spam.php

   1 <?php
   2 // $Id: spam.php,v 1.63 2006/12/10 02:31:18 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] = $default;
 174                 }
 175         }
 176
 177         return TRUE;
 178 }
 179
 180 // ---------------------
 181 // Area pickup
 182
 183 // Pickup all of markup areas
 184 function area_pickup($string = '', $method = array())
 185 {
 186         $area = array();
 187
 188         // Anchor tags by preg_match_all()
 189         // [OK] <a href=></a>
 190         // [OK] <a href=  >Good site!</a>
 191         // [OK] <a href= "#" >test</a>
 192         // [OK] <a href="http://nasty.example.com">visit http://nasty.example.com/</a>
 193         // [OK] <a href=\'http://nasty.example.com/\' >discount foobar</a>
 194         // [NG] <a href="http://ng.example.com">visit http://ng.example.com _not_ended_
 195         if (isset($method['anchor'])) {
 196                 $areas = array();
 197                 preg_match_all('#<a\b[^>]*\bhref\b[^>]*>.*?</a\b[^>]*(>)#i',
 198                          $string, $areas, PREG_SET_ORDER | PREG_OFFSET_CAPTURE);
 199                 //var_dump(recursive_map('htmlspecialchars', $areas));
 200                 foreach(array_keys($areas) as $_area) {
 201                         $areas[$_area] =  array(
 202                                 $areas[$_area][0][1], // Area start (<a href>)
 203                                 $areas[$_area][1][1], // Area end   (</a>)
 204                         );
 205                 }
 206                 if (! empty($areas)) $area['anchor'] = $areas;
 207         }
 208
 209         // phpBB's "BBCode" by preg_match_all()
 210         // [OK] [url][/url]
 211         // [OK] [url]http://nasty.example.com/[/url]
 212         // [OK] [link]http://nasty.example.com/[/link]
 213         // [OK] [url=http://nasty.example.com]visit http://nasty.example.com/[/url]
 214         // [OK] [link http://nasty.example.com/]buy something[/link]
 215         if (isset($method['bbcode'])) {
 216                 $areas = array();
 217                 preg_match_all('#\[(url|link)\b[^\]]*\].*?\[/\1\b[^\]]*(\])#i',
 218                          $string, $areas, PREG_SET_ORDER | PREG_OFFSET_CAPTURE);
 219                 //var_dump(recursive_map('htmlspecialchars', $areas));
 220                 foreach(array_keys($areas) as $_area) {
 221                         $areas[$_area] = array(
 222                                 $areas[$_area][0][1], // Area start ([url])
 223                                 $areas[$_area][2][1], // Area end   ([/url])
 224                         );
 225                 }
 226                 if (! empty($areas)) $area['bbcode'] = $areas;
 227         }
 228
 229         // Various Wiki syntax
 230         // [text_or_uri>text_or_uri]
 231         // [text_or_uri:text_or_uri]
 232         // [text_or_uri|text_or_uri]
 233         // [text_or_uri->text_or_uri]
 234         // [text_or_uri text_or_uri] // MediaWiki
 235         // MediaWiki: [http://nasty.example.com/ visit http://nasty.example.com/]
 236
 237         return $area;
 238 }
 239
 240 // If in doubt, it's a little doubtful
 241 function area_measure($areas, & $array, $belief = -1, $a_key = 'area', $o_key = 'offset')
 242 {
 243         if (! is_array($areas) || ! is_array($array)) return;
 244
 245         $areas_keys = array_keys($areas);
 246         foreach(array_keys($array) as $u_index) {
 247                 $offset = isset($array[$u_index][$o_key]) ?
 248                         intval($array[$u_index][$o_key]) : 0;
 249                 foreach($areas_keys as $a_index) {
 250                         if (isset($array[$u_index][$a_key])) {
 251                                 $offset_s = intval($areas[$a_index][0]);
 252                                 $offset_e = intval($areas[$a_index][1]);
 253                                 // [Area => inside <= Area]
 254                                 if ($offset_s < $offset && $offset < $offset_e) {
 255                                         $array[$u_index][$a_key] += $belief;
 256                                 }
 257                         }
 258                 }
 259         }
 260 }
 261
 262 // ---------------------
 263 // Spam-uri pickup
 264
 265 // Domain exposure callback (See spam_uri_pickup_preprocess())
 266 // http://victim.example.org/?foo+site:nasty.example.com+bar
 267 // => http://nasty.example.com/?refer=victim.example.org
 268 // NOTE: 'refer=' is not so good for (at this time).
 269 // Consider about using IP address of the victim, try to avoid that.
 270 function _preg_replace_callback_domain_exposure($matches = array())
 271 {
 272         $result = '';
 273
 274         // Preserve the victim URI as a complicity or ...
 275         if (isset($matches[5])) {
 276                 $result =
 277                         $matches[1] . '://' .   // scheme
 278                         $matches[2] . '/' .             // victim.example.org
 279                         $matches[3];                    // The rest of all (before victim)
 280         }
 281
 282         // Flipped URI
 283         if (isset($matches[4])) {
 284                 $result =
 285                         $matches[1] . '://' .   // scheme
 286                         $matches[4] .                   // nasty.example.com
 287                         '/?refer=' . strtolower($matches[2]) .  // victim.example.org
 288                         ' ' . $result;
 289         }
 290
 291         return $result;
 292 }
 293
 294 // Preprocess: rawurldecode() and adding space(s) and something
 295 // to detect/count some URIs _if possible_
 296 // NOTE: It's maybe danger to var_dump(result). [e.g. 'javascript:']
 297 // [OK] http://victim.example.org/go?http%3A%2F%2Fnasty.example.org
 298 // [OK] http://victim.example.org/http://nasty.example.org
 299 function spam_uri_pickup_preprocess($string = '')
 300 {
 301         if (! is_string($string)) return '';
 302
 303         $string = rawurldecode($string);
 304
 305         // Domain exposure (See _preg_replace_callback_domain_exposure())
 306         $string = preg_replace_callback(
 307                 array(
 308                         // Something Google: http://www.google.com/supported_domains
 309                         '#(http)://([a-z0-9.]+\.google\.[a-z]{2,3}(?:\.[a-z]{2})?)/' .
 310                         '([a-z0-9?=&.%_+-]+)' .         // ?query=foo+
 311                         '\bsite:([a-z0-9.%_-]+)' .      // site:nasty.example.com
 312                         //'()' .        // Preserve or remove?
 313                         '#i',
 314                 ),
 315                 '_preg_replace_callback_domain_exposure',
 316                 $string
 317         );
 318
 319         // URI exposure (uriuri => uri uri)
 320         $string = preg_replace(
 321                 array(
 322                         '#(?<! )(?:https?|ftp):/#i',
 323                 //      '#[a-z][a-z0-9.+-]{1,8}://#i',
 324                 //      '#[a-z][a-z0-9.+-]{1,8}://#i'
 325                 ),
 326                 ' $0',
 327                 $string
 328         );
 329
 330         return $string;
 331 }
 332
 333 // Main function of spam-uri pickup
 334 function spam_uri_pickup($string = '', $area = array())
 335 {
 336         if (! is_array($area) || empty($area)) {
 337                 $area = array(
 338                                 'anchor' => TRUE,
 339                                 'bbcode' => TRUE,
 340                         );
 341         }
 342
 343         $string = spam_uri_pickup_preprocess($string);
 344
 345         $array  = uri_pickup($string);
 346
 347         // Area elevation for '(especially external)link' intension
 348         if (! empty($array)) {
 349                 $areas = area_pickup($string, $area);
 350                 if (! empty($areas)) {
 351                         $area_shadow = array();
 352                         foreach(array_keys($array) as $key){
 353                                 $area_shadow[$key] = & $array[$key]['area'];
 354                                 $area_shadow[$key]['anchor'] = 0;
 355                                 $area_shadow[$key]['bbcode'] = 0;
 356                         }
 357                         if (isset($areas['anchor'])) {
 358                                 area_measure($areas['anchor'], $area_shadow, 1, 'anchor');
 359                         }
 360                         if (isset($areas['bbcode'])) {
 361                                 area_measure($areas['bbcode'], $area_shadow, 1, 'bbcode');
 362                         }
 363                 }
 364         }
 365
 366         // Remove 'offset's for area_measure()
 367         foreach(array_keys($array) as $key)
 368                 unset($array[$key]['area']['offset']);
 369
 370         return $array;
 371 }
 372
 373
 374 // ---------------------
 375 // Normalization
 376
 377 // Scheme normalization: Renaming the schemes
 378 // snntp://example.org =>  nntps://example.org
 379 // NOTE: Keep the static lists simple. See also port_normalize().
 380 function scheme_normalize($scheme = '', $considerd_harmfull = TRUE)
 381 {
 382         // Abbreviations considerable they don't have link intension
 383         static $abbrevs = array(
 384                 'ttp'   => 'http',
 385                 'ttps'  => 'https',
 386         );
 387
 388         // Alias => normalized
 389         static $aliases = array(
 390                 'pop'   => 'pop3',
 391                 'news'  => 'nntp',
 392                 'imap4' => 'imap',
 393                 'snntp' => 'nntps',
 394                 'snews' => 'nntps',
 395                 'spop3' => 'pop3s',
 396                 'pops'  => 'pop3s',
 397         );
 398
 399         $scheme = strtolower(trim($scheme));
 400         if (isset($abbrevs[$scheme])) {
 401                 if ($considerd_harmfull) {
 402                         $scheme = $abbrevs[$scheme];
 403                 } else {
 404                         $scheme = '';
 405                 }
 406         }
 407         if (isset($aliases[$scheme])) $scheme = $aliases[$scheme];
 408
 409         return $scheme;
 410 }
 411
 412 // Port normalization: Suppress the (redundant) default port
 413 // HTTP://example.org:80/ => http://example.org/
 414 // HTTP://example.org:8080/ => http://example.org:8080/
 415 // HTTPS://example.org:443/ => https://example.org/
 416 function port_normalize($port, $scheme, $scheme_normalize = TRUE)
 417 {
 418         // Schemes that users _maybe_ want to add protocol-handlers
 419         // to their web browsers. (and attackers _maybe_ want to use ...)
 420         // Reference: http://www.iana.org/assignments/port-numbers
 421         static $array = array(
 422                 // scheme => default port
 423                 'ftp'     =>    21,
 424                 'ssh'     =>    22,
 425                 'telnet'  =>    23,
 426                 'smtp'    =>    25,
 427                 'tftp'    =>    69,
 428                 'gopher'  =>    70,
 429                 'finger'  =>    79,
 430                 'http'    =>    80,
 431                 'pop3'    =>   110,
 432                 'sftp'    =>   115,
 433                 'nntp'    =>   119,
 434                 'imap'    =>   143,
 435                 'irc'     =>   194,
 436                 'wais'    =>   210,
 437                 'https'   =>   443,
 438                 'nntps'   =>   563,
 439                 'rsync'   =>   873,
 440                 'ftps'    =>   990,
 441                 'telnets' =>   992,
 442                 'imaps'   =>   993,
 443                 'ircs'    =>   994,
 444                 'pop3s'   =>   995,
 445                 'mysql'   =>  3306,
 446         );
 447
 448         $port = trim($port);
 449         if ($port === '') return $port;
 450
 451         if ($scheme_normalize) $scheme = scheme_normalize($scheme);
 452         if (isset($array[$scheme]) && $port == $array[$scheme])
 453                 $port = ''; // Ignore the defaults
 454
 455         return $port;
 456 }
 457
 458 // Path normalization
 459 // http://example.org => http://example.org/
 460 // http://example.org#hoge => http://example.org/#hoge
 461 // http://example.org/path/a/b/./c////./d => http://example.org/path/a/b/c/d
 462 // http://example.org/path/../../a/../back => http://example.org/back
 463 function path_normalize($path = '', $divider = '/', $addroot = TRUE)
 464 {
 465         if (! is_string($path) || $path == '')
 466                 return $addroot ? $divider : '';
 467
 468         $path = trim($path);
 469         $last = ($path[strlen($path) - 1] == $divider) ? $divider : '';
 470         $array = explode($divider, $path);
 471
 472         // Remove paddings
 473         foreach(array_keys($array) as $key) {
 474                 if ($array[$key] == '' || $array[$key] == '.')
 475                          unset($array[$key]);
 476         }
 477         // Back-track
 478         $tmp = array();
 479         foreach($array as $value) {
 480                 if ($value == '..') {
 481                         array_pop($tmp);
 482                 } else {
 483                         array_push($tmp, $value);
 484                 }
 485         }
 486         $array = & $tmp;
 487
 488         $path = $addroot ? $divider : '';
 489         if (! empty($array)) $path .= implode($divider, $array) . $last;
 490
 491         return $path;
 492 }
 493
 494 // DirectoryIndex normalize (Destructive and rough)
 495 function file_normalize($string = 'index.html.en')
 496 {
 497         static $array = array(
 498                 'index'                 => TRUE,        // Some system can omit the suffix
 499                 'index.htm'             => TRUE,
 500                 'index.html'    => TRUE,
 501                 'index.shtml'   => TRUE,
 502                 'index.jsp'             => TRUE,
 503                 'index.php'             => TRUE,
 504                 'index.php3'    => TRUE,
 505                 'index.php4'    => TRUE,
 506                 //'index.pl'    => TRUE,
 507                 //'index.py'    => TRUE,
 508                 //'index.rb'    => TRUE,
 509                 'index.cgi'             => TRUE,
 510                 'default.htm'   => TRUE,
 511                 'default.html'  => TRUE,
 512                 'default.asp'   => TRUE,
 513                 'default.aspx'  => TRUE,
 514         );
 515
 516         // Content-negothiation filter:
 517         // Roughly removing ISO 639 -like
 518         // 2-letter suffixes (See RFC3066)
 519         $matches = array();
 520         if (preg_match('/(.*)\.[a-z][a-z](?:-[a-z][a-z])?$/i', $string, $matches)) {
 521                 $_string = $matches[1];
 522         } else {
 523                 $_string = & $string;
 524         }
 525
 526         if (isset($array[strtolower($_string)])) {
 527                 return '';
 528         } else {
 529                 return $string;
 530         }
 531 }
 532
 533 // Sort query-strings if possible (Destructive and rough)
 534 // [OK] &&&&f=d&b&d&c&a=0dd  =>  a=0dd&b&c&d&f=d
 535 // [OK] nothing==&eg=dummy&eg=padding&eg=foobar  =>  eg=foobar
 536 function query_normalize($string = '', $equal = FALSE, $equal_cutempty = TRUE)
 537 {
 538         $array = explode('&', $string);
 539
 540         // Remove '&' paddings
 541         foreach(array_keys($array) as $key) {
 542                 if ($array[$key] == '') {
 543                          unset($array[$key]);
 544                 }
 545         }
 546
 547         // Consider '='-sepalated input and paddings
 548         if ($equal) {
 549                 $equals = $not_equals = array();
 550                 foreach ($array as $part) {
 551                         if (strpos($part, '=') === FALSE) {
 552                                  $not_equals[] = $part;
 553                         } else {
 554                                 list($key, $value) = explode('=', $part, 2);
 555                                 $value = ltrim($value, '=');
 556                                 if (! $equal_cutempty || $value != '') {
 557                                         $equals[$key] = $value;
 558                                 }
 559                         }
 560                 }
 561
 562                 $array = & $not_equals;
 563                 foreach ($equals as $key => $value) {
 564                         $array[] = $key . '=' . $value;
 565                 }
 566                 unset($equals);
 567         }
 568
 569         natsort($array);
 570         return implode('&', $array);
 571 }
 572
 573 // ---------------------
 574 // Part One : Checker
 575
 576 function generate_glob_regex($string = '', $divider = '/')
 577 {
 578         static $from = array(
 579                          1 => '*',
 580                         11 => '?',
 581         //              22 => '[',      // Maybe cause regex compilation error (e.g. '[]')
 582         //              23 => ']',      //
 583                 );
 584         static $mid = array(
 585                          1 => '_AST_',
 586                         11 => '_QUE_',
 587         //              22 => '_RBR_',
 588         //              23 => '_LBR_',
 589                 );
 590         static $to = array(
 591                          1 => '.*',
 592                         11 => '.',
 593         //              22 => '[',
 594         //              23 => ']',
 595                 );
 596
 597         if (is_array($string)) {
 598                 // Recurse
 599                 return '(?:' .
 600                         implode('|',    // OR
 601                                 array_map('generate_glob_regex',
 602                                         $string,
 603                                         array_pad(array(), count($string), $divider)
 604                                 )
 605                         ) .
 606                 ')';
 607         } else {
 608                 $string = str_replace($from, $mid, $string); // Hide
 609                 $string = preg_quote($string, $divider);
 610                 $string = str_replace($mid, $to, $string);   // Unhide
 611                 return $string;
 612         }
 613 }
 614
 615 // TODO: Ignore list
 616 // TODO: preg_grep() ?
 617 // TODO: Multi list
 618 function is_badhost($hosts = '', $asap = TRUE)
 619 {
 620         static $regex;
 621
 622         if (! isset($regex)) {
 623                 $regex = array();
 624                 $regex['badhost'] = array();
 625
 626                 // Sample
 627                 if (TRUE) {
 628                         $blocklist['badhost'] = array(
 629                                 //'*',                  // Deny all uri
 630                                 //'10.20.*.*',  // 10.20.example.com also matches
 631                                 //'*.blogspot.com',     // Blog services subdomains
 632                                 //array('blogspot.com', '*.blogspot.com')
 633                         );
 634                         foreach ($blocklist['badhost'] as $part) {
 635                                 $regex['badhost'][] = '/^' . generate_glob_regex($part) . '$/i';
 636                         }
 637                 }
 638
 639                 // Load
 640                 if (file_exists(SPAM_INI_FILE)) {
 641                         $blocklist = array();
 642                         require(SPAM_INI_FILE);
 643                         foreach ($blocklist['badhost'] as $part) {
 644                                 $regex['badhost'][] = '/^' . generate_glob_regex($part) . '$/i';
 645                         }
 646                 }
 647         }
 648         //var_dump($regex);
 649
 650         $result = 0;
 651         if (! is_array($hosts)) $hosts = array($hosts);
 652
 653         foreach($hosts as $host) {
 654                 if (! is_string($host)) $host = '';
 655
 656                 // badhost
 657                 foreach ($regex['badhost'] as $_regex) {
 658                         if (preg_match($_regex, $host)) {
 659                                 ++$result;
 660                                 if ($asap) {
 661                                         return $result;
 662                                 } else {
 663                                         break;
 664                                 }
 665                         }
 666                 }
 667         }
 668
 669         return $result;
 670 }
 671
 672 // Default (enabled) methods and thresholds
 673 function check_uri_spam_method($times = 1, $t_area = 0, $rule = TRUE)
 674 {
 675         $times  = intval($times);
 676         $t_area = intval($t_area);
 677
 678         // Thresholds
 679         $method = array(
 680                 'quantity' => 8 * $times,       // Allow N URIs
 681                 'non_uniq' => 3 * $times,       // Allow N duped (and normalized) URIs
 682         );
 683
 684         // Areas
 685         $area = array(
 686                 //'total' => $t_area,   // Allow N areas total, enabled below
 687                 'anchor'  => $t_area,   // Inside <a href> HTML tag
 688                 'bbcode'  => $t_area,   // Inside [url] or [link] BBCode
 689         );
 690
 691         // Rules
 692         $rules = array(
 693                 'asap'     => FALSE,    // Quit As Soon As Possible
 694                 'uniqhost' => TRUE,     // Show uniq host (at block notification mail)
 695                 'badhost'  => TRUE,     // Check badhost
 696         );
 697
 698         // Remove unused
 699         foreach (array_keys($method) as $key) {
 700                 if ($method[$key] < 0) unset($method[$key]);
 701         }
 702         foreach (array_keys($area) as $key) {
 703                 if ($area[$key] < 0) unset($area[$key]);
 704         }
 705         $area  = empty($area) ? array() : array('area' => $area);
 706         $rules = $rule ? $rules : array();
 707
 708         return $method + $area + $rules;
 709 }
 710
 711 // TODO: Simplify $progress data structure
 712 // TODO: Simplify. !empty(['is_spam']) just means $is_spam
 713 // Simple/fast spam check
 714 function check_uri_spam($target = '', $method = array())
 715 {
 716         $is_spam  = FALSE;
 717         if (! is_array($method) || empty($method)) {
 718                 $method = check_uri_spam_method();
 719         }
 720         $asap = isset($method['asap']) ? $method['asap'] : TRUE;
 721
 722         $progress = array(
 723                 'sum' =>  array(
 724                         'quantity'    => 0,
 725                         'uniqhost'    => 0,
 726                         'non_uniq'    => 0,
 727                         'badhost'     => 0,
 728                         'area_total'  => 0,
 729                         'area_anchor' => 0,
 730                         'area_bbcode' => 0,
 731                         ),
 732                 'is_spam' => array(),
 733                 'method' => & $method,
 734         );
 735
 736
 737         if (is_array($target)) {
 738                 // Recurse
 739                 foreach($target as $str) {
 740                         list($_is_spam, $_progress) = check_uri_spam($str, $method);
 741                         $is_spam = $is_spam || $_is_spam;
 742                         foreach (array_keys($_progress['sum']) as $key) {
 743                                 $progress['sum'][$key] += $_progress['sum'][$key];
 744                         }
 745                         foreach(array_keys($_progress['is_spam']) as $key) {
 746                                 $progress['is_spam'][$key] = TRUE;
 747                         }
 748                         if ($is_spam && $asap) break;
 749                 }
 750         } else {
 751                 $pickups = spam_uri_pickup($target);
 752                 if (! empty($pickups)) {
 753                         $progress['sum']['quantity'] += count($pickups);
 754
 755                         // URI quantity
 756                         if ((! $is_spam || ! $asap) && isset($method['quantity']) &&
 757                                 $progress['sum']['quantity'] > $method['quantity']) {
 758                                 $is_spam = TRUE;
 759                                 $progress['is_spam']['quantity'] = TRUE;
 760                         }
 761                         //var_dump($method['quantity'], $is_spam);
 762
 763                         // Using invalid area
 764                         if ((! $is_spam || ! $asap) && isset($method['area'])) {
 765                                 foreach($pickups as $pickup) {
 766                                         foreach ($pickup['area'] as $key => $value) {
 767                                                 if ($key == 'offset') continue;
 768                                                 $p_key = 'area_' . $key;
 769                                                 $progress['sum']['area_total'] += $value;
 770                                                 $progress['sum'][$p_key]       += $value;
 771                                                 if (isset($method['area']['total']) &&
 772                                                                 $progress['sum']['area_total'] > $method['area']['total']) {
 773                                                         $is_spam = TRUE;
 774                                                         $progress['is_spam']['area_total'] = TRUE;
 775                                                         if ($is_spam && $asap) break;
 776                                                 }
 777                                                 if(isset($method['area'][$key]) &&
 778                                                                 $progress['sum'][$p_key] > $method['area'][$key]) {
 779                                                         $is_spam = TRUE;
 780                                                         $progress['is_spam'][$p_key] = TRUE;
 781                                                         if ($is_spam && $asap) break;
 782                                                 }
 783                                         }
 784                                         if ($is_spam && $asap) break;
 785                                 }
 786                         }
 787                         //var_dump($method['area'], $is_spam);
 788
 789
 790                         // URI uniqueness (and removing non-uniques)
 791                         if ((! $is_spam || ! $asap) && isset($method['non_uniq'])) {
 792
 793                                 // Destructive normalize of URIs
 794                                 uri_array_normalize($pickups);
 795
 796                                 $uris = array();
 797                                 foreach (array_keys($pickups) as $key) {
 798                                         $uris[$key] = uri_array_implode($pickups[$key]);
 799                                 }
 800                                 $count = count($uris);
 801                                 $uris  = array_unique($uris);
 802                                 $progress['sum']['non_uniq'] += $count - count($uris);
 803                                 if ($progress['sum']['non_uniq'] > $method['non_uniq']) {
 804                                         $is_spam = TRUE;
 805                                         $progress['is_spam']['non_uniq'] = TRUE;
 806                                 }
 807                                 if (! $asap || ! $is_spam) {
 808                                         foreach (array_diff(array_keys($pickups),
 809                                                 array_keys($uris)) as $remove) {
 810                                                 unset($pickups[$remove]);
 811                                         }
 812                                 }
 813                                 unset($uris);
 814                         }
 815                         //var_dump($method['non_uniq'], $is_spam);
 816
 817                         // Unique host
 818                         $hosts = array();
 819                         foreach ($pickups as $pickup) {
 820                                 $hosts[] = & $pickup['host'];
 821                         }
 822                         $hosts = array_unique($hosts);
 823                         $progress['sum']['uniqhost'] += count($hosts);
 824                         //var_dump($method['uniqhost'], $is_spam);
 825
 826                         // Bad host
 827                         if ((! $is_spam || ! $asap) && isset($method['badhost'])) {
 828                                 $count = is_badhost($hosts, $asap);
 829                                 $progress['sum']['badhost'] += $count;
 830                                 if ($count !== 0) {
 831                                         $progress['is_spam']['badhost'] = TRUE;
 832                                         $is_spam = TRUE;
 833                                 }
 834                         }
 835                         //var_dump($method['badhost'], $is_spam);
 836                 }
 837         }
 838
 839         return array($is_spam, $progress);
 840 }
 841
 842 // ---------------------
 843 // Reporting
 844
 845 // TODO: Don't show unused $method!
 846 // Summarize $progress (blocked only)
 847 function summarize_spam_progress($progress = array(), $blockedonly = FALSE)
 848 {
 849         $method = $progress['method'];
 850         if (isset($method['area'])) {
 851                 foreach(array_keys($method['area']) as $key) {
 852                         $method['area_' . $key] = TRUE;
 853                 }
 854         }
 855
 856         if ($blockedonly) {
 857                 $tmp = array_keys($progress['is_spam']);
 858         } else {
 859                 $tmp = array();
 860                 foreach ($progress['sum'] as $key => $value) {
 861                         if (isset($method[$key])) {
 862                                 $tmp[] = $key . '(' . $value . ')';
 863                         }
 864                 }
 865         }
 866
 867         return implode(', ', $tmp);
 868 }
 869
 870 // ---------------------
 871 // Exit
 872
 873 // Common bahavior for blocking
 874 // NOTE: Call this function from various blocking feature, to disgueise the reason 'why blocked'
 875 function spam_exit()
 876 {
 877         die("\n");
 878 }
 879
 880
 881 // ---------------------
 882 // Simple filtering
 883
 884 // TODO: Record them
 885 // Simple/fast spam filter ($target: 'a string' or an array())
 886 function pkwk_spamfilter($action, $page, $target = array('title' => ''), $method = array())
 887 {
 888         global $notify;
 889
 890         list($is_spam, $progress) = check_uri_spam($target, $method);
 891
 892         if ($is_spam) {
 893                 // Mail to administrator(s)
 894                 if ($notify) pkwk_spamnotify($action, $page, $target, $progress, $method);
 895                 // End
 896                 spam_exit();
 897         }
 898 }
 899
 900 // ---------------------
 901 // PukiWiki original
 902
 903 // Mail to administrator(s)
 904 function pkwk_spamnotify($action, $page, $target = array('title' => ''), $progress = array(), $method = array())
 905 {
 906         global $notify_subject;
 907
 908         $asap = isset($method['asap']) ? $method['asap'] : TRUE;
 909
 910         $footer['ACTION']  = 'Blocked by: ' . summarize_spam_progress($progress, TRUE);
 911
 912         if (! $asap) {
 913                 $footer['METRICS'] = summarize_spam_progress($progress);
 914         }
 915         $footer['COMMENT'] = $action;
 916         $footer['PAGE']    = '[blocked] ' . $page;
 917         $footer['URI']     = get_script_uri() . '?' . rawurlencode($page);
 918         $footer['USER_AGENT']  = TRUE;
 919         $footer['REMOTE_ADDR'] = TRUE;
 920         pkwk_mail_notify($notify_subject,  var_export($target, TRUE), $footer);
 921 }
 922
 923 ?>