name: Provision dataset

on:
  workflow_dispatch:
    inputs:
      dataset:
        description: Dataset to download, convert, and upload as a tarball
        type: choice
        options:
          - kitti
          - euroc
          - tum
          - tartan
        default: kitti
      force_download:
        description: Re-download raw archives even when they already exist
        type: boolean
        default: false
      dry_run:
        description: Build the tarball but skip S3 upload
        type: boolean
        default: false

concurrency:
  group: provision-dataset-${{ github.event.inputs.dataset }}
  cancel-in-progress: false

permissions:
  contents: read

env:
  FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: true
  AWS_DEFAULT_REGION: ${{ vars.AWS_DEFAULT_REGION }}
  S3_DATASETS_BUCKET: ${{ vars.S3_DATASETS_BUCKET }}

jobs:
  provision:
    name: "Provision ${{ github.event.inputs.dataset }}"
    if: github.ref == 'refs/heads/main'
    runs-on: [gpu]
    timeout-minutes: 360
    steps:
      - uses: actions/checkout@v5
        with:
          persist-credentials: false

      - name: Build CI tools image
        env:
          AWS_CLI_PUBLIC_KEY: ${{ vars.AWS_CLI_PUBLIC_KEY }}
        run: |
          keyfile="$RUNNER_TEMP/aws-cli-public-key.asc"
          if [ -z "${AWS_CLI_PUBLIC_KEY:-}" ]; then
            echo "AWS_CLI_PUBLIC_KEY repository variable is not configured" >&2
            exit 1
          fi
          printf '%s\n' "$AWS_CLI_PUBLIC_KEY" > "$keyfile"
          DOCKER_BUILDKIT=1 docker build -f scripts/Dockerfile.ci . --network host --tag cuvslam-ci:local \
            --secret id=aws_cli_public_key,src="$keyfile"

      - name: Verify TartanGround provisioning architecture
        if: github.event.inputs.dataset == 'tartan'
        env:
          RUNNER_ARCH: ${{ runner.arch }}
        run: |
          if [ "$RUNNER_ARCH" != X64 ]; then
            echo "TartanGround provisioning requires an x86_64 runner; got $RUNNER_ARCH." >&2
            exit 1
          fi

      - name: Build TartanGround provisioning image
        if: github.event.inputs.dataset == 'tartan'
        run: |
          DOCKER_BUILDKIT=1 docker build -f scripts/Dockerfile.dataset-provision . --network host \
            --build-arg BASE_IMAGE=cuvslam-ci:local \
            --tag cuvslam-dataset-provision:local

      - name: Seed S3 dataset tarball
        env:
          DATASET: ${{ github.event.inputs.dataset }}
          FORCE_DOWNLOAD: ${{ github.event.inputs.force_download }}
          DRY_RUN: ${{ github.event.inputs.dry_run }}
          AWS_ACCESS_KEY_ID: ${{ github.event.inputs.dry_run != 'true' && secrets.AWS_S3_ACCESS_KEY_ID || '' }}
          AWS_SECRET_ACCESS_KEY: ${{ github.event.inputs.dry_run != 'true' && secrets.AWS_S3_SECRET_ACCESS_KEY || '' }}
        run: |
          host_work="$RUNNER_TEMP/provision-work-${GITHUB_RUN_ID}"
          container_work=/tmp/cuvslam-provision
          provision_image=cuvslam-ci:local
          if [ "$DATASET" = tartan ]; then
            provision_image=cuvslam-dataset-provision:local
          fi
          mkdir -p "$host_work"
          docker run --rm --network host \
            -e FORCE_DOWNLOAD -e DRY_RUN -e AWS_DEFAULT_REGION -e S3_DATASETS_BUCKET \
            -e AWS_ACCESS_KEY_ID -e AWS_SECRET_ACCESS_KEY \
            -e PROVISION_WORK_DIR="$container_work" \
            -v "$(pwd):/cuvslam:ro" \
            -v "$host_work:$container_work" \
            -w /cuvslam \
            "$provision_image" ./scripts/provision_dataset.sh "$DATASET"