ViewVC Help
View File | Revision Log | Show Annotations | View Changeset | Root Listing
root/BBClone/trunk/lib/robot.php
Revision: 320
Committed: Sun Nov 30 11:09:27 2014 UTC (10 years ago) by matthys
File size: 156195 byte(s)
Log Message:
Added TwitterBot detection and icon.

File Contents

# User Rev Content
1 joku 63 <?php
2     /* This file is part of BBClone (A PHP based Web Counter on Steroids)
3     *
4     * SVN FILE $Id$
5     *
6 joku 312 * Copyright (C) 2001-2015, the BBClone Team (see doc/authors.txt for details)
7 joku 63 *
8     * This program is free software: you can redistribute it and/or modify
9     * it under the terms of the GNU General Public License as published by
10     * the Free Software Foundation, either version 3 of the License, or
11     * (at your option) any later version.
12     *
13     * This program is distributed in the hope that it will be useful,
14     * but WITHOUT ANY WARRANTY; without even the implied warranty of
15     * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16     * GNU General Public License for more details.
17     *
18     * See doc/copying.txt for details
19     */
20    
21     /////////////////////
22     // Robot Detection //
23     /////////////////////
24    
25     $robot = array(
26     "1noon" => array(
27     "icon" => "1noon",
28     "title" => "1noon",
29     "rule" => array(
30     "1Noonbot[/ ]([0-9.]{1,10})" => "\\1",
31     "^Yeti$" => ""
32     ),
33     "uri" => ""
34     ),
35     "123people" => array(
36     "icon" => "123people",
37     "title" => "123people",
38     "rule" => array(
39     "123peoplebot[ /]([0-9.]{1,10})" => "\\1"
40     ),
41     "uri" => "http://www.123people.fr/"
42     ),
43     "123spider" => array(
44     "icon" => "robot",
45     "title" => "123Spider",
46     "rule" => array(
47     "123spider-Bot \(Version: ([0-9.]{1,10})" => "\\1"
48     ),
49     "uri" => "http://www.123spider.de/"
50     ),
51     "192com" => array(
52     "icon" => "robot",
53     "title" => "192.com",
54     "rule" => array(
55     "192.comAgent" => ""
56     ),
57     "uri" => "http://www.192.com/"
58     ),
59 matthys 298 "200please" => array(
60     "icon" => "200please",
61     "title" => "200please Crawler",
62     "rule" => array(
63     "200PleaseBot[ /]([0-9.]{1,10})" => "\\1"
64     ),
65     "uri" => "http://www.200please.com/bot"
66     ),
67 joku 63 "2dehands" => array(
68     "icon" => "2dehands",
69     "title" => "2deHands",
70     "rule" => array(
71     "2dehands\.nl" => ""
72     ),
73     "uri" => ""
74     ),
75     "80legs" => array(
76     "icon" => "80legs",
77     "title" => "80legs",
78     "rule" => array(
79     "008[/ ]([0-9.]{1,10}).*80legs" => "\\1"
80     ),
81     "uri" => "http://www.80legs.com/"
82     ),
83     "a1sitemap" => array(
84     "icon" => "a1sitemap",
85     "title" => "A1 Sitemap",
86     "rule" => array(
87     "^A1 Sitemap Generator[ /]([0-9.]{1,10})" => "\\1",
88     "miggibot[ /]([0-9.]{1,10})" => "\\1"
89     ),
90     "uri" => "http://www.micro-sys.dk/products/sitemap-generator/"
91     ),
92     "a2b" => array(
93     "icon" => "a2b",
94     "title" => "A2B",
95     "rule" => array(
96     "www\.a2b\.cc" => ""
97     ),
98     "uri" => "http://www.a2b.cc"
99     ),
100     "abacho" => array(
101     "icon" => "robot",
102     "title" => "Abacho",
103     "rule" => array(
104     "^ABACHOBot" => ""
105     ),
106     "uri" => ""
107     ),
108     "abcdatos" => array(
109     "icon" => "abcdatos",
110     "title" => "ABCdatos",
111     "rule" => array(
112     "^ABCdatos BotLink[ /]([0-9.]{1,10})" => "\\1"
113     ),
114     "uri" => "http://www.abcdatos.com/"
115     ),
116     "abot" => array(
117     "icon" => "abot",
118     "title" => "aBot",
119     "rule" => array(
120     "^abot[ /]([0-9.]{1,10})" => "\\1"
121     ),
122     "uri" => ""
123     ),
124     "about" => array(
125     "icon" => "about",
126     "title" => "About",
127     "rule" => array(
128     "Libby[_/ ]([0-9.]{1,10})" => "\\1",
129     "About[_/ ]([0-9.]{1,10})libwww-perl" => "\\1"
130     ),
131     "uri" => ""
132     ),
133     "aboutus" => array(
134     "icon" => "aboutus",
135     "title" => "AboutUs",
136     "rule" => array(
137     "AboutUsBot" => ""
138     ),
139     "uri" => "http://www.aboutus.org/"
140     ),
141     "ackerm" => array(
142     "icon" => "robot",
143     "title" => "Ackerm",
144     "rule" => array(
145     "www.ackerm.com" => ""
146     ),
147     "uri" => "http://www.ackerm.com/"
148     ),
149     "acoi" => array(
150     "icon" => "acoi",
151     "title" => "AcoiRobot",
152     "rule" => array(
153     "^AcoiRobot" => ""
154     ),
155     "uri" => "http://monetdb.cwi.nl/acoi/projects.html"
156     ),
157     "acoon" => array(
158     "icon" => "acoon",
159     "title" => "Acoon",
160     "rule" => array(
161     "Acoon[ \-]?Robot" => ""
162     ),
163     "uri" => ""
164     ),
165     "accoona" => array(
166     "icon" => "accoona",
167     "title" => "Accoona",
168     "rule" => array(
169     "Accoona-AI-Agent[/ ]([0-9.]{1,10})" => "\\1",
170     "^accoona" => ""
171     ),
172     "uri" => ""
173     ),
174     "acme" => array(
175     "icon" => "acme",
176     "title" => "Acme",
177     "rule" => array(
178     "^Acme\.Spider" => ""
179     ),
180     "uri" => "http://www.acme.com/java/software/Acme.Spider.html"
181     ),
182     "active" => array(
183     "icon" => "robot",
184     "title" => "ActiveBookmark",
185     "rule" => array(
186     "ActiveBookmark[/ ]([0-9.]{1,10})" => "\\1"
187     ),
188     "uri" => ""
189     ),
190     "admuncher" => array(
191     "icon" => "robot",
192     "title" => "Ad Muncher",
193     "rule" => array(
194     "Ad Muncher[/ v]*([0-9.]{1,10})" => "\\1"
195     ),
196     "uri" => ""
197     ),
198     "aesop" => array(
199     "icon" => "robot",
200     "title" => "Aesop",
201     "rule" => array(
202     "^AESOP_com_SpiderMan" => ""
203     ),
204     "uri" => "http://www.aesop.com"
205     ),
206     "agada" => array(
207     "icon" => "robot",
208     "title" => "Agada",
209     "rule" => array(
210     "^agadine[/ ]([0-9.]{1,10})" => "\\1"
211     ),
212     "uri" => ""
213     ),
214     "ahrefs" => array(
215     "icon" => "ahrefs",
216     "title" => "Ahrefs",
217     "rule" => array(
218     "AhrefsBot[ /]([0-9.]{1,10})" => "\\1"
219     ),
220     "uri" => "http://ahrefs.com/"
221     ),
222     "aibot" => array(
223     "icon" => "robot",
224     "title" => "Aibot",
225     "rule" => array(
226     "AIBOT[/ ]([0-9.]{1,10})" => "\\1"
227     ),
228     "uri" => ""
229     ),
230     "aihitbot" => array(
231     "icon" => "robot",
232     "title" => "aiHit",
233     "rule" => array(
234     "aiHitBot[/ ]([0-9.]{1,10})" => "\\1"
235     ),
236     "uri" => "http://www.aihit.com/"
237     ),
238     "aipbot" => array(
239     "icon" => "robot",
240     "title" => "Aipbot",
241     "rule" => array(
242     "aipbot[/ ]([0-9.]{1,10})" => "\\1"
243     ),
244     "uri" => ""
245     ),
246     "aleksika" => array(
247     "icon" => "aleksika",
248     "title" => "Aleksika",
249     "rule" => array(
250     "Aleksika Spider[/ ]([0-9.]{1,10})" => "\\1"
251     ),
252     "uri" => ""
253     ),
254     "alertsite" => array(
255     "icon" => "alertsite",
256     "title" => "AlertSite",
257     "rule" => array(
258     "ipd[ /]([0-9.]{1,10}).*Alertsite\.com" => "\\1"
259     ),
260     "uri" => "http://www.alertsite.com/index.html"
261     ),
262     "alexa" => array(
263     "icon" => "alexa",
264     "title" => "Alexa",
265     "rule" => array(
266     "^ia_archive" => ""
267     ),
268     "uri" => "http://www.alexa.com/"
269     ),
270     "almaden" => array(
271     "icon" => "almaden",
272     "title" => "IBM Crawler",
273     "rule" => array(
274     "www\.almaden\.ibm\.com/cs/crawler" => ""
275     ),
276     "uri" => "http://www.almaden.ibm.com/cs/crawler/"
277     ),
278     "altavista" => array(
279     "icon" => "altavista",
280     "title" => "Altavista",
281     "rule" => array(
282     "Scooter[ /\-]*[a-z]*([0-9.]{1,10})" => "\\1",
283     "AltaVista V([0-9.]{1,10})" => "\\1",
284     "AltaVista Intranet V([0-9.]{1,10})" => "\\1"
285     ),
286     "uri" => "http://www.altavista.com/"
287     ),
288     "amazon" => array(
289     "icon" => "amazon",
290     "title" => "Amazon",
291     "rule" => array(
292     "^(aranhabot|amzn_assoc)" => "",
293     "^NutchEC2Test" => ""
294     ),
295     "uri" => "http://www.amazon.com/"
296     ),
297     "amidalla" => array(
298     "icon" => "amidalla",
299     "title" => "Amidalla",
300     "rule" => array(
301     "^amibot" => ""
302     ),
303     "uri" => ""
304     ),
305     "amfibi" => array(
306     "icon" => "amfibi",
307     "title" => "Amfibi",
308     "rule" => array(
309     "Amfibibot[/ ]([0-9.]{1,10})" => "\\1",
310     "Amfibibot" => ""
311     ),
312     "uri" => ""
313     ),
314     "amphetadesk" => array(
315     "icon" => "robot",
316     "title" => "AmphetaDesk",
317     "rule" => array(
318     "AmphetaDesk[/ ]([0-9.]{1,10})" => "\\1"
319     ),
320     "uri" => ""
321     ),
322     "amphetameme" => array(
323     "icon" => "robot",
324     "title" => "Amphetameme",
325     "rule" => array(
326     "amphetameme[ \-]?crawler" => ""
327     ),
328     "uri" => ""
329     ),
330     "annomille" => array(
331     "icon" => "robot",
332     "title" => "AnnoMille",
333     "rule" => array(
334     "^AnnoMille( spider)?[/ ]([0-9.]{1,10})" => "\\2"
335     ),
336     "uri" => "http://www.annomille.it"
337     ),
338     "anonymouse" => array(
339     "icon" => "anonymouse",
340     "title" => "Anonymouse.org",
341     "rule" => array(
342     "http://Anonymouse.org" => ""
343     ),
344     "uri" => "http://anonymouse.org"
345     ),
346     "ansearch" => array(
347     "icon" => "robot",
348     "title" => "Ansearch",
349     "rule" => array(
350     "AnsearchBot[/ ]([0-9.]{1,10})" => "\\1"
351     ),
352     "uri" => ""
353     ),
354     "answerchase" => array(
355     "icon" => "answerchase",
356     "title" => "AnswerChase",
357     "rule" => array(
358     "AnswerChase( PROve)?[/ ]([0-9.]{1,10})" => "\\2"
359     ),
360     "uri" => "http://www.answerchase.com/"
361     ),
362     "antibot" => array(
363     "icon" => "robot",
364 matthys 205 "title" => "Antibot",
365 joku 63 "rule" => array(
366     "antibot-V([0-9.]{1,10})" => "\\1"
367     ),
368     "uri" => ""
369     ),
370     "aonde" => array(
371     "icon" => "aonde",
372     "title" => "Aonde",
373     "rule" => array(
374     "^AONDE-Spider" => ""
375     ),
376     "uri" => ""
377     ),
378     "aonline" => array(
379     "icon" => "robot",
380     "title" => "A-Online.at",
381     "rule" => array(
382     "^A-Online Search" => ""
383     ),
384     "uri" => "http://www.a-online.at/"
385     ),
386     "aol" => array(
387     "icon" => "aol",
388     "title" => "AOLserver",
389     "rule" => array(
390     "^AOLserver-Tcl[/ ]([0-9.]{1,10})" => "\\1",
391     "^AOLserver" => ""
392     ),
393     "uri" => ""
394     ),
395     "apachebench" => array(
396     "icon" => "robot",
397     "title" => "ApacheBench",
398     "rule" => array(
399     "ApacheBench[ /]([0-9.]{1,10})" => "\\1"
400     ),
401     "uri" => ""
402     ),
403     "apassion4jazz" => array(
404     "icon" => "robot",
405     "title" => "Passion 4 Jazz",
406     "rule" => array(
407     "^BebopBot[ /]([0-9.]{1,10})" => "\\1"
408     ),
409     "uri" => "http://www.apassion4jazz.net/bebopbot.html"
410     ),
411     "apexoo" => array(
412     "icon" => "robot",
413     "title" => "Apexoo",
414     "rule" => array(
415     "^Apexoo Spider ([0-9.]{1,10})" => "\\1"
416     ),
417     "uri" => "http://www.apexoo.com/"
418     ),
419     "apnoti" => array(
420     "icon" => "apnoti",
421 matthys 205 "title" => "Apnoti",
422 joku 63 "rule" => array(
423     "apnoti.*Robot[ /]v([0-9.]{1,10})" => "\\1"
424     ),
425     "uri" => "http://www.apnoti.com/"
426     ),
427     "aport" => array(
428     "icon" => "aport",
429     "title" => "Aport",
430     "rule" => array(
431     "^Aport" => ""
432     ),
433     "uri" => ""
434     ),
435     "appie" => array(
436     "icon" => "walhello",
437     "title" => "Walhello",
438     "rule" => array(
439     "appie[ /]([0-9.]{1,10})" => "\\1"
440     ),
441     "uri" => ""
442     ),
443     "arachmo" => array(
444     "icon" => "arachmo",
445     "title" => "Arachmo",
446     "rule" => array(
447     "compatible; Arachmo" => ""
448     ),
449     "uri" => ""
450     ),
451     "arexera" => array(
452     "icon" => "arexera",
453     "title" => "Arexera",
454     "rule" => array(
455     "^X-Crawler" => "",
456     "^TECOMAC-Crawler[ /]([0-9.]{1,10})" => "\\1"
457     ),
458     "uri" => "http://www.arexera.de/"
459     ),
460     "arianna" => array(
461     "icon" => "arianna",
462     "title" => "Arianna",
463     "rule" => array(
464     "^www.arianna.it" => ""
465     ),
466     "uri" => "http://www.arianna.it/"
467     ),
468     "artface" => array(
469     "icon" => "robot",
470     "title" => "Artface",
471     "rule" => array(
472     "^ArtfaceBot" => ""
473     ),
474     "uri" => ""
475     ),
476     "artviper" => array(
477     "icon" => "artviper",
478     "title" => "artViper",
479     "rule" => array(
480     "artViper" => ""
481     ),
482     "uri" => "http://www.artviper.net/"
483     ),
484     "asinfo" => array(
485     "icon" => "robot",
486     "title" => "Any Search Info",
487     "rule" => array(
488     "Sleek Spider[ /]([0-9.]{1,10})" => "\\1"
489     ),
490     "uri" => "http://search-info.com/"
491     ),
492     "ask" => array(
493     "icon" => "askjeeves",
494     "title" => "Ask Jeeves",
495     "rule" => array(
496     "Ask[ \-]?Jeeves" => "",
497     "teomaagent" => ""
498     ),
499     "uri" => ""
500     ),
501     "askaboutoil" => array(
502     "icon" => "robot",
503     "title" => "ASPseek",
504     "rule" => array(
505     "^AskAboutOil[ /]([0-9.]{1,10})" => "\\1"
506     ),
507     "uri" => "http://askaboutoil.com/"
508     ),
509     "asked" => array(
510     "icon" => "robot",
511 matthys 205 "title" => "AskEd!",
512 joku 63 "rule" => array(
513     "^asked[ /]Nutch[ \-]([0-9.]{1,10})" => "\\1"
514     ),
515     "uri" => "http://asked.jp"
516     ),
517     "aspseek" => array(
518     "icon" => "robot",
519     "title" => "ASPseek",
520     "rule" => array(
521     "^ASPseek[/ ]([0-9.]{1,10})" => "\\1"
522     ),
523     "uri" => ""
524     ),
525     "atlocal" => array(
526     "icon" => "robot",
527     "title" => "At Local",
528     "rule" => array(
529     "AtlocalBot[/ ]([0-9.]{1,10})" => "\\1"
530     ),
531     "uri" => "http://www.atlocal.com/"
532     ),
533     "atomz" => array(
534     "icon" => "atomz",
535     "title" => "Atomz",
536     "rule" => array(
537     "Atomz[/ ]([0-9.]{1,10})" => "\\1"
538     ),
539     "uri" => ""
540     ),
541     "axel" => array(
542     "icon" => "robot",
543     "title" => "Axel",
544     "rule" => array(
545     "^axel" => ""
546     ),
547     "uri" => ""
548     ),
549     "axmo" => array(
550     "icon" => "axmo",
551     "title" => "Axmo",
552     "rule" => array(
553     "AxmoRobot" => ""
554     ),
555     "uri" => ""
556     ),
557     "answerbus" => array(
558     "icon" => "answerbus",
559     "title" => "AnswerBus",
560     "rule" => array(
561     "answerbus" => ""
562     ),
563     "uri" => ""
564     ),
565     "automapit" => array(
566     "icon" => "robot",
567     "title" => "AutoMapIt",
568     "rule" => array(
569     "AutoMapIt[ /](Bot)?" => ""
570     ),
571     "uri" => "http://www.automapit.com/bot.html"
572     ),
573     "augurnfind" => array(
574     "icon" => "robot",
575     "title" => "Augurnfind",
576     "rule" => array(
577     "augurnfind[/ ][v\-]*([0-9.]{1,10})" => "\\1"
578     ),
579     "uri" => ""
580     ),
581     "awasu" => array(
582     "icon" => "awasu",
583     "title" => "Awasu",
584     "rule" => array(
585     "Awasu[/ ]([0-9a-z.]{1,10})" => "\\1"
586     ),
587     "uri" => ""
588     ),
589     "axxus" => array(
590     "icon" => "axxus",
591     "title" => "axxus.de",
592     "rule" => array(
593     "whoiam" => ""
594     ),
595     "uri" => "http://www.axxus.de/"
596     ),
597     "ayell" => array(
598     "icon" => "ayell",
599     "title" => "Ayell",
600     "rule" => array(
601     "^Eurobot[ /]([0-9.]{1,10})" => "\\1"
602     ),
603     "uri" => "http://eurobot.ayell.eu"
604     ),
605     "babaloo" => array(
606     "icon" => "babaloo",
607     "title" => "Babaloo",
608     "rule" => array(
609     "BabalooSpider[ /]([0-9.]{1,10})" => "\\1"
610     ),
611     "uri" => "http://www.babaloo.si"
612     ),
613     "babe" => array(
614     "icon" => "robot",
615     "title" => "ba.be",
616     "rule" => array(
617     "BACS http://www.ba.be" => ""
618     ),
619     "uri" => "http://www.ba.be/"
620     ),
621     "baidu" => array(
622     "icon" => "baidu",
623     "title" => "Baidu",
624     "rule" => array(
625     "Baiduspider[ /]([0-9.]{1,10})" => "\\1",
626     "Baiduspider" => ""
627     ),
628     "uri" => "http://www.baidu.com/search/spider.html"
629     ),
630     "bananatree" => array(
631     "icon" => "robot",
632     "title" => "BananaTree",
633     "rule" => array(
634     "www\.thebananatree\.org" => ""
635     ),
636     "uri" => "http://www.thebananatree.org/"
637     ),
638     "bdcindexer" => array(
639     "icon" => "robot",
640     "title" => "bdcindexer",
641     "rule" => array(
642     "bdcindexer_([0-9a-z.]{1,10})" => "\\1"
643     ),
644     "uri" => "http://www.business.com/"
645     ),
646     "bdfetch" => array(
647     "icon" => "robot",
648     "title" => "BDFetch",
649     "rule" => array(
650     "^BDFetch" => ""
651     ),
652     "uri" => ""
653     ),
654     "bdncentral" => array(
655     "icon" => "robot",
656     "title" => "Bdncentral",
657     "rule" => array(
658     "BDNcentral Crawler v([0-9.]{1,10})" => "\\1"
659     ),
660     "uri" => "http://www.bdncentral.com/robot.html"
661     ),
662     "beammachine" => array(
663     "icon" => "robot",
664     "title" => "BeamMachine",
665     "rule" => array(
666     "^BeamMachine[ /]([0-9.]{1,10})" => "\\1"
667     ),
668     "uri" => "http://www.beammachine.net/"
669     ),
670     "become" => array(
671     "icon" => "become",
672     "title" => "Become",
673     "rule" => array(
674     "Become(JP)?Bot[/ ]([0-9.]{1,10})" => "\\2",
675     "(BecomeBot|Exabot)@exava\.com\)$" => ""
676     ),
677     "uri" => "http://www.become.com/"
678     ),
679     "beebware" => array(
680     "icon" => "robot",
681     "title" => "Beebware",
682     "rule" => array(
683     "BeebwareDirectory[/ ]v?([0-9.]{1,10})" => "\\2"
684     ),
685     "uri" => ""
686     ),
687     "bigbrother" => array(
688     "icon" => "robot",
689     "title" => "Big Brother",
690     "rule" => array(
691     "^Big Brother" => ""
692     ),
693     "uri" => "http://pauillac.inria.fr/~fpottier/"
694     ),
695     "bigclique" => array(
696     "icon" => "robot",
697     "title" => "BigClique",
698     "rule" => array(
699     "^BigCliqueBOT[/ ]([0-9.]{1,10})" => "\\1"
700     ),
701     "uri" => "http://www.bigclique.com"
702     ),
703     "biglotron" => array(
704     "icon" => "robot",
705     "title" => "Biglotron",
706     "rule" => array(
707     "^BIGLOTRON" => ""
708     ),
709     "uri" => "http://www.bigclique.com"
710     ),
711     "bigsearch" => array(
712     "icon" => "bigsearch",
713     "title" => "Bigsearch",
714     "rule" => array(
715     "Bigsearch.ca[/ ]Nutch[- ]([0-9.]{1,10})" => "\\1"
716     ),
717     "uri" => ""
718     ),
719     "bilbo" => array(
720     "icon" => "robot",
721     "title" => "Bilbo",
722     "rule" => array(
723     "Bilbo[ /]([0-9.]{1,10})" => "\\1"
724     ),
725     "uri" => "http://home.broadpark.no/~tnilsen-1/Linux/Bilbo_-_Nessus_WEB/bilbo_-_nessus_web.html"
726     ),
727     "bilgi" => array(
728     "icon" => "bilgi",
729     "title" => "Bilgi",
730     "rule" => array(
731     "Bilgi(Beta)?Bot[ /]([0-9.]{1,10})" => "\\1"
732     ),
733     "uri" => "http://www.bilgi.com/"
734     ),
735     "bitacle" => array(
736     "icon" => "bitacle",
737     "title" => "Bitacle",
738     "rule" => array(
739     "Bitacle (ro)?bot[ \(/V\:]+([0-9.]{1,10})" => "\\2",
740     ),
741     "uri" => "http://bitacle.org/"
742     ),
743     "bitbeamer" => array(
744     "icon" => "bitbeamer",
745     "title" => "BitBeamer",
746     "rule" => array(
747     "BitBeamer/([0-9.]{1,10})" => "\\1"
748     ),
749     "uri" => ""
750     ),
751     "biz360" => array(
752     "icon" => "biz360",
753     "title" => "Biz360",
754     "rule" => array(
755     "^Biz360 spider" => ""
756     ),
757     "uri" => ""
758     ),
759     "backlinktest" => array(
760     "icon" => "backlinktest",
761     "title" => "BacklinkCrawler",
762     "rule" => array(
763     "BacklinkCrawler" => ""
764     ),
765     "uri" => "http://www.backlinktest.com/crawler.html"
766     ),
767     "blaizbee" => array(
768     "icon" => "blaiz-bee",
769     "title" => "Blaiz-Bee",
770     "rule" => array(
771     "Blaiz-Bee[ /]([0-9.]{1,10})" => ""
772     ),
773     "uri" => ""
774     ),
775     "blogbeat" => array(
776     "icon" => "robot",
777     "title" => "BlogBeat",
778     "rule" => array(
779     "BlogBeat/?([0-9.]{1,10})" => "\\1",
780     ),
781     "uri" => "http://www.blogbeat.nl"
782     ),
783     "blogbotde" => array(
784     "icon" => "robot",
785     "title" => "blogbot.de",
786     "rule" => array(
787     "Naamah[ /]([0-9.a-z]{1,10})[ /]Blogbot" => "\\1",
788     ),
789     "uri" => "http://blogbot.de/"
790     ),
791     "blogcensus" => array(
792     "icon" => "blogcensus",
793     "title" => "Blogcensus",
794     "rule" => array(
795     "^Blogcensus" => "",
796     ),
797     "uri" => "http://blogcensus.de/bot/"
798     ),
799     "blogdex" => array(
800     "icon" => "robot",
801     "title" => "Blogdex",
802     "rule" => array(
803     "Blogdex[ /]([0-9.]{1,10})" => "\\1",
804     ),
805     "uri" => ""
806     ),
807     "blogg" => array(
808     "icon" => "blogg",
809     "title" => "Blogg",
810     "rule" => array(
811     "^blogg\.de" => ""
812     ),
813     "uri" => ""
814     ),
815     "blogland" => array(
816     "icon" => "robot",
817     "title" => "BlogLand",
818     "rule" => array(
819     "BlogLand[/ ]([0-9.]{1,10})" => "\\1"
820     ),
821     "uri" => ""
822     ),
823     "bloglines" => array(
824     "icon" => "bloglines",
825     "title" => "Bloglines",
826     "rule" => array(
827     "Bloglines[ /]([0-9.]{1,10})" => "\\1",
828     "Bloglines" => ""
829     ),
830     "uri" => ""
831     ),
832     "blogmap" => array(
833     "icon" => "robot",
834     "title" => "Blogmap",
835     "rule" => array(
836     "blogmap" => ""
837     ),
838     "uri" => ""
839     ),
840     "blogosphere" => array(
841     "icon" => "robot",
842     "title" => "Blogosphere",
843     "rule" => array(
844     "Blogosphere" => ""
845     ),
846     "uri" => ""
847     ),
848     "blogpeople" => array(
849     "icon" => "robot",
850     "title" => "BlogPeople",
851     "rule" => array(
852     "BlogPeople" => ""
853     ),
854     "uri" => ""
855     ),
856     "blogpulse" => array(
857     "icon" => "blogpulse",
858     "title" => "Blogpulse",
859     "rule" => array(
860     "Blogpulse" => ""
861     ),
862     "uri" => ""
863     ),
864     "blogranking" => array(
865     "icon" => "blogranking",
866     "title" => "BlogRanking",
867     "rule" => array(
868     "^BlogRanking(/RSS checker)?" => ""
869     ),
870     "uri" => ""
871     ),
872     "blogs" => array(
873     "icon" => "blogs",
874     "title" => "Blo.gs",
875     "rule" => array(
876     "blo\.gs[ /]([0-9.]{1,10})" => "\\1",
877     "blo\.gs" => ""
878     ),
879     "uri" => ""
880     ),
881     "blogshares" => array(
882     "icon" => "blogshares",
883     "title" => "BlogShares",
884     "rule" => array(
885     "BlogShares[ /]V?([0-9.]{1,10})" => "\\1",
886     "(^| |\()Blogshares(\.com| |\))" => ""
887     ),
888     "uri" => ""
889     ),
890     "blogslife" => array(
891     "icon" => "robot",
892     "title" => "BlogsLife",
893     "rule" => array(
894     "Blogslive" => ""
895     ),
896     "uri" => ""
897     ),
898     "blogsnow" => array(
899     "icon" => "blogsnow",
900     "title" => "BlogsNow",
901     "rule" => array(
902     "blogsnowbot" => "",
903     "BlogsNow" => ""
904     ),
905     "uri" => ""
906     ),
907     "blogstreet" => array(
908     "icon" => "blogstreet",
909     "title" => "BlogStreet",
910     "rule" => array(
911     "^BlogStreetBot" => ""
912     ),
913     "uri" => ""
914     ),
915     "blogsurf" => array(
916     "icon" => "robot",
917     "title" => "BlogSurf",
918     "rule" => array(
919     "nomadscafe_ra[/ ]([0-9.]{1,10})" => "\\1"
920     ),
921     "uri" => ""
922     ),
923     "blogtick" => array(
924     "icon" => "robot",
925     "title" => "BlogTick",
926     "rule" => array(
927     "BlogTickServer" => ""
928     ),
929     "uri" => ""
930     ),
931     "blogwatcher" => array(
932     "icon" => "blogwatcher",
933     "title" => "Blogwatcher",
934     "rule" => array(
935     "blogWatcher_Spider[/ ]([0-9.]{1,10})" => "\\1"
936     ),
937     "uri" => ""
938     ),
939     "blogwise" => array(
940     "icon" => "blogwise",
941     "title" => "Blogwise",
942     "rule" => array(
943     "Blogwise\.com(-MetaChecker)?[/ ]([0-9.]{1,10})" => "\\2"
944     ),
945     "uri" => ""
946     ),
947     "boardreader" => array(
948     "icon" => "boardreader",
949     "title" => "BoardReader",
950     "rule" => array(
951     "BoardReader[ \-](Image|Favicon)[ \-]Fetcher[ /]+([0-9.]{1,10})" => "\\2"
952     ),
953     "uri" => "http://www.boardreader.com"
954     ),
955     "bobby" => array(
956     "icon" => "bobby",
957     "title" => "Bobby",
958     "rule" => array(
959     "bobby[ /]([0-9.]{1,10})" => "\\1"
960     ),
961     "uri" => ""
962     ),
963     "boitho" => array(
964     "icon" => "robot",
965     "title" => "Boitho",
966     "rule" => array(
967     "Boitho\.com[ \-](dc|robot)?[/ ]([0-9.]{1,10})" => "\\2"
968     ),
969     "uri" => ""
970     ),
971     "booch" => array(
972     "icon" => "robot",
973     "title" => "Booch",
974     "rule" => array(
975     "^booch[_ /]([0-9.]{1,10})" => "\\1"
976     ),
977     "uri" => ""
978     ),
979     "book" => array(
980     "icon" => "book",
981     "title" => "Bookmark",
982     "rule" => array(
983     "http://www\.bookmark\.ne\.jp" => ""
984     ),
985     "uri" => ""
986     ),
987     "bookdog" => array(
988     "icon" => "bookdog",
989     "title" => "Bookdog",
990     "rule" => array(
991     "^Bookdog[ /]([0-9.]{1,10})" => "\\1"
992     ),
993     "uri" => ""
994     ),
995     "bordermanager" => array(
996     "icon" => "bordermanager",
997     "title" => "Border Manager",
998     "rule" => array(
999     "BorderManager[ /]([0-9.]{1,10})" => "\\1"
1000     ),
1001     "uri" => ""
1002     ),
1003     "botonparade" => array(
1004     "icon" => "botonparade",
1005     "title" => "Bot on Parade",
1006     "rule" => array(
1007     "BotOnParade" => ""
1008     ),
1009     "uri" => "http://www.bots-on-para.de/bot.html"
1010     ),
1011     "botje" => array(
1012     "icon" => "botje",
1013     "title" => "Botje",
1014     "rule" => array(
1015     "Plukkie[ /]([0-9.]{1,10})" => "\\1"
1016     ),
1017     "uri" => "http://www.botje.com/"
1018     ),
1019     "bottomfeeder" => array(
1020     "icon" => "bottomfeeder",
1021     "title" => "BottomFeeder",
1022     "rule" => array(
1023     "BottomFeeder[ /]([0-9.]{1,10})" => "\\1"
1024     ),
1025     "uri" => ""
1026     ),
1027     "browseremulator" => array(
1028     "icon" => "browseremulator",
1029     "title" => "BrowserEmulator",
1030     "rule" => array(
1031     "BrowserEmulator[ /]([0-9.]{1,10})" => "\\1"
1032     ),
1033     "uri" => "http://www.dejavu.org/emulator.htm"
1034     ),
1035     "browsershots" => array(
1036     "icon" => "browsershots",
1037     "title" => "Browsershots",
1038     "rule" => array(
1039     "Browsershots URL Check" => ""
1040     ),
1041     "uri" => "http://browsershots.org"
1042     ),
1043     "browserspy" => array(
1044     "icon" => "robot",
1045     "title" => "BrowserSpy",
1046     "rule" => array(
1047     "BrowserSpy" => ""
1048     ),
1049     "uri" => ""
1050     ),
1051     "bruinbot" => array(
1052     "icon" => "robot",
1053     "title" => "BruinBot",
1054     "rule" => array(
1055     "BruinBot" => ""
1056     ),
1057     "uri" => ""
1058     ),
1059     "bruno" => array(
1060     "icon" => "robot",
1061     "title" => "Bruno",
1062     "rule" => array(
1063     "^Bruno" => ""
1064     ),
1065     "uri" => ""
1066     ),
1067     "btbot" => array(
1068     "icon" => "btbot",
1069     "title" => "BitTorrent",
1070     "rule" => array(
1071     "BTbot/([0-9.]{1,10})" => "\\1"
1072     ),
1073     "uri" => ""
1074     ),
1075     "builtwith" => array(
1076     "icon" => "builtwith",
1077     "title" => "BuiltWith",
1078     "rule" => array(
1079     "BuiltWith[/ ]([0-9.]{1,10})" => "\\1"
1080     ),
1081     "uri" => "http://builtwith.com/bot.html"
1082     ),
1083     "bulkfeeds" => array(
1084     "icon" => "robot",
1085     "title" => "Bulkfeeds",
1086     "rule" => array(
1087     "Bulkfeeds[/ ]([a-z0-9.]{1,10})" => "\\1"
1088     ),
1089     "uri" => ""
1090     ),
1091     "burf" => array(
1092     "icon" => "burf",
1093     "title" => "Burf.com",
1094     "rule" => array(
1095     "^Norbert the Spider" => ""
1096     ),
1097     "uri" => "http://www.burf.com/"
1098     ),
1099     "butch" => array(
1100     "icon" => "robot",
1101     "title" => "Butch",
1102     "rule" => array(
1103     "Butch(__| )?([a-z0-9.]{1,10})" => "\\2"
1104     ),
1105     "uri" => ""
1106     ),
1107     "buzzlogic" => array(
1108     "icon" => "buzzlogic",
1109     "title" => "Buzzlogic",
1110     "rule" => array(
1111     "BuzzBot[ /]([0-9.]{1,10})" => "\\1"
1112     ),
1113     "uri" => "http://www.buzzlogic.com"
1114     ),
1115     "catchbot" => array(
1116     "icon" => "catchbot",
1117     "title" => "CatchBot",
1118     "rule" => array(
1119     "^CatchBot[/ ]([0-9.]{1,10})" => "\\1"
1120     ),
1121     "uri" => "http://www.catchbot.com"
1122     ),
1123     "camdiscover" => array(
1124     "icon" => "robot",
1125     "title" => "Camdiscover",
1126     "rule" => array(
1127     "^Camcrawler" => ""
1128     ),
1129     "uri" => ""
1130     ),
1131     "cazoodle" => array(
1132     "icon" => "cazoodle",
1133     "title" => "Cazoodle",
1134     "rule" => array(
1135     "^CazoodleBot/(Nutch|CazoodleBot)[/ \-]([0-9.]{1,10})" => "\\2",
1136     "^CazoodleBot[/ ]([a-z0-9.]{1,10})" => "\\1"
1137     ),
1138     "uri" => "http://www.cazoodle.com/cazoodlebot"
1139     ),
1140     "ccbot" => array(
1141     "icon" => "robot",
1142     "title" => "ccBot",
1143     "rule" => array(
1144     "CCBot[/ ]([0-9.]{1,10})" => "\\1"
1145     ),
1146     "uri" => "http://www.commoncrawl.org/bot.html"
1147     ),
1148     "ccgcrawl" => array(
1149     "icon" => "robot",
1150     "title" => "CCGCrawl",
1151     "rule" => array(
1152     "CCGCrawl[/ ]([0-9.]{1,10})" => "\\1"
1153     ),
1154     "uri" => "http://www.myworkbase.com/bot.html"
1155     ),
1156     "cerberian" => array(
1157     "icon" => "robot",
1158     "title" => "Cerberian Drtrs",
1159     "rule" => array(
1160     "^Cerberian Drtrs" => ""
1161     ),
1162     "uri" => ""
1163     ),
1164     "cfnetwork" => array(
1165     "icon" => "robot",
1166     "title" => "Cerberian Drtrs",
1167     "rule" => array(
1168     "^CFNetwork[/ ]([0-9.]{1,10})" => "\\1"
1169     ),
1170     "uri" => "http://www.cfnetwork.be/"
1171     ),
1172     "charlotte" => array(
1173     "icon" => "charlotte",
1174     "title" => "Charlotte",
1175     "rule" => array(
1176     "Charlotte[/ ]([0-9a-z.]{1,10})" => "\\1"
1177     ),
1178     "uri" => ""
1179     ),
1180     "cirilizator" => array(
1181     "icon" => "cirilizator",
1182     "title" => "Cirilizator",
1183     "rule" => array(
1184     "Cirilizator[/ ]([0-9.]{1,10})" => "\\1"
1185     ),
1186     "uri" => ""
1187     ),
1188     "claria" => array(
1189     "icon" => "robot",
1190     "title" => "Claria",
1191     "rule" => array(
1192     "(Claria|Diamond)(Bot)?[ /]([0-9.]{1,10})" => "\\3",
1193     "(Claria|Diamond)(Bot)" => ""
1194     ),
1195     "uri" => ""
1196     ),
1197     "claymont" => array(
1198     "icon" => "claymont",
1199     "title" => "Claymont",
1200     "rule" => array(
1201     "claymont\.com" => "",
1202     "OliverPerry" => ""
1203     ),
1204     "uri" => "http://www.claymont.com"
1205     ),
1206     "clush" => array(
1207     "icon" => "clush",
1208     "title" => "Clush",
1209     "rule" => array(
1210     "Clus(tered-Search-|h)Bot[ /]([0-9.]{1,10})" => "\\2"
1211     ),
1212     "uri" => ""
1213     ),
1214     "cmsworldmap" => array(
1215     "icon" => "cmsworldmap",
1216     "title" => "CMS World Map",
1217     "rule" => array(
1218     "cmsworldmap" => ""
1219     ),
1220     "uri" => "http://www.cmsworldmap.com/"
1221     ),
1222     "cobion" => array(
1223     "icon" => "robot",
1224     "title" => "Cobion",
1225     "rule" => array(
1226     " (QXW03018|obot)\)" => ""
1227     ),
1228     "uri" => ""
1229     ),
1230     "coccoc" => array(
1231 matthys 287 "icon" => "coccoc",
1232 matthys 209 "title" => "Coccoc",
1233 joku 63 "rule" => array(
1234     "coccoc[ /]([0-9.]{1,10})" => "\\1"
1235     ),
1236 matthys 287 "uri" => "http://help.coccoc.com"
1237 joku 63 ),
1238     "coldfusion" => array(
1239     "icon" => "coldfusion",
1240     "title" => "ColdFusion",
1241     "rule" => array(
1242     "^coldfusion" => ""
1243     ),
1244     "uri" => ""
1245     ),
1246     "combine" => array(
1247     "icon" => "robot",
1248     "title" => "Combine",
1249     "rule" => array(
1250     "Combine[ /]([0-9.]{1,10})" => "\\1"
1251     ),
1252     "uri" => ""
1253     ),
1254     "combot" => array(
1255     "icon" => "robot",
1256     "title" => "comBot",
1257     "rule" => array(
1258     "^comBot[ /]([0-9.]{1,10})" => "\\1"
1259     ),
1260     "uri" => ""
1261     ),
1262     "comet" => array(
1263     "icon" => "comet",
1264     "title" => "Comet",
1265     "rule" => array(
1266     "cometsearch@cometsystems" => ""
1267     ),
1268     "uri" => ""
1269     ),
1270     "commerobo" => array(
1271     "icon" => "robot",
1272     "title" => "Commerobo",
1273     "rule" => array(
1274     "Commerobo[/ ]([0-9.]{1,10})" => "\\1"
1275     ),
1276     "uri" => ""
1277     ),
1278     "comrite" => array(
1279     "icon" => "robot",
1280     "title" => "ComRite",
1281     "rule" => array(
1282     "Comrite[/ ]([0-9.]{1,10})" => "\\1"
1283     ),
1284     "uri" => "http://www.comrite.com/"
1285     ),
1286     "convera" => array(
1287     "icon" => "convera",
1288     "title" => "Convera",
1289     "rule" => array(
1290     "Convera(MultiMedia)?Crawler[/ ]([0-9.]{1,10})" => "\\2",
1291     "Convera Internet Spider V([0-9.]{1,10})" => "\\1"
1292     ),
1293     "uri" => ""
1294     ),
1295     "coolbot" => array(
1296     "icon" => "robot",
1297     "title" => "CoolBot",
1298     "rule" => array(
1299     "^CoolBot" => ""
1300     ),
1301     "uri" => ""
1302     ),
1303     "cosmix" => array(
1304     "icon" => "robot",
1305     "title" => "Cosmix",
1306     "rule" => array(
1307     "^(voyager|cfetch|CosmixCrawler|carleson)[/ ]([0-9.]{1,10})" => "\\2",
1308     ),
1309     "uri" => ""
1310     ),
1311     "cosmos" => array(
1312     "icon" => "robot",
1313     "title" => "Cosmos",
1314     "rule" => array(
1315     "^cosmos" => ""
1316     ),
1317     "uri" => ""
1318     ),
1319     "cosmoty" => array(
1320     "icon" => "robot",
1321     "title" => "Cosmoty",
1322     "rule" => array(
1323     "^beautybot[/ ]([0-9.]{1,10})" => "\\1"
1324     ),
1325     "uri" => "http://www.uchoose.de/crawler/beautybot/"
1326     ),
1327     "covario" => array(
1328     "icon" => "covario",
1329     "title" => "Covario",
1330     "rule" => array(
1331     "Covario-IDS[/ ]([0-9.]{1,10})" => "\\1"
1332     ),
1333     "uri" => "http://www.covario.com/ids/"
1334     ),
1335     "creativecommons" => array(
1336     "icon" => "creativecommons",
1337     "title" => "Creative Commons",
1338     "rule" => array(
1339     "CreativeCommons[/ ]([0-9.]{1,6}(-dev)?)" => "\\1"
1340     ),
1341     "uri" => ""
1342     ),
1343     "cscrawler" => array(
1344     "icon" => "robot",
1345     "title" => "CsCrawler",
1346     "rule" => array(
1347     "CsCrawler" => ""
1348     ),
1349     "uri" => "http://www.kde.cs.uni-kassel.de/lehre/ss2005/googlespam/crawler.html"
1350     ),
1351 joku 111 "jigsaw" => array(
1352     "icon" => "jigsaw",
1353     "title" => "Jigsaw",
1354     "rule" => array(
1355     "Jigsaw[ /]([0-9.]{1,10})" => "\\1"
1356     ),
1357     "uri" => "http://jigsaw.w3.org/css-validator/"
1358     ),
1359 joku 63 "csscheck" => array(
1360     "icon" => "w3c",
1361     "title" => "W3C CSS Validator",
1362     "rule" => array(
1363     "CSS(Check|_Validator).+[ /]([0-9.]{1,10})" => "\\2"
1364     ),
1365     "uri" => "http://jigsaw.w3.org/css-validator/"
1366     ),
1367     "custo" => array(
1368     "icon" => "robot",
1369     "title" => "Custo",
1370     "rule" => array(
1371     "Custo[ /]([0-9.]{1,10})" => "\\1"
1372     ),
1373     "uri" => ""
1374     ),
1375     "cybernavi" => array(
1376     "icon" => "robot",
1377     "title" => "CyberNavi",
1378     "rule" => array(
1379     "CyberNavi_WebGet[ /]([0-9.]{1,10})" => "\\1"
1380     ),
1381     "uri" => ""
1382     ),
1383     "cyberpatrol" => array(
1384     "icon" => "cyberpatrol",
1385     "title" => "CyberPatrol",
1386     "rule" => array(
1387     "^CyberPatrol" => ""
1388     ),
1389     "uri" => "http://www.cyberpatrol.com/cyberpatrolcrawler.asp"
1390     ),
1391     "cyberz" => array(
1392     "icon" => "cyberz",
1393     "title" => "Cyberz",
1394     "rule" => array(
1395     "Cyberz Communication Agent" => ""
1396     ),
1397     "uri" => ""
1398     ),
1399     "cydral" => array(
1400     "icon" => "robot",
1401     "title" => "Cydral",
1402     "rule" => array(
1403     "CydralSpider[ /]([0-9.]{1,10})" => "\\1"
1404     ),
1405     "uri" => ""
1406     ),
1407     "cynthia" => array(
1408     "icon" => "cynthia",
1409     "title" => "Cynthia Says",
1410     "rule" => array(
1411     "Cynthia[ /]([0-9.]{1,10})" => "\\1"
1412     ),
1413     "uri" => ""
1414     ),
1415     "d4x" => array(
1416     "icon" => "d4x",
1417     "title" => "Downloader for X",
1418     "rule" => array(
1419     "Downloader for X[ /]([0-9.]{1,10})" => "\\1"
1420     ),
1421     "uri" => ""
1422     ),
1423     "da" => array(
1424     "icon" => "da",
1425     "title" => "DA",
1426     "rule" => array(
1427     "^DA[ /]([0-9.]{1,10})" => "\\1"
1428     ),
1429     "uri" => ""
1430     ),
1431     "daum" => array(
1432     "icon" => "daum",
1433     "title" => "DAUM",
1434     "rule" => array(
1435     "DAUMOA[ /]([0-9.]{1,10})" => "\\1",
1436     "DAUM Web Robot" => "",
1437     "Daum Communications Corp" => "",
1438     "EDI[ /]([0-9.]{1,10})" => "\\1",
1439     "Edacious.*Intelligent Web Robot" => "",
1440     "RaBot[/ ]([0-9.]{1,10}) Agent" => "\\1"
1441     ),
1442     "uri" => "http://ws.daum.net/"
1443     ),
1444     "daypop" => array(
1445     "icon" => "robot",
1446     "title" => "Daypop",
1447     "rule" => array(
1448     "daypopbot[/ ]([0-9.]{1,10})" => "\\1"
1449     ),
1450     "uri" => ""
1451     ),
1452     "delfi" => array(
1453     "icon" => "delfi",
1454     "title" => "Delfi",
1455     "rule" => array(
1456     "crawl at delfi dot lt" => ""
1457     ),
1458     "uri" => ""
1459     ),
1460     "depspid" => array(
1461     "icon" => "depspid",
1462     "title" => "DepSpid",
1463     "rule" => array(
1464     "DepSpid[/ ]([0-9.]{1,10})" => "\\1"
1465     ),
1466     "uri" => ""
1467     ),
1468     "devonagent" => array(
1469     "icon" => "devonagent",
1470     "title" => "DEVONagent",
1471     "rule" => array(
1472     "DEVONtech" => ""
1473     ),
1474     "uri" => ""
1475     ),
1476     "diffbot" => array(
1477     "icon" => "diffbot",
1478     "title" => "Diffbot",
1479     "rule" => array(
1480     " Diffbot" => ""
1481     ),
1482     "uri" => "http://www.diffbot.com"
1483     ),
1484     "diribot" => array(
1485     "icon" => "diribot",
1486     "title" => "Diri",
1487     "rule" => array(
1488     "diribot" => ""
1489     ),
1490     "uri" => "http://www.diri.bg"
1491     ),
1492     "directhit" => array(
1493     "icon" => "robot",
1494     "title" => "Direct Hit",
1495     "rule" => array(
1496     "EZResult -- Internet Search Engine" => ""
1497     ),
1498     "uri" => "http://www.directhit.com"
1499     ),
1500     "disco" => array(
1501     "icon" => "disco",
1502 matthys 209 "title" => "Disco",
1503 joku 63 "rule" => array(
1504     "disco/Nutch[/ -]([0-9.]{1,10})" => "\\1",
1505     "discobot[/ -]([0-9.]{1,10})" => "\\1",
1506     "disco-crawl@discoveryengine.com" => ""
1507     ),
1508     "uri" => "http://www.discoveryengine.com/"
1509     ),
1510     "discopump" => array(
1511     "icon" => "robot",
1512     "title" => "DISCo Pump",
1513     "rule" => array(
1514     "DISCo Pump[/ ]([0-9.]{1,10})" => "\\1"
1515     ),
1516     "uri" => ""
1517     ),
1518     "dkimreputation" => array(
1519     "icon" => "dkimreputation",
1520     "title" => "DKIM Reputation",
1521     "rule" => array(
1522     "DKIMRepBot[ /]([0-9.]{1,10})" => "\\1"
1523     ),
1524     "uri" => "http://www.dkim-reputation.org"
1525     ),
1526     "dnsdigger" => array(
1527     "icon" => "dnsdigger",
1528     "title" => "DNS-Digger",
1529     "rule" => array(
1530     "DNS-Digger-Explorer[ /]([0-9.]{1,10})" => "\\1"
1531     ),
1532     "uri" => "http://www.dnsdigger.com"
1533     ),
1534     "dnsvalidation" => array(
1535     "icon" => "dnsvalidation",
1536     "title" => "DNS Validation",
1537     "rule" => array(
1538     "DNSValidation[ /]([0-9.]{1,10})" => "\\1"
1539     ),
1540     "uri" => "http://www.dnsvalidation.com/"
1541     ),
1542     "doctorhtml" => array(
1543     "icon" => "robot",
1544     "title" => "DoctorHTML",
1545     "rule" => array(
1546     "Doctor[ \-]?HTML" => ""
1547     ),
1548     "uri" => ""
1549     ),
1550     "dolphin" => array(
1551     "icon" => "dolphin",
1552     "title" => "Dolphin",
1553     "rule" => array(
1554     "Dolphin[ /]([0-9.]{1,10})" => "\\1"
1555     ),
1556     "uri" => "http://telehouse.ru/crawler.html"
1557     ),
1558     "domaindatei" => array(
1559     "icon" => "robot",
1560     "title" => "Domaindatei",
1561     "rule" => array(
1562     "DomaindateiSpider[ /]([0-9.]{1,10})" => "\\1"
1563     ),
1564     "uri" => ""
1565     ),
1566     "dontbuylists" => array(
1567     "icon" => "dontbuylists",
1568     "title" => "Dontbuylists",
1569     "rule" => array(
1570     "DBLBot[ /]([0-9.]{1,10})" => "\\1"
1571     ),
1572     "uri" => "http://www.dontbuylists.com"
1573     ),
1574     "dotbot" => array(
1575     "icon" => "robot",
1576     "title" => "DotBot",
1577     "rule" => array(
1578     "DotBot[ /]([0-9.]{1,10})" => "\\1"
1579     ),
1580     "uri" => "http://www.dotnetdotcom.org"
1581     ),
1582     "doweb" => array(
1583     "icon" => "doweb",
1584     "title" => "DoWeb",
1585     "rule" => array(
1586     "^www.doweb.co.uk" => ""
1587     ),
1588     "uri" => "http://www.doweb.co.uk/"
1589     ),
1590     "downforeveryoneorjustme" => array(
1591     "icon" => "downforeveryoneorjustme",
1592     "title" => "Down for everyone or just me",
1593     "rule" => array(
1594     "downforeveryoneorjustme" => ""
1595     ),
1596     "uri" => "http://www.downforeveryoneorjustme.com"
1597     ),
1598     "downloadninja" => array(
1599     "icon" => "robot",
1600     "title" => "Download Ninja",
1601     "rule" => array(
1602     "Download Ninja[ /]([0-9.]{1,10})" => "\\1"
1603     ),
1604     "uri" => ""
1605     ),
1606     "drupal" => array(
1607     "icon" => "drupal",
1608     "title" => "Drupal",
1609     "rule" => array(
1610     "^Drupal" => ""
1611     ),
1612     "uri" => ""
1613     ),
1614     "dsns" => array(
1615     "icon" => "robot",
1616     "title" => "DSNS Scanner",
1617     "rule" => array(
1618     "^DSNS" => ""
1619     ),
1620     "uri" => ""
1621     ),
1622     "dtsagent" => array(
1623     "icon" => "robot",
1624     "title" => "DTS Agent",
1625     "rule" => array(
1626     "DTS Agent" => ""
1627     ),
1628     "uri" => ""
1629     ),
1630     "earthcom" => array(
1631     "icon" => "earthcom",
1632     "title" => "Earthcom",
1633     "rule" => array(
1634     "EARTHCOM\.info[/ ]([0-9a-z.]{1,10})" => "\\1"
1635     ),
1636     "uri" => ""
1637     ),
1638     "ebay" => array(
1639     "icon" => "ebay",
1640     "title" => "eBay",
1641     "rule" => array(
1642     "eBay Relevance Ad Crawler" => ""
1643     ),
1644     "uri" => ""
1645     ),
1646     "echo" => array(
1647     "icon" => "robot",
1648     "title" => "Echo.com",
1649     "rule" => array(
1650     "_TrueRobot[/ ]([0-9.]{1,10})" => "\\1"
1651     ),
1652     "uri" => "http://www.echo.com/"
1653     ),
1654     "eert" => array(
1655     "icon" => "robot",
1656 matthys 209 "title" => "Eert",
1657 joku 63 "rule" => array(
1658     "eert spdr[/ ]([0-9.]{1,10})" => "\\1"
1659     ),
1660     "uri" => "http://bot.eert.net"
1661     ),
1662     "eknip" => array(
1663     "icon" => "eknip",
1664     "title" => "E-Knip",
1665     "rule" => array(
1666     "eknip[ /]([0-9a-z.]{1,10})" => "\\1"
1667     ),
1668     "uri" => ""
1669     ),
1670     "zoominfo" => array(
1671     "icon" => "zoominfo",
1672 matthys 209 "title" => "ZoomInfo", //Previous called Eliyon
1673 joku 63 "rule" => array(
1674     "NextGenSearchBot[/ ]([0-9.]{1,10})" => "\\1"
1675     ),
1676     "uri" => "http://www.zoominfo.com/business/nextgensearchbot.html"
1677     ),
1678     "emeraldshield" => array(
1679     "icon" => "emeraldshield",
1680     "title" => "EmeraldShield",
1681     "rule" => array(
1682     "^EmeraldShield" => ""
1683     ),
1684     "uri" => ""
1685     ),
1686     "empas" => array(
1687     "icon" => "empas",
1688     "title" => "Empas",
1689     "rule" => array(
1690     "DigExt; empas\)$" => "",
1691     "^EMPAS[_\-]ROBOT" => ""
1692     ),
1693     "uri" => ""
1694     ),
1695     "entireweb" => array(
1696     "icon" => "entireweb",
1697     "title" => "Entireweb",
1698     "rule" => array(
1699     "Speedy[ ]?Spider" => ""
1700     ),
1701     "uri" => ""
1702     ),
1703     "envolk" => array(
1704     "icon" => "envolk",
1705     "title" => "Envolk",
1706     "rule" => array(
1707     "envolk\[ITS\]spider[/ ]([0-9.]{1,10})" => "\\1",
1708     "envolk[/ ]([0-9.]{1,10})" => "\\1"
1709     ),
1710     "uri" => ""
1711     ),
1712     "esnet" => array(
1713     "icon" => "robot",
1714     "title" => "ES.NET",
1715     "rule" => array(
1716     "ES.NET Crawler[ /]([0-9.]{1,10})" => "\\1"
1717     ),
1718     "uri" => ""
1719     ),
1720     "estyle" => array(
1721     "icon" => "estyle",
1722     "title" => "eStyle Search",
1723     "rule" => array(
1724     "eStyleSearch[ /]([0-9.]{1,10})" => "\\1"
1725     ),
1726     "uri" => ""
1727     ),
1728     "eurip" => array(
1729     "icon" => "robot",
1730     "title" => "Eurip",
1731     "rule" => array(
1732     "EuripBot[ /]([0-9.]{1,10})" => "\\1"
1733     ),
1734     "uri" => "http://www.eurip.com"
1735     ),
1736     "eurodirectory" => array(
1737     "icon" => "robot",
1738     "title" => "Euro Directory",
1739     "rule" => array(
1740     "www\.euro-directory\.com" => ""
1741     ),
1742     "uri" => "http://www.euro-directory.com/"
1743     ),
1744     "euroseek" => array(
1745     "icon" => "euroseek",
1746     "title" => "EuroSeek",
1747     "rule" => array(
1748     "Arachnoidea" => ""
1749     ),
1750     "uri" => ""
1751     ),
1752     "evaal" => array(
1753     "icon" => "evaal",
1754     "title" => "Evaal",
1755     "rule" => array(
1756     "^EvaalSE" => ""
1757     ),
1758     "uri" => "http://www.evaal.com/"
1759     ),
1760     "eventax" => array(
1761     "icon" => "eventax",
1762     "title" => "Eventax",
1763     "rule" => array(
1764     "^eventax[ /]([0-9.]{1,10})" => "\\1"
1765     ),
1766     "uri" => "http://www.eventax.de/"
1767     ),
1768     "everbee" => array(
1769     "icon" => "everbee",
1770     "title" => "Everbee",
1771     "rule" => array(
1772     "EverbeeCrawler" => ""
1773     ),
1774     "uri" => ""
1775     ),
1776     "everest" => array(
1777     "icon" => "everest",
1778     "title" => "Everest",
1779     "rule" => array(
1780     "Everest-Vulcan Inc.[ /]([0-9.]{1,10})" => "\\1"
1781     ),
1782     "uri" => ""
1783     ),
1784     "exabot" => array(
1785     "icon" => "exabot",
1786     "title" => "ExaBot",
1787     "rule" => array(
1788     "^NG[ /]([0-9.]{1,10})" => "\\1",
1789     "Exabot/([0-9.]{1,10})" => "\\1",
1790     "ExaBotTest/([0-9.]{1,10})" => "\\1",
1791     "ExaBot-(Test|Images)/([0-9.]{1,10})" => "\\2",
1792     "ExaBot-Thumbnails" => ""
1793     ),
1794     "uri" => "http://www.exabot.com/go/robot"
1795     ),
1796     "exactseek" => array(
1797     "icon" => "exactseek",
1798     "title" => "ExactSeek",
1799     "rule" => array(
1800     "^exactseek[ \-]?(pagereaper|crawler)[ \-]?([0-9.]{1,10})" => "\\2",
1801     "ExactSeek[ \.\-]?(Crawler|com)" => ""
1802     ),
1803     "uri" => "http://www.exactseek.com/"
1804     ),
1805     "excite" => array(
1806     "icon" => "excite",
1807     "title" => "Excite",
1808     "rule" => array(
1809     "Architext[ \-]?Spider" => ""
1810     ),
1811     "uri" => ""
1812     ),
1813     "execrawl" => array(
1814     "icon" => "robot",
1815     "title" => "Execrawl",
1816     "rule" => array(
1817     "Execrawl[ /]([0-9.]{1,10})" => "\\1",
1818     "Execrawl" => ""
1819     ),
1820     "uri" => ""
1821     ),
1822     "expertmonitor" => array(
1823     "icon" => "robot",
1824     "title" => "ExpertMonitor",
1825     "rule" => array(
1826     "^NetMonitor[ /]([0-9.]{1,10})" => "\\1"
1827     ),
1828     "uri" => ""
1829     ),
1830     "explorerrss" => array(
1831     "icon" => "robot",
1832     "title" => "Explorer RSS",
1833     "rule" => array(
1834     "^Windows-RSS-Platform[ /]([0-9.]{1,10})" => "\\1"
1835     ),
1836     "uri" => ""
1837     ),
1838     "ezooms" => array(
1839     "icon" => "robot",
1840     "title" => "Ezooms",
1841     "rule" => array(
1842     "ezooms[/ ]([0-9a-z.\-]{1,10})" => "\\1"
1843     ),
1844     "uri" => ""
1845     ),
1846     "facebook" => array(
1847     "icon" => "facebook",
1848     "title" => "Facebook",
1849     "rule" => array(
1850     "FacebookFeedParser[/ ]([0-9a-z.\-]{1,10})" => "\\1",
1851     "facebookexternalhit/([0-9.]{1,10})" => "\\1",
1852     "^facebook" => ""
1853     ),
1854     "uri" => "http://www.facebook.com/"
1855     ),
1856     "fast" => array(
1857     "icon" => "fast",
1858     "title" => "Fast",
1859     "rule" => array(
1860     "^FAST( Enterprise |-Web| MetaWeb )?Crawler[ /]([0-9.]{1,10})" => "\\2",
1861     "^FAST( Enterprise |-Web| MetaWeb | PartnerSite )?Crawler" => "",
1862     "^Fast Crawler" => "",
1863     "^libwww-perl[ /]([0-9.]{1,10}) FP[ /]([0-9.]{1,10})" => "\\2"
1864     ),
1865     "uri" => "http://www.alltheweb.com/"
1866     ),
1867     "fastbuzz" => array(
1868     "icon" => "fastbuzz",
1869     "title" => "Fastbuzz",
1870     "rule" => array(
1871     "^fastbuzz\.com" => ""
1872     ),
1873     "uri" => ""
1874     ),
1875     "favorg" => array(
1876     "icon" => "robot",
1877     "title" => "FavOrg",
1878     "rule" => array(
1879     "^FavOrg" => ""
1880     ),
1881     "uri" => ""
1882     ),
1883     "favorstar" => array(
1884     "icon" => "favorstar",
1885 matthys 209 "title" => "Favorstar",
1886 joku 63 "rule" => array(
1887     "favorstarbot[ /]([0-9.]{1,10})" => "\\1"
1888     ),
1889     "uri" => "http://favorstar.com/bot.html"
1890     ),
1891     "faxo" => array(
1892     "icon" => "robot",
1893     "title" => "Faxo",
1894     "rule" => array(
1895     "^Faxobot[ /]([0-9.]{1,10})" => "\\1"
1896     ),
1897     "uri" => "http://www.faxo.com/"
1898     ),
1899     "fdse" => array(
1900     "icon" => "robot",
1901     "title" => "FDSE Robot",
1902     "rule" => array(
1903     "FDSE[ \-]?robot" => ""
1904     ),
1905     "uri" => ""
1906     ),
1907     "feedback" => array(
1908     "icon" => "robot",
1909     "title" => "FeedBack",
1910     "rule" => array(
1911     "FeedBack[/ ]([0-9.]{1,10})" => "\\1"
1912     ),
1913     "uri" => ""
1914     ),
1915     "feedburner" => array(
1916     "icon" => "feedburner",
1917     "title" => "FeedBurner",
1918     "rule" => array(
1919     "^FeedBurner[/ ]([0-9.]{1,10})" => "\\1"
1920     ),
1921     "uri" => ""
1922     ),
1923     "feeddemon" => array(
1924     "icon" => "feeddemon",
1925     "title" => "FeedDemon",
1926     "rule" => array(
1927     "FeedDemon[/ ]([0-9.]{1,10})" => "\\1"
1928     ),
1929     "uri" => ""
1930     ),
1931     "feedfind" => array(
1932     "icon" => "feedfind",
1933     "title" => "FeedFind",
1934     "rule" => array(
1935     "Feed::Find[ /]([0-9.]{1,10})" => "\\1"
1936     ),
1937     "uri" => ""
1938     ),
1939     "feedonfeeds" => array(
1940     "icon" => "robot",
1941     "title" => "Feed On Feeds",
1942     "rule" => array(
1943     "FeedOnFeeds[/ ]([0-9.]{1,10})" => "\\1"
1944     ),
1945     "uri" => ""
1946     ),
1947     "feedparser" => array(
1948     "icon" => "robot",
1949     "title" => "Feedparser",
1950     "rule" => array(
1951     "UniversalFeedParser[/ ]([0-9a-z.\-]{1,10})" => "\\1",
1952     "FeedParser" => ""
1953     ),
1954     "uri" => ""
1955     ),
1956     "feedreader" => array(
1957     "icon" => "feedreader",
1958     "title" => "Feedreader",
1959     "rule" => array(
1960     "^Feedreader" => ""
1961     ),
1962     "uri" => ""
1963     ),
1964     "feedserver" => array(
1965     "icon" => "robot",
1966     "title" => "FeedServer",
1967     "rule" => array(
1968     "FeedServer[/ ]([0-9.]{1,10})" => "\\1"
1969     ),
1970     "uri" => ""
1971     ),
1972     "feedster" => array(
1973     "icon" => "feedster",
1974     "title" => "Feedster",
1975     "rule" => array(
1976     "Feedster Crawler[/ ]([0-9.]{1,10})" => "\\1"
1977     ),
1978     "uri" => ""
1979     ),
1980     "feedvalidator" => array(
1981     "icon" => "feedvalidator",
1982     "title" => "Feed Validator",
1983     "rule" => array(
1984     "^FeedValidator[/ ]([0-9.]{1,10})" => "\\1"
1985     ),
1986     "uri" => ""
1987     ),
1988     "fdm" => array(
1989     "icon" => "robot",
1990     "title" => "Free Download Manager",
1991     "rule" => array(
1992     "^FDM[/ ]([0-9a-z.]{1,10})" => "\\1"
1993     ),
1994     "uri" => ""
1995     ),
1996     "filangy" => array(
1997     "icon" => "robot",
1998     "title" => "Filangy",
1999     "rule" => array(
2000     "Filangy[/ ]([0-9.]{1,10})" => "\\1"
2001     ),
2002     "uri" => "http://www.filangy.com/"
2003     ),
2004     "findanisp" => array(
2005     "icon" => "robot",
2006     "title" => "FindAnISP",
2007     "rule" => array(
2008     "FindAnISP" => ""
2009     ),
2010     "uri" => "http://www.findanisp.com/"
2011     ),
2012     "findengines" => array(
2013     "icon" => "findengines",
2014     "title" => "FindEngines",
2015     "rule" => array(
2016     "FindEngines! Bot" => ""
2017     ),
2018     "uri" => ""
2019     ),
2020     "findexa" => array(
2021     "icon" => "findexa",
2022     "title" => "Findexa",
2023     "rule" => array(
2024     "Findexa Crawler" => ""
2025     ),
2026     "uri" => ""
2027     ),
2028     "findlinks" => array(
2029     "icon" => "findlinks",
2030     "title" => "FindLinks",
2031     "rule" => array(
2032     "findlinks[ /]([0-9.]{1,10})" => "\\1",
2033     "^FindLinks" => ""
2034     ),
2035     "uri" => ""
2036     ),
2037     "findoor" => array(
2038     "icon" => "findoor",
2039 matthys 209 "title" => "Findoor",
2040 joku 63 "rule" => array(
2041     "^findoor(-Bot)?" => "\\1"
2042     ),
2043     "uri" => ""
2044     ),
2045     "firefly" => array(
2046     "icon" => "firefly",
2047     "title" => "Firefly",
2048     "rule" => array(
2049     "Firefly" => ""
2050     ),
2051     "uri" => ""
2052     ),
2053     "flashget" => array(
2054     "icon" => "flashget",
2055     "title" => "FlashGet",
2056     "rule" => array(
2057     "^FlashGet" => ""
2058     ),
2059     "uri" => ""
2060     ),
2061     "flatland" => array(
2062     "icon" => "flatland",
2063     "title" => "Flatland",
2064     "rule" => array(
2065     "^flatlandbot" => ""
2066     ),
2067     "uri" => "http://www.flatlandindustries.com/flatlandbot"
2068     ),
2069     "flickbot" => array(
2070     "icon" => "flickbot",
2071     "title" => "FlickBot",
2072     "rule" => array(
2073     "FlickBot[ /]([0-9.]{1,10})" => "\\1"
2074     ),
2075     "uri" => ""
2076     ),
2077     "flightdeckreports" => array(
2078     "icon" => "flightdeckreports",
2079     "title" => "Flight Deck Reports",
2080     "rule" => array(
2081     "FlightDeckReportsBot[ /]([0-9.]{1,10})" => "\\1"
2082     ),
2083     "uri" => "http://www.flightdeckreports.com/"
2084     ),
2085     "forex" => array(
2086     "icon" => "robot",
2087     "title" => "Forex",
2088     "rule" => array(
2089     "^Forex Trading Network Organization" => ""
2090     ),
2091     "uri" => "http://www.netforex.org/"
2092     ),
2093     "freshmeat" => array(
2094     "icon" => "freshmeat",
2095 matthys 209 "title" => "Freshmeat",
2096 joku 63 "rule" => array(
2097     "fmII URL validator[ /]([0-9.]{1,10})" => "\\1",
2098     "freshmeat.net URL validator[ /]([0-9.]{1,10})" => "\\1"
2099     ),
2100     "uri" => "http://www.freshmeat.net/"
2101     ),
2102     "friend" => array(
2103     "icon" => "friend",
2104     "title" => "Friend",
2105     "rule" => array(
2106     "www\.friend\.fr" => ""
2107     ),
2108     "uri" => "http://www.friend.fr/"
2109     ),
2110     "frontier" => array(
2111     "icon" => "frontier",
2112     "title" => "Frontier",
2113     "rule" => array(
2114     "Frontier[ /]([0-9.]{1,10})" => "\\1"
2115     ),
2116     "uri" => ""
2117     ),
2118     "fyberspider" => array(
2119     "icon" => "robot",
2120     "title" => "FyberSpider",
2121     "rule" => array(
2122     "FyberSpider[ /]([0-9.]{1,10})" => "\\1"
2123     ),
2124     "uri" => "http://www.fyberspider.com/"
2125     ),
2126     "gaisbot" => array(
2127     "icon" => "gais",
2128     "title" => "Gaisbot",
2129     "rule" => array(
2130     "Gaisbot[ /]([0-9.]{1,10})" => "\\1"
2131     ),
2132     "uri" => ""
2133     ),
2134     "galaxy" => array(
2135     "icon" => "galaxy",
2136     "title" => "Galaxy",
2137     "rule" => array(
2138     "GalaxyBot[ /]([0-9.]{1,10})" => "\\1",
2139     "www.galaxy.com" => ""
2140     ),
2141     "uri" => "http://www.galaxy.com/"
2142     ),
2143     "gamespy" => array(
2144     "icon" => "gamespy",
2145     "title" => "GameSpy",
2146     "rule" => array(
2147     "GameSpyHTTP[ /]([0-9.]{1,10})" => "\\1"
2148     ),
2149     "uri" => ""
2150     ),
2151     "garlik" => array(
2152     "icon" => "garlik",
2153     "title" => "Garlik",
2154     "rule" => array(
2155     "GarlikCrawler[ /]([0-9.]{1,10})" => "\\1"
2156     ),
2157     "uri" => "http://garlik.com/"
2158     ),
2159     "gdesktop" => array(
2160     "icon" => "gdesktop",
2161     "title" => "Google Desktop",
2162     "rule" => array(
2163     "compatible; Google Desktop" => ""
2164     ),
2165     "uri" => ""
2166     ),
2167     "genome" => array(
2168     "icon" => "robot",
2169     "title" => "Genome Machine",
2170     "rule" => array(
2171     "Genome[ \-]?Machine" => ""
2172     ),
2173     "uri" => ""
2174     ),
2175     "geohash" => array(
2176     "icon" => "geohash",
2177     "title" => "Geohash",
2178     "rule" => array(
2179     "GeoHasher(/Nutch)[ /-]([0-9.]{1,10})" => "\\2"
2180     ),
2181     "uri" => "http://geohash.org/"
2182     ),
2183     "geona" => array(
2184     "icon" => "robot",
2185     "title" => "Geona",
2186     "rule" => array(
2187     "GeonaBot[ /]([0-9.]{1,10})" => "\\1"
2188     ),
2189     "uri" => ""
2190     ),
2191     "geoblog" => array(
2192     "icon" => "robot",
2193     "title" => "The World as a Blog",
2194     "rule" => array(
2195     "The World as a Blog" => ""
2196     ),
2197     "uri" => ""
2198     ),
2199     "geourl" => array(
2200     "icon" => "geourl",
2201     "title" => "GeoUrl",
2202     "rule" => array(
2203     "geourl[ /]([0-9.]{1,10})" => "\\1",
2204     "^GeoURLBot[ /]([0-9.]{1,10})" => "\\1",
2205     ),
2206     "uri" => ""
2207     ),
2208     "getnetwise" => array(
2209     "icon" => "robot",
2210     "title" => "GetNetWise",
2211     "rule" => array(
2212     " Crayon Crawler" => ""
2213     ),
2214     "uri" => ""
2215     ),
2216     "getright" => array(
2217     "icon" => "getright",
2218     "title" => "GetRight",
2219     "rule" => array(
2220     "GetRight[ /]([0-9.]{1,10})" => "\\1"
2221     ),
2222     "uri" => ""
2223     ),
2224     "getsmart" => array(
2225     "icon" => "getsmart",
2226     "title" => "GetSmart",
2227     "rule" => array(
2228     "GetSmart[ /]([0-9.]{1,10})" => "\\1"
2229     ),
2230     "uri" => ""
2231     ),
2232     "ginger" => array(
2233     "icon" => "ginger",
2234     "title" => "GingerCrawler",
2235     "rule" => array(
2236     "GingerCrawler[ /]([0-9.]{1,10})" => "\\1"
2237     ),
2238     "uri" => "http://www.gingersoftware.com/crawler_agent.htm"
2239     ),
2240     "gigabot" => array(
2241     "icon" => "gigablast",
2242     "title" => "Gigablast",
2243     "rule" => array(
2244     "(Gigabot|Sitesearch)[/ ]([0-9.]{1,10})" => "\\2",
2245     "GigabotSiteSearch[/ ]([0-9.]{1,10})" => "\\1",
2246     ),
2247     "uri" => ""
2248     ),
2249     "girafabot" => array(
2250     "icon" => "girafa",
2251     "title" => "Girafa",
2252     "rule" => array(
2253     "Girafabot" => ""
2254     ),
2255     "uri" => ""
2256     ),
2257     "globalspec" => array(
2258     "icon" => "globalspec",
2259     "title" => "GlobalSpec",
2260     "rule" => array(
2261     "Ocelli[ /]([0-9.]{1,10})" => "\\1"
2262     ),
2263     "uri" => ""
2264     ),
2265     "glucose" => array(
2266     "icon" => "glucose",
2267     "title" => "Glucose",
2268     "rule" => array(
2269     "glucose[ /]([0-9a-z.\-]{1,10})" => "\\1"
2270     ),
2271     "uri" => ""
2272     ),
2273     "gnomit" => array(
2274     "icon" => "gnomit",
2275     "title" => "Gnomit",
2276     "rule" => array(
2277     "gnomit[ /]([0-9.]{1,10})" => "\\1"
2278     ),
2279     "uri" => "http://www.gnomit.com/",
2280     ),
2281     "goforit" => array(
2282     "icon" => "goforit",
2283     "title" => "GoForIt",
2284     "rule" => array(
2285     "^GoForIt\.com" => "",
2286     "^GOFORITBOT" => ""
2287     ),
2288     "uri" => ""
2289     ),
2290     "goguides" => array(
2291     "icon" => "robot",
2292     "title" => "GoGuides",
2293     "rule" => array(
2294     "^GoGuidesBot[ /]([0-9.]{1,10})" => "\\1"
2295     ),
2296     "uri" => "http://www.goguides.org/"
2297     ),
2298     // Needs to be before google for correct detection
2299     "mediapartners" => array(
2300     "icon" => "google",
2301     "title" => "Mediapartners",
2302     "rule" => array (
2303     "Mediapartners-Google[ /]([0-9.]{1,10})" => "\\1",
2304     "Mediapartners-Google" => ""
2305     ),
2306     "uri" => ""
2307     ),
2308     "google" => array(
2309     "icon" => "google",
2310     "title" => "Google",
2311     "rule" => array (
2312     "Googl(e|ebot)(-Image)?/([0-9.]{1,10})" => "\\3",
2313     "Googl(e|ebot)(-Image)?/" => "",
2314     "^gsa-crawler" => "",
2315     "Google Web Preview" => ""
2316     ),
2317     "uri" => "http://www.google.com/"
2318     ),
2319     "googlesitemaps" => array(
2320     "icon" => "google",
2321     "title" => "Google-Sitemaps",
2322     "rule" => array (
2323     "Googl(e|ebot)(-Sitemaps)?/([0-9.]{1,10})" => "\\3",
2324     "GSiteCrawler[ /v]*([0-9.a-z]{1,10})" => "\\1",
2325     "Googl(e|ebot)(-Sitemaps)?/" => ""
2326     ),
2327     "uri" => ""
2328     ),
2329     "googlemobile" => array(
2330     "icon" => "google",
2331     "title" => "Google-Mobile",
2332     "rule" => array (
2333     "Googl(e|ebot)(-Mobile)?/([0-9.]{1,10})" => "\\3",
2334     "Googl(e|ebot)(-Mobile)?/" => ""
2335     ),
2336     "uri" => ""
2337     ),
2338     "googleads" => array(
2339     "icon" => "google",
2340     "title" => "Google-AdsBot",
2341     "rule" => array (
2342     "^AdsBot-Google" => "",
2343     ),
2344     "uri" => ""
2345     ),
2346     "googlefeeds" => array(
2347     "icon" => "google",
2348     "title" => "Google-Feedfetcher",
2349     "rule" => array (
2350     "^Feedfetcher-Google" => "",
2351     ),
2352     "uri" => ""
2353     ),
2354     "googlverification" => array(
2355     "icon" => "google",
2356     "title" => "Google-Site-Verification",
2357     "rule" => array (
2358     "Google-Site-Verification[ /]([0-9.]{1,10})" => "\\1"
2359     ),
2360     "uri" => ""
2361     ),
2362 matthys 207 // Needs to be AFTER Google detection
2363     "goo" => array(
2364     "icon" => "goo",
2365     "title" => "Goo",
2366     "rule" => array (
2367     "(gazz|ichiro|mog(et|imogi))[ /]([0-9.]{1,10})" => "\\3",
2368     "DoCoMo[ /]([0-9.]{1,10})" => "\\1"
2369     ),
2370     "uri" => ""
2371     ),
2372 joku 63 "goongee" => array(
2373     "icon" => "robot",
2374     "title" => "GoonGee",
2375     "rule" => array (
2376     "^Big Fish[ /]v?([0-9.]{1,10})" => "\\2"
2377     ),
2378     "uri" => "http://www.goongee.com/"
2379     ),
2380     "gpost" => array(
2381     "icon" => "gpost",
2382     "title" => "GPost",
2383     "rule" => array(
2384     "^GPostbot" => ""
2385     ),
2386     "uri" => ""
2387     ),
2388 matthys 309 "grapeshot" => array(
2389     "icon" => "grapeshot",
2390     "title" => "Grapeshot",
2391     "rule" => array(
2392     "GrapeshotCrawler[/ ]([0-9.]{1,10})" => "\\1"
2393     ),
2394     "uri" => "http://www.grapeshot.co.uk/crawler.php"
2395     ),
2396 joku 63 "gregarius" => array(
2397     "icon" => "robot",
2398     "title" => "Gregarius",
2399     "rule" => array(
2400     "^Gregarius[/ ]([0-9.]{1,10})" => ""
2401     ),
2402     "uri" => ""
2403     ),
2404     "grub" => array(
2405     "icon" => "grub",
2406     "title" => "Grub",
2407     "rule" => array(
2408     "grub[ \-]?client[ /\-]{1,5}([0-9.]{1,10})" => "\\1",
2409     "grub crawler" => "",
2410     "grub[/ ]([0-9.]{1,10})" => "\\1",
2411     "^GrubNG([0-9. ]{1,10})?" => "\\1"
2412     ),
2413     "uri" => "http://www.grub.org/"
2414     ),
2415     "gulliver" => array(
2416     "icon" => "robot",
2417     "title" => "Gulliver",
2418     "rule" => array(
2419     "Gulliver" => ""
2420     ),
2421     "uri" => ""
2422     ),
2423     "guruji" => array(
2424     "icon" => "guruji",
2425     "title" => "Guruji",
2426     "rule" => array(
2427     "GurujiBot[/ ]([0-9.]{1,10})" => "\\1"
2428     ),
2429     "uri" => "http://www.guruji.com/"
2430     ),
2431     "gush" => array(
2432     "icon" => "robot",
2433     "title" => "Gush",
2434     "rule" => array(
2435     "^Gush[/ ]([0-9.]{1,10})" => "\\1"
2436     ),
2437     "uri" => ""
2438     ),
2439     "gziptester" => array(
2440     "icon" => "robot",
2441     "title" => "Gzip Tester",
2442     "rule" => array(
2443     "g(id)?zip[ \-]?test(er)?" => ""
2444     ),
2445     "uri" => ""
2446     ),
2447     "hanzoweb" => array(
2448     "icon" => "hanzoweb",
2449     "title" => "Hanzoweb",
2450     "rule" => array(
2451     "^Hanzoweb" => ""
2452     ),
2453     "uri" => ""
2454     ),
2455     "harbot" => array(
2456     "icon" => "harbot",
2457     "title" => "Harbot",
2458     "rule" => array(
2459     "^Harbot GateStation" => ""
2460     ),
2461     "uri" => ""
2462     ),
2463     "hatena" => array(
2464     "icon" => "hatena",
2465     "title" => "Hatena",
2466     "rule" => array(
2467     "Hatena (Antenna|Bookmark|Pagetitle Agent)[ /]([0-9.]{1,10})" => "\\2"
2468     ),
2469     "uri" => ""
2470     ),
2471     "htmlparser" => array(
2472     "icon" => "htmlparser",
2473     "title" => "HTMLParser",
2474     "rule" => array(
2475     "HTMLParser[ /]([0-9.]{1,10})" => "\\1"
2476     ),
2477     "uri" => "http://htmlparser.sourceforge.net/"
2478     ),
2479     "helix" => array(
2480     "icon" => "helix",
2481     "title" => "Heritrix",
2482     "rule" => array(
2483     "^helix[ /]([0-9.]{1,10})" => "\\1"
2484     ),
2485     "uri" => "http://www.sitesearch.ca/helix/"
2486     ),
2487     "heritrix" => array(
2488     "icon" => "heritrix",
2489     "title" => "Heritrix",
2490     "rule" => array(
2491     "heritrix[ /]([0-9.]{1,10})" => "\\1",
2492     "archive.org_bot" => "",
2493     "InternetArchive[ /]([0-9.a-z]{1,10})" => "\\1"
2494     ),
2495     "uri" => "http://archive.org"
2496     ),
2497     "hiddenmarket" => array(
2498     "icon" => "robot",
2499     "title" => "HiddenMarket",
2500     "rule" => array(
2501     "HiddenMarket[ /\-]([0-9.]{1,10})" => "\\1"
2502     ),
2503     "uri" => ""
2504     ),
2505     "honda" => array(
2506     "icon" => "robot",
2507     "title" => "Honda",
2508     "rule" => array(
2509     "Honda-Search[ /]([0-9.]{1,10})" => "\\1"
2510     ),
2511     "uri" => "http://www.honda-search.com"
2512     ),
2513     "hoowwwer" => array(
2514     "icon" => "hoowwwer",
2515     "title" => "HooWWWer",
2516     "rule" => array(
2517     "HooWWWer[ /]([0-9.]{1,10})" => "\\1"
2518     ),
2519     "uri" => ""
2520     ),
2521     "hotzonu" => array(
2522     "icon" => "hotzonu",
2523     "title" => "Hotzonu",
2524     "rule" => array(
2525     "Hotzonu[ /]([0-9.]{1,10})" => "\\1"
2526     ),
2527     "uri" => ""
2528     ),
2529     "houxou" => array(
2530     "icon" => "robot",
2531     "title" => "Houxou",
2532     "rule" => array(
2533     "HouxouCrawler[ /]Nutch.([0-9.]{1,10})" => "\\1",
2534     "HouxouCrawler" => ""
2535     ),
2536     "uri" => ""
2537     ),
2538     "htdig" => array(
2539     "icon" => "htdig",
2540     "title" => "ht://Dig",
2541     "rule" => array(
2542     "htdig[ /]([0-9.]{1,10})" => "\\1",
2543     "htdig" => "\\1"
2544     ),
2545     "uri" => ""
2546     ),
2547     "html2jpg" => array(
2548     "icon" => "html2jpg",
2549     "title" => "HTML2JPG",
2550     "rule" => array(
2551     "^HTML2JPG" => ""
2552     ),
2553     "uri" => ""
2554     ),
2555     "httperf" => array(
2556     "icon" => "robot",
2557     "title" => "HTTPerf",
2558     "rule" => array(
2559     "httperf[ /]([0-9.]{1,10})" => "\\1"
2560     ),
2561     "uri" => ""
2562     ),
2563     "httpunit" => array(
2564     "icon" => "httpunit",
2565     "title" => "HttpUnit",
2566     "rule" => array(
2567     "httpunit[ /]([0-9.]{1,10})" => "\\1"
2568     ),
2569     "uri" => ""
2570     ),
2571     "httrack" => array(
2572     "icon" => "httrack",
2573     "title" => "HTTrack",
2574     "rule" => array(
2575     "HTTrack[ /]([0-9.]{1,10})" => "\\1"
2576     ),
2577     "uri" => ""
2578     ),
2579     "huaweisymantec" => array(
2580     "icon" => "huaweisymantec",
2581     "title" => "HuaweiSymantec",
2582     "rule" => array(
2583     "HuaweiSymantecSpider[ /]([0-9.]{1,10})" => "\\1"
2584     ),
2585     "uri" => "http://www.huaweisymantec.com/en/IRL/spider/"
2586     ),
2587     "hungary" => array(
2588     "icon" => "hungary",
2589     "title" => "Hungary",
2590     "rule" => array(
2591     "HuRob[ /]([0-9.]{1,10})" => "\\1"
2592     ),
2593     "uri" => ""
2594     ),
2595 joku 261 "i18n" => array(
2596     "icon" => "w3c",
2597     "title" => "W3C I18n Checker",
2598     "rule" => array(
2599     "^W3C_I18n-Checker[ /]([0-9.]{1,10})" => "\\1"
2600 joku 169 ),
2601 joku 261 "uri" => "http://validator.w3.org/services"
2602     ),
2603 joku 63 "iakke" => array(
2604     "icon" => "iakke",
2605 matthys 209 "title" => "Iakké",
2606 joku 63 "rule" => array(
2607     "^iakke_bot[ /]v([0-9.]{1,10})" => "\\1"
2608     ),
2609     "uri" => "http://fr.iakke.com/"
2610     ),
2611     "iask" => array(
2612     "icon" => "iask",
2613     "title" => "IAsk",
2614     "rule" => array(
2615     "iaskspider[ /]([0-9.]{1,10})" => "\\1",
2616     "^iaskspider" => ""
2617     ),
2618     "uri" => "http://iask.com"
2619     ),
2620     "icc" => array(
2621     "icon" => "icc",
2622     "title" => "ICC-Crawler",
2623     "rule" => array(
2624     "^ICC-Crawler" => ""
2625     ),
2626     "uri" => "http://kc.nict.go.jp/icc/crawl.html"
2627     ),
2628     "icerocket" => array(
2629     "icon" => "icerocket",
2630     "title" => "Icerocket",
2631     "rule" => array(
2632     "BlogzIce[ /]([0-9.]{1,10})" => "\\1",
2633     "BlogSearch[ /]([0-9.]{1,10})" => "\\1"
2634     ),
2635     "uri" => ""
2636     ),
2637     "icjobs" => array(
2638     "icon" => "icjobs",
2639     "title" => "iCjobs",
2640     "rule" => array(
2641     "iCjobs[ /]([0-9.]{1,10})" => "\\1"
2642     ),
2643     "uri" => "http://www.icjobs.de/"
2644     ),
2645     "icra" => array(
2646     "icon" => "icra",
2647     "title" => "ICRA",
2648     "rule" => array(
2649     "^ICRA_Semantic_spider[ /]([0-9.]{1,10})" => "\\1"
2650     ),
2651     "uri" => "http://www.icra.org"
2652     ),
2653     "iknow" => array(
2654     "icon" => "robot",
2655     "title" => "I know",
2656     "rule" => array(
2657     "Comaneci_bot[ /]([0-9.]{1,10})" => "\\1"
2658     ),
2659     "uri" => ""
2660     ),
2661     "ilial" => array(
2662     "icon" => "robot",
2663     "title" => "Ilial",
2664     "rule" => array(
2665     "ilial[ /]Nutch[ \-]([0-9.]{1,10})" => "\\1"
2666     ),
2667     "uri" => ""
2668     ),
2669     "ilocal" => array(
2670     "icon" => "ilocal",
2671     "title" => "ilocal",
2672     "rule" => array(
2673     "^ilocal" => "",
2674     ),
2675     "uri" => "http://www.ilocal.nl"
2676     ),
2677     "ilse" => array(
2678     "icon" => "ilse",
2679     "title" => "Ilse",
2680     "rule" => array(
2681     "I(NGRID|lseRobot|lseBot)[ /]([0-9.]{1,10})" => "\\1"
2682     ),
2683     "uri" => "http://ilse.nl/"
2684     ),
2685     "iltrovatore" => array(
2686     "icon" => "iltrovatore",
2687     "title" => "IlTrovatore",
2688     "rule" => array(
2689     "iltrovatore-setaccio[ /]([0-9.]{1,10})" => "\\1",
2690     "Iltrovatore-Setaccio" => "",
2691     "iltrovatore[ /]([0-9.]{1,10})" => "\\1"
2692     ),
2693     "uri" => ""
2694     ),
2695     "indylibrary" => array(
2696     "icon" => "robot",
2697     "title" => "Indy Library",
2698     "rule" => array(
2699     "Indy[ \-]?Library" => ""
2700     ),
2701     "uri" => ""
2702     ),
2703     "inela" => array(
2704     "icon" => "robot",
2705     "title" => "Inela",
2706     "rule" => array(
2707     "InelaBot[ /]([0-9.]{1,10})" => ""
2708     ),
2709     "uri" => "http://inelegant.org/bot"
2710     ),
2711     "ineturl" => array(
2712     "icon" => "robot",
2713     "title" => "InetURL",
2714     "rule" => array(
2715     "InetURL.?[ /]([0-9.]{1,10})" => ""
2716     ),
2717     "uri" => ""
2718     ),
2719     "infoart" => array(
2720     "icon" => "robot",
2721     "title" => "InfoArt",
2722     "rule" => array(
2723     "InfoArt crawler" => ""
2724     ),
2725     "uri" => ""
2726     ),
2727     "infomine" => array(
2728     "icon" => "infomine",
2729     "title" => "INFOMINE",
2730     "rule" => array(
2731     "^DataFountains/DMOZ" => "",
2732     "^INFOMINE[ /]([0-9.]{1,10})" => "\\1"
2733     ),
2734     "uri" => "http://infomine.ucr.edu/"
2735     ),
2736     "infoseek" => array(
2737     "icon" => "infoseek",
2738     "title" => "Infoseek",
2739     "rule" => array(
2740     "SideWinder[ /]?([0-9a-z.]{1,10})" => "\\1",
2741     "Infoseek" => ""
2742     ),
2743     "uri" => ""
2744     ),
2745     "inktomi" => array(
2746     "icon" => "inktomi",
2747     "title" => "Inktomi",
2748     "rule" => array(
2749     "slurp@inktomi\.com" => ""
2750     ),
2751     "uri" => ""
2752     ),
2753     "innerprise" => array(
2754     "icon" => "robot",
2755     "title" => "Innerprise",
2756     "rule" => array(
2757     "^InnerpriseBot[ /]([0-9.]{1,10})" => "\\1",
2758     "URL[ _]Spider[ _]Pro[ /]([0-9.+]{1,10})" => "\\1",
2759     "^ES[ .]NET[ _]Crawler[ /]([0-9.]{1,10})" => "\\1"
2760     ),
2761     "uri" => "http://www.innerprise.com/"
2762     ),
2763     "inria" => array(
2764     "icon" => "inria",
2765     "title" => "Inria",
2766     "rule" => array(
2767     "^xyro_" => ""
2768     ),
2769     "uri" => ""
2770     ),
2771     "insitor" => array(
2772     "icon" => "insitor",
2773     "title" => "Insitor",
2774     "rule" => array(
2775     "^Insitor(,|\.|naut)" => ""
2776     ),
2777     "uri" => "http://www.insitor.com/"
2778     ),
2779     "internetninja" => array(
2780     "icon" => "robot",
2781     "title" => "Internet Ninja",
2782     "rule" => array(
2783     "^Internet Ninja[ /]([0-9.]{1,10})" => "\\1"
2784     ),
2785     "uri" => ""
2786     ),
2787     "internetseer" => array(
2788     "icon" => "internetseer",
2789     "title" => "InternetSeer",
2790     "rule" => array(
2791     "^InternetSeer\.com" => ""
2792     ),
2793     "uri" => ""
2794     ),
2795     "interseek" => array(
2796     "icon" => "interseek",
2797     "title" => "Interseek",
2798     "rule" => array(
2799     "Interseek.com" => ""
2800     ),
2801     "uri" => "http://www.interseek.com"
2802     ),
2803     "intravnews" => array(
2804     "icon" => "intravnews",
2805     "title" => "IntraVnews",
2806     "rule" => array(
2807     "IntraVnews[ /]([0-9.]{1,10})" => "\\1"
2808     ),
2809     "uri" => "http://www.intravnews.com/"
2810     ),
2811     "ip2location" => array(
2812     "icon" => "ip2location",
2813     "title" => "IP2LocationBot",
2814     "rule" => array(
2815     "^IP2(Map|Location)Bot[ /]([0-9.]{1,10})" => "\\2"
2816     ),
2817     "uri" => "http://www.ip2location.com"
2818     ),
2819     "ipworks" => array(
2820     "icon" => "ipworks",
2821     "title" => "IP*Works",
2822     "rule" => array(
2823     "^IP\*Works\! V([0-9.]{1,10})" => "\\1"
2824     ),
2825     "uri" => "http://www.nsoftware.com/ipworks/"
2826     ),
2827     "irca" => array(
2828     "icon" => "robot",
2829     "title" => "Novell iChain Cool Solutions caching",
2830     "rule" => array(
2831 matthys 209 "^ICRA_(label_generator|Semantic_spider)[ /]([0-9.]{1,10})" => "\\2",
2832     "^Mozilla[/ ]([0-9.]{1,10})[/ ]\(compatible[ ;]*ICS" => "\\1"
2833 joku 63 ),
2834     "uri" => "http://www.icra.org"
2835     ),
2836     "irvine" => array(
2837     "icon" => "robot",
2838     "title" => "Irvine",
2839     "rule" => array(
2840     "Irvine[ /]([0-9.]{1,10})" => "\\1"
2841     ),
2842     "uri" => ""
2843     ),
2844     "ips-agent" => array(
2845     "icon" => "robot",
2846 matthys 209 "title" => "IPS-agent",
2847 joku 63 "rule" => array(
2848     "ips-agent" => ""
2849     ),
2850     "uri" => ""
2851     ),
2852     "isara" => array(
2853     "icon" => "isara",
2854 matthys 209 "title" => "Isara",
2855 joku 63 "rule" => array(
2856     "Isara[ -]([0-9.]{1,10})" => "\\1"
2857     ),
2858     "uri" => "http://www.isara.org/"
2859     ),
2860     "isspider" => array(
2861     "icon" => "robot",
2862     "title" => "ISSpider",
2863     "rule" => array(
2864     "ISSpider[ /\-]([0-9.]{1,10})" => "\\1"
2865     ),
2866     "uri" => ""
2867     ),
2868     "it2media" => array(
2869     "icon" => "it2media",
2870     "title" => "IT2media",
2871     "rule" => array(
2872     "www.adressendeutschland.de" => ""
2873     ),
2874     "uri" => "http://www.adressendeutschland.de/"
2875     ),
2876     "ivia" => array(
2877     "icon" => "robot",
2878     "title" => "iVia",
2879     "rule" => array(
2880     "iVia Site Checker.?[ /]([0-9.]{1,10})" => "\\1"
2881     ),
2882     "uri" => ""
2883     ),
2884     "jeteye" => array(
2885     "icon" => "jeteye",
2886     "title" => "Jeteye",
2887     "rule" => array(
2888     "Jetbot[ /]([0-9.]{1,10})" => "\\1"
2889     ),
2890     "uri" => ""
2891     ),
2892     "jobsde" => array(
2893     "icon" => "jobsde",
2894     "title" => "jobs.de",
2895     "rule" => array(
2896     "www\.jobs\.de" => "",
2897     "jobs.de-Robot" => ""
2898     ),
2899     "uri" => "http://www.jobs.de/"
2900     ),
2901     "jpluck" => array(
2902     "icon" => "robot",
2903     "title" => "Jpluck",
2904     "rule" => array(
2905     "JPluck[ /]([0-9a-z.]{1,10})" => "\\1"
2906     ),
2907     "uri" => ""
2908     ),
2909     "jxta" => array(
2910     "icon" => "robot",
2911     "title" => "Jxta",
2912     "rule" => array(
2913     "falcon[ /]([0-9.]{1,10})" => "\\1"
2914     ),
2915     "uri" => ""
2916     ),
2917     "jyte" => array(
2918     "icon" => "jyte",
2919     "title" => "Jyte",
2920     "rule" => array(
2921     "jyte_fetcher[ /]([0-9.]{1,10})" => "\\1"
2922     ),
2923     "uri" => ""
2924     ),
2925     "jyxo" => array(
2926     "icon" => "jyxo",
2927     "title" => "Jyxo",
2928     "rule" => array(
2929     "Jyxobot[ /]([0-9.]{1,10})" => "\\1"
2930     ),
2931     "uri" => ""
2932     ),
2933     "kalooga" => array(
2934     "icon" => "kalooga",
2935     "title" => "Kalooga",
2936     "rule" => array(
2937     "kalooga[ /-]([0-9.]{1,10})" => "\\1",
2938     "^kalooga" => "",
2939     "KaloogaBot" => ""
2940     ),
2941     "uri" => "http://www.kalooga.com"
2942     ),
2943     "keywen" => array(
2944     "icon" => "keywen",
2945     "title" => "Keywen",
2946     "rule" => array(
2947     "EasyDL[ /]([0-9.]{1,10})" => "\\1"
2948     ),
2949     "uri" => ""
2950     ),
2951     "kilomonkey" => array(
2952     "icon" => "robot",
2953     "title" => "Kilomonkey",
2954     "rule" => array(
2955     "kmky-not-a-bot[ /]([0-9.]{1,10})" => "\\1"
2956     ),
2957     "uri" => "http://www.kilomonkey.com/notabot.txt"
2958     ),
2959     "kinja" => array(
2960     "icon" => "kinja",
2961     "title" => "Kinja",
2962     "rule" => array(
2963     "kinjabot[ /]([0-9.]{1,10})" => "\\1",
2964     "^kinjabot" => ""
2965     ),
2966     "uri" => ""
2967     ),
2968     "kiwi" => array(
2969     "icon" => "robot",
2970     "title" => "KiwiStatus",
2971     "rule" => array(
2972     "KiwiStatus[ /]([0-9.]{1,10})" => "\\1",
2973     "KiwiStatus" => ""
2974     ),
2975     "uri" => "http://www.nzs.com/kiei-status/"
2976     ),
2977     "lachesis" => array(
2978     "icon" => "robot",
2979     "title" => "Lachesis",
2980     "rule" => array(
2981     "lachesis" => ""
2982     ),
2983     "uri" => ""
2984     ),
2985     "lanshan" => array(
2986     "icon" => "robot",
2987     "title" => "Lachesis",
2988     "rule" => array(
2989     "lanshanbot[/ ]([0-9.]{1,10})" => "\\1"
2990     ),
2991     "uri" => ""
2992     ),
2993     "lapozz" => array(
2994     "icon" => "lapozz",
2995     "title" => "Lapozz",
2996     "rule" => array(
2997     "LapozzBot[/ ]?([0-9.]{1,10})" => "\\1"
2998     ),
2999     "uri" => ""
3000     ),
3001     "larbin" => array(
3002     "icon" => "robot",
3003     "title" => "Larbin",
3004     "rule" => array(
3005     "larbin[_/ ]?([0-9.]{1,10})" => "\\1"
3006     ),
3007     "uri" => ""
3008     ),
3009     "laurion" => array(
3010     "icon" => "robot",
3011     "title" => "Laurion",
3012     "rule" => array(
3013     "^IPiumBot" => ""
3014     ),
3015     "uri" => "http://www.laurion.com/"
3016     ),
3017     "leechget" => array(
3018     "icon" => "leechget",
3019     "title" => "LeechGet",
3020     "rule" => array(
3021     "^LeechGet[ /]([0-9.]{1,10})" => "\\1"
3022     ),
3023     "uri" => ""
3024     ),
3025     "linguaseek" => array(
3026     "icon" => "linguaseek",
3027     "title" => "Linguaseek",
3028     "rule" => array(
3029     "^LinguaBot[ /]v([0-9.]{1,10})" => "\\1"
3030     ),
3031     "uri" => "http://www.linguaseek.com/"
3032     ),
3033     "linguee" => array(
3034     "icon" => "linguee",
3035     "title" => "Linguee",
3036     "rule" => array(
3037     "^Linguee Bot" => ""
3038     ),
3039     "uri" => "http://www.linguee.com/"
3040     ),
3041     "linkguard" => array(
3042     "icon" => "robot",
3043     "title" => "Linkguard",
3044     "rule" => array(
3045     "Linkguard Online[ /]([0-9.]{1,10})" => "\\1"
3046     ),
3047     "uri" => "http://www.linkguard.com/"
3048     ),
3049     "linklint" => array(
3050     "icon" => "linklint",
3051     "title" => "Linklint",
3052     "rule" => array(
3053     "^Linklint[-\-a-z /]*([0-9.]{1,10})" => "\\1"
3054     ),
3055     "uri" => "http://www.linklint.org/"
3056     ),
3057     "linkman" => array(
3058     "icon" => "linkman",
3059     "title" => "Linkman",
3060     "rule" => array(
3061     "\(compatible; Linkman\)" => ""
3062     ),
3063     "uri" => ""
3064     ),
3065     // Needs to be before linkcheck for correct detection
3066     "linkchecker" => array(
3067     "icon" => "w3c",
3068     "title" => "W3C Link Checker",
3069     "rule" => array(
3070     "^W3C-checklink[ /]([0-9.]{1,10})" => "\\1"
3071     ),
3072     "uri" => "http://validator.w3.org/checklink"
3073     ),
3074     "linkcheck" => array(
3075     "icon" => "linkcheck",
3076     "title" => "Linkcheck",
3077     "rule" => array(
3078     "checklink[ /]([0-9.]{1,10})" => "\\1",
3079     "Link[ \-]?(Chec(k|ker)|Val(et|idator))" => "",
3080     "Adaxas Spider" => "",
3081     "Agent-SharewarePlazaFileCheckBot[ /]([0-9.]{1,10})" => "\\1",
3082     "NetMechanic V([0-9.]{1,10})" => "\\1",
3083     "^InfoLink" => "",
3084     "InternetLinkAgent" => "",
3085     "; SPENG\)" => "",
3086     "SharewarePlazaFileCheckBot" => "",
3087     "fileboost.net" => "",
3088     "^billbot" => ""
3089     ),
3090     "uri" => ""
3091     ),
3092     "linkru" => array(
3093     "icon" => "robot",
3094     "title" => "Link.RU",
3095     "rule" => array(
3096     "^Link.RU bot" => ""
3097     ),
3098     "uri" => ""
3099     ),
3100     "linkssql" => array(
3101     "icon" => "robot",
3102     "title" => "Links SQL",
3103     "rule" => array(
3104     "links sql" => ""
3105     ),
3106     "uri" => ""
3107     ),
3108     "linksweeper" => array(
3109     "icon" => "robot",
3110     "title" => "Link Sweeper",
3111     "rule" => array(
3112     "LinkSweeper[ /]([0-9.]{1,10})" => "\\1"
3113     ),
3114     "uri" => ""
3115     ),
3116     "linkwalker" => array(
3117     "icon" => "robot",
3118     "title" => "Link Walker",
3119     "rule" => array(
3120     "^LinkWalker" => ""
3121     ),
3122     "uri" => ""
3123     ),
3124     "lipperhey" => array(
3125     "icon" => "lipperhey",
3126     "title" => "Lipperhey",
3127     "rule" => array(
3128     "Lipperhey Spider" => ""
3129     ),
3130     "uri" => "http://www.lipperhey.com"
3131     ),
3132     "livedoor" => array(
3133     "icon" => "livedoor",
3134     "title" => "Livedoor",
3135     "rule" => array(
3136     "^Livedoor( SF( - California Crawl)?|Checkers)[ /]" => ""
3137     ),
3138     "uri" => ""
3139     ),
3140     "livejournal" => array(
3141     "icon" => "livejournal",
3142     "title" => "Live Journal",
3143     "rule" => array(
3144     "^LiveJournal\.com" => ""
3145     ),
3146     "uri" => "http://www.livejournal.com"
3147     ),
3148     "ljpic" => array(
3149     "icon" => "ljpic",
3150     "title" => "ljpic",
3151     "rule" => array(
3152     "LjSEEK Picture-Bot[ /]+([0-9.]{1,10})" => "\\1"
3153     ),
3154     "uri" => "http://www.ljpic.com"
3155     ),
3156     "lmspider" => array(
3157     "icon" => "robot",
3158     "title" => "Lmspider",
3159     "rule" => array(
3160     "^lmspider" => ""
3161     ),
3162     "uri" => ""
3163     ),
3164     "locators" => array(
3165     "icon" => "robot",
3166     "title" => "Locaters",
3167     "rule" => array(
3168     "^FiNDoBot[/ ]([0-9a-z.]{1,10})" => "\\1"
3169     ),
3170     "uri" => ""
3171     ),
3172     "look" => array(
3173     "icon" => "look",
3174     "title" => "Look",
3175     "rule" => array(
3176     "www\.look\.com" => "",
3177     "Lookbot" => ""
3178     ),
3179     "uri" => ""
3180     ),
3181     "looksmart" => array(
3182     "icon" => "looksmart",
3183     "title" => "LookSmart",
3184     "rule" => array(
3185     "^Martini" => "",
3186     "^MantraAgent" => "",
3187     "FurlBot" => "",
3188     "looksmart-sv-fw" => ""
3189     ),
3190     "uri" => ""
3191     ),
3192     "loop" => array(
3193     "icon" => "loop",
3194     "title" => "LOOP",
3195     "rule" => array(
3196     "NetResearchServer[ /]([0-9.]{1,10})" => "\\1"
3197     ),
3198     "uri" => ""
3199     ),
3200     "lotkyll" => array(
3201     "icon" => "robot",
3202     "title" => "Lotkyll",
3203     "rule" => array(
3204     "Lotkyll" => ""
3205     ),
3206     "uri" => ""
3207     ),
3208     "lwp" => array(
3209     "icon" => "robot",
3210     "title" => "lwp",
3211     "rule" => array(
3212     "lwp(-trivial|::simple)[ /]([0-9.]{1,10})" => "\\2"
3213     ),
3214     "uri" => ""
3215     ),
3216     "lycos" => array(
3217     "icon" => "lycos",
3218     "title" => "Lycos",
3219     "rule" => array(
3220     "Lycos_Spider_" => ""
3221     ),
3222     "uri" => ""
3223     ),
3224     "magpierss" => array(
3225     "icon" => "rss",
3226     "title" => "MagpieRSS",
3227     "rule" => array(
3228     "MagpieRSS" => ""
3229     ),
3230     "uri" => "http://magpierss.sourceforge.net/"
3231     ),
3232     "mailru" => array(
3233     "icon" => "mailru",
3234 matthys 209 "title" => "Mail.Ru",
3235 joku 63 "rule" => array(
3236 matthys 195 "^Mail.Ru[ /]([0-9.]{1,10})" => "\\1",
3237     "Mail.RU_Bot[ /]([0-9.]{1,10})" => "\\1"
3238 joku 63 ),
3239 matthys 209 "uri" => "http://www.mail.ru/"
3240 joku 63 ),
3241     "mailsweeper" => array(
3242     "icon" => "robot",
3243     "title" => "Mail Sweeper",
3244     "rule" => array(
3245     "Mail[ \-]?Sweeper" => ""
3246     ),
3247     "uri" => ""
3248     ),
3249     "marvin" => array(
3250     "icon" => "robot",
3251     "title" => "Marvin",
3252     "rule" => array(
3253     "^Marvin" => ""
3254     ),
3255     "uri" => ""
3256     ),
3257     "matkurja" => array(
3258     "icon" => "matkurja",
3259     "title" => "Mat'Kurja",
3260     "rule" => array(
3261     "Mosad[ /]([0-9.]{1,10})" => "\\1"
3262     ),
3263     "uri" => ""
3264     ),
3265     "mavicanet" => array(
3266     "icon" => "mavicanet",
3267     "title" => "Mavicanet",
3268     "rule" => array(
3269     "Mavicanet robot" => ""
3270     ),
3271     "uri" => ""
3272     ),
3273 matthys 292 "meanpath" => array(
3274     "icon" => "meanpath",
3275 matthys 290 "title" => "Meanpath",
3276 matthys 280 "rule" => array(
3277     "meanpathbot[ /]([0-9.]{1,10})" => "\\1"
3278     ),
3279     "uri" => "http://www.meanpath.com/meanpathbot.html"
3280 matthys 275 ),
3281 matthys 280 "mediater" => array(
3282 joku 63 "icon" => "robot",
3283     "title" => "Mediater",
3284     "rule" => array (
3285     "^libwww[ /]([0-9.]{1,10})" => "\\1"
3286 matthys 280 ),
3287     "uri" => ""
3288 joku 63 ),
3289 matthys 305 "memory" => array(
3290     "icon" => "memory",
3291     "title" => "internet Memory",
3292     "rule" => array (
3293     "memoryBot[ /]([0-9.]{1,10})" => "\\1"
3294     ),
3295     "uri" => "http://internetmemory.org/en/"
3296     ),
3297 joku 63 "mercator" => array(
3298     "icon" => "robot",
3299     "title" => "Mercator",
3300     "rule" => array(
3301     "Mercator" => ""
3302     ),
3303     "uri" => ""
3304     ),
3305     "metacarta" => array(
3306     "icon" => "robot",
3307     "title" => "Metacarta",
3308     "rule" => array(
3309     "^RRC (crawler_admin@bigfoot.com)" => "",
3310     "^flunky" => "",
3311     "^Mozilla.*\(samualt9@bigfoot.com\)$" => ""
3312     ),
3313     "uri" => "http://www.metacarta.com"
3314     ),
3315     "metager" => array(
3316     "icon" => "metager",
3317     "title" => "MetaGer",
3318     "rule" => array(
3319     "MetaGer" => ""
3320     ),
3321     "uri" => ""
3322     ),
3323     "metamark" => array(
3324     "icon" => "robot",
3325     "title" => "Metamark",
3326     "rule" => array(
3327     "^XRL[ /]([0-9.a-z]{1,10})" => "\\1"
3328     ),
3329     "uri" => "http://metamark.net"
3330     ),
3331     "metamedic" => array(
3332     "icon" => "metamedic",
3333     "title" => "MetaMedic",
3334     "rule" => array(
3335     "MediBot[ /]([0-9.]{1,10})" => "\\1"
3336     ),
3337     "uri" => ""
3338     ),
3339     "mirago" => array(
3340     "icon" => "mirago",
3341     "title" => "Mirago",
3342     "rule" => array(
3343     "Mirago" => ""
3344     ),
3345     "uri" => ""
3346     ),
3347     "misterwong" => array(
3348     "icon" => "misterwong",
3349     "title" => "Mister Wong",
3350     "rule" => array(
3351     "Mister Wong-Bot[ /]([0-9.]{1,10})" => "\\1"
3352     ),
3353     "uri" => "http://www.mister-wong.de/"
3354     ),
3355     "miva" => array(
3356     "icon" => "miva",
3357     "title" => "Miva",
3358     "rule" => array(
3359     "AlgoFeedback@miva\.com" => ""
3360     ),
3361     "uri" => "http://www.miva.com/"
3362     ),
3363     "mj12" => array(
3364     "icon" => "mj12",
3365     "title" => "Majestic-12",
3366     "rule" => array(
3367     "Mj12bot[ /]v?([0-9.]{1,10})" => "\\1",
3368     "MJ12bot \(mini\)[ /]([0-9.]{1,10})" => "\\1"
3369     ),
3370     "uri" => "http://majestic12.co.uk/bot.php"
3371     ),
3372     "mlbot" => array(
3373     "icon" => "robot",
3374     "title" => "MLBot",
3375     "rule" => array(
3376     "^MLBot" => ""
3377     ),
3378     "uri" => "http://www.metadatalabs.com/mlbot"
3379     ),
3380     "mnogo" => array(
3381     "icon" => "mnogo",
3382     "title" => "mnoGoSearch",
3383     "rule" => array(
3384     "Mnogosearch[ /\-]([0-9.]{1,10})" => "\\1"
3385     ),
3386     "uri" => "http://www.mnogosearch.org/"
3387     ),
3388     "mobileok" => array(
3389     "icon" => "mobileok",
3390     "title" => "W3C mobileOK",
3391     "rule" => array(
3392     "^W3C-mobileOK" => ""
3393     ),
3394     "uri" => "http://www.w3.org/2006/07/mobileok-ddc"
3395     ),
3396 matthys 296 "mojeek" => array(
3397     "icon" => "mojeek",
3398     "title" => "Mojeek",
3399 joku 63 "rule" => array(
3400     "MojeekBot[ /]([0-9.]{1,10})" => "\\1"
3401     ),
3402 matthys 284 "uri" => "https://www.mojeek.com/bot.html"
3403 joku 63 ),
3404     "momspider" => array(
3405     "icon" => "robot",
3406     "title" => "MOM Spider",
3407     "rule" => array(
3408     "MOMspider[ /]([0-9.]{1,10})" => "\\1"
3409     ),
3410     "uri" => ""
3411     ),
3412     "moreover" => array(
3413     "icon" => "moreover",
3414     "title" => "Moreover",
3415     "rule" => array(
3416     "^Moreoverbot[ /]([0-9.]{1,10})" => "\\1"
3417     ),
3418     "uri" => ""
3419     ),
3420     "movabletype" => array(
3421     "icon" => "movabletype",
3422     "title" => "Movable Type",
3423     "rule" => array(
3424     "MovableType[ /]([0-9.]{1,10})" => "\\1"
3425     ),
3426     "uri" => ""
3427     ),
3428     "mozdex" => array(
3429     "icon" => "mozdex",
3430     "title" => "MozDex",
3431     "rule" => array(
3432     "mozDex[ /]([0-9.]{1,6}(-dev)?)" => "\\1"
3433     ),
3434     "uri" => ""
3435     ),
3436     "mqbot" => array(
3437     "icon" => "robot",
3438     "title" => "MQbot",
3439     "rule" => array(
3440     "MQbot" => ""
3441     ),
3442     "uri" => ""
3443     ),
3444     "msnbot" => array(
3445     "icon" => "bing",
3446 matthys 205 "title" => "Bing",
3447 joku 63 "rule" => array(
3448     "MSN(BOT|PTC)[ /]([0-9.]{1,10})" => "\\2",
3449     "MS Search ([0-9.]{1,10}) Robot" => "\\1",
3450     "MSNBOT-(MEDIA|PRODUCTS|NewsBlogs)[ /]([0-9.]{1,10})" => "\\2",
3451     "bingbot[ /]([0-9.]{1,10})" => "\\1"
3452     ),
3453     "uri" => ""
3454     ),
3455     "msnbot-mobile" => array(
3456     "icon" => "msn",
3457     "title" => "MSN Mobile",
3458     "rule" => array(
3459     "MSNBOT-(MOBILE|)[ /]([0-9.]{1,10})" => "\\2",
3460     "MSNBOT_Mobile" => "",
3461     "MSMOBOT" => ""
3462     ),
3463     "uri" => ""
3464     ),
3465     "mslivebot" => array(
3466     "icon" => "livesearch",
3467     "title" => "MS Live Search",
3468     "rule" => array(
3469     "MSNBOT-(Academic)[ /]([0-9.]{1,10})" => "\\2"
3470     ),
3471     "uri" => ""
3472     ),
3473     "msproxy" => array(
3474     "icon" => "robot",
3475     "title" => "MSProxy",
3476     "rule" => array(
3477     "MSProxy[ /]([0-9.]{1,10})" => "\\1"
3478     ),
3479     "uri" => ""
3480     ),
3481     "msrbot" => array(
3482     "icon" => "robot",
3483     "title" => "MSRBOT",
3484     "rule" => array(
3485     "^MSRBOT" => ""
3486     ),
3487     "uri" => "http://research.microsoft.com/research/sv/msrbot/"
3488     ),
3489     "mswebdav" => array(
3490     "icon" => "robot",
3491     "title" => "MS-WebDAV",
3492     "rule" => array(
3493     "Microsoft[ \-]?WebDAV[ \-]?MiniRedir" => ""
3494     ),
3495     "uri" => ""
3496     ),
3497     "mticon" => array(
3498     "icon" => "robot",
3499     "title" => "MTIcon",
3500     "rule" => array(
3501     "MTIcon[/ ]([0-9.]{1,10})" => "\\1"
3502     ),
3503     "uri" => ""
3504     ),
3505     "multipage-validator" => array(
3506     "icon" => "multipage-validator",
3507     "title" => "Multipage Validator",
3508     "rule" => array(
3509     "^W3C_Multipage_Validator[ /]([0-9.]{1,10})" => "\\1"
3510     ),
3511     "uri" => "http://www.validator.ca/"
3512     ),
3513     "myrss" => array(
3514     "icon" => "rss",
3515     "title" => "MyRSS",
3516     "rule" => array(
3517     "MyRSS.jp[/ ]([0-9.]{1,10})" => "\\1"
3518     ),
3519     "uri" => ""
3520     ),
3521     "multimap" => array(
3522     "icon" => "robot",
3523     "title" => "Multimap",
3524     "rule" => array(
3525     "Multimap Geotag Blog Parser[/ ]([0-9.]{1,10})" => "\\1"
3526     ),
3527     "uri" => ""
3528     ),
3529     "najdi" => array(
3530     "icon" => "najdi",
3531     "title" => "Najdi.si",
3532     "rule" => array(
3533     "Najdi.si" => ""
3534     ),
3535     "uri" => "http://www.najdi.si"
3536     ),
3537     "nameprotect" => array(
3538     "icon" => "nameprotect",
3539     "title" => "Name Protect",
3540     "rule" => array(
3541     "NPBot" => ""
3542     ),
3543     "uri" => ""
3544     ),
3545     "nationaldirectory" => array(
3546     "icon" => "robot",
3547     "title" => "National Directory",
3548     "rule" => array(
3549     "NationalDirectory-WebSpider[ /]([0-9.]{1,10})" => "\\1"
3550     ),
3551     "uri" => ""
3552     ),
3553     "natsumican" => array(
3554     "icon" => "robot",
3555     "title" => "Natsu Mican",
3556     "rule" => array(
3557     "NATSU[ \-]MICAN[/ ]([0-9a-z.]{1,10})" => "\\1"
3558     ),
3559     "uri" => ""
3560     ),
3561     "naverbot" => array(
3562     "icon" => "naverbot",
3563     "title" => "Naver",
3564     "rule" => array(
3565     "NaverBot([_\-]dloader)?[/ \-]([0-9.]{1,10})" => "\\2",
3566     "Naver(Bot)?" => "",
3567     "^nabot" => ""
3568     ),
3569     "uri" => ""
3570     ),
3571     "navisso" => array(
3572     "icon" => "navisso",
3573     "title" => "Navisso",
3574     "rule" => array(
3575     "Navisso(Bot)?" => ""
3576     ),
3577     "uri" => "http://www.navisso.com"
3578     ),
3579     "neofonie" => array(
3580     "icon" => "robot",
3581 matthys 205 "title" => "Neofonie",
3582 joku 63 "rule" => array(
3583     "www.neofonie.de" => ""
3584     ),
3585     "uri" => "http://www.neofonie.de/loesungen/search/robot.html"
3586     ),
3587     "neomo" => array(
3588     "icon" => "robot",
3589     "title" => "Neomo",
3590     "rule" => array(
3591     "Francis[ /]([0-9.]{1,10})" => "\\1"
3592     ),
3593     "uri" => ""
3594     ),
3595     "nessus" => array(
3596     "icon" => "nessus",
3597     "title" => "Nessus",
3598     "rule" => array(
3599     "Nessus\)$" => ""
3600     ),
3601     "uri" => ""
3602     ),
3603 matthys 279 "nerdybot" => array(
3604 matthys 280 "icon" => "robot",
3605     "title" => "NerdyBot",
3606     "rule" => array(
3607     "NerdyBot" => ""
3608     ),
3609     "uri" => "http://nerdybot.com/"
3610 matthys 279 ),
3611 joku 63 "nerdbynature" => array(
3612     "icon" => "nerdbynature",
3613     "title" => "NerdByNature",
3614     "rule" => array(
3615     "NerdByNature.Bot" => ""
3616     ),
3617     "uri" => "http://www.nerdbynature.net/bot"
3618     ),
3619     "netants" => array(
3620     "icon" => "netants",
3621     "title" => "NetAnts",
3622     "rule" => array(
3623     "NetAnts[ /]([0-9.]{1,10})" => "\\1"
3624     ),
3625     "uri" => ""
3626     ),
3627     "netcraft" => array(
3628     "icon" => "netcraft",
3629     "title" => "Netcraft",
3630     "rule" => array(
3631     "netcraft" => ""
3632     ),
3633     "uri" => ""
3634     ),
3635     "netinfo" => array(
3636     "icon" => "netinfo",
3637     "title" => "Netinfo",
3638     "rule" => array(
3639     "^NetinfoBot[ /]([0-9.]{1,10})" => "\\1"
3640     ),
3641     "uri" => "http://netinfo.bg/"
3642     ),
3643     "netluchs" => array(
3644     "icon" => "netluchs",
3645     "title" => "Netluchs",
3646     "rule" => array(
3647     "Netluchs[ /]([0-9.a-z]{1,10})" => "\\1"
3648     ),
3649     "uri" => "http://www.netluchs.de/"
3650     ),
3651     "netmechanic" => array(
3652     "icon" => "netmechanic",
3653     "title" => "NetMechanic",
3654     "rule" => array(
3655     "NetMechanic[ /V]{1,5}([0-9.]{1,10})" => "\\1"
3656     ),
3657     "uri" => ""
3658     ),
3659     "netnose" => array(
3660     "icon" => "netnose",
3661     "title" => "NetNose",
3662     "rule" => array(
3663     "NetNose[ \-]Crawler[/ ]([0-9.]{1,10})" => "\\1"
3664     ),
3665     "uri" => ""
3666     ),
3667     "netoskop" => array(
3668     "icon" => "robot",
3669     "title" => "Netoskop",
3670     "rule" => array(
3671     "netoskop" => ""
3672     ),
3673     "uri" => ""
3674     ),
3675     "netpromoter" => array(
3676     "icon" => "netpromoter",
3677     "title" => "NetPromoter",
3678     "rule" => array(
3679     "NetPromoter Spider" => ""
3680     ),
3681     "uri" => "http://www.net-promoter.com/"
3682     ),
3683     "netprospector" => array(
3684     "icon" => "robot",
3685     "title" => "Netprospector",
3686     "rule" => array(
3687     "^netprospector" => ""
3688     )
3689     ),
3690     "netpumper" => array(
3691     "icon" => "robot",
3692     "title" => "Netpumper",
3693     "rule" => array(
3694     "^NetPumper[/ ]([0-9.]{1,10})" => "\\1"
3695     ),
3696     "uri" => ""
3697     ),
3698     "netscapeproxy" => array(
3699     "icon" => "netscape",
3700     "title" => "Netscape Proxy",
3701     "rule" => array(
3702     "Netscape-Proxy[/ ]([0-9.]{1,10})" => "\\1"
3703     ),
3704     "uri" => ""
3705     ),
3706     "netspective" => array(
3707     "icon" => "robot",
3708     "title" => "NetSpective",
3709     "rule" => array(
3710     "^WebFilter Robot ([0-9.]{1,10})" => "\\1"
3711     ),
3712     "uri" => ""
3713     ),
3714     "netvibes" => array(
3715     "icon" => "netvibes",
3716     "title" => " Netvibes",
3717     "rule" => array(
3718     "^Netvibes" => "\\1"
3719     ),
3720     "uri" => ""
3721     ),
3722     "newsfire" => array(
3723     "icon" => "newsfire",
3724     "title" => "NewsFire",
3725     "rule" => array(
3726     "NewsFire[/ ]([0-9.]{1,10})" => "\\1"
3727     ),
3728     "uri" => ""
3729     ),
3730     "newsgator" => array(
3731     "icon" => "newsgator",
3732     "title" => "NewsGator",
3733     "rule" => array(
3734     "NewsGato(r|rOnline)[/ ]([0-9.]{1,10})" => "\\2"
3735     ),
3736     "uri" => ""
3737     ),
3738     "newzcrawler" => array(
3739     "icon" => "newzcrawler",
3740     "title" => "NewzCrawler",
3741     "rule" => array(
3742     "NewzCrawler[/ ]([0-9.]{1,10})" => "\\1"
3743     ),
3744     "uri" => ""
3745     ),
3746     "nextopia" => array(
3747     "icon" => "newzcrawler",
3748     "title" => "NewzCrawler",
3749     "rule" => array(
3750     "^NextopiaBOT.*[v ]([0-9.]{1,10})" => "\\1"
3751     ),
3752     "uri" => ""
3753     ),
3754     "ngsearch" => array(
3755     "icon" => "ngsearch",
3756     "title" => "NG Search",
3757     "rule" => array(
3758     "NG-Search[/ ]([0-9.]{1,10})" => "\\1"
3759     ),
3760     "uri" => ""
3761     ),
3762     "nimble" => array(
3763     "icon" => "robot",
3764     "title" => "Nimble",
3765     "rule" => array(
3766     "NimbleCrawler[/ ]([0-9.]{1,10})" => "\\1"
3767     ),
3768     "uri" => ""
3769     ),
3770     "nmap" => array(
3771     "icon" => "nmap",
3772     "title" => "Nmap",
3773     "rule" => array(
3774     "Nmap Scripting Engine" => ""
3775     ),
3776     "uri" => "http://nmap.org/book/nse.html"
3777     ),
3778     "nusearch" => array(
3779     "icon" => "robot",
3780     "title" => "NuSearch",
3781     "rule" => array(
3782     "^nuSearch" => ""
3783     ),
3784     "uri" => "http://www.nusearch.com/"
3785     ),
3786     "noago" => array(
3787     "icon" => "noago",
3788     "title" => "Noago",
3789     "rule" => array(
3790     "Noago Spider" => ""
3791     ),
3792     "uri" => "http://www.noago.com/"
3793     ),
3794     "noviforum" => array(
3795     "icon" => "noviforum",
3796     "title" => "Noviforum",
3797     "rule" => array(
3798     "TridentSpider[/ ]?([0-9.]{1,10})" => "\\1"
3799     ),
3800     "uri" => ""
3801     ),
3802     "noxtrum" => array(
3803     "icon" => "noxtrum",
3804     "title" => "noXtrum",
3805     "rule" => array(
3806     "noxtrumbot[/ ]?([0-9.]{1,10})" => "\\1"
3807     ),
3808     "uri" => ""
3809     ),
3810     "noyona" => array(
3811     "icon" => "robot",
3812     "title" => "Noyona",
3813     "rule" => array(
3814     "noyona.([0-9._]{1,10})" => "\\1"
3815     ),
3816     "uri" => "http://noyona.com/"
3817     ),
3818     "nsauditor" => array(
3819     "icon" => "nsauditor",
3820     "title" => "Nsauditor",
3821     "rule" => array(
3822     "Nsauditor[ /]([0-9.]{1,10})" => "\\1"
3823     ),
3824     "uri" => "http://www.nsauditor.com/"
3825     ),
3826     "obidosbot" => array(
3827     "icon" => "robot",
3828     "title" => "Bookwatch",
3829     "rule" => array(
3830     "obidos[ \-]?bot" => ""
3831     ),
3832     "uri" => ""
3833     ),
3834     "objectssearch" => array(
3835     "icon" => "robot",
3836     "title" => "Objects Search",
3837     "rule" => array(
3838     "ObjectsSearch[ /]([0-9.]{1,10})" => "\\1"
3839     ),
3840     "uri" => ""
3841     ),
3842     "obot" => array(
3843 matthys 297 "icon" => "ibm",
3844     "title" => "IBM (ONLY) Crawler",
3845 joku 63 "rule" => array(
3846 matthys 297 "oBot[ /]([0-9.]{1,10})" => "\\1"
3847 joku 63 ),
3848 matthys 297 "uri" => "http://filterdb.iss.net/crawler/"
3849     // Previous from "ONLY Solutions GmbH" – that is what the O stands for
3850 joku 63 ),
3851     "ocawa" => array(
3852     "icon" => "ocawa",
3853     "title" => "Ocawa",
3854     "rule" => array(
3855     "Ocawa[ /]([A-Z.a-z][0-9.]{1,10})" => "\\1",
3856     "Ocawa" => ""
3857     ),
3858     "uri" => "http://www.ocawa.com/"
3859     ),
3860     "octora" => array(
3861     "icon" => "octora",
3862     "title" => "Octora",
3863     "rule" => array(
3864     "^Octora (Beta)?" => ""
3865     ),
3866     "uri" => "http://www.octora.com/"
3867     ),
3868     "offlineexplorer" => array(
3869     "icon" => "robot",
3870     "title" => "OfflineExplorer",
3871     "rule" => array(
3872     "^Offline Explorer[ /]([0-9.]{1,10})" => "\\1"
3873     ),
3874     "uri" => ""
3875     ),
3876     "omea" => array(
3877     "icon" => "omea",
3878     "title" => "Omea Reader",
3879     "rule" => array(
3880     "Omea Reader[ /]([0-9.]{1,10})" => "\\1"
3881     ),
3882     "uri" => ""
3883     ),
3884     "onet" => array(
3885     "icon" => "onet",
3886     "title" => "Onet",
3887     "rule" => array(
3888     "OnetSzukaj[ /]([0-9.]{1,10})" => "\\1",
3889     "Onet\.pl" => "",
3890     "inktomi.search.onet" => ""
3891     ),
3892     "uri" => "http://www.onet.pl"
3893     ),
3894     "online24" => array(
3895     "icon" => "robot",
3896 matthys 209 "title" => "Online24",
3897 joku 63 "rule" => array(
3898     "^Online24-Bot .* ([0-9.]{1,10})" => "\\1"
3899     ),
3900     "uri" => "http://www.online24.de"
3901     ),
3902     "onlinewebcheck" => array(
3903     "icon" => "csehtmlvalidator",
3904     "title" => "CSE HTML Validator",
3905     "rule" => array(
3906     "onlinewebcheck" => ""
3907     ),
3908     "uri" => "http://www.onlinewebcheck.com"
3909     ),
3910     "onsearch" => array(
3911     "icon" => "onsearch",
3912 matthys 209 "title" => "OnSearch",
3913 joku 63 "rule" => array(
3914     "^onCHECK-Robot" => ""
3915     ),
3916     "uri" => "http://www.onsearch.de"
3917     ),
3918     "onto" => array(
3919     "icon" => "robot",
3920     "title" => "OntoSpider",
3921     "rule" => array(
3922     "^OntoSpider[ /]([0-9.]{1,10})" => "\\1"
3923     ),
3924     "uri" => "http://ontospider.i-n.info/"
3925     ),
3926     "openacoon" => array(
3927     "icon" => "robot",
3928     "title" => "OpenAcoon",
3929     "rule" => array(
3930     "^OpenAcoon v([0-9.]{1,10})" => "\\1"
3931     ),
3932     "uri" => "http://www.openacoon.de/"
3933     ),
3934     "openfind" => array(
3935     "icon" => "openfind",
3936     "title" => "Openfind",
3937     "rule" => array(
3938     "openbot[ /]([0-9.]{1,10})" => "\\1",
3939     "Openfind Robot[ /]([0-9.A-Z]{1,10})" => "\\1"
3940     ),
3941     "uri" => ""
3942     ),
3943     "opentagger" => array(
3944     "icon" => "robot",
3945     "title" => "OpenTagger",
3946     "rule" => array(
3947     "^OpenTaggerBot" => ""
3948     ),
3949     "uri" => "http://www.opentagger.com/opentaggerbot.htm"
3950     ),
3951     "opentext" => array(
3952     "icon" => "opentext",
3953     "title" => "OpenText",
3954     "rule" => array(
3955     "^OpenTextSiteCrawler[ /]([0-9.]{1,10})" => "\\1"
3956     ),
3957     "uri" => "http://www.opentext.net/"
3958     ),
3959     "openweb" => array(
3960     "icon" => "robot",
3961     "title" => "OpenWebSpider",
3962     "rule" => array(
3963     "^OpenWebSpider[ /]([0-9.]{1,10})" => "\\1",
3964     "^OpenWebSpider v([0-9.A-Z]{1,10})" => "\\1"
3965     ),
3966     "uri" => "http://www.openwebspider.org"
3967     ),
3968     "openx" => array(
3969     "icon" => "openx",
3970     "title" => "OpenX Spider",
3971     "rule" => array(
3972     "OpenX" => ""
3973     ),
3974     "uri" => "http://www.openx.org"
3975     ),
3976     "orange" => array(
3977     "icon" => "orange",
3978     "title" => "Orange",
3979     "rule" => array(
3980     "OrangeBot-Mobile[ /]([0-9.]{1,10})" => "\\1"
3981     ),
3982     "uri" => "http://orange.com/"
3983     ),
3984     "organica" => array(
3985     "icon" => "robot",
3986     "title" => "Organica",
3987     "rule" => array(
3988     "crawler@organica\.us" => ""
3989     ),
3990     "uri" => ""
3991     ),
3992     "outfox" => array(
3993     "icon" => "robot",
3994     "title" => "Outfox Melon",
3995     "rule" => array(
3996     "OutfoxMelonBot[ /]([0-9.]{1,10})" => "\\1",
3997     "OutfoxBot[ /]([0-9.]{1,10})" => "\\1"
3998     ),
3999     "uri" => ""
4000     ),
4001     "overture" => array(
4002     "icon" => "overture",
4003     "title" => "Overture",
4004     "rule" => array(
4005     "Overture[ \-]?WebCrawler" => ""
4006     ),
4007     "uri" => ""
4008     ),
4009     "page2rss" => array(
4010     "icon" => "page2rss",
4011     "title" => "Page2RSS",
4012     "rule" => array(
4013     "Page2RSS[ /]([0-9.]{1,10})" => "\\1"
4014     ),
4015     "uri" => "http://page2rss.com"
4016     ),
4017     "pagebytes" => array(
4018     "icon" => "robot",
4019     "title" => "PageBites",
4020     "rule" => array(
4021     "^PageBitesHyperBot[ /]([0-9.]{1,10})" => "\\1"
4022     ),
4023     "uri" => ""
4024     ),
4025 matthys 282 "pagesinventory" => array(
4026     "icon" => "pagesinventory",
4027     "title" => "PagesInventory",
4028     "rule" => array(
4029     "^PagesInventory" => ""
4030     ),
4031     "uri" => "http://www.pagesinventory.com"
4032     ),
4033 joku 63 "pagesjaunes" => array(
4034     "icon" => "pagesjaunes",
4035     "title" => "Pages Jaunes",
4036     "rule" => array(
4037     "PJBot[ /]([0-9.]{1,10})" => "\\1"
4038     ),
4039     "uri" => "http://www.pagesjaunes.fr/"
4040     ),
4041     "pagevalet" => array(
4042     "icon" => "wdg",
4043     "title" => "WDG Page Valet",
4044     "rule" => array(
4045     "^Page Valet[ /]([0-9.]{1,10})" => "\\1"
4046     ),
4047     "uri" => "http://www.htmlhelp.com/tools/validator/"
4048     ),
4049     "panopeabot" => array(
4050     "icon" => "robot",
4051     "title" => "PanopeaBot",
4052     "rule" => array(
4053     "PanopeaBot[/ ]([0-9.]{1,10})" => "\\1"
4054     ),
4055     "uri" => ""
4056     ),
4057     "parchmenthill" => array(
4058     "icon" => "robot",
4059     "title" => "Parchment Hill",
4060     "rule" => array(
4061     "ParchBot[ /]([0-9.]{1,10})" => "\\1"
4062     ),
4063     "uri" => "http://www.parchmenthill.com/search.htm"
4064     ),
4065     "pascient" => array(
4066     "icon" => "pascient",
4067     "title" => "Pascient",
4068     "rule" => array(
4069     "panscient.com" => ""
4070     ),
4071     "uri" => "http://www.panscient.com/"
4072     ),
4073     "peerbot" => array(
4074     "icon" => "peerbot",
4075     "title" => "Peerbot",
4076     "rule" => array(
4077     "^PEERbot" => ""
4078     ),
4079     "uri" => ""
4080     ),
4081     "php" => array(
4082     "icon" => "php",
4083     "title" => "PHP",
4084     "rule" => array(
4085     "^PHP[ /]([0-9.]{1,10})" => "\\1"
4086     ),
4087     "uri" => ""
4088     ),
4089     "phpcrawl" => array(
4090     "icon" => "robot",
4091     "title" => "PHPCrawl",
4092     "rule" => array(
4093     "^PHPCrawl" => ""
4094     ),
4095     "uri" => "http://phpcrawl.cuab.de/"
4096     ),
4097     "phpdig" => array(
4098     "icon" => "robot",
4099     "title" => "PhpDig",
4100     "rule" => array(
4101     "^PhpDig[ /]([0-9.]{1,10})" => "\\1"
4102     ),
4103     "uri" => "http://www.finbot.com/"
4104     ),
4105     "phpversiontracker" => array(
4106     "icon" => "phpversiontracker",
4107     "title" => "PHP version tracker",
4108     "rule" => array(
4109     "^PHP version tracker" => ""
4110     ),
4111     "uri" => "http://www.nexen.net/phpversion/bot.php"
4112     ),
4113     "pictureofinternet" => array(
4114     "icon" => "robot",
4115     "title" => "PictureOfInternet",
4116     "rule" => array(
4117     "^PictureOfInternet[ /]([0-9.]{1,10})" => ""
4118     ),
4119     "uri" => ""
4120     ),
4121     "pingdom" => array(
4122     "icon" => "pingdom",
4123     "title" => "Pingdom",
4124     "rule" => array(
4125     "^Pingdom GIGRIB v([0-9.]{1,10})" => "\\1",
4126     "^Pingdom" => ""
4127     ),
4128     "uri" => "http://www.pingdom.com/"
4129     ),
4130     "pinseri" => array(
4131     "icon" => "pinseri",
4132     "title" => "Pinseri",
4133     "rule" => array(
4134     "www\.pinseri\.com/bloglist" => ""
4135     ),
4136     "uri" => ""
4137     ),
4138     "plagger" => array(
4139     "icon" => "plagger",
4140     "title" => "Plagger",
4141     "rule" => array(
4142     "Plagger[ /]([0-9.]{1,10})" => "\\1"
4143     ),
4144     "uri" => "http://www.plugger.org"
4145     ),
4146     "planet" => array(
4147     "icon" => "planet",
4148     "title" => "Planet",
4149     "rule" => array(
4150     "Planet[ /]([0-9.]{1,10})" => "\\1"
4151     ),
4152     "uri" => ""
4153     ),
4154     "plantynet" => array(
4155     "icon" => "robot",
4156     "title" => "PlantyNet",
4157     "rule" => array(
4158     "PlantyNet_WebRobot[_ /]V?([0-9.]{1,10})" => "\\1"
4159     ),
4160     "uri" => ""
4161     ),
4162     "pluck" => array(
4163     "icon" => "pluck",
4164     "title" => "Pluck",
4165     "rule" => array(
4166     "PluckFeedCrawler[ /]([0-9.]{1,10})" => "\\1"
4167     ),
4168     "uri" => ""
4169     ),
4170     "plsearch" => array(
4171     "icon" => "plsearch",
4172     "title" => "PlanetSearch",
4173     "rule" => array(
4174     "fido[ /]([0-9.]{1,10}) Harvest" => "\\1"
4175     ),
4176     "uri" => ""
4177     ),
4178     "poe" => array(
4179     "icon" => "robot",
4180     "title" => "POE-Component",
4181     "rule" => array(
4182     "^POE-Component-Client-HTTP[/ ]([0-9.]{1,10})" => "\\1"
4183     ),
4184     "uri" => ""
4185     ),
4186     "pogodak" => array(
4187     "icon" => "pogodak",
4188     "title" => "Pogodak",
4189     "rule" => array(
4190     "Pogodak\.hr[/ ]?([0-9.]{1,10})" => "\\1"
4191     ),
4192     "uri" => ""
4193     ),
4194     "poodle" => array(
4195     "icon" => "robot",
4196     "title" => "Poodle predictor",
4197     "rule" => array(
4198     "P(oo|ooo)dle[ \-]?predictor[ \-]?([0-9.]{1,10})" => "\\1",
4199     "P(oo|ooo)dle[ \-]?predictor" => ""
4200     ),
4201     "uri" => "http://www.gritechnologies.com/tools/spider.go"
4202     ),
4203     "pompos" => array(
4204     "icon" => "pompos",
4205     "title" => "Pompos",
4206     "rule" => array(
4207     "Pompos[ /]([0-9.]{1,10})" => "\\1"
4208     ),
4209     "uri" => ""
4210     ),
4211     "popdex" => array(
4212     "icon" => "robot",
4213     "title" => "Popdexter",
4214     "rule" => array(
4215     "Popdexter" => ""
4216     ),
4217     "uri" => ""
4218     ),
4219     "postrank" => array(
4220     "icon" => "postrank",
4221     "title" => "PostRank",
4222     "rule" => array(
4223     "^PostRank[ /]+([0-9.]{1,10})" => "\\1"
4224     ),
4225     "uri" => "http://www.postrank.com/"
4226     ),
4227     "powermarks" => array(
4228     "icon" => "robot",
4229     "title" => "Powermarks",
4230     "rule" => array(
4231     "Powermarks[ /]([0-9.]{1,10})" => "\\1"
4232     ),
4233     "uri" => ""
4234     ),
4235     "probe" => array(
4236     "icon" => "robot",
4237     "title" => "PROBE!",
4238     "rule" => array(
4239     "^PROBE!" => ""
4240     ),
4241     "uri" => ""
4242     ),
4243     "projecthoneypot" => array(
4244     "icon" => "projecthoneypot",
4245     "title" => "Project Honeypot",
4246     "rule" => array(
4247     "^PHPot Verispider[ /v]+([0-9.]{1,10})" => "\\1",
4248     "projecthoneypot" => ""
4249     ),
4250     "uri" => "http://www.projecthoneypot.org"
4251     ),
4252     "proxycache" => array(
4253     "icon" => "robot",
4254     "title" => "Proxy Cache",
4255     "rule" => array(
4256     "^Mozilla/[0-9.]{1,10} \(compatible\;\)$" => ""
4257     ),
4258     "uri" => ""
4259     ),
4260     "proxyhunter" => array(
4261     "icon" => "robot",
4262     "title" => "ProxyHunter",
4263     "rule" => array(
4264     "ProxyHunter" => ""
4265     ),
4266     "uri" => ""
4267     ),
4268     "psbot" => array(
4269     "icon" => "picsearch",
4270     "title" => "PicSearch",
4271     "rule" => array(
4272     "^psbot" => ""
4273     ),
4274     "uri" => ""
4275     ),
4276     "pubsub" => array(
4277     "icon" => "pubsub",
4278     "title" => "PubSub",
4279     "rule" => array(
4280     "^PubSub-RSS-Reader[ /]([0-9.]{1,10})" => "\\1",
4281     "^PubSub\.com" => ""
4282     ),
4283     "uri" => ""
4284     ),
4285     "pukiwiki" => array(
4286     "icon" => "pukiwiki",
4287     "title" => "PukiWiki",
4288     "rule" => array(
4289     "PukiWiki[ /]([0-9.]{1,10})" => "\\1"
4290     ),
4291     "uri" => ""
4292     ),
4293     "pwebotxy" => array(
4294     "icon" => "pwebotxy",
4295     "title" => "PWeBot/X.Y",
4296     "rule" => array(
4297     "^PWeBot[ /]([0-9.]{1,10})" => "\\1"
4298     ),
4299     "uri" => "http://www.programacionweb.net/robot.php"
4300     ),
4301     "pxys" => array(
4302     "icon" => "robot",
4303     "title" => "PXYS",
4304     "rule" => array(
4305     "^pxys" => ""
4306     ),
4307     "uri" => ""
4308     ),
4309     "qango" => array(
4310     "icon" => "qango",
4311     "title" => "Qango",
4312     "rule" => array(
4313     "^Qango.com" => ""
4314     ),
4315     "uri" => "http://www.quango.com/"
4316     ),
4317     "qihoo" => array(
4318     "icon" => "qihoo",
4319     "title" => "Qihoo",
4320     "rule" => array(
4321     "QihooBot[ /]([0-9.]{1,10})" => "\\1"
4322     ),
4323     "uri" => "http://www.qihoo.com/"
4324     ),
4325     "qseero" => array(
4326     "icon" => "robot",
4327     "title" => "Qseero",
4328     "rule" => array(
4329     "Qseero v([0-9.]{1,10})" => "\\1"
4330     ),
4331     "uri" => "http://www.q0.com/"
4332     ),
4333     "quantcast" => array(
4334     "icon" => "robot",
4335     "title" => "Quantcast",
4336     "rule" => array(
4337     "Quantcastbot[ /]([0-9.]{1,10})" => "\\1"
4338     ),
4339     "uri" => "http://www.quantcast.com/"
4340     ),
4341     "quepasa" => array(
4342     "icon" => "quepasa",
4343     "title" => "Quepasa",
4344     "rule" => array(
4345     "Quepasa[ \-]?Creep" => ""
4346     ),
4347     "uri" => ""
4348     ),
4349     "questfinder" => array(
4350     "icon" => "robot",
4351     "title" => "QuestFinder",
4352     "rule" => array(
4353     "www\.questfinder\.com" => ""
4354     ),
4355     "uri" => ""
4356     ),
4357     "qweery" => array(
4358     "icon" => "robot",
4359     "title" => "Qweery",
4360     "rule" => array(
4361     "^QweeryBot[ /]([0-9.]{1,10})" => "\\1"
4362     ),
4363     "uri" => "http://qweerybot.qweery.nl"
4364     ),
4365     "racaicrawler" => array(
4366     "icon" => "robot",
4367     "title" => "RacaiCrawler",
4368     "rule" => array(
4369     "RacaiCrawler[ -]([0-9.]{1,10})" => "\\1"
4370     )
4371     ),
4372     "radian6" => array(
4373     "icon" => "robot",
4374     "title" => "Radian6",
4375     "rule" => array(
4376     "www\.radian6\.com" => ""
4377     ),
4378     "uri" => "http://www.radian6.com"
4379     ),
4380     "rambler" => array(
4381     "icon" => "rambler",
4382     "title" => "Rambler",
4383     "rule" => array(
4384     "StackRambler[ /]([0-9.]{1,10})" => "\\1"
4385     ),
4386     "uri" => ""
4387     ),
4388     "ramiba" => array(
4389     "icon" => "robot",
4390 matthys 209 "title" => "Ramiba",
4391 joku 63 "rule" => array(
4392     "^ramiba(-bot)?" => "\\1"
4393     ),
4394     "uri" => ""
4395     ),
4396     "rankur" => array(
4397     "icon" => "rankur",
4398     "title" => "Rankur",
4399     "rule" => array(
4400     "^RankurBot/Rankur([0-9.]{1,10})" => "\\1"
4401     ),
4402     "uri" => "http://rankur.com/",
4403     ),
4404 joku 261 "red" => array(
4405     "icon" => "red",
4406     "title" => "RED",
4407     "rule" => array(
4408     "RED[ /]([0-9.]{1,10})" => "\\1"
4409     ),
4410     "uri" => "http://redbot.org/"
4411 joku 181 ),
4412 joku 63 "rediff" => array(
4413     "icon" => "rediff",
4414 matthys 209 "title" => "Rediff",
4415 joku 63 "rule" => array(
4416     "^RedBot/redbot[ /\-]([0-9.]{1,10})" => "\\1"
4417     ),
4418     "uri" => "http://www.rediff.com",
4419     ),
4420     "repia" => array(
4421     "icon" => "robot",
4422     "title" => "Repia",
4423     "rule" => array(
4424     "webmaster@repia\.com" => ""
4425     ),
4426     "uri" => ""
4427     ),
4428     "robotgenius" => array(
4429     "icon" => "robot",
4430     "title" => "Robot Genius",
4431     "rule" => array(
4432     "^robotgenius" => ""
4433     ),
4434     "uri" => "http://robotgenius.net",
4435     ),
4436     "robozilla" => array(
4437     "icon" => "robot",
4438     "title" => "Robozilla",
4439     "rule" => array(
4440     "Robozilla" => ""
4441     ),
4442     "uri" => ""
4443     ),
4444 matthys 302 "rogerbot" => array(
4445 matthys 306 "icon" => "moz",
4446     "title" => "Moz Rogerbot",
4447 matthys 302 "rule" => array(
4448     "rogerbot[ /]([0-9.]{1,10})" => "\\1"
4449     ),
4450     "uri" => "http://moz.com/help/guides/moz-procedures/who-is-rogerbot"
4451     ),
4452 joku 63 "rojo" => array(
4453     "icon" => "rojo",
4454     "title" => "Rojo",
4455     "rule" => array(
4456     "Rojo[ /]([0-9.]{1,10})" => "\\1"
4457     ),
4458     "uri" => ""
4459     ),
4460     "rss-atom" => array(
4461     "icon" => "rss",
4462     "title" => "RSS / Atom",
4463     "rule" => array(
4464     "^Apple-PubSub" => "",
4465     "^AppleSyndication" => ""
4466     ),
4467     "uri" => ""
4468     ),
4469     "rssbot" => array(
4470     "icon" => "rss",
4471 matthys 209 "title" => "RSS-bot",
4472 joku 63 "rule" => array(
4473     "rss-bot[ /]([0-9.]{1,10})" => "\\1"
4474     ),
4475     "uri" => ""
4476     ),
4477     "rssbandit" => array(
4478     "icon" => "rssbandit",
4479     "title" => "RssBandit",
4480     "rule" => array(
4481     "RssBandit[ /]([0-9.]{1,10})" => "\\1"
4482     ),
4483     "uri" => ""
4484     ),
4485     "rssimages" => array(
4486     "icon" => "rss",
4487 matthys 209 "title" => "RssImages",
4488 joku 63 "rule" => array(
4489     "rssImagesBot[ /]([0-9.]{1,10})" => "\\1"
4490     ),
4491     "uri" => ""
4492     ),
4493     "rssmicro" => array(
4494     "icon" => "rssmicro",
4495     "title" => "RSSMicro",
4496     "rule" => array(
4497     "RSSMicro\.com" => ""
4498     ),
4499     "uri" => "http://www.rssmicro.com"
4500     ),
4501     "rssowl" => array(
4502     "icon" => "rssowl",
4503     "title" => "RSSOwl",
4504     "rule" => array(
4505     "RSSOwl[ /]([0-9a-z.]{1,10})" => "\\1"
4506     ),
4507     "uri" => ""
4508     ),
4509     "rssreader" => array(
4510     "icon" => "rss",
4511     "title" => "RssReader",
4512     "rule" => array(
4513     "RssReader[ /]([0-9.]{1,10})" => ""
4514     ),
4515     "uri" => ""
4516     ),
4517     "rtgi" => array(
4518     "icon" => "rtgi",
4519 matthys 209 "title" => "Rtgi",
4520 joku 63 "rule" => array(
4521     "RTGI" => ""
4522     ),
4523     "uri" => "http://rtgi.fr/"
4524     ),
4525     "rufusbot" => array(
4526     "icon" => "robot",
4527     "title" => "RufusBot",
4528     "rule" => array(
4529     "RufusBot" => ""
4530     ),
4531     "uri" => ""
4532     ),
4533 matthys 304 "runet" => array(
4534     "icon" => "runet",
4535     "title" => "iTrack RuNet Crawler",
4536     "rule" => array(
4537     "Runet-Research-Crawler" => ""
4538     ),
4539     "uri" => "http://www.itrack.ru/research/cmsrate/"
4540     ),
4541 joku 63 "runnk" => array(
4542     "icon" => "robot",
4543     "title" => "Runnk",
4544     "rule" => array(
4545     "Runnk RSS finder" => ""
4546     ),
4547     "uri" => "http://www.runnk.com/il/law"
4548     ),
4549     "sagool" => array(
4550     "icon" => "robot",
4551     "title" => "Sagool",
4552     "rule" => array(
4553     "MaSagool" => ""
4554     ),
4555     "uri" => "http://sagool.jp/"
4556     ),
4557     "sanszbot" => array(
4558     "icon" => "robot",
4559     "title" => "Sansz",
4560     "rule" => array(
4561     "SanszBot" => ""
4562     ),
4563     "uri" => ""
4564     ),
4565     "saucereader" => array(
4566     "icon" => "saucereader",
4567     "title" => "Sauce Reader",
4568     "rule" => array(
4569     "Sauce[ ]?Reader[ /]([0-9.]{1,10})" => "\\1"
4570     ),
4571     "uri" => ""
4572     ),
4573     "sbider" => array(
4574     "icon" => "sbider",
4575     "title" => "SBIder",
4576     "rule" => array(
4577     "SBIder[/ ]([0-9.]{1,10})" => "\\1",
4578     "SBIder[/ ]SBIder.([0-9.]{1,10})" => "\\1"
4579     ),
4580     "uri" => ""
4581     ),
4582     "scirus" => array(
4583     "icon" => "robot",
4584     "title" => "Scirus",
4585     "rule" => array(
4586     "FAST-WebCrawler/[0-9a-z.]{1,10}/Scirus" => ""
4587     ),
4588     "uri" => ""
4589     ),
4590     "scoutjet" => array(
4591     "icon" => "scoutjet",
4592     "title" => "ScoutJet",
4593     "rule" => array(
4594     "ScoutJet" => ""
4595     ),
4596     "uri" => "http://www.scoutjet.com/"
4597     ),
4598 joku 261 "scrapy" => array(
4599     "icon" => "scrapy",
4600     "title" => "Scrapy",
4601     "rule" => array(
4602     "Scrapy[ /]([0-9.]{1,10})" => "\\1"
4603     ),
4604     "uri" => "http://scrapy.org"
4605     ),
4606 joku 63 "scrubby" => array(
4607     "icon" => "scrubby",
4608     "title" => "Scrubby",
4609     "rule" => array(
4610     "Scrubby[ /]([0-9.]{1,10})" => "\\1"
4611     ),
4612     "uri" => ""
4613     ),
4614     "sdm" => array(
4615     "icon" => "sdm",
4616     "title" => "SUN Download Manager",
4617     "rule" => array(
4618     "Sun Download Manager[/ ]([0-9.]{1,10})" => "\\1"
4619     ),
4620     "uri" => ""
4621     ),
4622     "sealinks" => array(
4623     "icon" => "robot",
4624     "title" => "Sea Links",
4625     "rule" => array(
4626     "SEA-Links( HTML-Scanner Pingoo\!)?[ /]([0-9.]{1,10})" => "\\2"
4627     ),
4628     "uri" => ""
4629     ),
4630     "search17" => array(
4631     "icon" => "robot",
4632     "title" => "Search 17",
4633     "rule" => array(
4634     "Search17Bot[ /]([0-9.]{1,10})" => "\\1"
4635     ),
4636     "uri" => "http://www.search17.com/bot.php"
4637     ),
4638     "search2" => array(
4639     "icon" => "search2",
4640     "title" => "Search2.net",
4641     "rule" => array(
4642     "^S2Bot[ /]([0-9.]{1,10})" => "\\1"
4643     ),
4644     "uri" => "http://search2.net/"
4645     ),
4646     "searchbot" => array(
4647     "icon" => "robot",
4648     "title" => "Searchbot",
4649     "rule" => array(
4650     "Searchbot[ /]([0-9.]{1,10})" => "\\1"
4651     ),
4652     "uri" => ""
4653     ),
4654     "searchch" => array(
4655     "icon" => "robot",
4656     "title" => "Search.ch",
4657     "rule" => array(
4658     "search\.ch[ /]?V?([0-9.]{1,10})" => "\\1"
4659     ),
4660     "uri" => ""
4661     ),
4662     "searchengineworld" => array(
4663     "icon" => "robot",
4664     "title" => "SearchEngineWorld",
4665     "rule" => array(
4666     "searchengineworld" => ""
4667     ),
4668     "uri" => "http://www.searchengineworld.com/"
4669     ),
4670     "searchhippo" => array(
4671     "icon" => "searchhippo",
4672     "title" => "Searchhippo",
4673     "rule" => array(
4674     "searchhippo" => ""
4675     ),
4676     "uri" => "http://www.searchhippo.com/"
4677     ),
4678     "searchthruus" => array(
4679     "icon" => "robot",
4680     "title" => "SearchThruUs",
4681     "rule" => array(
4682     "www\.unitek-systems\.co\.uk[ /]([0-9.]{1,10})" => "\\1"
4683     ),
4684     "uri" => ""
4685     ),
4686     "securecomputing" => array(
4687     "icon" => "robot",
4688     "title" => "Secure Computing",
4689     "rule" => array(
4690     "securecomputing" => ""
4691     ),
4692     "uri" => ""
4693     ),
4694     "seekport" => array(
4695     "icon" => "seekport",
4696     "title" => "Seekport",
4697     "rule" => array(
4698     "Seekbot[ /]([0-9.]{1,10})" => "\\1"
4699     ),
4700     "uri" => ""
4701     ),
4702     "semanticdiscovery" => array(
4703     "icon" => "robot",
4704     "title" => "Semantic Discovery",
4705     "rule" => array(
4706     "semanticdiscovery[ /]([0-9.]{1,10})" => "\\1"
4707     ),
4708     "uri" => ""
4709     ),
4710 matthys 319 "semrush" => array(
4711     "icon" => "semrush",
4712     "title" => "SEMrush",
4713     "rule" => array(
4714     "SemrushBot[ /]([0-9.]{1,10})" => "\\1"
4715     ),
4716     "uri" => "http://www.semrush.com/bot.html"
4717     ),
4718 joku 63 "sengine" => array(
4719     "icon" => "sengine",
4720     "title" => "Sengine",
4721     "rule" => array(
4722     "netEstate NE Crawler" => ""
4723     ),
4724     "uri" => "http://www.sengine.info/"
4725     ),
4726     "sensis" => array(
4727     "icon" => "sensis",
4728     "title" => "Sensis",
4729     "rule" => array(
4730     "^Sensis(.com.au)? Web Crawler" => ""
4731     ),
4732     "uri" => "http://sensis.com.au"
4733     ),
4734     "seokicks" => array(
4735     "icon" => "seokicks",
4736     "title" => "SEOkicks",
4737     "rule" => array(
4738     "SEOkicks-Robot" => ""
4739     ),
4740     "uri" => "http://www.seokicks.de/"
4741     ),
4742     "seoprofiler" => array(
4743     "icon" => "seoprofiler",
4744     "title" => "SEOprofiler",
4745     "rule" => array(
4746     "spbot[ /]([0-9.]{1,10})" => "\\1"
4747     ),
4748     "uri" => "http://www.seoprofiler.com/"
4749     ),
4750     "setooz" => array(
4751     "icon" => "setooz",
4752     "title" => "Setooz",
4753     "rule" => array(
4754     "OOZBOT[ /]([0-9.]{1,10})" => "\\1"
4755     ),
4756     "uri" => "http://www.setooz.com/oozbot.html"
4757     ),
4758     "seznam" => array(
4759     "icon" => "seznam",
4760     "title" => "Seznam",
4761     "rule" => array(
4762     "SeznamBot[ /]([0-9.]{1,10})" => "\\1",
4763     "Seznam" => ""
4764     ),
4765     "uri" => "http://www.seznam.cz"
4766     ),
4767     "sharpreader" => array(
4768     "icon" => "sharpreader",
4769     "title" => "SharpReader",
4770     "rule" => array(
4771     "SharpReader[ /]([0-9.]{1,10})" => "\\1"
4772     ),
4773     "uri" => ""
4774     ),
4775     "shelob" => array(
4776     "icon" => "shelob",
4777     "title" => "Sherlock Spider",
4778     "rule" => array(
4779     "^Shelob" => ""
4780     ),
4781     "uri" => "http://www.gmx.net"
4782     ),
4783     "sherlockspider" => array(
4784     "icon" => "robot",
4785     "title" => "Sherlock Spider",
4786     "rule" => array(
4787     "sherlock_spider" => ""
4788     ),
4789     "uri" => ""
4790     ),
4791     "shim" => array(
4792     "icon" => "robot",
4793     "title" => "Shim Crawler",
4794     "rule" => array(
4795     "shim[ \-]crawler" => ""
4796     ),
4797     "uri" => ""
4798     ),
4799     "shopwiki" => array(
4800     "icon" => "shopwiki",
4801     "title" => "ShopWiki",
4802     "rule" => array(
4803     "^ShopWiki[ /]([0-9.]{1,10})" => "\\1"
4804     ),
4805     "uri" => ""
4806     ),
4807     "shoula" => array(
4808     "icon" => "robot",
4809     "title" => "Shoula",
4810     "rule" => array(
4811     "^Shoula.com Crawler ([0-9.]{1,10})" => "\\1"
4812     ),
4813     "uri" => "http://www.shoula.com"
4814     ),
4815     "siege" => array(
4816     "icon" => "robot",
4817     "title" => "Siege",
4818     "rule" => array(
4819     "Siege[ /]([0-9.]{1,10})" => "\\1"
4820     ),
4821     "uri" => ""
4822     ),
4823     "siets" => array(
4824     "icon" => "robot",
4825     "title" => "Siets",
4826     "rule" => array(
4827     "SietsCrawler[ /]([0-9.]{1,10})" => "\\1"
4828     ),
4829     "uri" => ""
4830     ),
4831     "simpy" => array(
4832     "icon" => "simpy",
4833     "title" => "Simpy",
4834     "rule" => array(
4835     "^(argus|simpy)[ /]([0-9.]{1,10})" => "\\2",
4836     ),
4837     "uri" => ""
4838     ),
4839     "singingfish" => array(
4840     "icon" => "singingfish",
4841     "title" => "SingingFish",
4842     "rule" => array(
4843     "asterias[ /]([0-9.]{1,10})" => "\\1",
4844     "Asterias Crawler v([0-9.]{1,10})" => "\\1",
4845     "asterias" => ""
4846     ),
4847     "uri" => ""
4848     ),
4849     "sirketce" => array(
4850     "icon" => "robot",
4851     "title" => "Sirketce",
4852     "rule" => array(
4853     "Sirketcebot[ /v]+([0-9.]{1,10})" => "\\1"
4854     ),
4855     "uri" => "http://www.sirketce.com/bot.html"
4856     ),
4857     "sirobot" => array(
4858     "icon" => "robot",
4859     "title" => "SiroBot",
4860     "rule" => array(
4861     "sirobot" => ""
4862     ),
4863     "uri" => ""
4864     ),
4865     "sistrix" => array(
4866     "icon" => "sistrix",
4867     "title" => "SISTRIX",
4868     "rule" => array(
4869     "SISTRIX Crawler" => ""
4870     ),
4871     "uri" => "http://crawler.sistrix.net/"
4872     ),
4873     "sitebar" => array(
4874     "icon" => "sitebar",
4875     "title" => "SiteBar",
4876     "rule" => array(
4877     "SiteBar[ /]([0-9.]{1,10})" => "\\1"
4878     ),
4879     "uri" => ""
4880     ),
4881     "sitebot" => array(
4882     "icon" => "sitebot",
4883     "title" => "SiteBot",
4884     "rule" => array(
4885     "SiteBot[ /]([0-9.]{1,10})" => "\\1"
4886     ),
4887     "uri" => "http://www.sitebot.org/robot/"
4888     ),
4889 matthys 273 "siteexplorer" => array(
4890     "icon" => "siteexplorer",
4891     "title" => "SiteExplorer",
4892     "rule" => array(
4893     "SiteExplorer[ /]([0-9a-z.]{1,10})" => "\\1"
4894     ),
4895     "uri" => "http://siteexplorer.info/"
4896     ),
4897 joku 63 "sitesell" => array(
4898     "icon" => "sitesell",
4899     "title" => "SiteSell",
4900     "rule" => array(
4901     "SBIder[/ ]([0-9a-z.\-]{1,10})" => "\\1"
4902     ),
4903     "uri" => ""
4904     ),
4905     "sitespider" => array(
4906     "icon" => "robot",
4907     "title" => "SiteSpider",
4908     "rule" => array(
4909     "^SiteSpider" => ""
4910     ),
4911     "uri" => ""
4912     ),
4913     "sitesucker" => array(
4914     "icon" => "sitesucker",
4915     "title" => "SiteSucker",
4916     "rule" => array(
4917     "Sitesucker[ /]([0-9.]{1,10})" => "\\1"
4918     ),
4919     "uri" => "http://www.sitesucker.us/"
4920     ),
4921     "sitidi" => array(
4922     "icon" => "robot",
4923     "title" => "SitiDi",
4924     "rule" => array(
4925     "SitiDiBot[ /]([0-9.]{1,10})" => "\\1"
4926     ),
4927     "uri" => ""
4928     ),
4929     "skaffe" => array(
4930     "icon" => "robot",
4931     "title" => "Skaffe",
4932     "rule" => array(
4933     "Skampy[ /]([0-9.\-]{1,10})" => "\\1"
4934     ),
4935     "uri" => "http://www.skaffe.com"
4936     ),
4937     "skizzle" => array(
4938     "icon" => "skizzle",
4939     "title" => "Skizzle",
4940     "rule" => array(
4941     "SKIZZLE! Distributed Internet Spider[ /v]+([0-9a-z.\-]{1,10})" => "\\1"
4942     ),
4943     "uri" => "http://www.skizzle.com"
4944     ),
4945     "slider" => array(
4946     "icon" => "robot",
4947     "title" => "Slider",
4948     "rule" => array(
4949     "^Slider[ /]([0-9.]{1,10})" => "\\1"
4950     ),
4951     "uri" => ""
4952     ),
4953     "slugch" => array(
4954     "icon" => "robot",
4955 matthys 209 "title" => "Slugch",
4956 joku 63 "rule" => array(
4957     "^slug\.ch crawl ([0-9a-z.\-]{1,10})" => "\\1"
4958     ),
4959     "uri" => ""
4960     ),
4961     "smartware" => array(
4962     "icon" => "robot",
4963     "title" => "SmartWareSoft",
4964     "rule" => array(
4965     "^SWSBot-Images[ /]([0-9.]{1,10})" => "\\1"
4966     ),
4967     "uri" => "http://www.smartwaresoft.com/swsbot12.html"
4968     ),
4969     "snookit" => array(
4970     "icon" => "snookit",
4971     "title" => "Snookit",
4972     "rule" => array(
4973     "^snookit" => ""
4974     ),
4975     "uri" => "http://www.snookit.com/"
4976     ),
4977     "snoopy" => array(
4978     "icon" => "robot",
4979     "title" => "Snoopy",
4980     "rule" => array(
4981     "^Snoopy.+([0-9.]{1,10})" => "\\1",
4982     "sna-([0-9.]{1,10})" => "\\1"
4983     ),
4984     "uri" => "http://snoopy.sourceforge.net/"
4985     ),
4986     "snyke" => array(
4987     "icon" => "robot",
4988     "title" => "Snyke",
4989     "rule" => array(
4990     "^SnykeBot[ /]([0-9.]{1,10})" => "\\1"
4991     ),
4992     "uri" => "http://www.snyke.com"
4993     ),
4994     "socialradar" => array(
4995     "icon" => "socialradar",
4996     "title" => "Social Radar",
4997     "rule" => array(
4998     "Socialradarbot[ /]([0-9.]{1,10})" => "\\1"
4999     ),
5000     "uri" => "http://infegy.com/"
5001     ),
5002     "soegning" => array(
5003     "icon" => "soegning",
5004 matthys 209 "title" => "Søgning",
5005 joku 63 "rule" => array(
5006     "soegning\.dk[/ ]spider[ /]([0-9.]{1,10})" => "\\1"
5007     ),
5008     "uri" => ""
5009     ),
5010     "soft411" => array(
5011     "icon" => "soft411",
5012     "title" => "Soft411",
5013     "rule" => array(
5014     "SOFT411 Directory" => ""
5015     ),
5016     "uri" => ""
5017     ),
5018     "sogou" => array(
5019     "icon" => "sogou",
5020     "title" => "Sogou",
5021     "rule" => array(
5022     "Sogou web spider[ /]([0-9.]{1,10})" => "\\1"
5023     ),
5024     "uri" => "http://www.sogou.com/docs/help/webmasters.htm#07"
5025     ),
5026     "sohu" => array(
5027     "icon" => "robot",
5028     "title" => "Sohu",
5029     "rule" => array(
5030     "sohu[ \-](agent|search)" => ""
5031     ),
5032     "uri" => ""
5033     ),
5034     "somewhere" => array(
5035     "icon" => "robot",
5036 matthys 209 "title" => "Somewhere.com",
5037 joku 63 "rule" => array(
5038     "Mozilla\@somewhere\.com" => "\\1"
5039     ),
5040     "uri" => "http://www.somewhere.com/"
5041     ),
5042     "sopheus" => array(
5043     "icon" => "robot",
5044     "title" => "Sopheus",
5045     "rule" => array(
5046     "Sopheus Project[ /]([0-9.]{1,10})" => "\\1"
5047     ),
5048     "uri" => "http://www.thenetplanet.com"
5049     ),
5050     "soso" => array(
5051     "icon" => "soso",
5052     "title" => "Soso",
5053     "rule" => array(
5054     "^Soso(image)?spider" => ""
5055     ),
5056     "uri" => "http://help.soso.com/soso-image-spider.htm"
5057     ),
5058     "souppot" => array(
5059     "icon" => "robot",
5060     "title" => "SoupPot",
5061     "rule" => array(
5062     "SoupPotBot" => ""
5063     ),
5064     "uri" => ""
5065     ),
5066     "specificmedia" => array(
5067     "icon" => "specificmedia",
5068     "title" => "Specific Media",
5069     "rule" => array(
5070     "^SMBot[ /]([0-9.]{1,10})" => "\\1"
5071     ),
5072     "uri" => ""
5073     ),
5074     "sphider" => array(
5075     "icon" => "sphider",
5076     "title" => "Sphider",
5077     "rule" => array(
5078     "Sphider" => ""
5079     ),
5080     "uri" => "http://www.sphider.eu"
5081     ),
5082     "spinn3r" => array(
5083     "icon" => "robot",
5084     "title" => "Spinn3r",
5085     "rule" => array(
5086     "Spinn3r[ /]([0-9.]{1,10})" => "\\1"
5087     ),
5088     "uri" => "http://spinn3r.com/robot"
5089     ),
5090     "sproose" => array(
5091     "icon" => "robot",
5092 matthys 209 "title" => "Sproose",
5093 joku 63 "rule" => array(
5094     "^sproose[ /]([0-9a-z.]{1,10})" => "\\1"
5095     ),
5096     "uri" => "http://www.sproose.com/bot.html"
5097     ),
5098     "spurlbot" => array(
5099     "icon" => "robot",
5100     "title" => "SpurlBot",
5101     "rule" => array(
5102     "SpurlBot[/ ]([0-9.]{1,10})" => "\\1"
5103     ),
5104     "uri" => ""
5105     ),
5106     "stardownloader" => array(
5107     "icon" => "stardownloader",
5108     "title" => "Star Downloader",
5109     "rule" => array(
5110     "^Star Downloader( Pro)?" => ""
5111     ),
5112     "uri" => ""
5113     ),
5114     "steeler" => array(
5115     "icon" => "robot",
5116     "title" => "Steeler",
5117     "rule" => array(
5118     "Steeler[ /]([0-9.]{1,10})" => "\\1"
5119     ),
5120     "uri" => ""
5121     ),
5122     "strategicboard" => array(
5123     "icon" => "strategicboard",
5124     "title" => "Strategic Board",
5125     "rule" => array(
5126     "Strategic Board Bot" => ""
5127     ),
5128     "uri" => "http://www.strategicboard.com"
5129     ),
5130     "suchbaer" => array(
5131     "icon" => "robot",
5132 matthys 209 "title" => "Suchbaer.de",
5133 joku 63 "rule" => array(
5134     "^suchbaer\.de" => ""
5135     ),
5136     "uri" => "http://www.suchbaer.de/"
5137     ),
5138     "suchbot" => array(
5139     "icon" => "robot",
5140 matthys 209 "title" => "Suchbot",
5141 joku 63 "rule" => array(
5142     "^suchbot" => ""
5143     ),
5144     "uri" => ""
5145     ),
5146     "suchende" => array(
5147     "icon" => "suchende",
5148 matthys 209 "title" => "Suchen.de",
5149 joku 63 "rule" => array(
5150     "^gonzo([0-9]{1,2}).*www.suchen.de" => "\\1",
5151     "^gonzo/([0-9.]{1,10})" => "\\1"
5152     ),
5153     "uri" => "http://www.suchen.de/"
5154     ),
5155     "suchknecht" => array(
5156     "icon" => "robot",
5157     "title" => "Suchknecht",
5158     "rule" => array(
5159     "^Suchknecht.at-Robot" => ""
5160     ),
5161     "uri" => "http://www.suchknecht.at/"
5162     ),
5163     "suchpad" => array(
5164     "icon" => "robot",
5165 matthys 209 "title" => "Suchpad",
5166 joku 63 "rule" => array(
5167     "^suchpadbot[ /]([0-9.]{1,10})" => "\\1"
5168     ),
5169     "uri" => "http://www.suchpad.de"
5170     ),
5171     "sunrise" => array(
5172     "icon" => "sunrise",
5173     "title" => "Sunrise",
5174     "rule" => array(
5175     "^Sunrise[ /]([0-9a-z.]{1,10})" => "\\1"
5176     ),
5177     "uri" => ""
5178     ),
5179     "superbot" => array(
5180     "icon" => "superbot",
5181     "title" => "SuperBot",
5182     "rule" => array(
5183     "SuperBot[ /]([0-9.]{1,10})" => "\\1"
5184     ),
5185     "uri" => ""
5186     ),
5187     "surfcontrol" => array(
5188     "icon" => "robot",
5189     "title" => "SurfControl",
5190     "rule" => array(
5191     "SurfControl" => "",
5192     "ScSpider[ /]([0-9.]{1,10})" => "\\1"
5193     ),
5194     "uri" => ""
5195     ),
5196     "surfnet" => array(
5197     "icon" => "robot",
5198     "title" => "SURFnet",
5199     "rule" => array(
5200     "AVSearch[ \-]([0-9.]{1,10})" => "\\1"
5201     ),
5202     "uri" => ""
5203     ),
5204     "surfsafely" => array(
5205     "icon" => "robot",
5206     "title" => "Surfsafely",
5207     "rule" => array(
5208     "Submission Spider at surfsafely.com" => ""
5209     ),
5210     "uri" => "http://www.surfsafely.com"
5211     ),
5212     "surphace" => array(
5213     "icon" => "spherescout",
5214     "title" => "Surphace",
5215     "rule" => array(
5216     "Sphere Scout.+([0-9.]{1,10})" => "\\1",//Previous called Sphere Scout
5217     "Surphace.+v([0-9.]{1,10})" => "\\1"
5218     ),
5219     "uri" => "http://www.surphace.com"
5220     ),
5221     "surveybot" => array(
5222     "icon" => "surveybot",
5223     "title" => "Whois Survey",
5224     "rule" => array(
5225     "SurveyBot[ /]([0-9.]{1,10})" => "\\1"
5226     ),
5227     "uri" => ""
5228     ),
5229     "swoogle" => array(
5230     "icon" => "robot",
5231     "title" => "Swoogle",
5232     "rule" => array(
5233     "^Swooglebot[ /]([0-9.]{1,10})" => "\\1"
5234     ),
5235     "uri" => "http://swoogle.umbc.edu/swooglebot.htm"
5236     ),
5237     "swse" => array(
5238     "icon" => "robot",
5239     "title" => "SWSE",
5240     "rule" => array(
5241     "sw\.deri\.org" => ""
5242     ),
5243     "uri" => "http://sw.deri.org/2006/04/multicrawler/robots.html"
5244     ),
5245     "swish-e" => array(
5246     "icon" => "swish-e",
5247     "title" => "Swish-e",
5248     "rule" => array(
5249     "^swish-e" => ""
5250     ),
5251     "uri" => "http://swish-e.org/"
5252     ),
5253     "sygol" => array(
5254     "icon" => "sygol",
5255     "title" => "Sygol",
5256     "rule" => array(
5257     "www.sygol.(com|net)" => ""
5258     ),
5259     "uri" => "http://www.sygol.com"
5260     ),
5261     "syllabs" => array(
5262     "icon" => "syllabs",
5263     "title" => "Syllabs",
5264     "rule" => array(
5265     "Focal[-]([0-9.]{1,10})" => "\\1"
5266     ),
5267     "uri" => "http://www.syllabs.com/crawler.html"
5268     ),
5269     "synapse" => array(
5270     "icon" => "robot",
5271     "title" => "Synapse",
5272     "rule" => array(
5273     " Synapse\)" => ""
5274     ),
5275     "uri" => "http://ws.apache.org/synapse/"
5276     ),
5277     "sync2it" => array(
5278     "icon" => "robot",
5279 matthys 209 "title" => "Sync2it",
5280 joku 63 "rule" => array(
5281     "^\!Susie" => ""
5282     ),
5283     "uri" => "http://www.sync2it.com/bms/susie.php"
5284     ),
5285     "syncit" => array(
5286     "icon" => "robot",
5287 matthys 209 "title" => "Syncit",
5288 joku 63 "rule" => array(
5289     "^SyncIT[ /]([0-9.]{1,10})" => "\\1"
5290     ),
5291     "uri" => "http://www.syncit.com/"
5292     ),
5293     "syndic8" => array(
5294     "icon" => "syndic8",
5295     "title" => "Syndic8",
5296     "rule" => array(
5297     "Syndic8[ /]([0-9.]{1,10})" => "\\1"
5298     ),
5299     "uri" => "http://www.syndic8.com/"
5300     ),
5301     "syndicatie" => array(
5302     "icon" => "robot",
5303     "title" => "Syndicatie.nl",
5304     "rule" => array(
5305     "Syndicatie\.nl robot v ([0-9.]{1,10})" => "\\1",
5306     "Syndicatie\.nl robot;" => ""
5307     ),
5308     "uri" => ""
5309     ),
5310     "synomia" => array(
5311     "icon" => "robot",
5312     "title" => "Synomia",
5313     "rule" => array(
5314     "^SynoBot" => ""
5315     ),
5316     "uri" => ""
5317     ),
5318     "synoo" => array(
5319     "icon" => "robot",
5320     "title" => "SynooBot",
5321     "rule" => array(
5322     "SynooBot[ /]([0-9.]{1,10})" => "\\1",
5323     ),
5324     "uri" => ""
5325     ),
5326     "szukacz" => array(
5327     "icon" => "szukacz",
5328     "title" => "Szukacz",
5329     "rule" => array(
5330     "Szukacz[ /]([0-9.]{1,10})" => "\\1"
5331     ),
5332     "uri" => ""
5333     ),
5334     "tagoo" => array(
5335     "icon" => "tagoobot",
5336     "title" => "Tagoo",
5337     "rule" => array(
5338     "Tagoobot[ /]([0-9.]{1,10})" => "\\1"
5339     ),
5340     "uri" => "http://www.tagoo.ru"
5341     ),
5342     "tagword" => array(
5343     "icon" => "tagword",
5344     "title" => "Tagword",
5345     "rule" => array(
5346     "^Tagword" => ""
5347     ),
5348     "uri" => "http://tagword.com/dmoz_survey.php"
5349     ),
5350     "trailfire" => array(
5351     "icon" => "trailfire",
5352     "title" => "Trailfire",
5353     "rule" => array(
5354     "^Trailfire-bot[ /]([0-9.]{1,10})" => "\\1"
5355     ),
5356     "uri" => "http://trailfire.com"
5357     ),
5358     "tamu" => array(
5359     "icon" => "tamu",
5360     "title" => "Tamu Crawler",
5361     "rule" => array(
5362     "IRLbot[ /]([0-9.]{1,10})" => "\\1",
5363     "TAMU_CS_IRL_CRAWLER[ /]([0-9.]{1,10})" => "\\1"
5364     ),
5365     "uri" => "http://irl.cs.tamu.edu/crawler/"
5366     ),
5367     "taptubot" => array(
5368     "icon" => "taptubot",
5369     "title" => "Taptu",
5370     "rule" => array(
5371     "taptubot" => ""
5372     ),
5373     "uri" => "http://www.taptu.com/corp/taptubot"
5374     ),
5375     "targetseek" => array(
5376     "icon" => "robot",
5377     "title" => "TargetSeek",
5378     "rule" => array(
5379     "TargetSeek[ /]([0-9.]{1,10})" => "\\1"
5380     ),
5381     "uri" => "http://www.targetgroups.net/TargetSeek.html"
5382     ),
5383     "taw" => array(
5384     "icon" => "taw",
5385     "title" => "TAW",
5386     "rule" => array(
5387     "TAW[ /]([0-9.]{1,10})" => "\\1"
5388     ),
5389     "uri" => "http://www.tawdis.net"
5390     ),
5391     "tcd" => array(
5392     "icon" => "tcd",
5393     "title" => "Trinity College Dublin",
5394     "rule" => array(
5395     "^TCDBOT/Nutch-([0-9.]{1,10})" => "\\1"
5396     ),
5397     "uri" => "http://www.tcd.ie"
5398     ),
5399     "technorati" => array(
5400     "icon" => "technorati",
5401     "title" => "Technorati",
5402     "rule" => array(
5403     "Technoratibot[ /]([0-9.]{1,10})" => "\\1"
5404     ),
5405     "uri" => ""
5406     ),
5407     "teleport" => array(
5408     "icon" => "teleport",
5409     "title" => "Teleport",
5410     "rule" => array(
5411     "Teleport[ \-]?Pro" => ""
5412     ),
5413     "uri" => ""
5414     ),
5415     "terrar" => array(
5416     "icon" => "robot",
5417     "title" => "Terrar",
5418     "rule" => array(
5419     "^Fresh Search :: Terrar" => ""
5420     ),
5421     "uri" => ""
5422     ),
5423     "theophrastus" => array(
5424     "icon" => "robot",
5425     "title" => "Theophrastus",
5426     "rule" => array(
5427     "Theophrastus[ /]([0-9.]{1,10})" => "\\1"
5428     ),
5429     "uri" => "http://users.cs.cf.ac.uk/N.A.Smith/theophrastus.php"
5430     ),
5431     "thumbnailscz" => array(
5432     "icon" => "robot",
5433 matthys 209 "title" => "Thumbnails.cz",
5434 joku 63 "rule" => array(
5435     "^thumbnail\.cz robot[ /]([0-9.]{1,10})" => "\\1"
5436     ),
5437     "uri" => ""
5438     ),
5439     "thumbshot" => array(
5440     "icon" => "robot",
5441 matthys 209 "title" => "Thumbshots",
5442 joku 63 "rule" => array(
5443     "^thumbshots.*(Version: |v)([0-9.]{2,10})e" => "\\1",
5444     "^thumbshots-de" => ""
5445     ),
5446     "uri" => "http://www.thumbshots.de"
5447     ),
5448     "thunderstone" => array(
5449     "icon" => "thunderstone",
5450     "title" => "Thunderstone",
5451     "rule" => array(
5452     "T-H-U-N-D-E-R-S-T-O-N-E" => "",
5453     "search.thunderstone.com" => ""
5454     ),
5455     "uri" => "http://search.thunderstone.com/"
5456     ),
5457     "timbobot" => array(
5458     "icon" => "robot",
5459 matthys 209 "title" => "TimboBot",
5460 joku 63 "rule" => array(
5461     "timboBot" => ""
5462     ),
5463     "uri" => ""
5464     ),
5465     "topsy" => array(
5466     "icon" => "topsy",
5467     "title" => "Topsy",
5468     "rule" => array(
5469     "Butterfly[ /]([0-9.]{1,10})" => "\\1"
5470     ),
5471     "uri" => "http://labs.topsy.com/butterfly.html"
5472     ),
5473     "trayce" => array(
5474     "icon" => "robot",
5475 matthys 209 "title" => "Trayce",
5476 joku 63 "rule" => array(
5477     "traycebot[ /]([0-9a-z.\-]{1,10})" => "\\1"
5478     ),
5479     "uri" => ""
5480     ),
5481     "tricus" => array(
5482     "icon" => "robot",
5483     "title" => "Tricus",
5484     "rule" => array(
5485     "B_l_i_t_z_B_O_T_@_t_r_i_c_u_s_\._c_o_m" => ""
5486     ),
5487     "uri" => ""
5488     ),
5489     "topicblogs" => array(
5490     "icon" => "robot",
5491     "title" => "Topicblogs",
5492     "rule" => array(
5493     "topicblogs[ /]([0-9.]{1,10})" => "\\1"
5494     ),
5495     "uri" => ""
5496     ),
5497     "touche" => array(
5498     "icon" => "touche",
5499 matthys 209 "title" => "Touché",
5500 joku 63 "rule" => array(
5501     "^Touche" => ""
5502     ),
5503     "uri" => "http://touche.com.ve"
5504     ),
5505     "tuezilla" => array(
5506     "icon" => "robot",
5507 matthys 209 "title" => "TÃœzilla",
5508 joku 63 "rule" => array(
5509     "tuezilla.de" => ""
5510     ),
5511     "uri" => "http://tuezilla.de/t_st-odp-entries-agent.html"
5512     ),
5513     "turnitin" => array(
5514     "icon" => "turnitin",
5515     "title" => "Turnitin",
5516     "rule" => array(
5517     "TurnitinBot[ /]([0-9.]{1,10})" => "\\1"
5518     ),
5519     "uri" => ""
5520     ),
5521     "tutorgig" => array(
5522     "icon" => "robot",
5523     "title" => "TutorGig",
5524     "rule" => array(
5525     "TutorGig(Bot)?[ /]([0-9.]{1,10})" => "\\2"
5526     ),
5527     "uri" => ""
5528     ),
5529     "twiceler" => array(
5530     "icon" => "cuill",
5531 matthys 209 "title" => "Cuill",
5532 joku 63 "rule" => array(
5533     "Twiceler[ /-]([0-9.]{1,10})" => "\\1",
5534     "Twiceler" => ""
5535     ),
5536     "uri" => "http://www.cuill.com/twiceler/robot.html"
5537     ),
5538     "twingly" => array(
5539     "icon" => "twingly",
5540     "title" => "Twingly",
5541     "rule" => array(
5542     "Twingly Recon" => ""
5543     ),
5544     "uri" => "http://www.twingly.com/"
5545     ),
5546 matthys 320 "twitter" => array(
5547     "icon" => "twitter",
5548     "title" => "Twitter",
5549     "rule" => array(
5550     "Twitterbot[ /-]([0-9.]{1,10})" => "\\1"
5551     ),
5552     "uri" => "https://dev.twitter.com/cards/getting-started#crawling"
5553     ),
5554 joku 63 "typepad" => array(
5555     "icon" => "typepad",
5556     "title" => "TypePad",
5557     "rule" => array(
5558     "TypePad/([0-9a-z.]{1,10})" => "\\1"
5559     ),
5560     "uri" => ""
5561     ),
5562     "udmsearch" => array(
5563     "icon" => "robot",
5564     "title" => "UdmSearch",
5565     "rule" => array(
5566     "UdmSearch[/ ]([0-9.]{1,10})" => "\\1"
5567     ),
5568     "uri" => ""
5569     ),
5570     "ukwizz" => array(
5571     "icon" => "ukwizz",
5572     "title" => "UKWizz",
5573     "rule" => array(
5574     "^Mackster.*ukwizz" => ""
5575     ),
5576     "uri" => "http://www.ukwizz.com"
5577     ),
5578     "ultraseek" => array(
5579     "icon" => "robot",
5580     "title" => "Ultraseek",
5581     "rule" => array(
5582     "Ultraseek" => ""
5583     ),
5584     "uri" => ""
5585     ),
5586     "ultraspider" => array(
5587     "icon" => "robot",
5588     "title" => "UltraSpider",
5589     "rule" => array(
5590     "UltraSpider3000[/ ]([0-9.]{1,10})" => "\\1"
5591     ),
5592     "uri" => "http://www.search.ch"
5593     ),
5594     "umai" => array(
5595     "icon" => "robot",
5596 matthys 209 "title" => "Umai",
5597 joku 63 "rule" => array(
5598     "umai[/ ]([0-9.]{1,10})" => "\\1"
5599     ),
5600     "uri" => ""
5601     ),
5602     "unchaos" => array(
5603     "icon" => "robot",
5604     "title" => "Unchaos",
5605     "rule" => array(
5606     "unchaos_crawler[_ /]([0-9.]{1,10})" => "\\1",
5607     "unchaos" => ""
5608     ),
5609     "uri" => ""
5610     ),
5611     "unido" => array(
5612     "icon" => "robot",
5613 matthys 209 "title" => "UNIdo",
5614 joku 63 "rule" => array(
5615     "^unido-bot" => "\\1"
5616     ),
5617     "uri" => "http://mobicom.cs.uni-dortmund.de/bot.html"
5618     ),
5619     "updated" => array(
5620     "icon" => "robot",
5621     "title" => "Updated",
5622     "rule" => array(
5623     "updated[ /]([0-9a-z.]{1,10})" => "\\1"
5624     ),
5625     "uri" => ""
5626     ),
5627     "uptimebot" => array(
5628     "icon" => "robot",
5629     "title" => "UptimeBot",
5630     "rule" => array(
5631     "^UptimeBot" => ""
5632     ),
5633     "uri" => "http://www.uptimebot.com/"
5634     ),
5635     "urifetch" => array(
5636     "icon" => "robot",
5637     "title" => "URI::Fetch",
5638     "rule" => array(
5639     "^URI::Fetch[ /]([0-9.]{1,10})" => "\\1"
5640     ),
5641     "uri" => ""
5642     ),
5643     "urlbase" => array(
5644     "icon" => "robot",
5645     "title" => "URLBase",
5646     "rule" => array(
5647     "URLBase[ /]([0-9.]{1,10})" => "\\1"
5648     ),
5649     "uri" => ""
5650     ),
5651     "urlblaze" => array(
5652     "icon" => "robot",
5653     "title" => "URLBlaze",
5654     "rule" => array(
5655     "^URLBlaze" => ""
5656     ),
5657     "uri" => ""
5658     ),
5659     "urlcontr" => array(
5660     "icon" => "robot",
5661     "title" => "MS URL Control",
5662     "rule" => array(
5663     "Microsoft URL[ \-]?Control" => ""
5664     ),
5665     "uri" => ""
5666     ),
5667     "urlgetfile" => array(
5668     "icon" => "robot",
5669     "title" => "URLGetFile",
5670     "rule" => array(
5671     "^URLGetFile" => ""
5672     ),
5673     "uri" => ""
5674     ),
5675     "urlscope" => array(
5676     "icon" => "robot",
5677     "title" => "UrlScope",
5678     "rule" => array(
5679     "UrlScope" => ""
5680     ),
5681     "uri" => ""
5682     ),
5683     "urltrends" => array(
5684     "icon" => "urltrends",
5685 matthys 209 "title" => "UrlTrends",
5686 joku 63 "rule" => array(
5687     "Snappy/([0-9.]{1,10})" => "\\1",
5688     ),
5689     "uri" => ""
5690     ),
5691     "usww" => array(
5692     "icon" => "usww",
5693 matthys 209 "title" => "USWW",
5694 joku 63 "rule" => array(
5695     "usww\.com" => "",
5696     "Mozilla/5\.0 URL-Spider" => ""
5697     ),
5698     "uri" => "http://www.usww.com/"
5699     ),
5700     "usyd" => array(
5701     "icon" => "robot",
5702     "title" => "USyd-NLP-Spider",
5703     "rule" => array(
5704     "^USyd-NLP-Spider" => ""
5705     ),
5706     "uri" => "http://www.it.usyd.edu.au/~vinci/bot.html"
5707     ),
5708     "vagabondo" => array(
5709     "icon" => "wiseguys",
5710     "title" => "WiseGuys",
5711     "rule" => array(
5712     "Vagabondo[ /]([0-9.]{1,10})" => "\\1",
5713     "Vagabondo-WAP[ /]([0-9.]{1,10})" => "\\1"
5714     ),
5715     "uri" => ""
5716     ),
5717     "unicorn" => array(
5718     "icon" => "w3c",
5719     "title" => "W3C Unicorn",
5720     "rule" => array(
5721     "Unicorn[ /]([0-9.]{1,10})" => "\\1",
5722     ),
5723     "uri" => "http://validator.w3.org/unicorn/"
5724     ),
5725     "validator" => array(
5726     "icon" => "w3c",
5727     "title" => "W3C Validator",
5728     "rule" => array(
5729     "W3C_Validator[ /]([0-9.]{1,10})" => "\\1"
5730     ),
5731     "uri" => "http://validator.w3.org/"
5732     ),
5733     "validator.nu" => array(
5734     "icon" => "validatornu",
5735     "title" => "Validator.nu",
5736     "rule" => array(
5737     "^Validator.nu[ /]([0-9.]{1,10})" => "\\1",
5738     "^Validator.nu/LV" => "\\1"
5739     ),
5740     "uri" => "http://validator.nu/"
5741     ),
5742     "verity" => array(
5743     "icon" => "robot",
5744     "title" => "Verity",
5745     "rule" => array(
5746     "^vspider[ /]([0-9.]{1,10})" => "\\1",
5747     "^vspider" => ""
5748     ),
5749     "uri" => "http://www.verity.com/"
5750     ),
5751     "versionsproject" => array(
5752     "icon" => "robot",
5753     "title" => "Versions-project",
5754     "rule" => array(
5755     "InfoFly[ /]([0-9.]{1,10})" => "\\1"
5756     ),
5757     "uri" => "http://www.versions-project.org/"
5758     ),
5759     "verticalmatch" => array(
5760     "icon" => "robot",
5761     "title" => "VerticalMatch",
5762     "rule" => array(
5763     "^VMBot[ /]([0-9.]{1,10})" => "\\1"
5764     ),
5765     "uri" => "http://www.VerticalMatch.com/"
5766     ),
5767     "verzamelgids" => array(
5768     "icon" => "robot",
5769     "title" => "Verzamelgids",
5770     "rule" => array(
5771     "Verzamelgids[ /]([0-9.]{1,10})" => "\\1"
5772     ),
5773     "uri" => "http://www.verzamelgids.nl/"
5774     ),
5775     "vestris" => array(
5776     "icon" => "vestris",
5777     "title" => "Vestris",
5778     "rule" => array(
5779     "AlkalineBOT[ /]([0-9.]{1,10})" => "\\1"
5780     ),
5781     "uri" => "http://alkaline.vestris.com/"
5782     ),
5783     "vindex" => array(
5784     "icon" => "vindex",
5785     "title" => "Vindex",
5786     "rule" => array(
5787     "Vindex[ /]([0-9.]{1,10})" => "\\1"
5788     ),
5789     "uri" => ""
5790     ),
5791     "visvo" => array(
5792     "icon" => "robot",
5793     "title" => "Visvo",
5794     "rule" => array(
5795     "VisBot[ /]([0-9.]{1,10})" => "\\1"
5796     ),
5797     "uri" => "http://www.visvo.com"
5798     ),
5799     "voila" => array(
5800     "icon" => "voila",
5801     "title" => "Voila",
5802     "rule" => array(
5803     "VoilaBot[ /]?[a-z ]*([0-9.]{1,10})" => "\\1",
5804     "VoilaBot;[ /]([0-9.]{1,10})" => "\\1"
5805     ),
5806     "uri" => ""
5807     ),
5808     "vonna" => array(
5809     "icon" => "robot",
5810     "title" => "Vonna",
5811     "rule" => array(
5812     "Vonna.com b o t" => ""
5813     ),
5814     "uri" => ""
5815     ),
5816     "vortex" => array(
5817     "icon" => "robot",
5818     "title" => "Vortex",
5819     "rule" => array(
5820     "Vortex[ /]([0-9.]{1,10})" => "\\1"
5821     ),
5822     "uri" => "http://marty.anstey.ca/robots/vortex/"
5823     ),
5824     "w3sitesearch" => array(
5825     "icon" => "w3sitesearch",
5826     "title" => "W3SiteSearch",
5827     "rule" => array(
5828     "^W3SiteSearch Crawler[\_v]*([0-9.]{1,10})" => "\\1"
5829     ),
5830     "uri" => "http://www.w3sitesearch.de"
5831     ),
5832     "wagger" => array(
5833     "icon" => "robot",
5834     "title" => "Wagger",
5835     "rule" => array(
5836     "^Waggr" => ""
5837     ),
5838     "uri" => "http://www.waggr.com/"
5839     ),
5840     "wanadoo" => array(
5841     "icon" => "robot",
5842     "title" => "Wanadoo",
5843     "rule" => array(
5844     "^SurferF3[ /]([0-9./]{1,10})" => "\\1"
5845     ),
5846     "uri" => "http://www.wanadoo.fr/"
5847     ),
5848     "wapalizer" => array(
5849     "icon" => "robot",
5850     "title" => "Wapalizer",
5851     "rule" => array(
5852     "wapalizer[ /]([0-9.]{1,10})" => "\\1"
5853     ),
5854     "uri" => "http://www.wapdrive.com/"
5855     ),
5856     "watson" => array(
5857     "icon" => "addy",
5858     "title" => "Dr.Watson",
5859     "rule" => array(
5860     "Watson[ /]([0-9.]{1,10})" => "\\1",
5861     "watson\.addy\.com" => ""
5862     ),
5863     "uri" => ""
5864     ),
5865     "wavefire" => array(
5866     "icon" => "robot",
5867     "title" => "Wavefire",
5868     "rule" => array(
5869     "^Wavefire[ /]([0-9.]{1,10})" => "\\1"
5870     ),
5871     "uri" => ""
5872     ),
5873     "waypath" => array(
5874     "icon" => "waypath",
5875     "title" => "Waypath",
5876     "rule" => array(
5877     "Waypath[ \-]?Scout" => "",
5878     "Waypath (development )?crawler" => ""
5879     ),
5880     "uri" => ""
5881     ),
5882     "wauuu" => array(
5883     "icon" => "wauuu",
5884     "title" => "Wauuu",
5885     "rule" => array(
5886     "Wauuu[/ ]([0-9.]{1,10})" => "\\2"
5887     ),
5888     "uri" => "http://www.wauuu.com"
5889     ),
5890     "wdg" => array(
5891     "icon" => "wdg",
5892     "title" => "WDG Validator",
5893     "rule" => array(
5894     "^WDG_(Site)?Validator[/ ]([0-9.]{1,10})" => "\\2"
5895     ),
5896     "uri" => "http://www.htmlhelp.com/tools/validator/"
5897     ),
5898     "webagogo" => array(
5899     "icon" => "webagogo",
5900     "title" => "Webagogo",
5901     "rule" => array(
5902     "^Webagogo" => ""
5903     ),
5904     "uri" => "http://www.webagogo.be/"
5905     ),
5906     "webalta" => array(
5907     "icon" => "webalta",
5908     "title" => "WebAlta",
5909     "rule" => array(
5910     "^WebAlta( Crawler)?[/ ]([0-9.]{1,10})" => "\\2"
5911     ),
5912     "uri" => "http://www.webalta.net/ru/about_webmaster.html"
5913     ),
5914     "webdigity" => array(
5915     "icon" => "robot",
5916     "title" => "Webdigity Whois Service",
5917     "rule" => array(
5918     "^webdigity whois service" => ""
5919     ),
5920     "uri" => "http://www.webdigity.com/ws"
5921     ),
5922     "webelixir" => array(
5923     "icon" => "webelixir",
5924     "title" => "Webelixir",
5925     "rule" => array(
5926     "^webelixir" => ""
5927     ),
5928     "uri" => "http://www.webelixir.net/"
5929     ),
5930     "webbotru" => array(
5931     "icon" => "robot",
5932     "title" => "Webbot.ru",
5933     "rule" => array(
5934     " Webbot[/ ]([0-9.]{1,10})" => "\\1"
5935     ),
5936     "uri" => "http://www.webbot.ru/bot.html"
5937     ),
5938     "webcapture" => array(
5939     "icon" => "robot",
5940     "title" => "WebCapture",
5941     "rule" => array(
5942     "WebCapture[/ ]([0-9.]{1,10})" => "\\1"
5943     ),
5944     "uri" => ""
5945     ),
5946     "webcollage" => array(
5947     "icon" => "robot",
5948     "title" => "Webcollage",
5949     "rule" => array(
5950     "webcollage" => ""
5951     ),
5952     "uri" => ""
5953     ),
5954     "webcopier" => array(
5955     "icon" => "webcopier",
5956     "title" => "WebCopier",
5957     "rule" => array(
5958     "WebCopier[/ ]v?([0-9.]{1,10})" => "\\1"
5959     ),
5960     "uri" => ""
5961     ),
5962     "webcorp" => array(
5963     "icon" => "webcorp",
5964     "title" => "WebCorp",
5965     "rule" => array(
5966     "WebCorp[/ ]+([0-9.]{1,10})" => "\\1"
5967     ),
5968     "uri" => "http://www.webcorp.org.uk/"
5969     ),
5970     "webcrawl" => array(
5971     "icon" => "robot",
5972     "title" => "WebCrawl",
5973     "rule" => array(
5974     "webcrawl\.net" => ""
5975     ),
5976     "uri" => ""
5977     ),
5978     "webdownloader" => array(
5979     "icon" => "robot",
5980     "title" => "Web Downloader",
5981     "rule" => array(
5982     "Web Downloader[/ ]([0-9.]{1,10})" => "\\1"
5983     ),
5984     "uri" => ""
5985     ),
5986     "webfetch" => array(
5987     "icon" => "robot",
5988 matthys 209 "title" => "WebFetch",
5989 joku 63 "rule" => array(
5990     "^webfetch[/ ]([0-9.]{1,10})" => "\\1"
5991     ),
5992     "uri" => ""
5993     ),
5994     "webfind" => array(
5995     "icon" => "robot",
5996 matthys 209 "title" => "WebFind",
5997 joku 63 "rule" => array(
5998     "^WebFindBot" => ""
5999     ),
6000     "uri" => "http://www.web-find.com"
6001     ),
6002     "webglimpse" => array(
6003     "icon" => "robot",
6004     "title" => "Webglimpse",
6005     "rule" => array(
6006     "^Webglimpse[/ ]([0-9.]{1,10})" => "\\1"
6007     ),
6008     "uri" => "http://webglimpse.net"
6009     ),
6010     "webgobbler" => array(
6011     "icon" => "robot",
6012     "title" => "webGobbler",
6013     "rule" => array(
6014     "^webGobbler[/ ]([0-9.]{1,10})" => "\\1"
6015     ),
6016     "uri" => ""
6017     ),
6018     "webimages" => array(
6019     "icon" => "webimages",
6020     "title" => "WebImages",
6021     "rule" => array(
6022     "^WebImages[/ ]([0-9.]{1,10})" => "\\1"
6023     ),
6024     "uri" => "http://herbert.groot.jebbink.nl/"
6025     ),
6026     "weblight" => array(
6027     "icon" => "robot",
6028     "title" => "WebLight",
6029     "rule" => array(
6030     "^WebLight[/ ]([0-9.]{1,10})" => "\\1"
6031     ),
6032     "uri" => "http://www.illumit.com/Products/weblight/"
6033     ),
6034     "weblinks" => array(
6035     "icon" => "robot",
6036     "title" => "WebLink's",
6037     "rule" => array(
6038     "^Weblink.s checker" => ""
6039     ),
6040     "uri" => ""
6041     ),
6042     "webmeasurement" => array(
6043     "icon" => "robot",
6044     "title" => "Webmeasurement",
6045     "rule" => array(
6046     "^webmeasurement-bot" => ""
6047     ),
6048     "uri" => "http://rvs.informatik.uni-leipzig.de"
6049     ),
6050     "webminer" => array(
6051     "icon" => "robot",
6052     "title" => "WebMiner",
6053     "rule" => array(
6054     "^WebMiner[/ ]([0-9.]{1,10})" => "\\1"
6055     ),
6056     "uri" => ""
6057     ),
6058     "webmin" => array(
6059     "icon" => "webmin",
6060     "title" => "Webmin",
6061     "rule" => array(
6062     "^webmin" => ""
6063     ),
6064     "uri" => ""
6065     ),
6066     "webmon" => array(
6067     "icon" => "webmon",
6068     "title" => "Webmon",
6069     "rule" => array(
6070     "WebMon[ /]([0-9.]{1,10})" => "\\1"
6071     ),
6072     "uri" => "http://www.markwell.btinternet.co.uk/webmon/"
6073     ),
6074     "webox" => array(
6075     "icon" => "robot",
6076     "title" => " WeBoX",
6077     "rule" => array(
6078     "WeBoX[/ ]([0-9.]{1,10})" => "\\1"
6079     ),
6080     "uri" => ""
6081     ),
6082     "webpatrol" => array(
6083     "icon" => "webpatrol",
6084     "title" => "WebPatrol",
6085     "rule" => array(
6086     "^WebPatrol[ /]([0-9.]{1,10})" => "\\1"
6087     ),
6088     "uri" => "http://soft.macfeeling.com/WebPatrol.html"
6089     ),
6090     "webpix" => array(
6091     "icon" => "webpix",
6092     "title" => "WebPix",
6093     "rule" => array(
6094     "WebPix[/ ]([0-9.]{1,10})" => "\\1"
6095     ),
6096     "uri" => ""
6097     ),
6098     "webrace" => array(
6099     "icon" => "robot",
6100     "title" => "WebRACE",
6101     "rule" => array(
6102     "^WebRACE[/ ]([0-9.]{1,10})" => "\\1"
6103     ),
6104     "uri" => ""
6105     ),
6106     "webreaper" => array(
6107     "icon" => "webreaper",
6108     "title" => "WebReaper",
6109     "rule" => array(
6110     "^WebReaper " => ""
6111     ),
6112     "uri" => "http://www.webreaper.net/"
6113     ),
6114     "webresult" => array(
6115     "icon" => "robot",
6116     "title" => "Webresult",
6117     "rule" => array(
6118     "Der webresult\.de Robot" => ""
6119     ),
6120     "uri" => ""
6121     ),
6122     "webring" => array(
6123     "icon" => "robot",
6124     "title" => "Webring Checker",
6125     "rule" => array(
6126     "WebRingChecker[/ ]([0-9.]{1,10})" => "\\1"
6127     ),
6128     "uri" => ""
6129     ),
6130     "webripper" => array(
6131     "icon" => "webripper",
6132     "title" => "WebRipper",
6133     "rule" => array(
6134     "^WebRipper" => ""
6135     ),
6136     "uri" => "http://calluna-software.com"
6137     ),
6138     "webscaled" => array(
6139     "icon" => "webscaled",
6140     "title" => "Webscaled",
6141     "rule" => array(
6142     "Swarm" => ""
6143     ),
6144     "uri" => "http://webscaled.com/"
6145     ),
6146     "websearchau" => array(
6147     "icon" => "websearchau",
6148     "title" => "WebSearch.COM.AU",
6149     "rule" => array(
6150     "WebSearch.COM.AU[/ ]+([0-9.]{1,10})" => "\\1"
6151     ),
6152     "uri" => "http://WebSearch.com.au/"
6153     ),
6154     "websearchbench" => array(
6155     "icon" => "robot",
6156     "title" => "WebSearchBench",
6157     "rule" => array(
6158     "WebSearchBench WebCrawler[v/ ]+([0-9.]{1,10})" => "\\1"
6159     ),
6160     "uri" => "http://websearchbench.cs.uni-dortmund.de/"
6161     ),
6162     "websense" => array(
6163     "icon" => "websense",
6164     "title" => "Websense",
6165     "rule" => array(
6166     "(Sqworm|websense|Konqueror/3\.(0|1)(-rc[1-6])?; i686 Linux; 2002[0-9]{4})" => ""
6167     ),
6168     "uri" => ""
6169     ),
6170     "websiteshadow" => array(
6171     "icon" => "websiteshadow",
6172     "title" => "Websiteshadow",
6173     "rule" => array(
6174     "Websiteshadowbot[ /]([0-9.]{1,10})" => "\\1"
6175     ),
6176     "uri" => "http://websiteshadow.com"
6177     ),
6178     "websiteworth" => array(
6179     "icon" => "robot",
6180     "title" => "WebsiteWorth",
6181     "rule" => array(
6182     "WebsiteWorth[v/ ]+([0-9.]{1,10})" => "\\1"
6183     ),
6184     "uri" => "http://directory.sootle.com/website-worth/tata.php"
6185     ),
6186     "websquash" => array(
6187     "icon" => "websquash",
6188     "title" => "Websquash",
6189     "rule" => array(
6190     "webs(quash\.com|ite[ \-]?Monitor)" => ""
6191     ),
6192     "uri" => ""
6193     ),
6194     "webstripper" => array(
6195     "icon" => "robot",
6196     "title" => "WebStripper",
6197     "rule" => array(
6198     "WebStripper[ /]([0-9.]{1,10})" => "\\1"
6199     ),
6200     "uri" => ""
6201     ),
6202     "webzip" => array(
6203     "icon" => "webzip",
6204     "title" => "WebZIP",
6205     "rule" => array(
6206     "Web[ \-]?ZIP[ /]([0-9.]{1,10})" => "\\1"
6207     ),
6208     "uri" => ""
6209     ),
6210     "wep" => array(
6211     "icon" => "robot",
6212     "title" => "WEP Search",
6213     "rule" => array(
6214     "WEP Search[ /]([0-9.]{1,10})" => "\\1"
6215     ),
6216     "uri" => ""
6217     ),
6218     "westwind" => array(
6219     "icon" => "robot",
6220     "title" => "West Wind Internet Protocols",
6221     "rule" => array(
6222     "^West Wind Internet Protocols[ /]([0-9.]{1,10})" => "\\1"
6223     ),
6224     "uri" => "http://www.west-wind.com/wwipstuff.asp"
6225     ),
6226     "wget" => array(
6227     "icon" => "wget",
6228     "title" => "Wget",
6229     "rule" => array(
6230 joku 308 "Wget[ /]([0-9.]{1,10})" => "\\1",
6231     "Wget" => ""
6232 joku 63 ),
6233     "uri" => ""
6234     ),
6235     "whizbang" => array(
6236     "icon" => "whizbang",
6237     "title" => "WhizBang",
6238     "rule" => array(
6239     "WhizBang" => ""
6240     ),
6241     "uri" => "http://www.whizbang.com/crawler/"
6242     ),
6243     "whois" => array(
6244     "icon" => "whois",
6245     "title" => "Who.is",
6246     "rule" => array(
6247     "Who.is Bot" => ""
6248     ),
6249     "uri" => "http://www.who.is/"
6250     ),
6251     "wikiwix" => array(
6252     "icon" => "wikiwix",
6253     "title" => "Wikiwix",
6254     "rule" => array(
6255     "wikiwix-bot-([0-9.]{1,10})" => "\\1"
6256     ),
6257     "uri" => "http://www.wikiwix.com/"
6258     ),
6259     "wingflyer" => array(
6260     "icon" => "robot",
6261     "title" => "WingFlyer",
6262     "rule" => array(
6263     "^WebFetch" => ""
6264     ),
6265     "uri" => "http://www.wingflyer.com/"
6266     ),
6267     "wininet" => array(
6268     "icon" => "robot",
6269     "title" => "WinInet",
6270     "rule" => array(
6271     "TeamSoft WinInet Component" => ""
6272     ),
6273     "uri" => "http://www.winsoft.sk/wininet.htm"
6274     ),
6275     "winhttp" => array(
6276     "icon" => "robot",
6277     "title" => "WinHTTP",
6278     "rule" => array(
6279     "WinHttp\.WinHttpRequest\.([0-9.]{1,10})" => "\\1"
6280     ),
6281     "uri" => ""
6282     ),
6283     "wire" => array(
6284     "icon" => "robot",
6285     "title" => "WIRE",
6286     "rule" => array(
6287     "^WIRE[ /]([0-9.]{1,10})" => "\\1"
6288     ),
6289     "uri" => ""
6290     ),
6291     "wmp" => array(
6292     "icon" => "robot",
6293     "title" => "WMP",
6294     "rule" => array(
6295     "^WMP" => ""
6296     ),
6297     "uri" => ""
6298     ),
6299     "woozweb" => array(
6300     "icon" => "woozweb",
6301     "title" => "Woozweb",
6302     "rule" => array(
6303     "woozweb" => ""
6304     ),
6305     "uri" => "http://www.woozweb.com/"
6306     ),
6307     "wordchamp" => array(
6308     "icon" => "robot",
6309     "title" => "WordChamp",
6310     "rule" => array(
6311     "^WordChampBot" => ""
6312     ),
6313     "uri" => "http://www.wordchamp.com/"
6314     ),
6315     "wordpress" => array(
6316     "icon" => "wordpress",
6317     "title" => "WordPress",
6318     "rule" => array(
6319     "WordPress[ /]([0-9.]{1,10})" => "\\1"
6320     ),
6321     "uri" => ""
6322     ),
6323     "worio" => array(
6324     "icon" => "worio",
6325     "title" => "Worio",
6326     "rule" => array(
6327     "woriobot" => ""
6328     ),
6329     "uri" => "http://worio.com/"
6330     ),
6331     "worldlight" => array(
6332     "icon" => "worldlight",
6333     "title" => "WorldLight",
6334     "rule" => array(
6335     "^WorldLight" => ""
6336     ),
6337     "uri" => ""
6338     ),
6339     "worqmada" => array(
6340     "icon" => "robot",
6341     "title" => "WorQmada",
6342     "rule" => array(
6343     "WorQmada[ /]([0-9.]{1,10})" => "\\1"
6344     ),
6345     "uri" => ""
6346     ),
6347     "wotbox" => array(
6348     "icon" => "wotbox",
6349     "title" => "Wotbox",
6350     "rule" => array(
6351     "Wotbox[ /]?[a-z]*([0-9.]{1,10})" => "\\1"
6352     ),
6353     "uri" => ""
6354     ),
6355     "wp" => array(
6356     "icon" => "wp",
6357     "title" => "Wirtualna Polska",
6358     "rule" => array(
6359     "NetSprint[ /\-]{1,4}([0-9.]{1,10})" => "\\1"
6360     ),
6361     "uri" => "http://wp.pl"
6362     ),
6363     "wsb" => array(
6364     "icon" => "robot",
6365     "title" => "WebSearchBench",
6366     "rule" => array(
6367     "WSB WebCrawler V([0-9.]{1,10})" => "\\1",
6368     "WSB " => ""
6369     ),
6370     "uri" => "http://websearchbench.cs.uni-dortmund.de/"
6371     ),
6372     "wume" => array(
6373     "icon" => "robot",
6374     "title" => "WUME Lab's",
6375     "rule" => array(
6376     "^wume_crawler[ /]([0-9.]{1,10})" => "\\1",
6377     ),
6378     "uri" => "http://wume.cse.lehigh.edu/~xiq204/crawler/"
6379     ),
6380     "wusage" => array(
6381     "icon" => "wusage",
6382     "title" => "Wusage",
6383     "rule" => array(
6384     "Wusage[ /]([0-9.]{1,10})" => "\\1",
6385     ),
6386     "uri" => "http://www.boutell.com/wusage/"
6387     ),
6388     "wwgrapevine" => array(
6389     "icon" => "wwgrapevine",
6390     "title" => "WWgrapevine",
6391     "rule" => array(
6392     "wwgrapevine[ /]([0-9.]{1,10})" => "\\1"
6393     ),
6394     "uri" => ""
6395     ),
6396     "wws" => array(
6397     "icon" => "robot",
6398     "title" => "WWSBOT",
6399     "rule" => array(
6400     "WWSBOT[ /]([0-9.]{1,10})" => "\\1",
6401     ),
6402     "uri" => "http://www.analyzer.nu"
6403     ),
6404     "www4mail" => array(
6405     "icon" => "robot",
6406 matthys 209 "title" => "WWW4mail",
6407 joku 63 "rule" => array(
6408     "^www4mail[ /]([0-9.]{1,10})" => "\\1"
6409     ),
6410     "uri" => "http://www.www4mail.org/"
6411     ),
6412     "wwwc" => array(
6413     "icon" => "wwwc",
6414     "title" => "WWWC",
6415     "rule" => array(
6416     "^WWWC[ /]([0-9.]{1,10})" => "\\1"
6417     ),
6418     "uri" => ""
6419     ),
6420     "wwwd" => array(
6421     "icon" => "robot",
6422     "title" => "WWWD",
6423     "rule" => array(
6424     "^WWWD[ /]([0-9.]{1,10})" => "\\1"
6425     ),
6426     "uri" => ""
6427     ),
6428     "wwweasel" => array(
6429     "icon" => "wwweasel",
6430     "title" => "WWWeasel",
6431     "rule" => array(
6432     "WWWeasel( Robot)?[/ ]v?([0-9.]{1,10})" => "\\2"
6433     ),
6434     "uri" => ""
6435     ),
6436     "wwwfi" => array(
6437     "icon" => "wwwfi",
6438     "title" => "www.fi",
6439     "rule" => array(
6440     "www\.fi crawler" => ""
6441     ),
6442     "uri" => "http://www.fi/"
6443     ),
6444     "wwwmechanize" => array(
6445     "icon" => "robot",
6446     "title" => "WWW-Mechanize",
6447     "rule" => array(
6448     "^WWW-Mechanize[/ ]([0-9.]{1,10})" => "\\1"
6449     ),
6450     "uri" => ""
6451     ),
6452     "wwwoffle" => array(
6453     "icon" => "robot",
6454     "title" => "WWWoffle",
6455     "rule" => array(
6456     "^wwwoffle[/ ]([0-9.]{1,10})" => "\\1"
6457     ),
6458     "uri" => ""
6459     ),
6460     "wwwster" => array(
6461     "icon" => "robot",
6462     "title" => "WWWster",
6463     "rule" => array(
6464     "^wwwster[/ ]([0-9.]{1,10})" => "\\1"
6465     ),
6466     "uri" => ""
6467     ),
6468     "wysigot" => array(
6469     "icon" => "wysigot",
6470     "title" => "Wysigot",
6471     "rule" => array(
6472     "Wysigot[/ ]([0-9.]{1,10})" => "\\1"
6473     ),
6474     "uri" => ""
6475     ),
6476     "xaldon" => array(
6477     "icon" => "xaldon",
6478     "title" => "Xaldon",
6479     "rule" => array(
6480     "Xaldon WebSpider" => ""
6481     ),
6482     "uri" => "http://www.xaldon.de/"
6483     ),
6484     "xenu" => array(
6485     "icon" => "xenu",
6486     "title" => "Xenu Link Sleuth",
6487     "rule" => array(
6488     "Xenu Link Sleuth[/ ]([0-9a-z.]{1,10})" => "\\1",
6489     "Xenu(&#039;s)? Link Sleuth[/ ]([0-9a-z.]{1,10})" => "\\1",
6490     "Xenu_Link_Sleuth_([0-9a-z.]{1,10})" => "\\1"
6491     ),
6492     "uri" => "http://home.snafu.de/tilman/xenulink.html"
6493     ),
6494     "xerka" => array(
6495     "icon" => "xerka",
6496     "title" => "Xerka",
6497     "rule" => array(
6498     "^Xerka WebBot v([0-9a-z.]{1,10})" => "\\1"
6499     ),
6500     "uri" => "http://www.diana-teknologia.com/www1/english/xerka.htm"
6501     ),
6502     "xirq" => array(
6503     "icon" => "robot",
6504     "title" => "XIRQ",
6505     "rule" => array(
6506     "^xirq[ /]([0-9a-z.]{1,10})" => "\\1"
6507     ),
6508     "uri" => "http://www.xirq.com"
6509     ),
6510     "xmlslurp" => array(
6511     "icon" => "robot",
6512     "title" => "XMLSlurp",
6513     "rule" => array(
6514     "^XMLSlurp[ /]([0-9a-z.]{1,10})" => "\\1"
6515     ),
6516     "uri" => ""
6517     ),
6518     "xml-sitemaps" => array(
6519     "icon" => "xml-sitemaps",
6520     "title" => "XML-Sitemaps",
6521     "rule" => array(
6522     "XML-Sitemaps[ /]([0-9a-z.]{1,10})" => "\\1"
6523     ),
6524     "uri" => ""
6525     ),
6526     "xmlrpc" => array(
6527     "icon" => "robot",
6528     "title" => "Trackback",
6529     "rule" => array(
6530     "XMLRPC" => ""
6531     ),
6532     "uri" => ""
6533     ),
6534 matthys 294 "xovi" => array(
6535     "icon" => "xovi",
6536     "title" => "Xovi",
6537     "rule" => array(
6538     "XoviBot[ /]([0-9a-z.]{1,10})" => "\\1"
6539     ),
6540     "uri" => "http://www.xovibot.net"
6541     ),
6542 joku 63 "yacy" => array(
6543     "icon" => "yacy",
6544     "title" => "Yacy",
6545     "rule" => array(
6546     "yacy\.net" => ""
6547     ),
6548     "uri" => ""
6549     ),
6550     "yahoo" => array(
6551     "icon" => "yahoo",
6552     "title" => "Yahoo",
6553     "rule" => array(
6554     "Yahoo(! ([a-z]{1,3} )?Slurp|-)" => "",
6555     "Yahoo-MMCrawler[/ ]([0-9a-z.]{1,10})" => "\\1",
6556     "Yahoo-VerticalCrawler-FormerWebCrawler[/ ]([0-9a-z.]{1,10})" => "\\1",
6557     "^AnzwersCrawl[/ ]([0-9a-z.]{1,10})" => "\\1",
6558     "Y!J(-BSC|-SRD)*[/ ]([0-9a-z.]{1,10})" => "\\2",
6559     "Y!OASIS/TEST" => "",
6560     "Harvest-NG[/ ]([0-9a-z.]{1,10})" => "\\1",
6561     "Y!J; for robot study" => "",
6562     "Yahoo Japan; for robot study" => ""
6563     ),
6564     "uri" => "http://www.yahoo.com"
6565     ),
6566     "yahoo-feed" => array(
6567     "icon" => "yahoo",
6568     "title" => "Yahoo Feedseeker",
6569     "rule" => array(
6570     "^YahooFeedSeeker[/ ]([0-9a-z.]{1,10})" => "\\1"
6571     ),
6572     "uri" => "http://www.yahoo.com"
6573     ),
6574     "yandex" => array(
6575     "icon" => "yandex",
6576     "title" => "Yandex",
6577     "rule" => array(
6578     "Yandex[/ ]([0-9.]{1,10})" => "\\1",
6579     "Yandex(Bot|Images|Media|Antivirus)[/ ]([0-9.]{1,10})" => "\\2"
6580     ),
6581     "uri" => "http://yandex.com/bots"
6582     ),
6583     "yanga" => array(
6584     "icon" => "yanga",
6585     "title" => "Yanga",
6586     "rule" => array (
6587     "^Yanga.*v([0-9.]{1,10})" => "\\1"
6588     ),
6589     "uri" => "http://www.yanga.co.uk/"
6590     ),
6591     "yap" => array(
6592     "icon" => "yap",
6593     "title" => "Yap",
6594     "rule" => array(
6595     "^YapBot[_/ ]v([0-9.]{1,10})" => "\\1"
6596     ),
6597     "uri" => "http://www.yapinc.com/"
6598     ),
6599     "yarienavoir" => array(
6600     "icon" => "robot",
6601     "title" => "Yarienavoir",
6602     "rule" => array(
6603     "^yarienavoir.net[/ ]([0-9.]{1,10})" => "\\1"
6604     ),
6605     "uri" => "http://www.yarienavoir.net/"
6606     ),
6607     "yell" => array(
6608     "icon" => "yell",
6609     "title" => "Yell",
6610     "rule" => array(
6611     "YellCrawl[ /]V?([0-9.]{1,10})" => "\\1",
6612     "Yellbot[ /]Nutch-([0-9.]{1,10})" => "\\1",
6613     ),
6614     "uri" => ""
6615     ),
6616     "youdao" => array(
6617     "icon" => "youdao",
6618     "title" => "Youdao",
6619     "rule" => array(
6620     "YodaoBot-(Image|Mobile)?[ /]([0-9.]{1,10})" => "\\2",
6621     "YoudaoBot[ /]([0-9.]{1,10})" => "\\1"
6622     ),
6623     "uri" => "http://www.youdao.com/help/webmaster/spider/"
6624     ),
6625     "yoogli" => array(
6626     "icon" => "yoogli",
6627     "title" => "Yoogli",
6628     "rule" => array(
6629     "yoogliFetchAgent[ /]([0-9.]{1,10})" => "\\1"
6630     ),
6631     "uri" => "http://www.yoogli.com"
6632     ),
6633     "yotta" => array(
6634     "icon" => "robot",
6635     "title" => "Yotta",
6636     "rule" => array(
6637     "Yotta(Shopping|Cars)_Bot[ /]([0-9.]{1,10})" => "\\2",
6638     "OmniExplorer_Bot[ /]([0-9.]{1,10})" => "\\1"
6639     ),
6640     "uri" => "http://www.yottacars.com"
6641     ),
6642     "yoono" => array(
6643     "icon" => "yoono",
6644     "title" => "Yoono",
6645     "rule" => array(
6646     "Yoono" => ""
6647     ),
6648     "uri" => "http://www.yoono.com/"
6649     ),
6650     "yowedo" => array(
6651     "icon" => "robot",
6652     "title" => "Yowedo",
6653     "rule" => array(
6654     "Yowedo[ /]([0-9.]{1,10})" => "\\1"
6655     ),
6656     "uri" => "http://yowedo.com/en/partners.html"
6657     ),
6658     "yuntis" => array(
6659     "icon" => "robot",
6660 matthys 209 "title" => "Yuntis",
6661 joku 63 "rule" => array(
6662     "Gulper Web Bot[ /]([0-9.]{1,10})" => "\\1"
6663     ),
6664     "uri" => "http://yuntis.ecsl.cs.sunysb.edu/help/robot/"
6665     ),
6666     "zao" => array(
6667     "icon" => "robot",
6668     "title" => "Zao",
6669     "rule" => array(
6670     "Zao[ /]([0-9.]{1,10})" => "\\1",
6671     "Zao-crawler" => ""
6672     ),
6673     "uri" => ""
6674     ),
6675     "zealbot" => array(
6676     "icon" => "zeal",
6677     "title" => "ZealBot",
6678     "rule" => array(
6679     "Zealbot[ /]([0-9.]{1,10})" => "\\1"
6680     ),
6681     "uri" => ""
6682     ),
6683     "zearchit" => array(
6684     "icon" => "zearchit",
6685     "title" => "Zearchit",
6686     "rule" => array(
6687     "Zearchit" => ""
6688     ),
6689     "uri" => "http://www.zearchit.de/"
6690     ),
6691     "zebz" => array(
6692     "icon" => "robot",
6693     "title" => "ze.bz",
6694     "rule" => array(
6695     "^ZeBot_(lseek\.net|www\.ze\.bz)" => ""
6696     ),
6697     "uri" => "http://www.ze.bz/"
6698     ),
6699     "zedzo" => array(
6700     "icon" => "robot",
6701     "title" => "Zedzo",
6702     "rule" => array(
6703     "zedzo.digest[ /]([0-9.]{1,10})" => "\\1"
6704     ),
6705     "uri" => "http://www.zedzo.com/"
6706     ),
6707     "zerx" => array(
6708     "icon" => "zerx",
6709     "title" => "Zerx",
6710     "rule" => array(
6711     "^zerxbot[ /](Version|v)*[ /]*([0-9.]{1,10})" => "\\2"
6712     ),
6713     "uri" => "http://www.zerx.com/"
6714     ),
6715     "zeus" => array(
6716     "icon" => "zeus",
6717     "title" => "Zeus",
6718     "rule" => array(
6719     "Zeus" => ""
6720     ),
6721     "uri" => "http://www.zeus.com"
6722     ),
6723     "zippp" => array(
6724     "icon" => "robot",
6725     "title" => "Zippp",
6726     "rule" => array(
6727     "ZipppBot[ /]([0-9.]{1,10})" => "\\1"
6728     ),
6729     "uri" => ""
6730     ),
6731     "zippy" => array(
6732     "icon" => "robot",
6733     "title" => "Zippy",
6734     "rule" => array(
6735     "^Zippy[ v/]*([0-9.]{1,10})" => "\\1"
6736     ),
6737     "uri" => "http://www.zippyfinder.com"
6738     ),
6739     "zoeky" => array(
6740     "icon" => "robot",
6741     "title" => "Zoeky",
6742     "rule" => array(
6743     "Zoekybot[ /]([0-9.]{1,10})" => "\\1"
6744     ),
6745     "uri" => ""
6746     ),
6747     "zoom" => array(
6748     "icon" => "zoom",
6749 matthys 209 "title" => "ZoomSpider",
6750 joku 63 "rule" => array(
6751     "^ZoomSpider" => ""
6752     ),
6753     "uri" => "http://www.wrensoft.com/"
6754     ),
6755     "zspider" => array(
6756     "icon" => "robot",
6757 matthys 209 "title" => "Zspider",
6758 joku 63 "rule" => array(
6759     "zspider[ /]([0-9.a-z]{1,10})" => "\\1"
6760     ),
6761     "uri" => "http://feedback.redkolibri.com/"
6762     ),
6763 matthys 281 "zumbot" => array(
6764     "icon" => "robot",
6765     "title" => "ZUM Search",
6766     "rule" => array(
6767     "ZumBot[ /]([0-9.]{1,10})" => "\\1"
6768     ),
6769     "uri" => "http://help.zum.com/inquiry"
6770     ),
6771 joku 63 "zyborg" => array(
6772     "icon" => "zyborg",
6773     "title" => "WiseNutBot",
6774     "rule" => array(
6775     "(WISE|Zy)bo(rg|t)[ /]([0-9.]{1,10})" => "\\3"
6776     ),
6777     "uri" => ""
6778     ),
6779     // Catch up for the originals, they got to stay in that order.
6780     "blogbot" => array(
6781     "icon" => "blogbot",
6782     "title" => "BlogBot",
6783     "rule" => array(
6784     "Blog[ \-]?Bot" => ""
6785     ),
6786     "uri" => "http://www.blogbot.com/"
6787     ),
6788     "centrum" => array(
6789     "icon" => "centrum",
6790     "title" => "Centrum",
6791     "rule" => array(
6792     "holmes[/ ]([0-9.]{1,10})" => "\\1",
6793     "^Centrum-checker" => ""
6794     ),
6795     "uri" => ""
6796     ),
6797     "httpclient" => array(
6798     "icon" => "robot",
6799     "title" => "HTTPClient",
6800     "rule" => array(
6801     "HTTP[ \-]?Client[ /]([0-9.]{1,10})" => "\\1",
6802     "HTTP[ \-]?Client" => ""
6803     ),
6804     "uri" => "http://www.innovation.ch/java/HTTPClient/"
6805     ),
6806     "incywincy" => array(
6807     "icon" => "robot",
6808     "title" => "IncyWincy",
6809     "rule" => array(
6810     "^IncyWincy[ /]([0-9.]{1,10})" => "\\1",
6811     "^IncyWincy" => ""
6812     ),
6813     "uri" => ""
6814     ),
6815     "java" => array(
6816     "icon" => "java",
6817     "title" => "Java",
6818     "rule" => array(
6819     "^java[ /]*([0-9.a-z]{1,10})" => "\\1"
6820     ),
6821     "uri" => ""
6822     ),
6823     "libfetch" => array(
6824     "icon" => "robot",
6825     "title" => "Libfetch",
6826     "rule" => array(
6827     "^(fetch )?libfetch[ /]([0-9.]{1,10})" => "\\2"
6828     ),
6829     "uri" => "http://www.freebsd.org/"
6830     ),
6831     "libwww" => array(
6832     "icon" => "libwww",
6833 matthys 209 "title" => "LibWWW",
6834 joku 63 "rule" => array(
6835     "^libww(w|w-perl|w-FM)[ /]([0-9.]{1,10})" => "\\2",
6836     "^libww(w|w-perl|w-FM)" => "",
6837     "MyApp.*libww(w|w-perl|w-FM)" => ""
6838     ),
6839     "uri" => ""
6840     ),
6841     "litefinder" => array(
6842     "icon" => "litefinder",
6843     "title" => "LiteFinder",
6844     "rule" => array(
6845     "LiteFinder[ /]([0-9.]{1,10})" => "\\1"
6846     ),
6847     "uri" => "http://www.litefinder.net/about.html"
6848     ),
6849     "nutchorg" => array(
6850     "icon" => "nutchorg",
6851     "title" => "Nutch",
6852     "rule" => array(
6853     "Nutc(hOrg|hCVS|h)?[ /]([0-9.]{1,10})" => "\\2",
6854     "Nutch" => ""
6855     ),
6856     "uri" => "http://lucene.apache.org/nutch/"
6857     ),
6858     "pythonurl" => array(
6859     "icon" => "pythonurl",
6860     "title" => "Python-url",
6861     "rule" => array(
6862     "Python[ \-]?urllib" => ""
6863     ),
6864     "uri" => ""
6865     ),
6866     // Know Robots as SPAM BOTS
6867     "SPAM" => array(
6868     "icon" => "robot",
6869     "title" => "SPAM",
6870     "rule" => array(
6871     "NASA Search[/ ]([0-9.]{1,10})" => "\\1",
6872     "^PHOTO CHECK" => "",
6873     "^FOTOCHECKER" => "",
6874     "^IPTC CHECK" => "",
6875     "^DataCha0s" => "",
6876     "^Mac Finder" => "",
6877     "^Missigua Locator[ /]([0-9.]{1,10})" => "\\1",
6878     "^Missouri College Browse" => "",
6879     "Email[ \-]?Siphon" => "",
6880     "atSpider" => "",
6881     "autoemailspider" => "",
6882     "^Demo Bot" => "",
6883     "^Program Shareware" => "",
6884     "^Snapbot" => "",
6885     "^snap.com" => "",
6886     "^Guestbook Auto Submitter" => "",
6887     "panscient.com" => "",
6888     ),
6889     "uri" => ""
6890     ),
6891     // Things we don't know by now
6892     "robot" => array(
6893     "icon" => "robot",
6894     "title" => "Robot",
6895     "rule" => array(
6896     "(robot|spider|harvest|(?<!orange)bot|(?<!msie)crawler)" => ""
6897     )
6898     )
6899     );
6900 joku 261 ?>

Properties

Name Value
svn:keywords Id