#!/bin/bash
# Check key pages for technical SEO issues
URLS=(
  "/"
  "/drogentest/"
  "/drogentest/urintest/"
  "/drogentest/speicheltest/"
  "/informationen/"
  "/einfach-aufgeklaert/"
  "/erklaert-aufgeklaert/"
  "/suchtberatung/"
  "/blog/"
  "/einfach-aufgeklaert/drogenpolitik/deutschland/"
  "/einfach-aufgeklaert/drogenpolitik/usa/kalifornien/"
  "/produkt/drugexpert-10-drogentest10-parameter-1-st/"
  "/substanz/thc-cannabis/"
  "/tag/cannabis/"
  "/category/ratgeber/"
  "/verkehrsmonitor/"
  "/ratgeber/nachweiszeiten-drogen-thc-kokain-amphetamin/"
  "/sitemap/"
)

for path in "${URLS[@]}"; do
  url="https://www.drogentest-online.de${path}"
  html=$(curl -s -L "$url" 2>/dev/null)
  
  title=$(echo "$html" | grep -oP '<title>\K[^<]+' | head -1)
  title_len=${#title}
  desc=$(echo "$html" | grep -oP 'meta name="description" content="\K[^"]+' | head -1)
  desc_len=${#desc}
  h1=$(echo "$html" | grep -oP '<h1[^>]*>\K[^<]+' | head -1)
  h1_count=$(echo "$html" | grep -cP '<h1[^>]*>')
  canonical=$(echo "$html" | grep -oP 'rel="canonical" href="\K[^"]+' | head -1)
  canonical_count=$(echo "$html" | grep -cP 'rel="canonical"')
  has_og=$(echo "$html" | grep -c 'og:title')
  has_twitter=$(echo "$html" | grep -c 'twitter:card')
  has_schema=$(echo "$html" | grep -c 'application/ld+json')
  has_robots=$(echo "$html" | grep -oP 'meta name=['\''"]robots['\''"] content="\K[^"]+' | head -1)
  has_hreflang=$(echo "$html" | grep -c 'hreflang=')
  has_kw=$(echo "$html" | grep -c 'meta name="keywords"')
  
  echo "URL: $path"
  echo "  Title ($title_len): $(echo $title | head -c 70)"
  echo "  Desc ($desc_len): $(echo $desc | head -c 80)"
  echo "  H1 ($h1_count): $h1"
  echo "  Canonical ($canonical_count): $(echo $canonical | sed 's|https://www.drogentest-online.de||')"
  echo "  OG:$has_og TW:$has_twitter Schema:$has_schema hreflang:$has_hreflang Keywords:$has_kw"
  [ -n "$has_robots" ] && echo "  Robots: $has_robots"
  echo ""
done