Param(
    [Parameter(Mandatory=$true)][string]$BaseUrl,
    [Parameter(Mandatory=$false)][string]$OutputPath = "sitemap.xml",
    [int]$MaxDepth = 3,
    [int]$RequestTimeout = 30
)

# Normalize base URL
if (-not $BaseUrl.StartsWith('http')) {
    $BaseUrl = "https://$BaseUrl"
}
$baseUri = [System.Uri]::new($BaseUrl)

# Set up collections
$visited = New-Object System.Collections.Generic.HashSet[string] ([System.StringComparer]::OrdinalIgnoreCase)
$foundUrls = New-Object System.Collections.Generic.List[string]

function Get-AbsoluteUrl {
    param(
        [string]$href,
        [System.Uri]$parentUri
    )
    try {
        $uri = $null
        if ([System.Uri]::IsWellFormedUriString($href, [System.UriKind]::Absolute)) {
            $uri = [System.Uri]::new($href)
        } else {
            $uri = New-Object System.Uri($parentUri, $href)
        }
        return $uri
    } catch {
        return $null
    }
}

function Extract-Links {
    param(
        [string]$html,
        [System.Uri]$parentUri
    )
    $links = @()
    # Use regex to pull href values from <a> tags
    $hrefPattern = '(?i)<a\s+(?:[^>]*?\s+)?href\s*=\s*["'']?([^"''\s>]+)'
    foreach ($match in [regex]::Matches($html, $hrefPattern)) {
        $rawHref = $match.Groups[1].Value
        $absUri = Get-AbsoluteUrl -href $rawHref -parentUri $parentUri
        if ($null -ne $absUri -and $absUri.Scheme -match '^https?$') {
            # Only keep URLs that belong to the same host
            if ($absUri.Host -ieq $baseUri.Host) {
                $links += $absUri.AbsoluteUri
            }
        }
    }
    return $links
}

function Crawl {
    param(
        [string]$url,
        [int]$depth
    )
    if ($depth -gt $MaxDepth) { return }
    if (-not $visited.Add($url)) { return } # already visited

    try {
        $response = Invoke-WebRequest -Uri $url -Method GET -TimeoutSec $RequestTimeout -ErrorAction Stop
        $foundUrls.Add($url) | Out-Null
        $pageLinks = Extract-Links -html $response.Content -parentUri ([System.Uri]::new($url))
        foreach ($link in $pageLinks) {
            Crawl -url $link -depth ($depth + 1)
        }
    } catch {
        # Silently ignore failures (404, 500, timeout, etc.)
    }
}

# Start crawling from base URL
Crawl -url $BaseUrl -depth 0

# Remove duplicates and sort
$uniqueUrls = $foundUrls | Sort-Object -Unique

# Build XML sitemap
$sitemapNs = "http://www.sitemaps.org/schemas/sitemap/0.9"
$xmlSettings = New-Object System.Xml.XmlWriterSettings
$xmlSettings.Indent = $true
$xmlSettings.Encoding = [System.Text.Encoding]::UTF8

$writer = [System.Xml.XmlWriter]::Create($OutputPath, $xmlSettings)
$writer.WriteStartDocument()
$writer.WriteStartElement("urlset", $sitemapNs)

$now = Get-Date -Format "yyyy-MM-ddTHH:mm:ssK"

foreach ($loc in $uniqueUrls) {
    $writer.WriteStartElement("url")
    $writer.WriteElementString("loc", $loc)
    $writer.WriteElementString("lastmod", $now)
    $writer.WriteEndElement() # url
}

$writer.WriteEndElement() # urlset
$writer.WriteEndDocument()
$writer.Flush()
$writer.Close()

Write-Host "Sitemap generated at $OutputPath with $($uniqueUrls.Count) URLs."