#!/bin/bash
# =====================================================================
# sitemap-test.sh - Test sitemap.xml
# =====================================================================
# This script tests that sitemap.xml is properly generated and contains
# all expected URLs
# =====================================================================

set -e

echo "=== Testing sitemap.xml ==="

# Get the base URL from the command line or use the default
if [ -z "$1" ]; then
  BASE_URL="http://localhost:8080"
else
  BASE_URL="$1"
fi

# Check if sitemap.xml exists
if ! curl -s "$BASE_URL/sitemap.xml" > /dev/null; then
  echo "❌ sitemap.xml not found at $BASE_URL/sitemap.xml"
  exit 1
fi

# Download sitemap.xml
echo "Downloading sitemap.xml from $BASE_URL/sitemap.xml"
SITEMAP=$(curl -s "$BASE_URL/sitemap.xml")

# Count URLs in sitemap
URL_COUNT=$(echo "$SITEMAP" | grep -c "<url>")
echo "Found $URL_COUNT URLs in sitemap.xml"

# Check if sitemap contains at least 10 URLs
if [ "$URL_COUNT" -lt 10 ]; then
  echo "❌ sitemap.xml contains fewer than 10 URLs ($URL_COUNT)"
  exit 1
fi

# Check if important pages are included
IMPORTANT_PAGES=(
  "$BASE_URL/index.html"
  "$BASE_URL/stories/index.html"
  "$BASE_URL/stories/open-source-success.html"
  "$BASE_URL/stories/viperwire.html"
  "$BASE_URL/one-pager-tools/csv-tool.html"
)

MISSING_PAGES=0
for page in "${IMPORTANT_PAGES[@]}"; do
  # Extract domain-relative path
  rel_path="${page#$BASE_URL/}"
  
  # Check if page is in sitemap
  if ! echo "$SITEMAP" | grep -q "$rel_path"; then
    echo "❌ Important page missing from sitemap: $rel_path"
    MISSING_PAGES=$((MISSING_PAGES + 1))
  fi
done

if [ "$MISSING_PAGES" -gt 0 ]; then
  echo "❌ $MISSING_PAGES important pages missing from sitemap.xml"
  exit 1
fi

echo "✅ sitemap.xml test passed"
exit 0