spam.php

   1 <?php
   2 // $Id: spam.php,v 1.40 2006/11/26 02:57:26 henoheno Exp $
   3 // Copyright (C) 2006 PukiWiki Developers Team
   4 // License: GPL v2 or (at your option) any later version
   5
   6 // Functions for Concept-work of spam-uri metrics
   7
   8 // Return an array of URIs in the $string
   9 // [OK] http://nasty.example.org#nasty_string
  10 // [OK] http://nasty.example.org:80/foo/xxx#nasty_string/bar
  11 // [OK] ftp://nasty.example.org:80/dfsdfs
  12 // [OK] ftp://cnn.example.com&story=breaking_news@10.0.0.1/top_story.htm (from RFC3986)
  13 function uri_pickup($string = '', $normalize = TRUE,
  14         $preserve_rawuri = FALSE, $preserve_chunk = TRUE)
  15 {
  16         // Not available for: IDN(ignored)
  17         $array = array();
  18         preg_match_all(
  19                 // scheme://userinfo@host:port/path/or/pathinfo/maybefile.and?query=string#fragment
  20                 // Refer RFC3986 (Regex below is not strict)
  21                 '#(\b[a-z][a-z0-9.+-]{1,8})://' .       // 1: Scheme
  22                 '(?:' .
  23                         '([^\s<>"\'\[\]/\#?@]*)' .              // 2: Userinfo (Username)
  24                 '@)?' .
  25                 '(' .
  26                         // 3: Host
  27                         '\[[0-9a-f:.]+\]' . '|' .                               // IPv6([colon-hex and dot]): RFC2732
  28                         '(?:[0-9]{1-3}\.){3}[0-9]{1-3}' . '|' . // IPv4(dot-decimal): 001.22.3.44
  29                         '[^\s<>"\'\[\]:/\#?]+' .                                // FQDN: foo.example.org
  30                 ')' .
  31                 '(?::([0-9]*))?' .                                      // 4: Port
  32                 '((?:/+[^\s<>"\'\[\]/\#]+)*/+)?' .      // 5: Directory path or path-info
  33                 '([^\s<>"\'\[\]\#]+)?' .                        // 6: File and query string
  34                 '(?:\#([a-z0-9._~%!$&\'()*+,;=:@-]*))?' .       // 7: Fragment
  35                 '#i',
  36                  $string, $array, PREG_SET_ORDER | PREG_OFFSET_CAPTURE
  37         );
  38         //var_dump(recursive_map('htmlspecialchars', $array));
  39
  40         // Shrink $array
  41         static $parts = array(
  42                 1 => 'scheme', 2 => 'userinfo', 3 => 'host', 4 => 'port',
  43                 5 => 'path', 6 => 'file', 7 => 'fragment'
  44         );
  45         $default = array('');
  46         foreach(array_keys($array) as $uri) {
  47                 array_rename_keys($array[$uri], $parts, TRUE, $default);
  48                 $offset = $array[$uri]['scheme'][1]; // Scheme's offset
  49
  50                 foreach(array_keys($array[$uri]) as $part) {
  51                         // Remove offsets for each part
  52                         $array[$uri][$part] = & $array[$uri][$part][0];
  53                 }
  54
  55                 if ($normalize) {
  56                         $array[$uri]['scheme'] = scheme_normalize($array[$uri]['scheme']);
  57                         //if ($array[$uri]['scheme'] === '') {
  58                         //      // Ignore
  59                         //      unset ($array[$uri]);
  60                         //      continue;
  61                         //}
  62
  63                         $array[$uri]['host']   = strtolower($array[$uri]['host']);
  64                         $array[$uri]['port']   = port_normalize($array[$uri]['port'], $array[$uri]['scheme'], FALSE);
  65                         $array[$uri]['path']   = path_normalize($array[$uri]['path']);
  66                         //$array[$uri]['uri']    = uri_array_implode($array[$uri]);
  67                         if ($preserve_rawuri) $array[$uri]['rawuri'] = & $array[$uri][0];
  68                 } else {
  69                         $array[$uri]['uri'] = & $array[$uri][0]; // Raw
  70                 }
  71                 unset($array[$uri][0]); // Matched string itself
  72                 if (! $preserve_chunk) {
  73                         unset(
  74                                 $array[$uri]['scheme'],
  75                                 $array[$uri]['userinfo'],
  76                                 $array[$uri]['host'],
  77                                 $array[$uri]['port'],
  78                                 $array[$uri]['path'],
  79                                 $array[$uri]['file'],
  80                                 $array[$uri]['fragment']
  81                         );
  82                 }
  83
  84                 $array[$uri]['area']['offset'] = $offset;
  85         }
  86
  87         return $array;
  88 }
  89
  90 // Domain exposure callback (See spam_uri_pickup_preprocess())
  91 // http://victim.example.org/?foo+site:nasty.example.com+bar
  92 // => http://nasty.example.com/?refer=victim.example.org
  93 // NOTE: 'refer=' is not so good for (at this time).
  94 // Consider about using IP address of the victim, try to avoid that.
  95 function _preg_replace_callback_domain_exposure($matches = array())
  96 {
  97         $result = '';
  98
  99         // Preserve the victim URI as a complicity or ...
 100         if (isset($matches[5])) {
 101                 $result =
 102                         $matches[1] . '://' .   // scheme
 103                         $matches[2] . '/' .             // victim.example.org
 104                         $matches[3];                    // The rest of all (before victim)
 105         }
 106
 107         // Flipped URI
 108         $result =
 109                 $matches[1] . '://' .   // scheme
 110                 $matches[4] .                   // nasty.example.com
 111                 '/?refer=' . strtolower($matches[2]) .  // victim.example.org
 112                 ' ' . $result;
 113
 114         return $result;
 115 }
 116
 117 // Preprocess: rawurldecode() and adding space(s) to detect/count some URIs _if possible_
 118 // NOTE: It's maybe danger to var_dump(result). [e.g. 'javascript:']
 119 // [OK] http://victim.example.org/go?http%3A%2F%2Fnasty.example.org
 120 // [OK] http://victim.example.org/http://nasty.example.org
 121 function spam_uri_pickup_preprocess($string = '')
 122 {
 123         if (! is_string($string)) return '';
 124
 125         $string = rawurldecode($string);
 126
 127         // Domain exposure (See _preg_replace_callback_domain_exposure())
 128         $string = preg_replace_callback(
 129                 array(
 130                         // Something Google: http://www.google.com/supported_domains
 131                         '#(http)://([a-z0-9.]+\.google\.[a-z]{2,3}(?:\.[a-z]{2})?)/' .
 132                         '([a-z0-9?=&.%_+-]+)' .         // ?query=foo+
 133                         '\bsite:([a-z0-9.%_-]+)' .      // site:nasty.example.com
 134                         '()' .  // Preserve?
 135                         '#i',
 136                 ),
 137                 '_preg_replace_callback_domain_exposure',
 138                 $string
 139         );
 140
 141         // URI exposure (uriuri => uri uri)
 142         $string = preg_replace(
 143                 array(
 144                         '#(?<! )(?:https?|ftp):/#',
 145                 //      '#[a-z][a-z0-9.+-]{1,8}://#i',
 146                 //      '#[a-z][a-z0-9.+-]{1,8}://#i'
 147                 ),
 148                 ' $0',
 149                 $string
 150         );
 151
 152         return $string;
 153 }
 154
 155 // Main function of spam-uri pickup
 156 function spam_uri_pickup($string = '', $area = array())
 157 {
 158         if (! is_array($area) || empty($area)) {
 159                 $area = array(
 160                                 'anchor' => TRUE,
 161                                 'bbcode' => TRUE,
 162                         );
 163         }
 164
 165         $string = spam_uri_pickup_preprocess($string);
 166
 167         $array  = uri_pickup($string);
 168
 169         // Area elevation for '(especially external)link' intension
 170         if (! empty($array)) {
 171
 172                 $area_shadow = array();
 173                 foreach(array_keys($array) as $key){
 174                         $area_shadow[$key] = & $array[$key]['area'];
 175                         $area_shadow[$key]['anchor'] = 0;
 176                         $area_shadow[$key]['bbcode'] = 0;
 177                 }
 178
 179                 // Anchor tags by preg_match_all()
 180                 // [OK] <a href="http://nasty.example.com">visit http://nasty.example.com/</a>
 181                 // [OK] <a href=\'http://nasty.example.com/\' >discount foobar</a>
 182                 // [NG] <a href="http://ng.example.com">visit http://ng.example.com _not_ended_
 183                 // [NG] <a href=  >Good site!</a> <a href= "#" >test</a>
 184                 if (isset($area['anchor'])) {
 185                         $areas = array();
 186                         preg_match_all('#<a\b[^>]*href[^>]*>.*?</a\b[^>]*(>)#i',
 187                                  $string, $areas, PREG_SET_ORDER | PREG_OFFSET_CAPTURE);
 188                         //var_dump(recursive_map('htmlspecialchars', $areas));
 189                         foreach(array_keys($areas) as $_area) {
 190                                 $areas[$_area] =  array(
 191                                         $areas[$_area][0][1], // Area start (<a href>)
 192                                         $areas[$_area][1][1], // Area end   (</a>)
 193                                 );
 194                         }
 195                         area_measure($areas, $area_shadow, -1, 'anchor');
 196                 }
 197
 198                 // phpBB's "BBCode" by preg_match_all()
 199                 // [url]http://nasty.example.com/[/url]
 200                 // [link]http://nasty.example.com/[/link]
 201                 // [url=http://nasty.example.com]visit http://nasty.example.com/[/url]
 202                 // [link http://nasty.example.com/]buy something[/link]
 203                 // ?? [url=][/url]
 204                 if (isset($area['bbcode'])) {
 205                         $areas = array();
 206                         preg_match_all('#\[(url|link)\b[^\]]*\].*?\[/\1\b[^\]]*(\])#i',
 207                                  $string, $areas, PREG_SET_ORDER | PREG_OFFSET_CAPTURE);
 208                         //var_dump(recursive_map('htmlspecialchars', $areas));
 209                         foreach(array_keys($areas) as $_area) {
 210                                 $areas[$_area] = array(
 211                                         $areas[$_area][0][1], // Area start ([url])
 212                                         $areas[$_area][2][1], // Area end   ([/url])
 213                                 );
 214                         }
 215                         area_measure($areas, $area_shadow, -1, 'bbcode');
 216                 }
 217
 218                 // Various Wiki syntax
 219                 // [text_or_uri>text_or_uri]
 220                 // [text_or_uri:text_or_uri]
 221                 // [text_or_uri|text_or_uri]
 222                 // [text_or_uri->text_or_uri]
 223                 // [text_or_uri text_or_uri] // MediaWiki
 224                 // MediaWiki: [http://nasty.example.com/ visit http://nasty.example.com/]
 225
 226                 // Remove 'offset's for area_measure()
 227                 //foreach(array_keys($array) as $key)
 228                 //      unset($array[$key]['offset']);
 229         }
 230
 231         return $array;
 232 }
 233
 234 // $array['something'] => $array['wanted']
 235 function array_rename_keys(& $array, $keys = array('from' => 'to'), $force = FALSE, $default = '')
 236 {
 237         if (! is_array($array) || ! is_array($keys))
 238                 return FALSE;
 239
 240         // Nondestructive test
 241         if (! $force)
 242                 foreach(array_keys($keys) as $from)
 243                         if (! isset($array[$from]))
 244                                 return FALSE;
 245
 246         foreach($keys as $from => $to) {
 247                 if ($from === $to) continue;
 248                 if (! $force || isset($array[$from])) {
 249                         $array[$to] = & $array[$from];
 250                         unset($array[$from]);
 251                 } else  {
 252                         $array[$to] = $default;
 253                 }
 254         }
 255
 256         return TRUE;
 257 }
 258
 259 // If in doubt, it's a little doubtful
 260 function area_measure($areas, & $array, $belief = -1, $a_key = 'area', $o_key = 'offset')
 261 {
 262         if (! is_array($areas) || ! is_array($array)) return;
 263
 264         $areas_keys = array_keys($areas);
 265         foreach(array_keys($array) as $u_index) {
 266                 $offset = isset($array[$u_index][$o_key]) ?
 267                         intval($array[$u_index][$o_key]) : 0;
 268                 foreach($areas_keys as $a_index) {
 269                         if (isset($array[$u_index][$a_key])) {
 270                                 $offset_s = intval($areas[$a_index][0]);
 271                                 $offset_e = intval($areas[$a_index][1]);
 272                                 // [Area => inside <= Area]
 273                                 if ($offset_s < $offset && $offset < $offset_e) {
 274                                         $array[$u_index][$a_key] += $belief;
 275                                 }
 276                         }
 277                 }
 278         }
 279 }
 280
 281
 282 // ---------------------
 283 // Part Two
 284
 285 // Scheme normalization: Renaming the schemes
 286 // snntp://example.org =>  nntps://example.org
 287 // NOTE: Keep the static lists simple. See also port_normalize().
 288 function scheme_normalize($scheme = '', $considerd_harmfull = TRUE)
 289 {
 290         // Abbreviations considerable they don't have link intension
 291         static $abbrevs = array(
 292                 'ttp'   => 'http',
 293                 'ttps'  => 'https',
 294         );
 295
 296         // Alias => normalized
 297         static $aliases = array(
 298                 'pop'   => 'pop3',
 299                 'news'  => 'nntp',
 300                 'imap4' => 'imap',
 301                 'snntp' => 'nntps',
 302                 'snews' => 'nntps',
 303                 'spop3' => 'pop3s',
 304                 'pops'  => 'pop3s',
 305         );
 306
 307         $scheme = strtolower(trim($scheme));
 308         if (isset($abbrevs[$scheme])) {
 309                 if ($considerd_harmfull) {
 310                         $scheme = $abbrevs[$scheme];
 311                 } else {
 312                         $scheme = '';
 313                 }
 314         }
 315         if (isset($aliases[$scheme])) $scheme = $aliases[$scheme];
 316
 317         return $scheme;
 318 }
 319
 320 // Port normalization: Suppress the (redundant) default port
 321 // HTTP://example.org:80/ => http://example.org/
 322 // HTTP://example.org:8080/ => http://example.org:8080/
 323 // HTTPS://example.org:443/ => https://example.org/
 324 function port_normalize($port, $scheme, $scheme_normalize = TRUE)
 325 {
 326         // Schemes that users _maybe_ want to add protocol-handlers
 327         // to their web browsers. (and attackers _maybe_ want to use ...)
 328         // Reference: http://www.iana.org/assignments/port-numbers
 329         static $array = array(
 330                 // scheme => default port
 331                 'ftp'     =>    21,
 332                 'ssh'     =>    22,
 333                 'telnet'  =>    23,
 334                 'smtp'    =>    25,
 335                 'tftp'    =>    69,
 336                 'gopher'  =>    70,
 337                 'finger'  =>    79,
 338                 'http'    =>    80,
 339                 'pop3'    =>   110,
 340                 'sftp'    =>   115,
 341                 'nntp'    =>   119,
 342                 'imap'    =>   143,
 343                 'irc'     =>   194,
 344                 'wais'    =>   210,
 345                 'https'   =>   443,
 346                 'nntps'   =>   563,
 347                 'rsync'   =>   873,
 348                 'ftps'    =>   990,
 349                 'telnets' =>   992,
 350                 'imaps'   =>   993,
 351                 'ircs'    =>   994,
 352                 'pop3s'   =>   995,
 353                 'mysql'   =>  3306,
 354         );
 355
 356         $port = trim($port);
 357         if ($port === '') return $port;
 358
 359         if ($scheme_normalize) $scheme = scheme_normalize($scheme);
 360         if (isset($array[$scheme]) && $port == $array[$scheme])
 361                 $port = ''; // Ignore the defaults
 362
 363         return $port;
 364 }
 365
 366 // Path normalization
 367 // http://example.org => http://example.org/
 368 // http://example.org#hoge => http://example.org/#hoge
 369 // http://example.org/path/a/b/./c////./d => http://example.org/path/a/b/c/d
 370 // http://example.org/path/../../a/../back => http://example.org/back
 371 function path_normalize($path = '', $divider = '/', $addroot = TRUE)
 372 {
 373         if (! is_string($path) || $path == '') {
 374                 $path = $addroot ? $divider : '';
 375         } else {
 376                 $path = trim($path);
 377                 $last = ($path[strlen($path) - 1] == $divider) ? $divider : '';
 378                 $array = explode($divider, $path);
 379
 380                 // Remove paddings
 381                 foreach(array_keys($array) as $key) {
 382                         if ($array[$key] == '' || $array[$key] == '.')
 383                                  unset($array[$key]);
 384                 }
 385                 // Back-track
 386                 $tmp = array();
 387                 foreach($array as $value) {
 388                         if ($value == '..') {
 389                                 array_pop($tmp);
 390                         } else {
 391                                 array_push($tmp, $value);
 392                         }
 393                 }
 394                 $array = & $tmp;
 395
 396                 $path = $addroot ? $divider : '';
 397                 if (! empty($array)) $path .= implode($divider, $array) . $last;
 398         }
 399
 400         return $path;
 401 }
 402
 403 // An URI array => An URI (See uri_pickup())
 404 function uri_array_implode($uri = array())
 405 {
 406         if (empty($uri) || ! is_array($uri)) return NULL;
 407
 408         $tmp = array();
 409         if (isset($uri['scheme']) && $uri['scheme'] !== '') {
 410                 $tmp[] = & $uri['scheme'];
 411                 $tmp[] = '://';
 412         }
 413         if (isset($uri['userinfo']) && $uri['userinfo'] !== '') {
 414                 $tmp[] = & $uri['userinfo'];
 415                 $tmp[] = '@';
 416         }
 417         if (isset($uri['host']) && $uri['host'] !== '') {
 418                 $tmp[] = & $uri['host'];
 419         }
 420         if (isset($uri['port']) && $uri['port'] !== '') {
 421                 $tmp[] = ':';
 422                 $tmp[] = & $uri['port'];
 423         }
 424         if (isset($uri['path']) && $uri['path'] !== '') {
 425                 $tmp[] = & $uri['path'];
 426         }
 427         if (isset($uri['file']) && $uri['file'] !== '') {
 428                 $tmp[] = & $uri['file'];
 429         }
 430         if (isset($uri['fragment']) && $uri['fragment'] !== '') {
 431                 $tmp[] = '#';
 432                 $tmp[] = & $uri['fragment'];
 433         }
 434
 435         return implode('', $tmp);
 436 }
 437
 438 // ---------------------
 439 // Part One : Checker
 440
 441 function generate_glob_regex($string = '', $divider = '/')
 442 {
 443         static $from = array(
 444                         0 => '*',
 445                         1 => '?',
 446                         2 => '\[',
 447                         3 => '\]',
 448                         4 => '[',
 449                         5 => ']',
 450                 );
 451         static $mid = array(
 452                         0 => '_AST_',
 453                         1 => '_QUE_',
 454                         2 => '_eRBR_',
 455                         3 => '_eLBR_',
 456                         4 => '_RBR_',
 457                         5 => '_LBR_',
 458                 );
 459         static $to = array(
 460                         0 => '.*',
 461                         1 => '.',
 462                         2 => '\[',
 463                         3 => '\]',
 464                         4 => '[',
 465                         5 => ']',
 466                 );
 467
 468         $string = str_replace($from, $mid, $string); // Hide
 469         $string = preg_quote($string, $divider);
 470         $string = str_replace($mid, $to, $string);   // Unhide
 471
 472         return $string;
 473 }
 474
 475 // TODO: Ignore list
 476 // TODO: require_or_include_once(another file)
 477 function is_badhost($hosts = '', $asap = TRUE)
 478 {
 479         static $blocklist_regex;
 480
 481         if (! isset($blocklist_regex)) {
 482                 $blocklist_regex = array();
 483                 $blocklist = array(
 484                         // Deny all uri
 485                         //'*',
 486
 487                         // IP address or ...
 488                         //'10.20.*.*',  // 10.20.example.com also matches
 489                         //'\[1\]',
 490
 491                         // Too much malicious sub-domains
 492                         //'*.blogspot.com',
 493
 494                         // 2006-11 dev
 495                         'wwwtahoo.com',
 496
 497                         // 2006-11 dev
 498                         '*.infogami.com',
 499
 500                         // 2006/11/19 17:50 dev
 501                         //'*.google0site.org',
 502                         //'*.bigpricesearch.org',
 503                         //'*.osfind.org',
 504                         //'*.bablomira.biz',
 505                 );
 506                 foreach ($blocklist as $part) {
 507                         $blocklist_regex[] = '#^' . generate_glob_regex($part, '#') . '$#i';
 508                 }
 509         }
 510
 511         $result = 0;
 512         if (! is_array($hosts)) $hosts = array($hosts);
 513         foreach($hosts as $host) {
 514                 if (! is_string($host)) $host = '';
 515                 foreach ($blocklist_regex as $regex) {
 516                         if (preg_match($regex, $host)) {
 517                                 ++$result;
 518                                 if ($asap) {
 519                                         return $result;
 520                                 } else {
 521                                         break; // Check next host
 522                                 }
 523                         }
 524                 }
 525         }
 526
 527         return $result;
 528 }
 529
 530 // Default method and threshold
 531 function check_uri_spam_method()
 532 {
 533         return array(
 534                 'quantity' => 8,                // Allow N URIs
 535                 'area'     => array(
 536                         'total'  => 0,          // Allow N areas
 537                         'anchor' => 0,          // <a href> HTML tag
 538                         'bbcode' => 0,          // [url] or [link] BBCode
 539                         ),
 540                 'non_uniq' => 3,                // Allow N duped (and normalized) URIs
 541                 'badhost'  => TRUE,
 542                 );
 543 }
 544
 545 // TODO return TRUE or FALSE!
 546 // Simple/fast spam check
 547 function check_uri_spam($target = '', $method = array(), $asap = TRUE)
 548 {
 549         $is_spam  = FALSE;
 550         $progress = array(
 551                 'quantity' => 0,
 552                 'area'     => array(
 553                         'total'  => 0,
 554                         'anchor' => 0,
 555                         'bbcode' => 0,
 556                         ),
 557                 'non_uniq' => 0,
 558                 'uniqhost' => 0,
 559                 'badhost'  => 0,
 560                 );
 561
 562         if (! is_array($method) || empty($method)) {
 563                 $method = check_uri_spam_method();
 564         }
 565
 566         if (is_array($target)) {
 567                 foreach($target as $str) {
 568                         // Recurse
 569                         list($is_spam, $_progress) = check_uri_spam($str, $method);
 570                         $progress['quantity'] += $_progress['quantity'];
 571                         $progress['area']['total']  += $_progress['area']['total'];
 572                         $progress['area']['anchor'] += $_progress['area']['anchor'];
 573                         $progress['area']['bbcode'] += $_progress['area']['bbcode'];
 574                         $progress['non_uniq'] += $_progress['non_uniq'];
 575                         $progress['uniqhost'] += $_progress['uniqhost'];
 576                         $progress['badhost']  += $_progress['badhost'];
 577                         if ($asap || $is_spam) break;
 578                 }
 579         } else {
 580                 $pickups = spam_uri_pickup($target);
 581                 $progress['quantity'] += count($pickups);
 582
 583                 if (! empty($pickups)) {
 584
 585                         // URI quantity
 586                         if ((! $is_spam || ! $asap) && isset($method['quantity']) &&
 587                                 $progress['quantity'] > $method['quantity']) {
 588                                 $is_spam = TRUE;
 589                         }
 590                         //var_dump($method['quantity'], $is_spam);
 591
 592                         // Using invalid area
 593                         if ((! $is_spam || ! $asap) && isset($method['area'])) {
 594                                 foreach($pickups as $pickup) {
 595                                         // Total
 596                                         $total = 0;
 597                                         foreach ($pickup['area'] as $key => $value) {
 598                                                 if ($key == 'offset') continue;
 599                                                 $progress['area']['total'] += $value;
 600                                                 $progress['area'][$key]    += $value;
 601                                                 if ($value < 0) {
 602                                                         $is_spam = TRUE;
 603                                                         if ($is_spam && $asap) break;
 604                                                 }
 605                                                 if ($asap) break;
 606                                         }
 607                                 }
 608                         }
 609                         //var_dump($method['area'], $is_spam);
 610
 611                         // URI uniqueness (and removing non-uniques)
 612                         if ((! $is_spam || ! $asap) && isset($method['non_uniq'])) {
 613                                 $uris = array();
 614                                 foreach ($pickups as $key => $pickup) {
 615                                         $uris[$key] = uri_array_implode($pickup);
 616                                 }
 617                                 $count = count($uris);
 618                                 $uris = array_unique($uris);
 619                                 $progress['non_uniq'] += $count - count($uris);
 620                                 if ($progress['non_uniq'] > $method['non_uniq']) {
 621                                         $is_spam = TRUE;
 622                                 }
 623                                 if (! $asap || ! $is_spam) {
 624                                         foreach (array_diff(array_keys($pickups),
 625                                                 array_keys($uris)) as $remove) {
 626                                                 unset($pickups[$remove]);
 627                                         }
 628                                 }
 629                                 unset($uris);
 630                                 //var_dump($uris, $pickups);
 631                         }
 632                         //var_dump($method['non_uniq'], $is_spam);
 633
 634                         // Unique host
 635                         $hosts = array();
 636                         foreach ($pickups as $pickup) {
 637                                 $hosts[] = & $pickup['host'];
 638                         }
 639                         $hosts = array_unique($hosts);
 640                         $progress['uniqhost'] += count($hosts);
 641                         //var_dump($method['uniqhost'], $is_spam);
 642
 643                         // Bad host
 644                         if ((! $is_spam || ! $asap) && isset($method['badhost'])) {
 645                                 $count = is_badhost($hosts, $asap);
 646                                 $progress['badhost'] += $count;
 647                                 if ($count !== 0) $is_spam = TRUE;
 648                         }
 649                         //var_dump($method['badhost'], $is_spam);
 650                 }
 651         }
 652
 653         return array($is_spam, $progress);
 654 }
 655
 656 // ---------------------
 657
 658 // Check User-Agent (not testing yet)
 659 function is_invalid_useragent($ua_name = '' /*, $ua_vars = ''*/ )
 660 {
 661         return $ua_name === '';
 662 }
 663
 664 // ---------------------
 665
 666 // TODO: Separate check-part(s) and mail part
 667 // TODO: Mail to administrator with more measurement data?
 668 // Simple/fast spam filter ($target: 'a string' or an array())
 669 function pkwk_spamfilter($action, $page, $target = array('title' => ''), $method = array())
 670 {
 671         $is_spam = FALSE;
 672
 673         //$is_spam =  is_invalid_useragent('NOTYET');
 674         if ($is_spam) {
 675                 $action .= ' (Invalid User-Agent)';
 676         } else {
 677                 list($is_spam) = check_uri_spam($target, $method);
 678         }
 679
 680         if ($is_spam) {
 681                 // Mail to administrator(s)
 682                 global $notify, $notify_subject;
 683                 if ($notify) {
 684                         $footer['ACTION'] = $action;
 685                         $footer['PAGE']   = '[blocked] ' . $page;
 686                         $footer['URI']    = get_script_uri() . '?' . rawurlencode($page);
 687                         $footer['USER_AGENT']  = TRUE;
 688                         $footer['REMOTE_ADDR'] = TRUE;
 689                         pkwk_mail_notify($notify_subject,  var_export($target, TRUE), $footer);
 690                         unset($footer);
 691                 }
 692         }
 693
 694         if ($is_spam) spam_exit();
 695 }
 696
 697 // ---------------------
 698
 699 // Common bahavior for blocking
 700 // NOTE: Call this function from various blocking feature, to disgueise the reason 'why blocked'
 701 function spam_exit()
 702 {
 703         die("\n");
 704 }
 705
 706 ?>