement a functionality to retrieve all requests for a specific username from the "request" Collection within the database.
    • getAllRequest utilizes the get_all_request function from utils.py to retrieve and return the list of requests for a specific username from the "request" Collection in the database.

• getBriefing

  • description
    • Retrieve briefing information based on the provided request_id.
    • Utilize the get_briefing function from utils.py to perform a functionality that retrieves all briefings for a specific request_id from the "briefing" Collection in the database.

• aiSummary

  • description

    • Generates an AI summary captions based on the provided request_id.

    • This API utilizes two methods: get_briefing in utils.py for querying all data from the database and chat_summary in utils.py for generating AI summaries based on the returned briefing data list.

    • To provide additional details on chat_summary, it involves using the OpenAI API key to generate messages based on a specified format prompt. This process utilizes the GPT-4-preview model and performs prompt engineering on the briefing data list retrieved earlier, adjusting hyperparameters such as "tone, writing style, temperature, top_p," and others.

• getCustom

  • description
    • this API retrieves the "custom" data for a specific username. The obtained custom_list is then used to generate an HTTP response in JSON format, encoded in UTF-8.
    • The get_custom in models.py process involves searching for the specified username within the "user" collection in the database and referring to it as a custom list.

• updateCustom

  • description
    • Utilizing the update_custom function in models.py, this API updates the "custom" field in the "user" table.
      • If the custom_list is successfully updated, it returns the message "Success to update Custom."
      • In case of exceptions or errors, it raises an HTTPException with a status code of 500 and the detail "Fail to update Custom."