spam.php

   1 <?php
   2 // $Id: spam.php,v 1.51 2006/12/02 09:19:51 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 if (! defined('SPAM_INI_FILE')) define('SPAM_INI_FILE', 'spam.ini.php');
   9
  10 // Return an array of URIs in the $string
  11 // [OK] http://nasty.example.org#nasty_string
  12 // [OK] http://nasty.example.org:80/foo/xxx#nasty_string/bar
  13 // [OK] ftp://nasty.example.org:80/dfsdfs
  14 // [OK] ftp://cnn.example.com&story=breaking_news@10.0.0.1/top_story.htm (from RFC3986)
  15 function uri_pickup($string = '', $normalize = TRUE,
  16         $preserve_rawuri = FALSE, $preserve_chunk = TRUE)
  17 {
  18         // Not available for: IDN(ignored)
  19         $array = array();
  20         preg_match_all(
  21                 // scheme://userinfo@host:port/path/or/pathinfo/maybefile.and?query=string#fragment
  22                 // Refer RFC3986 (Regex below is not strict)
  23                 '#(\b[a-z][a-z0-9.+-]{1,8})://' .       // 1: Scheme
  24                 '(?:' .
  25                         '([^\s<>"\'\[\]/\#?@]*)' .              // 2: Userinfo (Username)
  26                 '@)?' .
  27                 '(' .
  28                         // 3: Host
  29                         '\[[0-9a-f:.]+\]' . '|' .                               // IPv6([colon-hex and dot]): RFC2732
  30                         '(?:[0-9]{1-3}\.){3}[0-9]{1-3}' . '|' . // IPv4(dot-decimal): 001.22.3.44
  31                         '[^\s<>"\'\[\]:/\#?]+' .                                // FQDN: foo.example.org
  32                 ')' .
  33                 '(?::([0-9]*))?' .                                      // 4: Port
  34                 '((?:/+[^\s<>"\'\[\]/\#]+)*/+)?' .      // 5: Directory path or path-info
  35                 '([^\s<>"\'\[\]\#]+)?' .                        // 6: File and query string
  36                 '(?:\#([a-z0-9._~%!$&\'()*+,;=:@-]*))?' .       // 7: Fragment
  37                 '#i',
  38                  $string, $array, PREG_SET_ORDER | PREG_OFFSET_CAPTURE
  39         );
  40         //var_dump(recursive_map('htmlspecialchars', $array));
  41
  42         // Shrink $array
  43         static $parts = array(
  44                 1 => 'scheme', 2 => 'userinfo', 3 => 'host', 4 => 'port',
  45                 5 => 'path', 6 => 'file', 7 => 'fragment'
  46         );
  47         $default = array('');
  48         foreach(array_keys($array) as $uri) {
  49                 array_rename_keys($array[$uri], $parts, TRUE, $default);
  50                 $offset = $array[$uri]['scheme'][1]; // Scheme's offset
  51
  52                 foreach(array_keys($array[$uri]) as $part) {
  53                         // Remove offsets for each part
  54                         $array[$uri][$part] = & $array[$uri][$part][0];
  55                 }
  56
  57                 if ($normalize) {
  58                         $array[$uri]['scheme'] = scheme_normalize($array[$uri]['scheme']);
  59                         if ($array[$uri]['scheme'] === '') {
  60                                 unset ($array[$uri]);
  61                                 continue;
  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                         if ($preserve_rawuri) $array[$uri]['rawuri'] = & $array[$uri][0];
  67                 } else {
  68                         $array[$uri]['uri'] = & $array[$uri][0]; // Raw
  69                 }
  70                 unset($array[$uri][0]); // Matched string itself
  71                 if (! $preserve_chunk) {
  72                         unset(
  73                                 $array[$uri]['scheme'],
  74                                 $array[$uri]['userinfo'],
  75                                 $array[$uri]['host'],
  76                                 $array[$uri]['port'],
  77                                 $array[$uri]['path'],
  78                                 $array[$uri]['file'],
  79                                 $array[$uri]['fragment']
  80                         );
  81                 }
  82
  83                 $array[$uri]['area']['offset'] = $offset;
  84         }
  85
  86         return $array;
  87 }
  88
  89 // Domain exposure callback (See spam_uri_pickup_preprocess())
  90 // http://victim.example.org/?foo+site:nasty.example.com+bar
  91 // => http://nasty.example.com/?refer=victim.example.org
  92 // NOTE: 'refer=' is not so good for (at this time).
  93 // Consider about using IP address of the victim, try to avoid that.
  94 function _preg_replace_callback_domain_exposure($matches = array())
  95 {
  96         $result = '';
  97
  98         // Preserve the victim URI as a complicity or ...
  99         if (isset($matches[5])) {
 100                 $result =
 101                         $matches[1] . '://' .   // scheme
 102                         $matches[2] . '/' .             // victim.example.org
 103                         $matches[3];                    // The rest of all (before victim)
 104         }
 105
 106         // Flipped URI
 107         $result =
 108                 $matches[1] . '://' .   // scheme
 109                 $matches[4] .                   // nasty.example.com
 110                 '/?refer=' . strtolower($matches[2]) .  // victim.example.org
 111                 ' ' . $result;
 112
 113         return $result;
 114 }
 115
 116 // Preprocess: rawurldecode() and adding space(s) and something
 117 // to detect/count some URIs _if possible_
 118 // NOTE: It's maybe danger to var_dump(result). [e.g. 'javascript:']
 119 // [OK] http://victim.example.org/go?http%3A%2F%2Fnasty.example.org
 120 // [OK] http://victim.example.org/http://nasty.example.org
 121 function spam_uri_pickup_preprocess($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):/#i',
 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]['area']['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)) return FALSE;
 238
 239         // Nondestructive test
 240         if (! $force)
 241                 foreach(array_keys($keys) as $from)
 242                         if (! isset($array[$from]))
 243                                 return FALSE;
 244
 245         foreach($keys as $from => $to) {
 246                 if ($from === $to) continue;
 247                 if (! $force || isset($array[$from])) {
 248                         $array[$to] = & $array[$from];
 249                         unset($array[$from]);
 250                 } else  {
 251                         $array[$to] = $default;
 252                 }
 253         }
 254
 255         return TRUE;
 256 }
 257
 258 // If in doubt, it's a little doubtful
 259 function area_measure($areas, & $array, $belief = -1, $a_key = 'area', $o_key = 'offset')
 260 {
 261         if (! is_array($areas) || ! is_array($array)) return;
 262
 263         $areas_keys = array_keys($areas);
 264         foreach(array_keys($array) as $u_index) {
 265                 $offset = isset($array[$u_index][$o_key]) ?
 266                         intval($array[$u_index][$o_key]) : 0;
 267                 foreach($areas_keys as $a_index) {
 268                         if (isset($array[$u_index][$a_key])) {
 269                                 $offset_s = intval($areas[$a_index][0]);
 270                                 $offset_e = intval($areas[$a_index][1]);
 271                                 // [Area => inside <= Area]
 272                                 if ($offset_s < $offset && $offset < $offset_e) {
 273                                         $array[$u_index][$a_key] += $belief;
 274                                 }
 275                         }
 276                 }
 277         }
 278 }
 279
 280
 281 // ---------------------
 282 // Part Two
 283
 284 // Scheme normalization: Renaming the schemes
 285 // snntp://example.org =>  nntps://example.org
 286 // NOTE: Keep the static lists simple. See also port_normalize().
 287 function scheme_normalize($scheme = '', $considerd_harmfull = TRUE)
 288 {
 289         // Abbreviations considerable they don't have link intension
 290         static $abbrevs = array(
 291                 'ttp'   => 'http',
 292                 'ttps'  => 'https',
 293         );
 294
 295         // Alias => normalized
 296         static $aliases = array(
 297                 'pop'   => 'pop3',
 298                 'news'  => 'nntp',
 299                 'imap4' => 'imap',
 300                 'snntp' => 'nntps',
 301                 'snews' => 'nntps',
 302                 'spop3' => 'pop3s',
 303                 'pops'  => 'pop3s',
 304         );
 305
 306         $scheme = strtolower(trim($scheme));
 307         if (isset($abbrevs[$scheme])) {
 308                 if ($considerd_harmfull) {
 309                         $scheme = $abbrevs[$scheme];
 310                 } else {
 311                         $scheme = '';
 312                 }
 313         }
 314         if (isset($aliases[$scheme])) $scheme = $aliases[$scheme];
 315
 316         return $scheme;
 317 }
 318
 319 // Port normalization: Suppress the (redundant) default port
 320 // HTTP://example.org:80/ => http://example.org/
 321 // HTTP://example.org:8080/ => http://example.org:8080/
 322 // HTTPS://example.org:443/ => https://example.org/
 323 function port_normalize($port, $scheme, $scheme_normalize = TRUE)
 324 {
 325         // Schemes that users _maybe_ want to add protocol-handlers
 326         // to their web browsers. (and attackers _maybe_ want to use ...)
 327         // Reference: http://www.iana.org/assignments/port-numbers
 328         static $array = array(
 329                 // scheme => default port
 330                 'ftp'     =>    21,
 331                 'ssh'     =>    22,
 332                 'telnet'  =>    23,
 333                 'smtp'    =>    25,
 334                 'tftp'    =>    69,
 335                 'gopher'  =>    70,
 336                 'finger'  =>    79,
 337                 'http'    =>    80,
 338                 'pop3'    =>   110,
 339                 'sftp'    =>   115,
 340                 'nntp'    =>   119,
 341                 'imap'    =>   143,
 342                 'irc'     =>   194,
 343                 'wais'    =>   210,
 344                 'https'   =>   443,
 345                 'nntps'   =>   563,
 346                 'rsync'   =>   873,
 347                 'ftps'    =>   990,
 348                 'telnets' =>   992,
 349                 'imaps'   =>   993,
 350                 'ircs'    =>   994,
 351                 'pop3s'   =>   995,
 352                 'mysql'   =>  3306,
 353         );
 354
 355         $port = trim($port);
 356         if ($port === '') return $port;
 357
 358         if ($scheme_normalize) $scheme = scheme_normalize($scheme);
 359         if (isset($array[$scheme]) && $port == $array[$scheme])
 360                 $port = ''; // Ignore the defaults
 361
 362         return $port;
 363 }
 364
 365 // Path normalization
 366 // http://example.org => http://example.org/
 367 // http://example.org#hoge => http://example.org/#hoge
 368 // http://example.org/path/a/b/./c////./d => http://example.org/path/a/b/c/d
 369 // http://example.org/path/../../a/../back => http://example.org/back
 370 function path_normalize($path = '', $divider = '/', $addroot = TRUE)
 371 {
 372         if (! is_string($path) || $path == '') {
 373                 $path = $addroot ? $divider : '';
 374         } else {
 375                 $path = trim($path);
 376                 $last = ($path[strlen($path) - 1] == $divider) ? $divider : '';
 377                 $array = explode($divider, $path);
 378
 379                 // Remove paddings
 380                 foreach(array_keys($array) as $key) {
 381                         if ($array[$key] == '' || $array[$key] == '.')
 382                                  unset($array[$key]);
 383                 }
 384                 // Back-track
 385                 $tmp = array();
 386                 foreach($array as $value) {
 387                         if ($value == '..') {
 388                                 array_pop($tmp);
 389                         } else {
 390                                 array_push($tmp, $value);
 391                         }
 392                 }
 393                 $array = & $tmp;
 394
 395                 $path = $addroot ? $divider : '';
 396                 if (! empty($array)) $path .= implode($divider, $array) . $last;
 397         }
 398
 399         return $path;
 400 }
 401
 402 // An URI array => An URI (See uri_pickup())
 403 function uri_array_implode($uri = array())
 404 {
 405         if (empty($uri) || ! is_array($uri)) return NULL;
 406
 407         $tmp = array();
 408         if (isset($uri['scheme']) && $uri['scheme'] !== '') {
 409                 $tmp[] = & $uri['scheme'];
 410                 $tmp[] = '://';
 411         }
 412         if (isset($uri['userinfo']) && $uri['userinfo'] !== '') {
 413                 $tmp[] = & $uri['userinfo'];
 414                 $tmp[] = '@';
 415         }
 416         if (isset($uri['host']) && $uri['host'] !== '') {
 417                 $tmp[] = & $uri['host'];
 418         }
 419         if (isset($uri['port']) && $uri['port'] !== '') {
 420                 $tmp[] = ':';
 421                 $tmp[] = & $uri['port'];
 422         }
 423         if (isset($uri['path']) && $uri['path'] !== '') {
 424                 $tmp[] = & $uri['path'];
 425         }
 426         if (isset($uri['file']) && $uri['file'] !== '') {
 427                 $tmp[] = & $uri['file'];
 428         }
 429         if (isset($uri['fragment']) && $uri['fragment'] !== '') {
 430                 $tmp[] = '#';
 431                 $tmp[] = & $uri['fragment'];
 432         }
 433
 434         return implode('', $tmp);
 435 }
 436
 437 // ---------------------
 438 // Part One : Checker
 439
 440 function generate_glob_regex($string = '', $divider = '/')
 441 {
 442         static $from = array(
 443                          1 => '*',
 444                         11 => '?',
 445         //              22 => '[',      // Maybe cause regex compilation error (e.g. '[]')
 446         //              23 => ']',      //
 447                 );
 448         static $mid = array(
 449                          1 => '_AST_',
 450                         11 => '_QUE_',
 451         //              22 => '_RBR_',
 452         //              23 => '_LBR_',
 453                 );
 454         static $to = array(
 455                          1 => '.*',
 456                         11 => '.',
 457         //              22 => '[',
 458         //              23 => ']',
 459                 );
 460
 461         if (is_array($string)) {
 462                 // Recurse
 463                 return '(?:' .
 464                         implode('|',    // OR
 465                                 array_map('generate_glob_regex',
 466                                         $string,
 467                                         array_pad(array(), count($string), $divider)
 468                                 )
 469                         ) .
 470                 ')';
 471         } else {
 472                 $string = str_replace($from, $mid, $string); // Hide
 473                 $string = preg_quote($string, $divider);
 474                 $string = str_replace($mid, $to, $string);   // Unhide
 475                 return $string;
 476         }
 477 }
 478
 479 // TODO: Ignore list
 480 // TODO: preg_grep() ?
 481 // TODO: Multi list
 482 function is_badhost($hosts = '', $asap = TRUE)
 483 {
 484         static $regex;
 485
 486         if (! isset($regex)) {
 487                 $regex = array();
 488                 $regex['badhost'] = array();
 489
 490                 // Sample
 491                 if (TRUE) {
 492                         $blocklist['badhost'] = array(
 493                                 //'*',                  // Deny all uri
 494                                 //'10.20.*.*',  // 10.20.example.com also matches
 495                                 //'*.blogspot.com',     // Blog services subdomains
 496                                 //array('blogspot.com', '*.blogspot.com')
 497                         );
 498                         foreach ($blocklist['badhost'] as $part) {
 499                                 $regex['badhost'][] = '/^' . generate_glob_regex($part) . '$/i';
 500                         }
 501                 }
 502
 503                 // Load
 504                 if (file_exists(SPAM_INI_FILE)) {
 505                         $blocklist = array();
 506                         require(SPAM_INI_FILE);
 507                         foreach ($blocklist['badhost'] as $part) {
 508                                 $regex['badhost'][] = '/^' . generate_glob_regex($part) . '$/i';
 509                         }
 510                 }
 511         }
 512         //var_dump($regex);
 513
 514         $result = 0;
 515         if (! is_array($hosts)) $hosts = array($hosts);
 516
 517         foreach($hosts as $host) {
 518                 if (! is_string($host)) $host = '';
 519
 520                 // badhost
 521                 foreach ($regex['badhost'] as $_regex) {
 522                         if (preg_match($_regex, $host)) {
 523                                 ++$result;
 524                                 if ($asap) {
 525                                         return $result;
 526                                 } else {
 527                                         break;
 528                                 }
 529                         }
 530                 }
 531         }
 532
 533         return $result;
 534 }
 535
 536 // Default (enabled) methods and thresholds
 537 function check_uri_spam_method($times = 1, $t_area = 0, $rule = TRUE)
 538 {
 539         $times  = intval($times);
 540         $t_area = intval($t_area);
 541
 542         // Thresholds
 543         $method = array(
 544                 'quantity' => 8 * $times,       // Allow N URIs
 545                 'non_uniq' => 3 * $times,       // Allow N duped (and normalized) URIs
 546         );
 547
 548         // Areas
 549         $area = array(
 550                 //'total'  => $t_area,  // Allow N areas total, enabled below
 551                 'anchor'   => $t_area,  // Inside <a href> HTML tag
 552                 'bbcode'   => $t_area,  // Inside [url] or [link] BBCode
 553         );
 554
 555         // Rules
 556         $rules = array(
 557                 'badhost'  => TRUE,     // Check badhost
 558         );
 559
 560         // Remove unused
 561         foreach (array_keys($method) as $key) {
 562                 if ($method[$key] < 0) unset($method[$key]);
 563         }
 564         foreach (array_keys($area) as $key) {
 565                 if ($area[$key] < 0) unset($area[$key]);
 566         }
 567         $area  = empty($area) ? array() : array('area' => $area);
 568         $rules = $rule ? $rules : array();
 569
 570         return $method + $area + $rules;
 571 }
 572
 573
 574 // TODO return TRUE or FALSE!
 575 // Simple/fast spam check
 576 function check_uri_spam($target = '', $method = array(), $asap = TRUE)
 577 {
 578         $is_spam  = FALSE;
 579         $progress = array(
 580                 'quantity'   => 0,
 581                 'area' => array(
 582                         'total'  => 0,
 583                         'anchor' => 0,
 584                         'bbcode' => 0,
 585                         ),
 586                 'non_uniq'   => 0,
 587                 'uniqhost'   => 0,
 588                 'badhost'    => 0,
 589                 '_action'    => array(),
 590                 );
 591
 592         if (! is_array($method) || empty($method)) {
 593                 $method = check_uri_spam_method();
 594         }
 595
 596         if (is_array($target)) {
 597                 // Recurse
 598                 foreach($target as $str) {
 599                         list($_is_spam, $_progress) = check_uri_spam($str, $method, $asap);
 600                         $is_spam = $is_spam || $_is_spam;
 601                         $progress['quantity']       += $_progress['quantity'];
 602                         $progress['area']['total']  += $_progress['area']['total'];
 603                         $progress['area']['anchor'] += $_progress['area']['anchor'];
 604                         $progress['area']['bbcode'] += $_progress['area']['bbcode'];
 605                         $progress['non_uniq']       += $_progress['non_uniq'];
 606                         $progress['uniqhost']       += $_progress['uniqhost'];
 607                         $progress['badhost']        += $_progress['badhost'];
 608                         foreach($_progress['_action'] as $key => $value) {
 609                                 if (is_array($value)) {
 610                                         foreach(array_keys($value) as $_key) {
 611                                                 $progress['_action'][$key][$_key] = TRUE;
 612                                         }
 613                                 } else {
 614                                         $progress['_action'][$key] = TRUE;
 615                                 }
 616                         }
 617                         if ($is_spam && $asap) break;
 618                 }
 619         } else {
 620                 $pickups = spam_uri_pickup($target);
 621                 if (! empty($pickups)) {
 622                         $progress['quantity'] += count($pickups);
 623
 624                         // URI quantity
 625                         if ((! $is_spam || ! $asap) && isset($method['quantity']) &&
 626                                 $progress['quantity'] > $method['quantity']) {
 627                                 $is_spam = TRUE;
 628                                 $progress['_action']['quantity'] = TRUE;
 629                         }
 630                         //var_dump($method['quantity'], $is_spam);
 631
 632                         // Using invalid area
 633                         if ((! $is_spam || ! $asap) && isset($method['area'])) {
 634                                 foreach($pickups as $pickup) {
 635                                         foreach ($pickup['area'] as $key => $value) {
 636                                                 if ($key == 'offset') continue;
 637                                                 $progress['area']['total'] += $value;
 638                                                 $progress['area'][$key]    += $value;
 639                                                 if (isset($method['area']['total']) &&
 640                                                                 $progress['area']['total'] > $method['area']['total']) {
 641                                                         $is_spam = TRUE;
 642                                                         $progress['_action']['area']['total'] = TRUE;
 643                                                         if ($is_spam && $asap) break;
 644                                                 }
 645                                                 if(isset($method['area'][$key]) &&
 646                                                                 $progress['area'][$key] > $method['area'][$key]) {
 647                                                         $is_spam = TRUE;
 648                                                         $progress['_action']['area'][$key] = TRUE;
 649                                                         if ($is_spam && $asap) break;
 650                                                 }
 651                                         }
 652                                         if ($is_spam && $asap) break;
 653                                 }
 654                         }
 655                         //var_dump($method['area'], $is_spam);
 656
 657                         // URI uniqueness (and removing non-uniques)
 658                         if ((! $is_spam || ! $asap) && isset($method['non_uniq'])) {
 659                                 $uris = array();
 660                                 foreach ($pickups as $key => $pickup) {
 661                                         $uris[$key] = uri_array_implode($pickup);
 662                                 }
 663                                 $count = count($uris);
 664                                 $uris = array_unique($uris);
 665                                 $progress['non_uniq'] += $count - count($uris);
 666                                 if ($progress['non_uniq'] > $method['non_uniq']) {
 667                                         $is_spam = TRUE;
 668                                         $progress['_action']['non_uniq'] = TRUE;
 669                                 }
 670                                 if (! $asap || ! $is_spam) {
 671                                         foreach (array_diff(array_keys($pickups),
 672                                                 array_keys($uris)) as $remove) {
 673                                                 unset($pickups[$remove]);
 674                                         }
 675                                 }
 676                                 unset($uris);
 677                         }
 678                         //var_dump($method['non_uniq'], $is_spam);
 679
 680                         // Unique host
 681                         $hosts = array();
 682                         foreach ($pickups as $pickup) {
 683                                 $hosts[] = & $pickup['host'];
 684                         }
 685                         $hosts = array_unique($hosts);
 686                         $progress['uniqhost'] += count($hosts);
 687                         //var_dump($method['uniqhost'], $is_spam);
 688
 689                         // Bad host
 690                         if ((! $is_spam || ! $asap) && isset($method['badhost'])) {
 691                                 $count = is_badhost($hosts, $asap);
 692                                 $progress['badhost'] += $count;
 693                                 if ($count !== 0) {
 694                                         $progress['_action']['badhost'] = TRUE;
 695                                         $is_spam = TRUE;
 696                                 }
 697                         }
 698                         //var_dump($method['badhost'], $is_spam);
 699                 }
 700         }
 701
 702         return array($is_spam, $progress);
 703 }
 704
 705 // ---------------------
 706 // Reporting
 707
 708 // TODO: Show all
 709 // Summarize $progress (blocked only)
 710 function summarize_check_uri_spam_progress($progress = array(), $shownum = TRUE)
 711 {
 712         //$list = check_uri_spam_method();
 713
 714         $tmp = array();
 715         foreach (array_keys($progress['_action']) as $_action) {
 716                 if (is_array($progress['_action'][$_action])) {
 717                         foreach (array_keys($progress['_action'][$_action]) as $_area) {
 718                                 $tmp[] = $_action . '=>' . $_area .
 719                                         ($shownum ? '(' . $progress[$_action][$_area] . ')' : '');
 720                         }
 721                 } else {
 722                         $tmp[] = $_action .
 723                                         ($shownum ? '(' . $progress[$_action] . ')' : '');
 724                 }
 725         }
 726
 727         return implode(', ', $tmp);
 728 }
 729
 730 // ---------------------
 731 // Exit
 732
 733 // Common bahavior for blocking
 734 // NOTE: Call this function from various blocking feature, to disgueise the reason 'why blocked'
 735 function spam_exit()
 736 {
 737         die("\n");
 738 }
 739
 740
 741 // ---------------------
 742 //
 743
 744 // Simple/fast spam filter ($target: 'a string' or an array())
 745 function pkwk_spamfilter($action, $page, $target = array('title' => ''), $method = array(), $asap = FALSE)
 746 {
 747         global $notify;
 748
 749         list($is_spam, $progress) = check_uri_spam($target, $method, $asap);
 750
 751         // Mail to administrator(s)
 752         if ($is_spam) {
 753                 if ($notify) {
 754                         pkwk_spamnotify($action, $page, $target, $progress);
 755                 }
 756                 spam_exit();
 757         }
 758 }
 759
 760 // ---------------------
 761 // PukiWiki original
 762
 763 // Mail to administrator(s)
 764 function pkwk_spamnotify($action, $page, $target = array('title' => ''), $progress = array())
 765 {
 766         global $notify_subject;
 767
 768         $footer['BLOCKED'] = 'Blocked by: ' .
 769                 summarize_check_uri_spam_progress($progress);
 770         $footer['ACTION'] = 'Blocked: ' . $action;
 771         $footer['PAGE']   = '[blocked] ' . $page;
 772         $footer['URI']    = get_script_uri() . '?' . rawurlencode($page);
 773         $footer['USER_AGENT']  = TRUE;
 774         $footer['REMOTE_ADDR'] = TRUE;
 775         pkwk_mail_notify($notify_subject,  var_export($target, TRUE), $footer);
 776 }
 777
 778 ?>