Autonomous advent of code

• Each day during advent of code this repo solves the latest puzzle and submits the solution - all without human intervention
• All code in this repo was written by sourcery-ai[bot]
• Everything was set up by manually writing five issues and asking sourcery-ai[bot] to implement them
• Since then everything runs fully autonomously

There are two main processes that work together to solve the puzzles:

• Four GitHub actions workflows download the puzzles, create issues for sourcery-ai[bot] to solve the puzzles, merge PRs, and submit solutions. All actions they perform are carried out by @SourceryAI which is a GitHub account not associated with a real person
• sourcery-ai[bot] solves the puzzles and opens pull requests with the solutions and answers

How does it work?

Here are the steps:

1. Twelve hours after each puzzle is published a GitHub actions workflow [^1]:
   • Download the puzzle and input
   • ROT13 encode the puzzle [^2]
   • Commit both to the puzzle directory and push to main
   • This triggers step 2
2. A GitHub action workflow is triggered when puzzles are pushed to main:
   • Create a GitHub issue to solve the part
   • Post an issue comment with content @sourcery-ai develop
   • This triggers step 3
3. sourcery-ai[bot] listens to the comment:
   • Runs a script to ROT13 decode the puzzle
   • Posts a plan to the issue to solve the puzzle
   • Opens a pull request to implement the solution and write the solution
   • This triggers step 4
4. A GitHub action workflow is triggered when pull requests are opened
   • It merges pull requests only if the author is sourcery-ai[bot]
   • This triggers step 5
5. A GitHub action workflow is triggered when answers are pushed to main
   • It submits the answer to advent of code
   • It updates the results below
   • If the answer is correct and it was for part 1, it downloads the second part of the puzzle and commits it
   • This triggers step 2

How sourcery-ai[bot] works:

• When a member of a repo comments on an issue with @sourcery-ai develop it will post a plan to the issue, open a pull request with the solution, and request a review from the commenter
• If any review comments are left on its pull request it will address them and push new commits

Results

Here are the puzzles attempted and whether they were successfully solved or not.

Try out Sourcery

sourcery-ai[bot] is an GitHub app with two main functions:

• Issue agent
  • Comment an issue with @sourcery-ai plan to post a comment with a planned implementation
  • Comment and issue with @sourcery-ai develop to open a pull request to resolve the issue 👈 the functionality on display in this repo
• Pull request agent
  • Automatically reviews pull requests
  • Generates pull request titles / bodies
  • Generates review guides with diagrams explaining the changes

The functionality in this repo (@sourcery-ai develop) is currently in alpha testing and will be generally available in the new year. As well as the functionality shown here it will:

• Run in GitHub / GitLab - cloud and self-hosted
• Integrate with issue trackers - Linear, Jira
• Resolve errors in production - Sentry, etc.

It will be most useful for handling routine maintenance and clearing your backlog by:

• Fixing bugs
• Implementing smaller features
• Automatically fixing production issues
• Writing documentation

If you like the sound of this, join our waitlist and tell us anything else you want it to do.

Triggering the merge-repos Workflow

To trigger the merge-repos workflow, you need to dispatch a repository event with the event type merge-repos. You can do this using the GitHub API or the gh CLI tool.

Example using gh CLI

gh api repos/:owner/:repo/dispatches --field event_type=merge-repos --field client_payload='{"source_repo": "source-repo-name", "target_repo": "target-repo-name"}'

Replace :owner with the repository owner, :repo with the repository name, source-repo-name with the name of the source repository, and target-repo-name with the name of the target repository.

Example using GitHub API

curl -X POST -H "Accept: application/vnd.github.v3+json" -H "Authorization: token YOUR_GITHUB_TOKEN" https://api.github.com/repos/:owner/:repo/dispatches -d '{"event_type":"merge-repos","client_payload":{"source_repo":"source-repo-name","target_repo":"target-repo-name"}}'

Replace YOUR_GITHUB_TOKEN with your GitHub token, :owner with the repository owner, :repo with the repository name, source-repo-name with the name of the source repository, and target-repo-name with the name of the target repository.

Rotating GH_TOKEN Regularly

To rotate the GH_TOKEN regularly, follow these steps:

1. Generate a new GitHub token with the required permissions.
2. Update the GitHub Secrets in your repository settings with the new token.
3. Update any local environment variables or configuration files that use the GH_TOKEN.
4. Monitor the usage of the new token to ensure it is working correctly.
5. Revoke the old token to minimize the risk of unauthorized access.

Monitoring the Usage of GH_TOKEN

To monitor the usage of the GH_TOKEN, follow these best practices:

1. Enable GitHub's security features, such as security alerts and vulnerability scanning, to detect any suspicious activity.
2. Regularly review the audit logs in your GitHub repository to track the usage of the GH_TOKEN.
3. Set up notifications for any unusual activity or changes in the repository.
4. Use third-party tools or services to monitor the usage of the GH_TOKEN and detect any potential security issues.
5. Periodically review and update the permissions assigned to the GH_TOKEN to ensure they are still necessary and appropriate.

Merging Code from Different Branches or Repositories

The repository contains GitHub Actions workflows that automatically merge code from different branches or repositories. These workflows use the gh pr merge command with the --squash, --auto, and --delete-branch options.

Example using gh pr merge

gh pr merge <pull-request-number> --squash --auto --delete-branch

Replace <pull-request-number> with the number of the pull request you want to merge.

Example using GitHub Actions Workflow

The following GitHub Actions workflow automatically merges pull requests created by the sourcery-ai[bot] user:

name: Auto-merge Sourcery AI Bot PRs

on:
  pull_request:
    types: [opened, synchronize, reopened]

permissions:
  contents: write

env:
  GH_TOKEN: ${{ secrets.GH_TOKEN }}

jobs:
  merge-sourcery-pr:
    runs-on: ubuntu-latest
    if: github.event.pull_request.user.login == 'sourcery-ai[bot]'
    steps:
      - name: Check out repository
        uses: actions/checkout@v4

      - name: Merge Pull Request
        run: |
          gh pr merge "${{ github.event.pull_request.number }}" \
            --squash \
            --auto \
            --delete-branch \
            --repo "${{ github.repository }}"

      - name: Merge develop into main
        run: |
          git fetch origin
          git checkout main
          git merge develop
          gh pr merge --squash --auto --delete-branch

Poker Coach UI

The Poker Coach UI is an interactive interface for hand analysis and game play. It helps users analyze their poker hands and provides feedback.

Installation

To install the Poker Coach UI, follow these steps:

1. Ensure you have Python 3.12 or higher installed.
2. Clone the repository:

   git clone https://github.com/sourcery-ai/test-aoc.git
   cd test-aoc

3. Install the required dependencies:

   pip install -r requirements.txt

Usage

To use the Poker Coach UI, run the following command:

python -m poker_coach.ui

This will open the Poker Coach UI window where you can enter your poker hand and analyze it.