Advent of Code

My solutions for Advent of Code challenges.

Quick Start (For Others Using This Repo)

⚠️ First Time Setup: See SETUP.md for detailed installation and configuration instructions.

Quick version:

1. Clone the repository
2. Copy and configure environment: cp .env.example .env (update paths)
3. Run make new to create fresh boilerplate (archives my solutions)
4. Add your session cookie to .session file (see Auto-Submission Setup)
5. Start the auto-fetch server: ./launchers/start_server_daemon.command
6. Write your solution in the solve() function
7. Run it: make run-dXpY (just runs your code)
8. Test it: make test-dXpY (runs tests, auto-submits if they pass)

Where X is the day number and Y is the part (1 or 2). Example: make test-d1p1

Note

These solutions are preserved exactly as they were when I first solved each puzzle. I stopped working on them the moment I got the correct answer. They are unoptimised and most certainly are inefficient, as I wish to document my progress authentically as I grow as an engineer.

I'll come back to these later when I'm more experienced and refactor them to see how much I've improved.

Structure

Each day follows a consistent pattern:

2025/
└── day01/
    ├── part1.go      # Part 1 solution
    ├── part2.go      # Part 2 solution
    ├── input.txt     # Your puzzle input
    ├── testcases.txt # Test cases + expected outputs
    └── problem.md    # Problem description (auto-converted from HTML)

Note: benchmark_test.go is generated on-demand when running benchmarks, not stored in the repo.

Auto-Fetch Server

The auto-fetch server handles everything automatically.

First time setup: See SETUP.md for configuration instructions.

Start the server:

# Daemon mode (runs in background, auto-restarts)
./launchers/start_server_daemon.command

# Foreground mode (legacy)
./start_server.command

# Or run manually
cd server && node server.js

Server management:

./launchers/server_status.command    # Check if running
./launchers/stop_server.command       # Stop daemon
./launchers/server_logs.command       # View logs

The server will:

• ⏰ Auto-fetch at 4:00:05 PM AEDT when puzzles unlock
• 📥 Download problem description → dayN/problem.md
• 📥 Download your personal input → dayN/input.txt
• 🤖 Use Claude SDK to extract and populate test cases → testcases.txt
• 🔄 Auto-fetch Part 2 when you complete Part 1
• 💤 Run continuously in the background

Manual fetch (server must be running):

make fetch        # Fetch today's puzzle
make fetch-<day>  # Fetch specific day (e.g., make fetch-4)

Note: Manual fetch commands require the server to be running for Claude-powered test case extraction.

Solving Workflow

1. Wait for auto-fetch (or run make fetch-<day>)
2. Implement your solution in the solve() function (part1.go or part2.go)
3. Iterate with make run-dXpY to see your answer without running tests
4. Submit with make test-dXpY when ready - runs tests, auto-submits if they pass

Testing vs Running

# Just run your solution (no tests, no submission)
make run-dXpY

# Test + auto-submit (runs tests, submits if they pass)
make test-dXpY

Where X is the day number and Y is the part (1 or 2). Examples: make run-d1p1, make test-d12p2

Use make run-dXpY when developing. Use make test-dXpY when ready to submit.

Note: Auto-submission is controlled by the autoSubmit constant in each solution file (default: true).

How Testing Works

The test harness will:

1. ✅ Run your solution against test cases from testcases.txt
2. ✅ If tests pass, run against your real input from input.txt
3. ✅ Auto-submit the answer to Advent of Code

Auto-Submission Setup

To enable auto-submission:

1. Log in to adventofcode.com
2. Open DevTools (F12) → Application/Storage → Cookies
3. Copy the session cookie value
4. Create a .session file in the project root:

   echo "your_session_cookie_here" > .session

The .session file is git-ignored for security.

Starting Fresh (make new)

The make new command archives existing solutions and creates fresh boilerplate:

make new

This will:

1. Archive current solutions to archive/2025_TIMESTAMP.tar.gz
2. Delete and recreate the 2025/ directory with fresh boilerplate for all 12 days
3. Create .session.example file if .session doesn't exist

Perfect for:

• Starting a new year
• Letting others use your setup
• Resetting to clean slate

Your old solutions are safely archived!

Benchmarks

# Benchmark individual solutions (where X is day, Y is part 1 or 2)
make bench-dXpY   # Examples: make bench-d1p1, make bench-d3p2

# Run all benchmarks and update README table
make bench-all

Or run directly:

cd 2025/day01 && go test -tags=part1 -bench=BenchmarkPart1 -benchmem
cd benchmark/cmd && go run main.go  # Run all and update README

2025 Results (Apple M4 Pro)

Day	Part 1	Part 2
1	⭐ 175.28 µs/op	⭐ 175.14 µs/op
2	⭐ 99.46 ms/op	⭐ 604.75 ms/op
3	⭐ 53.32 µs/op	⭐ 66.25 µs/op
4	⭐ 69.66 µs/op	⭐ 800.97 µs/op
5	⭐ 1.16 ms/op	⭐ 27.79 µs/op

Language

Solutions are written in Go.