spam/spam.php

   1 <?php
   2 // $Id: spam.php,v 1.124 2007/03/04 12:19:08 henoheno Exp $
   3 // Copyright (C) 2006-2007 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 // (PHP 4 >= 4.3.0): preg_match_all(PREG_OFFSET_CAPTURE): $method['uri_XXX'] related feature
   9
  10 if (! defined('SPAM_INI_FILE')) define('SPAM_INI_FILE', 'spam.ini.php');
  11
  12 // ---------------------
  13 // Compat etc
  14
  15 // (PHP 4 >= 4.2.0): var_export(): mail-reporting and dump related
  16 if (! function_exists('var_export')) {
  17         function var_export() {
  18                 return 'var_export() is not found on this server' . "\n";
  19         }
  20 }
  21
  22 // (PHP 4 >= 4.2.0): preg_grep() enables invert option
  23 function preg_grep_invert($pattern = '//', $input = array())
  24 {
  25         static $invert;
  26         if (! isset($invert)) $invert = defined('PREG_GREP_INVERT');
  27
  28         if ($invert) {
  29                 return preg_grep($pattern, $input, PREG_GREP_INVERT);
  30         } else {
  31                 $result = preg_grep($pattern, $input);
  32                 if ($result) {
  33                         return array_diff($input, preg_grep($pattern, $input));
  34                 } else {
  35                         return $input;
  36                 }
  37         }
  38 }
  39
  40 // ---------------------
  41 // URI pickup
  42
  43 // Return an array of URIs in the $string
  44 // [OK] http://nasty.example.org#nasty_string
  45 // [OK] http://nasty.example.org:80/foo/xxx#nasty_string/bar
  46 // [OK] ftp://nasty.example.org:80/dfsdfs
  47 // [OK] ftp://cnn.example.com&story=breaking_news@10.0.0.1/top_story.htm (from RFC3986)
  48 function uri_pickup($string = '', $normalize = TRUE,
  49         $preserve_rawuri = FALSE, $preserve_chunk = TRUE)
  50 {
  51         // Not available for: IDN(ignored)
  52         $array = array();
  53         preg_match_all(
  54                 // scheme://userinfo@host:port/path/or/pathinfo/maybefile.and?query=string#fragment
  55                 // Refer RFC3986 (Regex below is not strict)
  56                 '#(\b[a-z][a-z0-9.+-]{1,8}):/+' .       // 1: Scheme
  57                 '(?:' .
  58                         '([^\s<>"\'\[\]/\#?@]*)' .              // 2: Userinfo (Username)
  59                 '@)?' .
  60                 '(' .
  61                         // 3: Host
  62                         '\[[0-9a-f:.]+\]' . '|' .                               // IPv6([colon-hex and dot]): RFC2732
  63                         '(?:[0-9]{1-3}\.){3}[0-9]{1-3}' . '|' . // IPv4(dot-decimal): 001.22.3.44
  64                         '[a-z0-9.-]+' .                                                 // hostname(FQDN) : foo.example.org
  65                 ')' .
  66                 '(?::([0-9]*))?' .                                      // 4: Port
  67                 '((?:/+[^\s<>"\'\[\]/\#]+)*/+)?' .      // 5: Directory path or path-info
  68                 '([^\s<>"\'\[\]\#?]+)?' .                       // 6: File?
  69                 '(?:\?([^\s<>"\'\[\]\#]+))?' .          // 7: Query string
  70                 '(?:\#([a-z0-9._~%!$&\'()*+,;=:@-]*))?' .       // 8: Fragment
  71                 '#i',
  72                  $string, $array, PREG_SET_ORDER | PREG_OFFSET_CAPTURE
  73         );
  74
  75         // Shrink $array
  76         static $parts = array(
  77                 1 => 'scheme', 2 => 'userinfo', 3 => 'host', 4 => 'port',
  78                 5 => 'path', 6 => 'file', 7 => 'query', 8 => 'fragment'
  79         );
  80         $default = array('');
  81         foreach(array_keys($array) as $uri) {
  82                 $_uri = & $array[$uri];
  83                 array_rename_keys($_uri, $parts, TRUE, $default);
  84
  85                 $offset = $_uri['scheme'][1]; // Scheme's offset
  86                 foreach(array_keys($_uri) as $part) {
  87                         // Remove offsets for each part
  88                         $_uri[$part] = & $_uri[$part][0];
  89                 }
  90
  91                 if ($normalize) {
  92                         $_uri['scheme'] = scheme_normalize($_uri['scheme']);
  93                         if ($_uri['scheme'] === '') {
  94                                 unset($array[$uri]);
  95                                 continue;
  96                         }
  97                         $_uri['host']  = strtolower($_uri['host']);
  98                         $_uri['port']  = port_normalize($_uri['port'], $_uri['scheme'], FALSE);
  99                         $_uri['path']  = path_normalize($_uri['path']);
 100                         if ($preserve_rawuri) $_uri['rawuri'] = & $_uri[0];
 101
 102                         // DEBUG
 103                         //$_uri['uri'] = uri_array_implode($_uri);
 104                 } else {
 105                         $_uri['uri'] = & $_uri[0]; // Raw
 106                 }
 107                 unset($_uri[0]); // Matched string itself
 108                 if (! $preserve_chunk) {
 109                         unset(
 110                                 $_uri['scheme'],
 111                                 $_uri['userinfo'],
 112                                 $_uri['host'],
 113                                 $_uri['port'],
 114                                 $_uri['path'],
 115                                 $_uri['file'],
 116                                 $_uri['query'],
 117                                 $_uri['fragment']
 118                         );
 119                 }
 120
 121                 // Area offset for area_measure()
 122                 $_uri['area']['offset'] = $offset;
 123         }
 124
 125         return $array;
 126 }
 127
 128 // Destructive normalize of URI array
 129 // NOTE: Give me the uri_pickup() result with chunks
 130 function uri_array_normalize(& $pickups, $preserve = TRUE)
 131 {
 132         if (! is_array($pickups)) return $pickups;
 133
 134         foreach (array_keys($pickups) as $key) {
 135                 $_key = & $pickups[$key];
 136                 $_key['path']     = isset($_key['path']) ? strtolower($_key['path']) : '';
 137                 $_key['file']     = isset($_key['file']) ? file_normalize($_key['file']) : '';
 138                 $_key['query']    = isset($_key['query']) ? query_normalize(strtolower($_key['query']), TRUE) : '';
 139                 $_key['fragment'] = (isset($_key['fragment']) && $preserve) ?
 140                         strtolower($_key['fragment']) : ''; // Just ignore
 141         }
 142
 143         return $pickups;
 144 }
 145
 146 // An URI array => An URI (See uri_pickup())
 147 function uri_array_implode($uri = array())
 148 {
 149         if (empty($uri) || ! is_array($uri)) return NULL;
 150
 151         $tmp = array();
 152         if (isset($uri['scheme']) && $uri['scheme'] !== '') {
 153                 $tmp[] = & $uri['scheme'];
 154                 $tmp[] = '://';
 155         }
 156         if (isset($uri['userinfo']) && $uri['userinfo'] !== '') {
 157                 $tmp[] = & $uri['userinfo'];
 158                 $tmp[] = '@';
 159         }
 160         if (isset($uri['host']) && $uri['host'] !== '') {
 161                 $tmp[] = & $uri['host'];
 162         }
 163         if (isset($uri['port']) && $uri['port'] !== '') {
 164                 $tmp[] = ':';
 165                 $tmp[] = & $uri['port'];
 166         }
 167         if (isset($uri['path']) && $uri['path'] !== '') {
 168                 $tmp[] = & $uri['path'];
 169         }
 170         if (isset($uri['file']) && $uri['file'] !== '') {
 171                 $tmp[] = & $uri['file'];
 172         }
 173         if (isset($uri['query']) && $uri['query'] !== '') {
 174                 $tmp[] = '?';
 175                 $tmp[] = & $uri['query'];
 176         }
 177         if (isset($uri['fragment']) && $uri['fragment'] !== '') {
 178                 $tmp[] = '#';
 179                 $tmp[] = & $uri['fragment'];
 180         }
 181
 182         return implode('', $tmp);
 183 }
 184
 185 // $array['something'] => $array['wanted']
 186 function array_rename_keys(& $array, $keys = array('from' => 'to'), $force = FALSE, $default = '')
 187 {
 188         if (! is_array($array) || ! is_array($keys)) return FALSE;
 189
 190         // Nondestructive test
 191         if (! $force)
 192                 foreach(array_keys($keys) as $from)
 193                         if (! isset($array[$from]))
 194                                 return FALSE;
 195
 196         foreach($keys as $from => $to) {
 197                 if ($from === $to) continue;
 198                 if (! $force || isset($array[$from])) {
 199                         $array[$to] = & $array[$from];
 200                         unset($array[$from]);
 201                 } else  {
 202                         $array[$to] = $default;
 203                 }
 204         }
 205
 206         return TRUE;
 207 }
 208
 209 // ---------------------
 210 // Area pickup
 211
 212 // Pickup all of markup areas
 213 function area_pickup($string = '', $method = array())
 214 {
 215         $area = array();
 216         if (empty($method)) return $area;
 217
 218         // Anchor tag pair by preg_match and preg_match_all()
 219         // [OK] <a href></a>
 220         // [OK] <a href=  >Good site!</a>
 221         // [OK] <a href= "#" >test</a>
 222         // [OK] <a href="http://nasty.example.com">visit http://nasty.example.com/</a>
 223         // [OK] <a href=\'http://nasty.example.com/\' >discount foobar</a>
 224         // [NG] <a href="http://ng.example.com">visit http://ng.example.com _not_ended_
 225         $regex = '#<a\b[^>]*\bhref\b[^>]*>.*?</a\b[^>]*(>)#i';
 226         if (isset($method['area_anchor'])) {
 227                 $areas = array();
 228                 $count = isset($method['asap']) ?
 229                         preg_match($regex, $string) :
 230                         preg_match_all($regex, $string, $areas);
 231                 if (! empty($count)) $area['area_anchor'] = $count;
 232         }
 233         if (isset($method['uri_anchor'])) {
 234                 $areas = array();
 235                 preg_match_all($regex, $string, $areas, PREG_SET_ORDER | PREG_OFFSET_CAPTURE);
 236                 foreach(array_keys($areas) as $_area) {
 237                         $areas[$_area] =  array(
 238                                 $areas[$_area][0][1], // Area start (<a href>)
 239                                 $areas[$_area][1][1], // Area end   (</a>)
 240                         );
 241                 }
 242                 if (! empty($areas)) $area['uri_anchor'] = $areas;
 243         }
 244
 245         // phpBB's "BBCode" pair by preg_match and preg_match_all()
 246         // [OK] [url][/url]
 247         // [OK] [url]http://nasty.example.com/[/url]
 248         // [OK] [link]http://nasty.example.com/[/link]
 249         // [OK] [url=http://nasty.example.com]visit http://nasty.example.com/[/url]
 250         // [OK] [link http://nasty.example.com/]buy something[/link]
 251         $regex = '#\[(url|link)\b[^\]]*\].*?\[/\1\b[^\]]*(\])#i';
 252         if (isset($method['area_bbcode'])) {
 253                 $areas = array();
 254                 $count = isset($method['asap']) ?
 255                         preg_match($regex, $string) :
 256                         preg_match_all($regex, $string, $areas, PREG_SET_ORDER);
 257                 if (! empty($count)) $area['area_bbcode'] = $count;
 258         }
 259         if (isset($method['uri_bbcode'])) {
 260                 $areas = array();
 261                 preg_match_all($regex, $string, $areas, PREG_SET_ORDER | PREG_OFFSET_CAPTURE);
 262                 foreach(array_keys($areas) as $_area) {
 263                         $areas[$_area] = array(
 264                                 $areas[$_area][0][1], // Area start ([url])
 265                                 $areas[$_area][2][1], // Area end   ([/url])
 266                         );
 267                 }
 268                 if (! empty($areas)) $area['uri_bbcode'] = $areas;
 269         }
 270
 271         // Various Wiki syntax
 272         // [text_or_uri>text_or_uri]
 273         // [text_or_uri:text_or_uri]
 274         // [text_or_uri|text_or_uri]
 275         // [text_or_uri->text_or_uri]
 276         // [text_or_uri text_or_uri] // MediaWiki
 277         // MediaWiki: [http://nasty.example.com/ visit http://nasty.example.com/]
 278
 279         return $area;
 280 }
 281
 282 // If in doubt, it's a little doubtful
 283 // if (Area => inside <= Area) $brief += -1
 284 function area_measure($areas, & $array, $belief = -1, $a_key = 'area', $o_key = 'offset')
 285 {
 286         if (! is_array($areas) || ! is_array($array)) return;
 287
 288         $areas_keys = array_keys($areas);
 289         foreach(array_keys($array) as $u_index) {
 290                 $offset = isset($array[$u_index][$o_key]) ?
 291                         intval($array[$u_index][$o_key]) : 0;
 292                 foreach($areas_keys as $a_index) {
 293                         if (isset($array[$u_index][$a_key])) {
 294                                 $offset_s = intval($areas[$a_index][0]);
 295                                 $offset_e = intval($areas[$a_index][1]);
 296                                 // [Area => inside <= Area]
 297                                 if ($offset_s < $offset && $offset < $offset_e) {
 298                                         $array[$u_index][$a_key] += $belief;
 299                                 }
 300                         }
 301                 }
 302         }
 303 }
 304
 305 // ---------------------
 306 // Spam-uri pickup
 307
 308 // Domain exposure callback (See spam_uri_pickup_preprocess())
 309 // http://victim.example.org/?foo+site:nasty.example.com+bar
 310 // => http://nasty.example.com/?refer=victim.example.org
 311 // NOTE: 'refer=' is not so good for (at this time).
 312 // Consider about using IP address of the victim, try to avoid that.
 313 function _preg_replace_callback_domain_exposure($matches = array())
 314 {
 315         $result = '';
 316
 317         // Preserve the victim URI as a complicity or ...
 318         if (isset($matches[5])) {
 319                 $result =
 320                         $matches[1] . '://' .   // scheme
 321                         $matches[2] . '/' .             // victim.example.org
 322                         $matches[3];                    // The rest of all (before victim)
 323         }
 324
 325         // Flipped URI
 326         if (isset($matches[4])) {
 327                 $result =
 328                         $matches[1] . '://' .   // scheme
 329                         $matches[4] .                   // nasty.example.com
 330                         '/?refer=' . strtolower($matches[2]) .  // victim.example.org
 331                         ' ' . $result;
 332         }
 333
 334         return $result;
 335 }
 336
 337 // Preprocess: rawurldecode() and adding space(s) and something
 338 // to detect/count some URIs _if possible_
 339 // NOTE: It's maybe danger to var_dump(result). [e.g. 'javascript:']
 340 // [OK] http://victim.example.org/go?http%3A%2F%2Fnasty.example.org
 341 // [OK] http://victim.example.org/http://nasty.example.org
 342 // TODO: link.toolbot.com, urlx.org
 343 function spam_uri_pickup_preprocess($string = '')
 344 {
 345         if (! is_string($string)) return '';
 346
 347         $string = rawurldecode($string);
 348
 349         // Domain exposure (See _preg_replace_callback_domain_exposure())
 350         $string = preg_replace_callback(
 351                 array(
 352                         '#(http)://' .
 353                         '(' .
 354                                 // Something Google: http://www.google.com/supported_domains
 355                                 '(?:[a-z0-9.]+\.)?google\.[a-z]{2,3}(?:\.[a-z]{2})?' .
 356                                 '|' .
 357                                 // AltaVista
 358                                 '(?:[a-z0-9.]+\.)?altavista.com' .
 359
 360                         ')' .
 361                         '/' .
 362                         '([a-z0-9?=&.%_/\'\\\+-]+)' .                           // path/?query=foo+bar+
 363                         '\bsite:([a-z0-9.%_-]+\.[a-z0-9.%_-]+)' .       // site:nasty.example.com
 364                         //'()' .        // Preserve or remove?
 365                         '#i',
 366                 ),
 367                 '_preg_replace_callback_domain_exposure',
 368                 $string
 369         );
 370
 371         // URI exposure (uriuri => uri uri)
 372         $string = preg_replace(
 373                 array(
 374                         '#(?<! )(?:https?|ftp):/#i',
 375                 //      '#[a-z][a-z0-9.+-]{1,8}://#i',
 376                 //      '#[a-z][a-z0-9.+-]{1,8}://#i'
 377                 ),
 378                 ' $0',
 379                 $string
 380         );
 381
 382         return $string;
 383 }
 384
 385 // Main function of spam-uri pickup
 386 function spam_uri_pickup($string = '', $method = array())
 387 {
 388         if (! is_array($method) || empty($method)) {
 389                 $method = check_uri_spam_method();
 390         }
 391
 392         $string = spam_uri_pickup_preprocess($string);
 393
 394         $array  = uri_pickup($string);
 395
 396         // Area elevation of URIs, for '(especially external)link' intension
 397         if (! empty($array)) {
 398                 $_method = array();
 399                 if (isset($method['uri_anchor'])) $_method['uri_anchor'] = & $method['uri_anchor'];
 400                 if (isset($method['uri_bbcode'])) $_method['uri_bbcode'] = & $method['uri_bbcode'];
 401                 $areas = area_pickup($string, $_method, TRUE);
 402                 if (! empty($areas)) {
 403                         $area_shadow = array();
 404                         foreach (array_keys($array) as $key) {
 405                                 $area_shadow[$key] = & $array[$key]['area'];
 406                                 foreach (array_keys($_method) as $_key) {
 407                                         $area_shadow[$key][$_key] = 0;
 408                                 }
 409                         }
 410                         foreach (array_keys($_method) as $_key) {
 411                                 if (isset($areas[$_key])) {
 412                                         area_measure($areas[$_key], $area_shadow, 1, $_key);
 413                                 }
 414                         }
 415                 }
 416         }
 417
 418         // Remove 'offset's for area_measure()
 419         foreach(array_keys($array) as $key)
 420                 unset($array[$key]['area']['offset']);
 421
 422         return $array;
 423 }
 424
 425
 426 // ---------------------
 427 // Normalization
 428
 429 // Scheme normalization: Renaming the schemes
 430 // snntp://example.org =>  nntps://example.org
 431 // NOTE: Keep the static lists simple. See also port_normalize().
 432 function scheme_normalize($scheme = '', $considerd_harmfull = TRUE)
 433 {
 434         // Abbreviations considerable they don't have link intension
 435         static $abbrevs = array(
 436                 'ttp'   => 'http',
 437                 'ttps'  => 'https',
 438         );
 439
 440         // Alias => normalized
 441         static $aliases = array(
 442                 'pop'   => 'pop3',
 443                 'news'  => 'nntp',
 444                 'imap4' => 'imap',
 445                 'snntp' => 'nntps',
 446                 'snews' => 'nntps',
 447                 'spop3' => 'pop3s',
 448                 'pops'  => 'pop3s',
 449         );
 450
 451         $scheme = strtolower(trim($scheme));
 452         if (isset($abbrevs[$scheme])) {
 453                 if ($considerd_harmfull) {
 454                         $scheme = $abbrevs[$scheme];
 455                 } else {
 456                         $scheme = '';
 457                 }
 458         }
 459         if (isset($aliases[$scheme])) $scheme = $aliases[$scheme];
 460
 461         return $scheme;
 462 }
 463
 464 // Port normalization: Suppress the (redundant) default port
 465 // HTTP://example.org:80/ => http://example.org/
 466 // HTTP://example.org:8080/ => http://example.org:8080/
 467 // HTTPS://example.org:443/ => https://example.org/
 468 function port_normalize($port, $scheme, $scheme_normalize = TRUE)
 469 {
 470         // Schemes that users _maybe_ want to add protocol-handlers
 471         // to their web browsers. (and attackers _maybe_ want to use ...)
 472         // Reference: http://www.iana.org/assignments/port-numbers
 473         static $array = array(
 474                 // scheme => default port
 475                 'ftp'     =>    21,
 476                 'ssh'     =>    22,
 477                 'telnet'  =>    23,
 478                 'smtp'    =>    25,
 479                 'tftp'    =>    69,
 480                 'gopher'  =>    70,
 481                 'finger'  =>    79,
 482                 'http'    =>    80,
 483                 'pop3'    =>   110,
 484                 'sftp'    =>   115,
 485                 'nntp'    =>   119,
 486                 'imap'    =>   143,
 487                 'irc'     =>   194,
 488                 'wais'    =>   210,
 489                 'https'   =>   443,
 490                 'nntps'   =>   563,
 491                 'rsync'   =>   873,
 492                 'ftps'    =>   990,
 493                 'telnets' =>   992,
 494                 'imaps'   =>   993,
 495                 'ircs'    =>   994,
 496                 'pop3s'   =>   995,
 497                 'mysql'   =>  3306,
 498         );
 499
 500         $port = trim($port);
 501         if ($port === '') return $port;
 502
 503         if ($scheme_normalize) $scheme = scheme_normalize($scheme);
 504         if (isset($array[$scheme]) && $port == $array[$scheme])
 505                 $port = ''; // Ignore the defaults
 506
 507         return $port;
 508 }
 509
 510 // Path normalization
 511 // http://example.org => http://example.org/
 512 // http://example.org#hoge => http://example.org/#hoge
 513 // http://example.org/path/a/b/./c////./d => http://example.org/path/a/b/c/d
 514 // http://example.org/path/../../a/../back => http://example.org/back
 515 function path_normalize($path = '', $divider = '/', $addroot = TRUE)
 516 {
 517         if (! is_string($path) || $path == '')
 518                 return $addroot ? $divider : '';
 519
 520         $path = trim($path);
 521         $last = ($path[strlen($path) - 1] == $divider) ? $divider : '';
 522         $array = explode($divider, $path);
 523
 524         // Remove paddings
 525         foreach(array_keys($array) as $key) {
 526                 if ($array[$key] == '' || $array[$key] == '.')
 527                          unset($array[$key]);
 528         }
 529         // Back-track
 530         $tmp = array();
 531         foreach($array as $value) {
 532                 if ($value == '..') {
 533                         array_pop($tmp);
 534                 } else {
 535                         array_push($tmp, $value);
 536                 }
 537         }
 538         $array = & $tmp;
 539
 540         $path = $addroot ? $divider : '';
 541         if (! empty($array)) $path .= implode($divider, $array) . $last;
 542
 543         return $path;
 544 }
 545
 546 // DirectoryIndex normalize (Destructive and rough)
 547 function file_normalize($string = 'index.html.en')
 548 {
 549         static $array = array(
 550                 'index'                 => TRUE,        // Some system can omit the suffix
 551                 'index.htm'             => TRUE,
 552                 'index.html'    => TRUE,
 553                 'index.shtml'   => TRUE,
 554                 'index.jsp'             => TRUE,
 555                 'index.php'             => TRUE,
 556                 'index.php3'    => TRUE,
 557                 'index.php4'    => TRUE,
 558                 //'index.pl'    => TRUE,
 559                 //'index.py'    => TRUE,
 560                 //'index.rb'    => TRUE,
 561                 'index.cgi'             => TRUE,
 562                 'default.htm'   => TRUE,
 563                 'default.html'  => TRUE,
 564                 'default.asp'   => TRUE,
 565                 'default.aspx'  => TRUE,
 566         );
 567
 568         // Content-negothiation filter:
 569         // Roughly removing ISO 639 -like
 570         // 2-letter suffixes (See RFC3066)
 571         $matches = array();
 572         if (preg_match('/(.*)\.[a-z][a-z](?:-[a-z][a-z])?$/i', $string, $matches)) {
 573                 $_string = $matches[1];
 574         } else {
 575                 $_string = & $string;
 576         }
 577
 578         if (isset($array[strtolower($_string)])) {
 579                 return '';
 580         } else {
 581                 return $string;
 582         }
 583 }
 584
 585 // Sort query-strings if possible (Destructive and rough)
 586 // [OK] &&&&f=d&b&d&c&a=0dd  =>  a=0dd&b&c&d&f=d
 587 // [OK] nothing==&eg=dummy&eg=padding&eg=foobar  =>  eg=foobar
 588 function query_normalize($string = '', $equal = FALSE, $equal_cutempty = TRUE)
 589 {
 590         $array = explode('&', $string);
 591
 592         // Remove '&' paddings
 593         foreach(array_keys($array) as $key) {
 594                 if ($array[$key] == '') {
 595                          unset($array[$key]);
 596                 }
 597         }
 598
 599         // Consider '='-sepalated input and paddings
 600         if ($equal) {
 601                 $equals = $not_equals = array();
 602                 foreach ($array as $part) {
 603                         if (strpos($part, '=') === FALSE) {
 604                                  $not_equals[] = $part;
 605                         } else {
 606                                 list($key, $value) = explode('=', $part, 2);
 607                                 $value = ltrim($value, '=');
 608                                 if (! $equal_cutempty || $value != '') {
 609                                         $equals[$key] = $value;
 610                                 }
 611                         }
 612                 }
 613
 614                 $array = & $not_equals;
 615                 foreach ($equals as $key => $value) {
 616                         $array[] = $key . '=' . $value;
 617                 }
 618                 unset($equals);
 619         }
 620
 621         natsort($array);
 622         return implode('&', $array);
 623 }
 624
 625 // ---------------------
 626 // Part One : Checker
 627
 628 // Rough implementation of globbing
 629 //
 630 // USAGE: $regex = '/^' . generate_glob_regex('*.txt', '/') . '$/i';
 631 //
 632 function generate_glob_regex($string = '', $divider = '/')
 633 {
 634         static $from = array(
 635                          1 => '*',
 636                         11 => '?',
 637         //              22 => '[',      // Maybe cause regex compilation error (e.g. '[]')
 638         //              23 => ']',      //
 639                 );
 640         static $mid = array(
 641                          1 => '_AST_',
 642                         11 => '_QUE_',
 643         //              22 => '_RBR_',
 644         //              23 => '_LBR_',
 645                 );
 646         static $to = array(
 647                          1 => '.*',
 648                         11 => '.',
 649         //              22 => '[',
 650         //              23 => ']',
 651                 );
 652
 653         if (is_array($string)) {
 654                 // Recurse
 655                 return '(?:' .
 656                         implode('|',    // OR
 657                                 array_map('generate_glob_regex',
 658                                         $string,
 659                                         array_pad(array(), count($string), $divider)
 660                                 )
 661                         ) .
 662                 ')';
 663         } else {
 664                 $string = str_replace($from, $mid, $string); // Hide
 665                 $string = preg_quote($string, $divider);
 666                 $string = str_replace($mid, $to, $string);   // Unhide
 667                 return $string;
 668         }
 669 }
 670
 671 function get_blocklist($list = '')
 672 {
 673         static $regexs;
 674
 675         if (! isset($regexs)) {
 676                 $regexs = array();
 677                 if (file_exists(SPAM_INI_FILE)) {
 678                         $blocklist = array();
 679                         include(SPAM_INI_FILE);
 680                         //      $blocklist['badhost'] = array(
 681                         //              '*.blogspot.com',       // Blog services's subdomains (only)
 682                         //              'IANA-examples' => '#^(?:.*\.)?example\.(?:com|net|org)$#',
 683                         //      );
 684                         foreach(array('goodhost', 'badhost') as $_list) {
 685                                 if (! isset($blocklist[$list])) continue;
 686                                 foreach ($blocklist[$_list] as $key => $value) {
 687                                         if (is_array($value)) {
 688                                                 $regexs[$_list][$key] = array();
 689                                                 foreach($value as $_key => $_value) {
 690                                                         if (is_string($_key)) {
 691                                                                  $regexs[$_list][$key][$_key] = $_value; // A regex
 692                                                         } else {
 693                                                                  $regexs[$_list][$key][] =
 694                                                                         '/^(?:www\.)?' . generate_glob_regex($_value, '/') . '$/i';
 695                                                         }
 696                                                 }
 697                                         } else {
 698                                                 if (is_string($key)) {
 699                                                         $regexs[$_list][$key] = $value; // A regex
 700                                                 } else {
 701                                                         $regexs[$_list][$value] =
 702                                                                 '/^(?:www\.)?' . generate_glob_regex($value, '/') . '$/i';
 703                                                 }
 704                                         }
 705                                 }
 706                         }
 707                 }
 708         }
 709
 710         if ($list == '') {
 711                 return $regexs; // ALL
 712         } else if (isset($regexs[$list])) {
 713                 return $regexs[$list];
 714         } else {
 715                 return array();
 716         }
 717 }
 718
 719 function is_badhost($hosts = array(), $asap = TRUE, & $remains)
 720 {
 721         $result = array();
 722         if (! is_array($hosts)) $hosts = array($hosts);
 723         foreach(array_keys($hosts) as $key) {
 724                 if (! is_string($hosts[$key])) unset($hosts[$key]);
 725         }
 726         if (empty($hosts)) return $result;
 727
 728         foreach (get_blocklist('goodhost') as $regex) {
 729                 $hosts = preg_grep_invert($regex, $hosts);
 730         }
 731         if (empty($hosts)) return $result;
 732
 733         $tmp = array();
 734         foreach (get_blocklist('badhost') as $label => $regex) {
 735                 if (is_array($regex)) {
 736                         $result[$label] = array();
 737                         foreach($regex as $_label => $_regex) {
 738                                 $_group = preg_grep($_regex, $hosts);
 739                                 if ($_group) {
 740                                         $result[$label][$_label] = $_group;
 741                                         $hosts = array_diff($hosts, $_group);
 742                                         if ($asap) break;
 743                                 }
 744                         }
 745                         if (empty($result[$label])) unset($result[$label]);
 746                 } else {
 747                         $_group = preg_grep($regex, $hosts);
 748                         if ($_group) {
 749                                 $result[$label] = $_group;
 750                                 $hosts = array_diff($hosts, $result[$label]);
 751                                 if ($asap) break;
 752                         }
 753                 }
 754         }
 755
 756         $remains = $hosts;
 757
 758         return $result;
 759 }
 760
 761 // Default (enabled) methods and thresholds (for content insertion)
 762 function check_uri_spam_method($times = 1, $t_area = 0, $rule = TRUE)
 763 {
 764         $times  = intval($times);
 765         $t_area = intval($t_area);
 766
 767         $positive = array(
 768                 // Thresholds
 769                 'quantity'     =>  8 * $times,  // Allow N URIs
 770                 'non_uniqhost' =>  3 * $times,  // Allow N duped (and normalized) Hosts
 771                 //'non_uniquri'=>  3 * $times,  // Allow N duped (and normalized) URIs
 772
 773                 // Areas
 774                 'area_anchor'  => $t_area,      // Using <a href> HTML tag
 775                 'area_bbcode'  => $t_area,      // Using [url] or [link] BBCode
 776                 //'uri_anchor' => $t_area,      // URI inside <a href> HTML tag
 777                 //'uri_bbcode' => $t_area,      // URI inside [url] or [link] BBCode
 778         );
 779         if ($rule) {
 780                 $bool = array(
 781                         // Rules
 782                         //'asap'   => TRUE,     // Quit or return As Soon As Possible
 783                         'uniqhost' => TRUE,     // Show uniq host (at block notification mail)
 784                         'badhost'  => TRUE,     // Check badhost
 785                 );
 786         } else {
 787                 $bool = array();
 788         }
 789
 790         // Remove non-$positive values
 791         foreach (array_keys($positive) as $key) {
 792                 if ($positive[$key] < 0) unset($positive[$key]);
 793         }
 794
 795         return $positive + $bool;
 796 }
 797
 798 // Simple/fast spam check
 799 function check_uri_spam($target = '', $method = array())
 800 {
 801         if (! is_array($method) || empty($method)) {
 802                 $method = check_uri_spam_method();
 803         }
 804         $progress = array(
 805                 'sum' => array(
 806                         'quantity'    => 0,
 807                         'uniqhost'    => 0,
 808                         'non_uniqhost'=> 0,
 809                         'non_uniquri' => 0,
 810                         'badhost'     => 0,
 811                         'area_anchor' => 0,
 812                         'area_bbcode' => 0,
 813                         'uri_anchor'  => 0,
 814                         'uri_bbcode'  => 0,
 815                 ),
 816                 'is_spam' => array(),
 817                 'method'  => & $method,
 818                 'remains' => array(),
 819         );
 820         $sum     = & $progress['sum'];
 821         $is_spam = & $progress['is_spam'];
 822         $remains = & $progress['remains'];
 823         $asap    = isset($method['asap']);
 824
 825         // Recurse
 826         if (is_array($target)) {
 827                 foreach($target as $str) {
 828                         // Recurse
 829                         $_progress = check_uri_spam($str, $method);
 830                         $_sum      = & $_progress['sum'];
 831                         $_is_spam  = & $_progress['is_spam'];
 832                         $_remains  = & $_progress['remains'];
 833                         foreach (array_keys($_sum) as $key) {
 834                                 $sum[$key] += $_sum[$key];
 835                         }
 836                         foreach (array_keys($_is_spam) as $key) {
 837                                 if (is_array($_is_spam[$key])) {
 838                                         // Marge keys (badhost)
 839                                         foreach(array_keys($_is_spam[$key]) as $_key) {
 840                                                 if (! isset($is_spam[$key][$_key])) {
 841                                                         $is_spam[$key][$_key] =  $_is_spam[$key][$_key];
 842                                                 } else {
 843                                                         $is_spam[$key][$_key] += $_is_spam[$key][$_key];
 844                                                 }
 845                                         }
 846                                 } else {
 847                                         $is_spam[$key] = TRUE;
 848                                 }
 849                         }
 850                         foreach ($_remains as $key=>$value) {
 851                                 foreach ($value as $_key=>$_value) {
 852                                         if (is_int($_key)) {
 853                                                 $remains[$key][]      = $_value;
 854                                         } else {
 855                                                 $remains[$key][$_key] = $_value;
 856                                         }
 857                                 }
 858                         }
 859                         if ($asap && $is_spam) break;
 860                 }
 861                 return $progress;
 862         }
 863
 864         // Area: There's HTML anchor tag
 865         if ((! $asap || ! $is_spam) && isset($method['area_anchor'])) {
 866                 $key = 'area_anchor';
 867                 $_asap = isset($method['asap']) ? array('asap' => TRUE) : array();
 868                 $result = area_pickup($target, array($key => TRUE) + $_asap);
 869                 if ($result) {
 870                         $sum[$key] = $result[$key];
 871                         if (isset($method[$key]) && $sum[$key] > $method[$key]) {
 872                                 $is_spam[$key] = TRUE;
 873                         }
 874                 }
 875         }
 876
 877         // Area: There's 'BBCode' linking tag
 878         if ((! $asap || ! $is_spam) && isset($method['area_bbcode'])) {
 879                 $key = 'area_bbcode';
 880                 $_asap = isset($method['asap']) ? array('asap' => TRUE) : array();
 881                 $result = area_pickup($target, array($key => TRUE) + $_asap);
 882                 if ($result) {
 883                         $sum[$key] = $result[$key];
 884                         if (isset($method[$key]) && $sum[$key] > $method[$key]) {
 885                                 $is_spam[$key] = TRUE;
 886                         }
 887                 }
 888         }
 889
 890         // Return if ...
 891         if ($asap && $is_spam) return $progress;
 892
 893         // URI: Pickup
 894         $pickups = spam_uri_pickup($target, $method);
 895         //$remains['uri_pickup'] = & $pickups;
 896
 897         // Return if ...
 898         if (empty($pickups)) return $progress;
 899
 900         // URI: Check quantity
 901         $sum['quantity'] += count($pickups);
 902                 // URI quantity
 903         if ((! $asap || ! $is_spam) && isset($method['quantity']) &&
 904                 $sum['quantity'] > $method['quantity']) {
 905                 $is_spam['quantity'] = TRUE;
 906         }
 907
 908         // URI: used inside HTML anchor tag pair
 909         if ((! $asap || ! $is_spam) && isset($method['uri_anchor'])) {
 910                 $key = 'uri_anchor';
 911                 foreach($pickups as $pickup) {
 912                         if (isset($pickup['area'][$key])) {
 913                                 $sum[$key] += $pickup['area'][$key];
 914                                 if(isset($method[$key]) &&
 915                                         $sum[$key] > $method[$key]) {
 916                                         $is_spam[$key] = TRUE;
 917                                         if ($asap && $is_spam) break;
 918                                 }
 919                                 if ($asap && $is_spam) break;
 920                         }
 921                 }
 922         }
 923
 924         // URI: used inside 'BBCode' pair
 925         if ((! $asap || ! $is_spam) && isset($method['uri_bbcode'])) {
 926                 $key = 'uri_bbcode';
 927                 foreach($pickups as $pickup) {
 928                         if (isset($pickup['area'][$key])) {
 929                                 $sum[$key] += $pickup['area'][$key];
 930                                 if(isset($method[$key]) &&
 931                                         $sum[$key] > $method[$key]) {
 932                                         $is_spam[$key] = TRUE;
 933                                         if ($asap && $is_spam) break;
 934                                 }
 935                                 if ($asap && $is_spam) break;
 936                         }
 937                 }
 938         }
 939
 940         // URI: Uniqueness (and removing non-uniques)
 941         if ((! $asap || ! $is_spam) && isset($method['non_uniquri'])) {
 942
 943                 // Destructive normalize of URIs
 944                 uri_array_normalize($pickups);
 945
 946                 $uris = array();
 947                 foreach (array_keys($pickups) as $key) {
 948                         $uris[$key] = uri_array_implode($pickups[$key]);
 949                 }
 950                 $count = count($uris);
 951                 $uris  = array_unique($uris);
 952                 $sum['non_uniquri'] += $count - count($uris);
 953                 if ($sum['non_uniquri'] > $method['non_uniquri']) {
 954                         $is_spam['non_uniquri'] = TRUE;
 955                 }
 956                 if (! $asap || ! $is_spam) {
 957                         foreach (array_diff(array_keys($pickups),
 958                                 array_keys($uris)) as $remove) {
 959                                 unset($pickups[$remove]);
 960                         }
 961                 }
 962                 unset($uris);
 963         }
 964
 965         // Return if ...
 966         if ($asap && $is_spam) return $progress;
 967
 968         // Host: Uniqueness (uniq / non-uniq)
 969         $hosts = array();
 970         foreach ($pickups as $pickup) $hosts[] = & $pickup['host'];
 971         $hosts = array_unique($hosts);
 972         //$remains['uniqhost'] = & $hosts;
 973         $sum['uniqhost'] += count($hosts);
 974         if ((! $asap || ! $is_spam) && isset($method['non_uniqhost'])) {
 975                 $sum['non_uniqhost'] = $sum['quantity'] - $sum['uniqhost'];
 976                 if ($sum['non_uniqhost'] > $method['non_uniqhost']) {
 977                         $is_spam['non_uniqhost'] = TRUE;
 978                 }
 979         }
 980
 981         // Return if ...
 982         if ($asap && $is_spam) return $progress;
 983
 984         // URI: Bad host
 985         if ((! $asap || ! $is_spam) && isset($method['badhost'])) {
 986                 $__remains = array();
 987                 $badhost = is_badhost($hosts, $asap, $__remains);
 988                 if (! $asap) {
 989                         if ($__remains) {
 990                                 $remains['badhost'] = array();
 991                                 foreach ($__remains as $value) {
 992                                         $remains['badhost'][$value] = TRUE;
 993                                 }
 994                         }
 995                 }
 996                 unset($__remains);
 997                 if (! empty($badhost)) {
 998                         $sum['badhost'] += array_count_leaves($badhost);
 999                         foreach(array_keys($badhost) as $keys) {
1000                                 $is_spam['badhost'][$keys] =
1001                                         array_count_leaves($badhost[$keys]);
1002                         }
1003                         unset($badhost);
1004                 }
1005         }
1006
1007         return $progress;
1008 }
1009
1010 // Count leaves
1011 function array_count_leaves($array = array(), $count_empty_array = FALSE)
1012 {
1013         if (! is_array($array) || (empty($array) && $count_empty_array))
1014                 return 1;
1015
1016         // Recurse
1017         $result = 0;
1018         foreach ($array as $part) {
1019                 $result += array_count_leaves($part, $count_empty_array);
1020         }
1021         return $result;
1022 }
1023
1024 // ---------------------
1025 // Reporting
1026
1027 // TODO: Don't show unused $method!
1028 // Summarize $progress (blocked only)
1029 function summarize_spam_progress($progress = array(), $blockedonly = FALSE)
1030 {
1031         if ($blockedonly) {
1032                 $tmp = array_keys($progress['is_spam']);
1033         } else {
1034                 $tmp = array();
1035                 $method = & $progress['method'];
1036                 if (isset($progress['sum'])) {
1037                         foreach ($progress['sum'] as $key => $value) {
1038                                 if (isset($method[$key])) {
1039                                         $tmp[] = $key . '(' . $value . ')';
1040                                 }
1041                         }
1042                 }
1043         }
1044
1045         return implode(', ', $tmp);
1046 }
1047
1048 // ---------------------
1049 // Exit
1050
1051 // Common bahavior for blocking
1052 // NOTE: Call this function from various blocking feature, to disgueise the reason 'why blocked'
1053 function spam_exit($mode = '', $data = array())
1054 {
1055         switch ($mode) {
1056                 case '':        echo("\n");     break;
1057                 case 'dump':
1058                         echo('<pre>' . "\n");
1059                         echo htmlspecialchars(var_export($data, TRUE));
1060                         echo('</pre>' . "\n");
1061                         break;
1062         };
1063
1064         // Force exit
1065         exit;
1066 }
1067
1068
1069 // ---------------------
1070 // Simple filtering
1071
1072 // TODO: Record them
1073 // Simple/fast spam filter ($target: 'a string' or an array())
1074 function pkwk_spamfilter($action, $page, $target = array('title' => ''), $method = array(), $exitmode = '')
1075 {
1076         $progress = check_uri_spam($target, $method);
1077
1078         if (! empty($progress['is_spam'])) {
1079                 // Mail to administrator(s)
1080                 pkwk_spamnotify($action, $page, $target, $progress, $method);
1081
1082                 // Exit
1083                 spam_exit($exitmode, $progress);
1084         }
1085 }
1086
1087 // ---------------------
1088 // PukiWiki original
1089
1090 // Mail to administrator(s)
1091 function pkwk_spamnotify($action, $page, $target = array('title' => ''), $progress = array(), $method = array())
1092 {
1093         global $notify, $notify_subject;
1094
1095         if (! $notify) return;
1096
1097         $asap = isset($method['asap']);
1098
1099         $summary['ACTION']  = 'Blocked by: ' . summarize_spam_progress($progress, TRUE);
1100         if (! $asap) {
1101                 $summary['METRICS'] = summarize_spam_progress($progress);
1102         }
1103         if (isset($progress['is_spam']['badhost'])) {
1104                 $badhost = array();
1105                 foreach($progress['is_spam']['badhost'] as $glob=>$number) {
1106                         $badhost[] = $glob . '(' . $number . ')';
1107                 }
1108                 $summary['DETAIL_BADHOST'] = implode(', ', $badhost);
1109         }
1110         if (! $asap && $progress['remains']['badhost']) {
1111                 $count = count($progress['remains']['badhost']);
1112                 $summary['DETAIL_NEUTRAL_HOST'] = $count .
1113                         ' (' .
1114                                 preg_replace(
1115                                         '/[^, a-z0-9.-]/i', '',
1116                                         implode(', ', array_keys($progress['remains']['badhost']))
1117                                 ) .
1118                         ')';
1119         }
1120         $summary['COMMENT'] = $action;
1121         $summary['PAGE']    = '[blocked] ' . (is_pagename($page) ? $page : '');
1122         $summary['URI']     = get_script_uri() . '?' . rawurlencode($page);
1123         $summary['USER_AGENT']  = TRUE;
1124         $summary['REMOTE_ADDR'] = TRUE;
1125         pkwk_mail_notify($notify_subject,  var_export($target, TRUE), $summary, TRUE);
1126 }
1127
1128 ?>