spam/spam.php

   1 <?php
   2 // $Id: spam.php,v 1.62 2006/12/09 12:18:52 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 // ---------------------
  11 // URI pickup
  12
  13 // Return an array of URIs in the $string
  14 // [OK] http://nasty.example.org#nasty_string
  15 // [OK] http://nasty.example.org:80/foo/xxx#nasty_string/bar
  16 // [OK] ftp://nasty.example.org:80/dfsdfs
  17 // [OK] ftp://cnn.example.com&story=breaking_news@10.0.0.1/top_story.htm (from RFC3986)
  18 function uri_pickup($string = '', $normalize = TRUE,
  19         $preserve_rawuri = FALSE, $preserve_chunk = TRUE)
  20 {
  21         // Not available for: IDN(ignored)
  22         $array = array();
  23         preg_match_all(
  24                 // scheme://userinfo@host:port/path/or/pathinfo/maybefile.and?query=string#fragment
  25                 // Refer RFC3986 (Regex below is not strict)
  26                 '#(\b[a-z][a-z0-9.+-]{1,8})://' .       // 1: Scheme
  27                 '(?:' .
  28                         '([^\s<>"\'\[\]/\#?@]*)' .              // 2: Userinfo (Username)
  29                 '@)?' .
  30                 '(' .
  31                         // 3: Host
  32                         '\[[0-9a-f:.]+\]' . '|' .                               // IPv6([colon-hex and dot]): RFC2732
  33                         '(?:[0-9]{1-3}\.){3}[0-9]{1-3}' . '|' . // IPv4(dot-decimal): 001.22.3.44
  34                         '[^\s<>"\'\[\]:/\#?]+' .                                // FQDN: foo.example.org
  35                 ')' .
  36                 '(?::([0-9]*))?' .                                      // 4: Port
  37                 '((?:/+[^\s<>"\'\[\]/\#]+)*/+)?' .      // 5: Directory path or path-info
  38                 '([^\s<>"\'\[\]\#?]+)?' .                       // 6: File?
  39                 '(?:\?([^\s<>"\'\[\]\#]+))?' .          // 7: Query string
  40                 '(?:\#([a-z0-9._~%!$&\'()*+,;=:@-]*))?' .       // 8: Fragment
  41                 '#i',
  42                  $string, $array, PREG_SET_ORDER | PREG_OFFSET_CAPTURE
  43         );
  44         //var_dump(recursive_map('htmlspecialchars', $array));
  45
  46         // Shrink $array
  47         static $parts = array(
  48                 1 => 'scheme', 2 => 'userinfo', 3 => 'host', 4 => 'port',
  49                 5 => 'path', 6 => 'file', 7 => 'query', 8 => 'fragment'
  50         );
  51         $default = array('');
  52         foreach(array_keys($array) as $uri) {
  53                 $_uri = & $array[$uri];
  54                 array_rename_keys($_uri, $parts, TRUE, $default);
  55
  56                 $offset = $_uri['scheme'][1]; // Scheme's offset
  57                 foreach(array_keys($_uri) as $part) {
  58                         // Remove offsets for each part
  59                         $_uri[$part] = & $_uri[$part][0];
  60                 }
  61
  62                 if ($normalize) {
  63                         $_uri['scheme'] = scheme_normalize($_uri['scheme']);
  64                         if ($_uri['scheme'] === '') {
  65                                 unset ($array[$uri]);
  66                                 continue;
  67                         }
  68                         $_uri['host']  = strtolower($_uri['host']);
  69                         $_uri['port']  = port_normalize($_uri['port'], $_uri['scheme'], FALSE);
  70                         $_uri['path']  = path_normalize($_uri['path']);
  71                         if ($preserve_rawuri) $_uri['rawuri'] = & $_uri[0];
  72
  73                         // DEBUG
  74                         //$_uri['uri'] = uri_array_implode($_uri);
  75                 } else {
  76                         $_uri['uri'] = & $_uri[0]; // Raw
  77                 }
  78                 unset($_uri[0]); // Matched string itself
  79                 if (! $preserve_chunk) {
  80                         unset(
  81                                 $_uri['scheme'],
  82                                 $_uri['userinfo'],
  83                                 $_uri['host'],
  84                                 $_uri['port'],
  85                                 $_uri['path'],
  86                                 $_uri['file'],
  87                                 $_uri['query'],
  88                                 $_uri['fragment']
  89                         );
  90                 }
  91
  92                 // Area offset for area_measure()
  93                 $_uri['area']['offset'] = $offset;
  94         }
  95
  96         return $array;
  97 }
  98
  99 // Destructive normalize of URI array
 100 // NOTE: Give me the uri_pickup() result with chunks
 101 function uri_array_normalize(& $pickups, $preserve = TRUE)
 102 {
 103         if (! is_array($pickups)) return $pickups;
 104
 105         foreach (array_keys($pickups) as $key) {
 106                 $_key = & $pickups[$key];
 107                 $_key['path']     = isset($_key['path']) ? strtolower($_key['path']) : '';
 108                 $_key['file']     = isset($_key['file']) ? file_normalize($_key['file']) : '';
 109                 $_key['query']    = isset($_key['query']) ? query_normalize(strtolower($_key['query']), TRUE) : '';
 110                 $_key['fragment'] = (isset($_key['fragment']) && $preserve) ?
 111                         strtolower($_key['fragment']) : ''; // Just ignore
 112         }
 113
 114         return $pickups;
 115 }
 116
 117 // An URI array => An URI (See uri_pickup())
 118 function uri_array_implode($uri = array())
 119 {
 120         if (empty($uri) || ! is_array($uri)) return NULL;
 121
 122         $tmp = array();
 123         if (isset($uri['scheme']) && $uri['scheme'] !== '') {
 124                 $tmp[] = & $uri['scheme'];
 125                 $tmp[] = '://';
 126         }
 127         if (isset($uri['userinfo']) && $uri['userinfo'] !== '') {
 128                 $tmp[] = & $uri['userinfo'];
 129                 $tmp[] = '@';
 130         }
 131         if (isset($uri['host']) && $uri['host'] !== '') {
 132                 $tmp[] = & $uri['host'];
 133         }
 134         if (isset($uri['port']) && $uri['port'] !== '') {
 135                 $tmp[] = ':';
 136                 $tmp[] = & $uri['port'];
 137         }
 138         if (isset($uri['path']) && $uri['path'] !== '') {
 139                 $tmp[] = & $uri['path'];
 140         }
 141         if (isset($uri['file']) && $uri['file'] !== '') {
 142                 $tmp[] = & $uri['file'];
 143         }
 144         if (isset($uri['query']) && $uri['query'] !== '') {
 145                 $tmp[] = '?';
 146                 $tmp[] = & $uri['query'];
 147         }
 148         if (isset($uri['fragment']) && $uri['fragment'] !== '') {
 149                 $tmp[] = '#';
 150                 $tmp[] = & $uri['fragment'];
 151         }
 152
 153         return implode('', $tmp);
 154 }
 155
 156 // $array['something'] => $array['wanted']
 157 function array_rename_keys(& $array, $keys = array('from' => 'to'), $force = FALSE, $default = '')
 158 {
 159         if (! is_array($array) || ! is_array($keys)) return FALSE;
 160
 161         // Nondestructive test
 162         if (! $force)
 163                 foreach(array_keys($keys) as $from)
 164                         if (! isset($array[$from]))
 165                                 return FALSE;
 166
 167         foreach($keys as $from => $to) {
 168                 if ($from === $to) continue;
 169                 if (! $force || isset($array[$from])) {
 170                         $array[$to] = & $array[$from];
 171                         unset($array[$from]);
 172                 } else  {
 173                         $array[$to] = $default;
 174                 }
 175         }
 176
 177         return TRUE;
 178 }
 179
 180 // ---------------------
 181 // Spam-uri pickup
 182
 183 // Domain exposure callback (See spam_uri_pickup_preprocess())
 184 // http://victim.example.org/?foo+site:nasty.example.com+bar
 185 // => http://nasty.example.com/?refer=victim.example.org
 186 // NOTE: 'refer=' is not so good for (at this time).
 187 // Consider about using IP address of the victim, try to avoid that.
 188 function _preg_replace_callback_domain_exposure($matches = array())
 189 {
 190         $result = '';
 191
 192         // Preserve the victim URI as a complicity or ...
 193         if (isset($matches[5])) {
 194                 $result =
 195                         $matches[1] . '://' .   // scheme
 196                         $matches[2] . '/' .             // victim.example.org
 197                         $matches[3];                    // The rest of all (before victim)
 198         }
 199
 200         // Flipped URI
 201         if (isset($matches[4])) {
 202                 $result =
 203                         $matches[1] . '://' .   // scheme
 204                         $matches[4] .                   // nasty.example.com
 205                         '/?refer=' . strtolower($matches[2]) .  // victim.example.org
 206                         ' ' . $result;
 207         }
 208
 209         return $result;
 210 }
 211
 212 // Preprocess: rawurldecode() and adding space(s) and something
 213 // to detect/count some URIs _if possible_
 214 // NOTE: It's maybe danger to var_dump(result). [e.g. 'javascript:']
 215 // [OK] http://victim.example.org/go?http%3A%2F%2Fnasty.example.org
 216 // [OK] http://victim.example.org/http://nasty.example.org
 217 function spam_uri_pickup_preprocess($string = '')
 218 {
 219         if (! is_string($string)) return '';
 220
 221         $string = rawurldecode($string);
 222
 223         // Domain exposure (See _preg_replace_callback_domain_exposure())
 224         $string = preg_replace_callback(
 225                 array(
 226                         // Something Google: http://www.google.com/supported_domains
 227                         '#(http)://([a-z0-9.]+\.google\.[a-z]{2,3}(?:\.[a-z]{2})?)/' .
 228                         '([a-z0-9?=&.%_+-]+)' .         // ?query=foo+
 229                         '\bsite:([a-z0-9.%_-]+)' .      // site:nasty.example.com
 230                         //'()' .        // Preserve or remove?
 231                         '#i',
 232                 ),
 233                 '_preg_replace_callback_domain_exposure',
 234                 $string
 235         );
 236
 237         // URI exposure (uriuri => uri uri)
 238         $string = preg_replace(
 239                 array(
 240                         '#(?<! )(?:https?|ftp):/#i',
 241                 //      '#[a-z][a-z0-9.+-]{1,8}://#i',
 242                 //      '#[a-z][a-z0-9.+-]{1,8}://#i'
 243                 ),
 244                 ' $0',
 245                 $string
 246         );
 247
 248         return $string;
 249 }
 250
 251 // Main function of spam-uri pickup
 252 function spam_uri_pickup($string = '', $area = array())
 253 {
 254         if (! is_array($area) || empty($area)) {
 255                 $area = array(
 256                                 'anchor' => TRUE,
 257                                 'bbcode' => TRUE,
 258                         );
 259         }
 260
 261         $string = spam_uri_pickup_preprocess($string);
 262
 263         $array  = uri_pickup($string);
 264
 265         // Area elevation for '(especially external)link' intension
 266         if (! empty($array)) {
 267
 268                 $area_shadow = array();
 269                 foreach(array_keys($array) as $key){
 270                         $area_shadow[$key] = & $array[$key]['area'];
 271                         $area_shadow[$key]['anchor'] = 0;
 272                         $area_shadow[$key]['bbcode'] = 0;
 273                 }
 274
 275                 // Anchor tags by preg_match_all()
 276                 // [OK] <a href="http://nasty.example.com">visit http://nasty.example.com/</a>
 277                 // [OK] <a href=\'http://nasty.example.com/\' >discount foobar</a>
 278                 // [NG] <a href="http://ng.example.com">visit http://ng.example.com _not_ended_
 279                 // [NG] <a href=  >Good site!</a> <a href= "#" >test</a>
 280                 if (isset($area['anchor'])) {
 281                         $areas = array();
 282                         preg_match_all('#<a\b[^>]*href[^>]*>.*?</a\b[^>]*(>)#i',
 283                                  $string, $areas, PREG_SET_ORDER | PREG_OFFSET_CAPTURE);
 284                         //var_dump(recursive_map('htmlspecialchars', $areas));
 285                         foreach(array_keys($areas) as $_area) {
 286                                 $areas[$_area] =  array(
 287                                         $areas[$_area][0][1], // Area start (<a href>)
 288                                         $areas[$_area][1][1], // Area end   (</a>)
 289                                 );
 290                         }
 291                         area_measure($areas, $area_shadow, 1, 'anchor');
 292                 }
 293
 294                 // phpBB's "BBCode" by preg_match_all()
 295                 // [url]http://nasty.example.com/[/url]
 296                 // [link]http://nasty.example.com/[/link]
 297                 // [url=http://nasty.example.com]visit http://nasty.example.com/[/url]
 298                 // [link http://nasty.example.com/]buy something[/link]
 299                 // ?? [url=][/url]
 300                 if (isset($area['bbcode'])) {
 301                         $areas = array();
 302                         preg_match_all('#\[(url|link)\b[^\]]*\].*?\[/\1\b[^\]]*(\])#i',
 303                                  $string, $areas, PREG_SET_ORDER | PREG_OFFSET_CAPTURE);
 304                         //var_dump(recursive_map('htmlspecialchars', $areas));
 305                         foreach(array_keys($areas) as $_area) {
 306                                 $areas[$_area] = array(
 307                                         $areas[$_area][0][1], // Area start ([url])
 308                                         $areas[$_area][2][1], // Area end   ([/url])
 309                                 );
 310                         }
 311                         area_measure($areas, $area_shadow, 1, 'bbcode');
 312                 }
 313
 314                 // Various Wiki syntax
 315                 // [text_or_uri>text_or_uri]
 316                 // [text_or_uri:text_or_uri]
 317                 // [text_or_uri|text_or_uri]
 318                 // [text_or_uri->text_or_uri]
 319                 // [text_or_uri text_or_uri] // MediaWiki
 320                 // MediaWiki: [http://nasty.example.com/ visit http://nasty.example.com/]
 321
 322                 // Remove 'offset's for area_measure()
 323                 foreach(array_keys($array) as $key)
 324                         unset($array[$key]['area']['offset']);
 325         }
 326
 327         return $array;
 328 }
 329
 330 // If in doubt, it's a little doubtful
 331 function area_measure($areas, & $array, $belief = -1, $a_key = 'area', $o_key = 'offset')
 332 {
 333         if (! is_array($areas) || ! is_array($array)) return;
 334
 335         $areas_keys = array_keys($areas);
 336         foreach(array_keys($array) as $u_index) {
 337                 $offset = isset($array[$u_index][$o_key]) ?
 338                         intval($array[$u_index][$o_key]) : 0;
 339                 foreach($areas_keys as $a_index) {
 340                         if (isset($array[$u_index][$a_key])) {
 341                                 $offset_s = intval($areas[$a_index][0]);
 342                                 $offset_e = intval($areas[$a_index][1]);
 343                                 // [Area => inside <= Area]
 344                                 if ($offset_s < $offset && $offset < $offset_e) {
 345                                         $array[$u_index][$a_key] += $belief;
 346                                 }
 347                         }
 348                 }
 349         }
 350 }
 351
 352 // ---------------------
 353 // Normalization
 354
 355 // Scheme normalization: Renaming the schemes
 356 // snntp://example.org =>  nntps://example.org
 357 // NOTE: Keep the static lists simple. See also port_normalize().
 358 function scheme_normalize($scheme = '', $considerd_harmfull = TRUE)
 359 {
 360         // Abbreviations considerable they don't have link intension
 361         static $abbrevs = array(
 362                 'ttp'   => 'http',
 363                 'ttps'  => 'https',
 364         );
 365
 366         // Alias => normalized
 367         static $aliases = array(
 368                 'pop'   => 'pop3',
 369                 'news'  => 'nntp',
 370                 'imap4' => 'imap',
 371                 'snntp' => 'nntps',
 372                 'snews' => 'nntps',
 373                 'spop3' => 'pop3s',
 374                 'pops'  => 'pop3s',
 375         );
 376
 377         $scheme = strtolower(trim($scheme));
 378         if (isset($abbrevs[$scheme])) {
 379                 if ($considerd_harmfull) {
 380                         $scheme = $abbrevs[$scheme];
 381                 } else {
 382                         $scheme = '';
 383                 }
 384         }
 385         if (isset($aliases[$scheme])) $scheme = $aliases[$scheme];
 386
 387         return $scheme;
 388 }
 389
 390 // Port normalization: Suppress the (redundant) default port
 391 // HTTP://example.org:80/ => http://example.org/
 392 // HTTP://example.org:8080/ => http://example.org:8080/
 393 // HTTPS://example.org:443/ => https://example.org/
 394 function port_normalize($port, $scheme, $scheme_normalize = TRUE)
 395 {
 396         // Schemes that users _maybe_ want to add protocol-handlers
 397         // to their web browsers. (and attackers _maybe_ want to use ...)
 398         // Reference: http://www.iana.org/assignments/port-numbers
 399         static $array = array(
 400                 // scheme => default port
 401                 'ftp'     =>    21,
 402                 'ssh'     =>    22,
 403                 'telnet'  =>    23,
 404                 'smtp'    =>    25,
 405                 'tftp'    =>    69,
 406                 'gopher'  =>    70,
 407                 'finger'  =>    79,
 408                 'http'    =>    80,
 409                 'pop3'    =>   110,
 410                 'sftp'    =>   115,
 411                 'nntp'    =>   119,
 412                 'imap'    =>   143,
 413                 'irc'     =>   194,
 414                 'wais'    =>   210,
 415                 'https'   =>   443,
 416                 'nntps'   =>   563,
 417                 'rsync'   =>   873,
 418                 'ftps'    =>   990,
 419                 'telnets' =>   992,
 420                 'imaps'   =>   993,
 421                 'ircs'    =>   994,
 422                 'pop3s'   =>   995,
 423                 'mysql'   =>  3306,
 424         );
 425
 426         $port = trim($port);
 427         if ($port === '') return $port;
 428
 429         if ($scheme_normalize) $scheme = scheme_normalize($scheme);
 430         if (isset($array[$scheme]) && $port == $array[$scheme])
 431                 $port = ''; // Ignore the defaults
 432
 433         return $port;
 434 }
 435
 436 // Path normalization
 437 // http://example.org => http://example.org/
 438 // http://example.org#hoge => http://example.org/#hoge
 439 // http://example.org/path/a/b/./c////./d => http://example.org/path/a/b/c/d
 440 // http://example.org/path/../../a/../back => http://example.org/back
 441 function path_normalize($path = '', $divider = '/', $addroot = TRUE)
 442 {
 443         if (! is_string($path) || $path == '')
 444                 return $addroot ? $divider : '';
 445
 446         $path = trim($path);
 447         $last = ($path[strlen($path) - 1] == $divider) ? $divider : '';
 448         $array = explode($divider, $path);
 449
 450         // Remove paddings
 451         foreach(array_keys($array) as $key) {
 452                 if ($array[$key] == '' || $array[$key] == '.')
 453                          unset($array[$key]);
 454         }
 455         // Back-track
 456         $tmp = array();
 457         foreach($array as $value) {
 458                 if ($value == '..') {
 459                         array_pop($tmp);
 460                 } else {
 461                         array_push($tmp, $value);
 462                 }
 463         }
 464         $array = & $tmp;
 465
 466         $path = $addroot ? $divider : '';
 467         if (! empty($array)) $path .= implode($divider, $array) . $last;
 468
 469         return $path;
 470 }
 471
 472 // DirectoryIndex normalize (Destructive and rough)
 473 function file_normalize($string = 'index.html.en')
 474 {
 475         static $array = array(
 476                 'index'                 => TRUE,        // Some system can omit the suffix
 477                 'index.htm'             => TRUE,
 478                 'index.html'    => TRUE,
 479                 'index.shtml'   => TRUE,
 480                 'index.jsp'             => TRUE,
 481                 'index.php'             => TRUE,
 482                 'index.php3'    => TRUE,
 483                 'index.php4'    => TRUE,
 484                 //'index.pl'    => TRUE,
 485                 //'index.py'    => TRUE,
 486                 //'index.rb'    => TRUE,
 487                 'index.cgi'             => TRUE,
 488                 'default.htm'   => TRUE,
 489                 'default.html'  => TRUE,
 490                 'default.asp'   => TRUE,
 491                 'default.aspx'  => TRUE,
 492         );
 493
 494         // Content-negothiation filter:
 495         // Roughly removing ISO 639 -like
 496         // 2-letter suffixes (See RFC3066)
 497         $matches = array();
 498         if (preg_match('/(.*)\.[a-z][a-z](?:-[a-z][a-z])?$/i', $string, $matches)) {
 499                 $_string = $matches[1];
 500         } else {
 501                 $_string = & $string;
 502         }
 503
 504         if (isset($array[strtolower($_string)])) {
 505                 return '';
 506         } else {
 507                 return $string;
 508         }
 509 }
 510
 511 // Sort query-strings if possible (Destructive and rough)
 512 // [OK] &&&&f=d&b&d&c&a=0dd  =>  a=0dd&b&c&d&f=d
 513 // [OK] nothing==&eg=dummy&eg=padding&eg=foobar  =>  eg=foobar
 514 function query_normalize($string = '', $equal = FALSE, $equal_cutempty = TRUE)
 515 {
 516         $array = explode('&', $string);
 517
 518         // Remove '&' paddings
 519         foreach(array_keys($array) as $key) {
 520                 if ($array[$key] == '') {
 521                          unset($array[$key]);
 522                 }
 523         }
 524
 525         // Consider '='-sepalated input and paddings
 526         if ($equal) {
 527                 $equals = $not_equals = array();
 528                 foreach ($array as $part) {
 529                         if (strpos($part, '=') === FALSE) {
 530                                  $not_equals[] = $part;
 531                         } else {
 532                                 list($key, $value) = explode('=', $part, 2);
 533                                 $value = ltrim($value, '=');
 534                                 if (! $equal_cutempty || $value != '') {
 535                                         $equals[$key] = $value;
 536                                 }
 537                         }
 538                 }
 539
 540                 $array = & $not_equals;
 541                 foreach ($equals as $key => $value) {
 542                         $array[] = $key . '=' . $value;
 543                 }
 544                 unset($equals);
 545         }
 546
 547         natsort($array);
 548         return implode('&', $array);
 549 }
 550
 551 // ---------------------
 552 // Part One : Checker
 553
 554 function generate_glob_regex($string = '', $divider = '/')
 555 {
 556         static $from = array(
 557                          1 => '*',
 558                         11 => '?',
 559         //              22 => '[',      // Maybe cause regex compilation error (e.g. '[]')
 560         //              23 => ']',      //
 561                 );
 562         static $mid = array(
 563                          1 => '_AST_',
 564                         11 => '_QUE_',
 565         //              22 => '_RBR_',
 566         //              23 => '_LBR_',
 567                 );
 568         static $to = array(
 569                          1 => '.*',
 570                         11 => '.',
 571         //              22 => '[',
 572         //              23 => ']',
 573                 );
 574
 575         if (is_array($string)) {
 576                 // Recurse
 577                 return '(?:' .
 578                         implode('|',    // OR
 579                                 array_map('generate_glob_regex',
 580                                         $string,
 581                                         array_pad(array(), count($string), $divider)
 582                                 )
 583                         ) .
 584                 ')';
 585         } else {
 586                 $string = str_replace($from, $mid, $string); // Hide
 587                 $string = preg_quote($string, $divider);
 588                 $string = str_replace($mid, $to, $string);   // Unhide
 589                 return $string;
 590         }
 591 }
 592
 593 // TODO: Ignore list
 594 // TODO: preg_grep() ?
 595 // TODO: Multi list
 596 function is_badhost($hosts = '', $asap = TRUE)
 597 {
 598         static $regex;
 599
 600         if (! isset($regex)) {
 601                 $regex = array();
 602                 $regex['badhost'] = array();
 603
 604                 // Sample
 605                 if (TRUE) {
 606                         $blocklist['badhost'] = array(
 607                                 //'*',                  // Deny all uri
 608                                 //'10.20.*.*',  // 10.20.example.com also matches
 609                                 //'*.blogspot.com',     // Blog services subdomains
 610                                 //array('blogspot.com', '*.blogspot.com')
 611                         );
 612                         foreach ($blocklist['badhost'] as $part) {
 613                                 $regex['badhost'][] = '/^' . generate_glob_regex($part) . '$/i';
 614                         }
 615                 }
 616
 617                 // Load
 618                 if (file_exists(SPAM_INI_FILE)) {
 619                         $blocklist = array();
 620                         require(SPAM_INI_FILE);
 621                         foreach ($blocklist['badhost'] as $part) {
 622                                 $regex['badhost'][] = '/^' . generate_glob_regex($part) . '$/i';
 623                         }
 624                 }
 625         }
 626         //var_dump($regex);
 627
 628         $result = 0;
 629         if (! is_array($hosts)) $hosts = array($hosts);
 630
 631         foreach($hosts as $host) {
 632                 if (! is_string($host)) $host = '';
 633
 634                 // badhost
 635                 foreach ($regex['badhost'] as $_regex) {
 636                         if (preg_match($_regex, $host)) {
 637                                 ++$result;
 638                                 if ($asap) {
 639                                         return $result;
 640                                 } else {
 641                                         break;
 642                                 }
 643                         }
 644                 }
 645         }
 646
 647         return $result;
 648 }
 649
 650 // Default (enabled) methods and thresholds
 651 function check_uri_spam_method($times = 1, $t_area = 0, $rule = TRUE)
 652 {
 653         $times  = intval($times);
 654         $t_area = intval($t_area);
 655
 656         // Thresholds
 657         $method = array(
 658                 'quantity' => 8 * $times,       // Allow N URIs
 659                 'non_uniq' => 3 * $times,       // Allow N duped (and normalized) URIs
 660         );
 661
 662         // Areas
 663         $area = array(
 664                 //'total' => $t_area,   // Allow N areas total, enabled below
 665                 'anchor'  => $t_area,   // Inside <a href> HTML tag
 666                 'bbcode'  => $t_area,   // Inside [url] or [link] BBCode
 667         );
 668
 669         // Rules
 670         $rules = array(
 671                 'asap'     => FALSE,    // Quit As Soon As Possible
 672                 'uniqhost' => TRUE,     // Show uniq host (at block notification mail)
 673                 'badhost'  => TRUE,     // Check badhost
 674         );
 675
 676         // Remove unused
 677         foreach (array_keys($method) as $key) {
 678                 if ($method[$key] < 0) unset($method[$key]);
 679         }
 680         foreach (array_keys($area) as $key) {
 681                 if ($area[$key] < 0) unset($area[$key]);
 682         }
 683         $area  = empty($area) ? array() : array('area' => $area);
 684         $rules = $rule ? $rules : array();
 685
 686         return $method + $area + $rules;
 687 }
 688
 689 // TODO: Simplify $progress data structure
 690 // TODO: Simplify. !empty(['is_spam']) just means $is_spam
 691 // Simple/fast spam check
 692 function check_uri_spam($target = '', $method = array())
 693 {
 694         $is_spam  = FALSE;
 695         if (! is_array($method) || empty($method)) {
 696                 $method = check_uri_spam_method();
 697         }
 698         $asap = isset($method['asap']) ? $method['asap'] : TRUE;
 699
 700         $progress = array(
 701                 'sum' =>  array(
 702                         'quantity'    => 0,
 703                         'uniqhost'    => 0,
 704                         'non_uniq'    => 0,
 705                         'badhost'     => 0,
 706                         'area_total'  => 0,
 707                         'area_anchor' => 0,
 708                         'area_bbcode' => 0,
 709                         ),
 710                 'is_spam' => array(),
 711                 'method' => & $method,
 712         );
 713
 714
 715         if (is_array($target)) {
 716                 // Recurse
 717                 foreach($target as $str) {
 718                         list($_is_spam, $_progress) = check_uri_spam($str, $method);
 719                         $is_spam = $is_spam || $_is_spam;
 720                         foreach (array_keys($_progress['sum']) as $key) {
 721                                 $progress['sum'][$key] += $_progress['sum'][$key];
 722                         }
 723                         foreach(array_keys($_progress['is_spam']) as $key) {
 724                                 $progress['is_spam'][$key] = TRUE;
 725                         }
 726                         if ($is_spam && $asap) break;
 727                 }
 728         } else {
 729                 $pickups = spam_uri_pickup($target);
 730                 if (! empty($pickups)) {
 731                         $progress['sum']['quantity'] += count($pickups);
 732
 733                         // URI quantity
 734                         if ((! $is_spam || ! $asap) && isset($method['quantity']) &&
 735                                 $progress['sum']['quantity'] > $method['quantity']) {
 736                                 $is_spam = TRUE;
 737                                 $progress['is_spam']['quantity'] = TRUE;
 738                         }
 739                         //var_dump($method['quantity'], $is_spam);
 740
 741                         // Using invalid area
 742                         if ((! $is_spam || ! $asap) && isset($method['area'])) {
 743                                 foreach($pickups as $pickup) {
 744                                         foreach ($pickup['area'] as $key => $value) {
 745                                                 if ($key == 'offset') continue;
 746                                                 $p_key = 'area_' . $key;
 747                                                 $progress['sum']['area_total'] += $value;
 748                                                 $progress['sum'][$p_key]       += $value;
 749                                                 if (isset($method['area']['total']) &&
 750                                                                 $progress['sum']['area_total'] > $method['area']['total']) {
 751                                                         $is_spam = TRUE;
 752                                                         $progress['is_spam']['area_total'] = TRUE;
 753                                                         if ($is_spam && $asap) break;
 754                                                 }
 755                                                 if(isset($method['area'][$key]) &&
 756                                                                 $progress['sum'][$p_key] > $method['area'][$key]) {
 757                                                         $is_spam = TRUE;
 758                                                         $progress['is_spam'][$p_key] = TRUE;
 759                                                         if ($is_spam && $asap) break;
 760                                                 }
 761                                         }
 762                                         if ($is_spam && $asap) break;
 763                                 }
 764                         }
 765                         //var_dump($method['area'], $is_spam);
 766
 767
 768                         // URI uniqueness (and removing non-uniques)
 769                         if ((! $is_spam || ! $asap) && isset($method['non_uniq'])) {
 770
 771                                 // Destructive normalize of URIs
 772                                 uri_array_normalize($pickups);
 773
 774                                 $uris = array();
 775                                 foreach (array_keys($pickups) as $key) {
 776                                         $uris[$key] = uri_array_implode($pickups[$key]);
 777                                 }
 778                                 $count = count($uris);
 779                                 $uris  = array_unique($uris);
 780                                 $progress['sum']['non_uniq'] += $count - count($uris);
 781                                 if ($progress['sum']['non_uniq'] > $method['non_uniq']) {
 782                                         $is_spam = TRUE;
 783                                         $progress['is_spam']['non_uniq'] = TRUE;
 784                                 }
 785                                 if (! $asap || ! $is_spam) {
 786                                         foreach (array_diff(array_keys($pickups),
 787                                                 array_keys($uris)) as $remove) {
 788                                                 unset($pickups[$remove]);
 789                                         }
 790                                 }
 791                                 unset($uris);
 792                         }
 793                         //var_dump($method['non_uniq'], $is_spam);
 794
 795                         // Unique host
 796                         $hosts = array();
 797                         foreach ($pickups as $pickup) {
 798                                 $hosts[] = & $pickup['host'];
 799                         }
 800                         $hosts = array_unique($hosts);
 801                         $progress['sum']['uniqhost'] += count($hosts);
 802                         //var_dump($method['uniqhost'], $is_spam);
 803
 804                         // Bad host
 805                         if ((! $is_spam || ! $asap) && isset($method['badhost'])) {
 806                                 $count = is_badhost($hosts, $asap);
 807                                 $progress['sum']['badhost'] += $count;
 808                                 if ($count !== 0) {
 809                                         $progress['is_spam']['badhost'] = TRUE;
 810                                         $is_spam = TRUE;
 811                                 }
 812                         }
 813                         //var_dump($method['badhost'], $is_spam);
 814                 }
 815         }
 816
 817         return array($is_spam, $progress);
 818 }
 819
 820 // ---------------------
 821 // Reporting
 822
 823 // TODO: Don't show unused $method!
 824 // Summarize $progress (blocked only)
 825 function summarize_spam_progress($progress = array(), $blockedonly = FALSE)
 826 {
 827         $method = $progress['method'];
 828         if (isset($method['area'])) {
 829                 foreach(array_keys($method['area']) as $key) {
 830                         $method['area_' . $key] = TRUE;
 831                 }
 832         }
 833
 834         if ($blockedonly) {
 835                 $tmp = array_keys($progress['is_spam']);
 836         } else {
 837                 $tmp = array();
 838                 foreach ($progress['sum'] as $key => $value) {
 839                         if (isset($method[$key])) {
 840                                 $tmp[] = $key . '(' . $value . ')';
 841                         }
 842                 }
 843         }
 844
 845         return implode(', ', $tmp);
 846 }
 847
 848 // ---------------------
 849 // Exit
 850
 851 // Common bahavior for blocking
 852 // NOTE: Call this function from various blocking feature, to disgueise the reason 'why blocked'
 853 function spam_exit()
 854 {
 855         die("\n");
 856 }
 857
 858
 859 // ---------------------
 860 // Simple filtering
 861
 862 // TODO: Record them
 863 // Simple/fast spam filter ($target: 'a string' or an array())
 864 function pkwk_spamfilter($action, $page, $target = array('title' => ''), $method = array())
 865 {
 866         global $notify;
 867
 868         list($is_spam, $progress) = check_uri_spam($target, $method);
 869
 870         if ($is_spam) {
 871                 // Mail to administrator(s)
 872                 if ($notify) pkwk_spamnotify($action, $page, $target, $progress, $method);
 873                 // End
 874                 spam_exit();
 875         }
 876 }
 877
 878 // ---------------------
 879 // PukiWiki original
 880
 881 // Mail to administrator(s)
 882 function pkwk_spamnotify($action, $page, $target = array('title' => ''), $progress = array(), $method = array())
 883 {
 884         global $notify_subject;
 885
 886         $asap = isset($method['asap']) ? $method['asap'] : TRUE;
 887
 888         $footer['ACTION']  = 'Blocked by: ' . summarize_spam_progress($progress, TRUE);
 889
 890         if (! $asap) {
 891                 $footer['METRICS'] = summarize_spam_progress($progress);
 892         }
 893         $footer['COMMENT'] = $action;
 894         $footer['PAGE']    = '[blocked] ' . $page;
 895         $footer['URI']     = get_script_uri() . '?' . rawurlencode($page);
 896         $footer['USER_AGENT']  = TRUE;
 897         $footer['REMOTE_ADDR'] = TRUE;
 898         pkwk_mail_notify($notify_subject,  var_export($target, TRUE), $footer);
 899 }
 900
 901 ?>