# Drop a PDF (or a .url/.txt file with a link to a PDF) in _incoming/ and commit it. This job turns it into a finished audiobook folder and commits that back.
# No bot is involved: every step is a deterministic script in _pipeline/.
name: audiobook

on:
  push:
    branches: [main, 'test/**']       # test/** lets the pipeline be proven on a branch without touching main
    paths: ['_incoming/*.pdf', '_incoming/*.PDF', '_incoming/*.url', '_incoming/*.URL', '_incoming/*.txt']   # a PDF or a link file landing in _incoming/ starts a run
  workflow_dispatch:
    inputs:
      pdf:
        description: 'PDF or .url/.txt link-file path inside the repo (empty = everything waiting in _incoming/; a dry run with nothing waiting uses the Clark book as a regression sample)'
        required: false
        default: ''
      voice_endpoint:
        description: 'Testing only: use this voice URL instead of the default (for example to prove a failed voice run commits nothing)'
        required: false
        default: ''
      dry_run:
        description: 'true = text stages only, voice reachability check, push-permission test, commit nothing'
        type: boolean
        default: true

permissions:
  contents: write

concurrency:
  group: audiobook
  cancel-in-progress: false           # one book at a time; a second drop waits

jobs:
  build:
    runs-on: ubuntu-latest
    timeout-minutes: 720              # long books render for hours
    # The commit-back commit carries [skip ci] and removes the PDF from _incoming/, so it cannot start another run.
    if: ${{ !contains(github.event.head_commit.message, '[skip ci]') }}
    env:
      AUDIOBOOK_VOICE_ENDPOINT: ${{ (github.event_name == 'workflow_dispatch' && inputs.voice_endpoint) || vars.AUDIOBOOK_VOICE_ENDPOINT || 'http://192.168.1.202:8090/tts' }}
      AUDIOBOOK_WORK_DIR: /tmp/audiobook-work
      AUDIOBOOK_REQUIRE_OCR: '1'
      BRANCH: ${{ github.ref_name }}
      DRY_RUN: ${{ github.event_name == 'workflow_dispatch' && inputs.dry_run || 'false' }}
      PDF_INPUT: ${{ github.event_name == 'workflow_dispatch' && inputs.pdf || '' }}
    steps:
      - name: Check out the repository
        timeout-minutes: 15           # a stalled fetch fails the run instead of hanging it for hours
        uses: actions/checkout@v4
        with:
          fetch-depth: 100                # enough history to rebase onto a main that moved during a long render
          filter: blob:none               # do not download every book's audio and PDFs on every run
          sparse-checkout: |
            /_pipeline/
            /_incoming/
            /.gitea/
          sparse-checkout-cone-mode: false
      - name: Install tools (fresh container every run)
        run: bash _pipeline/ci/install.sh
      - name: Unit tests (dry runs only)
        if: ${{ env.DRY_RUN == 'true' }}
        run: |
          .venv/bin/pip install -q --disable-pip-version-check reportlab pillow
          .venv/bin/python -W ignore -m unittest discover -s _pipeline/tests 2>&1 | grep -v -E '^\s+(data|read) =' | tail -40
          test "${PIPESTATUS[0]}" -eq 0
      - name: Voice server reachable from this job
        run: |
          echo "voice endpoint: $AUDIOBOOK_VOICE_ENDPOINT"
          base="${AUDIOBOOK_VOICE_ENDPOINT%/tts}"
          curl -sS --max-time 20 "$base/api/model-info" && echo || echo "WARNING: voice server not reachable from this job (the Process step will stop with a clear message if it is needed)"
      - name: Process
        run: bash _pipeline/ci/run.sh