action.yml

name: "Convert PowerPoint to Video with Voiceover"
description: "Automatically create a video from a PowerPoint presentation, using AI-generated voiceovers for each slide based on the notes section."

inputs:
  pptx_file:
    description: "The path to the PowerPoint (.pptx) file to convert."
    required: true
  openai_api_key:
    description: "The OpenAI API key to use for TTS generation."
    required: true
  model:
    description: "The TTS model to use for speech synthesis (default: tts-1-hd)."
    required: false
    default: "tts-1-hd"
  voice:
    description: "The voice profile to use for TTS generation (default: echo)."
    required: false
    default: "echo"
  response_format:
    description: "The audio format of the output file (default: mp3)."
    required: false
    default: "mp3"
  speed:
    description: "The speed of the synthesized speech (0.25 to 4.0, default: 1.0)."
    required: false
    default: "1.0"
  output_file:
    description: "The output video file name (default: output.mp4)."
    required: false
    default: "output.mp4"

runs:
  using: "composite"
  steps:
    - name: Check out repository
      uses: actions/checkout@v4

    - name: Set up Python
      uses: actions/setup-python@v5
      with:
        python-version: "3.x"

    - name: Install system dependencies
      shell: bash
      run: |
        sudo apt-get update
        sudo apt-get install -y ffmpeg libreoffice poppler-utils

    - name: Install Python dependencies
      shell: bash
      run: |
        python -m pip install --upgrade pip
        pip install -r requirements.txt

    - name: Generate video from PowerPoint
      shell: bash
      run: |
        export OPENAI_API_KEY=${{ inputs.openai_api_key }}
        python scripts/pptx2video.py ${{ inputs.pptx_file }} \
          --model ${{ inputs.model }} \
          --voice ${{ inputs.voice }} \
          --format ${{ inputs.response_format }} \
          --speed ${{ inputs.speed }}

    - name: Upload video artifact
      uses: actions/upload-artifact@v4.4.0
      with:
        name: video-output
        path: ${{ inputs.output_file }}

outputs:
  video_file:
    description: "The path to the generated video file."
    value: ${{ inputs.output_file }}