# robots.txt
# @url: https://www.opalesurfcasting.net
# @generator: SPIP 3.2.19
# @template: squelettes/robots.txt.html

# ------------------------------------------------------------
# 1) Règles SPIP standards (inchangées)
# ------------------------------------------------------------

User-agent: *
Allow: /edupict/www/images/
Allow: /local/cache-css/
Allow: /local/cache-js/
Allow: /local/cache-vignettes/
Allow: /local/cache-TeX/
Allow: /local/cache-gd2/
Allow: /plugins-dist/medias/prive/vignettes/
Allow: /squelettes/css/
Allow: /squelettes/js/
Allow: /squelettes/tarteaucitron/
Allow: /squelettes/inclure/lamaillebanner/
Allow: /squelettes/NAVPICS/
Allow: /squelettes/epnjs/dist/epn.js
Allow: /squelettes/puce.gif
Allow: /geoportail/

Disallow: /local/
Disallow: /plugins-dist/
Disallow: /lib/
Disallow: /plugins/
Disallow: /prive/
Disallow: /squelettes-dist/
Disallow: /squelettes/

# Exceptions SPIP
Allow: /prive/javascript/SearchHighlight.js
Allow: /plugins/auto/saisies/v3.56.6/css/saisies.css
Allow: /plugins/auto/saisies/v3.56.6/javascript/saisies.js

# ------------------------------------------------------------
# 2) Règles images pour Googlebot-Image (inchangé)
# ------------------------------------------------------------

User-agent: Googlebot-Image
Disallow: /IMG/jpg/la_rochelle_pallice_pont_de_re.jpg
Disallow: /IMG/merlan.jpg
Disallow: /IMG/jpg/merlan-oye-plage.jpg
Disallow: /local/cache-vignettes/L300xH225/xmerlan-oye-plage-8ab70.jpg.pagespeed.ic.YAamfNHyvXrXDBFSUtII.jpg

# ------------------------------------------------------------
# 3) Baiduspider-image (inchangé)
# ------------------------------------------------------------

User-agent: Baiduspider-image
Allow: /edupict/www/images/
Allow: /local/cache-css/
Allow: /local/cache-js/
Allow: /local/cache-vignettes/
Allow: /local/cache-TeX/
Allow: /plugins-dist/medias/prive/vignettes/
Allow: /squelettes/css/
Allow: /squelettes/mes_fonctions_geo_carto.js
Allow: /IMG/logo_opale_surfcasting.jpg

Disallow: /local/
Disallow: /plugins-dist/
Disallow: /lib/
Disallow: /plugins/
Disallow: /prive/
Disallow: /squelettes-dist/
Disallow: /squelettes/
Disallow: /IMG/jpg/
Disallow: /IMG/*.jpg$
Disallow: /wetterzentrale/

# ------------------------------------------------------------
# 4) Crawlers indésirables classiques (scrapers, SEO spammers)
# ------------------------------------------------------------

User-agent: WebsiteOutlook
User-agent: CuteStat
User-agent: VelenPublicWebCrawler
User-agent: MixnodeCache
User-agent: Ltx71
User-agent: TurnitinBot
User-agent: Seobility
User-agent: The Knowledge AI
User-agent: YisouSpider
User-agent: serpstatbot
User-agent: DataForSeoBot
Disallow: /

# ------------------------------------------------------------
# 5) IA – BLOQUÉES (entraînement, index massifs, scrapers IA)
# ------------------------------------------------------------

# OpenAI (entraînement + search)
User-agent: GPTBot
User-agent: CCBot
User-agent: OAI-SearchBot

# Anthropic / Claude
User-agent: ClaudeBot
User-agent: Claude-Web
User-agent: Claude-SearchBot
User-agent: Claude-User

# Perplexity – crawlers
User-agent: PerplexityBot
User-agent: PPLCrawl
User-agent: PPLScout

# Mistral
User-agent: MistralBot

# Google AI (hors indexation classique)
User-agent: Google-Extended
User-agent: Google-AI
User-agent: GoogleOther
User-agent: GoogleOther-Image
User-agent: GoogleOther-Video

# Amazon / Meta / autres scrapers IA
User-agent: Amazonbot
User-agent: Bytespider
User-agent: img2dataset
User-agent: cohere-ai

# Commercial NLP scrapers
User-agent: Diffbot
User-agent: AwarioRssBot
User-agent: AwarioSmartBot
User-agent: FriendlyCrawler
User-agent: Meltwater
User-agent: magpie-crawler
User-agent: omgili
User-agent: omgilibot
User-agent: peer39_crawler
User-agent: peer39_crawler/1.0
User-agent: PiplBot
User-agent: scoop.it
User-agent: Seekr
User-agent: YouBot

Disallow: /

# ------------------------------------------------------------
# 6) IA – AUTORISÉES (trafic humain réel)
# ------------------------------------------------------------

# ChatGPT – accès déclenché par un utilisateur (Browser Mode)
User-agent: ChatGPT-User
Allow: /

# Perplexity – trafic humain via WebView / App
User-agent: Perplexity-User
Allow: /

User-agent: PerplexityUser
Allow: /

User-agent: Perplexity-WebView
Allow: /

User-agent: PerplexityApp
Allow: /

# ------------------------------------------------------------
# 7) Sitemap
# ------------------------------------------------------------
Sitemap: https://www.opalesurfcasting.net/sitemap.xml