# /robots.txt file for http://www.joke.co.uk/

User-agent: *
Disallow: /starwars/halloween1/		# out of date
Disallow: /halloween1/			# out of date
Disallow: /facemask/			# out of date
Disallow: /christmas/			# out of date
Disallow: /acatalog/			# out of date
Disallow: /epages/			# out of date
Disallow: /powerrangers/		# out of date
Disallow: /ftp/				# out of date
Disallow: /products/			# out of date
Disallow: /management/			# god knows what this is!
Disallow: /military/			# god knows what this is!
Disallow: /black_dog/			# god knows what this is!
Disallow: farea.htm			# god knows what this is!
Disallow: fetch.shtml			# old joke site
Disallow: login.asp			# old joke site
Disallow: default.asp			# old joke site
Disallow: showPart_large.asp		# old joke site
Disallow: showPart_largeDD.asp		# old joke site
Disallow: prodpage.asp		# old joke site
Disallow: prodpagesearch.asp		# old joke site
Disallow: /links/			# no index no follow
Disallow: /assets/			# no index no follow
Disallow: /storefront/keywordsearch/	# disallow to stop bots indexing search
Disallow: /storefront/viewprofile/	# disallow to stop bots indexing stuff
Disallow: /storefront/checkout/		# disallow to stop bots indexing checkout
Disallow: /storefront/main/		# disallow to stop bots indexing checkout
Disallow: /*.pdf$			# not to crawl .pdf's
Disallow: /*jsessionid=		# dont index pages with session ids

User-agent: http://www.almaden.ibm.com/cs/crawler
Disallow: /

User-agent: e-SocietyRobot
Disallow: /