Markdown Link Validator.yml

name: Markdown Link Validator

permissions:
  contents: read

on:
  workflow_dispatch:

jobs:
  build:
    runs-on: windows-latest
    steps:
      - name: Check out the repository code
        uses: actions/checkout@v4

      - name: Validate the links
        shell: pwsh
        run: |
          $ErrorActionPreference = 'Stop'
          Add-Type -TypeDefinition @'
          using System.Collections.Generic;
          using System.Collections.Concurrent;
          public sealed class LinkReport(string originalLink, string fileFoundIn, int statusCode, string redirectedLink, bool redirectMatch)
          {
              public string OriginalLink { get; set; } = originalLink;
              public string FileFoundIn { get; set; } = fileFoundIn;
              public int StatusCode { get; set; } = statusCode;
              public string RedirectedLink { get; set; } = redirectedLink;
              public bool RedirectMatch { get; set; } = redirectMatch;
          }
          public static class Variables
          {
              // List to store instances of LinkReport
              public static ConcurrentBag<LinkReport> LinkReports = new ConcurrentBag<LinkReport>();
          }
          '@

          [string[]]$MarkdownFilePaths = @('.\README.md', '.\Rationale.md')

          # Iterate over each file simultaneously
          $MarkdownFilePaths | ForEach-Object -Parallel {

              [string]$MarkdownContent = Get-Content -Path $_ -Raw -Force

              # Define regex patterns for HTML and Markdown links
              [string]$RegexForHTMLBasedLinks = '="(https://[^"]+)"'
              [string]$RegexForMarkDownBasedLinks = '\(https://[^)]+\)'

              # Initialize a HashSet to collect unique links
              [System.Collections.Generic.HashSet[string]]$AggregatedLinks = @()

              # Find all HTML-based link matches and add them to the HashSet
              $MatchesHTML = [regex]::Matches($MarkdownContent, $RegexForHTMLBasedLinks)
              foreach ($Match in $MatchesHTML) {
                  $Url = $Match.Groups[1].Value
                  [System.Void]$AggregatedLinks.Add($Url)
              }

              # Find all Markdown-based link matches and add them to the HashSet
              $MatchesMarkdown = [regex]::Matches($MarkdownContent, $RegexForMarkDownBasedLinks)
              foreach ($Match in $MatchesMarkdown) {
                  $Url = $Match.Value.Trim('(', ')')
                  [System.Void]$AggregatedLinks.Add($Url)
              }

              # Check each link for validity and redirection
              foreach ($Link in $AggregatedLinks) {
                  # Default values in case the link is unreachable
                  $statusCode = 0
                  $RedirectedLink = $Link
                  $RedirectMatch = $true

                  try {
                      Write-Host "Checking link: $Link" -ForegroundColor Yellow

                      # Attempt to send an HTTP request to check the link, allowing automatic redirection
                      $Response = Invoke-WebRequest -Uri $Link -Method Head

                      # Check if the response was successful
                      if ($null -ne $Response) {
                          # Capture the final status code and redirected URL
                          $statusCode = $Response.StatusCode

                          # Check if the response has a final redirected URL
                          if ($Response.BaseResponse.ResponseUri) {
                              $RedirectedLink = $Response.BaseResponse.ResponseUri.AbsoluteUri
                          }

                          # Determine if the final redirected link matches the original link
                          $RedirectMatch = ($Link -eq $RedirectedLink)
                      }
                      else {
                          # If the link is unreachable, set a status code indicating failure (e.g., 404 or 403)
                          $statusCode = 404
                          $RedirectedLink = 'Unreachable'
                          $RedirectMatch = $false
                      }

                  }
                  catch {
                      # If the link is unreachable, set a status code indicating failure (e.g., 404 or 403)
                      $statusCode = 404
                      $RedirectedLink = 'Unreachable'
                      $RedirectMatch = $false
                  }

                  Write-Host "Status code: $statusCode - Match Result: $RedirectMatch"

                  $LinkReport = [LinkReport]::new($Link, $_, $statusCode, $RedirectedLink, $RedirectMatch)
                  [Variables]::LinkReports.Add($LinkReport)
              }

          } -ThrottleLimit 25

          Write-Host 'Sorting the data'

          # Sort the result so unreachable links appear at the top
          $SortedResult = [Variables]::LinkReports | Sort-Object -Property RedirectMatch

          # Adding header for the markdown summary
          [string]$outputString = "# Link Validation Report 🚀`n`n<br>`n`n"
          $outputString += "- Total Links Checked: $($SortedResult.Count) 📉`n`n"
          $outputString += "- Total number of unreachable links: $($SortedResult.Where({$_.StatusCode -eq 404}).Count) 🙈`n`n<br>`n"
          $outputString += "`n| 🦄 Original Link | 📁 File Found In | 🗿 Status Code | 🎯 Redirected Link | 👀 Redirect Match |"
          $outputString += "`n|:-------------|:------------|:----------:|:---------------|:-------------:|"

          # https://docs.github.com/en/actions/writing-workflows/choosing-what-your-workflow-does/workflow-commands-for-github-actions#adding-a-job-summary
          # https://github.blog/news-insights/product-news/supercharging-github-actions-with-job-summaries/

          Write-Host 'Adding data to the markdown summary'

          foreach ($LinkReport in $SortedResult) {
              # Append result to the output string
              $outputString += "`n| $($LinkReport.OriginalLink) | $($LinkReport.FileFoundIn) | $($LinkReport.StatusCode) | $($LinkReport.RedirectedLink) | $($LinkReport.RedirectMatch) $($($LinkReport.RedirectMatch) ? '✅' : '❌') |"
          }

          Set-Content -Path $env:GITHUB_STEP_SUMMARY -Value $outputString -Force