| 1 |
<?php |
| 2 |
/* This file is part of BBClone (A PHP based Web Counter on Steroids) |
| 3 |
* |
| 4 |
* SVN FILE $Id$ |
| 5 |
* |
| 6 |
* Copyright (C) 2001-2014, the BBClone Team (see doc/authors.txt for details) |
| 7 |
* |
| 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 |
// Log Processes // |
| 23 |
/////////////////// |
| 24 |
|
| 25 |
// Checking where we came from |
| 26 |
if (!defined("_MARK_PAGE")) return; |
| 27 |
|
| 28 |
// used by usort() |
| 29 |
function bbc_sort_time_sc($row_a, $row_b) { |
| 30 |
if ($row_a['time'] == $row_b['time']) return 0; |
| 31 |
return ($row_a['time'] > $row_b['time'] ) ? 1 : -1; |
| 32 |
} |
| 33 |
|
| 34 |
// return the key of a value |
| 35 |
function bbc_get_key($array, $str) { |
| 36 |
reset($array); |
| 37 |
|
| 38 |
while (list($idx, $val) = each($array)) { |
| 39 |
if ($val == $str) return $idx; |
| 40 |
} |
| 41 |
return 0; |
| 42 |
} |
| 43 |
|
| 44 |
// purge host and referrer stats at request |
| 45 |
function bbc_purge_single() { |
| 46 |
global $access; |
| 47 |
|
| 48 |
foreach (array("host", "key", "referer") as $cat) { |
| 49 |
reset($access[$cat]); |
| 50 |
|
| 51 |
while (list($key, $score) = each($access[$cat])) { |
| 52 |
if ($score == 1) { |
| 53 |
unset($access[$cat][$key]); |
| 54 |
($cat == "referer") ? ++$access[$cat]['not_specified'] : ""; |
| 55 |
} |
| 56 |
} |
| 57 |
} |
| 58 |
} |
| 59 |
|
| 60 |
// records the hosts that visited us most. Note, that we only pick up hostnames and strip |
| 61 |
// any sort of prefix because else the listing would become rather useless |
| 62 |
|
| 63 |
if ((empty($BBC_IGNORE_BOTS)) || (!isset($client['robot']))) { |
| 64 |
$is_num = ($client['dns'] == $client['ip']) ? 1 : 0; |
| 65 |
$host = (!$is_num) ? explode(".", strtolower($client['dns'])) : trim(substr($client['ip'], 0, |
| 66 |
strrpos($client['ip'], "."))).". -"; |
| 67 |
$parts = (!$is_num) ? count($host) : 0; |
| 68 |
// these ones can have less than 3 characters as hostname |
| 69 |
$glob = array("org", "com", "edu", "mil", "net", "gov", "int"); |
| 70 |
|
| 71 |
if ($parts > 2) { |
| 72 |
$host = ((!in_array(trim($host[($parts - 1)]), $glob)) && (((strlen(trim($host[($parts - 2)])) < 3)) || |
| 73 |
(in_array($host[($parts - 2)], $glob)))) ? |
| 74 |
trim(implode(".", array_splice($host, -3))) : trim(implode(".", array_splice($host, -2))); |
| 75 |
} |
| 76 |
else $host = (!$is_num) ? trim(implode(".", $host)) : $host; |
| 77 |
} |
| 78 |
|
| 79 |
function bbc_update_host_stat($client) { |
| 80 |
global $access, $BBC_IGNORE_BOTS, $BBC_DOUBLE_TLD; |
| 81 |
|
| 82 |
if ((empty($BBC_IGNORE_BOTS)) || (!isset($client['robot']))) { |
| 83 |
$is_num = ($client['dns'] == $client['ip']) ? 1 : 0; |
| 84 |
if ($is_num) { |
| 85 |
// Numeric |
| 86 |
if (strpos($client['ip'], ":") === false) { |
| 87 |
// IPv4 |
| 88 |
$host = trim(substr($client['ip'], 0, strrpos($client['ip'], "."))).". -"; |
| 89 |
} else { |
| 90 |
// IPv6 |
| 91 |
$parts = explode( ":", $client['ip']); |
| 92 |
$parts = array_slice( $parts, 0, 4); |
| 93 |
$host = implode( ":", $parts) . ": -"; |
| 94 |
} |
| 95 |
} else { |
| 96 |
// DNS name |
| 97 |
$double_toplevels = explode(",", $BBC_DOUBLE_TLD); |
| 98 |
$parts = explode( ".", $client['dns']); |
| 99 |
$lastparts = implode( ".", array_slice( $parts, -2, 2) ); |
| 100 |
if (in_array($lastparts, $double_toplevels)) { |
| 101 |
// We have a double tld like co.uk |
| 102 |
$host = implode( ".", array_slice( $parts, -4, 4) ); |
| 103 |
} else { |
| 104 |
// We have a normal tld |
| 105 |
$host = implode( ".", array_slice( $parts, -3, 3) ); |
| 106 |
} |
| 107 |
// Remove optional www* from start |
| 108 |
$host = preg_replace("/^www[0-9]*\./", "", $host); |
| 109 |
} |
| 110 |
} |
| 111 |
if ((empty($BBC_IGNORE_BOTS)) || (!isset($client['robot']))) { |
| 112 |
if (!isset($access['host'][$host])) $access['host'][$host] = 0; |
| 113 |
$access['host'][$host]++; |
| 114 |
} |
| 115 |
|
| 116 |
if (isset($access['host']['not_specified'])) unset($access['host']['not_specified']); |
| 117 |
} |
| 118 |
|
| 119 |
// the listing of the visited pages |
| 120 |
function bbc_update_visits($time, $page, $nr) { |
| 121 |
global $BBC_MAXVISIBLE, $last; |
| 122 |
|
| 123 |
$lv = count($last['traffic'][$nr]['views']) - 1; |
| 124 |
$last_time = substr($last['traffic'][$nr]['views'][$lv], 0, strpos($last['traffic'][$nr]['views'][$lv], "|")); |
| 125 |
$last_cnt = substr($last['traffic'][$nr]['views'][$lv], (strrpos($last['traffic'][$nr]['views'][$lv], "|") + 1)); |
| 126 |
$last_page = substr($last['traffic'][$nr]['views'][$lv], (strpos($last['traffic'][$nr]['views'][$lv], "|") + 1)); |
| 127 |
$last_page = substr($last_page, 0, strpos($last_page, "|")); |
| 128 |
$last['traffic'][$nr]['off'] = !empty($last['traffic'][$nr]['off']) ? $last['traffic'][$nr]['off'] : 0; |
| 129 |
|
| 130 |
if ((empty($last['traffic'][$nr]['views'])) || (!is_array($last['traffic'][$nr]['views']))) return; |
| 131 |
|
| 132 |
if (intval($last_page) === intval($page)) { |
| 133 |
$last['traffic'][$nr]['views'][$lv] = "$last_time|$last_page|".++$last_cnt; |
| 134 |
$last['traffic'][$nr]['off']++; |
| 135 |
} |
| 136 |
else $last['traffic'][$nr]['views'][] = "$time|$page|1"; |
| 137 |
|
| 138 |
sort($last['traffic'][$nr]['views']); |
| 139 |
|
| 140 |
// number of elements to be removed with array_splice() if necessary |
| 141 |
$lv = count($last['traffic'][$nr]['views']) - 1; |
| 142 |
$del = (($lv + 1) > $BBC_MAXVISIBLE) ? (($lv + 1) - $BBC_MAXVISIBLE) : false; |
| 143 |
$last['traffic'][$nr]['views'] = ($del !== false) ? array_splice($last['traffic'][$nr]['views'], $del) : |
| 144 |
$last['traffic'][$nr]['views']; |
| 145 |
} |
| 146 |
|
| 147 |
// The most visited pages ranking |
| 148 |
function bbc_update_page_stats($connect) { |
| 149 |
global $access, $last; |
| 150 |
|
| 151 |
$long_page = $connect['page']; |
| 152 |
$over_60 = (strlen($long_page) > 60) ? 1 : 0; |
| 153 |
$connect['page'] = $over_60 ? "...".substr($long_page, -57) : $long_page; |
| 154 |
|
| 155 |
// Fix oversized page titles |
| 156 |
if (($over_60) && (isset($access['page'][$long_page]['count']))) { |
| 157 |
$access['page'][($connect['page'])]['count'] = $access['page'][$long_page]['count']; |
| 158 |
$access['page'][($connect['page'])]['uri'] = $access['page'][$long_page]['uri']; |
| 159 |
unset($access['page'][$long_page]); |
| 160 |
} |
| 161 |
|
| 162 |
|
| 163 |
if (!isset($access['page'][($connect['page'])]['count'])) { |
| 164 |
$access['page'][($connect['page'])]['count'] = 0; |
| 165 |
} |
| 166 |
|
| 167 |
$access['page'][($connect['page'])]['count']++; |
| 168 |
$access['page'][($connect['page'])]['uri'] = $connect['uri']; |
| 169 |
|
| 170 |
$last['pages'] = ((empty($last['pages'])) || (!is_array($last['pages']))) ? array() : $last['pages']; |
| 171 |
|
| 172 |
if (($over_60) && (in_array($long_page, $last['pages']))) { |
| 173 |
$last['pages'][bbc_get_key($last['pages'], $long_page)] = $connect['page']; |
| 174 |
} |
| 175 |
if (!in_array($connect['page'], $last['pages'])) $last['pages'][] = $connect['page']; |
| 176 |
|
| 177 |
$connect['page'] = bbc_get_key($last['pages'], $connect['page']); |
| 178 |
|
| 179 |
if (isset($connect['uri'])) unset($connect['uri']); |
| 180 |
|
| 181 |
return $connect; |
| 182 |
} |
| 183 |
|
| 184 |
// Transfer the raw data from the main counters of var into $last. |
| 185 |
// Any new data (more recent than $BBC_MAXTIME) is used in the global stats |
| 186 |
function bbc_add_new_connections_to_old() { |
| 187 |
global $BBC_IGNORE_AGENT, $BBC_IGNORE_BOTS, $BBC_MAXTIME, $BBC_MAXVISIBLE, $BBC_NO_DNS, $BBC_NO_HITS, |
| 188 |
$BBC_PURGE_SINGLE, $access, $last; |
| 189 |
|
| 190 |
// Checking whether we have new connections |
| 191 |
if (!$new_access = bbc_counter_to_array()) return false; |
| 192 |
|
| 193 |
// cleanup if requested |
| 194 |
!empty($BBC_PURGE_SINGLE) ? bbc_purge_single() : ""; |
| 195 |
|
| 196 |
((!empty($access['time'])) && (is_array($access['time']))) ? bbc_time_offset() : ""; |
| 197 |
|
| 198 |
// Upgrade from older versions. We need to erase the "last" data. |
| 199 |
if (isset($access['last'])) unset($access['last']); |
| 200 |
|
| 201 |
//check for broken 0.4.2 referrer counting and apply fix if necessary |
| 202 |
if (isset($access) && !isset($access['bugs']['ref_fix']) && isset($access['referer']['not_specified'])) { |
| 203 |
bbc_fix_refer_stat(array_sum($access['referer'])); |
| 204 |
} |
| 205 |
|
| 206 |
// fix wrong browser assignments |
| 207 |
foreach (array("java", "wwwc", "libwww") as $what) { |
| 208 |
if (isset($access) && isset($access['stat']['browser'][$what])) { |
| 209 |
$access['stat']['robot'][$what] = $access['stat']['browser'][$what]; |
| 210 |
$access['stat']['os']['other'] -= $access['stat']['robot'][$what]; |
| 211 |
|
| 212 |
unset($access['stat']['browser'][$what]); |
| 213 |
} |
| 214 |
} |
| 215 |
|
| 216 |
$nb_new_access = (!empty($new_access) && is_array($new_access)) ? count($new_access) : 0; |
| 217 |
$nb_last_access = (!empty($last['traffic']) && is_array($last['traffic'])) ? count($last['traffic']) : 0; |
| 218 |
|
| 219 |
foreach ($new_access as $connect) { |
| 220 |
$connect = bbc_update_connect($connect); |
| 221 |
|
| 222 |
// the "last reset on" flag initialisation |
| 223 |
if ((!isset($access['time'])) && (!isset($access['time']['reset']))) { |
| 224 |
$access['time']['reset'] = $connect['time']; |
| 225 |
} |
| 226 |
|
| 227 |
// Stop processing if bots are completely ignored |
| 228 |
if ((!empty($BBC_IGNORE_BOTS)) && ($BBC_IGNORE_BOTS == 2)) { |
| 229 |
if (!empty($connect['robot'])) { |
| 230 |
--$nb_new_access; |
| 231 |
continue; |
| 232 |
} |
| 233 |
} |
| 234 |
// Omit referrers coming from robots |
| 235 |
$connect['referer'] = !empty($connect['robot']) ? "unknown" : $connect['referer']; |
| 236 |
|
| 237 |
$this_connect = $connect['time']; |
| 238 |
$last_connect = !empty($access['time']['last']) ? $access['time']['last'] : 0; |
| 239 |
|
| 240 |
// Hits as base for time stats if desired |
| 241 |
if (empty($BBC_NO_HITS)) bbc_update_time_stat($this_connect, $last_connect); |
| 242 |
|
| 243 |
// The script viewed |
| 244 |
$connect = isset($connect['page']) ? bbc_update_page_stats($connect) : $connect; |
| 245 |
$prev_recorded = 0; |
| 246 |
|
| 247 |
// Check if a similar connection has been recorded yet |
| 248 |
for($l = $nb_last_access - 1; ($l >= 0) && (($connect['time'] - $last['traffic'][$l]['time']) < $BBC_MAXTIME); |
| 249 |
$l--) { |
| 250 |
if (!empty($BBC_IGNORE_AGENT) ? ($connect['ip'] == $last['traffic'][$l]['ip']) : |
| 251 |
(($connect['ip'] == $last['traffic'][$l]['ip']) && ($connect['agent'] == $last['traffic'][$l]['agent']))) { |
| 252 |
$last['traffic'][$l]['page'] = $connect['page']; |
| 253 |
$last['traffic'][$l]['time'] = $this_connect; |
| 254 |
$last['traffic'][$l]['visits']++; |
| 255 |
$access['stat']['totalvisits']++; |
| 256 |
|
| 257 |
($BBC_MAXVISIBLE > 0) ? bbc_update_visits($connect['time'], $connect['page'], $l) : ""; |
| 258 |
|
| 259 |
$prev_recorded = 1; |
| 260 |
break; |
| 261 |
} |
| 262 |
} |
| 263 |
|
| 264 |
// Add new connection if it hasn't been recorded yet |
| 265 |
if (!$prev_recorded) { |
| 266 |
if (empty($access['stat']['totalvisits'])) $access['stat']['totalvisits'] = 0; |
| 267 |
if (empty($access['stat']['totalcount'])) $access['stat']['totalcount'] = 0; |
| 268 |
|
| 269 |
$connect['dns'] = !empty($BBC_NO_DNS) ? $connect['ip'] : bbc_clean(gethostbyaddr($connect['ip'])); |
| 270 |
$connect['ext'] = bbc_get_extension($connect['dns'], $connect['ip']); |
| 271 |
|
| 272 |
$last['traffic'][$nb_last_access] = bbc_update_access($connect); |
| 273 |
// Visit stats |
| 274 |
$last['traffic'][$nb_last_access]['views'][] = $last['traffic'][$nb_last_access]['time']."|" |
| 275 |
.$last['traffic'][$nb_last_access]['page']."|1"; |
| 276 |
|
| 277 |
// Unique visits as base for time stats if desired |
| 278 |
if (!empty($BBC_NO_HITS)) bbc_update_time_stat($this_connect, $last_connect); |
| 279 |
|
| 280 |
// Referrers collection will be updated all along with the keywords if available |
| 281 |
if (isset($connect['referer'])) { |
| 282 |
bbc_update_referer_stat($connect['referer']); |
| 283 |
$flt_search = bbc_get_keywords($connect['referer']); |
| 284 |
} |
| 285 |
// The search as a whole in $last |
| 286 |
$last['traffic'][$nb_last_access]['search'] = ($flt_search !== false) ? implode(" ", $flt_search) : "-"; |
| 287 |
|
| 288 |
// The host listing |
| 289 |
if ((isset($connect['dns'])) && (isset($connect['ip']))) { |
| 290 |
bbc_update_host_stat($last['traffic'][$nb_last_access]); |
| 291 |
} |
| 292 |
|
| 293 |
$access['stat']['totalvisits']++; |
| 294 |
$access['stat']['totalcount']++; |
| 295 |
$nb_last_access++; |
| 296 |
} |
| 297 |
} |
| 298 |
return $nb_new_access; |
| 299 |
} |
| 300 |
|
| 301 |
// Remove unnecessary connections from $last, that either exceed the $BBC_MAXVISIBLE limit or are |
| 302 |
// older than time() - $BBC_MAXTIME. |
| 303 |
function bbc_update_last_access() { |
| 304 |
global $last, $BBC_MAXTIME, $BBC_MAXVISIBLE, $BBC_TIMESTAMP, $BBC_TIME_OFFSET; |
| 305 |
|
| 306 |
if (($BBC_MAXVISIBLE <= 0) || (empty($last['traffic'])) || (!is_array($last['traffic']))) { |
| 307 |
$last['traffic'] = array(); |
| 308 |
return; |
| 309 |
} |
| 310 |
else { |
| 311 |
$nb_connect = count($last['traffic']); |
| 312 |
$ctime = $BBC_TIMESTAMP + ($BBC_TIME_OFFSET * 60); |
| 313 |
|
| 314 |
for ($k = $nb_connect - 1 - $BBC_MAXVISIBLE; $k >= 0; $k--) { |
| 315 |
if (($ctime - $last['traffic'][$k]['time']) > $BBC_MAXTIME) unset($last['traffic'][$k]); |
| 316 |
} |
| 317 |
usort($last['traffic'],"bbc_sort_time_sc"); |
| 318 |
} |
| 319 |
} |
| 320 |
?> |
