# Parameters - adjust these values as needed
$RootPath = "C:\Path\To\Site"          # Local directory containing the website files
$BaseUrl  = "https://example.com"      # Base URL of the site (no trailing slash)

# Gather files that should appear in the sitemap
$FilePatterns = @("*.html","*.htm","*.php","*.aspx","*.asp")
$Files = Get-ChildItem -Path $RootPath -Recurse -File -Include $FilePatterns

# Prepare XML writer settings
$settings = New-Object System.Xml.XmlWriterSettings
$settings.Indent = $true
$settings.Encoding = [System.Text.Encoding]::UTF8

# Create sitemap.xml in the root directory
$SitemapPath = Join-Path $RootPath "sitemap.xml"
$writer = [System.Xml.XmlWriter]::Create($SitemapPath, $settings)

# Write XML header and root element
$writer.WriteStartDocument()
$writer.WriteStartElement("urlset")
$writer.WriteAttributeString("xmlns", "http://www.sitemaps.org/schemas/sitemap/0.9")

foreach ($file in $Files) {
    # Compute relative path from site root
    $relative = $file.FullName.Substring($RootPath.Length).TrimStart('\','/')
    
    # Convert filesystem separators to URL separators and escape each segment
    $segments = $relative -split '[\\/]' | ForEach-Object { [System.Uri]::EscapeDataString($_) }
    $urlPath = ($segments -join '/')
    
    # Special handling for default documents (e.g., index.html at any folder)
    if ($urlPath -match '^(.*?)(index\.html?|default\.aspx?)$') {
        $urlPath = $matches[1].TrimEnd('/')
    }

    $loc = "$BaseUrl/$urlPath".TrimEnd('/')
    $lastMod = $file.LastWriteTimeUtc.ToString("yyyy-MM-ddTHH:mm:ssZ")

    # Write <url> entry
    $writer.WriteStartElement("url")
    $writer.WriteElementString("loc", $loc)
    $writer.WriteElementString("lastmod", $lastMod)
    $writer.WriteElementString("changefreq", "weekly")
    $writer.WriteElementString("priority", "0.5")
    $writer.WriteEndElement() # </url>
}

# Close root element and document
$writer.WriteEndElement() # </urlset>
$writer.WriteEndDocument()
$writer.Flush()
$writer.Close()

Write-Host "Sitemap generated at:" $SitemapPath