droncogene/chrome-ext-api

An api for storing, merging, fetching and transcribing (using OpenAI whisper API) video recorded in chunks by a chrome extension

HNGX Stage 5 task

Description

A backend api for a chrome extension that functions as a screen recorder. The api processes and stored the video files.

Technologies

• Python 3.11
• FastAPI
• Pymongo
• mongodb on docker
• poetry or pip for dependency management

Installation

• Clone the repo
• Install poetry
• Run poetry install to install dependencies
• Run poetry shell to activate the virtual environment
• Or use pip to install dependencies from requirements.txt: pip install -r requirements.txt
• Set environment variables: see .env.example
• Spin up a mongodb instance on docker: docker run -d -p 27017:27017 --name mongodb mongo:latest
• Spin up rabbitmq instance on docker: docker run -d -p 5672:5672 -p 15672:15672 -p 15692:15692 --name rabbitmq rabbitmq:latest
• Both rabbitmq and mongodb can also be installed manually without docker
• Install ffmpeg: sudo apt install ffmpeg, version 4.4.2 was used for this project
• Check here for instructions on installing ffmpeg on windows
• Run python3 main.py to start the server
• In separate terminals, run:
  • Run python3 merger.py to start the video merger service
  • Run python3 transribe.py to start the video transcription service
• You can run multiple instances (workers) of the merger and transcription services to speed up the video processing

Endpoints

• POST /upload/new: Start recording a new video

  • Request body:

  {
    file_type: str
  }
  

  • Response:

  {
    status_code: int,
    message: str,
    data: {
      file_id: str
    }
  }
  

• POST /upload/chunks: Upload a chunk of the video

  • Request body: formData with fields:

  {
    file_id: str,
    chunk: Blob,
    is_final: boolean
    chunk_num: int
  }
  

  • Response:

  {
    status_code: int,
    message: str,
    data: null
  }
  

• GET /videos/{file_id}: Get a video

  • Response: Video file

Enjoy!