action.yml

name: 'broken-links'
description: 'Detect broken links in html content (relative only)'
inputs:
  path:
    description: 'Local path to html content (working directory by default)'
    required: true
  prefix:
    description: 'Prefix expected by the html content (empty by default)'
    required: false
    default: ''
  root:
    description: 'Root page to start detection (index.htm or html by default)'
    required: false
    default: '/'
  port:
    description: 'Port use by the temporary http server'
    required: false
    default: 8000
outputs:
  result:
    description: 'Result of the detection'
    value: ${{ steps.scan.outputs.result }}
runs:
  using: "composite"
  steps:
      - id: scan
        name: Scan for broken links
        shell: bash
        run: |
          # Use a temp server to check the links
          echo ":rocket: Broken link detection started" >> $GITHUB_STEP_SUMMARY
          # Prepare temp files
          TMP_LOG="$(mktemp log.XXXXXXXXXX)"
          TMP_OUT="$(mktemp out.XXXXXXXXXX)"
          # Prepare serving directory
          if [ -z "${{ inputs.prefix }}" ]; then
            HTTPD_DIR="${{ inputs.path }}"
          else
            # Prepare symbolic link to content with prefix
            HTTPD_DIR="$(mktemp --directory www.XXXXXXXXXX)"
            pushd "${HTTPD_DIR}"; ln -s "../${{ inputs.path }}" "${{ inputs.prefix }}"; popd
          fi
          # Start a minimal http server to host the files
          HTTPD_PID=$(nohup python3 -m http.server --directory ${HTTPD_DIR} --bind 127.0.0.1 ${{ inputs.port }} > /dev/null 2>&1 & echo $!)
          # Ordely exit with trimmed output
          trap "kill -TERM ${HTTPD_PID} && rm -f ${TMP_LOG} ${TMP_OUT}" EXIT
          # Wait for the server
          for try in {1..3}; do
            nc -w 3 127.0.0.1 ${{ inputs.port }} > /dev/null 2>&1 && break || { sleep 1; }
            if [ $try = 3 ]; then
              echo ":x: Scan server failed to start" >> $GITHUB_STEP_SUMMARY;
              exit 1
            fi
          done
          # Recursively test all local
          RET=0
          wget --spider --recursive --no-verbose -o "${TMP_LOG}" \
          "http://127.0.0.1:${{ inputs.port }}${{ inputs.root }}"\
          && echo ":heavy_check_mark: Scan completed - no broken link found" >> $GITHUB_STEP_SUMMARY \
          || { echo ":x: Scan completed - some link are broken" >> $GITHUB_STEP_SUMMARY; RET=1; }
          # Parse output and send back as result
          sed -n -r -e '/Found/,/FINISHED/ p' "${TMP_LOG}" > "${TMP_OUT}"
          echo "result<<$(basename "${TMP_OUT}")" >> $GITHUB_OUTPUT
          cat "${TMP_OUT}" >> $GITHUB_OUTPUT
          echo "$(basename "${TMP_OUT}")" >> $GITHUB_OUTPUT
          # Print raw outpout log and exit
          cat "${TMP_LOG}"
          exit ${RET}