.github/workflows/build-llama-cpp-wheel.yml

name: Build Llama wheel in Docker

on:
  push:
    tags:
      - '*-llama'

jobs:
  build:
    runs-on: self-hosted
    permissions:
      contents: write
    steps:
      - name: Checkout code
        uses: actions/checkout@v2

      - name: Extract version from tag
        id: get_version
        run: echo "::set-output name=version::${GITHUB_REF#refs/tags/}"

      - name: Separate version and suffix
        id: separate
        run: |
          VERSION=$(echo ${{ steps.get_version.outputs.version }} | rev | cut -d'-' -f2- | rev)
          echo "::set-output name=prefix::$VERSION"

      - name: Build Docker image
        run: docker build . -t artifact-builder --build-arg LLAMA_CPP_VERSION=${{ steps.separate.outputs.prefix }} -f Dockerfile-llama-cpp-wheel

      - name: Run Docker container
        run: docker run --name my-artifact-builder artifact-builder

      - name: Copy GPU artifact from Docker container
        run: |
          docker cp my-artifact-builder:/build/dists/llama_cpp_python-gpu-${{ steps.separate.outputs.prefix }}-cp38-cp38-linux_x86_64.whl ./llama_cpp_python-gpu-${{ steps.separate.outputs.prefix }}-cp38-cp38-linux_x86_64.whl

      - name: Upload artifacts
        uses: actions/upload-artifact@v3
        with:
          name: wheels
          path: |
            *.whl

  release:
    needs: build
    runs-on: self-hosted
    permissions:
      contents: write
    steps:
      - name: Checkout code
        uses: actions/checkout@v2

      - name: Download artifacts
        uses: actions/download-artifact@v3
        with:
          name: wheels

      - name: Release
        uses: softprops/action-gh-release@v1
        with:
          files: |
            *.whl
          token: ${{ secrets.GITHUB_TOKEN }}