Gymnasium/.github/workflows/run-tutorial.yml

name: Test Gymnasium Tutorials

on:
  # Run all tests when merging to main
  push:
    branches: [ main ]

  # Run tests only for modified tutorials in PRs
  pull_request:
    paths:
      - 'docs/tutorials/**/*.py'
      - '.github/workflows/test-tutorials.yml'

  # Allow manual trigger
  workflow_dispatch:

jobs:
  test-tutorials:
    runs-on: ubuntu-latest
    strategy:
      fail-fast: false # This ensures all matrix combinations run even if one fails
      matrix:
        python-version: ["3.10"]
        tutorial-group:
          - gymnasium_basics
          - training_agents

    timeout-minutes: 30  # Set a reasonable timeout for all tests

    steps:
    - uses: actions/checkout@v4

    - name: Set up Python ${{ matrix.python-version }}
      uses: actions/setup-python@v5
      with:
        python-version: ${{ matrix.python-version }}

    - name: Install dependencies
      run: |
        sudo apt-get update
        sudo apt-get install -y libglu1-mesa-dev libgl1-mesa-dev libosmesa6-dev xvfb unzip patchelf ffmpeg cmake swig

    - name: Install Gymnasium basics tutorial requirements
      if: matrix.tutorial-group == 'gymnasium_basics'
      run: |
        pip install ".[mujoco, box2d]"

    - name: Install Training Agents tutorial requirements
      if: matrix.tutorial-group == 'training_agents'
      run: |
        pip install ".[mujoco, toy_text, box2d]" torch seaborn matplotlib pandas tqdm

    - name: Get changed files
      id: changed-files
      uses: tj-actions/changed-files@v46
      with:
        files: docs/tutorials/**/*.py
      if: github.event_name == 'pull_request'

    - name: Patch tutorials
      run: |
        # Patch load_quadruped_model.py to use the built-in ant.xml instead of the missing mujoco_menagerie file
        if [ -f "docs/tutorials/gymnasium_basics/load_quadruped_model.py" ]; then
          sed -i 's|"./mujoco_menagerie/unitree_go1/scene.xml"|"ant.xml"|g' docs/tutorials/gymnasium_basics/load_quadruped_model.py
        fi

        # Patch mujoco_reinforce.py to reduce the number of episodes for CI
        if [ -f "docs/tutorials/training_agents/mujoco_reinforce.py" ]; then
          sed -i 's/total_num_episodes = int(5e3)/total_num_episodes = int(1e2)/g' docs/tutorials/training_agents/mujoco_reinforce.py
        fi

        # Patch frozenlake_q_learning.py to reduce the number of episodes and runs
        if [ -f "docs/tutorials/training_agents/frozenlake_q_learning.py" ]; then
          sed -i 's/total_episodes=2000/total_episodes=200/g' docs/tutorials/training_agents/frozenlake_q_learning.py
          sed -i 's/n_runs=20/n_runs=3/g' docs/tutorials/training_agents/frozenlake_q_learning.py
          sed -i 's/map_sizes = \[4, 7, 9, 11\]/map_sizes = \[4, 7\]/g' docs/tutorials/training_agents/frozenlake_q_learning.py
        fi

        # Patch vector_a2c.py to reduce the number of updates and environments
        if [ -f "docs/tutorials/training_agents/vector_a2c.py" ]; then
          sed -i 's/n_envs = 10/n_envs = 4/g' docs/tutorials/training_agents/vector_a2c.py
          sed -i 's/n_updates = 1000/n_updates = 100/g' docs/tutorials/training_agents/vector_a2c.py
        fi

        # Make sure we use 'rgb_array' render mode instead of 'human' everywhere to avoid display issues
        find docs/tutorials -name "*.py" -type f -exec sed -i 's/render_mode="human"/render_mode="rgb_array"/g' {} \;

    - name: Test tutorials (${{ matrix.tutorial-group }})
      id: run-tutorials
      run: |
        echo "::group::Running tutorials in ${{ matrix.tutorial-group }}"
        mkdir -p test-results

        # Determine which tutorials to test
        if [[ "${{ github.event_name }}" == "pull_request" ]]; then
          # Get the list of modified tutorial files in this group
          for file in ${{ steps.changed-files.outputs.all_changed_files }}; do
            if [[ $file == docs/tutorials/${{ matrix.tutorial-group }}/* && $file == *.py ]]; then
              echo $file >> tutorial_files.txt
            fi
          done

          # If no tutorials in this group were modified, skip this job
          if [ ! -f tutorial_files.txt ] || [ ! -s tutorial_files.txt ]; then
            echo "No tutorials modified in ${{ matrix.tutorial-group }} - skipping"
            exit 0
          fi
        else
          # Find all Python files in the tutorial group
          find docs/tutorials/${{ matrix.tutorial-group }} -name "*.py" -type f | sort > tutorial_files.txt
        fi

        # Initialize counters
        total=0
        passed=0
        failed=0

        # Run each tutorial with timeout - continue even if one fails
        while IFS= read -r tutorial; do
          # Clear separator for better readability
          echo ""
          echo "========================================================"
          echo "Running tutorial: $tutorial"
          echo "========================================================"
          total=$((total+1))

          # Set max time based on complexity (can be adjusted)
          max_time=300  # 5 minutes default

          # Create log file path
          log_file="test-results/$(basename "$tutorial").log"

          # Run the tutorial with timeout and record results
          start_time=$(date +%s)
          # Use set +e so the script continues even if the command fails
          set +e
          timeout $max_time python "$tutorial" > "$log_file" 2>&1
          exit_code=$?
          set -e
          end_time=$(date +%s)
          execution_time=$((end_time-start_time))

          # Output results to console immediately
          if [ $exit_code -eq 0 ]; then
            echo "✅ PASSED: $tutorial (${execution_time}s)"
            passed=$((passed+1))
            echo "$tutorial,pass,$execution_time" >> test-results/summary.csv
          elif [ $exit_code -eq 124 ]; then
            echo "⚠️ TIMEOUT: $tutorial (exceeded ${max_time}s)"
            failed=$((failed+1))
            echo "$tutorial,timeout,$max_time" >> test-results/summary.csv
            # Show the last output before timeout
            echo "Last output before timeout:"
            echo "----------------------------------------"
            tail -n 20 "$log_file"
            echo "----------------------------------------"
          else
            echo "❌ FAILED: $tutorial (${execution_time}s)"
            failed=$((failed+1))
            echo "$tutorial,fail,$execution_time" >> test-results/summary.csv
            # Show the error details
            echo "Error details:"
            echo "----------------------------------------"
            cat "$log_file"
            echo "----------------------------------------"
          fi

        done < tutorial_files.txt

        # Export the counters as outputs for later steps
        echo "total=$total" >> $GITHUB_OUTPUT
        echo "passed=$passed" >> $GITHUB_OUTPUT
        echo "failed=$failed" >> $GITHUB_OUTPUT

        echo "::endgroup::"

    - name: Upload test results
      if: always()
      uses: actions/upload-artifact@v4
      with:
        name: tutorial-test-results-${{ matrix.tutorial-group }}
        path: test-results/
        retention-days: 7

    - name: Check test results
      if: always()
      run: |
        # Read the outputs from the test step
        total="${{ steps.run-tutorials.outputs.total }}"
        failed="${{ steps.run-tutorials.outputs.failed }}"

        # Default to 0 if empty to avoid bash errors
        total=${total:-0}
        failed=${failed:-0}

        if [ "$total" -eq 0 ]; then
          echo "No tutorials were tested in this group."
        elif [ "$failed" -gt 0 ]; then
          echo "Warning: $failed out of $total tutorials failed."
          echo "Check the test results in the uploaded artifacts to see detailed error messages."
        else
          echo "Success: All $total tutorials passed."
        fi

        # Return exit code based on failures
        exit $failed