lifetype-1.2/class/data/validator/rules/urlformatrule.class.php

00001 <?php
00002 
00003 lt_include(PLOG_CLASS_PATH."class/data/validator/rules/rule.class.php");
00004 lt_include(PLOG_CLASS_PATH."class/net/dns.class.php");
00005 lt_include(PLOG_CLASS_PATH."class/net/http/httpvars.class.php");
00006 
00007 define( "EW_OK", 1 );
00008 
00009 define( "EW_ERR_URL_INVALID_PROTOCOL", 10 );
00010 define( "EW_ERR_URL_INVALID_USER", 20 );
00011 define( "EW_ERR_URL_INVALID_PASSWORD", 30 );
00012 define( "EW_ERR_URL_INVALID_SERVER", 40 );
00013 define( "EW_ERR_URL_INVALID_TLD", 50 );
00014 define( "EW_ERR_URL_INVALID_PORT", 60 );
00015 define( "EW_ERR_URL_INVALID_RESOURCE", 70 );
00016 define( "EW_ERR_URL_INVALID_QUERYSTRING", 80 );
00017 define( "EW_ERR_URL_INVALID_ANCHOR", 100 );
00018 
00019 define( "EW_ERR_URL_MISSING_PROTOCOL", 110 );
00020 define( "EW_ERR_URL_MISSING_USER", 120 );
00021 define( "EW_ERR_URL_MISSING_PASSWORD", 130 );
00022 define( "EW_ERR_URL_MISSING_SERVER", 140 );
00023 define( "EW_ERR_URL_MISSING_TLD", 150 );
00024 define( "EW_ERR_URL_MISSING_PORT", 160 );
00025 define( "EW_ERR_URL_MISSING_RESOURCE", 170 );
00026 define( "EW_ERR_URL_MISSING_QUERYSTRING", 180 );
00027 define( "EW_ERR_URL_MISSING_ANCHOR", 190 );
00028 
00029 
00030 
00045 class UrlFormatRule extends Rule
00046 {
00047     var $_options;
00048 
00049     function UrlFormatRule( $options = Array())
00050         {
00051         $this->Rule();
00052         $this->_options = $options;
00053     }
00054                 
00055     function validate( $value )
00056         {
00057         $result = $this->_ValURL( $value, $this->_options );
00058         return( $result['Result'] == EW_OK );
00059     }
00060 
00061     function _ValURL($value, $options = array())
00062     {
00063         $value = trim($value);
00064 
00065         if (!$value)
00066             return array('Result' => array(EW_ERR_URL_EMPTY_STRING), 'Value' => '');
00067 
00068         if(!isset($options['Require']))
00069             $options['Require'] = array();
00070         if(!isset($options['Forbid']))
00071             $options['Forbid'] = array();
00072         
00073             /* Set up default options */
00074         $options = array_merge(array(
00075                                        /* array('http', 'https', etc.) always lcase! */
00076                                    'AllowedProtocols' =>
00077                                    array('http', 'https', 'ftp', 'mailto',
00078                                          'file', 'news', 'gopher', 'telnet',
00079                                          'nntp'),
00080                                    'AllowBracks' => false, /* Allow square brackets in the query string ? */
00081                                    'Protocols' => array('http', 'https', 'ftp', 'mailto', 'file', 'news', 'gopher', 'telnet', 'nntp'), 
00082                                    'AssumeProtocol' => false, 
00083                                    ), $options);
00084 
00085             /* Setup default values for $options['Require]*/
00086         $options['Require'] = array_merge(array(
00087                                                 'Protocol' => true, 
00088                                                 'User' => false, 
00089                                                 'Password' => false, 
00090                                                 'Server' => true, 
00091                                                 'TLD' => false, 
00092                                                 'Port' => false, 
00093                                                 'Resource' => false, 
00094                                                 'QueryString' => false, 
00095                                                 'Anchor' => false, 
00096                                                 ), $options['Require']);
00097 
00098             /* Setup default values for $options['Forbid]*/
00099         $options['Forbid'] = array_merge(array(
00100                                                'Protocol' => false, 
00101                                                'User' => false, 
00102                                                'Password' => false, 
00103                                                'Server' => false, 
00104                                                'TLD' => false, 
00105                                                'Port' => false, 
00106                                                'Resource' => false, 
00107                                                'QueryString' => false, 
00108                                                'Anchor' => false, 
00109                                                ), $options['Forbid']);
00110 
00111             /* Create a container for the URL parts*/
00112         $url = array(
00113             'Protocol' => '', 
00114             'User' => '', 
00115             'Password' => '', 
00116             'Server' => '', 
00117             'Port' => '', 
00118             'Resource' => '', 
00119             'TLD' => '', 
00120             'QueryString' => '', 
00121             'Anchor' => '');
00122 
00123             /* Setup errorcodes for invalid elements */
00124         $errCodeInvalid = array(
00125             'Protocol' => EW_ERR_URL_INVALID_PROTOCOL, 
00126             'User' => EW_ERR_URL_INVALID_USER, 
00127             'Password' => EW_ERR_URL_INVALID_PASSWORD, 
00128             'Server' => EW_ERR_URL_INVALID_SERVER, 
00129             'TLD' => EW_ERR_URL_INVALID_TLD, 
00130             'Port' => EW_ERR_URL_INVALID_PORT, 
00131             'Resource' => EW_ERR_URL_INVALID_RESOURCE, 
00132             'QueryString' => EW_ERR_URL_INVALID_QUERYSTRING, 
00133             'Anchor' => EW_ERR_URL_INVALID_ANCHOR);
00134 
00135             /* Setup errorcodes for missing elements */
00136         $errCodeMissing = array(
00137             'Protocol' => EW_ERR_URL_MISSING_PROTOCOL, 
00138             'User' => EW_ERR_URL_MISSING_USER, 
00139             'Password' => EW_ERR_URL_MISSING_PASSWORD, 
00140             'Server' => EW_ERR_URL_MISSING_SERVER, 
00141             'TLD' => EW_ERR_URL_MISSING_TLD, 
00142             'Port' => EW_ERR_URL_MISSING_PORT, 
00143             'Resource' => EW_ERR_URL_MISSING_RESOURCE, 
00144             'QueryString' => EW_ERR_URL_MISSING_QUERYSTRING, 
00145             'Anchor' => EW_ERR_URL_MISSING_ANCHOR);
00146 
00147             /* set up some needed vars */
00148         extract($options);
00149         $errArr = array();
00150         $tmpValue = $value;
00151         $lcValue = strtolower($value);
00152 
00157         foreach ($Protocols as $key => $protocol)
00158         {
00159             if (strpos($lcValue, "$protocol:") === 0)
00160             {
00161                 $tmp = explode(':', $tmpValue, 2);
00162                 $url['Protocol'] = $tmp[0];
00163                 $tmpValue = $tmp[1];
00164 
00165                 if ($url['Protocol'] == 'mailto' || $url['Protocol'] == 'news')
00166                 {
00167 
00168                         /* Check for % that is NOT an escape sequence */
00169                     if (preg_match('/%[^a-f0-9]/i', $tmpValue) || preg_match("/^[^a-z0-9;&=+$,_.!*'()%~-]/i", $tmpValue))
00170                     {
00171                         $errArr[EW_ERR_URL_INVALID_PROTOCOL] = EW_ERR_URL_INVALID_PROTOCOL;
00172                     }
00173                 }
00174                 else
00175                 {
00176                     if (!(strpos($tmpValue, '//') === 0))
00177                     {
00178                         $errArr[EW_ERR_URL_INVALID_PROTOCOL] = EW_ERR_URL_INVALID_PROTOCOL;
00179                     }
00180                     else
00181                     {
00182                         $tmpValue = substr($tmpValue, 2);
00183                     }
00184                 }
00185             }
00186         }
00187 
00188         if (!$url['Protocol'])
00189         {
00190             if (strpos(strtolower($tmpValue), ('mailto:')) === 0 || strpos(strtolower($tmpValue), ('news:')) === 0)
00191                 $tmp = ':';
00192             else
00193                 $tmp = '://';
00194 
00195             $tmp = explode($tmp, $tmpValue, 2);
00196             if (count($tmp) == 2)
00197             {
00198                 $url['Protocol'] = strtolower($tmp[0]);
00199                 $tmpValue = $tmp[1];
00200             }
00201         }
00202 
00203         $tmp = explode('?', $tmpValue);
00204 
00205         if (count($tmp) > 1)
00206         {
00207             $tmpValue = $tmp[0];
00208             $url['QueryString'] = $tmp[1];
00209 
00210             $tmp = explode('#', $url['QueryString']);
00211             if (count($tmp) > 1)
00212             {
00213                 $url['QueryString'] = $tmp[0];
00214                 $url['Anchor'] = $tmp[1];
00215             }
00216         }
00217         else
00218         {
00219             $tmp = explode('#', $tmpValue);
00220             if (count($tmp) > 1)
00221             {
00222                 $tmpValue = $tmp[0];
00223                 $url['Anchor'] = $tmp[1];
00224             }
00225         }
00226 
00227         $tmp = explode('/', $tmpValue, 2);
00228         if (count($tmp) > 1)
00229         {
00230             $url['Server'] = strtolower($tmp[0]);
00231             $url['Resource'] = $tmp[1];
00232         }
00233         else
00234         {
00235             $url['Server'] = strtolower($tmpValue);
00236         }
00237 
00238             /* User / password */
00239         $tmp = explode('@', $url['Server']);
00240         if (count($tmp) > 1)
00241         {
00242             $url['User'] = $tmp[0];
00243             $url['Server'] = $tmp[1];
00244 
00245             if ($url['User'])
00246             {
00247                 $tmp = explode(':', $url['User']);
00248                 if (count($tmp) > 1)
00249                 {
00250                     $url['User'] = $tmp[0];
00251                     $url['Password'] = $tmp[1];
00252                 }
00253             }
00254         }
00255 
00256         $tmp = explode(':', $url['Server'], 2);
00257         if (count($tmp) > 1)
00258         {
00259             if ($tmp[0])
00260             {
00261                 $url['Server'] = $tmp[0];
00262                 $url['Port'] = $tmp[1];
00263 
00264             }
00265         }
00266 
00267         if (!$url['Protocol'] && !$url['Password'] && in_array(strtolower($url['User']), array('mail', 'news')))
00268         {
00269             $url['Protocol'] = strtolower($url['User']);
00270             $url['User'] = '';
00271 
00272         }
00273 
00274         if ($url['Protocol'] == 'mailto' && $url['Server'] && !$url['User'])
00275         {
00276             $url['User'] = $url['Server'];
00277             $url['Server'] = '';
00278         }
00279 
00284             /* Check the protocol */
00285         if ($url['Protocol'])
00286         {
00287             $tmp = preg_replace("/[^a-z0-9+-.]/", '', $url['Protocol']);
00288 
00289             if ($tmp != $url['Protocol'])
00290             {
00291                 $errArr[EW_ERR_URL_INVALID_PROTOCOL] = EW_ERR_URL_INVALID_PROTOCOL;
00292             }
00293 
00294             if (count($options['AllowedProtocols']))
00295                 if (!in_array($url['Protocol'], $options['AllowedProtocols']))
00296                     $errArr[EW_ERR_URL_INVALID_PROTOCOL] = EW_ERR_URL_INVALID_PROTOCOL;
00297 
00298         }
00299 
00300             /* check userinfo */
00301         if ($url['User'])
00302         {
00303                 /* Check for % that is NOT an escape sequence */
00304             if (preg_match('/%[^a-f0-9]/i', $url['User']) || preg_match("/[^a-z0-9;&=+$,_.!~*'()%-]/i", $url['User']))
00305             {
00306                 $errArr[EW_ERR_URL_INVALID_USER] = EW_ERR_URL_INVALID_USER;
00307                 $url['User'] = urlencode(urldecode($url['User']));
00308             }
00309         }
00310         if ($url['Password'])
00311         {
00312                 /* Check for % that is NOT an escape sequence */
00313             if (preg_match('/%[^a-f0-9]/i', $url['Password']) || preg_match("/[^a-z0-9;&=+$,_.!~*'()%-]/i", $url['Password']))
00314             {
00315                 $errArr[EW_ERR_URL_INVALID_PASSWORD] = EW_ERR_URL_INVALID_PASSWORD;
00316             }
00317             $url['Password'] = urlencode(urldecode($url['Password']));
00318         }
00319 
00320             //      userinfo      = *( unreserved | escaped |
00321             //                         ";" | ":" | "&" | "=" | "+" | "$" | "," )
00322             //      unreserved    = alphanum | mark
00323             //      mark          = "-" | "_" | "." | "!" | "~" | "*" | "'" |
00324             //                      "(" | ")"
00325 
00326             //      escaped       = "%" hex hex
00327             /* Check if the server part is an ip */
00328         if ($url['Server'])
00329         {
00330             if (!preg_match('/[^.0-9]/', $url['Server']))
00331             {
00332                 $ServerIsIP = true;
00333 
00334                 $ipErr = false;
00335 
00336                 $ipPart = explode('.', $url['Server']);
00337 
00338                 if ($ipPart[0] > 224 || $ipPart[0] == 0)
00339                 {
00340                     $errArr[EW_ERR_URL_INVALID_SERVER] = EW_ERR_URL_INVALID_SERVER;
00341                 }
00342                 else
00343                 {
00344                     for ($i = 1; $i < 4; $i ++)
00345                     {
00346                         $ipPart[$i] = (integer) $ipPart[$i];
00347                         if ($ipPart[$i] > 255)
00348                             $errArr[EW_ERR_URL_INVALID_SERVER] = EW_ERR_URL_INVALID_SERVER;
00349                     }
00350                 }
00351 
00361                 $url['Server'] = join('.', $ipPart);
00362             }
00363                 /* url is not an ip */
00364             else
00365             {
00366                 $ServerIsIP = false;
00367 
00368                 $serverParts = explode('.', $url['Server']);
00369 
00370                     /* check serverparts */
00371                 for ($i = 0; $i < count($serverParts); $i ++)
00372                 {
00373                     $tmp = preg_replace('/[^a-z0-9-]/', '', $serverParts[$i]);
00374                     
00375                         /* Check if it is a top-level server */
00376                     if ($i  && $i == count($serverParts) - 1)
00377                         $tmp = preg_replace('/^[^a-z]/', '', $tmp);
00378                     else
00379                         $tmp = preg_replace('/^[^a-z0-9]/', '', $serverParts[$i]);
00380                     
00381                     $tmp = preg_replace('/[^a-z0-9]$/', '', $tmp);
00382 
00383                     if ($serverParts[$i] == '' || $tmp != $serverParts[$i])
00384                     {
00385                         if ($tmp != '')
00386                             $serverParts[$i] = $tmp;
00387                         else
00388                             unset($serverParts[$i]);
00389 
00390                         $errArr[EW_ERR_URL_INVALID_SERVER] = EW_ERR_URL_INVALID_SERVER;
00391 
00392                     }
00393                 }
00394 
00395                 if (count($serverParts) < 2)
00396                 {
00397                     if ($Require['TLD'])
00398                     {
00399                         $errArr[EW_ERR_URL_MISSING_TLD] = EW_ERR_URL_MISSING_TLD;
00400                     }
00401                 }
00402                 else
00403                 {
00404 
00405                     $url['TLD'] = $serverParts[count($serverParts) - 1];
00406                 }
00407 
00408                 $url['Server'] = join('.', $serverParts);
00409             }
00410         }
00411 
00412             /* Check the Port */
00413         if ($url['Port'])
00414         {
00415             $tmp = (integer) $url['Port'];
00416             if ($url['Port'] != (string) $tmp)
00417             {
00418                 $errArr[EW_ERR_URL_INVALID_PORT] = EW_ERR_URL_INVALID_PORT;
00419 
00420                 $url['Port'] = '';
00421             }
00422             else
00423             {
00424                 $url['Port'] = $tmp;
00425                 if ($url['Port'] > 65535)
00426                     $errArr[EW_ERR_URL_INVALID_PORT] = EW_ERR_URL_INVALID_PORT;
00427             }
00428 
00429         }
00430 
00431             /* Check the resource */
00432             //path          = [ abs_path | opaque_part ]
00433             //path_segments = segment *( "/" segment )
00434             //segment       = *pchar *( ";" param )
00435             //param         = *pchar
00436             //pchar         = unreserved | escaped |
00437             //                ":" | "@" | "&" | "=" | "+" | "$" | ","
00438 
00439         if ($url['Resource'])
00440         {
00441             $resourceParts = explode('/', $url['Resource']);
00442 
00443             if ($resourceParts[count($resourceParts) - 1] == '')
00444                 array_pop($resourceParts);
00445 
00446             if ($resourceParts[0] == '')
00447                 unset($resourceParts[0]);
00448 
00449             foreach ($resourceParts as $key => $part)
00450             {
00451                 if ($part == '')
00452                 {
00453                     $errArr[EW_ERR_URL_INVALID_RESOURCE] = EW_ERR_URL_INVALID_RESOURCE;
00454                     unset($resourceParts[$key]);
00455                 }
00456 
00457                     /* Check for % that is NOT an escape sequence || invalid chars*/
00458                 elseif (preg_match('/%[^a-f0-9]/i', $part) || preg_match("/[^@a-z0-9_.!~*'()$+&,%:=;?-]/i", $part))
00459                 {
00460                     $errArr[EW_ERR_URL_INVALID_RESOURCE] = EW_ERR_URL_INVALID_RESOURCE;
00461                     $resourceParts[$key] = urlencode(urldecode($part));
00462                 }
00463 
00464                     /* check for invalid chars */
00465 
00466             }
00467             $url['Resource'] = join('/', $resourceParts);
00468         }
00469 
00470         if ($url['QueryString'])
00471         {
00472 
00473                 /* Check for % NOT part of an escape sequence || invalid chars */
00474             $tmp = $options['AllowBracks'] ? 
00475                 "^a-z0-9_.!~*'()%;\/?:@&=+$,\[\]-" : 
00476                 "^a-z0-9_.!~*'()%;\/?:@&=+$,-"; 
00477 
00478             if (preg_match('/%[^a-f0-9]/i', $url['QueryString']) || preg_match("/[$tmp]+/i", $url['QueryString']))
00479             {
00480                 $errArr[EW_ERR_URL_INVALID_QUERYSTRING] = EW_ERR_URL_INVALID_QUERYSTRING;
00481                 $url['QueryString'] = $url['QueryString'];
00482             }
00483 
00484         }
00485         if ($url['Anchor'])
00486         {
00487             if (preg_match('/%[^a-f0-9][a-f0-9]?/i', $url['Anchor']) || //
00488                 preg_match("/[^a-z0-9-_.!~*'()%;\/?:@&=+$,]/i", $url['Anchor']))
00489             {
00490                 $errArr[EW_ERR_URL_INVALID_ANCHOR] = EW_ERR_URL_INVALID_ANCHOR;
00491                 $url['Anchor'] = $url['Anchor'];
00492             }
00493 
00494         }
00495         foreach ($url as $partName => $notused)
00496         {
00497             if ($partName == 'TLD' && $ServerIsIP)
00498                 continue;
00499 
00500             if ($Require[$partName] && !$url[$partName])
00501                 $errArr[$errCodeMissing[$partName]] = $errCodeMissing[$partName];
00502 
00503             if ($Forbid[$partName] && $url[$partName])
00504                 $errArr[$errCodeMissing[$partName]] = $errCodeInvalid[$partName];
00505         }
00506 
00507             /* Construct an estimate of what the value should've been */
00508         if ($options['AssumeProtocol'] && !$url['Protocol'] && ($url['Server'] || (!$url['Server'] && !$url['Resource'])))
00509             $url['Protocol'] = $options['AssumeProtocol'];
00510 
00511         $value = $url['Protocol'];
00512 
00513         if ($url['Protocol'])
00514         {
00515             if ($url['Protocol'] == 'mailto' | $url['Protocol'] == 'mailto')
00516                 $value.= ':';
00517             else
00518                 $value.= '://';
00519         }
00520 
00521         if ($url['User'])
00522         {
00523             if ($url['Password'])
00524                 $value.= "{$url['User']}:{$url['Password']}";
00525             else
00526                 $value.= "{$url['User']}";
00527 
00528             if ($url['Server'])
00529                 $value.= '@';
00530         }
00531 
00532         $value.= $url['Server'];
00533 
00534         if ($url['Port'])
00535             $value.= ":{$url['Port']}";
00536 
00537         if ($url['Server'] && $url['Resource'])
00538             $value.= "/";
00539 
00540         $value.= $url['Resource'];
00541 
00542         if ($url['QueryString'])
00543             $value.= "?{$url['QueryString']}";
00544 
00545         if ($url['Anchor'])
00546             $value.= "#{$url['Anchor']}";
00547 
00548         $r = array('Result' => count($errArr) ? $errArr : EW_OK, 'Value' => $value, 'URLParts' => $url);
00549 
00550         return $r;
00551 
00552     }
00553 
00554 }
00555 ?>