rabizao/oka-repository

Ongoing project, not ready for production yet

OKA Repository

Install API

sudo apt install python3.8-venv python3.8-dev python3.8-distutils redis-server # For Debian-like systems.
git clone https://github.com/rabizao/oka-repository.git
cd oka-repository/backend
python3.8 -m venv venv
source venv/bin/activate
pip install -r requirements.txt

Start API:

cd backend/
source venv/bin/activate    
python backend.py

Access documentation:

http://localhost:5000/docs or http://localhost:5000/docs/swagger

Start celery workers:

cd backend/
celery worker -A celery_worker.celery --loglevel=info

Install frontend:

sudo apt install npm
cd frontend/
npm install

Start frontend:

npm start

oka-repository's Issues

ZIP download of partitioned data has no training parttions

tests_backend doesn't work on a fresh cloned okarepo from github

FileNotFoundError: [Errno 2] No such file or directory: '... oka-repository/backend/app/media/c210268f-fcb7-4311-8806-883156b0b0ae.zip'

Suggestion to reduce clutter in post pages

When the name of dataset is automatically generated, it competes for attention and space with the visual history.
Out of much creativity for now, I think we could collapse the history in this specific case (automatic name) like an accordion, so the user have to click to expand it.
Another possibility is to visually shorten the name when it is automatic, requiring also some intervention like a click to be shown.
Or even the two ways together, so the user can decide which one they prefer (just by clicking one, another, or both), and the site would remember which he chose and keep it as default for the next posts yet to be seen.

Tratar ARFF defeituoso no backend

Apesar de:
Upload successful. You can navigate while we process your datasets

Celery:
"arff.BadLayout: Invalid layout of the ARFF file, at line 0."

Deveria tratar e alertar usuário (no caso é um arff vazio)

DuplicateEntry when a new user add the same dataset as other