W3C HTML Sitemap-based Site Validator

validate-site-html.ps1

param ([Parameter(Mandatory=$True)][string]$SitemapUrl,
	   [int]$Max=0)	

function Validate($url) {
	
	if(-not $url.StartsWith("http")) {
		$parsed = [System.Uri]$SitemapUrl
		$url = "{0}://{1}{2}" -f $parsed.Scheme, $parsed.Host, $url
	}

	try { 
		$test = Invoke-WebRequest -Uri $url -MaximumRedirection 0 -ErrorAction Ignore 
		if ($test.StatusCode -ge 300) {
			Write-Warning ("{0}: HTTP {1}" -f $url, $test.StatusCode)
		} else {
			$result = Invoke-RestMethod -Uri ("http://validator.w3.org/check?uri={0}&output=json" -f [System.Web.HttpUtility]::UrlEncode($url)) 
				
			$errors = $result.messages | Where-Object { $_.subtype -eq "error" }
			$warnings = $result.messages | Where-Object { $_.subtype -eq "warning" }
			$message = ("{0}: {1} Error{2} {3} Warning{4}" -f $url, $errors.Count, @{$true="s";$false=""}[$errors.Count -gt 0], $warnings.Count, @{$true="s";$false=""}[$warnings.Count -gt 0])	
		
			if ($warnings.Count -gt 0 -or $errors.Count -gt 0) 
			{
				Write-Warning $message
			} else {
				Write-Host $message
			}
			
		}
	} catch {
		Write-Warning ("{0} {1}" -f $url, $_.Exception.Message)
	} #This is bad at handling 404's
}

[xml]$sitemap = (New-Object System.Net.WebClient).DownloadString($SitemapUrl)
$urls = $sitemap.urlset.get_ChildNodes() | foreach { $_.loc }
$i = 0

while($i -lt $Max -and $i -lt $urls.Count)
{
	Validate $urls[$i]
	$i = $i+1
}