_sync-folder.yml

name: _sync-folder

# Reusable: gate (superset) + LFS guard + hub-sync for one folder → a uv-scripts Hub repo.
# Per-repo callers (sync-<repo>.yml) invoke this; the logic lives here only.
# Defaults mirror a top-level folder to uv-scripts/<folder> as a dataset. The optional
# repo_id / repo_type let a folder map to a differently-named repo or a non-dataset repo
# (e.g. the org profile card: org-card/ → uv-scripts/README, a static Space).
on:
  workflow_call:
    inputs:
      folder:
        description: local top-level folder to mirror
        required: true
        type: string
      repo_id:
        description: target Hub repo id (default uv-scripts/<folder>)
        required: false
        type: string
        default: ''
      repo_type:
        description: Hub repo type (dataset | space | model)
        required: false
        type: string
        default: dataset
    secrets:
      HF_TOKEN:
        required: true

jobs:
  sync:
    runs-on: ubuntu-latest
    env:
      REPO_ID: ${{ inputs.repo_id != '' && inputs.repo_id || format('uv-scripts/{0}', inputs.folder) }}
    steps:
      - uses: actions/checkout@v4

      - name: Gate — folder must be a superset of the Hub repo (guards --delete="*")
        env:
          HF_TOKEN: ${{ secrets.HF_TOKEN }}
        run: bash ./tools/verify-superset.sh "${{ inputs.folder }}" "$REPO_ID" "${{ inputs.repo_type }}"

      - name: Guard — no LFS pointers (would corrupt Hub binaries)
        run: |
          if grep -rIl '^version https://git-lfs' "${{ inputs.folder }}/"; then
            echo "::error::LFS pointer file found in ${{ inputs.folder }}/ — would upload pointer text and corrupt the Hub asset"
            exit 1
          fi

      - uses: huggingface/hub-sync@v0.2.0
        with:
          github_repo_id: ${{ github.repository }}
          huggingface_repo_id: ${{ env.REPO_ID }}
          repo_type: ${{ inputs.repo_type }}
          subdirectory: ${{ inputs.folder }}
          hf_token: ${{ secrets.HF_TOKEN }}