# Robots.txt optimisé 2025 - Riman.fr
# Basé sur la configuration actuelle + optimisations SEO/IA
# Dernière mise à jour: 2025

# ========================================
# RÈGLES GÉNÉRALES (basées sur la prod actuelle)
# ========================================

User-agent: *
Disallow: /wp/wp-admin/
Disallow: /wp/wp-login.php
Disallow: /?s=
Disallow: /?rest_route=
Disallow: /feed/
Disallow: /comments/

# ========================================
# MOTEURS DE RECHERCHE PRINCIPAUX
# ========================================

# Google
User-agent: Googlebot
Crawl-delay: 1
Allow: /

# Bing
User-agent: Bingbot
Crawl-delay: 1
Allow: /

# Yandex
User-agent: YandexBot
Crawl-delay: 2
Allow: /

# ========================================
# SERVICES D'INTELLIGENCE ARTIFICIELLE
# ========================================

# ChatGPT / OpenAI
User-agent: GPTBot
Crawl-delay: 2
Allow: /

# Google Bard / Gemini
User-agent: Google-Extended
Crawl-delay: 2
Allow: /

# Microsoft Copilot
User-agent: msnbot-WebCrawler
Crawl-delay: 2
Allow: /

# Anthropic Claude
User-agent: Claude-Web
Crawl-delay: 2
Allow: /

# Perplexity AI
User-agent: PerplexityBot
Crawl-delay: 2
Allow: /

# Meta AI (Facebook)
User-agent: facebookexternalhit
Crawl-delay: 1
Allow: /

# ========================================
# RÉSEAUX SOCIAUX
# ========================================

# LinkedIn
User-agent: LinkedInBot
Crawl-delay: 1
Allow: /

# Twitter/X
User-agent: Twitterbot
Crawl-delay: 1
Allow: /

# Pinterest
User-agent: Pinterestbot
Crawl-delay: 1
Allow: /

# WhatsApp
User-agent: WhatsApp
Crawl-delay: 1
Allow: /

# ========================================
# PROTECTION AVANCÉE (Bedrock + WordPress)
# ========================================

# Structure Bedrock
Disallow: /app/plugins/
Disallow: /app/themes/
Disallow: /app/cache/
Disallow: /app/debug.log
Disallow: /app/ewww/
Disallow: /app/wflogs/
Disallow: /app/updraft/
Disallow: /app/upgrade/
Disallow: /app/upgrade-temp-backup/

# Uploads sensibles
Disallow: /app/uploads/2024/
Disallow: /app/uploads/batch_imports/
Disallow: /app/uploads/candidatures/
Disallow: /app/uploads/import_temp/
Disallow: /app/uploads/riman-import-logs/

# Fichiers techniques
Disallow: /config/
Disallow: /vendor/
Disallow: /node_modules/
Disallow: /dist/
Disallow: /test-marker.html

# Fichiers WordPress sensibles
Disallow: /wp-config.php
Disallow: /wp-config-sample.php
Disallow: /wp-cron.php
Disallow: /wp-register.php
Disallow: /wp-signup.php
Disallow: /wp-trackback.php
Disallow: /xmlrpc.php
Disallow: /readme.html
Disallow: /license.txt

# Fichiers techniques
Disallow: /*.log$
Disallow: /*.sql$
Disallow: /*.bak$
Disallow: /*.backup$
Disallow: /*.tmp$
Disallow: /*.temp$

# Paramètres d'URL marketing
Disallow: /*?utm_*
Disallow: /*?fbclid=*
Disallow: /*?gclid=*
Disallow: /*?msclkid=*
Disallow: /*?ref=*

# ========================================
# AUTORISATIONS IMPORTANTES
# ========================================

# Ressources publiques (thèmes et uploads)
Allow: /app/themes/*/style.css
Allow: /app/themes/*/js/
Allow: /app/themes/*/css/
Allow: /app/themes/*/assets/

# Images et médias publics
Allow: /app/uploads/*.jpg
Allow: /app/uploads/*.jpeg
Allow: /app/uploads/*.png
Allow: /app/uploads/*.gif
Allow: /app/uploads/*.webp
Allow: /app/uploads/*.svg
Allow: /app/uploads/*.pdf
Allow: /app/uploads/*.css
Allow: /app/uploads/*.js

# ========================================
# BOTS MALVEILLANTS À BLOQUER
# ========================================

User-agent: AhrefsBot
Disallow: /

User-agent: SemrushBot
Disallow: /

User-agent: MJ12bot
Disallow: /

User-agent: DotBot
Disallow: /

User-agent: BLEXBot
Disallow: /

User-agent: PetalBot
Disallow: /

User-agent: SeoCheckBot
Disallow: /

User-agent: SeobilityBot
Disallow: /

# ========================================
# SITEMAP
# ========================================

Sitemap: https://www.riman.fr/sitemap_index.xml
Sitemap: https://www.riman.fr/sitemap.xml

# ========================================
# NOTES
# ========================================

# Ce robots.txt combine:
# - Configuration actuelle fonctionnelle de www.riman.fr
# - Optimisations SEO 2025
# - Support des services d'IA modernes
# - Protection contre le scraping malveillant
# - Structure Bedrock WordPress