# AWSTATS SEARCH ENGINES DATABASE #------------------------------------------------------- # If you want to add a Search Engine to extend AWStats database detection capabilities, # you must add an entry in SearchEnginesSearchIDOrder and in SearchEnginesHashIDLib. # An entry if known in SearchEnginesKnownUrl is also welcome. #------------------------------------------------------- # $Revision: 1.5 $ - $Author: eldy $ - $Date: 2002/09/30 13:45:51 $ #package AWSSE; # SearchEnginesSearchIDOrder # This list is used to know in which order to search Search Engines IDs (Most # frequent one are first in this list to increase detect speed). # Note: Browsers IDs are in lower case and ' ' and '+' are changed into '_' #----------------------------------------------------------------- @SearchEnginesSearchIDOrder=( # Japanese Search Engines # Added by Ryu 2002.10.05 "biglobe\.ne\.jp", "goo\.ne\.jp", "nifty\.com", "msn\.co\.jp", "search\.odn\.ne\.jp", "excite\.co\.jp", "naver\.co\.jp", "fresyeye\.com", "tocc\.co\.jp", "infoseek\.co\.jp", # Major internationnal search engines "google\.", "msn\.", "voila\.", "yahoo\.", "lycos\.", "altavista\.", "search\.terra\.", "alltheweb\.com", "netscape\.", "dmoz\.org", "search\.aol\.co", "www\.search\.com", "overture\.com", # Replace "goto\.com","Goto.com", # Minor internationnal search engines "northernlight\.", "hotbot\.", "kvasir\.", "webcrawler\.", "metacrawler\.", "go2net\.com", "(^|\.)go\.com", "euroseek\.", "excite\.", "lokace\.", "spray\.", "netfind\.aol\.com", "recherche\.aol\.fr", "nbci\.com/search", "askjeeves\.", "mamma\.", "dejanews\.", "search\.dogpile\.com", "wisenut\.com", # Minor brazilian search engines "engine\.exe", "miner\.bol\.com\.br", # Minor danish search-engines "opasia\.dk", "danielsen\.com", # Minor dutch search engines "ilse\.","vindex\.", # Minor english search engines "splut\.","ukplus\.","mirago\.","ukindex\.co\.uk","ukdirectory\.", # Minor finnish search engines "haku\.www\.fi", # Minor french search engines "nomade\.fr/","ctrouve\.","francite\.","\.lbb\.org","rechercher\.libertysurf\.fr", # Minor german search engines "fireball\.de","infoseek\.de","suche\.web\.de","meta\.ger", # Minor hungarian search engines "heureka\.hu","vizsla\.origo\.hu/katalogus?","vizsla\.origo\.hu","lapkereso\.hu","goliat\.hu","index\.hu","wahoo\.hu","freeweb\.hu","webmania\.hu","search\.internetto\.hu", # Minor italian search engines "virgilio\.it", # Minor norvegian search engines "sok\.start\.no", # Minor swedish search engines "evreka\.passagen\.se", # Minor czech search engines "atlas\.cz","seznam\.cz","quick\.cz","centrum\.cz","najdi\.to","redbox\.cz", # Other "search\..*com" ); # SearchEnginesHashIDLib # List of search engines names # "match_string_in_url_that_identify_engine", "search_engine_name", #----------------------------------------------------------------- %SearchEnginesHashIDLib=( # Japanese Search Engines # Added by Ryu 2002.10.05 "biglobe\.ne\.jp","Biglobe", "goo\.ne\.jp","Goo", "nifty\.com","Nifty", "msn\.co\.jp","MSN Japan", "search\.odn\.ne\.jp","ODN", "excite\.co\.jp","Excite Japan", "naver\.co\.jp","Naver", "fresheye\.com","Fresheye", "tocc\.co\.jp","TOCC/Search", "infoseek\.co\.jp","Infoseek", # Major internationnal search engines "google\.","Google", "msn\.","MSN", "voila\.", "Voila", "yahoo\.","Yahoo", "lycos\.","Lycos", "altavista\.","AltaVista", "search\.terra\.","Terra", "alltheweb\.com","AllTheWeb", "netscape\.","Netscape", "dmoz\.org","DMOZ", "search\.aol\.co","AOL", "www\.search\.com","Search.com", "overture\.com","Overture", # Replace "goto\.com","Goto.com", # Minor internationnal search engines "northernlight\.","NorthernLight", "hotbot\.","Hotbot", "kvasir\.","Kvasir", "webcrawler\.","WebCrawler", "metacrawler\.","MetaCrawler (Metamoteur)", "go2net\.com","Go2Net (Metamoteur)", "(^|\.)go\.com","Go.com", "euroseek\.","Euroseek", "excite\.","Excite", "lokace\.", "Lokace", "spray\.","Spray", "netfind\.aol\.com","AOL", "recherche\.aol\.fr","AOL", "nbci\.com/search","NBCI", "askjeeves\.","Ask Jeeves", "mamma\.","Mamma", "dejanews\.","DejaNews", "search\.dogpile\.com","Dogpile", "wisenut\.com","WISENut", # Minor brazilian search engines "engine\.exe","Cade", "miner\.bol\.com\.br","Meta Miner", # Minor danish search-engines "opasia\.dk","Opasia", "danielsen\.com","Thor (danielsen.com)", # Minor dutch search engines "ilse\.","Ilse","vindex\.","Vindex\.nl", # Minor english search engines "splut\.","Splut", "ukplus\.", "UKPlus", "mirago\.", "Mirago", "ukindex\.co\.uk", "UKIndex", "ukdirectory\.","UK Directory", # Minor finnish search engines "haku\.www\.fi","Ihmemaa", # Minor french search engines "nomade\.fr/","Nomade", "ctrouve\.","C'est trouvé", "francite\.","Francité", "\.lbb\.org", "LBB", "rechercher\.libertysurf\.fr", "Libertysurf", # Minor german search engines "fireball\.de","Fireball", "infoseek\.de","Infoseek", "suche\.web\.de","Web.de", "meta\.ger","MetaGer", # Minor hungarian search engines "heureka\.hu","Heureka", "vizsla\.origo\.hu/katalogus?","Origo-Vizsla-Katalógus", "vizsla\.origo\.hu","Origo-Vizsla", "lapkereso\.hu","Startlapkereső", "goliat\.hu","Góliát", "index\.hu","Index", "wahoo\.hu","Wahoo", "freeweb\.hu","FreeWeb", "webmania\.hu","webmania.hu", "search\.internetto\.hu","Internetto Kereső", # Minor italian search engines "virgilio\.it","Virgilio", # Minor norvegian search engines "sok\.start\.no","start.no", # Minor swedish search engines "evreka\.passagen\.se","Evreka", # Minor czech search engines "atlas\.cz","Atlas.cz", "seznam\.cz","Seznam.cz", "quick\.cz","Quick.cz", "centrum\.cz","Centrum.cz","najdi\.to","Najdi.to","redbox\.cz","RedBox.cz", # Other "search\..*com","Other search engines" ); # SearchEnginesKnownUrl # Search engines known rules to extract keywords from a referrer URL #------------------------------------------------- %SearchEnginesKnownUrl=( # Added by Ryu 2002.10.05 # Japanese Search Engines "biglobe\.","q=", "goo\.","MT=", "asearch\.nifty\.","q=", "www\.nifty\.","Text=", "msn\.co\.jp\/results\.asp","q=", "msn\.co\.jp\/spbasic\.html","MT=", "odn\.","Querystring=", "naver\.co\.jp","query=", "fresheye.com","kw=", "tocc\.co\.jp","QRY=", "infoseek\.co\.jp","qt=", # Most common search engines "yahoo\.","p=", "altavista\.","q=", "msn\.","q=", "voila\.","kw=", "lycos\.","query=", "search\.terra\.","query=", "google\.","(p|q)=", "alltheweb\.com","q(|uery)=", "netscape\.","search=", "northernlight\.","qr=", "dmoz\.org","search=", "search\.aol\.co","query=", "www\.search\.com","q=", "kvasir\.", "q=", # Others "askjeeves\.","ask=", "hotbot\.","mt=", "metacrawler\.","general=", "go2net\.com","general=", "(^|\.)go\.com","qt=", "overture\.com","keywords=", "euroseek\.","query=", "excite\.","search=", "spray\.","string=", "nbci\.com/search","keyword=", "mamma\.","query=", "search\.dogpile\.com", "q=", "wisenut\.com","query=", "virgilio\.it","qs=", "webcrawler","searchText=", "engine\.exe","p1=", "miner\.bol\.com\.br","q=", # Minor brazilian search engines "opasia\.dk","q=", "danielsen\.com","q=", # Minor danish search engines "ilse\.","search_for=", "vindex\.","in=", # Minor dutch search engines "splut\.","pattern=", "ukplus\.", "search=", "mirago\.", "txtSearch=", # Minor english search engines "ukindex\.co\.uk", "stext=", "ukdirectory\.","k=", # Minor english search engines "haku\.www\.fi","w=", # Minor finnish search engines "nomade\.fr/","s=", "francite\.","name=", # Minor french search engines "fireball\.de","q=", "infoseek\.de","qt=", "suche\.web\.de","su=", # Minor german search engines "heureka\.hu","heureka=", "vizsla\.origo\.hu/katalogus?","q=", "vizsla\.origo\.hu","search=", "lapkereso\.hu","keres.php", "goliat\.hu","KERESES=", "index\.hu","search.php3", "wahoo\.hu","q=", "freeweb\.hu","KERESES=", "search\.internetto\.hu","searchstr=", #Minor hungarian search engines "sok\.start\.no", "q=", # Minor norvegian search engines "evreka\.passagen\.se","q=", # Minor swedish search engines "atlas\.cz","searchtext=", "seznam\.cz","w=", "ftxt\.quick\.cz","query=", "centrum\.cz","q=", "najdi\.to","dotaz=", "redbox.cz","srch=" # Minor czech search engines ); # If no rules are known, this will be used to clean URL of not keyword parameters. # Words after "look=" are added by Ryu for Japanese Search Engines on 2002.10.05. @WordsToCleanSearchUrl= ("act=","annuaire=","btng=","categoria=","cfg=","cof=","cou=","cp=","dd=","domain=","dt=","dw=","enc=","exec=","geo=","hc=","height=","hl=","hq=","hs=","id=","kl=","lang=","loc=","lr=","matchmode=","medor=","message=","meta=","mode=","order=","page=","par=","pays=","pg=","pos=","prg=","qc=","refer=","sa=","safe=","sc=","sort=","src=","start=","style=","stype=","sum=","tag=","temp=","theme=","url=","user=","width=","what=","\\.x=","\\.y=","y=","look=","target=","collection=","showSummary=","perPage=","next=","DB=","OPE=","Max=","base=","submit=","SearchType=","SESSIONID=","QUERYRULE=","DISPLAYMIN=","RELURLSWITCH=","SORT=","start=","direct=","relwd=","lk=","svx=","nh=","internet=","DC=","submit0=","DEST=","where=","CCM=","NRS=","UNI=","UD0=","FRS=","sv=","rf=","oq=","col=","act\.search="); # Never put the following exclusion ("ask=","claus=","general=","kw=","keyword=","keywords=","MT","p=","q=","qr=","qt=","query=","s=","search=","searchText=","string=","su=","w=") because they are strings that contain keywords we're looking for. 1;