name: "Setup OSV Scanner DB"
description: "Setup the osv-scanner offline vulnerability database, restoring from cache if available"
author: "Timo Behrendt <t.behrendt@t00n.de>"
branding:
  icon: "database"
  color: "blue"

inputs:
  cache-dir:
    description: "Path used as OSV_SCANNER_LOCAL_DB_CACHE_DIRECTORY (default: ${{runner.temp}}/osv-scanner)"
    required: false
    default: "${{ runner.temp }}/osv-scanner"
  ecosystems:
    description: "Comma-separated list of OSV ecosystems to download. Empty downloads all ecosystems."
    required: false
    default: ""

outputs:
  cache-dir:
    description: "Path to the osv-scanner local DB cache directory"
    value: ${{ inputs.cache-dir }}

runs:
  using: "composite"
  steps:
    - id: current-date
      shell: bash
      run: |
        echo "current-date=$(date +%Y-%m-%d)" >> $GITHUB_OUTPUT
    - id: restore-db
      uses: actions/cache@55cc8345863c7cc4c66a329aec7e433d2d1c52a9 # v6.1.0
      with:
        path: ${{ inputs.cache-dir }}
        key: osv-scanner-db-${{ steps.current-date.outputs.current-date }}
        restore-keys: |
          osv-scanner-db-${{ steps.current-date.outputs.current-date }}
    - name: Download offline databases
      if: steps.restore-db.outputs.cache-hit != 'true'
      shell: bash
      run: |
        set -euo pipefail

        CACHE_DIR="${{ inputs.cache-dir }}"
        DB_DIR="${CACHE_DIR}/osv-scanner"
        mkdir -p "${DB_DIR}"

        ECOSYSTEMS_INPUT="${{ inputs.ecosystems }}"
        if [ -n "${ECOSYSTEMS_INPUT}" ]; then
          # shellcheck disable=SC2001
          ECOSYSTEMS=$(echo "${ECOSYSTEMS_INPUT}" | tr ',' '\n' | sed 's/^[[:space:]]*//;s/[[:space:]]*$//' | grep -v '^$' || true)
        else
          ECOSYSTEMS=$(curl -fsSL https://osv-vulnerabilities.storage.googleapis.com/ecosystems.txt | grep -v '^$' | grep -v '^\[EMPTY\]$' || true)
        fi

        if [ -z "${ECOSYSTEMS}" ]; then
          echo "No ecosystems to download" >&2
          exit 1
        fi

        download_one() {
          local ecosystem="$1"
          local encoded
          encoded=$(printf '%s' "${ecosystem}" | jq -sRr @uri)
          local dest="${DB_DIR}/${ecosystem}"
          mkdir -p "${dest}"
          echo "Downloading ${ecosystem}..."
          curl -fsSL "https://osv-vulnerabilities.storage.googleapis.com/${encoded}/all.zip" -o "${dest}/all.zip"
        }

        max_jobs=8
        pids=()
        while IFS= read -r ecosystem; do
          [ -z "${ecosystem}" ] && continue
          while [ "$(jobs -rp | wc -l)" -ge "${max_jobs}" ]; do
            sleep 0.2
          done
          download_one "${ecosystem}" &
          pids+=("$!")
        done <<< "${ECOSYSTEMS}"

        fail=0
        for pid in "${pids[@]}"; do
          wait "${pid}" || fail=1
        done
        if [ "${fail}" -ne 0 ]; then
          echo "One or more ecosystem database downloads failed" >&2
          exit 1
        fi

        echo "Downloaded $(find "${DB_DIR}" -name all.zip | wc -l) ecosystem database(s) to ${DB_DIR}"
    - name: Export DB cache directory
      shell: bash
      run: |
        echo "OSV_SCANNER_LOCAL_DB_CACHE_DIRECTORY=${{ inputs.cache-dir }}" >> "$GITHUB_ENV"