Ticket #1885: xtc_check_agent.inc.php

File xtc_check_agent.inc.php, 10.8 KB (added by d.schneider@…, 5 years ago)
Line 
1<?php
2/* -----------------------------------------------------------------------------------------
3 $Id: xtc_check_agent.inc.php 974 2005-06-07 12:40:29Z mz $
4
5 XT-Commerce - community made shopping
6 http://www.xt-commerce.com
7
8 Copyright (c) 2003 XT-Commerce
9 -----------------------------------------------------------------------------------------
10 based on:
11 (c) 2000-2001 The Exchange Project (earlier name of osCommerce)
12 (c) 2002-2003 osCommerce(html_output.php,v 1.52 2003/03/19); www.oscommerce.com
13 (c) 2003 nextcommerce (xtc_href_link.inc.php,v 1.3 2003/08/13); www.nextcommerce.org
14
15 Released under the GNU General Public License
16 ---------------------------------------------------------------------------------------*/
17
18
19function xtc_check_agent()
20{
21if (CHECK_CLIENT_AGENT=='true') {
22 $Robots = array (
23 "antibot",
24 "appie",
25 "architext",
26 "bjaaland",
27 "digout4u",
28 "echo",
29 "fast-webcrawler",
30 "ferret",
31 "googlebot",
32 "gulliver",
33 "harvest",
34 "htdig",
35 "ia_archiver",
36 "jeeves",
37 "jennybot",
38 "linkwalker",
39 "lycos",
40 "mercator",
41 "moget",
42 "muscatferret",
43 "myweb",
44 "netcraft",
45 "nomad",
46 "petersnews",
47 "scooter",
48 "slurp",
49 "unlost_web_crawler",
50 "voila",
51 "voyager",
52 "webbase",
53 "weblayers",
54 "wget",
55 "wisenutbot",
56 "acme.spider",
57 "ahoythehomepagefinder",
58 "alkaline",
59 "arachnophilia",
60 "aretha",
61 "ariadne",
62 "arks",
63 "aspider",
64 "atn.txt",
65 "atomz",
66 "auresys",
67 "backrub",
68 "bigbrother",
69 "blackwidow",
70 "blindekuh",
71 "bloodhound",
72 "brightnet",
73 "bspider",
74 "cactvschemistryspider",
75 "cassandra",
76 "cgireader",
77 "checkbot",
78 "churl",
79 "cmc",
80 "collective",
81 "combine",
82 "conceptbot",
83 "coolbot",
84 "core",
85 "cosmos",
86 "cruiser",
87 "cusco",
88 "cyberspyder",
89 "deweb",
90 "dienstspider",
91 "digger",
92 "diibot",
93 "directhit",
94 "dnabot",
95 "download_express",
96 "dragonbot",
97 "dwcp",
98 "e-collector",
99 "ebiness",
100 "eit",
101 "elfinbot",
102 "emacs",
103 "emcspider",
104 "esther",
105 "evliyacelebi",
106 "nzexplorer",
107 "fdse",
108 "felix",
109 "fetchrover",
110 "fido",
111 "finnish",
112 "fireball",
113 "fouineur",
114 "francoroute",
115 "freecrawl",
116 "funnelweb",
117 "gama",
118 "gazz",
119 "gcreep",
120 "getbot",
121 "geturl",
122 "golem",
123 "grapnel",
124 "griffon",
125 "gromit",
126 "hambot",
127 "havindex",
128 "hometown",
129 "htmlgobble",
130 "hyperdecontextualizer",
131 "iajabot",
132 "ibm",
133 "iconoclast",
134 "ilse",
135 "imagelock",
136 "incywincy",
137 "informant",
138 "infoseek",
139 "infoseeksidewinder",
140 "infospider",
141 "inspectorwww",
142 "intelliagent",
143 "irobot",
144 "iron33",
145 "israelisearch",
146 "javabee",
147 "jbot",
148 "jcrawler",
149 "jobo",
150 "jobot",
151 "joebot",
152 "jubii",
153 "jumpstation",
154 "katipo",
155 "kdd",
156 "kilroy",
157 "ko_yappo_robot",
158 "labelgrabber.txt",
159 "larbin",
160 "legs",
161 "linkidator",
162 "linkscan",
163 "lockon",
164 "logo_gif",
165 "macworm",
166 "magpie",
167 "marvin",
168 "mattie",
169 "mediafox",
170 "merzscope",
171 "meshexplorer",
172 "mindcrawler",
173 "momspider",
174 "monster",
175 "motor",
176 "mwdsearch",
177 "netcarta",
178 "netmechanic",
179 "netscoop",
180 "newscan-online",
181 "nhse",
182 "northstar",
183 "occam",
184 "octopus",
185 "openfind",
186 "orb_search",
187 "packrat",
188 "pageboy",
189 "parasite",
190 "patric",
191 "pegasus",
192 "perignator",
193 "perlcrawler",
194 "phantom",
195 "piltdownman",
196 "pimptrain",
197 "pioneer",
198 "pitkow",
199 "pjspider",
200 "pka",
201 "plumtreewebaccessor",
202 "poppi",
203 "portalb",
204 "puu",
205 "python",
206 "raven",
207 "rbse",
208 "resumerobot",
209 "rhcs",
210 "roadrunner",
211 "robbie",
212 "robi",
213 "robofox",
214 "robozilla",
215 "roverbot",
216 "rules",
217 "safetynetrobot",
218 "search_au",
219 "searchprocess",
220 "senrigan",
221 "sgscout",
222 "shaggy",
223 "shaihulud",
224 "sift",
225 "simbot",
226 "site-valet",
227 "sitegrabber",
228 "sitetech",
229 "slcrawler",
230 "smartspider",
231 "snooper",
232 "solbot",
233 "spanner",
234 "speedy",
235 "spider_monkey",
236 "spiderbot",
237 "spiderline",
238 "spiderman",
239 "spiderview",
240 "spry",
241 "ssearcher",
242 "suke",
243 "suntek",
244 "sven",
245 "tach_bw",
246 "tarantula",
247 "tarspider",
248 "techbot",
249 "templeton",
250 "teoma_agent1",
251 "titin",
252 "titan",
253 "tkwww",
254 "tlspider",
255 "ucsd",
256 "udmsearch",
257 "urlck",
258 "valkyrie",
259 "victoria",
260 "visionsearch",
261 "vwbot",
262 "w3index",
263 "w3m2",
264 "wallpaper",
265 "wanderer",
266 "wapspider",
267 "webbandit",
268 "webcatcher",
269 "webcopy",
270 "webfetcher",
271 "webfoot",
272 "weblinker",
273 "webmirror",
274 "webmoose",
275 "webquest",
276 "webreader",
277 "webreaper",
278 "websnarf",
279 "webspider",
280 "webvac",
281 "webwalk",
282 "webwalker",
283 "webwatch",
284 "whatuseek",
285 "whowhere",
286 "wired-digital",
287 "wmir",
288 "wolp",
289 "wombat",
290 "worm",
291 "wwwc",
292 "wz101",
293 "xget",
294 "awbot",
295 "bobby",
296 "boris",
297 "bumblebee",
298 "cscrawler",
299 "daviesbot",
300 "ezresult",
301 "gigabot",
302 "gnodspider",
303 "internetseer",
304 "justview",
305 "linkbot",
306 "linkchecker",
307 "nederland.zoek",
308 "perman",
309 "pompos",
310 "pooodle",
311 "redalert",
312 "shoutcast",
313 "slysearch",
314 "ultraseek",
315 "webcompass",
316 "yandex",
317 "robot",
318 "yahoo",
319 "bot",
320 "psbot",
321 "crawl"
322 );
323
324
325 $botID = strtolower($_SERVER['HTTP_USER_AGENT']);
326 $botID2 = strtolower(getenv("HTTP_USER_AGENT"));
327 for ($i = 0; $i < count($Robots); $i++)
328 {
329
330 if (strstr($botID, $Robots[$i]) or strstr($botID2, $Robots[$i]))
331 {
332 return 1;
333 }
334
335 }
336 return 0;
337} else {
338return 0;
339}
340
341}
342
343
344
345//hd7b
346function xtc_return_agent(){
347 $Robots = array (
348 "antibot",
349 "appie",
350 "architext",
351 "bjaaland",
352 "digout4u",
353 "echo",
354 "fast-webcrawler",
355 "ferret",
356 "googlebot",
357 "gulliver",
358 "harvest",
359 "htdig",
360 "ia_archiver",
361 "jeeves",
362 "jennybot",
363 "linkwalker",
364 "lycos",
365 "mercator",
366 "moget",
367 "muscatferret",
368 "myweb",
369 "netcraft",
370 "nomad",
371 "petersnews",
372 "scooter",
373 "slurp",
374 "unlost_web_crawler",
375 "voila",
376 "voyager",
377 "webbase",
378 "weblayers",
379 "wget",
380 "wisenutbot",
381 "acme.spider",
382 "ahoythehomepagefinder",
383 "alkaline",
384 "arachnophilia",
385 "aretha",
386 "ariadne",
387 "arks",
388 "aspider",
389 "atn.txt",
390 "atomz",
391 "auresys",
392 "backrub",
393 "bigbrother",
394 "blackwidow",
395 "blindekuh",
396 "bloodhound",
397 "brightnet",
398 "bspider",
399 "cactvschemistryspider",
400 "cassandra",
401 "cgireader",
402 "checkbot",
403 "churl",
404 "cmc",
405 "collective",
406 "combine",
407 "conceptbot",
408 "coolbot",
409 "core",
410 "cosmos",
411 "cruiser",
412 "cusco",
413 "cyberspyder",
414 "deweb",
415 "dienstspider",
416 "digger",
417 "diibot",
418 "directhit",
419 "dnabot",
420 "download_express",
421 "dragonbot",
422 "dwcp",
423 "e-collector",
424 "ebiness",
425 "eit",
426 "elfinbot",
427 "emacs",
428 "emcspider",
429 "esther",
430 "evliyacelebi",
431 "nzexplorer",
432 "fdse",
433 "felix",
434 "fetchrover",
435 "fido",
436 "finnish",
437 "fireball",
438 "fouineur",
439 "francoroute",
440 "freecrawl",
441 "funnelweb",
442 "gama",
443 "gazz",
444 "gcreep",
445 "getbot",
446 "geturl",
447 "golem",
448 "grapnel",
449 "griffon",
450 "gromit",
451 "hambot",
452 "havindex",
453 "hometown",
454 "htmlgobble",
455 "hyperdecontextualizer",
456 "iajabot",
457 "ibm",
458 "iconoclast",
459 "ilse",
460 "imagelock",
461 "incywincy",
462 "informant",
463 "infoseek",
464 "infoseeksidewinder",
465 "infospider",
466 "inspectorwww",
467 "intelliagent",
468 "irobot",
469 "iron33",
470 "israelisearch",
471 "javabee",
472 "jbot",
473 "jcrawler",
474 "jobo",
475 "jobot",
476 "joebot",
477 "jubii",
478 "jumpstation",
479 "katipo",
480 "kdd",
481 "kilroy",
482 "ko_yappo_robot",
483 "labelgrabber.txt",
484 "larbin",
485 "legs",
486 "linkidator",
487 "linkscan",
488 "lockon",
489 "logo_gif",
490 "macworm",
491 "magpie",
492 "marvin",
493 "mattie",
494 "mediafox",
495 "merzscope",
496 "meshexplorer",
497 "mindcrawler",
498 "momspider",
499 "monster",
500 "motor",
501 "mwdsearch",
502 "netcarta",
503 "netmechanic",
504 "netscoop",
505 "newscan-online",
506 "nhse",
507 "northstar",
508 "occam",
509 "octopus",
510 "openfind",
511 "orb_search",
512 "packrat",
513 "pageboy",
514 "parasite",
515 "patric",
516 "pegasus",
517 "perignator",
518 "perlcrawler",
519 "phantom",
520 "piltdownman",
521 "pimptrain",
522 "pioneer",
523 "pitkow",
524 "pjspider",
525 "pka",
526 "plumtreewebaccessor",
527 "poppi",
528 "portalb",
529 "puu",
530 "python",
531 "raven",
532 "rbse",
533 "resumerobot",
534 "rhcs",
535 "roadrunner",
536 "robbie",
537 "robi",
538 "robofox",
539 "robozilla",
540 "roverbot",
541 "rules",
542 "safetynetrobot",
543 "search_au",
544 "searchprocess",
545 "senrigan",
546 "sgscout",
547 "shaggy",
548 "shaihulud",
549 "sift",
550 "simbot",
551 "site-valet",
552 "sitegrabber",
553 "sitetech",
554 "slcrawler",
555 "smartspider",
556 "snooper",
557 "solbot",
558 "spanner",
559 "speedy",
560 "spider_monkey",
561 "spiderbot",
562 "spiderline",
563 "spiderman",
564 "spiderview",
565 "spry",
566 "ssearcher",
567 "suke",
568 "suntek",
569 "sven",
570 "tach_bw",
571 "tarantula",
572 "tarspider",
573 "techbot",
574 "templeton",
575 "teoma_agent1",
576 "titin",
577 "titan",
578 "tkwww",
579 "tlspider",
580 "ucsd",
581 "udmsearch",
582 "urlck",
583 "valkyrie",
584 "victoria",
585 "visionsearch",
586 "vwbot",
587 "w3index",
588 "w3m2",
589 "wallpaper",
590 "wanderer",
591 "wapspider",
592 "webbandit",
593 "webcatcher",
594 "webcopy",
595 "webfetcher",
596 "webfoot",
597 "weblinker",
598 "webmirror",
599 "webmoose",
600 "webquest",
601 "webreader",
602 "webreaper",
603 "websnarf",
604 "webspider",
605 "webvac",
606 "webwalk",
607 "webwalker",
608 "webwatch",
609 "whatuseek",
610 "whowhere",
611 "wired-digital",
612 "wmir",
613 "wolp",
614 "wombat",
615 "worm",
616 "wwwc",
617 "wz101",
618 "xget",
619 "awbot",
620 "bobby",
621 "boris",
622 "bumblebee",
623 "cscrawler",
624 "daviesbot",
625 "ezresult",
626 "gigabot",
627 "gnodspider",
628 "internetseer",
629 "justview",
630 "linkbot",
631 "linkchecker",
632 "nederland.zoek",
633 "perman",
634 "pompos",
635 "pooodle",
636 "redalert",
637 "shoutcast",
638 "slysearch",
639 "ultraseek",
640 "webcompass",
641 "yandex",
642 "robot",
643 "yahoo",
644 "bot",
645 "psbot",
646 "crawl"
647 );
648
649 $botID = strtolower($_SERVER['HTTP_USER_AGENT']);
650 $botID2 = strtolower(getenv("HTTP_USER_AGENT"));
651 for ($i = 0; $i < count($Robots); $i++){
652 if (strstr($botID, $Robots[$i]) or strstr($botID2, $Robots[$i])) {
653 return $Robots[$i];
654 }
655 }
656 return "bot?!";
657}
658
659?>