# Prevent Crawling of WordPress JSON API Endpoints
User-agent: *
Disallow: /wp-json/
Disallowp: /?rest_route=

# Block Search URLs /search/ and /?s=
User-agent: *
Disallow: /search/
Disallow: /?s=

#Block Parameters
User-agent: *
Disallow: *?s=*
Disallow: *?p=*
Disallow: *&p=*
Disallow: *&preview=*

#Block Feed
User-agent: *
Disallow: /feed/
Disallow: /feed/$
Disallow: /comments/feed
Disallow: */feed
Disallow: */feed$
Disallow: /?feed=
Disallow: /wp-feed

#Block Spam Directories
User-agent: *
Disallow: /trackback/
Disallow: */comments$
Disallow: */trackback
Disallow: */trackback$
Disallow: /wp-comments
Disallow: /wp-trackback
Disallow: */replytocom=

# Block archive.org bots
User-agent: ia_archiver
Disallow: /
User-agent: archive.org_bot
Disallow: /
User-agent: ia_archiver-web.archive.org
Disallow: /

# Block Chatgpt
User-agent: CCBot
Disallow: /

# Block Cart Page
User-agent: *
Disallow: /cart/

# Block Checkout Page
User-agent: *
Disallow: /checkout/

# Block My Account Page
User-agent: *
Disallow: /my-account/

# Block Login Page
User-agent: *
Disallow: /login/

# Block Woocommerce Parameters
User-agent: *
Disallow: /*?orderby=price
Disallow: /*?orderby=rating
Disallow: /*?orderby=date
Disallow: /*?orderby=price-desc
Disallow: /*?orderby=popularity
Disallow: /*?filter
Disallow: /*?orderby=title
Disallow: /*?orderby=desc
Disallow: /*add-to-cart=*
Disallow: /*add_to_wishlist=*
Disallow: /*?paged=&count=*
Disallow: /*?count=*

# Yoast Sitemap Link
Sitemap: https://vitahaute.com/post-sitemap.xml
Sitemap: https://vitahaute.com/page-sitemap.xml

# News Sitemap Link
Sitemap: https://vitahaute.com/sitemap-news.xml

# Product Sitemap Link
Sitemap: https://vitahaute.com/product-sitemap.xml

# Block Ahrefs Crawler
User-agent: AhrefsBot
Disallow: /

# Block Semrush Crawler
User-agent: SemrushBot
Disallow: /
User-agent: SemrushBot-SA
Disallow: /
User-agent: SemrushBot-BA
Disallow: /
User-agent: SemrushBot-SI
Disallow: /
User-agent: SemrushBot-SWA
Disallow: /
User-agent: SemrushBot-CT
Disallow: /
User-agent: SemrushBot-BM
Disallow: /

# Block Moz Crawler
User-agent: rogerbot
Disallow: /

# Block Majestic Crawler
User-agent: MJ12bot
Disallow: /

# Block Xenu Crawler
User-agent: Xenu
Disallow: /

# Allow Google Bot
User-agent: Googlebot
Allow: /

# Allow Google Images Bot
User-agent: Googlebot-Image
Allow: /wp-content/uploads/

# Allow Google Media Partners Bot
User-agent: Mediapartners-Google
Allow: /

# Allow Google AdsBot Bot
User-agent: AdsBot-Google
Allow: /

# Allow Google Mobile Bot
User-agent: AdsBot-Google-Mobile
Allow: /

# Allow Bing Bot
User-agent: Bingbot
Allow: /

# Allow MSN Bot
User-agent: Msnbot
Allow: /

# Allow Yandex Bot
User-agent: Yandex
Allow: /

# Allow MSNBot Media Bot
User-agent: msnbot-media
Allow: /wp-content/uploads/

# Allow Apple Bot
User-agent: Applebot
Allow: /

# Allow Yandex Images Bot
User-agent: YandexImages
Allow: /wp-content/uploads/

# Allow Yahoo Search (Slurp bot)
User-agent: Slurp
Allow: /

# Allow DuckDuckGo Bot
User-agent: DuckDuckBot
Allow: /

# Allow Qwant Bot
User-agent: Qwantify
Allow: /

# Allow Baidu/Sogou/Soso/Youdao Bot
User-agent: Baiduspider
Allow: /
User-agent: Baiduspider/2.0
Allow: /
User-agent: Baiduspider-video
Allow: /
User-agent: Baiduspider-image
Allow: /
User-agent: Sogou spider
Allow: /
User-agent: Sogou web spider
Allow: /
User-agent: Sosospider
Allow: /
User-agent: Sosospider+
Allow: /
User-agent: Sosospider/2.0
Allow: /
User-agent: yodao
Allow: /
User-agent: youdao
Allow: /
User-agent: YoudaoBot
Allow: /
User-agent: YoudaoBot/1.0
Allow: /

# Allow Seznam Bot
User-agent: seznambot
Allow: /

# Allow Naver Bot
User-agent: Naverbot
Allow: /

# Allow Facebook Bot
User-agent: Facebook
Allow: /
User-agent: facebookplatform/1.0
Allow: /
User-agent: facebookexternalhit
Allow: /
User-agent: facebookexternalhit/1.0
Allow: /
User-agent: facebookexternalhit/1.1
Allow: /
User-agent: Facebookscraper
Allow: /
User-agent: Facebot/1.0
Allow: /
User-agent: Visionutils/0.2
Allow: /
User-agent: datagnionbot/1.0
Allow: /

# Allow Instagram Bot
User-agent: Instagrambot
Allow: /

# Allow Telegram Bot
User-agent: TelegramBot
Allow: /

# Allow Whatsapp Bot
User-agent: WhatsApp Bot
Allow: /

# Allow Twitter Bot
User-agent: Twitterbot
Allow: /

# Allow Linkedin Bot
User-agent: LinkedInBot
Allow: /
User-agent: LinkedInBot/1.0
Allow: /

# Allow Discord Bot
User-agent: Discordbot
Allow: /

# Allow Pinterest Bot
User-agent: Pinterest Bot
Allow: /
User-agent: Pinterest/0.1
Allow: /
User-agent: Pinterest/0.2
Allow: /

# Allow Webp Images
User-agent: *
Allow: /*.webp$

# Allow Jpg Images
User-agent: *
Allow: /*.jpg$

# Allow Png Images
User-agent: *
Allow: /*.png$

# Allow Gif Images
User-agent: *
Allow: /*.gif$

# Block Scrapper Bots
User-agent: DotBot
Disallow: /
User-agent: GiftGhostBot
Disallow: /
User-agent: Seznam
Disallow: /
User-agent: PaperLiBot
Disallow: /
User-agent: Genieo
Disallow: /
User-agent: Dataprovider/6.101
Disallow: /
User-agent: DataproviderSiteExplorer
Disallow: /
User-agent: Dazoobot/1.0
Disallow: /
User-agent: Diffbot
Disallow: /
User-agent: DomainStatsBot/1.0
Disallow: /
User-agent: DotBot/1.1
Disallow: /
User-agent: dubaiindex
Disallow: /
User-agent: eCommerceBot
Disallow: /
User-agent: ExpertSearchSpider
Disallow: /
User-agent: Feedbin
Disallow: /
User-agent: Fetch/2.0a
Disallow: /
User-agent: FFbot/1.0
Disallow: /
User-agent: focusbot/1.1
Disallow: /
User-agent: HuaweiSymantecSpider
Disallow: /
User-agent: HuaweiSymantecSpider/1.0
Disallow: /
User-agent: JobdiggerSpider
Disallow: /
User-agent: LemurWebCrawler
Disallow: /
User-agent: LipperheyLinkExplorer
Disallow: /
User-agent: LSSRocketCrawler/1.0
Disallow: /
User-agent: LYT.SRv1.5
Disallow: /
User-agent: MiaDev/0.0.1
Disallow: /
User-agent: Najdi.si/3.1
Disallow: /
User-agent: BountiiBot
Disallow: /
User-agent: Experibot_v1
Disallow: /
User-agent: bixocrawler
Disallow: /
User-agent: bixocrawler TestCrawler
Disallow: /
User-agent: Crawler4j
Disallow: /
User-agent: Crowsnest/0.5
Disallow: /
User-agent: CukBot
Disallow: /
User-agent: Dataprovider/6.92
Disallow: /
User-agent: DBLBot/1.0
Disallow: /
User-agent: Diffbot/0.1
Disallow: /
User-agent: Digg Deeper/v1
Disallow: /
User-agent: discobot/1.0
Disallow: /
User-agent: discobot/1.1
Disallow: /
User-agent: discobot/2.0
Disallow: /
User-agent: discoverybot/2.0
Disallow: /
User-agent: Dlvr.it/1.0
Disallow: /
User-agent: DomainStatsBot/1.0
Disallow: /
User-agent: drupact/0.7
Disallow: /
User-agent: Ezooms/1.0
Disallow: /
User-agent: fastbot crawler beta 2.0
Disallow: /
User-agent: fastbot crawler beta 4.0
Disallow: /
User-agent: feedly social
Disallow: /
User-agent: Feedly/1.0
Disallow: /
User-agent: FeedlyBot/1.0
Disallow: /
User-agent: Feedspot
Disallow: /
User-agent: Feedspotbot/1.0
Disallow: /
User-agent: Clickagy Intelligence Bot v2
Disallow: /
User-agent: classbot
Disallow: /
User-agent: CISPA Vulnerability Notification
Disallow: /
User-agent: CirrusExplorer/1.1
Disallow: /
User-agent: Checksem/Nutch-1.10
Disallow: /
User-agent: CatchBot/5.0
Disallow: /
User-agent: CatchBot/3.0
Disallow: /
User-agent: CatchBot/2.0
Disallow: /
User-agent: CatchBot/1.0
Disallow: /
User-agent: CamontSpider/1.0
Disallow: /
User-agent: Buzzbot/1.0
Disallow: /
User-agent: Buzzbot
Disallow: /
User-agent: BusinessSeek.biz_Spider
Disallow: /
User-agent: BUbiNG
Disallow: /
User-agent: 008/0.85
Disallow: /
User-agent: 008/0.83
Disallow: /
User-agent: 008/0.71
Disallow: /
User-agent: ^Nail
Disallow: /
User-agent: FyberSpider/1.3
Disallow: /
User-agent: findlinks/1.1.6-beta5
Disallow: /
User-agent: g2reader-bot/1.0
Disallow: /
User-agent: findlinks/1.1.6-beta6
Disallow: /
User-agent: findlinks/2.0
Disallow: /
User-agent: findlinks/2.0.1
Disallow: /
User-agent: findlinks/2.0.2
Disallow: /
User-agent: findlinks/2.0.4
Disallow: /
User-agent: findlinks/2.0.5
Disallow: /
User-agent: findlinks/2.0.9
Disallow: /
User-agent: findlinks/2.1
Disallow: /
User-agent: findlinks/2.1.5
Disallow: /
User-agent: findlinks/2.1.3
Disallow: /
User-agent: findlinks/2.2
Disallow: /
User-agent: findlinks/2.5
Disallow: /
User-agent: findlinks/2.6
Disallow: /
User-agent: FFbot/1.0
Disallow: /
User-agent: findlinks/1.0
Disallow: /
User-agent: findlinks/1.1.3-beta8
Disallow: /
User-agent: findlinks/1.1.3-beta9
Disallow: /
User-agent: findlinks/1.1.4-beta7
Disallow: /
User-agent: findlinks/1.1.6-beta1
Disallow: /
User-agent: findlinks/1.1.6-beta1 Yacy
Disallow: /
User-agent: findlinks/1.1.6-beta2
Disallow: /
User-agent: findlinks/1.1.6-beta3
Disallow: /
User-agent: findlinks/1.1.6-beta4
Disallow: /
User-agent: bixo
Disallow: /
User-agent: bixolabs/1.0
Disallow: /
User-agent: Crawlera/1.10.2
Disallow: /
User-agent: Dataprovider Site Explorer
Disallow: /

# Block Php Files
User-agent: *
Disallow: /*.php$

# Block Html Files
User-agent: *
Disallow: /*.html$

# Block DOCX Files
User-agent: *
Disallow: /*.docx$

# Block PDF Files
User-agent: *
Disallow: /*.pdf$