Setup

1. Make sure you have npm installed (tested with version 6.14.4, if you have a lower version then try updating it if you are unable to build the project).
2. Run the following command in your console to install dependencies:

npm i
npm i --global webpack ts-node

3. Install Redis - on Linux this should be possible through your package manager, but on Windows use an Ubuntu WSL environment

Building

This project uses webpack for compilation in order to provide benefits like ES2018 syntax, TypeScript, and .tsx/.jsx file support. Running the following command will make webpack watch files for changes and automatically recompile when you make changes:

npm run dev

To create a minified production build instead, run:

npm run build

Running

1. Build the javascript files using the steps above
2. Install python 3.8+ (tested with 3.8.5)
3. Install PostgreSQL and start running the postgres server
4. Create a database in Postgres called "outcropsketch"
5. In the root folder of the repository, create a new python virtual environment with python -m venv env
6. Install python requirements with pip install -r requirements.txt
7. Create a .env file in the root directory of the repo to fill in with sensitive information like database credentials. Use the default one below as a template and replace values with ones for your configuration:

DJANGO_SECRET_KEY=123456 # Secret key for django server to use, value doesn't matter except in production
GOOGLE_CREDENTIAL_FILE=outcropsketch-123456789.json # Name of Google service account credential json file located in the webserver folder
POSTGRES_HOST=127.0.0.1
POSTGRES_PORT=5432
POSTGRES_USER=postgres
POSTGRES_PASSWORD=password

7. Run sudo service redis-server start to use Redis as a Celery message broker
8. Run Celery in a background process:

cd webserver
celery -A outcropsketch worker

(on Windows, you will need to use WSL or add -P solo to the end of this command as Celery does not officially support Windows) 9. Run the server by running python webserver/manage.py runserver and visiting 127.0.0.1:8000 in your browser

File Conversion for Outcrop Sketch .json files

1. If it doesn't already exist, create a folder named input in the root directory of the project
2. Put all the .json files you want to convert in the input folder
3. Run the following command to convert all .json files into image and .csv files:

npm run convert-json

4. Check the converted/x folder to find the source images, and converted/y for .csv label files.
5. Run the following command to convert the .csv files into .png files with labels:

python utils/csv_to_png.py

6. Check the converted/StructureTypeLabels folder for the structure types, and converted/StructureTypeLabels for surface types.