syft/task.d/generate/cpe-index.yaml

version: "3"

vars:
  CPE_CACHE_DIR: "syft/pkg/cataloger/internal/cpegenerate/dictionary/index-generator/.cpe-cache"
  CPE_CACHE_REGISTRY: "ghcr.io/anchore/syft/cpe-cache:latest"
  CPE_INDEX_OUTPUT: "syft/pkg/cataloger/internal/cpegenerate/dictionary/data/cpe-index.json"
  CPE_GENERATOR_DIR: "syft/pkg/cataloger/internal/cpegenerate/dictionary/index-generator"

tasks:
  cache:pull:
    desc: Pull CPE cache from ORAS registry (ghcr.io/anchore/syft/cpe-cache:latest)
    # deps: [tools]
    cmds:
      - cmd: |
          set -eu
          echo "Pulling CPE cache from ORAS registry..."
          mkdir -p {{ .CPE_CACHE_DIR }}

          # pull compressed files from ORAS
          {{ .ORAS }} pull {{ .CPE_CACHE_REGISTRY }} --output {{ .CPE_CACHE_DIR }} || {
            exit_code=$?
            if [ $exit_code -eq 1 ]; then
              echo "No existing cache found in registry (this is normal for first run)"
              exit 0
            else
              exit $exit_code
            fi
          }

          # handle nested directory structure from old pushes (if exists)
          # files might be at .cpe-cache/syft/pkg/.../cpe-cache/*.json.zst
          nested_cache=$(find {{ .CPE_CACHE_DIR }} -type d -name ".cpe-cache" ! -path {{ .CPE_CACHE_DIR }} | head -1)
          if [ -n "$nested_cache" ]; then
            echo "Found nested cache structure, moving files to correct location..."
            mv "$nested_cache"/*.json.zst {{ .CPE_CACHE_DIR }}/ 2>/dev/null || true
            # clean up nested directories
            rm -rf {{ .CPE_CACHE_DIR }}/syft 2>/dev/null || true
          fi

          # decompress all .json.zst files to .json
          echo "Decompressing cache files..."
          decompressed_count=0
          for zst_file in {{ .CPE_CACHE_DIR }}/*.json.zst; do
            # skip if no .zst files found (glob didn't match)
            if [ ! -f "$zst_file" ]; then
              echo "No compressed files to decompress"
              break
            fi

            # decompress to .json (removing .zst extension)
            json_file="${zst_file%.zst}"
            echo "  Decompressing $(basename "$zst_file")..."
            zstd -d -q -f "$zst_file" -o "$json_file"

            # remove compressed file
            rm "$zst_file"
            decompressed_count=$((decompressed_count + 1))
          done

          if [ $decompressed_count -gt 0 ]; then
            echo "Decompressed $decompressed_count file(s) successfully"
          fi
        silent: false

  cache:update:
    desc: Fetch incremental CPE updates from NVD API and update local cache
    # deps: [tools]
    dir: "{{ .CPE_GENERATOR_DIR }}"
    cmds:
      - cmd: |
          echo "Updating CPE cache from NVD Products API..."
          go run . -cache-only
        silent: false
    sources:
      - "{{ .CPE_GENERATOR_DIR }}/*.go"
    generates:
      - "{{ .CPE_CACHE_DIR }}/metadata.json"
      - "{{ .CPE_CACHE_DIR }}/products/*.json"

  cache:push:
    desc: Push updated CPE cache to ORAS registry
    # deps: [tools]
    cmds:
      - cmd: |
          set -eu
          echo "Pushing CPE cache to ORAS registry..."

          if [ ! -d "{{ .CPE_CACHE_DIR }}" ]; then
            echo "No cache directory found"
            exit 1
          fi

          # store absolute path to ORAS before changing directory
          oras_bin="$(pwd)/{{ .ORAS }}"

          # change to cache directory to avoid including full paths in ORAS push
          cd {{ .CPE_CACHE_DIR }}

          # find all JSON files (basenames only)
          json_files=$(find . -maxdepth 1 -type f -name "*.json" -exec basename {} \;)
          if [ -z "$json_files" ]; then
            echo "No cache files to push"
            exit 1
          fi

          # compress each JSON file to .json.zst
          echo "Compressing cache files..."
          compressed_files=""
          for json_file in $json_files; do
            zst_file="${json_file}.zst"
            echo "  Compressing $(basename "$json_file")..."
            zstd -q -f "$json_file" -o "$zst_file"
            compressed_files="$compressed_files $zst_file"
          done

          # push compressed files to ORAS (from cache directory, so only basenames are used)
          echo "Pushing compressed files to registry..."
          "$oras_bin" push {{ .CPE_CACHE_REGISTRY }} $compressed_files \
            --annotation org.opencontainers.image.source=https://github.com/{{ .OWNER }}/{{ .PROJECT }} \
            --annotation org.opencontainers.image.created=$(date -u +"%Y-%m-%dT%H:%M:%SZ")

          # clean up compressed files
          echo "Cleaning up compressed files..."
          for zst_file in $compressed_files; do
            rm "$zst_file"
          done

          echo "Cache pushed successfully"
        silent: false

  build:
    desc: Generate cpe-index.json from existing local cache (does not pull/push/update)
    dir: "{{ .CPE_GENERATOR_DIR }}"
    cmds:
      - cmd: |
          echo "Generating CPE index from local cache..."
          go run . -o ../data/cpe-index.json
          echo "CPE index generated successfully"
        silent: false
    sources:
      - "{{ .CPE_GENERATOR_DIR }}/*.go"
      - "{{ .CPE_CACHE_DIR }}/metadata.json"
      - "{{ .CPE_CACHE_DIR }}/products/*.json"
    generates:
      - "{{ .CPE_INDEX_OUTPUT }}"

  cache:clean:
    desc: Remove local CPE cache directory
    dir: "{{ .CPE_GENERATOR_DIR }}"
    cmds:
      - rm -rf .cpe-cache
      - echo "CPE cache cleaned"