action.yaml

name: Check links
description: Check the links of the provided site and upload a helpful CSV of results.
author: 'DAKLabb'
inputs:
  url:
    description: The url of the site to check.
    required: true
  site-name:
    description: The name of the site being checked.
    required: false
    default: ''
  report-name:
    required: false
    default: Report
  check-external:
    description: Whether to check external links.
    required: false
    default: 'false'
  include-warnings:
    description: Whether to include warnings.
    required: false
    default: 'true'
  debug-config:
    description: Set to true to log the contents of the linkchecker config.
    required: false
    default: 'false'

runs:
  using: "composite"
  steps:
    - name: Confirm site is accessible
      shell: bash
      run: |
        set +e
        wget --server-response -q --delete-after ${{ inputs.url }}
        if [ $? -ne 0 ]; then
          echo "::error::URL: ${{ inputs.url }} not accessible."
          exit 1
        fi
    - name: Install link checker
      shell: bash
      run: pip install linkchecker
    - name: Setup config file
      shell: bash
      run: echo -e "[csv]\nseparator=,\n[output]\nstatus=0" > lcconfig
    - name: Configure warnings
      if: inputs.include-warnings == 'false'
      shell: bash
      run: echo "warnings=0" >> lcconfig
    - name: Configure external checks
      if: inputs.check-external == 'true'
      shell: bash
      run: echo -e "[filtering]\ncheckextern=1" >> lcconfig
    - name: Log config for testing
      if: inputs.debug-config == 'true'
      shell: bash
      run: cat lcconfig
    - name: Check ${{ inputs.site-name }} Links
      shell: bash
      id: check
      run: |
        set +e
        linkchecker https://${{ inputs.url }} --config lcconfig -F csv/utf-8/link_report.csv
        if [ $? -ne 0 ]; then
          echo "The link check failed, there should be a report"
          echo "report-generated=true" >> $GITHUB_OUTPUT
        fi

        echo "Scan done, generating summary"

        started=false
        while read p; do
          let "ln++"
          # skip comments (top and bottom)
          if [[ $p == \#* ]]; then
            continue
          fi
          # make sure first line after comments is skipped
          if [[ "$started" = false ]]; then
            started=true
            continue
          fi
          
          IFS=',' read -r -a array <<< "$p"

          if [[ ${array[10]} == '' ]]; then
            echo "::error::URL: ${array[0]} on page:${array[1]} is returning a ${array[3]}"
          else
            echo "::error::URL: ${array[0]} linked from '${array[10]}' on page:${array[1]} is returning a ${array[3]}"
          fi

        done < link_report.csv

    - name: Upload report
      if: always() && steps.check.outputs.report-generated == 'true'
      uses: actions/upload-artifact@v4
      with:
        name: ${{ inputs.report-name }}
        path: link_report.csv