0.6.2-RC1/lib/referrer.php

<?php
/* This file is part of BBClone (A PHP based Web Counter on Steroids)
 * 
 * SVN FILE $Id$
 *  
 * Copyright (C) 2001-2013, the BBClone Team (see doc/authors.txt for details)
 *
 * This program is free software: you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation, either version 3 of the License, or
 * (at your option) any later version.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
 * GNU General Public License for more details.
 *
 * See doc/copying.txt for details
 */

//////////////
// Referrer //
//////////////

// referer stats
function bbc_sum_item($keep, $remove) {
  global $access;

  $access['referer'][$keep] += $access['referer'][$remove];

  unset($access['referer'][$remove]);
}

function bbc_update_referer_stat($referer) {
  global $access;

  if ($referer == "ignored") {
    $access['referer']['ignored'] = !isset($access['referer']['ignored']) ? 1 : ++$access['referer']['ignored'];
    return;
  }

  $referer_light = substr(strstr($referer, "://"), 3);

  if (empty($referer_light)) {
    $access['referer']['not_specified'] = !isset($access['referer']['not_specified']) ? 1 :
                                          ++$access['referer']['not_specified'];
    return;
  }

  if (($qm = strpos($referer_light, "?")) !== false) $referer_light = substr($referer_light, 0, $qm);
  if ((($parent = dirname($referer_light)) == ".") || ($parent === false)) {
    $referer_light = (substr($referer_light, -1) == "/") ? $referer_light : $referer_light."/";
  }
  else $referer_light = $parent."/";

  // compare whether we got a "www.*" equivalent recorded (or missing)
  $prefix = substr($referer_light, 0, ($tmp = strpos($referer_light, ".")));
  $suffix = substr($referer_light, ++$tmp);
  $result = ($prefix != "www") ? "www.".$referer_light : $suffix;
  // patterns for old records without trailing slashes
  $ref_no_slash = substr($referer_light, 0, -1);
  $res_no_slash = substr($result, 0, -1);

  // neither recorded with "www." nor without, seems to be our 1st visit ;)
  if (!isset($access['referer'][$referer_light]) && !isset($access['referer'][$result])) {
    $access['referer'][$referer_light] = 1;
  }
  // if there are old records without trailing slashes we will add them to our score too...
  elseif (!isset($access['referer'][$referer_light]) && isset($access['referer'][$result])) {
    isset($access['referer'][$res_no_slash]) ? bbc_sum_item($result, $res_no_slash) : "";

    ++$access['referer'][$result];
  }
  // same here...
  elseif (isset($access['referer'][$referer_light]) && !isset($access['referer'][$result])) {
    isset($access['referer'][$res_no_slash]) ? bbc_sum_item($referer_light, $res_no_slash) : "";

    ++$access['referer'][$referer_light];
  }
  // Now we got both of them, let's continue with the one we got most of
  else {
    if ($access['referer'][$referer_light] < $access['referer'][$result]) {
      bbc_sum_item($result, $referer_light);

      ++$access['referer'][$result];
    }
    else {
      bbc_sum_item($referer_light, $result);

      ++$access['referer'][$referer_light];
    }
  }
}

function bbc_fix_refer_stat($refsum) {
  global $access;

  $access['bugs']['ref_fix'] = 1;
  $access['referer']['not_specified'] = $access['stat']['totalcount'] - $refsum + $access['referer']['not_specified'];
}

function bbc_get_sep($query, $array) {
  // puts the query into an array

  foreach ($array as $match) {
    $has_sep = (strpos($query, $match) !== false) ? true : false;
    $pool = $has_sep ? explode($match, $query) : array($query);

    for ($i = 0, $max = count($pool); $i < $max; $i++) {
      // unsafe decoded characters could hose our stats
      $pool[$i] = preg_replace("%^[`\'\"<>@\^\!\?/\(\)\[\]\{\}|+*~#;,.:_\-]+%", "", $pool[$i]);
      $pool[$i] = preg_replace("%[`\'\"<>@\^\!\?/\(\)\[\]\{\}|+*~#;,.:_\-]+$%", "", $pool[$i]);

      if (empty($pool[$i]) || (strlen($pool[$i]) < 2)) {
        unset($pool[$i]);
        continue;
      }
    }
    if ($has_sep) return array_values($pool);
  }
  return array_values($pool);
}

function bbc_get_search($array) {
  // turns variable assignments to an associative array
  $result = false;
  $query = array(
    "^as_(ep|o|e)?q=",
    "^q(_(a(ll|ny)|phrase|not)|s|t|u(ery)?)?=",
    "^s(u|2f|p\-q|earch(_?for)?|tring|zukaj)?=",
    "^k(w|e(reses|y(word)?s?))=",
    "^b(egriff|uscar?)=",
    "^w(d|ords?)?=",
    "^te(rms?|xt)=",
    "^mi?t=",
    "^heureka=",
    "^p=",
    "^r(eq)?=",
    "/search/web/",
    "^userQuery=",
    "^v[aeop]="
  );

  foreach ($array as $string) {
    $string = urldecode($string);

    // skip empty GET variables
    if (substr($string, -1) == "=") continue;

    foreach ($query as $key) {
      preg_match(":$key:", $string, $matches);
      if (empty($matches)) continue;

      $par = $matches[0];
      $pos = strpos($string, $par);
      $term = substr($string, ($pos + strlen($par)));

      if (strlen($term) < 2) {
        $matches = array();
        continue;
      }

      if (($par[0] == "q") || ($par[0] == "s")) return $term;
    }
    $result = (!empty($matches)) ? $term : $result;
  }
  return $result;
}

function bbc_get_keywords($ref) {
  global $BBC_CUSTOM_CHARSET, $access;

  $var_sep =  array("&", "|");
  $word_sep = array( "+", " ", "/");
  $match = array(
    "ara", "busca", "pesquis", "search", "srch", "seek", "zoek", "result", "szuka", "cherch", "such", "find",
    "trouve", "trova", "pursuit", "keres", "katalogus", "alltheinternet.com", "mamma.com", "baidu.com", "heureka.hu",
    "kartoo.com", "ask.com", "aport.ru", "google", "yahoo"
  );

  foreach ($match as $key) {
    // if string occurs at the beginning strpos() returns integer 0, if it can't be
    // found at all, however, it returns boolean false => definition required which
    // considers 0 as true
    $is_search = (strpos(strtolower($ref), $key) !== false) ? true : false;

    if ($is_search) break;
  }

  if (!$is_search) return false;

  $ref = str_replace("&amp;", "&", urldecode($ref));
  $is_query = strrpos($ref, "?");
  $ref = ($is_query !== false) ? substr($ref, ++$is_query) : substr($ref, (strpos($ref, "://") + 3));
  $get_vars = bbc_get_sep($ref, $var_sep);
  $raw_search = bbc_get_search($get_vars);

  if ($raw_search === false) return false;

  // Conversion of keywords, if applicable
  $from = extension_loaded("mbstring") ? bbc_get_encoding($raw_search) : false;
  $char = (!empty($BBC_CUSTOM_CHARSET)) ? $BBC_CUSTOM_CHARSET : false;
  $raw_search = (($from !== false) || extension_loaded("recode")) ?
                bbc_convert_lang($raw_search, $from, $char) : $raw_search;
  $flt_search = bbc_get_sep($raw_search, $word_sep);

  for ($i = 0, $j = count($flt_search); $i < $j; $i++) {
    // Filter search engine cache indicator
    if ((strlen($flt_search[$i]) > 50) || (strlen($flt_search[$i]) < 2) ||
        (preg_match("#^(cache|tbn)\:[a-z0-9_\-]{8,16}\:#", $flt_search[$i]))) {
      unset($flt_search[$i]);
      continue;
    }

    // strtolower messes up UTF-8 so we leave things case sensitive if it's
    // requested as charset
    $flt_search[$i] = (!$char || (stristr($char, "UTF") === false)) ?
                      strtolower(bbc_clean($flt_search[$i])) : bbc_clean($flt_search[$i]);

    $access['key'][($flt_search[$i])] = !isset($access['key'][($flt_search[$i])]) ? 1 :
                                        ++$access['key'][($flt_search[$i])];
  }
  return (!empty($flt_search) ? $flt_search : false);
}
?>
Revision:	16
Committed:	Thu Nov 21 13:07:39 2013 UTC (11 years ago) by matthys
Original Path:	*trunk/lib/referrer.php*
File size:	7912 byte(s)
Log Message:	Cleanup code.
#	User	Rev	Content
1	matthys	15	<?php
2			/* This file is part of BBClone (A PHP based Web Counter on Steroids)
3			*
4	matthys	16	* SVN FILE $Id$
5	matthys	15	*
6	matthys	16	* Copyright (C) 2001-2013, the BBClone Team (see doc/authors.txt for details)
7	matthys	15	*
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			// Referrer //
23			//////////////
24
25			// referer stats
26			function bbc_sum_item($keep, $remove) {
27			global $access;
28
29			$access['referer'][$keep] += $access['referer'][$remove];
30
31			unset($access['referer'][$remove]);
32			}
33
34			function bbc_update_referer_stat($referer) {
35			global $access;
36
37			if ($referer == "ignored") {
38			$access['referer']['ignored'] = !isset($access['referer']['ignored']) ? 1 : ++$access['referer']['ignored'];
39			return;
40			}
41
42			$referer_light = substr(strstr($referer, "://"), 3);
43
44			if (empty($referer_light)) {
45			$access['referer']['not_specified'] = !isset($access['referer']['not_specified']) ? 1 :
46			++$access['referer']['not_specified'];
47			return;
48			}
49
50			if (($qm = strpos($referer_light, "?")) !== false) $referer_light = substr($referer_light, 0, $qm);
51			if ((($parent = dirname($referer_light)) == ".") \|\| ($parent === false)) {
52			$referer_light = (substr($referer_light, -1) == "/") ? $referer_light : $referer_light."/";
53			}
54			else $referer_light = $parent."/";
55
56			// compare whether we got a "www.*" equivalent recorded (or missing)
57			$prefix = substr($referer_light, 0, ($tmp = strpos($referer_light, ".")));
58			$suffix = substr($referer_light, ++$tmp);
59			$result = ($prefix != "www") ? "www.".$referer_light : $suffix;
60			// patterns for old records without trailing slashes
61			$ref_no_slash = substr($referer_light, 0, -1);
62			$res_no_slash = substr($result, 0, -1);
63
64			// neither recorded with "www." nor without, seems to be our 1st visit ;)
65			if (!isset($access['referer'][$referer_light]) && !isset($access['referer'][$result])) {
66			$access['referer'][$referer_light] = 1;
67			}
68			// if there are old records without trailing slashes we will add them to our score too...
69			elseif (!isset($access['referer'][$referer_light]) && isset($access['referer'][$result])) {
70			isset($access['referer'][$res_no_slash]) ? bbc_sum_item($result, $res_no_slash) : "";
71
72			++$access['referer'][$result];
73			}
74			// same here...
75			elseif (isset($access['referer'][$referer_light]) && !isset($access['referer'][$result])) {
76			isset($access['referer'][$res_no_slash]) ? bbc_sum_item($referer_light, $res_no_slash) : "";
77
78			++$access['referer'][$referer_light];
79			}
80			// Now we got both of them, let's continue with the one we got most of
81			else {
82			if ($access['referer'][$referer_light] < $access['referer'][$result]) {
83			bbc_sum_item($result, $referer_light);
84
85			++$access['referer'][$result];
86			}
87			else {
88			bbc_sum_item($referer_light, $result);
89
90			++$access['referer'][$referer_light];
91			}
92			}
93			}
94
95			function bbc_fix_refer_stat($refsum) {
96			global $access;
97
98			$access['bugs']['ref_fix'] = 1;
99			$access['referer']['not_specified'] = $access['stat']['totalcount'] - $refsum + $access['referer']['not_specified'];
100			}
101
102			function bbc_get_sep($query, $array) {
103			// puts the query into an array
104
105			foreach ($array as $match) {
106			$has_sep = (strpos($query, $match) !== false) ? true : false;
107			$pool = $has_sep ? explode($match, $query) : array($query);
108
109			for ($i = 0, $max = count($pool); $i < $max; $i++) {
110			// unsafe decoded characters could hose our stats
111			$pool[$i] = preg_replace("%^[`\'\"<>@\^\!\?/\(\)\[\]\{\}\|+*~#;,.:_\-]+%", "", $pool[$i]);
112			$pool[$i] = preg_replace("%[`\'\"<>@\^\!\?/\(\)\[\]\{\}\|+*~#;,.:_\-]+$%", "", $pool[$i]);
113
114			if (empty($pool[$i]) \|\| (strlen($pool[$i]) < 2)) {
115			unset($pool[$i]);
116			continue;
117			}
118			}
119			if ($has_sep) return array_values($pool);
120			}
121			return array_values($pool);
122			}
123
124			function bbc_get_search($array) {
125			// turns variable assignments to an associative array
126			$result = false;
127			$query = array(
128			"^as_(ep\|o\|e)?q=",
129			"^q(_(a(ll\|ny)\|phrase\|not)\|s\|t\|u(ery)?)?=",
130			"^s(u\|2f\|p\-q\|earch(_?for)?\|tring\|zukaj)?=",
131			"^k(w\|e(reses\|y(word)?s?))=",
132			"^b(egriff\|uscar?)=",
133			"^w(d\|ords?)?=",
134			"^te(rms?\|xt)=",
135			"^mi?t=",
136			"^heureka=",
137			"^p=",
138			"^r(eq)?=",
139			"/search/web/",
140			"^userQuery=",
141			"^v[aeop]="
142			);
143
144			foreach ($array as $string) {
145			$string = urldecode($string);
146
147			// skip empty GET variables
148			if (substr($string, -1) == "=") continue;
149
150			foreach ($query as $key) {
151			preg_match(":$key:", $string, $matches);
152			if (empty($matches)) continue;
153
154			$par = $matches[0];
155			$pos = strpos($string, $par);
156			$term = substr($string, ($pos + strlen($par)));
157
158			if (strlen($term) < 2) {
159			$matches = array();
160			continue;
161			}
162
163			if (($par[0] == "q") \|\| ($par[0] == "s")) return $term;
164			}
165			$result = (!empty($matches)) ? $term : $result;
166			}
167			return $result;
168			}
169
170			function bbc_get_keywords($ref) {
171			global $BBC_CUSTOM_CHARSET, $access;
172
173			$var_sep = array("&", "\|");
174			$word_sep = array( "+", " ", "/");
175			$match = array(
176			"ara", "busca", "pesquis", "search", "srch", "seek", "zoek", "result", "szuka", "cherch", "such", "find",
177			"trouve", "trova", "pursuit", "keres", "katalogus", "alltheinternet.com", "mamma.com", "baidu.com", "heureka.hu",
178			"kartoo.com", "ask.com", "aport.ru", "google", "yahoo"
179			);
180
181			foreach ($match as $key) {
182			// if string occurs at the beginning strpos() returns integer 0, if it can't be
183			// found at all, however, it returns boolean false => definition required which
184			// considers 0 as true
185			$is_search = (strpos(strtolower($ref), $key) !== false) ? true : false;
186
187			if ($is_search) break;
188			}
189
190			if (!$is_search) return false;
191
192			$ref = str_replace("&", "&", urldecode($ref));
193			$is_query = strrpos($ref, "?");
194			$ref = ($is_query !== false) ? substr($ref, ++$is_query) : substr($ref, (strpos($ref, "://") + 3));
195			$get_vars = bbc_get_sep($ref, $var_sep);
196			$raw_search = bbc_get_search($get_vars);
197
198			if ($raw_search === false) return false;
199
200			// Conversion of keywords, if applicable
201			$from = extension_loaded("mbstring") ? bbc_get_encoding($raw_search) : false;
202			$char = (!empty($BBC_CUSTOM_CHARSET)) ? $BBC_CUSTOM_CHARSET : false;
203			$raw_search = (($from !== false) \|\| extension_loaded("recode")) ?
204			bbc_convert_lang($raw_search, $from, $char) : $raw_search;
205			$flt_search = bbc_get_sep($raw_search, $word_sep);
206
207			for ($i = 0, $j = count($flt_search); $i < $j; $i++) {
208			// Filter search engine cache indicator
209			if ((strlen($flt_search[$i]) > 50) \|\| (strlen($flt_search[$i]) < 2) \|\|
210			(preg_match("#^(cache\|tbn)\:[a-z0-9_\-]{8,16}\:#", $flt_search[$i]))) {
211			unset($flt_search[$i]);
212			continue;
213			}
214
215			// strtolower messes up UTF-8 so we leave things case sensitive if it's
216			// requested as charset
217			$flt_search[$i] = (!$char \|\| (stristr($char, "UTF") === false)) ?
218			strtolower(bbc_clean($flt_search[$i])) : bbc_clean($flt_search[$i]);
219
220			$access['key'][($flt_search[$i])] = !isset($access['key'][($flt_search[$i])]) ? 1 :
221			++$access['key'][($flt_search[$i])];
222			}
223			return (!empty($flt_search) ? $flt_search : false);
224			}
225			?>