param(
    [Parameter(Mandatory=$true)]
    [string]$BaseUrl,
    
    [int]$MaxDepth = 3,
    
    [string]$OutputPath = "sitemap.xml"
)

# Normalize base URL (ensure trailing slash for relative resolution)
if (-not $BaseUrl.EndsWith('/')) { $BaseUrl = "$BaseUrl/" }

# Create collections for tracking
$visited = New-Object System.Collections.Generic.HashSet[string]
$queue = New-Object System.Collections.Generic.Queue[object]

# Enqueue the start URL with depth 0
$queue.Enqueue([pscustomobject]@{ Url = $BaseUrl; Depth = 0 })

while ($queue.Count -gt 0) {
    $item = $queue.Dequeue()
    $url = $item.Url
    $depth = $item.Depth

    # Skip if already visited or beyond depth limit
    if ($visited.Contains($url) -or $depth -gt $MaxDepth) { continue }

    try {
        # Attempt a GET request (HTML needed for link extraction)
        $response = Invoke-WebRequest -Uri $url -UseBasicParsing -Method Get -ErrorAction Stop
    } catch {
        # On error (404, timeout, etc.), just skip processing this URL
        continue
    }

    # Mark as visited
    $null = $visited.Add($url)

    # Only parse HTML content-type pages for further links
    if ($response.Content -and $response.ContentType -match 'text/html') {
        # Simple regex to extract href values; may capture duplicates/invalid values
        $hrefs = [regex]::Matches($response.Content, 'href\s*=\s*["'']([^"''>#]+)["'']', 'IgnoreCase') | 
                 ForEach-Object { $_.Groups[1].Value.Trim() }

        foreach ($href in $hrefs) {
            # Skip empty, mailto:, javascript:, tel:, #
            if ([string]::IsNullOrWhiteSpace($href) -or $href -match '^\s*(mailto:|javascript:|tel:|#)') { continue }

            # Resolve relative URLs against the current page URL
            try {
                $resolved = (New-Object System.Uri($url, $href)).AbsoluteUri
            } catch {
                continue
            }

            # Ensure we stay within the original domain (optional but typical for sitemaps)
            try {
                $baseHost = (New-Object System.Uri($BaseUrl)).Host
                $resolvedHost = (New-Object System.Uri($resolved)).Host
                if ($resolvedHost -ne $baseHost) { continue }
            } catch { continue }

            # Enqueue for later crawling if not visited yet
            if (-not $visited.Contains($resolved)) {
                $queue.Enqueue([pscustomobject]@{ Url = $resolved; Depth = $depth + 1 })
            }
        }
    }
}

# Build the sitemap XML
$ns = "http://www.sitemaps.org/schemas/sitemap/0.9"
$xmlDoc = New-Object System.Xml.Linq.XDocument
$urlset = New-Object System.Xml.Linq.XElement ([System.Xml.Linq.XName]::Get("urlset", $ns))

foreach ($loc in $visited) {
    $urlElement = New-Object System.Xml.Linq.XElement ([System.Xml.Linq.XName]::Get("url", $ns))
    $locElement = New-Object System.Xml.Linq.XElement ([System.Xml.Linq.XName]::Get("loc", $ns), $loc)
    $urlElement.Add($locElement)
    $urlset.Add($urlElement)
}

$xmlDoc.Add($urlset)

# Save to file
try {
    $xmlDoc.Save($OutputPath)
    Write-Host "Sitemap generated at '$OutputPath' with $($visited.Count) URLs."
} catch {
    Write-Error "Failed to write sitemap file: $_"
}