Type Changes Bot

Requirements

• Python 3.11+
• Pip
• A running Redis-compatible server
• A running database server (MySQL is known to work)
• A working JDK installation (JDK 17.0.13 is known to work)

See pyproject.toml for more specific requirements.

Setup

Basic Installation

This section is applicable to the first node.

1. Clone repository from GitHub, and cd into it.
2. Create a Python virtual environment, python -mvenv venv
3. Activate the virtual environment, source venv/bin/activate
4. Install minimum necessary packages, with pip install -e .
5. Generate a secret key and copy to the clipboard: python manage.py shell -c "from django.core.management import utils; print(utils.get_random_secret_key())".
6. Copy env.template to .env and edit according to your local configuration (see also Configuration).
7. Create migration files for the database: python manage.py makemigrations
8. Create database tables: python manage.py migrate
9. Collect static files: python manage.py collectstatic
10. Create an administrative user, python manage.py createsuperuser

On all remaining nodes, complete steps 1--4, copying the .env file and make any necessary changes.

GumTree Installation

On all nodes which will run the worker, complete the following additional steps to install gumtree.

1. Activate the virtual environment, source venv/bin/activate
2. Install the gumtree support packages, pip install -e .[gumtree_support].
3. Initialize GumTree submodules, git submodule update --init --recursive.
4. Navigate to the gumtree directory, cd gumtree
5. Build gumtree, ./gradlew build
6. Navigate to the gumtree parser directory (cd ../gumtree-parser), and pre-build the tree-sitter parsers by running the following commands:
   • ./tree-sitter-parser.py ../test-data/py/arg-type/with.py python
   • ./tree-sitter-parser.py ../test-data/php/arg-definition/with.php php
   • ./tree-sitter-parser.py ../test-data/r/args-default/with.r r
   • ./tree-sitter-parser.py ../test-data/rb/local/with.rb ruby
   • ./tree-sitter-parser.py ../test-data/ts/arg-type/with.ts typescript

Worker Service Installation

This section is applicable to all nodes running the worker service.

1. Copy the file template-typechangebot-worker.service to typechangebot-worker.service.
2. Edit the file, taking special care to set the WorkingDirectory setting to the installation directory, and the User setting to the user which will the worker will be run as.
3. Copy typechangebot-worker.service to the /etc/systemd/system/ directory.
4. Reload the systemd daemon with systemctl daemon-reload.
5. Enable and start the service with systemctl enable --now typechangebot-worker.

Web Server Installation

Configure your webserver/WSGI installation following vendor directions. If using Ubuntu and Apache, Digital Ocean's guide may be helpful.

Configuration

The following environment variable must be configured in a .env file. This file must be available on all nodes, and all nodes must be uniformly configured.

Basic

• SECRET_KEY Necessary configuration parameter to ensure application security. See above.
• ALLOWED_HOSTS Comma-separated list of allowed host names, see also Django documentation
• URL_ROOT Path from domain root to app, normally /, but may be /tcbot or similar based on your configuration.
• DATA_DIR Absolute path to directory containing stored repositories.

GitHub Application

The following must be collected from GitHub's application management UI.

• GITHUB_APP_KEY Separate newlines with literal \n
• GITHUB_APP_ID
• GITHUB_APP_CLIENT_SECRET

Gum Tree

• GUMTREE_DIR Location of GumTree binaries. Note, the example configuration will only need TODO replaced with the path to the installation directory.
• GUMTREE_TREE_SITTER_DIR Location of the GumTree treesitter parser. Note, the example configuration will only need TODO replaced with the path to the installation directory.

Database

• DATABASE_URL a dj-database-url compatible database URL. Make sure to install the relevant database drivers (necessary packages for MySQL are available with the mysql optional dependency set).

Task Queue

• CELERY_BROKER_URL URL for Redis access. See also Celery documentation. It is recommended that you use a database number specific to this project.