ViewVC Help
View File | Revision Log | Show Annotations | View Changeset | Root Listing
root/BBClone/trunk/lib/robot.php
Revision: 402
Committed: Sun Nov 7 11:52:51 2021 UTC (3 years ago) by matthys
File size: 159768 byte(s)
Log Message:
Added Neevabot detection

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 394 * Copyright (C) 2001-2021, 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 matthys 401 "^NutchEC2Test" => "",
294     "Amazonbot[/ ]([0-9.]{1,10})" => "\\1"
295 joku 63 ),
296     "uri" => "http://www.amazon.com/"
297     ),
298     "amidalla" => array(
299     "icon" => "amidalla",
300     "title" => "Amidalla",
301     "rule" => array(
302     "^amibot" => ""
303     ),
304     "uri" => ""
305     ),
306     "amfibi" => array(
307     "icon" => "amfibi",
308     "title" => "Amfibi",
309     "rule" => array(
310     "Amfibibot[/ ]([0-9.]{1,10})" => "\\1",
311     "Amfibibot" => ""
312     ),
313     "uri" => ""
314     ),
315     "amphetadesk" => array(
316     "icon" => "robot",
317     "title" => "AmphetaDesk",
318     "rule" => array(
319     "AmphetaDesk[/ ]([0-9.]{1,10})" => "\\1"
320     ),
321     "uri" => ""
322     ),
323     "amphetameme" => array(
324     "icon" => "robot",
325     "title" => "Amphetameme",
326     "rule" => array(
327     "amphetameme[ \-]?crawler" => ""
328     ),
329     "uri" => ""
330     ),
331     "annomille" => array(
332     "icon" => "robot",
333     "title" => "AnnoMille",
334     "rule" => array(
335     "^AnnoMille( spider)?[/ ]([0-9.]{1,10})" => "\\2"
336     ),
337     "uri" => "http://www.annomille.it"
338     ),
339     "anonymouse" => array(
340     "icon" => "anonymouse",
341     "title" => "Anonymouse.org",
342     "rule" => array(
343     "http://Anonymouse.org" => ""
344     ),
345     "uri" => "http://anonymouse.org"
346     ),
347     "ansearch" => array(
348     "icon" => "robot",
349     "title" => "Ansearch",
350     "rule" => array(
351     "AnsearchBot[/ ]([0-9.]{1,10})" => "\\1"
352     ),
353     "uri" => ""
354     ),
355     "answerchase" => array(
356     "icon" => "answerchase",
357     "title" => "AnswerChase",
358     "rule" => array(
359     "AnswerChase( PROve)?[/ ]([0-9.]{1,10})" => "\\2"
360     ),
361     "uri" => "http://www.answerchase.com/"
362     ),
363     "antibot" => array(
364     "icon" => "robot",
365 matthys 205 "title" => "Antibot",
366 joku 63 "rule" => array(
367     "antibot-V([0-9.]{1,10})" => "\\1"
368     ),
369     "uri" => ""
370     ),
371     "aonde" => array(
372     "icon" => "aonde",
373     "title" => "Aonde",
374     "rule" => array(
375     "^AONDE-Spider" => ""
376     ),
377     "uri" => ""
378     ),
379     "aonline" => array(
380     "icon" => "robot",
381     "title" => "A-Online.at",
382     "rule" => array(
383     "^A-Online Search" => ""
384     ),
385     "uri" => "http://www.a-online.at/"
386     ),
387     "aol" => array(
388     "icon" => "aol",
389     "title" => "AOLserver",
390     "rule" => array(
391     "^AOLserver-Tcl[/ ]([0-9.]{1,10})" => "\\1",
392     "^AOLserver" => ""
393     ),
394     "uri" => ""
395     ),
396     "apachebench" => array(
397     "icon" => "robot",
398     "title" => "ApacheBench",
399     "rule" => array(
400     "ApacheBench[ /]([0-9.]{1,10})" => "\\1"
401     ),
402     "uri" => ""
403     ),
404     "apassion4jazz" => array(
405     "icon" => "robot",
406     "title" => "Passion 4 Jazz",
407     "rule" => array(
408     "^BebopBot[ /]([0-9.]{1,10})" => "\\1"
409     ),
410     "uri" => "http://www.apassion4jazz.net/bebopbot.html"
411     ),
412     "apexoo" => array(
413     "icon" => "robot",
414     "title" => "Apexoo",
415     "rule" => array(
416     "^Apexoo Spider ([0-9.]{1,10})" => "\\1"
417     ),
418     "uri" => "http://www.apexoo.com/"
419     ),
420     "apnoti" => array(
421     "icon" => "apnoti",
422 matthys 205 "title" => "Apnoti",
423 joku 63 "rule" => array(
424     "apnoti.*Robot[ /]v([0-9.]{1,10})" => "\\1"
425     ),
426     "uri" => "http://www.apnoti.com/"
427     ),
428     "aport" => array(
429     "icon" => "aport",
430     "title" => "Aport",
431     "rule" => array(
432     "^Aport" => ""
433     ),
434     "uri" => ""
435     ),
436     "appie" => array(
437     "icon" => "walhello",
438     "title" => "Walhello",
439     "rule" => array(
440     "appie[ /]([0-9.]{1,10})" => "\\1"
441     ),
442     "uri" => ""
443     ),
444 matthys 389 "apple" => array(
445     "icon" => "apple",
446     "title" => "Apple",
447     "rule" => array(
448     "Applebot[ /]([0-9.]{1,10})" => "\\1"
449     ),
450     "uri" => "http://www.apple.com/go/applebot"
451     ),
452    
453 joku 63 "arachmo" => array(
454     "icon" => "arachmo",
455     "title" => "Arachmo",
456     "rule" => array(
457     "compatible; Arachmo" => ""
458     ),
459     "uri" => ""
460     ),
461     "arexera" => array(
462     "icon" => "arexera",
463     "title" => "Arexera",
464     "rule" => array(
465     "^X-Crawler" => "",
466     "^TECOMAC-Crawler[ /]([0-9.]{1,10})" => "\\1"
467     ),
468     "uri" => "http://www.arexera.de/"
469     ),
470     "arianna" => array(
471     "icon" => "arianna",
472     "title" => "Arianna",
473     "rule" => array(
474     "^www.arianna.it" => ""
475     ),
476     "uri" => "http://www.arianna.it/"
477     ),
478     "artface" => array(
479     "icon" => "robot",
480     "title" => "Artface",
481     "rule" => array(
482     "^ArtfaceBot" => ""
483     ),
484     "uri" => ""
485     ),
486     "artviper" => array(
487     "icon" => "artviper",
488     "title" => "artViper",
489     "rule" => array(
490     "artViper" => ""
491     ),
492     "uri" => "http://www.artviper.net/"
493     ),
494     "asinfo" => array(
495     "icon" => "robot",
496     "title" => "Any Search Info",
497     "rule" => array(
498     "Sleek Spider[ /]([0-9.]{1,10})" => "\\1"
499     ),
500     "uri" => "http://search-info.com/"
501     ),
502     "ask" => array(
503     "icon" => "askjeeves",
504     "title" => "Ask Jeeves",
505     "rule" => array(
506     "Ask[ \-]?Jeeves" => "",
507     "teomaagent" => ""
508     ),
509     "uri" => ""
510     ),
511     "askaboutoil" => array(
512     "icon" => "robot",
513     "title" => "ASPseek",
514     "rule" => array(
515     "^AskAboutOil[ /]([0-9.]{1,10})" => "\\1"
516     ),
517     "uri" => "http://askaboutoil.com/"
518     ),
519     "asked" => array(
520     "icon" => "robot",
521 matthys 205 "title" => "AskEd!",
522 joku 63 "rule" => array(
523     "^asked[ /]Nutch[ \-]([0-9.]{1,10})" => "\\1"
524     ),
525     "uri" => "http://asked.jp"
526     ),
527     "aspseek" => array(
528     "icon" => "robot",
529     "title" => "ASPseek",
530     "rule" => array(
531     "^ASPseek[/ ]([0-9.]{1,10})" => "\\1"
532     ),
533     "uri" => ""
534     ),
535     "atlocal" => array(
536     "icon" => "robot",
537     "title" => "At Local",
538     "rule" => array(
539     "AtlocalBot[/ ]([0-9.]{1,10})" => "\\1"
540     ),
541     "uri" => "http://www.atlocal.com/"
542     ),
543     "atomz" => array(
544     "icon" => "atomz",
545     "title" => "Atomz",
546     "rule" => array(
547     "Atomz[/ ]([0-9.]{1,10})" => "\\1"
548     ),
549     "uri" => ""
550     ),
551 matthys 345 "avira" => array(
552     "icon" => "avira",
553     "title" => "Avira SafeSearch",
554     "rule" => array(
555     "^SafeSearch microdata crawler" => ""
556     ),
557     "uri" => "https://safesearch.avira.com"
558     ),
559 joku 63 "axel" => array(
560     "icon" => "robot",
561     "title" => "Axel",
562     "rule" => array(
563     "^axel" => ""
564     ),
565     "uri" => ""
566     ),
567     "axmo" => array(
568     "icon" => "axmo",
569     "title" => "Axmo",
570     "rule" => array(
571     "AxmoRobot" => ""
572     ),
573     "uri" => ""
574     ),
575     "answerbus" => array(
576     "icon" => "answerbus",
577     "title" => "AnswerBus",
578     "rule" => array(
579     "answerbus" => ""
580     ),
581     "uri" => ""
582     ),
583     "automapit" => array(
584     "icon" => "robot",
585     "title" => "AutoMapIt",
586     "rule" => array(
587     "AutoMapIt[ /](Bot)?" => ""
588     ),
589     "uri" => "http://www.automapit.com/bot.html"
590     ),
591     "augurnfind" => array(
592     "icon" => "robot",
593     "title" => "Augurnfind",
594     "rule" => array(
595     "augurnfind[/ ][v\-]*([0-9.]{1,10})" => "\\1"
596     ),
597     "uri" => ""
598     ),
599     "awasu" => array(
600     "icon" => "awasu",
601     "title" => "Awasu",
602     "rule" => array(
603     "Awasu[/ ]([0-9a-z.]{1,10})" => "\\1"
604     ),
605     "uri" => ""
606     ),
607     "axxus" => array(
608     "icon" => "axxus",
609     "title" => "axxus.de",
610     "rule" => array(
611     "whoiam" => ""
612     ),
613     "uri" => "http://www.axxus.de/"
614     ),
615     "ayell" => array(
616     "icon" => "ayell",
617     "title" => "Ayell",
618     "rule" => array(
619     "^Eurobot[ /]([0-9.]{1,10})" => "\\1"
620     ),
621     "uri" => "http://eurobot.ayell.eu"
622     ),
623     "babaloo" => array(
624     "icon" => "babaloo",
625     "title" => "Babaloo",
626     "rule" => array(
627     "BabalooSpider[ /]([0-9.]{1,10})" => "\\1"
628     ),
629     "uri" => "http://www.babaloo.si"
630     ),
631     "babe" => array(
632     "icon" => "robot",
633     "title" => "ba.be",
634     "rule" => array(
635     "BACS http://www.ba.be" => ""
636     ),
637     "uri" => "http://www.ba.be/"
638     ),
639     "baidu" => array(
640     "icon" => "baidu",
641     "title" => "Baidu",
642     "rule" => array(
643     "Baiduspider[ /]([0-9.]{1,10})" => "\\1",
644     "Baiduspider" => ""
645     ),
646     "uri" => "http://www.baidu.com/search/spider.html"
647     ),
648     "bananatree" => array(
649     "icon" => "robot",
650     "title" => "BananaTree",
651     "rule" => array(
652     "www\.thebananatree\.org" => ""
653     ),
654     "uri" => "http://www.thebananatree.org/"
655     ),
656     "bdcindexer" => array(
657     "icon" => "robot",
658     "title" => "bdcindexer",
659     "rule" => array(
660     "bdcindexer_([0-9a-z.]{1,10})" => "\\1"
661     ),
662     "uri" => "http://www.business.com/"
663     ),
664     "bdfetch" => array(
665     "icon" => "robot",
666     "title" => "BDFetch",
667     "rule" => array(
668     "^BDFetch" => ""
669     ),
670     "uri" => ""
671     ),
672     "bdncentral" => array(
673     "icon" => "robot",
674     "title" => "Bdncentral",
675     "rule" => array(
676     "BDNcentral Crawler v([0-9.]{1,10})" => "\\1"
677     ),
678     "uri" => "http://www.bdncentral.com/robot.html"
679     ),
680     "beammachine" => array(
681     "icon" => "robot",
682     "title" => "BeamMachine",
683     "rule" => array(
684     "^BeamMachine[ /]([0-9.]{1,10})" => "\\1"
685     ),
686     "uri" => "http://www.beammachine.net/"
687     ),
688     "become" => array(
689     "icon" => "become",
690     "title" => "Become",
691     "rule" => array(
692     "Become(JP)?Bot[/ ]([0-9.]{1,10})" => "\\2",
693     "(BecomeBot|Exabot)@exava\.com\)$" => ""
694     ),
695     "uri" => "http://www.become.com/"
696     ),
697     "beebware" => array(
698     "icon" => "robot",
699     "title" => "Beebware",
700     "rule" => array(
701     "BeebwareDirectory[/ ]v?([0-9.]{1,10})" => "\\2"
702     ),
703     "uri" => ""
704     ),
705     "bigbrother" => array(
706     "icon" => "robot",
707     "title" => "Big Brother",
708     "rule" => array(
709     "^Big Brother" => ""
710     ),
711     "uri" => "http://pauillac.inria.fr/~fpottier/"
712     ),
713     "bigclique" => array(
714     "icon" => "robot",
715     "title" => "BigClique",
716     "rule" => array(
717     "^BigCliqueBOT[/ ]([0-9.]{1,10})" => "\\1"
718     ),
719     "uri" => "http://www.bigclique.com"
720     ),
721     "biglotron" => array(
722     "icon" => "robot",
723     "title" => "Biglotron",
724     "rule" => array(
725     "^BIGLOTRON" => ""
726     ),
727     "uri" => "http://www.bigclique.com"
728     ),
729     "bigsearch" => array(
730     "icon" => "bigsearch",
731     "title" => "Bigsearch",
732     "rule" => array(
733     "Bigsearch.ca[/ ]Nutch[- ]([0-9.]{1,10})" => "\\1"
734     ),
735     "uri" => ""
736     ),
737     "bilbo" => array(
738     "icon" => "robot",
739     "title" => "Bilbo",
740     "rule" => array(
741     "Bilbo[ /]([0-9.]{1,10})" => "\\1"
742     ),
743     "uri" => "http://home.broadpark.no/~tnilsen-1/Linux/Bilbo_-_Nessus_WEB/bilbo_-_nessus_web.html"
744     ),
745     "bilgi" => array(
746     "icon" => "bilgi",
747     "title" => "Bilgi",
748     "rule" => array(
749     "Bilgi(Beta)?Bot[ /]([0-9.]{1,10})" => "\\1"
750     ),
751     "uri" => "http://www.bilgi.com/"
752     ),
753     "bitacle" => array(
754     "icon" => "bitacle",
755     "title" => "Bitacle",
756     "rule" => array(
757     "Bitacle (ro)?bot[ \(/V\:]+([0-9.]{1,10})" => "\\2",
758     ),
759     "uri" => "http://bitacle.org/"
760     ),
761     "bitbeamer" => array(
762     "icon" => "bitbeamer",
763     "title" => "BitBeamer",
764     "rule" => array(
765     "BitBeamer/([0-9.]{1,10})" => "\\1"
766     ),
767     "uri" => ""
768     ),
769     "biz360" => array(
770     "icon" => "biz360",
771     "title" => "Biz360",
772     "rule" => array(
773     "^Biz360 spider" => ""
774     ),
775     "uri" => ""
776     ),
777     "backlinktest" => array(
778     "icon" => "backlinktest",
779     "title" => "BacklinkCrawler",
780     "rule" => array(
781     "BacklinkCrawler" => ""
782     ),
783     "uri" => "http://www.backlinktest.com/crawler.html"
784     ),
785     "blaizbee" => array(
786     "icon" => "blaiz-bee",
787     "title" => "Blaiz-Bee",
788     "rule" => array(
789     "Blaiz-Bee[ /]([0-9.]{1,10})" => ""
790     ),
791     "uri" => ""
792     ),
793     "blogbeat" => array(
794     "icon" => "robot",
795     "title" => "BlogBeat",
796     "rule" => array(
797     "BlogBeat/?([0-9.]{1,10})" => "\\1",
798     ),
799     "uri" => "http://www.blogbeat.nl"
800     ),
801     "blogbotde" => array(
802     "icon" => "robot",
803     "title" => "blogbot.de",
804     "rule" => array(
805     "Naamah[ /]([0-9.a-z]{1,10})[ /]Blogbot" => "\\1",
806     ),
807     "uri" => "http://blogbot.de/"
808     ),
809     "blogcensus" => array(
810     "icon" => "blogcensus",
811     "title" => "Blogcensus",
812     "rule" => array(
813     "^Blogcensus" => "",
814     ),
815     "uri" => "http://blogcensus.de/bot/"
816     ),
817     "blogdex" => array(
818     "icon" => "robot",
819     "title" => "Blogdex",
820     "rule" => array(
821     "Blogdex[ /]([0-9.]{1,10})" => "\\1",
822     ),
823     "uri" => ""
824     ),
825     "blogg" => array(
826     "icon" => "blogg",
827     "title" => "Blogg",
828     "rule" => array(
829     "^blogg\.de" => ""
830     ),
831     "uri" => ""
832     ),
833     "blogland" => array(
834     "icon" => "robot",
835     "title" => "BlogLand",
836     "rule" => array(
837     "BlogLand[/ ]([0-9.]{1,10})" => "\\1"
838     ),
839     "uri" => ""
840     ),
841     "bloglines" => array(
842     "icon" => "bloglines",
843     "title" => "Bloglines",
844     "rule" => array(
845     "Bloglines[ /]([0-9.]{1,10})" => "\\1",
846     "Bloglines" => ""
847     ),
848     "uri" => ""
849     ),
850     "blogmap" => array(
851     "icon" => "robot",
852     "title" => "Blogmap",
853     "rule" => array(
854     "blogmap" => ""
855     ),
856     "uri" => ""
857     ),
858     "blogosphere" => array(
859     "icon" => "robot",
860     "title" => "Blogosphere",
861     "rule" => array(
862     "Blogosphere" => ""
863     ),
864     "uri" => ""
865     ),
866     "blogpeople" => array(
867     "icon" => "robot",
868     "title" => "BlogPeople",
869     "rule" => array(
870     "BlogPeople" => ""
871     ),
872     "uri" => ""
873     ),
874     "blogpulse" => array(
875     "icon" => "blogpulse",
876     "title" => "Blogpulse",
877     "rule" => array(
878     "Blogpulse" => ""
879     ),
880     "uri" => ""
881     ),
882     "blogranking" => array(
883     "icon" => "blogranking",
884     "title" => "BlogRanking",
885     "rule" => array(
886     "^BlogRanking(/RSS checker)?" => ""
887     ),
888     "uri" => ""
889     ),
890     "blogs" => array(
891     "icon" => "blogs",
892     "title" => "Blo.gs",
893     "rule" => array(
894     "blo\.gs[ /]([0-9.]{1,10})" => "\\1",
895     "blo\.gs" => ""
896     ),
897     "uri" => ""
898     ),
899     "blogshares" => array(
900     "icon" => "blogshares",
901     "title" => "BlogShares",
902     "rule" => array(
903     "BlogShares[ /]V?([0-9.]{1,10})" => "\\1",
904     "(^| |\()Blogshares(\.com| |\))" => ""
905     ),
906     "uri" => ""
907     ),
908     "blogslife" => array(
909     "icon" => "robot",
910     "title" => "BlogsLife",
911     "rule" => array(
912     "Blogslive" => ""
913     ),
914     "uri" => ""
915     ),
916     "blogsnow" => array(
917     "icon" => "blogsnow",
918     "title" => "BlogsNow",
919     "rule" => array(
920     "blogsnowbot" => "",
921     "BlogsNow" => ""
922     ),
923     "uri" => ""
924     ),
925     "blogstreet" => array(
926     "icon" => "blogstreet",
927     "title" => "BlogStreet",
928     "rule" => array(
929     "^BlogStreetBot" => ""
930     ),
931     "uri" => ""
932     ),
933     "blogsurf" => array(
934     "icon" => "robot",
935     "title" => "BlogSurf",
936     "rule" => array(
937     "nomadscafe_ra[/ ]([0-9.]{1,10})" => "\\1"
938     ),
939     "uri" => ""
940     ),
941     "blogtick" => array(
942     "icon" => "robot",
943     "title" => "BlogTick",
944     "rule" => array(
945     "BlogTickServer" => ""
946     ),
947     "uri" => ""
948     ),
949     "blogwatcher" => array(
950     "icon" => "blogwatcher",
951     "title" => "Blogwatcher",
952     "rule" => array(
953     "blogWatcher_Spider[/ ]([0-9.]{1,10})" => "\\1"
954     ),
955     "uri" => ""
956     ),
957     "blogwise" => array(
958     "icon" => "blogwise",
959     "title" => "Blogwise",
960     "rule" => array(
961     "Blogwise\.com(-MetaChecker)?[/ ]([0-9.]{1,10})" => "\\2"
962     ),
963     "uri" => ""
964     ),
965     "boardreader" => array(
966     "icon" => "boardreader",
967     "title" => "BoardReader",
968     "rule" => array(
969     "BoardReader[ \-](Image|Favicon)[ \-]Fetcher[ /]+([0-9.]{1,10})" => "\\2"
970     ),
971     "uri" => "http://www.boardreader.com"
972     ),
973     "bobby" => array(
974     "icon" => "bobby",
975     "title" => "Bobby",
976     "rule" => array(
977     "bobby[ /]([0-9.]{1,10})" => "\\1"
978     ),
979     "uri" => ""
980     ),
981     "boitho" => array(
982     "icon" => "robot",
983     "title" => "Boitho",
984     "rule" => array(
985     "Boitho\.com[ \-](dc|robot)?[/ ]([0-9.]{1,10})" => "\\2"
986     ),
987     "uri" => ""
988     ),
989     "booch" => array(
990     "icon" => "robot",
991     "title" => "Booch",
992     "rule" => array(
993     "^booch[_ /]([0-9.]{1,10})" => "\\1"
994     ),
995     "uri" => ""
996     ),
997     "book" => array(
998     "icon" => "book",
999     "title" => "Bookmark",
1000     "rule" => array(
1001     "http://www\.bookmark\.ne\.jp" => ""
1002     ),
1003     "uri" => ""
1004     ),
1005     "bookdog" => array(
1006     "icon" => "bookdog",
1007     "title" => "Bookdog",
1008     "rule" => array(
1009     "^Bookdog[ /]([0-9.]{1,10})" => "\\1"
1010     ),
1011     "uri" => ""
1012     ),
1013     "bordermanager" => array(
1014     "icon" => "bordermanager",
1015     "title" => "Border Manager",
1016     "rule" => array(
1017     "BorderManager[ /]([0-9.]{1,10})" => "\\1"
1018     ),
1019     "uri" => ""
1020     ),
1021     "botonparade" => array(
1022     "icon" => "botonparade",
1023     "title" => "Bot on Parade",
1024     "rule" => array(
1025     "BotOnParade" => ""
1026     ),
1027     "uri" => "http://www.bots-on-para.de/bot.html"
1028     ),
1029     "botje" => array(
1030     "icon" => "botje",
1031     "title" => "Botje",
1032     "rule" => array(
1033     "Plukkie[ /]([0-9.]{1,10})" => "\\1"
1034     ),
1035     "uri" => "http://www.botje.com/"
1036     ),
1037     "bottomfeeder" => array(
1038     "icon" => "bottomfeeder",
1039     "title" => "BottomFeeder",
1040     "rule" => array(
1041     "BottomFeeder[ /]([0-9.]{1,10})" => "\\1"
1042     ),
1043     "uri" => ""
1044     ),
1045     "browseremulator" => array(
1046     "icon" => "browseremulator",
1047     "title" => "BrowserEmulator",
1048     "rule" => array(
1049     "BrowserEmulator[ /]([0-9.]{1,10})" => "\\1"
1050     ),
1051     "uri" => "http://www.dejavu.org/emulator.htm"
1052     ),
1053     "browsershots" => array(
1054     "icon" => "browsershots",
1055     "title" => "Browsershots",
1056     "rule" => array(
1057     "Browsershots URL Check" => ""
1058     ),
1059     "uri" => "http://browsershots.org"
1060     ),
1061     "browserspy" => array(
1062     "icon" => "robot",
1063     "title" => "BrowserSpy",
1064     "rule" => array(
1065     "BrowserSpy" => ""
1066     ),
1067     "uri" => ""
1068     ),
1069     "bruinbot" => array(
1070     "icon" => "robot",
1071     "title" => "BruinBot",
1072     "rule" => array(
1073     "BruinBot" => ""
1074     ),
1075     "uri" => ""
1076     ),
1077     "bruno" => array(
1078     "icon" => "robot",
1079     "title" => "Bruno",
1080     "rule" => array(
1081     "^Bruno" => ""
1082     ),
1083     "uri" => ""
1084     ),
1085     "btbot" => array(
1086     "icon" => "btbot",
1087     "title" => "BitTorrent",
1088     "rule" => array(
1089     "BTbot/([0-9.]{1,10})" => "\\1"
1090     ),
1091     "uri" => ""
1092     ),
1093 joku 380 "bubing" => array(
1094     "icon" => "robot",
1095     "title" => "BUbiNG",
1096     "rule" => array(
1097     "BUbiNG" => ""
1098     ),
1099     "uri" => "http://law.di.unimi.it/BUbiNG.html"
1100     ),
1101 joku 63 "builtwith" => array(
1102     "icon" => "builtwith",
1103     "title" => "BuiltWith",
1104     "rule" => array(
1105     "BuiltWith[/ ]([0-9.]{1,10})" => "\\1"
1106     ),
1107     "uri" => "http://builtwith.com/bot.html"
1108     ),
1109     "bulkfeeds" => array(
1110     "icon" => "robot",
1111     "title" => "Bulkfeeds",
1112     "rule" => array(
1113     "Bulkfeeds[/ ]([a-z0-9.]{1,10})" => "\\1"
1114     ),
1115     "uri" => ""
1116     ),
1117     "burf" => array(
1118     "icon" => "burf",
1119     "title" => "Burf.com",
1120     "rule" => array(
1121     "^Norbert the Spider" => ""
1122     ),
1123     "uri" => "http://www.burf.com/"
1124     ),
1125     "butch" => array(
1126     "icon" => "robot",
1127     "title" => "Butch",
1128     "rule" => array(
1129     "Butch(__| )?([a-z0-9.]{1,10})" => "\\2"
1130     ),
1131     "uri" => ""
1132     ),
1133     "buzzlogic" => array(
1134     "icon" => "buzzlogic",
1135     "title" => "Buzzlogic",
1136     "rule" => array(
1137     "BuzzBot[ /]([0-9.]{1,10})" => "\\1"
1138     ),
1139     "uri" => "http://www.buzzlogic.com"
1140     ),
1141     "catchbot" => array(
1142     "icon" => "catchbot",
1143     "title" => "CatchBot",
1144     "rule" => array(
1145     "^CatchBot[/ ]([0-9.]{1,10})" => "\\1"
1146     ),
1147     "uri" => "http://www.catchbot.com"
1148     ),
1149     "camdiscover" => array(
1150     "icon" => "robot",
1151     "title" => "Camdiscover",
1152     "rule" => array(
1153     "^Camcrawler" => ""
1154     ),
1155     "uri" => ""
1156     ),
1157     "cazoodle" => array(
1158     "icon" => "cazoodle",
1159     "title" => "Cazoodle",
1160     "rule" => array(
1161     "^CazoodleBot/(Nutch|CazoodleBot)[/ \-]([0-9.]{1,10})" => "\\2",
1162     "^CazoodleBot[/ ]([a-z0-9.]{1,10})" => "\\1"
1163     ),
1164     "uri" => "http://www.cazoodle.com/cazoodlebot"
1165     ),
1166     "ccbot" => array(
1167     "icon" => "robot",
1168     "title" => "ccBot",
1169     "rule" => array(
1170     "CCBot[/ ]([0-9.]{1,10})" => "\\1"
1171     ),
1172     "uri" => "http://www.commoncrawl.org/bot.html"
1173     ),
1174     "ccgcrawl" => array(
1175     "icon" => "robot",
1176     "title" => "CCGCrawl",
1177     "rule" => array(
1178     "CCGCrawl[/ ]([0-9.]{1,10})" => "\\1"
1179     ),
1180     "uri" => "http://www.myworkbase.com/bot.html"
1181     ),
1182     "cerberian" => array(
1183     "icon" => "robot",
1184     "title" => "Cerberian Drtrs",
1185     "rule" => array(
1186     "^Cerberian Drtrs" => ""
1187     ),
1188     "uri" => ""
1189     ),
1190     "cfnetwork" => array(
1191     "icon" => "robot",
1192     "title" => "Cerberian Drtrs",
1193     "rule" => array(
1194     "^CFNetwork[/ ]([0-9.]{1,10})" => "\\1"
1195     ),
1196     "uri" => "http://www.cfnetwork.be/"
1197     ),
1198     "charlotte" => array(
1199     "icon" => "charlotte",
1200     "title" => "Charlotte",
1201     "rule" => array(
1202     "Charlotte[/ ]([0-9a-z.]{1,10})" => "\\1"
1203     ),
1204     "uri" => ""
1205     ),
1206     "cirilizator" => array(
1207     "icon" => "cirilizator",
1208     "title" => "Cirilizator",
1209     "rule" => array(
1210     "Cirilizator[/ ]([0-9.]{1,10})" => "\\1"
1211     ),
1212     "uri" => ""
1213     ),
1214     "claria" => array(
1215     "icon" => "robot",
1216     "title" => "Claria",
1217     "rule" => array(
1218     "(Claria|Diamond)(Bot)?[ /]([0-9.]{1,10})" => "\\3",
1219     "(Claria|Diamond)(Bot)" => ""
1220     ),
1221     "uri" => ""
1222     ),
1223     "claymont" => array(
1224     "icon" => "claymont",
1225     "title" => "Claymont",
1226     "rule" => array(
1227     "claymont\.com" => "",
1228     "OliverPerry" => ""
1229     ),
1230     "uri" => "http://www.claymont.com"
1231     ),
1232 joku 377 "clipinc" => array(
1233     "icon" => "clipinc",
1234     "title" => "ClipInc",
1235     "rule" => array(
1236     "ClipInc" => ""
1237     ),
1238     "uri" => "http://tobit.software"
1239     ),
1240 joku 63 "clush" => array(
1241     "icon" => "clush",
1242     "title" => "Clush",
1243     "rule" => array(
1244     "Clus(tered-Search-|h)Bot[ /]([0-9.]{1,10})" => "\\2"
1245     ),
1246     "uri" => ""
1247     ),
1248     "cmsworldmap" => array(
1249     "icon" => "cmsworldmap",
1250     "title" => "CMS World Map",
1251     "rule" => array(
1252     "cmsworldmap" => ""
1253     ),
1254     "uri" => "http://www.cmsworldmap.com/"
1255     ),
1256     "cobion" => array(
1257     "icon" => "robot",
1258     "title" => "Cobion",
1259     "rule" => array(
1260     " (QXW03018|obot)\)" => ""
1261     ),
1262     "uri" => ""
1263     ),
1264     "coccoc" => array(
1265 matthys 287 "icon" => "coccoc",
1266 matthys 209 "title" => "Coccoc",
1267 joku 63 "rule" => array(
1268 matthys 400 "coccoc[ /]([0-9.]{1,10})" => "\\1",
1269     "coccocbot-web[ /]([0-9.]{1,10})" => "\\1"
1270 joku 63 ),
1271 matthys 287 "uri" => "http://help.coccoc.com"
1272 joku 63 ),
1273     "coldfusion" => array(
1274     "icon" => "coldfusion",
1275     "title" => "ColdFusion",
1276     "rule" => array(
1277     "^coldfusion" => ""
1278     ),
1279     "uri" => ""
1280     ),
1281     "combine" => array(
1282     "icon" => "robot",
1283     "title" => "Combine",
1284     "rule" => array(
1285     "Combine[ /]([0-9.]{1,10})" => "\\1"
1286     ),
1287     "uri" => ""
1288     ),
1289     "combot" => array(
1290     "icon" => "robot",
1291     "title" => "comBot",
1292     "rule" => array(
1293     "^comBot[ /]([0-9.]{1,10})" => "\\1"
1294     ),
1295     "uri" => ""
1296     ),
1297     "comet" => array(
1298     "icon" => "comet",
1299     "title" => "Comet",
1300     "rule" => array(
1301     "cometsearch@cometsystems" => ""
1302     ),
1303     "uri" => ""
1304     ),
1305     "commerobo" => array(
1306     "icon" => "robot",
1307     "title" => "Commerobo",
1308     "rule" => array(
1309     "Commerobo[/ ]([0-9.]{1,10})" => "\\1"
1310     ),
1311     "uri" => ""
1312     ),
1313     "comrite" => array(
1314     "icon" => "robot",
1315     "title" => "ComRite",
1316     "rule" => array(
1317     "Comrite[/ ]([0-9.]{1,10})" => "\\1"
1318     ),
1319     "uri" => "http://www.comrite.com/"
1320     ),
1321     "convera" => array(
1322     "icon" => "convera",
1323     "title" => "Convera",
1324     "rule" => array(
1325     "Convera(MultiMedia)?Crawler[/ ]([0-9.]{1,10})" => "\\2",
1326     "Convera Internet Spider V([0-9.]{1,10})" => "\\1"
1327     ),
1328     "uri" => ""
1329     ),
1330     "coolbot" => array(
1331     "icon" => "robot",
1332     "title" => "CoolBot",
1333     "rule" => array(
1334     "^CoolBot" => ""
1335     ),
1336     "uri" => ""
1337     ),
1338     "cosmix" => array(
1339     "icon" => "robot",
1340     "title" => "Cosmix",
1341     "rule" => array(
1342     "^(voyager|cfetch|CosmixCrawler|carleson)[/ ]([0-9.]{1,10})" => "\\2",
1343     ),
1344     "uri" => ""
1345     ),
1346     "cosmos" => array(
1347     "icon" => "robot",
1348     "title" => "Cosmos",
1349     "rule" => array(
1350     "^cosmos" => ""
1351     ),
1352     "uri" => ""
1353     ),
1354     "cosmoty" => array(
1355     "icon" => "robot",
1356     "title" => "Cosmoty",
1357     "rule" => array(
1358     "^beautybot[/ ]([0-9.]{1,10})" => "\\1"
1359     ),
1360     "uri" => "http://www.uchoose.de/crawler/beautybot/"
1361     ),
1362     "covario" => array(
1363     "icon" => "covario",
1364     "title" => "Covario",
1365     "rule" => array(
1366     "Covario-IDS[/ ]([0-9.]{1,10})" => "\\1"
1367     ),
1368     "uri" => "http://www.covario.com/ids/"
1369     ),
1370     "creativecommons" => array(
1371     "icon" => "creativecommons",
1372     "title" => "Creative Commons",
1373     "rule" => array(
1374     "CreativeCommons[/ ]([0-9.]{1,6}(-dev)?)" => "\\1"
1375     ),
1376     "uri" => ""
1377     ),
1378     "cscrawler" => array(
1379     "icon" => "robot",
1380     "title" => "CsCrawler",
1381     "rule" => array(
1382     "CsCrawler" => ""
1383     ),
1384     "uri" => "http://www.kde.cs.uni-kassel.de/lehre/ss2005/googlespam/crawler.html"
1385     ),
1386 joku 111 "jigsaw" => array(
1387     "icon" => "jigsaw",
1388     "title" => "Jigsaw",
1389     "rule" => array(
1390     "Jigsaw[ /]([0-9.]{1,10})" => "\\1"
1391     ),
1392     "uri" => "http://jigsaw.w3.org/css-validator/"
1393     ),
1394 joku 63 "csscheck" => array(
1395     "icon" => "w3c",
1396     "title" => "W3C CSS Validator",
1397     "rule" => array(
1398     "CSS(Check|_Validator).+[ /]([0-9.]{1,10})" => "\\2"
1399     ),
1400     "uri" => "http://jigsaw.w3.org/css-validator/"
1401     ),
1402     "custo" => array(
1403     "icon" => "robot",
1404     "title" => "Custo",
1405     "rule" => array(
1406     "Custo[ /]([0-9.]{1,10})" => "\\1"
1407     ),
1408     "uri" => ""
1409     ),
1410     "cybernavi" => array(
1411     "icon" => "robot",
1412     "title" => "CyberNavi",
1413     "rule" => array(
1414     "CyberNavi_WebGet[ /]([0-9.]{1,10})" => "\\1"
1415     ),
1416     "uri" => ""
1417     ),
1418     "cyberpatrol" => array(
1419     "icon" => "cyberpatrol",
1420     "title" => "CyberPatrol",
1421     "rule" => array(
1422     "^CyberPatrol" => ""
1423     ),
1424     "uri" => "http://www.cyberpatrol.com/cyberpatrolcrawler.asp"
1425     ),
1426     "cyberz" => array(
1427     "icon" => "cyberz",
1428     "title" => "Cyberz",
1429     "rule" => array(
1430     "Cyberz Communication Agent" => ""
1431     ),
1432     "uri" => ""
1433     ),
1434     "cydral" => array(
1435     "icon" => "robot",
1436     "title" => "Cydral",
1437     "rule" => array(
1438     "CydralSpider[ /]([0-9.]{1,10})" => "\\1"
1439     ),
1440     "uri" => ""
1441     ),
1442     "cynthia" => array(
1443     "icon" => "cynthia",
1444     "title" => "Cynthia Says",
1445     "rule" => array(
1446     "Cynthia[ /]([0-9.]{1,10})" => "\\1"
1447     ),
1448     "uri" => ""
1449     ),
1450     "d4x" => array(
1451     "icon" => "d4x",
1452     "title" => "Downloader for X",
1453     "rule" => array(
1454     "Downloader for X[ /]([0-9.]{1,10})" => "\\1"
1455     ),
1456     "uri" => ""
1457     ),
1458     "da" => array(
1459     "icon" => "da",
1460     "title" => "DA",
1461     "rule" => array(
1462     "^DA[ /]([0-9.]{1,10})" => "\\1"
1463     ),
1464     "uri" => ""
1465     ),
1466     "daum" => array(
1467     "icon" => "daum",
1468     "title" => "DAUM",
1469     "rule" => array(
1470     "DAUMOA[ /]([0-9.]{1,10})" => "\\1",
1471 joku 379 "Daum[ /]([0-9.]{1,10})" => "\\1",
1472 joku 63 "DAUM Web Robot" => "",
1473 joku 339 "daum.net" => "",
1474 joku 63 "Daum Communications Corp" => "",
1475     "EDI[ /]([0-9.]{1,10})" => "\\1",
1476     "Edacious.*Intelligent Web Robot" => "",
1477     "RaBot[/ ]([0-9.]{1,10}) Agent" => "\\1"
1478     ),
1479     "uri" => "http://ws.daum.net/"
1480     ),
1481     "daypop" => array(
1482     "icon" => "robot",
1483     "title" => "Daypop",
1484     "rule" => array(
1485     "daypopbot[/ ]([0-9.]{1,10})" => "\\1"
1486     ),
1487     "uri" => ""
1488     ),
1489     "delfi" => array(
1490     "icon" => "delfi",
1491     "title" => "Delfi",
1492     "rule" => array(
1493     "crawl at delfi dot lt" => ""
1494     ),
1495     "uri" => ""
1496     ),
1497     "depspid" => array(
1498     "icon" => "depspid",
1499     "title" => "DepSpid",
1500     "rule" => array(
1501     "DepSpid[/ ]([0-9.]{1,10})" => "\\1"
1502     ),
1503     "uri" => ""
1504     ),
1505     "devonagent" => array(
1506     "icon" => "devonagent",
1507     "title" => "DEVONagent",
1508     "rule" => array(
1509     "DEVONtech" => ""
1510     ),
1511     "uri" => ""
1512     ),
1513     "diffbot" => array(
1514     "icon" => "diffbot",
1515     "title" => "Diffbot",
1516     "rule" => array(
1517     " Diffbot" => ""
1518     ),
1519     "uri" => "http://www.diffbot.com"
1520     ),
1521     "diribot" => array(
1522     "icon" => "diribot",
1523     "title" => "Diri",
1524     "rule" => array(
1525     "diribot" => ""
1526     ),
1527     "uri" => "http://www.diri.bg"
1528     ),
1529     "directhit" => array(
1530     "icon" => "robot",
1531     "title" => "Direct Hit",
1532     "rule" => array(
1533     "EZResult -- Internet Search Engine" => ""
1534     ),
1535     "uri" => "http://www.directhit.com"
1536     ),
1537     "disco" => array(
1538     "icon" => "disco",
1539 matthys 209 "title" => "Disco",
1540 joku 63 "rule" => array(
1541     "disco/Nutch[/ -]([0-9.]{1,10})" => "\\1",
1542     "discobot[/ -]([0-9.]{1,10})" => "\\1",
1543     "disco-crawl@discoveryengine.com" => ""
1544     ),
1545     "uri" => "http://www.discoveryengine.com/"
1546     ),
1547     "discopump" => array(
1548     "icon" => "robot",
1549     "title" => "DISCo Pump",
1550     "rule" => array(
1551     "DISCo Pump[/ ]([0-9.]{1,10})" => "\\1"
1552     ),
1553     "uri" => ""
1554     ),
1555     "dkimreputation" => array(
1556     "icon" => "dkimreputation",
1557     "title" => "DKIM Reputation",
1558     "rule" => array(
1559     "DKIMRepBot[ /]([0-9.]{1,10})" => "\\1"
1560     ),
1561     "uri" => "http://www.dkim-reputation.org"
1562     ),
1563     "dnsdigger" => array(
1564     "icon" => "dnsdigger",
1565     "title" => "DNS-Digger",
1566     "rule" => array(
1567     "DNS-Digger-Explorer[ /]([0-9.]{1,10})" => "\\1"
1568     ),
1569     "uri" => "http://www.dnsdigger.com"
1570     ),
1571     "dnsvalidation" => array(
1572     "icon" => "dnsvalidation",
1573     "title" => "DNS Validation",
1574     "rule" => array(
1575     "DNSValidation[ /]([0-9.]{1,10})" => "\\1"
1576     ),
1577     "uri" => "http://www.dnsvalidation.com/"
1578     ),
1579     "doctorhtml" => array(
1580     "icon" => "robot",
1581     "title" => "DoctorHTML",
1582     "rule" => array(
1583     "Doctor[ \-]?HTML" => ""
1584     ),
1585     "uri" => ""
1586     ),
1587     "dolphin" => array(
1588     "icon" => "dolphin",
1589     "title" => "Dolphin",
1590     "rule" => array(
1591     "Dolphin[ /]([0-9.]{1,10})" => "\\1"
1592     ),
1593     "uri" => "http://telehouse.ru/crawler.html"
1594     ),
1595     "domaindatei" => array(
1596     "icon" => "robot",
1597     "title" => "Domaindatei",
1598     "rule" => array(
1599     "DomaindateiSpider[ /]([0-9.]{1,10})" => "\\1"
1600     ),
1601     "uri" => ""
1602     ),
1603     "dontbuylists" => array(
1604     "icon" => "dontbuylists",
1605     "title" => "Dontbuylists",
1606     "rule" => array(
1607     "DBLBot[ /]([0-9.]{1,10})" => "\\1"
1608     ),
1609     "uri" => "http://www.dontbuylists.com"
1610     ),
1611     "dotbot" => array(
1612     "icon" => "robot",
1613     "title" => "DotBot",
1614     "rule" => array(
1615     "DotBot[ /]([0-9.]{1,10})" => "\\1"
1616     ),
1617     "uri" => "http://www.dotnetdotcom.org"
1618     ),
1619     "doweb" => array(
1620     "icon" => "doweb",
1621     "title" => "DoWeb",
1622     "rule" => array(
1623     "^www.doweb.co.uk" => ""
1624     ),
1625     "uri" => "http://www.doweb.co.uk/"
1626     ),
1627     "downforeveryoneorjustme" => array(
1628     "icon" => "downforeveryoneorjustme",
1629     "title" => "Down for everyone or just me",
1630     "rule" => array(
1631     "downforeveryoneorjustme" => ""
1632     ),
1633     "uri" => "http://www.downforeveryoneorjustme.com"
1634     ),
1635     "downloadninja" => array(
1636     "icon" => "robot",
1637     "title" => "Download Ninja",
1638     "rule" => array(
1639     "Download Ninja[ /]([0-9.]{1,10})" => "\\1"
1640     ),
1641     "uri" => ""
1642     ),
1643     "drupal" => array(
1644     "icon" => "drupal",
1645     "title" => "Drupal",
1646     "rule" => array(
1647     "^Drupal" => ""
1648     ),
1649     "uri" => ""
1650     ),
1651     "dsns" => array(
1652     "icon" => "robot",
1653     "title" => "DSNS Scanner",
1654     "rule" => array(
1655     "^DSNS" => ""
1656     ),
1657     "uri" => ""
1658     ),
1659     "dtsagent" => array(
1660     "icon" => "robot",
1661     "title" => "DTS Agent",
1662     "rule" => array(
1663     "DTS Agent" => ""
1664     ),
1665     "uri" => ""
1666     ),
1667 matthys 342 "duckduckgo" => array(
1668     "icon" => "duckduckgo",
1669     "title" => "DuckDuckGo",
1670     "rule" => array(
1671 matthys 397 "DuckDuckGo-Favicons-Bot[ /]([0-9.]{1,10})" => "\\1",
1672     "DuckDuckBot-Https[ /]([0-9.]{1,10})" => "\\1"
1673 matthys 342 ),
1674     "uri" => "http://duckduckgo.com"
1675     ),
1676 joku 63 "earthcom" => array(
1677     "icon" => "earthcom",
1678     "title" => "Earthcom",
1679     "rule" => array(
1680     "EARTHCOM\.info[/ ]([0-9a-z.]{1,10})" => "\\1"
1681     ),
1682     "uri" => ""
1683     ),
1684     "ebay" => array(
1685     "icon" => "ebay",
1686     "title" => "eBay",
1687     "rule" => array(
1688     "eBay Relevance Ad Crawler" => ""
1689     ),
1690     "uri" => ""
1691     ),
1692     "echo" => array(
1693     "icon" => "robot",
1694     "title" => "Echo.com",
1695     "rule" => array(
1696     "_TrueRobot[/ ]([0-9.]{1,10})" => "\\1"
1697     ),
1698     "uri" => "http://www.echo.com/"
1699     ),
1700     "eert" => array(
1701     "icon" => "robot",
1702 matthys 209 "title" => "Eert",
1703 joku 63 "rule" => array(
1704     "eert spdr[/ ]([0-9.]{1,10})" => "\\1"
1705     ),
1706     "uri" => "http://bot.eert.net"
1707     ),
1708     "eknip" => array(
1709     "icon" => "eknip",
1710     "title" => "E-Knip",
1711     "rule" => array(
1712     "eknip[ /]([0-9a-z.]{1,10})" => "\\1"
1713     ),
1714     "uri" => ""
1715     ),
1716     "zoominfo" => array(
1717     "icon" => "zoominfo",
1718 matthys 209 "title" => "ZoomInfo", //Previous called Eliyon
1719 joku 63 "rule" => array(
1720     "NextGenSearchBot[/ ]([0-9.]{1,10})" => "\\1"
1721     ),
1722     "uri" => "http://www.zoominfo.com/business/nextgensearchbot.html"
1723     ),
1724     "emeraldshield" => array(
1725     "icon" => "emeraldshield",
1726     "title" => "EmeraldShield",
1727     "rule" => array(
1728     "^EmeraldShield" => ""
1729     ),
1730     "uri" => ""
1731     ),
1732     "empas" => array(
1733     "icon" => "empas",
1734     "title" => "Empas",
1735     "rule" => array(
1736     "DigExt; empas\)$" => "",
1737     "^EMPAS[_\-]ROBOT" => ""
1738     ),
1739     "uri" => ""
1740     ),
1741     "entireweb" => array(
1742     "icon" => "entireweb",
1743     "title" => "Entireweb",
1744     "rule" => array(
1745     "Speedy[ ]?Spider" => ""
1746     ),
1747     "uri" => ""
1748     ),
1749     "envolk" => array(
1750     "icon" => "envolk",
1751     "title" => "Envolk",
1752     "rule" => array(
1753     "envolk\[ITS\]spider[/ ]([0-9.]{1,10})" => "\\1",
1754     "envolk[/ ]([0-9.]{1,10})" => "\\1"
1755     ),
1756     "uri" => ""
1757     ),
1758     "esnet" => array(
1759     "icon" => "robot",
1760     "title" => "ES.NET",
1761     "rule" => array(
1762     "ES.NET Crawler[ /]([0-9.]{1,10})" => "\\1"
1763     ),
1764     "uri" => ""
1765     ),
1766     "estyle" => array(
1767     "icon" => "estyle",
1768     "title" => "eStyle Search",
1769     "rule" => array(
1770     "eStyleSearch[ /]([0-9.]{1,10})" => "\\1"
1771     ),
1772     "uri" => ""
1773     ),
1774     "eurip" => array(
1775     "icon" => "robot",
1776     "title" => "Eurip",
1777     "rule" => array(
1778     "EuripBot[ /]([0-9.]{1,10})" => "\\1"
1779     ),
1780     "uri" => "http://www.eurip.com"
1781     ),
1782     "eurodirectory" => array(
1783     "icon" => "robot",
1784     "title" => "Euro Directory",
1785     "rule" => array(
1786     "www\.euro-directory\.com" => ""
1787     ),
1788     "uri" => "http://www.euro-directory.com/"
1789     ),
1790     "euroseek" => array(
1791     "icon" => "euroseek",
1792     "title" => "EuroSeek",
1793     "rule" => array(
1794     "Arachnoidea" => ""
1795     ),
1796     "uri" => ""
1797     ),
1798     "evaal" => array(
1799     "icon" => "evaal",
1800     "title" => "Evaal",
1801     "rule" => array(
1802     "^EvaalSE" => ""
1803     ),
1804     "uri" => "http://www.evaal.com/"
1805     ),
1806     "eventax" => array(
1807     "icon" => "eventax",
1808     "title" => "Eventax",
1809     "rule" => array(
1810     "^eventax[ /]([0-9.]{1,10})" => "\\1"
1811     ),
1812     "uri" => "http://www.eventax.de/"
1813     ),
1814     "everbee" => array(
1815     "icon" => "everbee",
1816     "title" => "Everbee",
1817     "rule" => array(
1818     "EverbeeCrawler" => ""
1819     ),
1820     "uri" => ""
1821     ),
1822     "everest" => array(
1823     "icon" => "everest",
1824     "title" => "Everest",
1825     "rule" => array(
1826     "Everest-Vulcan Inc.[ /]([0-9.]{1,10})" => "\\1"
1827     ),
1828     "uri" => ""
1829     ),
1830     "exabot" => array(
1831     "icon" => "exabot",
1832     "title" => "ExaBot",
1833     "rule" => array(
1834     "^NG[ /]([0-9.]{1,10})" => "\\1",
1835     "Exabot/([0-9.]{1,10})" => "\\1",
1836     "ExaBotTest/([0-9.]{1,10})" => "\\1",
1837     "ExaBot-(Test|Images)/([0-9.]{1,10})" => "\\2",
1838     "ExaBot-Thumbnails" => ""
1839     ),
1840     "uri" => "http://www.exabot.com/go/robot"
1841     ),
1842     "exactseek" => array(
1843     "icon" => "exactseek",
1844     "title" => "ExactSeek",
1845     "rule" => array(
1846     "^exactseek[ \-]?(pagereaper|crawler)[ \-]?([0-9.]{1,10})" => "\\2",
1847     "ExactSeek[ \.\-]?(Crawler|com)" => ""
1848     ),
1849     "uri" => "http://www.exactseek.com/"
1850     ),
1851     "excite" => array(
1852     "icon" => "excite",
1853     "title" => "Excite",
1854     "rule" => array(
1855     "Architext[ \-]?Spider" => ""
1856     ),
1857     "uri" => ""
1858     ),
1859     "execrawl" => array(
1860     "icon" => "robot",
1861     "title" => "Execrawl",
1862     "rule" => array(
1863     "Execrawl[ /]([0-9.]{1,10})" => "\\1",
1864     "Execrawl" => ""
1865     ),
1866     "uri" => ""
1867     ),
1868     "expertmonitor" => array(
1869     "icon" => "robot",
1870     "title" => "ExpertMonitor",
1871     "rule" => array(
1872     "^NetMonitor[ /]([0-9.]{1,10})" => "\\1"
1873     ),
1874     "uri" => ""
1875     ),
1876     "explorerrss" => array(
1877     "icon" => "robot",
1878     "title" => "Explorer RSS",
1879     "rule" => array(
1880     "^Windows-RSS-Platform[ /]([0-9.]{1,10})" => "\\1"
1881     ),
1882     "uri" => ""
1883     ),
1884     "ezooms" => array(
1885     "icon" => "robot",
1886     "title" => "Ezooms",
1887     "rule" => array(
1888     "ezooms[/ ]([0-9a-z.\-]{1,10})" => "\\1"
1889     ),
1890     "uri" => ""
1891     ),
1892     "facebook" => array(
1893     "icon" => "facebook",
1894     "title" => "Facebook",
1895     "rule" => array(
1896     "FacebookFeedParser[/ ]([0-9a-z.\-]{1,10})" => "\\1",
1897     "facebookexternalhit/([0-9.]{1,10})" => "\\1",
1898     "^facebook" => ""
1899     ),
1900     "uri" => "http://www.facebook.com/"
1901     ),
1902     "fast" => array(
1903     "icon" => "fast",
1904     "title" => "Fast",
1905     "rule" => array(
1906     "^FAST( Enterprise |-Web| MetaWeb )?Crawler[ /]([0-9.]{1,10})" => "\\2",
1907     "^FAST( Enterprise |-Web| MetaWeb | PartnerSite )?Crawler" => "",
1908     "^Fast Crawler" => "",
1909     "^libwww-perl[ /]([0-9.]{1,10}) FP[ /]([0-9.]{1,10})" => "\\2"
1910     ),
1911     "uri" => "http://www.alltheweb.com/"
1912     ),
1913     "fastbuzz" => array(
1914     "icon" => "fastbuzz",
1915     "title" => "Fastbuzz",
1916     "rule" => array(
1917     "^fastbuzz\.com" => ""
1918     ),
1919     "uri" => ""
1920     ),
1921     "favorg" => array(
1922     "icon" => "robot",
1923     "title" => "FavOrg",
1924     "rule" => array(
1925     "^FavOrg" => ""
1926     ),
1927     "uri" => ""
1928     ),
1929     "favorstar" => array(
1930     "icon" => "favorstar",
1931 matthys 209 "title" => "Favorstar",
1932 joku 63 "rule" => array(
1933     "favorstarbot[ /]([0-9.]{1,10})" => "\\1"
1934     ),
1935     "uri" => "http://favorstar.com/bot.html"
1936     ),
1937     "faxo" => array(
1938     "icon" => "robot",
1939     "title" => "Faxo",
1940     "rule" => array(
1941     "^Faxobot[ /]([0-9.]{1,10})" => "\\1"
1942     ),
1943     "uri" => "http://www.faxo.com/"
1944     ),
1945     "fdse" => array(
1946     "icon" => "robot",
1947     "title" => "FDSE Robot",
1948     "rule" => array(
1949     "FDSE[ \-]?robot" => ""
1950     ),
1951     "uri" => ""
1952     ),
1953     "feedback" => array(
1954     "icon" => "robot",
1955     "title" => "FeedBack",
1956     "rule" => array(
1957     "FeedBack[/ ]([0-9.]{1,10})" => "\\1"
1958     ),
1959     "uri" => ""
1960     ),
1961     "feedburner" => array(
1962     "icon" => "feedburner",
1963     "title" => "FeedBurner",
1964     "rule" => array(
1965     "^FeedBurner[/ ]([0-9.]{1,10})" => "\\1"
1966     ),
1967     "uri" => ""
1968     ),
1969     "feeddemon" => array(
1970     "icon" => "feeddemon",
1971     "title" => "FeedDemon",
1972     "rule" => array(
1973     "FeedDemon[/ ]([0-9.]{1,10})" => "\\1"
1974     ),
1975     "uri" => ""
1976     ),
1977     "feedfind" => array(
1978     "icon" => "feedfind",
1979     "title" => "FeedFind",
1980     "rule" => array(
1981     "Feed::Find[ /]([0-9.]{1,10})" => "\\1"
1982     ),
1983     "uri" => ""
1984     ),
1985     "feedonfeeds" => array(
1986     "icon" => "robot",
1987     "title" => "Feed On Feeds",
1988     "rule" => array(
1989     "FeedOnFeeds[/ ]([0-9.]{1,10})" => "\\1"
1990     ),
1991     "uri" => ""
1992     ),
1993     "feedparser" => array(
1994     "icon" => "robot",
1995     "title" => "Feedparser",
1996     "rule" => array(
1997     "UniversalFeedParser[/ ]([0-9a-z.\-]{1,10})" => "\\1",
1998     "FeedParser" => ""
1999     ),
2000     "uri" => ""
2001     ),
2002     "feedreader" => array(
2003     "icon" => "feedreader",
2004     "title" => "Feedreader",
2005     "rule" => array(
2006     "^Feedreader" => ""
2007     ),
2008     "uri" => ""
2009     ),
2010     "feedserver" => array(
2011     "icon" => "robot",
2012     "title" => "FeedServer",
2013     "rule" => array(
2014     "FeedServer[/ ]([0-9.]{1,10})" => "\\1"
2015     ),
2016     "uri" => ""
2017     ),
2018     "feedster" => array(
2019     "icon" => "feedster",
2020     "title" => "Feedster",
2021     "rule" => array(
2022     "Feedster Crawler[/ ]([0-9.]{1,10})" => "\\1"
2023     ),
2024     "uri" => ""
2025     ),
2026     "feedvalidator" => array(
2027     "icon" => "feedvalidator",
2028     "title" => "Feed Validator",
2029     "rule" => array(
2030     "^FeedValidator[/ ]([0-9.]{1,10})" => "\\1"
2031     ),
2032     "uri" => ""
2033     ),
2034     "fdm" => array(
2035     "icon" => "robot",
2036     "title" => "Free Download Manager",
2037     "rule" => array(
2038     "^FDM[/ ]([0-9a-z.]{1,10})" => "\\1"
2039     ),
2040     "uri" => ""
2041     ),
2042     "filangy" => array(
2043     "icon" => "robot",
2044     "title" => "Filangy",
2045     "rule" => array(
2046     "Filangy[/ ]([0-9.]{1,10})" => "\\1"
2047     ),
2048     "uri" => "http://www.filangy.com/"
2049     ),
2050     "findanisp" => array(
2051     "icon" => "robot",
2052     "title" => "FindAnISP",
2053     "rule" => array(
2054     "FindAnISP" => ""
2055     ),
2056     "uri" => "http://www.findanisp.com/"
2057     ),
2058     "findengines" => array(
2059     "icon" => "findengines",
2060     "title" => "FindEngines",
2061     "rule" => array(
2062     "FindEngines! Bot" => ""
2063     ),
2064     "uri" => ""
2065     ),
2066     "findexa" => array(
2067     "icon" => "findexa",
2068     "title" => "Findexa",
2069     "rule" => array(
2070     "Findexa Crawler" => ""
2071     ),
2072     "uri" => ""
2073     ),
2074     "findlinks" => array(
2075     "icon" => "findlinks",
2076     "title" => "FindLinks",
2077     "rule" => array(
2078     "findlinks[ /]([0-9.]{1,10})" => "\\1",
2079     "^FindLinks" => ""
2080     ),
2081     "uri" => ""
2082     ),
2083     "findoor" => array(
2084     "icon" => "findoor",
2085 matthys 209 "title" => "Findoor",
2086 joku 63 "rule" => array(
2087     "^findoor(-Bot)?" => "\\1"
2088     ),
2089     "uri" => ""
2090     ),
2091     "firefly" => array(
2092     "icon" => "firefly",
2093     "title" => "Firefly",
2094     "rule" => array(
2095     "Firefly" => ""
2096     ),
2097     "uri" => ""
2098     ),
2099     "flashget" => array(
2100     "icon" => "flashget",
2101     "title" => "FlashGet",
2102     "rule" => array(
2103     "^FlashGet" => ""
2104     ),
2105     "uri" => ""
2106     ),
2107     "flatland" => array(
2108     "icon" => "flatland",
2109     "title" => "Flatland",
2110     "rule" => array(
2111     "^flatlandbot" => ""
2112     ),
2113     "uri" => "http://www.flatlandindustries.com/flatlandbot"
2114     ),
2115     "flickbot" => array(
2116     "icon" => "flickbot",
2117     "title" => "FlickBot",
2118     "rule" => array(
2119     "FlickBot[ /]([0-9.]{1,10})" => "\\1"
2120     ),
2121     "uri" => ""
2122     ),
2123     "flightdeckreports" => array(
2124     "icon" => "flightdeckreports",
2125     "title" => "Flight Deck Reports",
2126     "rule" => array(
2127     "FlightDeckReportsBot[ /]([0-9.]{1,10})" => "\\1"
2128     ),
2129     "uri" => "http://www.flightdeckreports.com/"
2130     ),
2131     "forex" => array(
2132     "icon" => "robot",
2133     "title" => "Forex",
2134     "rule" => array(
2135     "^Forex Trading Network Organization" => ""
2136     ),
2137     "uri" => "http://www.netforex.org/"
2138     ),
2139     "freshmeat" => array(
2140     "icon" => "freshmeat",
2141 matthys 209 "title" => "Freshmeat",
2142 joku 63 "rule" => array(
2143     "fmII URL validator[ /]([0-9.]{1,10})" => "\\1",
2144     "freshmeat.net URL validator[ /]([0-9.]{1,10})" => "\\1"
2145     ),
2146     "uri" => "http://www.freshmeat.net/"
2147     ),
2148     "friend" => array(
2149     "icon" => "friend",
2150     "title" => "Friend",
2151     "rule" => array(
2152     "www\.friend\.fr" => ""
2153     ),
2154     "uri" => "http://www.friend.fr/"
2155     ),
2156     "frontier" => array(
2157     "icon" => "frontier",
2158     "title" => "Frontier",
2159     "rule" => array(
2160     "Frontier[ /]([0-9.]{1,10})" => "\\1"
2161     ),
2162     "uri" => ""
2163     ),
2164     "fyberspider" => array(
2165     "icon" => "robot",
2166     "title" => "FyberSpider",
2167     "rule" => array(
2168     "FyberSpider[ /]([0-9.]{1,10})" => "\\1"
2169     ),
2170     "uri" => "http://www.fyberspider.com/"
2171     ),
2172     "gaisbot" => array(
2173     "icon" => "gais",
2174     "title" => "Gaisbot",
2175     "rule" => array(
2176     "Gaisbot[ /]([0-9.]{1,10})" => "\\1"
2177     ),
2178     "uri" => ""
2179     ),
2180     "galaxy" => array(
2181     "icon" => "galaxy",
2182     "title" => "Galaxy",
2183     "rule" => array(
2184     "GalaxyBot[ /]([0-9.]{1,10})" => "\\1",
2185     "www.galaxy.com" => ""
2186     ),
2187     "uri" => "http://www.galaxy.com/"
2188     ),
2189     "gamespy" => array(
2190     "icon" => "gamespy",
2191     "title" => "GameSpy",
2192     "rule" => array(
2193     "GameSpyHTTP[ /]([0-9.]{1,10})" => "\\1"
2194     ),
2195     "uri" => ""
2196     ),
2197     "garlik" => array(
2198     "icon" => "garlik",
2199     "title" => "Garlik",
2200     "rule" => array(
2201     "GarlikCrawler[ /]([0-9.]{1,10})" => "\\1"
2202     ),
2203     "uri" => "http://garlik.com/"
2204     ),
2205     "gdesktop" => array(
2206     "icon" => "gdesktop",
2207     "title" => "Google Desktop",
2208     "rule" => array(
2209     "compatible; Google Desktop" => ""
2210     ),
2211     "uri" => ""
2212     ),
2213     "genome" => array(
2214     "icon" => "robot",
2215     "title" => "Genome Machine",
2216     "rule" => array(
2217     "Genome[ \-]?Machine" => ""
2218     ),
2219     "uri" => ""
2220     ),
2221     "geohash" => array(
2222     "icon" => "geohash",
2223     "title" => "Geohash",
2224     "rule" => array(
2225     "GeoHasher(/Nutch)[ /-]([0-9.]{1,10})" => "\\2"
2226     ),
2227     "uri" => "http://geohash.org/"
2228     ),
2229     "geona" => array(
2230     "icon" => "robot",
2231     "title" => "Geona",
2232     "rule" => array(
2233     "GeonaBot[ /]([0-9.]{1,10})" => "\\1"
2234     ),
2235     "uri" => ""
2236     ),
2237     "geoblog" => array(
2238     "icon" => "robot",
2239     "title" => "The World as a Blog",
2240     "rule" => array(
2241     "The World as a Blog" => ""
2242     ),
2243     "uri" => ""
2244     ),
2245     "geourl" => array(
2246     "icon" => "geourl",
2247     "title" => "GeoUrl",
2248     "rule" => array(
2249     "geourl[ /]([0-9.]{1,10})" => "\\1",
2250     "^GeoURLBot[ /]([0-9.]{1,10})" => "\\1",
2251     ),
2252     "uri" => ""
2253     ),
2254     "getnetwise" => array(
2255     "icon" => "robot",
2256     "title" => "GetNetWise",
2257     "rule" => array(
2258     " Crayon Crawler" => ""
2259     ),
2260     "uri" => ""
2261     ),
2262     "getright" => array(
2263     "icon" => "getright",
2264     "title" => "GetRight",
2265     "rule" => array(
2266     "GetRight[ /]([0-9.]{1,10})" => "\\1"
2267     ),
2268     "uri" => ""
2269     ),
2270     "getsmart" => array(
2271     "icon" => "getsmart",
2272     "title" => "GetSmart",
2273     "rule" => array(
2274     "GetSmart[ /]([0-9.]{1,10})" => "\\1"
2275     ),
2276     "uri" => ""
2277     ),
2278     "gigabot" => array(
2279     "icon" => "gigablast",
2280     "title" => "Gigablast",
2281     "rule" => array(
2282     "(Gigabot|Sitesearch)[/ ]([0-9.]{1,10})" => "\\2",
2283     "GigabotSiteSearch[/ ]([0-9.]{1,10})" => "\\1",
2284 joku 336 "GigablastOpenSource" => ""
2285 joku 63 ),
2286     "uri" => ""
2287 matthys 328 ),
2288     "gimme" => array(
2289     "icon" => "gimme",
2290     "title" => "gimmeUSA",
2291     "rule" => array(
2292     "Gimme60bot[ /]([0-9.]{1,10})" => "\\1"
2293     ),
2294     "uri" => "http://gimme60.com"
2295 joku 63 ),
2296 matthys 328 "ginger" => array(
2297     "icon" => "ginger",
2298     "title" => "GingerCrawler",
2299     "rule" => array(
2300     "GingerCrawler[ /]([0-9.]{1,10})" => "\\1"
2301     ),
2302     "uri" => "http://www.gingersoftware.com/crawler_agent.htm"
2303     ),
2304 joku 63 "girafabot" => array(
2305     "icon" => "girafa",
2306     "title" => "Girafa",
2307     "rule" => array(
2308     "Girafabot" => ""
2309     ),
2310     "uri" => ""
2311     ),
2312     "globalspec" => array(
2313     "icon" => "globalspec",
2314     "title" => "GlobalSpec",
2315     "rule" => array(
2316     "Ocelli[ /]([0-9.]{1,10})" => "\\1"
2317     ),
2318     "uri" => ""
2319     ),
2320     "glucose" => array(
2321     "icon" => "glucose",
2322     "title" => "Glucose",
2323     "rule" => array(
2324     "glucose[ /]([0-9a-z.\-]{1,10})" => "\\1"
2325     ),
2326     "uri" => ""
2327     ),
2328     "gnomit" => array(
2329     "icon" => "gnomit",
2330     "title" => "Gnomit",
2331     "rule" => array(
2332     "gnomit[ /]([0-9.]{1,10})" => "\\1"
2333     ),
2334     "uri" => "http://www.gnomit.com/",
2335     ),
2336     "goforit" => array(
2337     "icon" => "goforit",
2338     "title" => "GoForIt",
2339     "rule" => array(
2340     "^GoForIt\.com" => "",
2341     "^GOFORITBOT" => ""
2342     ),
2343     "uri" => ""
2344     ),
2345     "goguides" => array(
2346     "icon" => "robot",
2347     "title" => "GoGuides",
2348     "rule" => array(
2349     "^GoGuidesBot[ /]([0-9.]{1,10})" => "\\1"
2350     ),
2351     "uri" => "http://www.goguides.org/"
2352     ),
2353     // Needs to be before google for correct detection
2354     "mediapartners" => array(
2355     "icon" => "google",
2356     "title" => "Mediapartners",
2357     "rule" => array (
2358     "Mediapartners-Google[ /]([0-9.]{1,10})" => "\\1",
2359     "Mediapartners-Google" => ""
2360     ),
2361     "uri" => ""
2362     ),
2363     "google" => array(
2364     "icon" => "google",
2365     "title" => "Google",
2366     "rule" => array (
2367     "Googl(e|ebot)(-Image)?/([0-9.]{1,10})" => "\\3",
2368     "Googl(e|ebot)(-Image)?/" => "",
2369     "^gsa-crawler" => "",
2370     "Google Web Preview" => ""
2371     ),
2372     "uri" => "http://www.google.com/"
2373     ),
2374     "googlesitemaps" => array(
2375     "icon" => "google",
2376     "title" => "Google-Sitemaps",
2377     "rule" => array (
2378     "Googl(e|ebot)(-Sitemaps)?/([0-9.]{1,10})" => "\\3",
2379     "GSiteCrawler[ /v]*([0-9.a-z]{1,10})" => "\\1",
2380     "Googl(e|ebot)(-Sitemaps)?/" => ""
2381     ),
2382     "uri" => ""
2383     ),
2384     "googlemobile" => array(
2385     "icon" => "google",
2386     "title" => "Google-Mobile",
2387     "rule" => array (
2388     "Googl(e|ebot)(-Mobile)?/([0-9.]{1,10})" => "\\3",
2389     "Googl(e|ebot)(-Mobile)?/" => ""
2390     ),
2391     "uri" => ""
2392     ),
2393     "googleads" => array(
2394     "icon" => "google",
2395     "title" => "Google-AdsBot",
2396     "rule" => array (
2397     "^AdsBot-Google" => "",
2398     ),
2399     "uri" => ""
2400     ),
2401     "googlefeeds" => array(
2402     "icon" => "google",
2403     "title" => "Google-Feedfetcher",
2404     "rule" => array (
2405     "^Feedfetcher-Google" => "",
2406     ),
2407     "uri" => ""
2408     ),
2409     "googlverification" => array(
2410     "icon" => "google",
2411     "title" => "Google-Site-Verification",
2412     "rule" => array (
2413     "Google-Site-Verification[ /]([0-9.]{1,10})" => "\\1"
2414     ),
2415     "uri" => ""
2416     ),
2417 matthys 207 // Needs to be AFTER Google detection
2418     "goo" => array(
2419     "icon" => "goo",
2420     "title" => "Goo",
2421     "rule" => array (
2422     "(gazz|ichiro|mog(et|imogi))[ /]([0-9.]{1,10})" => "\\3",
2423     "DoCoMo[ /]([0-9.]{1,10})" => "\\1"
2424     ),
2425     "uri" => ""
2426     ),
2427 joku 63 "goongee" => array(
2428     "icon" => "robot",
2429     "title" => "GoonGee",
2430     "rule" => array (
2431     "^Big Fish[ /]v?([0-9.]{1,10})" => "\\2"
2432     ),
2433     "uri" => "http://www.goongee.com/"
2434     ),
2435     "gpost" => array(
2436     "icon" => "gpost",
2437     "title" => "GPost",
2438     "rule" => array(
2439     "^GPostbot" => ""
2440     ),
2441     "uri" => ""
2442     ),
2443 matthys 309 "grapeshot" => array(
2444     "icon" => "grapeshot",
2445     "title" => "Grapeshot",
2446     "rule" => array(
2447     "GrapeshotCrawler[/ ]([0-9.]{1,10})" => "\\1"
2448     ),
2449     "uri" => "http://www.grapeshot.co.uk/crawler.php"
2450     ),
2451 joku 63 "gregarius" => array(
2452     "icon" => "robot",
2453     "title" => "Gregarius",
2454     "rule" => array(
2455     "^Gregarius[/ ]([0-9.]{1,10})" => ""
2456     ),
2457     "uri" => ""
2458     ),
2459     "grub" => array(
2460     "icon" => "grub",
2461     "title" => "Grub",
2462     "rule" => array(
2463     "grub[ \-]?client[ /\-]{1,5}([0-9.]{1,10})" => "\\1",
2464     "grub crawler" => "",
2465     "grub[/ ]([0-9.]{1,10})" => "\\1",
2466     "^GrubNG([0-9. ]{1,10})?" => "\\1"
2467     ),
2468     "uri" => "http://www.grub.org/"
2469     ),
2470     "gulliver" => array(
2471     "icon" => "robot",
2472     "title" => "Gulliver",
2473     "rule" => array(
2474     "Gulliver" => ""
2475     ),
2476     "uri" => ""
2477     ),
2478     "guruji" => array(
2479     "icon" => "guruji",
2480     "title" => "Guruji",
2481     "rule" => array(
2482     "GurujiBot[/ ]([0-9.]{1,10})" => "\\1"
2483     ),
2484     "uri" => "http://www.guruji.com/"
2485     ),
2486     "gush" => array(
2487     "icon" => "robot",
2488     "title" => "Gush",
2489     "rule" => array(
2490     "^Gush[/ ]([0-9.]{1,10})" => "\\1"
2491     ),
2492     "uri" => ""
2493     ),
2494     "gziptester" => array(
2495     "icon" => "robot",
2496     "title" => "Gzip Tester",
2497     "rule" => array(
2498     "g(id)?zip[ \-]?test(er)?" => ""
2499     ),
2500     "uri" => ""
2501     ),
2502     "hanzoweb" => array(
2503     "icon" => "hanzoweb",
2504     "title" => "Hanzoweb",
2505     "rule" => array(
2506     "^Hanzoweb" => ""
2507     ),
2508     "uri" => ""
2509     ),
2510     "harbot" => array(
2511     "icon" => "harbot",
2512     "title" => "Harbot",
2513     "rule" => array(
2514     "^Harbot GateStation" => ""
2515     ),
2516     "uri" => ""
2517     ),
2518     "hatena" => array(
2519     "icon" => "hatena",
2520     "title" => "Hatena",
2521     "rule" => array(
2522     "Hatena (Antenna|Bookmark|Pagetitle Agent)[ /]([0-9.]{1,10})" => "\\2"
2523     ),
2524     "uri" => ""
2525     ),
2526     "htmlparser" => array(
2527     "icon" => "htmlparser",
2528     "title" => "HTMLParser",
2529     "rule" => array(
2530     "HTMLParser[ /]([0-9.]{1,10})" => "\\1"
2531     ),
2532     "uri" => "http://htmlparser.sourceforge.net/"
2533     ),
2534     "helix" => array(
2535     "icon" => "helix",
2536     "title" => "Heritrix",
2537     "rule" => array(
2538     "^helix[ /]([0-9.]{1,10})" => "\\1"
2539     ),
2540     "uri" => "http://www.sitesearch.ca/helix/"
2541     ),
2542     "heritrix" => array(
2543     "icon" => "heritrix",
2544     "title" => "Heritrix",
2545     "rule" => array(
2546     "heritrix[ /]([0-9.]{1,10})" => "\\1",
2547     "archive.org_bot" => "",
2548     "InternetArchive[ /]([0-9.a-z]{1,10})" => "\\1"
2549     ),
2550     "uri" => "http://archive.org"
2551     ),
2552     "hiddenmarket" => array(
2553     "icon" => "robot",
2554     "title" => "HiddenMarket",
2555     "rule" => array(
2556     "HiddenMarket[ /\-]([0-9.]{1,10})" => "\\1"
2557     ),
2558     "uri" => ""
2559     ),
2560     "honda" => array(
2561     "icon" => "robot",
2562     "title" => "Honda",
2563     "rule" => array(
2564     "Honda-Search[ /]([0-9.]{1,10})" => "\\1"
2565     ),
2566     "uri" => "http://www.honda-search.com"
2567     ),
2568     "hoowwwer" => array(
2569     "icon" => "hoowwwer",
2570     "title" => "HooWWWer",
2571     "rule" => array(
2572     "HooWWWer[ /]([0-9.]{1,10})" => "\\1"
2573     ),
2574     "uri" => ""
2575     ),
2576     "hotzonu" => array(
2577     "icon" => "hotzonu",
2578     "title" => "Hotzonu",
2579     "rule" => array(
2580     "Hotzonu[ /]([0-9.]{1,10})" => "\\1"
2581     ),
2582     "uri" => ""
2583     ),
2584     "houxou" => array(
2585     "icon" => "robot",
2586     "title" => "Houxou",
2587     "rule" => array(
2588     "HouxouCrawler[ /]Nutch.([0-9.]{1,10})" => "\\1",
2589     "HouxouCrawler" => ""
2590     ),
2591     "uri" => ""
2592     ),
2593     "htdig" => array(
2594     "icon" => "htdig",
2595     "title" => "ht://Dig",
2596     "rule" => array(
2597     "htdig[ /]([0-9.]{1,10})" => "\\1",
2598     "htdig" => "\\1"
2599     ),
2600     "uri" => ""
2601     ),
2602     "html2jpg" => array(
2603     "icon" => "html2jpg",
2604     "title" => "HTML2JPG",
2605     "rule" => array(
2606     "^HTML2JPG" => ""
2607     ),
2608     "uri" => ""
2609     ),
2610     "httperf" => array(
2611     "icon" => "robot",
2612     "title" => "HTTPerf",
2613     "rule" => array(
2614     "httperf[ /]([0-9.]{1,10})" => "\\1"
2615     ),
2616     "uri" => ""
2617     ),
2618     "httpunit" => array(
2619     "icon" => "httpunit",
2620     "title" => "HttpUnit",
2621     "rule" => array(
2622     "httpunit[ /]([0-9.]{1,10})" => "\\1"
2623     ),
2624     "uri" => ""
2625     ),
2626     "httrack" => array(
2627     "icon" => "httrack",
2628     "title" => "HTTrack",
2629     "rule" => array(
2630     "HTTrack[ /]([0-9.]{1,10})" => "\\1"
2631     ),
2632     "uri" => ""
2633     ),
2634     "huaweisymantec" => array(
2635     "icon" => "huaweisymantec",
2636     "title" => "HuaweiSymantec",
2637     "rule" => array(
2638     "HuaweiSymantecSpider[ /]([0-9.]{1,10})" => "\\1"
2639     ),
2640     "uri" => "http://www.huaweisymantec.com/en/IRL/spider/"
2641     ),
2642     "hungary" => array(
2643     "icon" => "hungary",
2644     "title" => "Hungary",
2645     "rule" => array(
2646     "HuRob[ /]([0-9.]{1,10})" => "\\1"
2647     ),
2648     "uri" => ""
2649     ),
2650 joku 261 "i18n" => array(
2651     "icon" => "w3c",
2652     "title" => "W3C I18n Checker",
2653     "rule" => array(
2654     "^W3C_I18n-Checker[ /]([0-9.]{1,10})" => "\\1"
2655 joku 169 ),
2656 joku 261 "uri" => "http://validator.w3.org/services"
2657     ),
2658 joku 63 "iakke" => array(
2659     "icon" => "iakke",
2660 matthys 209 "title" => "Iakké",
2661 joku 63 "rule" => array(
2662     "^iakke_bot[ /]v([0-9.]{1,10})" => "\\1"
2663     ),
2664     "uri" => "http://fr.iakke.com/"
2665     ),
2666     "iask" => array(
2667     "icon" => "iask",
2668     "title" => "IAsk",
2669     "rule" => array(
2670     "iaskspider[ /]([0-9.]{1,10})" => "\\1",
2671     "^iaskspider" => ""
2672     ),
2673     "uri" => "http://iask.com"
2674     ),
2675     "icc" => array(
2676     "icon" => "icc",
2677     "title" => "ICC-Crawler",
2678     "rule" => array(
2679     "^ICC-Crawler" => ""
2680     ),
2681     "uri" => "http://kc.nict.go.jp/icc/crawl.html"
2682     ),
2683     "icerocket" => array(
2684     "icon" => "icerocket",
2685     "title" => "Icerocket",
2686     "rule" => array(
2687     "BlogzIce[ /]([0-9.]{1,10})" => "\\1",
2688     "BlogSearch[ /]([0-9.]{1,10})" => "\\1"
2689     ),
2690     "uri" => ""
2691     ),
2692     "icjobs" => array(
2693     "icon" => "icjobs",
2694     "title" => "iCjobs",
2695     "rule" => array(
2696     "iCjobs[ /]([0-9.]{1,10})" => "\\1"
2697     ),
2698     "uri" => "http://www.icjobs.de/"
2699     ),
2700     "icra" => array(
2701     "icon" => "icra",
2702     "title" => "ICRA",
2703     "rule" => array(
2704     "^ICRA_Semantic_spider[ /]([0-9.]{1,10})" => "\\1"
2705     ),
2706     "uri" => "http://www.icra.org"
2707     ),
2708     "iknow" => array(
2709     "icon" => "robot",
2710     "title" => "I know",
2711     "rule" => array(
2712     "Comaneci_bot[ /]([0-9.]{1,10})" => "\\1"
2713     ),
2714     "uri" => ""
2715     ),
2716     "ilial" => array(
2717     "icon" => "robot",
2718     "title" => "Ilial",
2719     "rule" => array(
2720     "ilial[ /]Nutch[ \-]([0-9.]{1,10})" => "\\1"
2721     ),
2722     "uri" => ""
2723     ),
2724     "ilocal" => array(
2725     "icon" => "ilocal",
2726     "title" => "ilocal",
2727     "rule" => array(
2728     "^ilocal" => "",
2729     ),
2730     "uri" => "http://www.ilocal.nl"
2731     ),
2732     "ilse" => array(
2733     "icon" => "ilse",
2734     "title" => "Ilse",
2735     "rule" => array(
2736     "I(NGRID|lseRobot|lseBot)[ /]([0-9.]{1,10})" => "\\1"
2737     ),
2738     "uri" => "http://ilse.nl/"
2739     ),
2740     "iltrovatore" => array(
2741     "icon" => "iltrovatore",
2742     "title" => "IlTrovatore",
2743     "rule" => array(
2744     "iltrovatore-setaccio[ /]([0-9.]{1,10})" => "\\1",
2745     "Iltrovatore-Setaccio" => "",
2746     "iltrovatore[ /]([0-9.]{1,10})" => "\\1"
2747     ),
2748     "uri" => ""
2749     ),
2750     "indylibrary" => array(
2751     "icon" => "robot",
2752     "title" => "Indy Library",
2753     "rule" => array(
2754     "Indy[ \-]?Library" => ""
2755     ),
2756     "uri" => ""
2757     ),
2758     "inela" => array(
2759     "icon" => "robot",
2760     "title" => "Inela",
2761     "rule" => array(
2762     "InelaBot[ /]([0-9.]{1,10})" => ""
2763     ),
2764     "uri" => "http://inelegant.org/bot"
2765     ),
2766     "ineturl" => array(
2767     "icon" => "robot",
2768     "title" => "InetURL",
2769     "rule" => array(
2770     "InetURL.?[ /]([0-9.]{1,10})" => ""
2771     ),
2772     "uri" => ""
2773     ),
2774     "infoart" => array(
2775     "icon" => "robot",
2776     "title" => "InfoArt",
2777     "rule" => array(
2778     "InfoArt crawler" => ""
2779     ),
2780     "uri" => ""
2781     ),
2782     "infomine" => array(
2783     "icon" => "infomine",
2784     "title" => "INFOMINE",
2785     "rule" => array(
2786     "^DataFountains/DMOZ" => "",
2787     "^INFOMINE[ /]([0-9.]{1,10})" => "\\1"
2788     ),
2789     "uri" => "http://infomine.ucr.edu/"
2790     ),
2791     "infoseek" => array(
2792     "icon" => "infoseek",
2793     "title" => "Infoseek",
2794     "rule" => array(
2795     "SideWinder[ /]?([0-9a-z.]{1,10})" => "\\1",
2796     "Infoseek" => ""
2797     ),
2798     "uri" => ""
2799     ),
2800     "inktomi" => array(
2801     "icon" => "inktomi",
2802     "title" => "Inktomi",
2803     "rule" => array(
2804     "slurp@inktomi\.com" => ""
2805     ),
2806     "uri" => ""
2807     ),
2808     "innerprise" => array(
2809     "icon" => "robot",
2810     "title" => "Innerprise",
2811     "rule" => array(
2812     "^InnerpriseBot[ /]([0-9.]{1,10})" => "\\1",
2813     "URL[ _]Spider[ _]Pro[ /]([0-9.+]{1,10})" => "\\1",
2814     "^ES[ .]NET[ _]Crawler[ /]([0-9.]{1,10})" => "\\1"
2815     ),
2816     "uri" => "http://www.innerprise.com/"
2817     ),
2818     "inria" => array(
2819     "icon" => "inria",
2820     "title" => "Inria",
2821     "rule" => array(
2822     "^xyro_" => ""
2823     ),
2824     "uri" => ""
2825     ),
2826     "insitor" => array(
2827     "icon" => "insitor",
2828     "title" => "Insitor",
2829     "rule" => array(
2830     "^Insitor(,|\.|naut)" => ""
2831     ),
2832     "uri" => "http://www.insitor.com/"
2833     ),
2834     "internetninja" => array(
2835     "icon" => "robot",
2836     "title" => "Internet Ninja",
2837     "rule" => array(
2838     "^Internet Ninja[ /]([0-9.]{1,10})" => "\\1"
2839     ),
2840     "uri" => ""
2841     ),
2842     "internetseer" => array(
2843     "icon" => "internetseer",
2844     "title" => "InternetSeer",
2845     "rule" => array(
2846     "^InternetSeer\.com" => ""
2847     ),
2848     "uri" => ""
2849     ),
2850     "interseek" => array(
2851     "icon" => "interseek",
2852     "title" => "Interseek",
2853     "rule" => array(
2854     "Interseek.com" => ""
2855     ),
2856     "uri" => "http://www.interseek.com"
2857     ),
2858     "intravnews" => array(
2859     "icon" => "intravnews",
2860     "title" => "IntraVnews",
2861     "rule" => array(
2862     "IntraVnews[ /]([0-9.]{1,10})" => "\\1"
2863     ),
2864     "uri" => "http://www.intravnews.com/"
2865     ),
2866     "ip2location" => array(
2867     "icon" => "ip2location",
2868     "title" => "IP2LocationBot",
2869     "rule" => array(
2870     "^IP2(Map|Location)Bot[ /]([0-9.]{1,10})" => "\\2"
2871     ),
2872     "uri" => "http://www.ip2location.com"
2873     ),
2874     "ipworks" => array(
2875     "icon" => "ipworks",
2876     "title" => "IP*Works",
2877     "rule" => array(
2878     "^IP\*Works\! V([0-9.]{1,10})" => "\\1"
2879     ),
2880     "uri" => "http://www.nsoftware.com/ipworks/"
2881     ),
2882     "irca" => array(
2883     "icon" => "robot",
2884     "title" => "Novell iChain Cool Solutions caching",
2885     "rule" => array(
2886 matthys 209 "^ICRA_(label_generator|Semantic_spider)[ /]([0-9.]{1,10})" => "\\2",
2887     "^Mozilla[/ ]([0-9.]{1,10})[/ ]\(compatible[ ;]*ICS" => "\\1"
2888 joku 63 ),
2889     "uri" => "http://www.icra.org"
2890     ),
2891     "irvine" => array(
2892     "icon" => "robot",
2893     "title" => "Irvine",
2894     "rule" => array(
2895     "Irvine[ /]([0-9.]{1,10})" => "\\1"
2896     ),
2897     "uri" => ""
2898     ),
2899     "ips-agent" => array(
2900     "icon" => "robot",
2901 matthys 209 "title" => "IPS-agent",
2902 joku 63 "rule" => array(
2903     "ips-agent" => ""
2904     ),
2905     "uri" => ""
2906     ),
2907     "isara" => array(
2908     "icon" => "isara",
2909 matthys 209 "title" => "Isara",
2910 joku 63 "rule" => array(
2911     "Isara[ -]([0-9.]{1,10})" => "\\1"
2912     ),
2913     "uri" => "http://www.isara.org/"
2914     ),
2915     "isspider" => array(
2916     "icon" => "robot",
2917     "title" => "ISSpider",
2918     "rule" => array(
2919     "ISSpider[ /\-]([0-9.]{1,10})" => "\\1"
2920     ),
2921     "uri" => ""
2922     ),
2923     "it2media" => array(
2924     "icon" => "it2media",
2925     "title" => "IT2media",
2926     "rule" => array(
2927     "www.adressendeutschland.de" => ""
2928     ),
2929     "uri" => "http://www.adressendeutschland.de/"
2930     ),
2931     "ivia" => array(
2932     "icon" => "robot",
2933     "title" => "iVia",
2934     "rule" => array(
2935     "iVia Site Checker.?[ /]([0-9.]{1,10})" => "\\1"
2936     ),
2937     "uri" => ""
2938     ),
2939     "jeteye" => array(
2940     "icon" => "jeteye",
2941     "title" => "Jeteye",
2942     "rule" => array(
2943     "Jetbot[ /]([0-9.]{1,10})" => "\\1"
2944     ),
2945     "uri" => ""
2946     ),
2947     "jobsde" => array(
2948     "icon" => "jobsde",
2949     "title" => "jobs.de",
2950     "rule" => array(
2951     "www\.jobs\.de" => "",
2952     "jobs.de-Robot" => ""
2953     ),
2954     "uri" => "http://www.jobs.de/"
2955     ),
2956     "jpluck" => array(
2957     "icon" => "robot",
2958     "title" => "Jpluck",
2959     "rule" => array(
2960     "JPluck[ /]([0-9a-z.]{1,10})" => "\\1"
2961     ),
2962     "uri" => ""
2963     ),
2964     "jxta" => array(
2965     "icon" => "robot",
2966     "title" => "Jxta",
2967     "rule" => array(
2968     "falcon[ /]([0-9.]{1,10})" => "\\1"
2969     ),
2970     "uri" => ""
2971     ),
2972     "jyte" => array(
2973     "icon" => "jyte",
2974     "title" => "Jyte",
2975     "rule" => array(
2976     "jyte_fetcher[ /]([0-9.]{1,10})" => "\\1"
2977     ),
2978     "uri" => ""
2979     ),
2980     "jyxo" => array(
2981     "icon" => "jyxo",
2982     "title" => "Jyxo",
2983     "rule" => array(
2984     "Jyxobot[ /]([0-9.]{1,10})" => "\\1"
2985     ),
2986     "uri" => ""
2987     ),
2988     "kalooga" => array(
2989     "icon" => "kalooga",
2990     "title" => "Kalooga",
2991     "rule" => array(
2992     "kalooga[ /-]([0-9.]{1,10})" => "\\1",
2993     "^kalooga" => "",
2994     "KaloogaBot" => ""
2995     ),
2996     "uri" => "http://www.kalooga.com"
2997     ),
2998     "keywen" => array(
2999     "icon" => "keywen",
3000     "title" => "Keywen",
3001     "rule" => array(
3002     "EasyDL[ /]([0-9.]{1,10})" => "\\1"
3003     ),
3004     "uri" => ""
3005     ),
3006     "kilomonkey" => array(
3007     "icon" => "robot",
3008     "title" => "Kilomonkey",
3009     "rule" => array(
3010     "kmky-not-a-bot[ /]([0-9.]{1,10})" => "\\1"
3011     ),
3012     "uri" => "http://www.kilomonkey.com/notabot.txt"
3013     ),
3014     "kinja" => array(
3015     "icon" => "kinja",
3016     "title" => "Kinja",
3017     "rule" => array(
3018     "kinjabot[ /]([0-9.]{1,10})" => "\\1",
3019     "^kinjabot" => ""
3020     ),
3021     "uri" => ""
3022     ),
3023     "kiwi" => array(
3024     "icon" => "robot",
3025     "title" => "KiwiStatus",
3026     "rule" => array(
3027     "KiwiStatus[ /]([0-9.]{1,10})" => "\\1",
3028     "KiwiStatus" => ""
3029     ),
3030     "uri" => "http://www.nzs.com/kiei-status/"
3031     ),
3032     "lachesis" => array(
3033     "icon" => "robot",
3034     "title" => "Lachesis",
3035     "rule" => array(
3036     "lachesis" => ""
3037     ),
3038     "uri" => ""
3039     ),
3040     "lanshan" => array(
3041     "icon" => "robot",
3042     "title" => "Lachesis",
3043     "rule" => array(
3044     "lanshanbot[/ ]([0-9.]{1,10})" => "\\1"
3045     ),
3046     "uri" => ""
3047     ),
3048     "lapozz" => array(
3049     "icon" => "lapozz",
3050     "title" => "Lapozz",
3051     "rule" => array(
3052     "LapozzBot[/ ]?([0-9.]{1,10})" => "\\1"
3053     ),
3054     "uri" => ""
3055     ),
3056     "larbin" => array(
3057     "icon" => "robot",
3058     "title" => "Larbin",
3059     "rule" => array(
3060     "larbin[_/ ]?([0-9.]{1,10})" => "\\1"
3061     ),
3062     "uri" => ""
3063     ),
3064     "laurion" => array(
3065     "icon" => "robot",
3066     "title" => "Laurion",
3067     "rule" => array(
3068     "^IPiumBot" => ""
3069     ),
3070     "uri" => "http://www.laurion.com/"
3071     ),
3072     "leechget" => array(
3073     "icon" => "leechget",
3074     "title" => "LeechGet",
3075     "rule" => array(
3076     "^LeechGet[ /]([0-9.]{1,10})" => "\\1"
3077     ),
3078     "uri" => ""
3079     ),
3080     "linguaseek" => array(
3081     "icon" => "linguaseek",
3082     "title" => "Linguaseek",
3083     "rule" => array(
3084     "^LinguaBot[ /]v([0-9.]{1,10})" => "\\1"
3085     ),
3086     "uri" => "http://www.linguaseek.com/"
3087     ),
3088     "linguee" => array(
3089     "icon" => "linguee",
3090     "title" => "Linguee",
3091     "rule" => array(
3092     "^Linguee Bot" => ""
3093     ),
3094     "uri" => "http://www.linguee.com/"
3095     ),
3096 matthys 355 "linkapedia" => array(
3097     "icon" => "linkapedia",
3098     "title" => "Linkapedia",
3099     "rule" => array(
3100     "^linkapedia" => ""
3101     ),
3102     "uri" => "http://www.linkapedia.com"
3103     ),
3104 matthys 346 "linkdex" => array(
3105     "icon" => "linkdex",
3106     "title" => "Linkdex",
3107     "rule" => array(
3108     "linkdexbot[ /]([0-9.]{1,10})" => "\\1"
3109     ),
3110     "uri" => "http://www.linkdex.com/bots/"
3111     ),
3112 matthys 366 "linkedin" => array(
3113     "icon" => "linkedin",
3114     "title" => "LinkedIn",
3115     "rule" => array(
3116     "LinkedInBot[ /]([0-9.]{1,10})" => "\\1"
3117     ),
3118     "uri" => "http://www.linkedin.com"
3119     ),
3120 joku 63 "linkguard" => array(
3121     "icon" => "robot",
3122     "title" => "Linkguard",
3123     "rule" => array(
3124     "Linkguard Online[ /]([0-9.]{1,10})" => "\\1"
3125     ),
3126     "uri" => "http://www.linkguard.com/"
3127     ),
3128     "linklint" => array(
3129     "icon" => "linklint",
3130     "title" => "Linklint",
3131     "rule" => array(
3132     "^Linklint[-\-a-z /]*([0-9.]{1,10})" => "\\1"
3133     ),
3134     "uri" => "http://www.linklint.org/"
3135     ),
3136     "linkman" => array(
3137     "icon" => "linkman",
3138     "title" => "Linkman",
3139     "rule" => array(
3140     "\(compatible; Linkman\)" => ""
3141     ),
3142     "uri" => ""
3143     ),
3144     // Needs to be before linkcheck for correct detection
3145     "linkchecker" => array(
3146     "icon" => "w3c",
3147     "title" => "W3C Link Checker",
3148     "rule" => array(
3149     "^W3C-checklink[ /]([0-9.]{1,10})" => "\\1"
3150     ),
3151     "uri" => "http://validator.w3.org/checklink"
3152     ),
3153     "linkcheck" => array(
3154     "icon" => "linkcheck",
3155     "title" => "Linkcheck",
3156     "rule" => array(
3157     "checklink[ /]([0-9.]{1,10})" => "\\1",
3158     "Link[ \-]?(Chec(k|ker)|Val(et|idator))" => "",
3159     "Adaxas Spider" => "",
3160     "Agent-SharewarePlazaFileCheckBot[ /]([0-9.]{1,10})" => "\\1",
3161     "NetMechanic V([0-9.]{1,10})" => "\\1",
3162     "^InfoLink" => "",
3163     "InternetLinkAgent" => "",
3164     "; SPENG\)" => "",
3165     "SharewarePlazaFileCheckBot" => "",
3166     "fileboost.net" => "",
3167     "^billbot" => ""
3168     ),
3169     "uri" => ""
3170     ),
3171     "linkru" => array(
3172     "icon" => "robot",
3173     "title" => "Link.RU",
3174     "rule" => array(
3175     "^Link.RU bot" => ""
3176     ),
3177     "uri" => ""
3178     ),
3179     "linkssql" => array(
3180     "icon" => "robot",
3181     "title" => "Links SQL",
3182     "rule" => array(
3183     "links sql" => ""
3184     ),
3185     "uri" => ""
3186     ),
3187     "linksweeper" => array(
3188     "icon" => "robot",
3189     "title" => "Link Sweeper",
3190     "rule" => array(
3191     "LinkSweeper[ /]([0-9.]{1,10})" => "\\1"
3192     ),
3193     "uri" => ""
3194     ),
3195     "linkwalker" => array(
3196     "icon" => "robot",
3197     "title" => "Link Walker",
3198     "rule" => array(
3199     "^LinkWalker" => ""
3200     ),
3201     "uri" => ""
3202     ),
3203     "lipperhey" => array(
3204     "icon" => "lipperhey",
3205     "title" => "Lipperhey",
3206     "rule" => array(
3207     "Lipperhey Spider" => ""
3208     ),
3209     "uri" => "http://www.lipperhey.com"
3210     ),
3211     "livedoor" => array(
3212     "icon" => "livedoor",
3213     "title" => "Livedoor",
3214     "rule" => array(
3215     "^Livedoor( SF( - California Crawl)?|Checkers)[ /]" => ""
3216     ),
3217     "uri" => ""
3218     ),
3219     "livejournal" => array(
3220     "icon" => "livejournal",
3221     "title" => "Live Journal",
3222     "rule" => array(
3223     "^LiveJournal\.com" => ""
3224     ),
3225     "uri" => "http://www.livejournal.com"
3226     ),
3227     "ljpic" => array(
3228     "icon" => "ljpic",
3229     "title" => "ljpic",
3230     "rule" => array(
3231     "LjSEEK Picture-Bot[ /]+([0-9.]{1,10})" => "\\1"
3232     ),
3233     "uri" => "http://www.ljpic.com"
3234     ),
3235     "lmspider" => array(
3236     "icon" => "robot",
3237     "title" => "Lmspider",
3238     "rule" => array(
3239     "^lmspider" => ""
3240     ),
3241     "uri" => ""
3242     ),
3243     "locators" => array(
3244     "icon" => "robot",
3245     "title" => "Locaters",
3246     "rule" => array(
3247     "^FiNDoBot[/ ]([0-9a-z.]{1,10})" => "\\1"
3248     ),
3249     "uri" => ""
3250     ),
3251     "look" => array(
3252     "icon" => "look",
3253     "title" => "Look",
3254     "rule" => array(
3255     "www\.look\.com" => "",
3256     "Lookbot" => ""
3257     ),
3258     "uri" => ""
3259     ),
3260     "looksmart" => array(
3261     "icon" => "looksmart",
3262     "title" => "LookSmart",
3263     "rule" => array(
3264     "^Martini" => "",
3265     "^MantraAgent" => "",
3266     "FurlBot" => "",
3267     "looksmart-sv-fw" => ""
3268     ),
3269     "uri" => ""
3270     ),
3271     "loop" => array(
3272     "icon" => "loop",
3273     "title" => "LOOP",
3274     "rule" => array(
3275     "NetResearchServer[ /]([0-9.]{1,10})" => "\\1"
3276     ),
3277     "uri" => ""
3278     ),
3279     "lotkyll" => array(
3280     "icon" => "robot",
3281     "title" => "Lotkyll",
3282     "rule" => array(
3283     "Lotkyll" => ""
3284     ),
3285     "uri" => ""
3286     ),
3287 joku 335 "ltx71" => array(
3288     "icon" => "robot",
3289     "title" => "LTX71",
3290     "rule" => array(
3291     "ltx71" => ""
3292     ),
3293     "uri" => "http://ltx71.com/"
3294     ),
3295 joku 63 "lwp" => array(
3296     "icon" => "robot",
3297     "title" => "lwp",
3298     "rule" => array(
3299     "lwp(-trivial|::simple)[ /]([0-9.]{1,10})" => "\\2"
3300     ),
3301     "uri" => ""
3302     ),
3303     "lycos" => array(
3304     "icon" => "lycos",
3305     "title" => "Lycos",
3306     "rule" => array(
3307     "Lycos_Spider_" => ""
3308     ),
3309     "uri" => ""
3310     ),
3311     "magpierss" => array(
3312     "icon" => "rss",
3313     "title" => "MagpieRSS",
3314     "rule" => array(
3315     "MagpieRSS" => ""
3316     ),
3317     "uri" => "http://magpierss.sourceforge.net/"
3318     ),
3319     "mailru" => array(
3320     "icon" => "mailru",
3321 matthys 209 "title" => "Mail.Ru",
3322 joku 63 "rule" => array(
3323 matthys 195 "^Mail.Ru[ /]([0-9.]{1,10})" => "\\1",
3324     "Mail.RU_Bot[ /]([0-9.]{1,10})" => "\\1"
3325 joku 63 ),
3326 matthys 209 "uri" => "http://www.mail.ru/"
3327 joku 63 ),
3328     "mailsweeper" => array(
3329     "icon" => "robot",
3330     "title" => "Mail Sweeper",
3331     "rule" => array(
3332     "Mail[ \-]?Sweeper" => ""
3333     ),
3334     "uri" => ""
3335     ),
3336     "marvin" => array(
3337     "icon" => "robot",
3338     "title" => "Marvin",
3339     "rule" => array(
3340     "^Marvin" => ""
3341     ),
3342     "uri" => ""
3343     ),
3344     "matkurja" => array(
3345     "icon" => "matkurja",
3346     "title" => "Mat'Kurja",
3347     "rule" => array(
3348     "Mosad[ /]([0-9.]{1,10})" => "\\1"
3349     ),
3350     "uri" => ""
3351     ),
3352     "mavicanet" => array(
3353     "icon" => "mavicanet",
3354     "title" => "Mavicanet",
3355     "rule" => array(
3356     "Mavicanet robot" => ""
3357     ),
3358     "uri" => ""
3359     ),
3360 matthys 292 "meanpath" => array(
3361     "icon" => "meanpath",
3362 matthys 290 "title" => "Meanpath",
3363 matthys 280 "rule" => array(
3364     "meanpathbot[ /]([0-9.]{1,10})" => "\\1"
3365     ),
3366     "uri" => "http://www.meanpath.com/meanpathbot.html"
3367 matthys 275 ),
3368 matthys 280 "mediater" => array(
3369 joku 63 "icon" => "robot",
3370     "title" => "Mediater",
3371     "rule" => array (
3372     "^libwww[ /]([0-9.]{1,10})" => "\\1"
3373 matthys 280 ),
3374     "uri" => ""
3375 joku 63 ),
3376 matthys 305 "memory" => array(
3377     "icon" => "memory",
3378     "title" => "internet Memory",
3379     "rule" => array (
3380     "memoryBot[ /]([0-9.]{1,10})" => "\\1"
3381     ),
3382     "uri" => "http://internetmemory.org/en/"
3383     ),
3384 joku 63 "mercator" => array(
3385     "icon" => "robot",
3386     "title" => "Mercator",
3387     "rule" => array(
3388     "Mercator" => ""
3389     ),
3390     "uri" => ""
3391     ),
3392     "metacarta" => array(
3393     "icon" => "robot",
3394     "title" => "Metacarta",
3395     "rule" => array(
3396     "^RRC (crawler_admin@bigfoot.com)" => "",
3397     "^flunky" => "",
3398     "^Mozilla.*\(samualt9@bigfoot.com\)$" => ""
3399     ),
3400     "uri" => "http://www.metacarta.com"
3401     ),
3402     "metager" => array(
3403     "icon" => "metager",
3404     "title" => "MetaGer",
3405     "rule" => array(
3406     "MetaGer" => ""
3407     ),
3408     "uri" => ""
3409     ),
3410     "metamark" => array(
3411     "icon" => "robot",
3412     "title" => "Metamark",
3413     "rule" => array(
3414     "^XRL[ /]([0-9.a-z]{1,10})" => "\\1"
3415     ),
3416     "uri" => "http://metamark.net"
3417     ),
3418     "metamedic" => array(
3419     "icon" => "metamedic",
3420     "title" => "MetaMedic",
3421     "rule" => array(
3422     "MediBot[ /]([0-9.]{1,10})" => "\\1"
3423     ),
3424     "uri" => ""
3425     ),
3426     "mirago" => array(
3427     "icon" => "mirago",
3428     "title" => "Mirago",
3429     "rule" => array(
3430     "Mirago" => ""
3431     ),
3432     "uri" => ""
3433     ),
3434     "misterwong" => array(
3435     "icon" => "misterwong",
3436     "title" => "Mister Wong",
3437     "rule" => array(
3438     "Mister Wong-Bot[ /]([0-9.]{1,10})" => "\\1"
3439     ),
3440     "uri" => "http://www.mister-wong.de/"
3441     ),
3442     "miva" => array(
3443     "icon" => "miva",
3444     "title" => "Miva",
3445     "rule" => array(
3446     "AlgoFeedback@miva\.com" => ""
3447     ),
3448     "uri" => "http://www.miva.com/"
3449     ),
3450     "mj12" => array(
3451     "icon" => "mj12",
3452     "title" => "Majestic-12",
3453     "rule" => array(
3454     "Mj12bot[ /]v?([0-9.]{1,10})" => "\\1",
3455     "MJ12bot \(mini\)[ /]([0-9.]{1,10})" => "\\1"
3456     ),
3457     "uri" => "http://majestic12.co.uk/bot.php"
3458     ),
3459     "mlbot" => array(
3460     "icon" => "robot",
3461     "title" => "MLBot",
3462     "rule" => array(
3463     "^MLBot" => ""
3464     ),
3465     "uri" => "http://www.metadatalabs.com/mlbot"
3466     ),
3467     "mnogo" => array(
3468     "icon" => "mnogo",
3469     "title" => "mnoGoSearch",
3470     "rule" => array(
3471     "Mnogosearch[ /\-]([0-9.]{1,10})" => "\\1"
3472     ),
3473     "uri" => "http://www.mnogosearch.org/"
3474     ),
3475     "mobileok" => array(
3476     "icon" => "mobileok",
3477     "title" => "W3C mobileOK",
3478     "rule" => array(
3479     "^W3C-mobileOK" => ""
3480     ),
3481     "uri" => "http://www.w3.org/2006/07/mobileok-ddc"
3482     ),
3483 matthys 296 "mojeek" => array(
3484     "icon" => "mojeek",
3485     "title" => "Mojeek",
3486 joku 63 "rule" => array(
3487     "MojeekBot[ /]([0-9.]{1,10})" => "\\1"
3488     ),
3489 matthys 284 "uri" => "https://www.mojeek.com/bot.html"
3490 joku 63 ),
3491     "momspider" => array(
3492     "icon" => "robot",
3493     "title" => "MOM Spider",
3494     "rule" => array(
3495     "MOMspider[ /]([0-9.]{1,10})" => "\\1"
3496     ),
3497     "uri" => ""
3498     ),
3499     "moreover" => array(
3500     "icon" => "moreover",
3501     "title" => "Moreover",
3502     "rule" => array(
3503     "^Moreoverbot[ /]([0-9.]{1,10})" => "\\1"
3504     ),
3505     "uri" => ""
3506     ),
3507     "movabletype" => array(
3508     "icon" => "movabletype",
3509     "title" => "Movable Type",
3510     "rule" => array(
3511     "MovableType[ /]([0-9.]{1,10})" => "\\1"
3512     ),
3513     "uri" => ""
3514     ),
3515     "mozdex" => array(
3516     "icon" => "mozdex",
3517     "title" => "MozDex",
3518     "rule" => array(
3519     "mozDex[ /]([0-9.]{1,6}(-dev)?)" => "\\1"
3520     ),
3521     "uri" => ""
3522     ),
3523     "mqbot" => array(
3524     "icon" => "robot",
3525     "title" => "MQbot",
3526     "rule" => array(
3527     "MQbot" => ""
3528     ),
3529     "uri" => ""
3530     ),
3531     "msnbot" => array(
3532     "icon" => "bing",
3533 matthys 205 "title" => "Bing",
3534 joku 63 "rule" => array(
3535     "MSN(BOT|PTC)[ /]([0-9.]{1,10})" => "\\2",
3536     "MS Search ([0-9.]{1,10}) Robot" => "\\1",
3537     "MSNBOT-(MEDIA|PRODUCTS|NewsBlogs)[ /]([0-9.]{1,10})" => "\\2",
3538     "bingbot[ /]([0-9.]{1,10})" => "\\1"
3539     ),
3540     "uri" => ""
3541     ),
3542     "msnbot-mobile" => array(
3543     "icon" => "msn",
3544     "title" => "MSN Mobile",
3545     "rule" => array(
3546     "MSNBOT-(MOBILE|)[ /]([0-9.]{1,10})" => "\\2",
3547     "MSNBOT_Mobile" => "",
3548     "MSMOBOT" => ""
3549     ),
3550     "uri" => ""
3551     ),
3552     "mslivebot" => array(
3553     "icon" => "livesearch",
3554     "title" => "MS Live Search",
3555     "rule" => array(
3556     "MSNBOT-(Academic)[ /]([0-9.]{1,10})" => "\\2"
3557     ),
3558     "uri" => ""
3559     ),
3560     "msproxy" => array(
3561     "icon" => "robot",
3562     "title" => "MSProxy",
3563     "rule" => array(
3564     "MSProxy[ /]([0-9.]{1,10})" => "\\1"
3565     ),
3566     "uri" => ""
3567     ),
3568     "msrbot" => array(
3569     "icon" => "robot",
3570     "title" => "MSRBOT",
3571     "rule" => array(
3572     "^MSRBOT" => ""
3573     ),
3574     "uri" => "http://research.microsoft.com/research/sv/msrbot/"
3575     ),
3576     "mswebdav" => array(
3577     "icon" => "robot",
3578     "title" => "MS-WebDAV",
3579     "rule" => array(
3580     "Microsoft[ \-]?WebDAV[ \-]?MiniRedir" => ""
3581     ),
3582     "uri" => ""
3583     ),
3584     "mticon" => array(
3585     "icon" => "robot",
3586     "title" => "MTIcon",
3587     "rule" => array(
3588     "MTIcon[/ ]([0-9.]{1,10})" => "\\1"
3589     ),
3590     "uri" => ""
3591     ),
3592     "multipage-validator" => array(
3593     "icon" => "multipage-validator",
3594     "title" => "Multipage Validator",
3595     "rule" => array(
3596     "^W3C_Multipage_Validator[ /]([0-9.]{1,10})" => "\\1"
3597     ),
3598     "uri" => "http://www.validator.ca/"
3599     ),
3600     "myrss" => array(
3601     "icon" => "rss",
3602     "title" => "MyRSS",
3603     "rule" => array(
3604     "MyRSS.jp[/ ]([0-9.]{1,10})" => "\\1"
3605     ),
3606     "uri" => ""
3607     ),
3608     "multimap" => array(
3609     "icon" => "robot",
3610     "title" => "Multimap",
3611     "rule" => array(
3612     "Multimap Geotag Blog Parser[/ ]([0-9.]{1,10})" => "\\1"
3613     ),
3614     "uri" => ""
3615     ),
3616     "najdi" => array(
3617     "icon" => "najdi",
3618     "title" => "Najdi.si",
3619     "rule" => array(
3620     "Najdi.si" => ""
3621     ),
3622     "uri" => "http://www.najdi.si"
3623     ),
3624     "nameprotect" => array(
3625     "icon" => "nameprotect",
3626     "title" => "Name Protect",
3627     "rule" => array(
3628     "NPBot" => ""
3629     ),
3630     "uri" => ""
3631     ),
3632     "nationaldirectory" => array(
3633     "icon" => "robot",
3634     "title" => "National Directory",
3635     "rule" => array(
3636     "NationalDirectory-WebSpider[ /]([0-9.]{1,10})" => "\\1"
3637     ),
3638     "uri" => ""
3639     ),
3640     "natsumican" => array(
3641     "icon" => "robot",
3642     "title" => "Natsu Mican",
3643     "rule" => array(
3644     "NATSU[ \-]MICAN[/ ]([0-9a-z.]{1,10})" => "\\1"
3645     ),
3646     "uri" => ""
3647     ),
3648     "naverbot" => array(
3649     "icon" => "naverbot",
3650     "title" => "Naver",
3651     "rule" => array(
3652     "NaverBot([_\-]dloader)?[/ \-]([0-9.]{1,10})" => "\\2",
3653     "Naver(Bot)?" => "",
3654     "^nabot" => ""
3655     ),
3656     "uri" => ""
3657     ),
3658     "navisso" => array(
3659     "icon" => "navisso",
3660     "title" => "Navisso",
3661     "rule" => array(
3662     "Navisso(Bot)?" => ""
3663     ),
3664     "uri" => "http://www.navisso.com"
3665     ),
3666 matthys 402 "neevabot" => array(
3667     "icon" => "robot",
3668     "title" => "Neeva",
3669     "rule" => array(
3670     "Neevabot[ /]([0-9.]{1,10})" => "\\1"
3671     ),
3672     "uri" => "https://neeva.com/neevabot"
3673     ),
3674 joku 63 "neofonie" => array(
3675     "icon" => "robot",
3676 matthys 205 "title" => "Neofonie",
3677 joku 63 "rule" => array(
3678     "www.neofonie.de" => ""
3679     ),
3680     "uri" => "http://www.neofonie.de/loesungen/search/robot.html"
3681     ),
3682     "neomo" => array(
3683     "icon" => "robot",
3684     "title" => "Neomo",
3685     "rule" => array(
3686     "Francis[ /]([0-9.]{1,10})" => "\\1"
3687     ),
3688     "uri" => ""
3689     ),
3690     "nessus" => array(
3691     "icon" => "nessus",
3692     "title" => "Nessus",
3693     "rule" => array(
3694     "Nessus\)$" => ""
3695     ),
3696     "uri" => ""
3697     ),
3698 matthys 279 "nerdybot" => array(
3699 matthys 280 "icon" => "robot",
3700     "title" => "NerdyBot",
3701     "rule" => array(
3702     "NerdyBot" => ""
3703     ),
3704     "uri" => "http://nerdybot.com/"
3705 matthys 279 ),
3706 joku 63 "nerdbynature" => array(
3707     "icon" => "nerdbynature",
3708     "title" => "NerdByNature",
3709     "rule" => array(
3710     "NerdByNature.Bot" => ""
3711     ),
3712     "uri" => "http://www.nerdbynature.net/bot"
3713     ),
3714     "netants" => array(
3715     "icon" => "netants",
3716     "title" => "NetAnts",
3717     "rule" => array(
3718     "NetAnts[ /]([0-9.]{1,10})" => "\\1"
3719     ),
3720     "uri" => ""
3721     ),
3722     "netcraft" => array(
3723     "icon" => "netcraft",
3724     "title" => "Netcraft",
3725     "rule" => array(
3726     "netcraft" => ""
3727     ),
3728     "uri" => ""
3729     ),
3730     "netinfo" => array(
3731     "icon" => "netinfo",
3732     "title" => "Netinfo",
3733     "rule" => array(
3734     "^NetinfoBot[ /]([0-9.]{1,10})" => "\\1"
3735     ),
3736     "uri" => "http://netinfo.bg/"
3737     ),
3738     "netluchs" => array(
3739     "icon" => "netluchs",
3740     "title" => "Netluchs",
3741     "rule" => array(
3742     "Netluchs[ /]([0-9.a-z]{1,10})" => "\\1"
3743     ),
3744     "uri" => "http://www.netluchs.de/"
3745     ),
3746     "netmechanic" => array(
3747     "icon" => "netmechanic",
3748     "title" => "NetMechanic",
3749     "rule" => array(
3750     "NetMechanic[ /V]{1,5}([0-9.]{1,10})" => "\\1"
3751     ),
3752     "uri" => ""
3753     ),
3754     "netnose" => array(
3755     "icon" => "netnose",
3756     "title" => "NetNose",
3757     "rule" => array(
3758     "NetNose[ \-]Crawler[/ ]([0-9.]{1,10})" => "\\1"
3759     ),
3760     "uri" => ""
3761     ),
3762     "netoskop" => array(
3763     "icon" => "robot",
3764     "title" => "Netoskop",
3765     "rule" => array(
3766     "netoskop" => ""
3767     ),
3768     "uri" => ""
3769     ),
3770     "netpromoter" => array(
3771     "icon" => "netpromoter",
3772     "title" => "NetPromoter",
3773     "rule" => array(
3774     "NetPromoter Spider" => ""
3775     ),
3776     "uri" => "http://www.net-promoter.com/"
3777     ),
3778     "netprospector" => array(
3779     "icon" => "robot",
3780     "title" => "Netprospector",
3781     "rule" => array(
3782     "^netprospector" => ""
3783     )
3784     ),
3785     "netpumper" => array(
3786     "icon" => "robot",
3787     "title" => "Netpumper",
3788     "rule" => array(
3789     "^NetPumper[/ ]([0-9.]{1,10})" => "\\1"
3790     ),
3791     "uri" => ""
3792     ),
3793     "netscapeproxy" => array(
3794     "icon" => "netscape",
3795     "title" => "Netscape Proxy",
3796     "rule" => array(
3797     "Netscape-Proxy[/ ]([0-9.]{1,10})" => "\\1"
3798     ),
3799     "uri" => ""
3800     ),
3801     "netspective" => array(
3802     "icon" => "robot",
3803     "title" => "NetSpective",
3804     "rule" => array(
3805     "^WebFilter Robot ([0-9.]{1,10})" => "\\1"
3806     ),
3807     "uri" => ""
3808     ),
3809     "netvibes" => array(
3810     "icon" => "netvibes",
3811     "title" => " Netvibes",
3812     "rule" => array(
3813     "^Netvibes" => "\\1"
3814     ),
3815     "uri" => ""
3816     ),
3817     "newsfire" => array(
3818     "icon" => "newsfire",
3819     "title" => "NewsFire",
3820     "rule" => array(
3821     "NewsFire[/ ]([0-9.]{1,10})" => "\\1"
3822     ),
3823     "uri" => ""
3824     ),
3825     "newsgator" => array(
3826     "icon" => "newsgator",
3827     "title" => "NewsGator",
3828     "rule" => array(
3829     "NewsGato(r|rOnline)[/ ]([0-9.]{1,10})" => "\\2"
3830     ),
3831     "uri" => ""
3832     ),
3833     "newzcrawler" => array(
3834     "icon" => "newzcrawler",
3835     "title" => "NewzCrawler",
3836     "rule" => array(
3837     "NewzCrawler[/ ]([0-9.]{1,10})" => "\\1"
3838     ),
3839     "uri" => ""
3840     ),
3841     "nextopia" => array(
3842     "icon" => "newzcrawler",
3843     "title" => "NewzCrawler",
3844     "rule" => array(
3845     "^NextopiaBOT.*[v ]([0-9.]{1,10})" => "\\1"
3846     ),
3847     "uri" => ""
3848     ),
3849     "ngsearch" => array(
3850     "icon" => "ngsearch",
3851     "title" => "NG Search",
3852     "rule" => array(
3853     "NG-Search[/ ]([0-9.]{1,10})" => "\\1"
3854     ),
3855     "uri" => ""
3856     ),
3857     "nimble" => array(
3858     "icon" => "robot",
3859     "title" => "Nimble",
3860     "rule" => array(
3861     "NimbleCrawler[/ ]([0-9.]{1,10})" => "\\1"
3862     ),
3863     "uri" => ""
3864     ),
3865     "nmap" => array(
3866     "icon" => "nmap",
3867     "title" => "Nmap",
3868     "rule" => array(
3869     "Nmap Scripting Engine" => ""
3870     ),
3871     "uri" => "http://nmap.org/book/nse.html"
3872     ),
3873     "nusearch" => array(
3874     "icon" => "robot",
3875     "title" => "NuSearch",
3876     "rule" => array(
3877     "^nuSearch" => ""
3878     ),
3879     "uri" => "http://www.nusearch.com/"
3880     ),
3881     "noago" => array(
3882     "icon" => "noago",
3883     "title" => "Noago",
3884     "rule" => array(
3885     "Noago Spider" => ""
3886     ),
3887     "uri" => "http://www.noago.com/"
3888     ),
3889     "noviforum" => array(
3890     "icon" => "noviforum",
3891     "title" => "Noviforum",
3892     "rule" => array(
3893     "TridentSpider[/ ]?([0-9.]{1,10})" => "\\1"
3894     ),
3895     "uri" => ""
3896     ),
3897     "noxtrum" => array(
3898     "icon" => "noxtrum",
3899     "title" => "noXtrum",
3900     "rule" => array(
3901     "noxtrumbot[/ ]?([0-9.]{1,10})" => "\\1"
3902     ),
3903     "uri" => ""
3904     ),
3905     "noyona" => array(
3906     "icon" => "robot",
3907     "title" => "Noyona",
3908     "rule" => array(
3909     "noyona.([0-9._]{1,10})" => "\\1"
3910     ),
3911     "uri" => "http://noyona.com/"
3912     ),
3913     "nsauditor" => array(
3914     "icon" => "nsauditor",
3915     "title" => "Nsauditor",
3916     "rule" => array(
3917     "Nsauditor[ /]([0-9.]{1,10})" => "\\1"
3918     ),
3919     "uri" => "http://www.nsauditor.com/"
3920     ),
3921     "obidosbot" => array(
3922     "icon" => "robot",
3923     "title" => "Bookwatch",
3924     "rule" => array(
3925     "obidos[ \-]?bot" => ""
3926     ),
3927     "uri" => ""
3928     ),
3929     "objectssearch" => array(
3930     "icon" => "robot",
3931     "title" => "Objects Search",
3932     "rule" => array(
3933     "ObjectsSearch[ /]([0-9.]{1,10})" => "\\1"
3934     ),
3935     "uri" => ""
3936     ),
3937     "obot" => array(
3938 matthys 297 "icon" => "ibm",
3939     "title" => "IBM (ONLY) Crawler",
3940 joku 63 "rule" => array(
3941 matthys 297 "oBot[ /]([0-9.]{1,10})" => "\\1"
3942 joku 63 ),
3943 matthys 297 "uri" => "http://filterdb.iss.net/crawler/"
3944     // Previous from "ONLY Solutions GmbH" – that is what the O stands for
3945 joku 63 ),
3946     "ocawa" => array(
3947     "icon" => "ocawa",
3948     "title" => "Ocawa",
3949     "rule" => array(
3950     "Ocawa[ /]([A-Z.a-z][0-9.]{1,10})" => "\\1",
3951     "Ocawa" => ""
3952     ),
3953     "uri" => "http://www.ocawa.com/"
3954     ),
3955     "octora" => array(
3956     "icon" => "octora",
3957     "title" => "Octora",
3958     "rule" => array(
3959     "^Octora (Beta)?" => ""
3960     ),
3961     "uri" => "http://www.octora.com/"
3962     ),
3963     "offlineexplorer" => array(
3964     "icon" => "robot",
3965     "title" => "OfflineExplorer",
3966     "rule" => array(
3967     "^Offline Explorer[ /]([0-9.]{1,10})" => "\\1"
3968     ),
3969     "uri" => ""
3970     ),
3971     "omea" => array(
3972     "icon" => "omea",
3973     "title" => "Omea Reader",
3974     "rule" => array(
3975     "Omea Reader[ /]([0-9.]{1,10})" => "\\1"
3976     ),
3977     "uri" => ""
3978     ),
3979     "onet" => array(
3980     "icon" => "onet",
3981     "title" => "Onet",
3982     "rule" => array(
3983     "OnetSzukaj[ /]([0-9.]{1,10})" => "\\1",
3984     "Onet\.pl" => "",
3985     "inktomi.search.onet" => ""
3986     ),
3987     "uri" => "http://www.onet.pl"
3988     ),
3989     "online24" => array(
3990     "icon" => "robot",
3991 matthys 209 "title" => "Online24",
3992 joku 63 "rule" => array(
3993     "^Online24-Bot .* ([0-9.]{1,10})" => "\\1"
3994     ),
3995     "uri" => "http://www.online24.de"
3996     ),
3997     "onlinewebcheck" => array(
3998     "icon" => "csehtmlvalidator",
3999     "title" => "CSE HTML Validator",
4000     "rule" => array(
4001     "onlinewebcheck" => ""
4002     ),
4003     "uri" => "http://www.onlinewebcheck.com"
4004     ),
4005     "onsearch" => array(
4006     "icon" => "onsearch",
4007 matthys 209 "title" => "OnSearch",
4008 joku 63 "rule" => array(
4009     "^onCHECK-Robot" => ""
4010     ),
4011     "uri" => "http://www.onsearch.de"
4012     ),
4013     "onto" => array(
4014     "icon" => "robot",
4015     "title" => "OntoSpider",
4016     "rule" => array(
4017     "^OntoSpider[ /]([0-9.]{1,10})" => "\\1"
4018     ),
4019     "uri" => "http://ontospider.i-n.info/"
4020     ),
4021     "openacoon" => array(
4022     "icon" => "robot",
4023     "title" => "OpenAcoon",
4024     "rule" => array(
4025     "^OpenAcoon v([0-9.]{1,10})" => "\\1"
4026     ),
4027     "uri" => "http://www.openacoon.de/"
4028     ),
4029     "openfind" => array(
4030     "icon" => "openfind",
4031     "title" => "Openfind",
4032     "rule" => array(
4033     "openbot[ /]([0-9.]{1,10})" => "\\1",
4034     "Openfind Robot[ /]([0-9.A-Z]{1,10})" => "\\1"
4035     ),
4036     "uri" => ""
4037     ),
4038     "opentagger" => array(
4039     "icon" => "robot",
4040     "title" => "OpenTagger",
4041     "rule" => array(
4042     "^OpenTaggerBot" => ""
4043     ),
4044     "uri" => "http://www.opentagger.com/opentaggerbot.htm"
4045     ),
4046     "opentext" => array(
4047     "icon" => "opentext",
4048     "title" => "OpenText",
4049     "rule" => array(
4050     "^OpenTextSiteCrawler[ /]([0-9.]{1,10})" => "\\1"
4051     ),
4052     "uri" => "http://www.opentext.net/"
4053     ),
4054     "openweb" => array(
4055     "icon" => "robot",
4056     "title" => "OpenWebSpider",
4057     "rule" => array(
4058     "^OpenWebSpider[ /]([0-9.]{1,10})" => "\\1",
4059     "^OpenWebSpider v([0-9.A-Z]{1,10})" => "\\1"
4060     ),
4061     "uri" => "http://www.openwebspider.org"
4062     ),
4063     "openx" => array(
4064     "icon" => "openx",
4065     "title" => "OpenX Spider",
4066     "rule" => array(
4067     "OpenX" => ""
4068     ),
4069     "uri" => "http://www.openx.org"
4070     ),
4071     "orange" => array(
4072     "icon" => "orange",
4073     "title" => "Orange",
4074     "rule" => array(
4075 joku 338 "OrangeBot-Mobile[ /]([0-9.]{1,10})" => "\\1",
4076     "OrangeBot[ /]([0-9.]{1,10})" => "\\1"
4077 joku 63 ),
4078     "uri" => "http://orange.com/"
4079     ),
4080     "organica" => array(
4081     "icon" => "robot",
4082     "title" => "Organica",
4083     "rule" => array(
4084     "crawler@organica\.us" => ""
4085     ),
4086     "uri" => ""
4087     ),
4088     "outfox" => array(
4089     "icon" => "robot",
4090     "title" => "Outfox Melon",
4091     "rule" => array(
4092     "OutfoxMelonBot[ /]([0-9.]{1,10})" => "\\1",
4093     "OutfoxBot[ /]([0-9.]{1,10})" => "\\1"
4094     ),
4095     "uri" => ""
4096     ),
4097     "overture" => array(
4098     "icon" => "overture",
4099     "title" => "Overture",
4100     "rule" => array(
4101     "Overture[ \-]?WebCrawler" => ""
4102     ),
4103     "uri" => ""
4104     ),
4105 joku 337 "owler" => array(
4106     "icon" => "robot",
4107     "title" => "Owler",
4108     "rule" => array(
4109     "Owler[ /]([0-9.]{1,10})" => "\\1",
4110     "Owler" => ""
4111     ),
4112     "uri" => ""
4113     ),
4114 joku 63 "page2rss" => array(
4115     "icon" => "page2rss",
4116     "title" => "Page2RSS",
4117     "rule" => array(
4118     "Page2RSS[ /]([0-9.]{1,10})" => "\\1"
4119     ),
4120     "uri" => "http://page2rss.com"
4121     ),
4122     "pagebytes" => array(
4123     "icon" => "robot",
4124     "title" => "PageBites",
4125     "rule" => array(
4126     "^PageBitesHyperBot[ /]([0-9.]{1,10})" => "\\1"
4127     ),
4128     "uri" => ""
4129     ),
4130 matthys 282 "pagesinventory" => array(
4131     "icon" => "pagesinventory",
4132     "title" => "PagesInventory",
4133     "rule" => array(
4134     "^PagesInventory" => ""
4135     ),
4136     "uri" => "http://www.pagesinventory.com"
4137     ),
4138 joku 63 "pagesjaunes" => array(
4139     "icon" => "pagesjaunes",
4140     "title" => "Pages Jaunes",
4141     "rule" => array(
4142     "PJBot[ /]([0-9.]{1,10})" => "\\1"
4143     ),
4144     "uri" => "http://www.pagesjaunes.fr/"
4145     ),
4146     "pagevalet" => array(
4147     "icon" => "wdg",
4148     "title" => "WDG Page Valet",
4149     "rule" => array(
4150     "^Page Valet[ /]([0-9.]{1,10})" => "\\1"
4151     ),
4152     "uri" => "http://www.htmlhelp.com/tools/validator/"
4153     ),
4154     "panopeabot" => array(
4155     "icon" => "robot",
4156     "title" => "PanopeaBot",
4157     "rule" => array(
4158     "PanopeaBot[/ ]([0-9.]{1,10})" => "\\1"
4159     ),
4160     "uri" => ""
4161     ),
4162     "parchmenthill" => array(
4163     "icon" => "robot",
4164     "title" => "Parchment Hill",
4165     "rule" => array(
4166     "ParchBot[ /]([0-9.]{1,10})" => "\\1"
4167     ),
4168     "uri" => "http://www.parchmenthill.com/search.htm"
4169     ),
4170     "pascient" => array(
4171     "icon" => "pascient",
4172     "title" => "Pascient",
4173     "rule" => array(
4174     "panscient.com" => ""
4175     ),
4176     "uri" => "http://www.panscient.com/"
4177     ),
4178     "peerbot" => array(
4179     "icon" => "peerbot",
4180     "title" => "Peerbot",
4181     "rule" => array(
4182     "^PEERbot" => ""
4183     ),
4184     "uri" => ""
4185     ),
4186 matthys 389 "petalbot" => array(
4187     "icon" => "huawei",
4188     "title" => "PetalBot", // Huawei Search - also called Aspiegel
4189     "rule" => array(
4190     "PetalBot" => "",
4191     "AspiegelBot" => ""
4192     ),
4193     "uri" => "https://aspiegel.com/petalbot"
4194     ),
4195 joku 63 "php" => array(
4196     "icon" => "php",
4197     "title" => "PHP",
4198     "rule" => array(
4199     "^PHP[ /]([0-9.]{1,10})" => "\\1"
4200     ),
4201     "uri" => ""
4202     ),
4203     "phpcrawl" => array(
4204     "icon" => "robot",
4205     "title" => "PHPCrawl",
4206     "rule" => array(
4207     "^PHPCrawl" => ""
4208     ),
4209     "uri" => "http://phpcrawl.cuab.de/"
4210     ),
4211     "phpdig" => array(
4212     "icon" => "robot",
4213     "title" => "PhpDig",
4214     "rule" => array(
4215     "^PhpDig[ /]([0-9.]{1,10})" => "\\1"
4216     ),
4217     "uri" => "http://www.finbot.com/"
4218     ),
4219     "phpversiontracker" => array(
4220     "icon" => "phpversiontracker",
4221     "title" => "PHP version tracker",
4222     "rule" => array(
4223     "^PHP version tracker" => ""
4224     ),
4225     "uri" => "http://www.nexen.net/phpversion/bot.php"
4226     ),
4227     "pictureofinternet" => array(
4228     "icon" => "robot",
4229     "title" => "PictureOfInternet",
4230     "rule" => array(
4231     "^PictureOfInternet[ /]([0-9.]{1,10})" => ""
4232     ),
4233     "uri" => ""
4234     ),
4235     "pingdom" => array(
4236     "icon" => "pingdom",
4237     "title" => "Pingdom",
4238     "rule" => array(
4239     "^Pingdom GIGRIB v([0-9.]{1,10})" => "\\1",
4240     "^Pingdom" => ""
4241     ),
4242     "uri" => "http://www.pingdom.com/"
4243     ),
4244     "pinseri" => array(
4245     "icon" => "pinseri",
4246     "title" => "Pinseri",
4247     "rule" => array(
4248     "www\.pinseri\.com/bloglist" => ""
4249     ),
4250     "uri" => ""
4251     ),
4252     "plagger" => array(
4253     "icon" => "plagger",
4254     "title" => "Plagger",
4255     "rule" => array(
4256     "Plagger[ /]([0-9.]{1,10})" => "\\1"
4257     ),
4258     "uri" => "http://www.plugger.org"
4259     ),
4260     "planet" => array(
4261     "icon" => "planet",
4262     "title" => "Planet",
4263     "rule" => array(
4264     "Planet[ /]([0-9.]{1,10})" => "\\1"
4265     ),
4266     "uri" => ""
4267     ),
4268     "plantynet" => array(
4269     "icon" => "robot",
4270     "title" => "PlantyNet",
4271     "rule" => array(
4272     "PlantyNet_WebRobot[_ /]V?([0-9.]{1,10})" => "\\1"
4273     ),
4274     "uri" => ""
4275     ),
4276     "pluck" => array(
4277     "icon" => "pluck",
4278     "title" => "Pluck",
4279     "rule" => array(
4280     "PluckFeedCrawler[ /]([0-9.]{1,10})" => "\\1"
4281     ),
4282     "uri" => ""
4283     ),
4284     "plsearch" => array(
4285     "icon" => "plsearch",
4286     "title" => "PlanetSearch",
4287     "rule" => array(
4288     "fido[ /]([0-9.]{1,10}) Harvest" => "\\1"
4289     ),
4290     "uri" => ""
4291     ),
4292     "poe" => array(
4293     "icon" => "robot",
4294     "title" => "POE-Component",
4295     "rule" => array(
4296     "^POE-Component-Client-HTTP[/ ]([0-9.]{1,10})" => "\\1"
4297     ),
4298     "uri" => ""
4299     ),
4300     "pogodak" => array(
4301     "icon" => "pogodak",
4302     "title" => "Pogodak",
4303     "rule" => array(
4304     "Pogodak\.hr[/ ]?([0-9.]{1,10})" => "\\1"
4305     ),
4306     "uri" => ""
4307     ),
4308     "poodle" => array(
4309     "icon" => "robot",
4310     "title" => "Poodle predictor",
4311     "rule" => array(
4312     "P(oo|ooo)dle[ \-]?predictor[ \-]?([0-9.]{1,10})" => "\\1",
4313     "P(oo|ooo)dle[ \-]?predictor" => ""
4314     ),
4315     "uri" => "http://www.gritechnologies.com/tools/spider.go"
4316     ),
4317     "pompos" => array(
4318     "icon" => "pompos",
4319     "title" => "Pompos",
4320     "rule" => array(
4321     "Pompos[ /]([0-9.]{1,10})" => "\\1"
4322     ),
4323     "uri" => ""
4324     ),
4325     "popdex" => array(
4326     "icon" => "robot",
4327     "title" => "Popdexter",
4328     "rule" => array(
4329     "Popdexter" => ""
4330     ),
4331     "uri" => ""
4332     ),
4333     "postrank" => array(
4334     "icon" => "postrank",
4335     "title" => "PostRank",
4336     "rule" => array(
4337     "^PostRank[ /]+([0-9.]{1,10})" => "\\1"
4338     ),
4339     "uri" => "http://www.postrank.com/"
4340     ),
4341     "powermarks" => array(
4342     "icon" => "robot",
4343     "title" => "Powermarks",
4344     "rule" => array(
4345     "Powermarks[ /]([0-9.]{1,10})" => "\\1"
4346     ),
4347     "uri" => ""
4348     ),
4349 matthys 366 "pinterest" => array(
4350     "icon" => "pinterest",
4351 joku 63 "title" => "PROBE!",
4352     "rule" => array(
4353 matthys 366 "Pinterest[ /]([0-9.]{1,10})" => "\\1"
4354 joku 63 ),
4355 matthys 366 "uri" => "http://www.pinterest.com/"
4356 joku 63 ),
4357 matthys 366 "probe" => array(
4358     "icon" => "robot",
4359     "title" => "PROBE!",
4360     "rule" => array(
4361     "^PROBE!" => ""
4362     ),
4363     "uri" => ""
4364     ),
4365 joku 63 "projecthoneypot" => array(
4366     "icon" => "projecthoneypot",
4367     "title" => "Project Honeypot",
4368     "rule" => array(
4369     "^PHPot Verispider[ /v]+([0-9.]{1,10})" => "\\1",
4370     "projecthoneypot" => ""
4371     ),
4372     "uri" => "http://www.projecthoneypot.org"
4373     ),
4374     "proxycache" => array(
4375     "icon" => "robot",
4376     "title" => "Proxy Cache",
4377     "rule" => array(
4378     "^Mozilla/[0-9.]{1,10} \(compatible\;\)$" => ""
4379     ),
4380     "uri" => ""
4381     ),
4382     "proxyhunter" => array(
4383     "icon" => "robot",
4384     "title" => "ProxyHunter",
4385     "rule" => array(
4386     "ProxyHunter" => ""
4387     ),
4388     "uri" => ""
4389     ),
4390     "psbot" => array(
4391     "icon" => "picsearch",
4392     "title" => "PicSearch",
4393     "rule" => array(
4394     "^psbot" => ""
4395     ),
4396     "uri" => ""
4397     ),
4398     "pubsub" => array(
4399     "icon" => "pubsub",
4400     "title" => "PubSub",
4401     "rule" => array(
4402     "^PubSub-RSS-Reader[ /]([0-9.]{1,10})" => "\\1",
4403     "^PubSub\.com" => ""
4404     ),
4405     "uri" => ""
4406     ),
4407 joku 330 "pubarch" => array(
4408 joku 334 "icon" => "robot",
4409 joku 330 "title" => "PubArchive",
4410     "rule" => array(
4411     "publiclibraryarchive.org" => ""
4412     ),
4413     "uri" => ""
4414     ),
4415 joku 63 "pukiwiki" => array(
4416     "icon" => "pukiwiki",
4417     "title" => "PukiWiki",
4418     "rule" => array(
4419     "PukiWiki[ /]([0-9.]{1,10})" => "\\1"
4420     ),
4421     "uri" => ""
4422     ),
4423     "pwebotxy" => array(
4424     "icon" => "pwebotxy",
4425     "title" => "PWeBot/X.Y",
4426     "rule" => array(
4427     "^PWeBot[ /]([0-9.]{1,10})" => "\\1"
4428     ),
4429     "uri" => "http://www.programacionweb.net/robot.php"
4430     ),
4431     "pxys" => array(
4432     "icon" => "robot",
4433     "title" => "PXYS",
4434     "rule" => array(
4435     "^pxys" => ""
4436     ),
4437     "uri" => ""
4438     ),
4439     "qango" => array(
4440     "icon" => "qango",
4441     "title" => "Qango",
4442     "rule" => array(
4443     "^Qango.com" => ""
4444     ),
4445     "uri" => "http://www.quango.com/"
4446     ),
4447     "qihoo" => array(
4448     "icon" => "qihoo",
4449     "title" => "Qihoo",
4450     "rule" => array(
4451     "QihooBot[ /]([0-9.]{1,10})" => "\\1"
4452     ),
4453     "uri" => "http://www.qihoo.com/"
4454     ),
4455     "qseero" => array(
4456     "icon" => "robot",
4457     "title" => "Qseero",
4458     "rule" => array(
4459     "Qseero v([0-9.]{1,10})" => "\\1"
4460     ),
4461     "uri" => "http://www.q0.com/"
4462     ),
4463     "quantcast" => array(
4464     "icon" => "robot",
4465     "title" => "Quantcast",
4466     "rule" => array(
4467     "Quantcastbot[ /]([0-9.]{1,10})" => "\\1"
4468     ),
4469     "uri" => "http://www.quantcast.com/"
4470     ),
4471     "quepasa" => array(
4472     "icon" => "quepasa",
4473     "title" => "Quepasa",
4474     "rule" => array(
4475     "Quepasa[ \-]?Creep" => ""
4476     ),
4477     "uri" => ""
4478     ),
4479     "questfinder" => array(
4480     "icon" => "robot",
4481     "title" => "QuestFinder",
4482     "rule" => array(
4483     "www\.questfinder\.com" => ""
4484     ),
4485     "uri" => ""
4486     ),
4487 matthys 351 "qwantify" => array(
4488     "icon" => "qwantify",
4489     "title" => "Qwantify",
4490     "rule" => array(
4491     "Qwantify[ /]([0-9.]{1,10})" => "\\1"
4492     ),
4493     "uri" => "https://www.qwant.com/"
4494     ),
4495 joku 63 "qweery" => array(
4496     "icon" => "robot",
4497     "title" => "Qweery",
4498     "rule" => array(
4499     "^QweeryBot[ /]([0-9.]{1,10})" => "\\1"
4500     ),
4501     "uri" => "http://qweerybot.qweery.nl"
4502     ),
4503     "racaicrawler" => array(
4504     "icon" => "robot",
4505     "title" => "RacaiCrawler",
4506     "rule" => array(
4507     "RacaiCrawler[ -]([0-9.]{1,10})" => "\\1"
4508     )
4509     ),
4510     "radian6" => array(
4511     "icon" => "robot",
4512     "title" => "Radian6",
4513     "rule" => array(
4514     "www\.radian6\.com" => ""
4515     ),
4516     "uri" => "http://www.radian6.com"
4517     ),
4518 joku 375 "radiofx" => array(
4519     "icon" => "radiofx",
4520     "title" => "Radio.fx",
4521     "rule" => array(
4522     "Radio.fx" => ""
4523     ),
4524     "uri" => "https://de.tobit.softwarw"
4525     ),
4526 joku 63 "rambler" => array(
4527     "icon" => "rambler",
4528     "title" => "Rambler",
4529     "rule" => array(
4530     "StackRambler[ /]([0-9.]{1,10})" => "\\1"
4531     ),
4532     "uri" => ""
4533     ),
4534     "ramiba" => array(
4535     "icon" => "robot",
4536 matthys 209 "title" => "Ramiba",
4537 joku 63 "rule" => array(
4538     "^ramiba(-bot)?" => "\\1"
4539     ),
4540     "uri" => ""
4541     ),
4542     "rankur" => array(
4543     "icon" => "rankur",
4544     "title" => "Rankur",
4545     "rule" => array(
4546     "^RankurBot/Rankur([0-9.]{1,10})" => "\\1"
4547     ),
4548     "uri" => "http://rankur.com/",
4549     ),
4550 joku 261 "red" => array(
4551     "icon" => "red",
4552     "title" => "RED",
4553     "rule" => array(
4554     "RED[ /]([0-9.]{1,10})" => "\\1"
4555     ),
4556     "uri" => "http://redbot.org/"
4557 joku 181 ),
4558 joku 63 "rediff" => array(
4559     "icon" => "rediff",
4560 matthys 209 "title" => "Rediff",
4561 joku 63 "rule" => array(
4562     "^RedBot/redbot[ /\-]([0-9.]{1,10})" => "\\1"
4563     ),
4564     "uri" => "http://www.rediff.com",
4565     ),
4566     "repia" => array(
4567     "icon" => "robot",
4568     "title" => "Repia",
4569     "rule" => array(
4570     "webmaster@repia\.com" => ""
4571     ),
4572     "uri" => ""
4573     ),
4574     "robotgenius" => array(
4575     "icon" => "robot",
4576     "title" => "Robot Genius",
4577     "rule" => array(
4578     "^robotgenius" => ""
4579     ),
4580     "uri" => "http://robotgenius.net",
4581     ),
4582     "robozilla" => array(
4583     "icon" => "robot",
4584     "title" => "Robozilla",
4585     "rule" => array(
4586     "Robozilla" => ""
4587     ),
4588     "uri" => ""
4589     ),
4590 matthys 302 "rogerbot" => array(
4591 matthys 306 "icon" => "moz",
4592     "title" => "Moz Rogerbot",
4593 matthys 302 "rule" => array(
4594     "rogerbot[ /]([0-9.]{1,10})" => "\\1"
4595     ),
4596     "uri" => "http://moz.com/help/guides/moz-procedures/who-is-rogerbot"
4597     ),
4598 joku 63 "rojo" => array(
4599     "icon" => "rojo",
4600     "title" => "Rojo",
4601     "rule" => array(
4602     "Rojo[ /]([0-9.]{1,10})" => "\\1"
4603     ),
4604     "uri" => ""
4605     ),
4606     "rss-atom" => array(
4607     "icon" => "rss",
4608     "title" => "RSS / Atom",
4609     "rule" => array(
4610     "^Apple-PubSub" => "",
4611     "^AppleSyndication" => ""
4612     ),
4613     "uri" => ""
4614     ),
4615     "rssbot" => array(
4616     "icon" => "rss",
4617 matthys 209 "title" => "RSS-bot",
4618 joku 63 "rule" => array(
4619     "rss-bot[ /]([0-9.]{1,10})" => "\\1"
4620     ),
4621     "uri" => ""
4622     ),
4623     "rssbandit" => array(
4624     "icon" => "rssbandit",
4625     "title" => "RssBandit",
4626     "rule" => array(
4627     "RssBandit[ /]([0-9.]{1,10})" => "\\1"
4628     ),
4629     "uri" => ""
4630     ),
4631     "rssimages" => array(
4632     "icon" => "rss",
4633 matthys 209 "title" => "RssImages",
4634 joku 63 "rule" => array(
4635     "rssImagesBot[ /]([0-9.]{1,10})" => "\\1"
4636     ),
4637     "uri" => ""
4638     ),
4639     "rssmicro" => array(
4640     "icon" => "rssmicro",
4641     "title" => "RSSMicro",
4642     "rule" => array(
4643     "RSSMicro\.com" => ""
4644     ),
4645     "uri" => "http://www.rssmicro.com"
4646     ),
4647     "rssowl" => array(
4648     "icon" => "rssowl",
4649     "title" => "RSSOwl",
4650     "rule" => array(
4651     "RSSOwl[ /]([0-9a-z.]{1,10})" => "\\1"
4652     ),
4653     "uri" => ""
4654     ),
4655     "rssreader" => array(
4656     "icon" => "rss",
4657     "title" => "RssReader",
4658     "rule" => array(
4659     "RssReader[ /]([0-9.]{1,10})" => ""
4660     ),
4661     "uri" => ""
4662     ),
4663     "rtgi" => array(
4664     "icon" => "rtgi",
4665 matthys 209 "title" => "Rtgi",
4666 joku 63 "rule" => array(
4667     "RTGI" => ""
4668     ),
4669     "uri" => "http://rtgi.fr/"
4670     ),
4671     "rufusbot" => array(
4672     "icon" => "robot",
4673     "title" => "RufusBot",
4674     "rule" => array(
4675     "RufusBot" => ""
4676     ),
4677     "uri" => ""
4678     ),
4679 matthys 304 "runet" => array(
4680     "icon" => "runet",
4681     "title" => "iTrack RuNet Crawler",
4682     "rule" => array(
4683     "Runet-Research-Crawler" => ""
4684     ),
4685     "uri" => "http://www.itrack.ru/research/cmsrate/"
4686     ),
4687 joku 63 "runnk" => array(
4688     "icon" => "robot",
4689     "title" => "Runnk",
4690     "rule" => array(
4691     "Runnk RSS finder" => ""
4692     ),
4693     "uri" => "http://www.runnk.com/il/law"
4694     ),
4695     "sagool" => array(
4696     "icon" => "robot",
4697     "title" => "Sagool",
4698     "rule" => array(
4699     "MaSagool" => ""
4700     ),
4701     "uri" => "http://sagool.jp/"
4702     ),
4703     "sanszbot" => array(
4704     "icon" => "robot",
4705     "title" => "Sansz",
4706     "rule" => array(
4707     "SanszBot" => ""
4708     ),
4709     "uri" => ""
4710     ),
4711     "saucereader" => array(
4712     "icon" => "saucereader",
4713     "title" => "Sauce Reader",
4714     "rule" => array(
4715     "Sauce[ ]?Reader[ /]([0-9.]{1,10})" => "\\1"
4716     ),
4717     "uri" => ""
4718     ),
4719     "sbider" => array(
4720     "icon" => "sbider",
4721     "title" => "SBIder",
4722     "rule" => array(
4723     "SBIder[/ ]([0-9.]{1,10})" => "\\1",
4724     "SBIder[/ ]SBIder.([0-9.]{1,10})" => "\\1"
4725     ),
4726     "uri" => ""
4727     ),
4728     "scirus" => array(
4729     "icon" => "robot",
4730     "title" => "Scirus",
4731     "rule" => array(
4732     "FAST-WebCrawler/[0-9a-z.]{1,10}/Scirus" => ""
4733     ),
4734     "uri" => ""
4735     ),
4736     "scoutjet" => array(
4737     "icon" => "scoutjet",
4738     "title" => "ScoutJet",
4739     "rule" => array(
4740     "ScoutJet" => ""
4741     ),
4742     "uri" => "http://www.scoutjet.com/"
4743     ),
4744 joku 261 "scrapy" => array(
4745     "icon" => "scrapy",
4746     "title" => "Scrapy",
4747     "rule" => array(
4748     "Scrapy[ /]([0-9.]{1,10})" => "\\1"
4749     ),
4750     "uri" => "http://scrapy.org"
4751     ),
4752 joku 63 "scrubby" => array(
4753     "icon" => "scrubby",
4754     "title" => "Scrubby",
4755     "rule" => array(
4756     "Scrubby[ /]([0-9.]{1,10})" => "\\1"
4757     ),
4758     "uri" => ""
4759     ),
4760     "sdm" => array(
4761     "icon" => "sdm",
4762     "title" => "SUN Download Manager",
4763     "rule" => array(
4764     "Sun Download Manager[/ ]([0-9.]{1,10})" => "\\1"
4765     ),
4766     "uri" => ""
4767     ),
4768     "sealinks" => array(
4769     "icon" => "robot",
4770     "title" => "Sea Links",
4771     "rule" => array(
4772     "SEA-Links( HTML-Scanner Pingoo\!)?[ /]([0-9.]{1,10})" => "\\2"
4773     ),
4774     "uri" => ""
4775     ),
4776     "search17" => array(
4777     "icon" => "robot",
4778     "title" => "Search 17",
4779     "rule" => array(
4780     "Search17Bot[ /]([0-9.]{1,10})" => "\\1"
4781     ),
4782     "uri" => "http://www.search17.com/bot.php"
4783     ),
4784     "search2" => array(
4785     "icon" => "search2",
4786     "title" => "Search2.net",
4787     "rule" => array(
4788     "^S2Bot[ /]([0-9.]{1,10})" => "\\1"
4789     ),
4790     "uri" => "http://search2.net/"
4791     ),
4792     "searchbot" => array(
4793     "icon" => "robot",
4794     "title" => "Searchbot",
4795     "rule" => array(
4796     "Searchbot[ /]([0-9.]{1,10})" => "\\1"
4797     ),
4798     "uri" => ""
4799     ),
4800     "searchch" => array(
4801     "icon" => "robot",
4802     "title" => "Search.ch",
4803     "rule" => array(
4804     "search\.ch[ /]?V?([0-9.]{1,10})" => "\\1"
4805     ),
4806     "uri" => ""
4807     ),
4808     "searchengineworld" => array(
4809     "icon" => "robot",
4810     "title" => "SearchEngineWorld",
4811     "rule" => array(
4812     "searchengineworld" => ""
4813     ),
4814     "uri" => "http://www.searchengineworld.com/"
4815     ),
4816     "searchhippo" => array(
4817     "icon" => "searchhippo",
4818     "title" => "Searchhippo",
4819     "rule" => array(
4820     "searchhippo" => ""
4821     ),
4822     "uri" => "http://www.searchhippo.com/"
4823     ),
4824     "searchthruus" => array(
4825     "icon" => "robot",
4826     "title" => "SearchThruUs",
4827     "rule" => array(
4828     "www\.unitek-systems\.co\.uk[ /]([0-9.]{1,10})" => "\\1"
4829     ),
4830     "uri" => ""
4831     ),
4832     "securecomputing" => array(
4833     "icon" => "robot",
4834     "title" => "Secure Computing",
4835     "rule" => array(
4836     "securecomputing" => ""
4837     ),
4838     "uri" => ""
4839     ),
4840     "seekport" => array(
4841     "icon" => "seekport",
4842     "title" => "Seekport",
4843     "rule" => array(
4844     "Seekbot[ /]([0-9.]{1,10})" => "\\1"
4845     ),
4846     "uri" => ""
4847     ),
4848     "semanticdiscovery" => array(
4849     "icon" => "robot",
4850     "title" => "Semantic Discovery",
4851     "rule" => array(
4852     "semanticdiscovery[ /]([0-9.]{1,10})" => "\\1"
4853     ),
4854     "uri" => ""
4855     ),
4856 matthys 319 "semrush" => array(
4857     "icon" => "semrush",
4858     "title" => "SEMrush",
4859     "rule" => array(
4860     "SemrushBot[ /]([0-9.]{1,10})" => "\\1"
4861     ),
4862     "uri" => "http://www.semrush.com/bot.html"
4863     ),
4864 joku 63 "sengine" => array(
4865     "icon" => "sengine",
4866     "title" => "Sengine",
4867     "rule" => array(
4868     "netEstate NE Crawler" => ""
4869     ),
4870     "uri" => "http://www.sengine.info/"
4871     ),
4872     "sensis" => array(
4873     "icon" => "sensis",
4874     "title" => "Sensis",
4875     "rule" => array(
4876     "^Sensis(.com.au)? Web Crawler" => ""
4877     ),
4878     "uri" => "http://sensis.com.au"
4879     ),
4880     "seokicks" => array(
4881     "icon" => "seokicks",
4882     "title" => "SEOkicks",
4883     "rule" => array(
4884     "SEOkicks-Robot" => ""
4885     ),
4886     "uri" => "http://www.seokicks.de/"
4887     ),
4888     "seoprofiler" => array(
4889     "icon" => "seoprofiler",
4890     "title" => "SEOprofiler",
4891     "rule" => array(
4892     "spbot[ /]([0-9.]{1,10})" => "\\1"
4893     ),
4894     "uri" => "http://www.seoprofiler.com/"
4895     ),
4896     "setooz" => array(
4897     "icon" => "setooz",
4898     "title" => "Setooz",
4899     "rule" => array(
4900     "OOZBOT[ /]([0-9.]{1,10})" => "\\1"
4901     ),
4902     "uri" => "http://www.setooz.com/oozbot.html"
4903     ),
4904     "seznam" => array(
4905     "icon" => "seznam",
4906     "title" => "Seznam",
4907     "rule" => array(
4908     "SeznamBot[ /]([0-9.]{1,10})" => "\\1",
4909     "Seznam" => ""
4910     ),
4911     "uri" => "http://www.seznam.cz"
4912     ),
4913     "sharpreader" => array(
4914     "icon" => "sharpreader",
4915     "title" => "SharpReader",
4916     "rule" => array(
4917     "SharpReader[ /]([0-9.]{1,10})" => "\\1"
4918     ),
4919     "uri" => ""
4920     ),
4921     "shelob" => array(
4922     "icon" => "shelob",
4923     "title" => "Sherlock Spider",
4924     "rule" => array(
4925     "^Shelob" => ""
4926     ),
4927     "uri" => "http://www.gmx.net"
4928     ),
4929     "sherlockspider" => array(
4930     "icon" => "robot",
4931     "title" => "Sherlock Spider",
4932     "rule" => array(
4933     "sherlock_spider" => ""
4934     ),
4935     "uri" => ""
4936     ),
4937     "shim" => array(
4938     "icon" => "robot",
4939     "title" => "Shim Crawler",
4940     "rule" => array(
4941     "shim[ \-]crawler" => ""
4942     ),
4943     "uri" => ""
4944     ),
4945     "shopwiki" => array(
4946     "icon" => "shopwiki",
4947     "title" => "ShopWiki",
4948     "rule" => array(
4949     "^ShopWiki[ /]([0-9.]{1,10})" => "\\1"
4950     ),
4951     "uri" => ""
4952     ),
4953     "shoula" => array(
4954     "icon" => "robot",
4955     "title" => "Shoula",
4956     "rule" => array(
4957     "^Shoula.com Crawler ([0-9.]{1,10})" => "\\1"
4958     ),
4959     "uri" => "http://www.shoula.com"
4960     ),
4961     "siege" => array(
4962     "icon" => "robot",
4963     "title" => "Siege",
4964     "rule" => array(
4965     "Siege[ /]([0-9.]{1,10})" => "\\1"
4966     ),
4967     "uri" => ""
4968     ),
4969     "siets" => array(
4970     "icon" => "robot",
4971     "title" => "Siets",
4972     "rule" => array(
4973     "SietsCrawler[ /]([0-9.]{1,10})" => "\\1"
4974     ),
4975     "uri" => ""
4976     ),
4977     "simpy" => array(
4978     "icon" => "simpy",
4979     "title" => "Simpy",
4980     "rule" => array(
4981     "^(argus|simpy)[ /]([0-9.]{1,10})" => "\\2",
4982     ),
4983     "uri" => ""
4984     ),
4985     "singingfish" => array(
4986     "icon" => "singingfish",
4987     "title" => "SingingFish",
4988     "rule" => array(
4989     "asterias[ /]([0-9.]{1,10})" => "\\1",
4990     "Asterias Crawler v([0-9.]{1,10})" => "\\1",
4991     "asterias" => ""
4992     ),
4993     "uri" => ""
4994     ),
4995     "sirketce" => array(
4996     "icon" => "robot",
4997     "title" => "Sirketce",
4998     "rule" => array(
4999     "Sirketcebot[ /v]+([0-9.]{1,10})" => "\\1"
5000     ),
5001     "uri" => "http://www.sirketce.com/bot.html"
5002     ),
5003     "sirobot" => array(
5004     "icon" => "robot",
5005     "title" => "SiroBot",
5006     "rule" => array(
5007     "sirobot" => ""
5008     ),
5009     "uri" => ""
5010     ),
5011     "sistrix" => array(
5012     "icon" => "sistrix",
5013     "title" => "SISTRIX",
5014     "rule" => array(
5015     "SISTRIX Crawler" => ""
5016     ),
5017     "uri" => "http://crawler.sistrix.net/"
5018     ),
5019     "sitebar" => array(
5020     "icon" => "sitebar",
5021     "title" => "SiteBar",
5022     "rule" => array(
5023     "SiteBar[ /]([0-9.]{1,10})" => "\\1"
5024     ),
5025     "uri" => ""
5026     ),
5027     "sitebot" => array(
5028     "icon" => "sitebot",
5029     "title" => "SiteBot",
5030     "rule" => array(
5031     "SiteBot[ /]([0-9.]{1,10})" => "\\1"
5032     ),
5033     "uri" => "http://www.sitebot.org/robot/"
5034     ),
5035 matthys 273 "siteexplorer" => array(
5036     "icon" => "siteexplorer",
5037     "title" => "SiteExplorer",
5038     "rule" => array(
5039     "SiteExplorer[ /]([0-9a-z.]{1,10})" => "\\1"
5040     ),
5041     "uri" => "http://siteexplorer.info/"
5042     ),
5043 joku 63 "sitesell" => array(
5044     "icon" => "sitesell",
5045     "title" => "SiteSell",
5046     "rule" => array(
5047     "SBIder[/ ]([0-9a-z.\-]{1,10})" => "\\1"
5048     ),
5049     "uri" => ""
5050     ),
5051     "sitespider" => array(
5052     "icon" => "robot",
5053     "title" => "SiteSpider",
5054     "rule" => array(
5055     "^SiteSpider" => ""
5056     ),
5057     "uri" => ""
5058     ),
5059     "sitesucker" => array(
5060     "icon" => "sitesucker",
5061     "title" => "SiteSucker",
5062     "rule" => array(
5063     "Sitesucker[ /]([0-9.]{1,10})" => "\\1"
5064     ),
5065     "uri" => "http://www.sitesucker.us/"
5066     ),
5067     "sitidi" => array(
5068     "icon" => "robot",
5069     "title" => "SitiDi",
5070     "rule" => array(
5071     "SitiDiBot[ /]([0-9.]{1,10})" => "\\1"
5072     ),
5073     "uri" => ""
5074     ),
5075 joku 329 "sixy" => array(
5076     "icon" => "robot",
5077     "title" => "Sixy.ch",
5078     "rule" => array(
5079     "sixy.ch[ /]([0-9a-z.]{1,10})" => "\\1"
5080     ),
5081     "uri" => "http://sixy.ch"
5082     ),
5083 joku 63 "skaffe" => array(
5084     "icon" => "robot",
5085     "title" => "Skaffe",
5086     "rule" => array(
5087     "Skampy[ /]([0-9.\-]{1,10})" => "\\1"
5088     ),
5089     "uri" => "http://www.skaffe.com"
5090     ),
5091     "skizzle" => array(
5092     "icon" => "skizzle",
5093     "title" => "Skizzle",
5094     "rule" => array(
5095     "SKIZZLE! Distributed Internet Spider[ /v]+([0-9a-z.\-]{1,10})" => "\\1"
5096     ),
5097     "uri" => "http://www.skizzle.com"
5098     ),
5099     "slider" => array(
5100     "icon" => "robot",
5101     "title" => "Slider",
5102     "rule" => array(
5103     "^Slider[ /]([0-9.]{1,10})" => "\\1"
5104     ),
5105     "uri" => ""
5106     ),
5107     "slugch" => array(
5108     "icon" => "robot",
5109 matthys 209 "title" => "Slugch",
5110 joku 63 "rule" => array(
5111     "^slug\.ch crawl ([0-9a-z.\-]{1,10})" => "\\1"
5112     ),
5113     "uri" => ""
5114     ),
5115     "smartware" => array(
5116     "icon" => "robot",
5117     "title" => "SmartWareSoft",
5118     "rule" => array(
5119     "^SWSBot-Images[ /]([0-9.]{1,10})" => "\\1"
5120     ),
5121     "uri" => "http://www.smartwaresoft.com/swsbot12.html"
5122     ),
5123     "snookit" => array(
5124     "icon" => "snookit",
5125     "title" => "Snookit",
5126     "rule" => array(
5127     "^snookit" => ""
5128     ),
5129     "uri" => "http://www.snookit.com/"
5130     ),
5131     "snoopy" => array(
5132     "icon" => "robot",
5133     "title" => "Snoopy",
5134     "rule" => array(
5135     "^Snoopy.+([0-9.]{1,10})" => "\\1",
5136     "sna-([0-9.]{1,10})" => "\\1"
5137     ),
5138     "uri" => "http://snoopy.sourceforge.net/"
5139     ),
5140     "snyke" => array(
5141     "icon" => "robot",
5142     "title" => "Snyke",
5143     "rule" => array(
5144     "^SnykeBot[ /]([0-9.]{1,10})" => "\\1"
5145     ),
5146     "uri" => "http://www.snyke.com"
5147     ),
5148     "socialradar" => array(
5149     "icon" => "socialradar",
5150     "title" => "Social Radar",
5151     "rule" => array(
5152     "Socialradarbot[ /]([0-9.]{1,10})" => "\\1"
5153     ),
5154     "uri" => "http://infegy.com/"
5155     ),
5156     "soegning" => array(
5157     "icon" => "soegning",
5158 matthys 209 "title" => "Søgning",
5159 joku 63 "rule" => array(
5160     "soegning\.dk[/ ]spider[ /]([0-9.]{1,10})" => "\\1"
5161     ),
5162     "uri" => ""
5163     ),
5164     "soft411" => array(
5165     "icon" => "soft411",
5166     "title" => "Soft411",
5167     "rule" => array(
5168     "SOFT411 Directory" => ""
5169     ),
5170     "uri" => ""
5171     ),
5172     "sogou" => array(
5173     "icon" => "sogou",
5174     "title" => "Sogou",
5175     "rule" => array(
5176     "Sogou web spider[ /]([0-9.]{1,10})" => "\\1"
5177     ),
5178     "uri" => "http://www.sogou.com/docs/help/webmasters.htm#07"
5179     ),
5180     "sohu" => array(
5181     "icon" => "robot",
5182     "title" => "Sohu",
5183     "rule" => array(
5184     "sohu[ \-](agent|search)" => ""
5185     ),
5186     "uri" => ""
5187     ),
5188     "somewhere" => array(
5189     "icon" => "robot",
5190 matthys 209 "title" => "Somewhere.com",
5191 joku 63 "rule" => array(
5192     "Mozilla\@somewhere\.com" => "\\1"
5193     ),
5194     "uri" => "http://www.somewhere.com/"
5195     ),
5196     "sopheus" => array(
5197     "icon" => "robot",
5198     "title" => "Sopheus",
5199     "rule" => array(
5200     "Sopheus Project[ /]([0-9.]{1,10})" => "\\1"
5201     ),
5202     "uri" => "http://www.thenetplanet.com"
5203     ),
5204     "soso" => array(
5205     "icon" => "soso",
5206     "title" => "Soso",
5207     "rule" => array(
5208     "^Soso(image)?spider" => ""
5209     ),
5210     "uri" => "http://help.soso.com/soso-image-spider.htm"
5211     ),
5212     "souppot" => array(
5213     "icon" => "robot",
5214     "title" => "SoupPot",
5215     "rule" => array(
5216     "SoupPotBot" => ""
5217     ),
5218     "uri" => ""
5219     ),
5220     "specificmedia" => array(
5221     "icon" => "specificmedia",
5222     "title" => "Specific Media",
5223     "rule" => array(
5224     "^SMBot[ /]([0-9.]{1,10})" => "\\1"
5225     ),
5226     "uri" => ""
5227     ),
5228     "sphider" => array(
5229     "icon" => "sphider",
5230     "title" => "Sphider",
5231     "rule" => array(
5232     "Sphider" => ""
5233     ),
5234     "uri" => "http://www.sphider.eu"
5235     ),
5236     "spinn3r" => array(
5237     "icon" => "robot",
5238     "title" => "Spinn3r",
5239     "rule" => array(
5240     "Spinn3r[ /]([0-9.]{1,10})" => "\\1"
5241     ),
5242     "uri" => "http://spinn3r.com/robot"
5243     ),
5244     "sproose" => array(
5245     "icon" => "robot",
5246 matthys 209 "title" => "Sproose",
5247 joku 63 "rule" => array(
5248     "^sproose[ /]([0-9a-z.]{1,10})" => "\\1"
5249     ),
5250     "uri" => "http://www.sproose.com/bot.html"
5251     ),
5252     "spurlbot" => array(
5253     "icon" => "robot",
5254     "title" => "SpurlBot",
5255     "rule" => array(
5256     "SpurlBot[/ ]([0-9.]{1,10})" => "\\1"
5257     ),
5258     "uri" => ""
5259     ),
5260     "stardownloader" => array(
5261     "icon" => "stardownloader",
5262     "title" => "Star Downloader",
5263     "rule" => array(
5264     "^Star Downloader( Pro)?" => ""
5265     ),
5266     "uri" => ""
5267     ),
5268     "steeler" => array(
5269     "icon" => "robot",
5270     "title" => "Steeler",
5271     "rule" => array(
5272     "Steeler[ /]([0-9.]{1,10})" => "\\1"
5273     ),
5274     "uri" => ""
5275     ),
5276     "strategicboard" => array(
5277     "icon" => "strategicboard",
5278     "title" => "Strategic Board",
5279     "rule" => array(
5280     "Strategic Board Bot" => ""
5281     ),
5282     "uri" => "http://www.strategicboard.com"
5283     ),
5284     "suchbaer" => array(
5285     "icon" => "robot",
5286 matthys 209 "title" => "Suchbaer.de",
5287 joku 63 "rule" => array(
5288     "^suchbaer\.de" => ""
5289     ),
5290     "uri" => "http://www.suchbaer.de/"
5291     ),
5292     "suchbot" => array(
5293     "icon" => "robot",
5294 matthys 209 "title" => "Suchbot",
5295 joku 63 "rule" => array(
5296     "^suchbot" => ""
5297     ),
5298     "uri" => ""
5299     ),
5300     "suchende" => array(
5301     "icon" => "suchende",
5302 matthys 209 "title" => "Suchen.de",
5303 joku 63 "rule" => array(
5304     "^gonzo([0-9]{1,2}).*www.suchen.de" => "\\1",
5305     "^gonzo/([0-9.]{1,10})" => "\\1"
5306     ),
5307     "uri" => "http://www.suchen.de/"
5308     ),
5309     "suchknecht" => array(
5310     "icon" => "robot",
5311     "title" => "Suchknecht",
5312     "rule" => array(
5313     "^Suchknecht.at-Robot" => ""
5314     ),
5315     "uri" => "http://www.suchknecht.at/"
5316     ),
5317     "suchpad" => array(
5318     "icon" => "robot",
5319 matthys 209 "title" => "Suchpad",
5320 joku 63 "rule" => array(
5321     "^suchpadbot[ /]([0-9.]{1,10})" => "\\1"
5322     ),
5323     "uri" => "http://www.suchpad.de"
5324     ),
5325     "sunrise" => array(
5326     "icon" => "sunrise",
5327     "title" => "Sunrise",
5328     "rule" => array(
5329     "^Sunrise[ /]([0-9a-z.]{1,10})" => "\\1"
5330     ),
5331     "uri" => ""
5332     ),
5333     "superbot" => array(
5334     "icon" => "superbot",
5335     "title" => "SuperBot",
5336     "rule" => array(
5337     "SuperBot[ /]([0-9.]{1,10})" => "\\1"
5338     ),
5339     "uri" => ""
5340     ),
5341     "surfcontrol" => array(
5342     "icon" => "robot",
5343     "title" => "SurfControl",
5344     "rule" => array(
5345     "SurfControl" => "",
5346     "ScSpider[ /]([0-9.]{1,10})" => "\\1"
5347     ),
5348     "uri" => ""
5349     ),
5350     "surfnet" => array(
5351     "icon" => "robot",
5352     "title" => "SURFnet",
5353     "rule" => array(
5354     "AVSearch[ \-]([0-9.]{1,10})" => "\\1"
5355     ),
5356     "uri" => ""
5357     ),
5358     "surfsafely" => array(
5359     "icon" => "robot",
5360     "title" => "Surfsafely",
5361     "rule" => array(
5362     "Submission Spider at surfsafely.com" => ""
5363     ),
5364     "uri" => "http://www.surfsafely.com"
5365     ),
5366     "surphace" => array(
5367     "icon" => "spherescout",
5368     "title" => "Surphace",
5369     "rule" => array(
5370     "Sphere Scout.+([0-9.]{1,10})" => "\\1",//Previous called Sphere Scout
5371     "Surphace.+v([0-9.]{1,10})" => "\\1"
5372     ),
5373     "uri" => "http://www.surphace.com"
5374     ),
5375     "surveybot" => array(
5376     "icon" => "surveybot",
5377     "title" => "Whois Survey",
5378     "rule" => array(
5379     "SurveyBot[ /]([0-9.]{1,10})" => "\\1"
5380     ),
5381     "uri" => ""
5382     ),
5383     "swoogle" => array(
5384     "icon" => "robot",
5385     "title" => "Swoogle",
5386     "rule" => array(
5387     "^Swooglebot[ /]([0-9.]{1,10})" => "\\1"
5388     ),
5389     "uri" => "http://swoogle.umbc.edu/swooglebot.htm"
5390     ),
5391     "swse" => array(
5392     "icon" => "robot",
5393     "title" => "SWSE",
5394     "rule" => array(
5395     "sw\.deri\.org" => ""
5396     ),
5397     "uri" => "http://sw.deri.org/2006/04/multicrawler/robots.html"
5398     ),
5399     "swish-e" => array(
5400     "icon" => "swish-e",
5401     "title" => "Swish-e",
5402     "rule" => array(
5403     "^swish-e" => ""
5404     ),
5405     "uri" => "http://swish-e.org/"
5406     ),
5407     "sygol" => array(
5408     "icon" => "sygol",
5409     "title" => "Sygol",
5410     "rule" => array(
5411     "www.sygol.(com|net)" => ""
5412     ),
5413     "uri" => "http://www.sygol.com"
5414     ),
5415     "syllabs" => array(
5416     "icon" => "syllabs",
5417     "title" => "Syllabs",
5418     "rule" => array(
5419     "Focal[-]([0-9.]{1,10})" => "\\1"
5420     ),
5421     "uri" => "http://www.syllabs.com/crawler.html"
5422     ),
5423     "synapse" => array(
5424     "icon" => "robot",
5425     "title" => "Synapse",
5426     "rule" => array(
5427     " Synapse\)" => ""
5428     ),
5429     "uri" => "http://ws.apache.org/synapse/"
5430     ),
5431     "sync2it" => array(
5432     "icon" => "robot",
5433 matthys 209 "title" => "Sync2it",
5434 joku 63 "rule" => array(
5435     "^\!Susie" => ""
5436     ),
5437     "uri" => "http://www.sync2it.com/bms/susie.php"
5438     ),
5439     "syncit" => array(
5440     "icon" => "robot",
5441 matthys 209 "title" => "Syncit",
5442 joku 63 "rule" => array(
5443     "^SyncIT[ /]([0-9.]{1,10})" => "\\1"
5444     ),
5445     "uri" => "http://www.syncit.com/"
5446     ),
5447     "syndic8" => array(
5448     "icon" => "syndic8",
5449     "title" => "Syndic8",
5450     "rule" => array(
5451     "Syndic8[ /]([0-9.]{1,10})" => "\\1"
5452     ),
5453     "uri" => "http://www.syndic8.com/"
5454     ),
5455     "syndicatie" => array(
5456     "icon" => "robot",
5457     "title" => "Syndicatie.nl",
5458     "rule" => array(
5459     "Syndicatie\.nl robot v ([0-9.]{1,10})" => "\\1",
5460     "Syndicatie\.nl robot;" => ""
5461     ),
5462     "uri" => ""
5463     ),
5464     "synomia" => array(
5465     "icon" => "robot",
5466     "title" => "Synomia",
5467     "rule" => array(
5468     "^SynoBot" => ""
5469     ),
5470     "uri" => ""
5471     ),
5472     "synoo" => array(
5473     "icon" => "robot",
5474     "title" => "SynooBot",
5475     "rule" => array(
5476     "SynooBot[ /]([0-9.]{1,10})" => "\\1",
5477     ),
5478     "uri" => ""
5479     ),
5480     "szukacz" => array(
5481     "icon" => "szukacz",
5482     "title" => "Szukacz",
5483     "rule" => array(
5484     "Szukacz[ /]([0-9.]{1,10})" => "\\1"
5485     ),
5486     "uri" => ""
5487     ),
5488     "tagoo" => array(
5489     "icon" => "tagoobot",
5490     "title" => "Tagoo",
5491     "rule" => array(
5492     "Tagoobot[ /]([0-9.]{1,10})" => "\\1"
5493     ),
5494     "uri" => "http://www.tagoo.ru"
5495     ),
5496     "tagword" => array(
5497     "icon" => "tagword",
5498     "title" => "Tagword",
5499     "rule" => array(
5500     "^Tagword" => ""
5501     ),
5502     "uri" => "http://tagword.com/dmoz_survey.php"
5503     ),
5504     "trailfire" => array(
5505     "icon" => "trailfire",
5506     "title" => "Trailfire",
5507     "rule" => array(
5508     "^Trailfire-bot[ /]([0-9.]{1,10})" => "\\1"
5509     ),
5510     "uri" => "http://trailfire.com"
5511     ),
5512     "tamu" => array(
5513     "icon" => "tamu",
5514     "title" => "Tamu Crawler",
5515     "rule" => array(
5516     "IRLbot[ /]([0-9.]{1,10})" => "\\1",
5517     "TAMU_CS_IRL_CRAWLER[ /]([0-9.]{1,10})" => "\\1"
5518     ),
5519     "uri" => "http://irl.cs.tamu.edu/crawler/"
5520     ),
5521     "taptubot" => array(
5522     "icon" => "taptubot",
5523     "title" => "Taptu",
5524     "rule" => array(
5525     "taptubot" => ""
5526     ),
5527     "uri" => "http://www.taptu.com/corp/taptubot"
5528     ),
5529     "targetseek" => array(
5530     "icon" => "robot",
5531     "title" => "TargetSeek",
5532     "rule" => array(
5533     "TargetSeek[ /]([0-9.]{1,10})" => "\\1"
5534     ),
5535     "uri" => "http://www.targetgroups.net/TargetSeek.html"
5536     ),
5537     "taw" => array(
5538     "icon" => "taw",
5539     "title" => "TAW",
5540     "rule" => array(
5541     "TAW[ /]([0-9.]{1,10})" => "\\1"
5542     ),
5543     "uri" => "http://www.tawdis.net"
5544     ),
5545     "tcd" => array(
5546     "icon" => "tcd",
5547     "title" => "Trinity College Dublin",
5548     "rule" => array(
5549     "^TCDBOT/Nutch-([0-9.]{1,10})" => "\\1"
5550     ),
5551     "uri" => "http://www.tcd.ie"
5552     ),
5553     "technorati" => array(
5554     "icon" => "technorati",
5555     "title" => "Technorati",
5556     "rule" => array(
5557     "Technoratibot[ /]([0-9.]{1,10})" => "\\1"
5558     ),
5559     "uri" => ""
5560     ),
5561     "teleport" => array(
5562     "icon" => "teleport",
5563     "title" => "Teleport",
5564     "rule" => array(
5565     "Teleport[ \-]?Pro" => ""
5566     ),
5567     "uri" => ""
5568     ),
5569     "terrar" => array(
5570     "icon" => "robot",
5571     "title" => "Terrar",
5572     "rule" => array(
5573     "^Fresh Search :: Terrar" => ""
5574     ),
5575     "uri" => ""
5576     ),
5577     "theophrastus" => array(
5578     "icon" => "robot",
5579     "title" => "Theophrastus",
5580     "rule" => array(
5581     "Theophrastus[ /]([0-9.]{1,10})" => "\\1"
5582     ),
5583     "uri" => "http://users.cs.cf.ac.uk/N.A.Smith/theophrastus.php"
5584     ),
5585     "thumbnailscz" => array(
5586     "icon" => "robot",
5587 matthys 209 "title" => "Thumbnails.cz",
5588 joku 63 "rule" => array(
5589     "^thumbnail\.cz robot[ /]([0-9.]{1,10})" => "\\1"
5590     ),
5591     "uri" => ""
5592     ),
5593     "thumbshot" => array(
5594     "icon" => "robot",
5595 matthys 209 "title" => "Thumbshots",
5596 joku 63 "rule" => array(
5597     "^thumbshots.*(Version: |v)([0-9.]{2,10})e" => "\\1",
5598     "^thumbshots-de" => ""
5599     ),
5600     "uri" => "http://www.thumbshots.de"
5601     ),
5602     "thunderstone" => array(
5603     "icon" => "thunderstone",
5604     "title" => "Thunderstone",
5605     "rule" => array(
5606     "T-H-U-N-D-E-R-S-T-O-N-E" => "",
5607     "search.thunderstone.com" => ""
5608     ),
5609     "uri" => "http://search.thunderstone.com/"
5610     ),
5611     "timbobot" => array(
5612     "icon" => "robot",
5613 matthys 209 "title" => "TimboBot",
5614 joku 63 "rule" => array(
5615     "timboBot" => ""
5616     ),
5617     "uri" => ""
5618     ),
5619     "topsy" => array(
5620     "icon" => "topsy",
5621     "title" => "Topsy",
5622     "rule" => array(
5623     "Butterfly[ /]([0-9.]{1,10})" => "\\1"
5624     ),
5625     "uri" => "http://labs.topsy.com/butterfly.html"
5626     ),
5627     "trayce" => array(
5628     "icon" => "robot",
5629 matthys 209 "title" => "Trayce",
5630 joku 63 "rule" => array(
5631     "traycebot[ /]([0-9a-z.\-]{1,10})" => "\\1"
5632     ),
5633     "uri" => ""
5634     ),
5635     "tricus" => array(
5636     "icon" => "robot",
5637     "title" => "Tricus",
5638     "rule" => array(
5639     "B_l_i_t_z_B_O_T_@_t_r_i_c_u_s_\._c_o_m" => ""
5640     ),
5641     "uri" => ""
5642     ),
5643     "topicblogs" => array(
5644     "icon" => "robot",
5645     "title" => "Topicblogs",
5646     "rule" => array(
5647     "topicblogs[ /]([0-9.]{1,10})" => "\\1"
5648     ),
5649     "uri" => ""
5650     ),
5651     "touche" => array(
5652     "icon" => "touche",
5653 matthys 209 "title" => "Touché",
5654 joku 63 "rule" => array(
5655     "^Touche" => ""
5656     ),
5657     "uri" => "http://touche.com.ve"
5658     ),
5659     "tuezilla" => array(
5660     "icon" => "robot",
5661 matthys 209 "title" => "TÃœzilla",
5662 joku 63 "rule" => array(
5663     "tuezilla.de" => ""
5664     ),
5665     "uri" => "http://tuezilla.de/t_st-odp-entries-agent.html"
5666     ),
5667     "turnitin" => array(
5668     "icon" => "turnitin",
5669     "title" => "Turnitin",
5670     "rule" => array(
5671     "TurnitinBot[ /]([0-9.]{1,10})" => "\\1"
5672     ),
5673     "uri" => ""
5674     ),
5675     "tutorgig" => array(
5676     "icon" => "robot",
5677     "title" => "TutorGig",
5678     "rule" => array(
5679     "TutorGig(Bot)?[ /]([0-9.]{1,10})" => "\\2"
5680     ),
5681     "uri" => ""
5682     ),
5683     "twiceler" => array(
5684     "icon" => "cuill",
5685 matthys 209 "title" => "Cuill",
5686 joku 63 "rule" => array(
5687     "Twiceler[ /-]([0-9.]{1,10})" => "\\1",
5688     "Twiceler" => ""
5689     ),
5690     "uri" => "http://www.cuill.com/twiceler/robot.html"
5691     ),
5692     "twingly" => array(
5693     "icon" => "twingly",
5694     "title" => "Twingly",
5695     "rule" => array(
5696     "Twingly Recon" => ""
5697     ),
5698     "uri" => "http://www.twingly.com/"
5699     ),
5700 matthys 320 "twitter" => array(
5701     "icon" => "twitter",
5702     "title" => "Twitter",
5703     "rule" => array(
5704     "Twitterbot[ /-]([0-9.]{1,10})" => "\\1"
5705     ),
5706     "uri" => "https://dev.twitter.com/cards/getting-started#crawling"
5707     ),
5708 joku 63 "typepad" => array(
5709     "icon" => "typepad",
5710     "title" => "TypePad",
5711     "rule" => array(
5712     "TypePad/([0-9a-z.]{1,10})" => "\\1"
5713     ),
5714     "uri" => ""
5715     ),
5716     "udmsearch" => array(
5717     "icon" => "robot",
5718     "title" => "UdmSearch",
5719     "rule" => array(
5720     "UdmSearch[/ ]([0-9.]{1,10})" => "\\1"
5721     ),
5722     "uri" => ""
5723     ),
5724     "ukwizz" => array(
5725     "icon" => "ukwizz",
5726     "title" => "UKWizz",
5727     "rule" => array(
5728     "^Mackster.*ukwizz" => ""
5729     ),
5730     "uri" => "http://www.ukwizz.com"
5731     ),
5732     "ultraseek" => array(
5733     "icon" => "robot",
5734     "title" => "Ultraseek",
5735     "rule" => array(
5736     "Ultraseek" => ""
5737     ),
5738     "uri" => ""
5739     ),
5740     "ultraspider" => array(
5741     "icon" => "robot",
5742     "title" => "UltraSpider",
5743     "rule" => array(
5744     "UltraSpider3000[/ ]([0-9.]{1,10})" => "\\1"
5745     ),
5746     "uri" => "http://www.search.ch"
5747     ),
5748     "umai" => array(
5749     "icon" => "robot",
5750 matthys 209 "title" => "Umai",
5751 joku 63 "rule" => array(
5752     "umai[/ ]([0-9.]{1,10})" => "\\1"
5753     ),
5754     "uri" => ""
5755     ),
5756     "unchaos" => array(
5757     "icon" => "robot",
5758     "title" => "Unchaos",
5759     "rule" => array(
5760     "unchaos_crawler[_ /]([0-9.]{1,10})" => "\\1",
5761     "unchaos" => ""
5762     ),
5763     "uri" => ""
5764     ),
5765     "unido" => array(
5766     "icon" => "robot",
5767 matthys 209 "title" => "UNIdo",
5768 joku 63 "rule" => array(
5769     "^unido-bot" => "\\1"
5770     ),
5771     "uri" => "http://mobicom.cs.uni-dortmund.de/bot.html"
5772     ),
5773     "updated" => array(
5774     "icon" => "robot",
5775     "title" => "Updated",
5776     "rule" => array(
5777     "updated[ /]([0-9a-z.]{1,10})" => "\\1"
5778     ),
5779     "uri" => ""
5780     ),
5781     "uptimebot" => array(
5782     "icon" => "robot",
5783     "title" => "UptimeBot",
5784     "rule" => array(
5785     "^UptimeBot" => ""
5786     ),
5787     "uri" => "http://www.uptimebot.com/"
5788     ),
5789     "urifetch" => array(
5790     "icon" => "robot",
5791     "title" => "URI::Fetch",
5792     "rule" => array(
5793     "^URI::Fetch[ /]([0-9.]{1,10})" => "\\1"
5794     ),
5795     "uri" => ""
5796     ),
5797     "urlbase" => array(
5798     "icon" => "robot",
5799     "title" => "URLBase",
5800     "rule" => array(
5801     "URLBase[ /]([0-9.]{1,10})" => "\\1"
5802     ),
5803     "uri" => ""
5804     ),
5805     "urlblaze" => array(
5806     "icon" => "robot",
5807     "title" => "URLBlaze",
5808     "rule" => array(
5809     "^URLBlaze" => ""
5810     ),
5811     "uri" => ""
5812     ),
5813     "urlcontr" => array(
5814     "icon" => "robot",
5815     "title" => "MS URL Control",
5816     "rule" => array(
5817     "Microsoft URL[ \-]?Control" => ""
5818     ),
5819     "uri" => ""
5820     ),
5821     "urlgetfile" => array(
5822     "icon" => "robot",
5823     "title" => "URLGetFile",
5824     "rule" => array(
5825     "^URLGetFile" => ""
5826     ),
5827     "uri" => ""
5828     ),
5829     "urlscope" => array(
5830     "icon" => "robot",
5831     "title" => "UrlScope",
5832     "rule" => array(
5833     "UrlScope" => ""
5834     ),
5835     "uri" => ""
5836     ),
5837     "urltrends" => array(
5838     "icon" => "urltrends",
5839 matthys 209 "title" => "UrlTrends",
5840 joku 63 "rule" => array(
5841     "Snappy/([0-9.]{1,10})" => "\\1",
5842     ),
5843     "uri" => ""
5844     ),
5845     "usww" => array(
5846     "icon" => "usww",
5847 matthys 209 "title" => "USWW",
5848 joku 63 "rule" => array(
5849     "usww\.com" => "",
5850     "Mozilla/5\.0 URL-Spider" => ""
5851     ),
5852     "uri" => "http://www.usww.com/"
5853     ),
5854     "usyd" => array(
5855     "icon" => "robot",
5856     "title" => "USyd-NLP-Spider",
5857     "rule" => array(
5858     "^USyd-NLP-Spider" => ""
5859     ),
5860     "uri" => "http://www.it.usyd.edu.au/~vinci/bot.html"
5861     ),
5862     "vagabondo" => array(
5863     "icon" => "wiseguys",
5864     "title" => "WiseGuys",
5865     "rule" => array(
5866     "Vagabondo[ /]([0-9.]{1,10})" => "\\1",
5867     "Vagabondo-WAP[ /]([0-9.]{1,10})" => "\\1"
5868     ),
5869     "uri" => ""
5870     ),
5871     "unicorn" => array(
5872     "icon" => "w3c",
5873     "title" => "W3C Unicorn",
5874     "rule" => array(
5875     "Unicorn[ /]([0-9.]{1,10})" => "\\1",
5876     ),
5877     "uri" => "http://validator.w3.org/unicorn/"
5878     ),
5879     "validator" => array(
5880     "icon" => "w3c",
5881     "title" => "W3C Validator",
5882     "rule" => array(
5883     "W3C_Validator[ /]([0-9.]{1,10})" => "\\1"
5884     ),
5885     "uri" => "http://validator.w3.org/"
5886     ),
5887     "validator.nu" => array(
5888     "icon" => "validatornu",
5889     "title" => "Validator.nu",
5890     "rule" => array(
5891     "^Validator.nu[ /]([0-9.]{1,10})" => "\\1",
5892     "^Validator.nu/LV" => "\\1"
5893     ),
5894     "uri" => "http://validator.nu/"
5895     ),
5896     "verity" => array(
5897     "icon" => "robot",
5898     "title" => "Verity",
5899     "rule" => array(
5900     "^vspider[ /]([0-9.]{1,10})" => "\\1",
5901     "^vspider" => ""
5902     ),
5903     "uri" => "http://www.verity.com/"
5904     ),
5905     "versionsproject" => array(
5906     "icon" => "robot",
5907     "title" => "Versions-project",
5908     "rule" => array(
5909     "InfoFly[ /]([0-9.]{1,10})" => "\\1"
5910     ),
5911     "uri" => "http://www.versions-project.org/"
5912     ),
5913     "verticalmatch" => array(
5914     "icon" => "robot",
5915     "title" => "VerticalMatch",
5916     "rule" => array(
5917     "^VMBot[ /]([0-9.]{1,10})" => "\\1"
5918     ),
5919     "uri" => "http://www.VerticalMatch.com/"
5920     ),
5921     "verzamelgids" => array(
5922     "icon" => "robot",
5923     "title" => "Verzamelgids",
5924     "rule" => array(
5925     "Verzamelgids[ /]([0-9.]{1,10})" => "\\1"
5926     ),
5927     "uri" => "http://www.verzamelgids.nl/"
5928     ),
5929     "vestris" => array(
5930     "icon" => "vestris",
5931     "title" => "Vestris",
5932     "rule" => array(
5933     "AlkalineBOT[ /]([0-9.]{1,10})" => "\\1"
5934     ),
5935     "uri" => "http://alkaline.vestris.com/"
5936     ),
5937     "vindex" => array(
5938     "icon" => "vindex",
5939     "title" => "Vindex",
5940     "rule" => array(
5941     "Vindex[ /]([0-9.]{1,10})" => "\\1"
5942     ),
5943     "uri" => ""
5944     ),
5945     "visvo" => array(
5946     "icon" => "robot",
5947     "title" => "Visvo",
5948     "rule" => array(
5949     "VisBot[ /]([0-9.]{1,10})" => "\\1"
5950     ),
5951     "uri" => "http://www.visvo.com"
5952     ),
5953     "voila" => array(
5954     "icon" => "voila",
5955     "title" => "Voila",
5956     "rule" => array(
5957     "VoilaBot[ /]?[a-z ]*([0-9.]{1,10})" => "\\1",
5958     "VoilaBot;[ /]([0-9.]{1,10})" => "\\1"
5959     ),
5960     "uri" => ""
5961     ),
5962     "vonna" => array(
5963     "icon" => "robot",
5964     "title" => "Vonna",
5965     "rule" => array(
5966     "Vonna.com b o t" => ""
5967     ),
5968     "uri" => ""
5969     ),
5970     "vortex" => array(
5971     "icon" => "robot",
5972     "title" => "Vortex",
5973     "rule" => array(
5974     "Vortex[ /]([0-9.]{1,10})" => "\\1"
5975     ),
5976     "uri" => "http://marty.anstey.ca/robots/vortex/"
5977     ),
5978     "w3sitesearch" => array(
5979     "icon" => "w3sitesearch",
5980     "title" => "W3SiteSearch",
5981     "rule" => array(
5982     "^W3SiteSearch Crawler[\_v]*([0-9.]{1,10})" => "\\1"
5983     ),
5984     "uri" => "http://www.w3sitesearch.de"
5985     ),
5986     "wagger" => array(
5987     "icon" => "robot",
5988     "title" => "Wagger",
5989     "rule" => array(
5990     "^Waggr" => ""
5991     ),
5992     "uri" => "http://www.waggr.com/"
5993     ),
5994     "wanadoo" => array(
5995     "icon" => "robot",
5996     "title" => "Wanadoo",
5997     "rule" => array(
5998     "^SurferF3[ /]([0-9./]{1,10})" => "\\1"
5999     ),
6000     "uri" => "http://www.wanadoo.fr/"
6001     ),
6002     "wapalizer" => array(
6003     "icon" => "robot",
6004     "title" => "Wapalizer",
6005     "rule" => array(
6006     "wapalizer[ /]([0-9.]{1,10})" => "\\1"
6007     ),
6008     "uri" => "http://www.wapdrive.com/"
6009     ),
6010     "watson" => array(
6011     "icon" => "addy",
6012     "title" => "Dr.Watson",
6013     "rule" => array(
6014     "Watson[ /]([0-9.]{1,10})" => "\\1",
6015     "watson\.addy\.com" => ""
6016     ),
6017     "uri" => ""
6018     ),
6019     "wavefire" => array(
6020     "icon" => "robot",
6021     "title" => "Wavefire",
6022     "rule" => array(
6023     "^Wavefire[ /]([0-9.]{1,10})" => "\\1"
6024     ),
6025     "uri" => ""
6026     ),
6027     "waypath" => array(
6028     "icon" => "waypath",
6029     "title" => "Waypath",
6030     "rule" => array(
6031     "Waypath[ \-]?Scout" => "",
6032     "Waypath (development )?crawler" => ""
6033     ),
6034     "uri" => ""
6035     ),
6036     "wauuu" => array(
6037     "icon" => "wauuu",
6038     "title" => "Wauuu",
6039     "rule" => array(
6040     "Wauuu[/ ]([0-9.]{1,10})" => "\\2"
6041     ),
6042     "uri" => "http://www.wauuu.com"
6043     ),
6044     "wdg" => array(
6045     "icon" => "wdg",
6046     "title" => "WDG Validator",
6047     "rule" => array(
6048     "^WDG_(Site)?Validator[/ ]([0-9.]{1,10})" => "\\2"
6049     ),
6050     "uri" => "http://www.htmlhelp.com/tools/validator/"
6051     ),
6052     "webagogo" => array(
6053     "icon" => "webagogo",
6054     "title" => "Webagogo",
6055     "rule" => array(
6056     "^Webagogo" => ""
6057     ),
6058     "uri" => "http://www.webagogo.be/"
6059     ),
6060     "webalta" => array(
6061     "icon" => "webalta",
6062     "title" => "WebAlta",
6063     "rule" => array(
6064     "^WebAlta( Crawler)?[/ ]([0-9.]{1,10})" => "\\2"
6065     ),
6066     "uri" => "http://www.webalta.net/ru/about_webmaster.html"
6067     ),
6068     "webdigity" => array(
6069     "icon" => "robot",
6070     "title" => "Webdigity Whois Service",
6071     "rule" => array(
6072     "^webdigity whois service" => ""
6073     ),
6074     "uri" => "http://www.webdigity.com/ws"
6075     ),
6076     "webelixir" => array(
6077     "icon" => "webelixir",
6078     "title" => "Webelixir",
6079     "rule" => array(
6080     "^webelixir" => ""
6081     ),
6082     "uri" => "http://www.webelixir.net/"
6083     ),
6084     "webbotru" => array(
6085     "icon" => "robot",
6086     "title" => "Webbot.ru",
6087     "rule" => array(
6088     " Webbot[/ ]([0-9.]{1,10})" => "\\1"
6089     ),
6090     "uri" => "http://www.webbot.ru/bot.html"
6091     ),
6092     "webcapture" => array(
6093     "icon" => "robot",
6094     "title" => "WebCapture",
6095     "rule" => array(
6096     "WebCapture[/ ]([0-9.]{1,10})" => "\\1"
6097     ),
6098     "uri" => ""
6099     ),
6100     "webcollage" => array(
6101     "icon" => "robot",
6102     "title" => "Webcollage",
6103     "rule" => array(
6104     "webcollage" => ""
6105     ),
6106     "uri" => ""
6107     ),
6108     "webcopier" => array(
6109     "icon" => "webcopier",
6110     "title" => "WebCopier",
6111     "rule" => array(
6112     "WebCopier[/ ]v?([0-9.]{1,10})" => "\\1"
6113     ),
6114     "uri" => ""
6115     ),
6116     "webcorp" => array(
6117     "icon" => "webcorp",
6118     "title" => "WebCorp",
6119     "rule" => array(
6120     "WebCorp[/ ]+([0-9.]{1,10})" => "\\1"
6121     ),
6122     "uri" => "http://www.webcorp.org.uk/"
6123     ),
6124     "webcrawl" => array(
6125     "icon" => "robot",
6126     "title" => "WebCrawl",
6127     "rule" => array(
6128     "webcrawl\.net" => ""
6129     ),
6130     "uri" => ""
6131     ),
6132     "webdownloader" => array(
6133     "icon" => "robot",
6134     "title" => "Web Downloader",
6135     "rule" => array(
6136     "Web Downloader[/ ]([0-9.]{1,10})" => "\\1"
6137     ),
6138     "uri" => ""
6139     ),
6140     "webfetch" => array(
6141     "icon" => "robot",
6142 matthys 209 "title" => "WebFetch",
6143 joku 63 "rule" => array(
6144     "^webfetch[/ ]([0-9.]{1,10})" => "\\1"
6145     ),
6146     "uri" => ""
6147     ),
6148     "webfind" => array(
6149     "icon" => "robot",
6150 matthys 209 "title" => "WebFind",
6151 joku 63 "rule" => array(
6152     "^WebFindBot" => ""
6153     ),
6154     "uri" => "http://www.web-find.com"
6155     ),
6156     "webglimpse" => array(
6157     "icon" => "robot",
6158     "title" => "Webglimpse",
6159     "rule" => array(
6160     "^Webglimpse[/ ]([0-9.]{1,10})" => "\\1"
6161     ),
6162     "uri" => "http://webglimpse.net"
6163     ),
6164     "webgobbler" => array(
6165     "icon" => "robot",
6166     "title" => "webGobbler",
6167     "rule" => array(
6168     "^webGobbler[/ ]([0-9.]{1,10})" => "\\1"
6169     ),
6170     "uri" => ""
6171     ),
6172     "webimages" => array(
6173     "icon" => "webimages",
6174     "title" => "WebImages",
6175     "rule" => array(
6176     "^WebImages[/ ]([0-9.]{1,10})" => "\\1"
6177     ),
6178     "uri" => "http://herbert.groot.jebbink.nl/"
6179     ),
6180     "weblight" => array(
6181     "icon" => "robot",
6182     "title" => "WebLight",
6183     "rule" => array(
6184     "^WebLight[/ ]([0-9.]{1,10})" => "\\1"
6185     ),
6186     "uri" => "http://www.illumit.com/Products/weblight/"
6187     ),
6188     "weblinks" => array(
6189     "icon" => "robot",
6190     "title" => "WebLink's",
6191     "rule" => array(
6192     "^Weblink.s checker" => ""
6193     ),
6194     "uri" => ""
6195     ),
6196     "webmeasurement" => array(
6197     "icon" => "robot",
6198     "title" => "Webmeasurement",
6199     "rule" => array(
6200     "^webmeasurement-bot" => ""
6201     ),
6202     "uri" => "http://rvs.informatik.uni-leipzig.de"
6203     ),
6204     "webminer" => array(
6205     "icon" => "robot",
6206     "title" => "WebMiner",
6207     "rule" => array(
6208     "^WebMiner[/ ]([0-9.]{1,10})" => "\\1"
6209     ),
6210     "uri" => ""
6211     ),
6212     "webmin" => array(
6213     "icon" => "webmin",
6214     "title" => "Webmin",
6215     "rule" => array(
6216     "^webmin" => ""
6217     ),
6218     "uri" => ""
6219     ),
6220     "webmon" => array(
6221     "icon" => "webmon",
6222     "title" => "Webmon",
6223     "rule" => array(
6224     "WebMon[ /]([0-9.]{1,10})" => "\\1"
6225     ),
6226     "uri" => "http://www.markwell.btinternet.co.uk/webmon/"
6227     ),
6228     "webox" => array(
6229     "icon" => "robot",
6230     "title" => " WeBoX",
6231     "rule" => array(
6232     "WeBoX[/ ]([0-9.]{1,10})" => "\\1"
6233     ),
6234     "uri" => ""
6235     ),
6236     "webpatrol" => array(
6237     "icon" => "webpatrol",
6238     "title" => "WebPatrol",
6239     "rule" => array(
6240     "^WebPatrol[ /]([0-9.]{1,10})" => "\\1"
6241     ),
6242     "uri" => "http://soft.macfeeling.com/WebPatrol.html"
6243     ),
6244     "webpix" => array(
6245     "icon" => "webpix",
6246     "title" => "WebPix",
6247     "rule" => array(
6248     "WebPix[/ ]([0-9.]{1,10})" => "\\1"
6249     ),
6250     "uri" => ""
6251     ),
6252     "webrace" => array(
6253     "icon" => "robot",
6254     "title" => "WebRACE",
6255     "rule" => array(
6256     "^WebRACE[/ ]([0-9.]{1,10})" => "\\1"
6257     ),
6258     "uri" => ""
6259     ),
6260     "webreaper" => array(
6261     "icon" => "webreaper",
6262     "title" => "WebReaper",
6263     "rule" => array(
6264     "^WebReaper " => ""
6265     ),
6266     "uri" => "http://www.webreaper.net/"
6267     ),
6268     "webresult" => array(
6269     "icon" => "robot",
6270     "title" => "Webresult",
6271     "rule" => array(
6272     "Der webresult\.de Robot" => ""
6273     ),
6274     "uri" => ""
6275     ),
6276     "webring" => array(
6277     "icon" => "robot",
6278     "title" => "Webring Checker",
6279     "rule" => array(
6280     "WebRingChecker[/ ]([0-9.]{1,10})" => "\\1"
6281     ),
6282     "uri" => ""
6283     ),
6284     "webripper" => array(
6285     "icon" => "webripper",
6286     "title" => "WebRipper",
6287     "rule" => array(
6288     "^WebRipper" => ""
6289     ),
6290     "uri" => "http://calluna-software.com"
6291     ),
6292     "webscaled" => array(
6293     "icon" => "webscaled",
6294     "title" => "Webscaled",
6295     "rule" => array(
6296     "Swarm" => ""
6297     ),
6298     "uri" => "http://webscaled.com/"
6299     ),
6300     "websearchau" => array(
6301     "icon" => "websearchau",
6302     "title" => "WebSearch.COM.AU",
6303     "rule" => array(
6304     "WebSearch.COM.AU[/ ]+([0-9.]{1,10})" => "\\1"
6305     ),
6306     "uri" => "http://WebSearch.com.au/"
6307     ),
6308     "websearchbench" => array(
6309     "icon" => "robot",
6310     "title" => "WebSearchBench",
6311     "rule" => array(
6312     "WebSearchBench WebCrawler[v/ ]+([0-9.]{1,10})" => "\\1"
6313     ),
6314     "uri" => "http://websearchbench.cs.uni-dortmund.de/"
6315     ),
6316     "websense" => array(
6317     "icon" => "websense",
6318     "title" => "Websense",
6319     "rule" => array(
6320     "(Sqworm|websense|Konqueror/3\.(0|1)(-rc[1-6])?; i686 Linux; 2002[0-9]{4})" => ""
6321     ),
6322     "uri" => ""
6323     ),
6324     "websiteshadow" => array(
6325     "icon" => "websiteshadow",
6326     "title" => "Websiteshadow",
6327     "rule" => array(
6328     "Websiteshadowbot[ /]([0-9.]{1,10})" => "\\1"
6329     ),
6330     "uri" => "http://websiteshadow.com"
6331     ),
6332     "websiteworth" => array(
6333     "icon" => "robot",
6334     "title" => "WebsiteWorth",
6335     "rule" => array(
6336     "WebsiteWorth[v/ ]+([0-9.]{1,10})" => "\\1"
6337     ),
6338     "uri" => "http://directory.sootle.com/website-worth/tata.php"
6339     ),
6340     "websquash" => array(
6341     "icon" => "websquash",
6342     "title" => "Websquash",
6343     "rule" => array(
6344     "webs(quash\.com|ite[ \-]?Monitor)" => ""
6345     ),
6346     "uri" => ""
6347     ),
6348     "webstripper" => array(
6349     "icon" => "robot",
6350     "title" => "WebStripper",
6351     "rule" => array(
6352     "WebStripper[ /]([0-9.]{1,10})" => "\\1"
6353     ),
6354     "uri" => ""
6355     ),
6356     "webzip" => array(
6357     "icon" => "webzip",
6358     "title" => "WebZIP",
6359     "rule" => array(
6360     "Web[ \-]?ZIP[ /]([0-9.]{1,10})" => "\\1"
6361     ),
6362     "uri" => ""
6363     ),
6364     "wep" => array(
6365     "icon" => "robot",
6366     "title" => "WEP Search",
6367     "rule" => array(
6368     "WEP Search[ /]([0-9.]{1,10})" => "\\1"
6369     ),
6370     "uri" => ""
6371     ),
6372     "westwind" => array(
6373     "icon" => "robot",
6374     "title" => "West Wind Internet Protocols",
6375     "rule" => array(
6376     "^West Wind Internet Protocols[ /]([0-9.]{1,10})" => "\\1"
6377     ),
6378     "uri" => "http://www.west-wind.com/wwipstuff.asp"
6379     ),
6380     "wget" => array(
6381     "icon" => "wget",
6382     "title" => "Wget",
6383     "rule" => array(
6384 joku 308 "Wget[ /]([0-9.]{1,10})" => "\\1",
6385     "Wget" => ""
6386 joku 63 ),
6387     "uri" => ""
6388     ),
6389     "whizbang" => array(
6390     "icon" => "whizbang",
6391     "title" => "WhizBang",
6392     "rule" => array(
6393     "WhizBang" => ""
6394     ),
6395     "uri" => "http://www.whizbang.com/crawler/"
6396     ),
6397     "whois" => array(
6398     "icon" => "whois",
6399     "title" => "Who.is",
6400     "rule" => array(
6401     "Who.is Bot" => ""
6402     ),
6403     "uri" => "http://www.who.is/"
6404     ),
6405     "wikiwix" => array(
6406     "icon" => "wikiwix",
6407     "title" => "Wikiwix",
6408     "rule" => array(
6409     "wikiwix-bot-([0-9.]{1,10})" => "\\1"
6410     ),
6411     "uri" => "http://www.wikiwix.com/"
6412     ),
6413     "wingflyer" => array(
6414     "icon" => "robot",
6415     "title" => "WingFlyer",
6416     "rule" => array(
6417     "^WebFetch" => ""
6418     ),
6419     "uri" => "http://www.wingflyer.com/"
6420     ),
6421     "wininet" => array(
6422     "icon" => "robot",
6423     "title" => "WinInet",
6424     "rule" => array(
6425     "TeamSoft WinInet Component" => ""
6426     ),
6427     "uri" => "http://www.winsoft.sk/wininet.htm"
6428     ),
6429     "winhttp" => array(
6430     "icon" => "robot",
6431     "title" => "WinHTTP",
6432     "rule" => array(
6433     "WinHttp\.WinHttpRequest\.([0-9.]{1,10})" => "\\1"
6434     ),
6435     "uri" => ""
6436     ),
6437     "wire" => array(
6438     "icon" => "robot",
6439     "title" => "WIRE",
6440     "rule" => array(
6441     "^WIRE[ /]([0-9.]{1,10})" => "\\1"
6442     ),
6443     "uri" => ""
6444     ),
6445     "wmp" => array(
6446     "icon" => "robot",
6447     "title" => "WMP",
6448     "rule" => array(
6449     "^WMP" => ""
6450     ),
6451     "uri" => ""
6452     ),
6453     "woozweb" => array(
6454     "icon" => "woozweb",
6455     "title" => "Woozweb",
6456     "rule" => array(
6457     "woozweb" => ""
6458     ),
6459     "uri" => "http://www.woozweb.com/"
6460     ),
6461     "wordchamp" => array(
6462     "icon" => "robot",
6463     "title" => "WordChamp",
6464     "rule" => array(
6465     "^WordChampBot" => ""
6466     ),
6467     "uri" => "http://www.wordchamp.com/"
6468     ),
6469     "wordpress" => array(
6470     "icon" => "wordpress",
6471     "title" => "WordPress",
6472     "rule" => array(
6473     "WordPress[ /]([0-9.]{1,10})" => "\\1"
6474     ),
6475     "uri" => ""
6476     ),
6477     "worio" => array(
6478     "icon" => "worio",
6479     "title" => "Worio",
6480     "rule" => array(
6481     "woriobot" => ""
6482     ),
6483     "uri" => "http://worio.com/"
6484     ),
6485     "worldlight" => array(
6486     "icon" => "worldlight",
6487     "title" => "WorldLight",
6488     "rule" => array(
6489     "^WorldLight" => ""
6490     ),
6491     "uri" => ""
6492     ),
6493     "worqmada" => array(
6494     "icon" => "robot",
6495     "title" => "WorQmada",
6496     "rule" => array(
6497     "WorQmada[ /]([0-9.]{1,10})" => "\\1"
6498     ),
6499     "uri" => ""
6500     ),
6501     "wotbox" => array(
6502     "icon" => "wotbox",
6503     "title" => "Wotbox",
6504     "rule" => array(
6505     "Wotbox[ /]?[a-z]*([0-9.]{1,10})" => "\\1"
6506     ),
6507     "uri" => ""
6508     ),
6509     "wp" => array(
6510     "icon" => "wp",
6511     "title" => "Wirtualna Polska",
6512     "rule" => array(
6513     "NetSprint[ /\-]{1,4}([0-9.]{1,10})" => "\\1"
6514     ),
6515     "uri" => "http://wp.pl"
6516     ),
6517     "wsb" => array(
6518     "icon" => "robot",
6519     "title" => "WebSearchBench",
6520     "rule" => array(
6521     "WSB WebCrawler V([0-9.]{1,10})" => "\\1",
6522     "WSB " => ""
6523     ),
6524     "uri" => "http://websearchbench.cs.uni-dortmund.de/"
6525     ),
6526     "wume" => array(
6527     "icon" => "robot",
6528     "title" => "WUME Lab's",
6529     "rule" => array(
6530     "^wume_crawler[ /]([0-9.]{1,10})" => "\\1",
6531     ),
6532     "uri" => "http://wume.cse.lehigh.edu/~xiq204/crawler/"
6533     ),
6534     "wusage" => array(
6535     "icon" => "wusage",
6536     "title" => "Wusage",
6537     "rule" => array(
6538     "Wusage[ /]([0-9.]{1,10})" => "\\1",
6539     ),
6540     "uri" => "http://www.boutell.com/wusage/"
6541     ),
6542     "wwgrapevine" => array(
6543     "icon" => "wwgrapevine",
6544     "title" => "WWgrapevine",
6545     "rule" => array(
6546     "wwgrapevine[ /]([0-9.]{1,10})" => "\\1"
6547     ),
6548     "uri" => ""
6549     ),
6550     "wws" => array(
6551     "icon" => "robot",
6552     "title" => "WWSBOT",
6553     "rule" => array(
6554     "WWSBOT[ /]([0-9.]{1,10})" => "\\1",
6555     ),
6556     "uri" => "http://www.analyzer.nu"
6557     ),
6558     "www4mail" => array(
6559     "icon" => "robot",
6560 matthys 209 "title" => "WWW4mail",
6561 joku 63 "rule" => array(
6562     "^www4mail[ /]([0-9.]{1,10})" => "\\1"
6563     ),
6564     "uri" => "http://www.www4mail.org/"
6565     ),
6566     "wwwc" => array(
6567     "icon" => "wwwc",
6568     "title" => "WWWC",
6569     "rule" => array(
6570     "^WWWC[ /]([0-9.]{1,10})" => "\\1"
6571     ),
6572     "uri" => ""
6573     ),
6574     "wwwd" => array(
6575     "icon" => "robot",
6576     "title" => "WWWD",
6577     "rule" => array(
6578     "^WWWD[ /]([0-9.]{1,10})" => "\\1"
6579     ),
6580     "uri" => ""
6581     ),
6582     "wwweasel" => array(
6583     "icon" => "wwweasel",
6584     "title" => "WWWeasel",
6585     "rule" => array(
6586     "WWWeasel( Robot)?[/ ]v?([0-9.]{1,10})" => "\\2"
6587     ),
6588     "uri" => ""
6589     ),
6590     "wwwfi" => array(
6591     "icon" => "wwwfi",
6592     "title" => "www.fi",
6593     "rule" => array(
6594     "www\.fi crawler" => ""
6595     ),
6596     "uri" => "http://www.fi/"
6597     ),
6598     "wwwmechanize" => array(
6599     "icon" => "robot",
6600     "title" => "WWW-Mechanize",
6601     "rule" => array(
6602     "^WWW-Mechanize[/ ]([0-9.]{1,10})" => "\\1"
6603     ),
6604     "uri" => ""
6605     ),
6606     "wwwoffle" => array(
6607     "icon" => "robot",
6608     "title" => "WWWoffle",
6609     "rule" => array(
6610     "^wwwoffle[/ ]([0-9.]{1,10})" => "\\1"
6611     ),
6612     "uri" => ""
6613     ),
6614     "wwwster" => array(
6615     "icon" => "robot",
6616     "title" => "WWWster",
6617     "rule" => array(
6618     "^wwwster[/ ]([0-9.]{1,10})" => "\\1"
6619     ),
6620     "uri" => ""
6621     ),
6622     "wysigot" => array(
6623     "icon" => "wysigot",
6624     "title" => "Wysigot",
6625     "rule" => array(
6626     "Wysigot[/ ]([0-9.]{1,10})" => "\\1"
6627     ),
6628     "uri" => ""
6629     ),
6630     "xaldon" => array(
6631     "icon" => "xaldon",
6632     "title" => "Xaldon",
6633     "rule" => array(
6634     "Xaldon WebSpider" => ""
6635     ),
6636     "uri" => "http://www.xaldon.de/"
6637     ),
6638     "xenu" => array(
6639     "icon" => "xenu",
6640     "title" => "Xenu Link Sleuth",
6641     "rule" => array(
6642     "Xenu Link Sleuth[/ ]([0-9a-z.]{1,10})" => "\\1",
6643     "Xenu(&#039;s)? Link Sleuth[/ ]([0-9a-z.]{1,10})" => "\\1",
6644     "Xenu_Link_Sleuth_([0-9a-z.]{1,10})" => "\\1"
6645     ),
6646     "uri" => "http://home.snafu.de/tilman/xenulink.html"
6647     ),
6648     "xerka" => array(
6649     "icon" => "xerka",
6650     "title" => "Xerka",
6651     "rule" => array(
6652     "^Xerka WebBot v([0-9a-z.]{1,10})" => "\\1"
6653     ),
6654     "uri" => "http://www.diana-teknologia.com/www1/english/xerka.htm"
6655     ),
6656     "xirq" => array(
6657     "icon" => "robot",
6658     "title" => "XIRQ",
6659     "rule" => array(
6660     "^xirq[ /]([0-9a-z.]{1,10})" => "\\1"
6661     ),
6662     "uri" => "http://www.xirq.com"
6663     ),
6664     "xmlslurp" => array(
6665     "icon" => "robot",
6666     "title" => "XMLSlurp",
6667     "rule" => array(
6668     "^XMLSlurp[ /]([0-9a-z.]{1,10})" => "\\1"
6669     ),
6670     "uri" => ""
6671     ),
6672     "xml-sitemaps" => array(
6673     "icon" => "xml-sitemaps",
6674     "title" => "XML-Sitemaps",
6675     "rule" => array(
6676     "XML-Sitemaps[ /]([0-9a-z.]{1,10})" => "\\1"
6677     ),
6678     "uri" => ""
6679     ),
6680     "xmlrpc" => array(
6681     "icon" => "robot",
6682     "title" => "Trackback",
6683     "rule" => array(
6684     "XMLRPC" => ""
6685     ),
6686     "uri" => ""
6687     ),
6688 matthys 294 "xovi" => array(
6689     "icon" => "xovi",
6690     "title" => "Xovi",
6691     "rule" => array(
6692     "XoviBot[ /]([0-9a-z.]{1,10})" => "\\1"
6693     ),
6694     "uri" => "http://www.xovibot.net"
6695     ),
6696 joku 63 "yacy" => array(
6697     "icon" => "yacy",
6698     "title" => "Yacy",
6699     "rule" => array(
6700     "yacy\.net" => ""
6701     ),
6702     "uri" => ""
6703     ),
6704     "yahoo" => array(
6705     "icon" => "yahoo",
6706     "title" => "Yahoo",
6707     "rule" => array(
6708     "Yahoo(! ([a-z]{1,3} )?Slurp|-)" => "",
6709     "Yahoo-MMCrawler[/ ]([0-9a-z.]{1,10})" => "\\1",
6710     "Yahoo-VerticalCrawler-FormerWebCrawler[/ ]([0-9a-z.]{1,10})" => "\\1",
6711     "^AnzwersCrawl[/ ]([0-9a-z.]{1,10})" => "\\1",
6712     "Y!J(-BSC|-SRD)*[/ ]([0-9a-z.]{1,10})" => "\\2",
6713     "Y!OASIS/TEST" => "",
6714     "Harvest-NG[/ ]([0-9a-z.]{1,10})" => "\\1",
6715     "Y!J; for robot study" => "",
6716     "Yahoo Japan; for robot study" => ""
6717     ),
6718     "uri" => "http://www.yahoo.com"
6719     ),
6720     "yahoo-feed" => array(
6721     "icon" => "yahoo",
6722     "title" => "Yahoo Feedseeker",
6723     "rule" => array(
6724     "^YahooFeedSeeker[/ ]([0-9a-z.]{1,10})" => "\\1"
6725     ),
6726     "uri" => "http://www.yahoo.com"
6727     ),
6728     "yandex" => array(
6729     "icon" => "yandex",
6730     "title" => "Yandex",
6731     "rule" => array(
6732     "Yandex[/ ]([0-9.]{1,10})" => "\\1",
6733     "Yandex(Bot|Images|Media|Antivirus)[/ ]([0-9.]{1,10})" => "\\2"
6734     ),
6735     "uri" => "http://yandex.com/bots"
6736     ),
6737     "yanga" => array(
6738     "icon" => "yanga",
6739     "title" => "Yanga",
6740     "rule" => array (
6741     "^Yanga.*v([0-9.]{1,10})" => "\\1"
6742     ),
6743     "uri" => "http://www.yanga.co.uk/"
6744     ),
6745     "yap" => array(
6746     "icon" => "yap",
6747     "title" => "Yap",
6748     "rule" => array(
6749     "^YapBot[_/ ]v([0-9.]{1,10})" => "\\1"
6750     ),
6751     "uri" => "http://www.yapinc.com/"
6752     ),
6753     "yarienavoir" => array(
6754     "icon" => "robot",
6755     "title" => "Yarienavoir",
6756     "rule" => array(
6757     "^yarienavoir.net[/ ]([0-9.]{1,10})" => "\\1"
6758     ),
6759     "uri" => "http://www.yarienavoir.net/"
6760     ),
6761     "yell" => array(
6762     "icon" => "yell",
6763     "title" => "Yell",
6764     "rule" => array(
6765     "YellCrawl[ /]V?([0-9.]{1,10})" => "\\1",
6766     "Yellbot[ /]Nutch-([0-9.]{1,10})" => "\\1",
6767     ),
6768     "uri" => ""
6769     ),
6770     "youdao" => array(
6771     "icon" => "youdao",
6772     "title" => "Youdao",
6773     "rule" => array(
6774     "YodaoBot-(Image|Mobile)?[ /]([0-9.]{1,10})" => "\\2",
6775     "YoudaoBot[ /]([0-9.]{1,10})" => "\\1"
6776     ),
6777     "uri" => "http://www.youdao.com/help/webmaster/spider/"
6778     ),
6779     "yoogli" => array(
6780     "icon" => "yoogli",
6781     "title" => "Yoogli",
6782     "rule" => array(
6783     "yoogliFetchAgent[ /]([0-9.]{1,10})" => "\\1"
6784     ),
6785     "uri" => "http://www.yoogli.com"
6786     ),
6787     "yotta" => array(
6788     "icon" => "robot",
6789     "title" => "Yotta",
6790     "rule" => array(
6791     "Yotta(Shopping|Cars)_Bot[ /]([0-9.]{1,10})" => "\\2",
6792     "OmniExplorer_Bot[ /]([0-9.]{1,10})" => "\\1"
6793     ),
6794     "uri" => "http://www.yottacars.com"
6795     ),
6796     "yoono" => array(
6797     "icon" => "yoono",
6798     "title" => "Yoono",
6799     "rule" => array(
6800     "Yoono" => ""
6801     ),
6802     "uri" => "http://www.yoono.com/"
6803     ),
6804     "yowedo" => array(
6805     "icon" => "robot",
6806     "title" => "Yowedo",
6807     "rule" => array(
6808     "Yowedo[ /]([0-9.]{1,10})" => "\\1"
6809     ),
6810     "uri" => "http://yowedo.com/en/partners.html"
6811     ),
6812     "yuntis" => array(
6813     "icon" => "robot",
6814 matthys 209 "title" => "Yuntis",
6815 joku 63 "rule" => array(
6816     "Gulper Web Bot[ /]([0-9.]{1,10})" => "\\1"
6817     ),
6818     "uri" => "http://yuntis.ecsl.cs.sunysb.edu/help/robot/"
6819     ),
6820     "zao" => array(
6821     "icon" => "robot",
6822     "title" => "Zao",
6823     "rule" => array(
6824     "Zao[ /]([0-9.]{1,10})" => "\\1",
6825     "Zao-crawler" => ""
6826     ),
6827     "uri" => ""
6828     ),
6829     "zealbot" => array(
6830     "icon" => "zeal",
6831     "title" => "ZealBot",
6832     "rule" => array(
6833     "Zealbot[ /]([0-9.]{1,10})" => "\\1"
6834     ),
6835     "uri" => ""
6836     ),
6837     "zearchit" => array(
6838     "icon" => "zearchit",
6839     "title" => "Zearchit",
6840     "rule" => array(
6841     "Zearchit" => ""
6842     ),
6843     "uri" => "http://www.zearchit.de/"
6844     ),
6845     "zebz" => array(
6846     "icon" => "robot",
6847     "title" => "ze.bz",
6848     "rule" => array(
6849     "^ZeBot_(lseek\.net|www\.ze\.bz)" => ""
6850     ),
6851     "uri" => "http://www.ze.bz/"
6852     ),
6853     "zedzo" => array(
6854     "icon" => "robot",
6855     "title" => "Zedzo",
6856     "rule" => array(
6857     "zedzo.digest[ /]([0-9.]{1,10})" => "\\1"
6858     ),
6859     "uri" => "http://www.zedzo.com/"
6860     ),
6861     "zerx" => array(
6862     "icon" => "zerx",
6863     "title" => "Zerx",
6864     "rule" => array(
6865     "^zerxbot[ /](Version|v)*[ /]*([0-9.]{1,10})" => "\\2"
6866     ),
6867     "uri" => "http://www.zerx.com/"
6868     ),
6869     "zeus" => array(
6870     "icon" => "zeus",
6871     "title" => "Zeus",
6872     "rule" => array(
6873     "Zeus" => ""
6874     ),
6875     "uri" => "http://www.zeus.com"
6876     ),
6877     "zippp" => array(
6878     "icon" => "robot",
6879     "title" => "Zippp",
6880     "rule" => array(
6881     "ZipppBot[ /]([0-9.]{1,10})" => "\\1"
6882     ),
6883     "uri" => ""
6884     ),
6885     "zippy" => array(
6886     "icon" => "robot",
6887     "title" => "Zippy",
6888     "rule" => array(
6889     "^Zippy[ v/]*([0-9.]{1,10})" => "\\1"
6890     ),
6891     "uri" => "http://www.zippyfinder.com"
6892     ),
6893     "zoeky" => array(
6894     "icon" => "robot",
6895     "title" => "Zoeky",
6896     "rule" => array(
6897     "Zoekybot[ /]([0-9.]{1,10})" => "\\1"
6898     ),
6899     "uri" => ""
6900     ),
6901     "zoom" => array(
6902     "icon" => "zoom",
6903 matthys 209 "title" => "ZoomSpider",
6904 joku 63 "rule" => array(
6905     "^ZoomSpider" => ""
6906     ),
6907     "uri" => "http://www.wrensoft.com/"
6908     ),
6909     "zspider" => array(
6910     "icon" => "robot",
6911 matthys 209 "title" => "Zspider",
6912 joku 63 "rule" => array(
6913     "zspider[ /]([0-9.a-z]{1,10})" => "\\1"
6914     ),
6915     "uri" => "http://feedback.redkolibri.com/"
6916     ),
6917 matthys 281 "zumbot" => array(
6918     "icon" => "robot",
6919     "title" => "ZUM Search",
6920     "rule" => array(
6921     "ZumBot[ /]([0-9.]{1,10})" => "\\1"
6922     ),
6923     "uri" => "http://help.zum.com/inquiry"
6924     ),
6925 joku 63 "zyborg" => array(
6926     "icon" => "zyborg",
6927     "title" => "WiseNutBot",
6928     "rule" => array(
6929     "(WISE|Zy)bo(rg|t)[ /]([0-9.]{1,10})" => "\\3"
6930     ),
6931     "uri" => ""
6932     ),
6933     // Catch up for the originals, they got to stay in that order.
6934     "blogbot" => array(
6935     "icon" => "blogbot",
6936     "title" => "BlogBot",
6937     "rule" => array(
6938     "Blog[ \-]?Bot" => ""
6939     ),
6940     "uri" => "http://www.blogbot.com/"
6941     ),
6942     "centrum" => array(
6943     "icon" => "centrum",
6944     "title" => "Centrum",
6945     "rule" => array(
6946     "holmes[/ ]([0-9.]{1,10})" => "\\1",
6947     "^Centrum-checker" => ""
6948     ),
6949     "uri" => ""
6950     ),
6951     "httpclient" => array(
6952     "icon" => "robot",
6953     "title" => "HTTPClient",
6954     "rule" => array(
6955     "HTTP[ \-]?Client[ /]([0-9.]{1,10})" => "\\1",
6956     "HTTP[ \-]?Client" => ""
6957     ),
6958     "uri" => "http://www.innovation.ch/java/HTTPClient/"
6959     ),
6960     "incywincy" => array(
6961     "icon" => "robot",
6962     "title" => "IncyWincy",
6963     "rule" => array(
6964     "^IncyWincy[ /]([0-9.]{1,10})" => "\\1",
6965     "^IncyWincy" => ""
6966     ),
6967     "uri" => ""
6968     ),
6969     "java" => array(
6970     "icon" => "java",
6971     "title" => "Java",
6972     "rule" => array(
6973     "^java[ /]*([0-9.a-z]{1,10})" => "\\1"
6974     ),
6975     "uri" => ""
6976     ),
6977     "libfetch" => array(
6978     "icon" => "robot",
6979     "title" => "Libfetch",
6980     "rule" => array(
6981     "^(fetch )?libfetch[ /]([0-9.]{1,10})" => "\\2"
6982     ),
6983     "uri" => "http://www.freebsd.org/"
6984     ),
6985     "libwww" => array(
6986     "icon" => "libwww",
6987 matthys 209 "title" => "LibWWW",
6988 joku 63 "rule" => array(
6989     "^libww(w|w-perl|w-FM)[ /]([0-9.]{1,10})" => "\\2",
6990     "^libww(w|w-perl|w-FM)" => "",
6991     "MyApp.*libww(w|w-perl|w-FM)" => ""
6992     ),
6993     "uri" => ""
6994     ),
6995     "litefinder" => array(
6996     "icon" => "litefinder",
6997     "title" => "LiteFinder",
6998     "rule" => array(
6999     "LiteFinder[ /]([0-9.]{1,10})" => "\\1"
7000     ),
7001     "uri" => "http://www.litefinder.net/about.html"
7002     ),
7003     "nutchorg" => array(
7004     "icon" => "nutchorg",
7005     "title" => "Nutch",
7006     "rule" => array(
7007     "Nutc(hOrg|hCVS|h)?[ /]([0-9.]{1,10})" => "\\2",
7008     "Nutch" => ""
7009     ),
7010     "uri" => "http://lucene.apache.org/nutch/"
7011     ),
7012     "pythonurl" => array(
7013     "icon" => "pythonurl",
7014     "title" => "Python-url",
7015     "rule" => array(
7016     "Python[ \-]?urllib" => ""
7017     ),
7018     "uri" => ""
7019     ),
7020     // Know Robots as SPAM BOTS
7021     "SPAM" => array(
7022     "icon" => "robot",
7023     "title" => "SPAM",
7024     "rule" => array(
7025     "NASA Search[/ ]([0-9.]{1,10})" => "\\1",
7026     "^PHOTO CHECK" => "",
7027     "^FOTOCHECKER" => "",
7028     "^IPTC CHECK" => "",
7029     "^DataCha0s" => "",
7030     "^Mac Finder" => "",
7031     "^Missigua Locator[ /]([0-9.]{1,10})" => "\\1",
7032     "^Missouri College Browse" => "",
7033     "Email[ \-]?Siphon" => "",
7034     "atSpider" => "",
7035     "autoemailspider" => "",
7036     "^Demo Bot" => "",
7037     "^Program Shareware" => "",
7038     "^Snapbot" => "",
7039     "^snap.com" => "",
7040     "^Guestbook Auto Submitter" => "",
7041     "panscient.com" => "",
7042     ),
7043     "uri" => ""
7044     ),
7045     // Things we don't know by now
7046     "robot" => array(
7047     "icon" => "robot",
7048     "title" => "Robot",
7049     "rule" => array(
7050     "(robot|spider|harvest|(?<!orange)bot|(?<!msie)crawler)" => ""
7051     )
7052     )
7053     );
7054 joku 261 ?>

Properties

Name Value
svn:keywords Id