param(
    [Parameter(Mandatory=$true,HelpMessage="URL of the sitemap.xml to process")]
    [string]$SitemapUrl,
    
    [Parameter(Mandatory=$false,HelpMessage="Path to output file (defaults to sitemap_urls.txt in current directory)")]
    [string]$OutputPath = "$(Get-Location)\sitemap_urls.txt"
)

function Get-UrlsFromSitemap {
    param(
        [xml]$XmlDoc,
        [string]$BaseUrl
    )
    $urls = @()

    # Detect if this is a sitemap index
    if ($XmlDoc.sitemapindex) {
        foreach ($sitemapNode in $XmlDoc.sitemapindex.sitemap) {
            $loc = $sitemapNode.loc -as [string]
            if ($loc) {
                try {
                    $subContent = Invoke-WebRequest -Uri $loc -UseBasicParsing -ErrorAction Stop
                    $subXml = [xml]$subContent.Content
                    $urls += Get-UrlsFromSitemap -XmlDoc $subXml -BaseUrl $loc
                } catch {
                    Write-Warning "Failed to retrieve nested sitemap at $loc : $_"
                }
            }
        }
    } elseif ($XmlDoc.urlset) {
        foreach ($urlNode in $XmlDoc.urlset.url) {
            $loc = $urlNode.loc -as [string]
            if ($loc) { $urls += $loc }
        }
    } else {
        Write-Warning "Unrecognized sitemap structure at $BaseUrl"
    }

    return $urls
}

try {
    Write-Host "Downloading sitemap from $SitemapUrl ..."
    $response = Invoke-WebRequest -Uri $SitemapUrl -UseBasicParsing -ErrorAction Stop
    $xmlDoc = [xml]$response.Content
} catch {
    Write-Error "Failed to download sitemap from $SitemapUrl : $_"
    exit 1
}

Write-Host "Parsing sitemap..."
$allUrls = Get-UrlsFromSitemap -XmlDoc $xmlDoc -BaseUrl $SitemapUrl

if ($allUrls.Count -eq 0) {
    Write-Warning "No URLs found in the provided sitemap."
} else {
    Write-Host "Found $($allUrls.Count) URLs. Writing to $OutputPath ..."
    try {
        $allUrls | Sort-Object -Unique | Out-File -FilePath $OutputPath -Encoding UTF8
        Write-Host "Successfully wrote URLs to $OutputPath"
    } catch {
        Write-Error "Failed to write output file: $_"
        exit 1
    }
}