Generate a Basic `sitemap.xml` for Static Sites

#!/bin/bash

# Configuration
BASE_URL="https://yourwebsite.com"
SITEMAP_FILE="sitemap.xml"
TARGET_DIR="." # Directory to scan for HTML files (e.g., 'public', 'dist')

echo "Generating sitemap.xml for $BASE_URL from directory $TARGET_DIR..."

# Start sitemap XML structure
echo '<?xml version="1.0" encoding="UTF-8"?>' > "$SITEMAP_FILE"
echo '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' >> "$SITEMAP_FILE"

# Find HTML files and add them to the sitemap
find "$TARGET_DIR" -type f -name "*.html" | while read -r file; do
  # Get path relative to TARGET_DIR, remove .html extension
  # Example: ./public/about/index.html -> about/index
  relative_path_no_ext=$(echo "$file" | sed "s|^$TARGET_DIR/||; s|\.html$||")
  
  # Determine the final URL segment
  final_url_segment=""
  if [[ "$relative_path_no_ext" == "index" ]]; then
    final_url_segment="" # 'index.html' at root or subdirectory means just the directory URL
  elif [[ "$relative_path_no_ext" =~ /index$ ]]; then
    # if it's 'about/index', it should be 'about'
    final_url_segment=$(echo "$relative_path_no_ext" | sed 's|/index$||')
  else
    final_url_segment="$relative_path_no_ext"
  fi

  # Construct the full URL
  if [[ -n "$final_url_segment" ]]; then
    url="${BASE_URL}/${final_url_segment}"
  else
    url="${BASE_URL}/" # For the root index.html or sub-directory index.html
  fi
  
  # Clean up any potential double slashes (e.g., if BASE_URL already ends with /)
  url=$(echo "$url" | sed 's|//|/|g')
  
  echo "  <url>" >> "$SITEMAP_FILE"
  echo "    <loc>$url</loc>" >> "$SITEMAP_FILE"
  echo "    <lastmod>$(date -r "$file" +"%Y-%m-%d")</lastmod>" >> "$SITEMAP_FILE"
  echo "    <changefreq>weekly</changefreq>" >> "$SITEMAP_FILE"
  echo "    <priority>0.8</priority>" >> "$SITEMAP_FILE"
  echo "  </url>" >> "$SITEMAP_FILE"
done

# End sitemap XML structure
echo '</urlset>' >> "$SITEMAP_FILE"

echo "Sitemap generated successfully at $SITEMAP_FILE"