@@ -1,13 +1,13 @@ |
||
| 1 | 1 | <?php |
| 2 | 2 | /** |
| 3 | - * TAttributeCollection classes |
|
| 4 | - * |
|
| 5 | - * @author Qiang Xue <[email protected]> |
|
| 6 | - * @link https://github.com/pradosoft/prado |
|
| 7 | - * @copyright Copyright © 2005-2015 The PRADO Group |
|
| 8 | - * @license https://github.com/pradosoft/prado/blob/master/COPYRIGHT |
|
| 9 | - * @package System.Collections |
|
| 10 | - */ |
|
| 3 | + * TAttributeCollection classes |
|
| 4 | + * |
|
| 5 | + * @author Qiang Xue <[email protected]> |
|
| 6 | + * @link https://github.com/pradosoft/prado |
|
| 7 | + * @copyright Copyright © 2005-2015 The PRADO Group |
|
| 8 | + * @license https://github.com/pradosoft/prado/blob/master/COPYRIGHT |
|
| 9 | + * @package System.Collections |
|
| 10 | + */ |
|
| 11 | 11 | |
| 12 | 12 | /** |
| 13 | 13 | * Includes TMap class |
@@ -1,13 +1,13 @@ |
||
| 1 | 1 | <?php |
| 2 | 2 | /** |
| 3 | - * TList, TListIterator classes |
|
| 4 | - * |
|
| 5 | - * @author Qiang Xue <[email protected]> |
|
| 6 | - * @link https://github.com/pradosoft/prado |
|
| 7 | - * @copyright Copyright © 2005-2015 The PRADO Group |
|
| 8 | - * @license https://github.com/pradosoft/prado/blob/master/COPYRIGHT |
|
| 9 | - * @package System.Collections |
|
| 10 | - */ |
|
| 3 | + * TList, TListIterator classes |
|
| 4 | + * |
|
| 5 | + * @author Qiang Xue <[email protected]> |
|
| 6 | + * @link https://github.com/pradosoft/prado |
|
| 7 | + * @copyright Copyright © 2005-2015 The PRADO Group |
|
| 8 | + * @license https://github.com/pradosoft/prado/blob/master/COPYRIGHT |
|
| 9 | + * @package System.Collections |
|
| 10 | + */ |
|
| 11 | 11 | |
| 12 | 12 | /** |
| 13 | 13 | * TList class |
@@ -1,19 +1,19 @@ |
||
| 1 | 1 | <?php |
| 2 | 2 | /** |
| 3 | - * Wsdl file. |
|
| 4 | - * |
|
| 5 | - * This program is free software; you can redistribute it and/or modify |
|
| 6 | - * it under the terms of the BSD License. |
|
| 7 | - * |
|
| 8 | - * Copyright(c) 2005 by Marcus Nyeholt. All rights reserved. |
|
| 9 | - * |
|
| 10 | - * To contact the author write to {@link mailto:[email protected] Marcus Nyeholt} |
|
| 11 | - * This file is part of the PRADO framework from {@link http://www.xisc.com} |
|
| 12 | - * |
|
| 13 | - * @author Marcus Nyeholt <[email protected]> |
|
| 14 | - * @author Wei Zhuo <weizhuo[at]gmail[dot]com> |
|
| 15 | - * @package System.Web.Services.SOAP |
|
| 16 | - */ |
|
| 3 | + * Wsdl file. |
|
| 4 | + * |
|
| 5 | + * This program is free software; you can redistribute it and/or modify |
|
| 6 | + * it under the terms of the BSD License. |
|
| 7 | + * |
|
| 8 | + * Copyright(c) 2005 by Marcus Nyeholt. All rights reserved. |
|
| 9 | + * |
|
| 10 | + * To contact the author write to {@link mailto:[email protected] Marcus Nyeholt} |
|
| 11 | + * This file is part of the PRADO framework from {@link http://www.xisc.com} |
|
| 12 | + * |
|
| 13 | + * @author Marcus Nyeholt <[email protected]> |
|
| 14 | + * @author Wei Zhuo <weizhuo[at]gmail[dot]com> |
|
| 15 | + * @package System.Web.Services.SOAP |
|
| 16 | + */ |
|
| 17 | 17 | |
| 18 | 18 | /** |
| 19 | 19 | * Contains the dom object used to build up the wsdl. The |
@@ -1,19 +1,19 @@ |
||
| 1 | 1 | <?php |
| 2 | 2 | /** |
| 3 | - * WsdlMessage file. |
|
| 4 | - * |
|
| 5 | - * This program is free software; you can redistribute it and/or modify |
|
| 6 | - * it under the terms of the BSD License. |
|
| 7 | - * |
|
| 8 | - * Copyright(c) 2005 by Marcus Nyeholt. All rights reserved. |
|
| 9 | - * |
|
| 10 | - * To contact the author write to {@link mailto:[email protected] Marcus Nyeholt} |
|
| 11 | - * This file is part of the PRADO framework from {@link http://www.xisc.com} |
|
| 12 | - * |
|
| 13 | - * @author Marcus Nyeholt <[email protected]> |
|
| 14 | - * @author Wei Zhuo <weizhuo[at]gmail[dot]com> |
|
| 15 | - * @package System.Web.Services.SOAP |
|
| 16 | - */ |
|
| 3 | + * WsdlMessage file. |
|
| 4 | + * |
|
| 5 | + * This program is free software; you can redistribute it and/or modify |
|
| 6 | + * it under the terms of the BSD License. |
|
| 7 | + * |
|
| 8 | + * Copyright(c) 2005 by Marcus Nyeholt. All rights reserved. |
|
| 9 | + * |
|
| 10 | + * To contact the author write to {@link mailto:[email protected] Marcus Nyeholt} |
|
| 11 | + * This file is part of the PRADO framework from {@link http://www.xisc.com} |
|
| 12 | + * |
|
| 13 | + * @author Marcus Nyeholt <[email protected]> |
|
| 14 | + * @author Wei Zhuo <weizhuo[at]gmail[dot]com> |
|
| 15 | + * @package System.Web.Services.SOAP |
|
| 16 | + */ |
|
| 17 | 17 | |
| 18 | 18 | /** |
| 19 | 19 | * Represents a WSDL message. This is bound to the portTypes |
@@ -73,8 +73,8 @@ discard block |
||
| 73 | 73 | * @return void |
| 74 | 74 | */ |
| 75 | 75 | public static function setEnabled($Enabled) { |
| 76 | - $instance = FirePHP::getInstance(true); |
|
| 77 | - $instance->setEnabled($Enabled); |
|
| 76 | + $instance = FirePHP::getInstance(true); |
|
| 77 | + $instance->setEnabled($Enabled); |
|
| 78 | 78 | } |
| 79 | 79 | |
| 80 | 80 | /** |
@@ -84,8 +84,8 @@ discard block |
||
| 84 | 84 | * @return boolean TRUE if enabled |
| 85 | 85 | */ |
| 86 | 86 | public static function getEnabled() { |
| 87 | - $instance = FirePHP::getInstance(true); |
|
| 88 | - return $instance->getEnabled(); |
|
| 87 | + $instance = FirePHP::getInstance(true); |
|
| 88 | + return $instance->getEnabled(); |
|
| 89 | 89 | } |
| 90 | 90 | |
| 91 | 91 | /** |
@@ -99,8 +99,8 @@ discard block |
||
| 99 | 99 | * @return void |
| 100 | 100 | */ |
| 101 | 101 | public static function setObjectFilter($Class, $Filter) { |
| 102 | - $instance = FirePHP::getInstance(true); |
|
| 103 | - $instance->setObjectFilter($Class, $Filter); |
|
| 102 | + $instance = FirePHP::getInstance(true); |
|
| 103 | + $instance->setObjectFilter($Class, $Filter); |
|
| 104 | 104 | } |
| 105 | 105 | |
| 106 | 106 | /** |
@@ -111,8 +111,8 @@ discard block |
||
| 111 | 111 | * @return void |
| 112 | 112 | */ |
| 113 | 113 | public static function setOptions($Options) { |
| 114 | - $instance = FirePHP::getInstance(true); |
|
| 115 | - $instance->setOptions($Options); |
|
| 114 | + $instance = FirePHP::getInstance(true); |
|
| 115 | + $instance->setOptions($Options); |
|
| 116 | 116 | } |
| 117 | 117 | |
| 118 | 118 | /** |
@@ -122,8 +122,8 @@ discard block |
||
| 122 | 122 | * @return array The options |
| 123 | 123 | */ |
| 124 | 124 | public static function getOptions() { |
| 125 | - $instance = FirePHP::getInstance(true); |
|
| 126 | - return $instance->getOptions(); |
|
| 125 | + $instance = FirePHP::getInstance(true); |
|
| 126 | + return $instance->getOptions(); |
|
| 127 | 127 | } |
| 128 | 128 | |
| 129 | 129 | /** |
@@ -136,9 +136,9 @@ discard block |
||
| 136 | 136 | */ |
| 137 | 137 | public static function send() |
| 138 | 138 | { |
| 139 | - $instance = FirePHP::getInstance(true); |
|
| 140 | - $args = func_get_args(); |
|
| 141 | - return call_user_func_array(array($instance,'fb'),$args); |
|
| 139 | + $instance = FirePHP::getInstance(true); |
|
| 140 | + $args = func_get_args(); |
|
| 141 | + return call_user_func_array(array($instance,'fb'),$args); |
|
| 142 | 142 | } |
| 143 | 143 | |
| 144 | 144 | /** |
@@ -153,8 +153,8 @@ discard block |
||
| 153 | 153 | * @return true |
| 154 | 154 | */ |
| 155 | 155 | public static function group($Name, $Options=null) { |
| 156 | - $instance = FirePHP::getInstance(true); |
|
| 157 | - return $instance->group($Name, $Options); |
|
| 156 | + $instance = FirePHP::getInstance(true); |
|
| 157 | + return $instance->group($Name, $Options); |
|
| 158 | 158 | } |
| 159 | 159 | |
| 160 | 160 | /** |
@@ -164,7 +164,7 @@ discard block |
||
| 164 | 164 | * @throws Exception |
| 165 | 165 | */ |
| 166 | 166 | public static function groupEnd() { |
| 167 | - return self::send(null, null, FirePHP::GROUP_END); |
|
| 167 | + return self::send(null, null, FirePHP::GROUP_END); |
|
| 168 | 168 | } |
| 169 | 169 | |
| 170 | 170 | /** |
@@ -177,7 +177,7 @@ discard block |
||
| 177 | 177 | * @throws Exception |
| 178 | 178 | */ |
| 179 | 179 | public static function log($Object, $Label=null) { |
| 180 | - return self::send($Object, $Label, FirePHP::LOG); |
|
| 180 | + return self::send($Object, $Label, FirePHP::LOG); |
|
| 181 | 181 | } |
| 182 | 182 | |
| 183 | 183 | /** |
@@ -190,7 +190,7 @@ discard block |
||
| 190 | 190 | * @throws Exception |
| 191 | 191 | */ |
| 192 | 192 | public static function info($Object, $Label=null) { |
| 193 | - return self::send($Object, $Label, FirePHP::INFO); |
|
| 193 | + return self::send($Object, $Label, FirePHP::INFO); |
|
| 194 | 194 | } |
| 195 | 195 | |
| 196 | 196 | /** |
@@ -203,7 +203,7 @@ discard block |
||
| 203 | 203 | * @throws Exception |
| 204 | 204 | */ |
| 205 | 205 | public static function warn($Object, $Label=null) { |
| 206 | - return self::send($Object, $Label, FirePHP::WARN); |
|
| 206 | + return self::send($Object, $Label, FirePHP::WARN); |
|
| 207 | 207 | } |
| 208 | 208 | |
| 209 | 209 | /** |
@@ -216,7 +216,7 @@ discard block |
||
| 216 | 216 | * @throws Exception |
| 217 | 217 | */ |
| 218 | 218 | public static function error($Object, $Label=null) { |
| 219 | - return self::send($Object, $Label, FirePHP::ERROR); |
|
| 219 | + return self::send($Object, $Label, FirePHP::ERROR); |
|
| 220 | 220 | } |
| 221 | 221 | |
| 222 | 222 | /** |
@@ -229,7 +229,7 @@ discard block |
||
| 229 | 229 | * @throws Exception |
| 230 | 230 | */ |
| 231 | 231 | public static function dump($Key, $Variable) { |
| 232 | - return self::send($Variable, $Key, FirePHP::DUMP); |
|
| 232 | + return self::send($Variable, $Key, FirePHP::DUMP); |
|
| 233 | 233 | } |
| 234 | 234 | |
| 235 | 235 | /** |
@@ -241,7 +241,7 @@ discard block |
||
| 241 | 241 | * @throws Exception |
| 242 | 242 | */ |
| 243 | 243 | public static function trace($Label) { |
| 244 | - return self::send($Label, FirePHP::TRACE); |
|
| 244 | + return self::send($Label, FirePHP::TRACE); |
|
| 245 | 245 | } |
| 246 | 246 | |
| 247 | 247 | /** |
@@ -254,7 +254,7 @@ discard block |
||
| 254 | 254 | * @throws Exception |
| 255 | 255 | */ |
| 256 | 256 | public static function table($Label, $Table) { |
| 257 | - return self::send($Table, $Label, FirePHP::TABLE); |
|
| 257 | + return self::send($Table, $Label, FirePHP::TABLE); |
|
| 258 | 258 | } |
| 259 | 259 | |
| 260 | 260 | } |
@@ -45,13 +45,13 @@ discard block |
||
| 45 | 45 | * @return string - encoded request |
| 46 | 46 | */ |
| 47 | 47 | function _recaptcha_qsencode ($data) { |
| 48 | - $req = ""; |
|
| 49 | - foreach ( $data as $key => $value ) |
|
| 50 | - $req .= $key . '=' . urlencode( stripslashes($value) ) . '&'; |
|
| 48 | + $req = ""; |
|
| 49 | + foreach ( $data as $key => $value ) |
|
| 50 | + $req .= $key . '=' . urlencode( stripslashes($value) ) . '&'; |
|
| 51 | 51 | |
| 52 | - // Cut the last '&' |
|
| 53 | - $req=substr($req,0,strlen($req)-1); |
|
| 54 | - return $req; |
|
| 52 | + // Cut the last '&' |
|
| 53 | + $req=substr($req,0,strlen($req)-1); |
|
| 54 | + return $req; |
|
| 55 | 55 | } |
| 56 | 56 | |
| 57 | 57 | |
@@ -66,29 +66,29 @@ discard block |
||
| 66 | 66 | */ |
| 67 | 67 | function _recaptcha_http_post($host, $path, $data, $port = 80) { |
| 68 | 68 | |
| 69 | - $req = _recaptcha_qsencode ($data); |
|
| 69 | + $req = _recaptcha_qsencode ($data); |
|
| 70 | 70 | |
| 71 | - $http_request = "POST $path HTTP/1.0\r\n"; |
|
| 72 | - $http_request .= "Host: $host\r\n"; |
|
| 73 | - $http_request .= "Content-Type: application/x-www-form-urlencoded;\r\n"; |
|
| 74 | - $http_request .= "Content-Length: " . strlen($req) . "\r\n"; |
|
| 75 | - $http_request .= "User-Agent: reCAPTCHA/PHP\r\n"; |
|
| 76 | - $http_request .= "\r\n"; |
|
| 77 | - $http_request .= $req; |
|
| 71 | + $http_request = "POST $path HTTP/1.0\r\n"; |
|
| 72 | + $http_request .= "Host: $host\r\n"; |
|
| 73 | + $http_request .= "Content-Type: application/x-www-form-urlencoded;\r\n"; |
|
| 74 | + $http_request .= "Content-Length: " . strlen($req) . "\r\n"; |
|
| 75 | + $http_request .= "User-Agent: reCAPTCHA/PHP\r\n"; |
|
| 76 | + $http_request .= "\r\n"; |
|
| 77 | + $http_request .= $req; |
|
| 78 | 78 | |
| 79 | - $response = ''; |
|
| 80 | - if( false == ( $fs = @fsockopen($host, $port, $errno, $errstr, 10) ) ) { |
|
| 81 | - die ('Could not open socket'); |
|
| 82 | - } |
|
| 79 | + $response = ''; |
|
| 80 | + if( false == ( $fs = @fsockopen($host, $port, $errno, $errstr, 10) ) ) { |
|
| 81 | + die ('Could not open socket'); |
|
| 82 | + } |
|
| 83 | 83 | |
| 84 | - fwrite($fs, $http_request); |
|
| 84 | + fwrite($fs, $http_request); |
|
| 85 | 85 | |
| 86 | - while ( !feof($fs) ) |
|
| 87 | - $response .= fgets($fs, 1160); // One TCP-IP packet |
|
| 88 | - fclose($fs); |
|
| 89 | - $response = explode("\r\n\r\n", $response, 2); |
|
| 86 | + while ( !feof($fs) ) |
|
| 87 | + $response .= fgets($fs, 1160); // One TCP-IP packet |
|
| 88 | + fclose($fs); |
|
| 89 | + $response = explode("\r\n\r\n", $response, 2); |
|
| 90 | 90 | |
| 91 | - return $response; |
|
| 91 | + return $response; |
|
| 92 | 92 | } |
| 93 | 93 | |
| 94 | 94 | |
@@ -100,7 +100,6 @@ discard block |
||
| 100 | 100 | * @param string $pubkey A public key for reCAPTCHA |
| 101 | 101 | * @param string $error The error given by reCAPTCHA (optional, default is null) |
| 102 | 102 | * @param boolean $use_ssl Should the request be made over ssl? (optional, default is false) |
| 103 | - |
|
| 104 | 103 | * @return string - The HTML to be embedded in the user's form. |
| 105 | 104 | */ |
| 106 | 105 | function recaptcha_get_html ($pubkey, $error = null, $use_ssl = false) |
@@ -110,16 +109,16 @@ discard block |
||
| 110 | 109 | } |
| 111 | 110 | |
| 112 | 111 | if ($use_ssl) { |
| 113 | - $server = RECAPTCHA_API_SECURE_SERVER; |
|
| 114 | - } else { |
|
| 115 | - $server = RECAPTCHA_API_SERVER; |
|
| 116 | - } |
|
| 112 | + $server = RECAPTCHA_API_SECURE_SERVER; |
|
| 113 | + } else { |
|
| 114 | + $server = RECAPTCHA_API_SERVER; |
|
| 115 | + } |
|
| 117 | 116 | |
| 118 | - $errorpart = ""; |
|
| 119 | - if ($error) { |
|
| 120 | - $errorpart = "&error=" . $error; |
|
| 121 | - } |
|
| 122 | - return '<script type="text/javascript" src="'. $server . '/challenge?k=' . $pubkey . $errorpart . '"></script> |
|
| 117 | + $errorpart = ""; |
|
| 118 | + if ($error) { |
|
| 119 | + $errorpart = "&error=" . $error; |
|
| 120 | + } |
|
| 121 | + return '<script type="text/javascript" src="'. $server . '/challenge?k=' . $pubkey . $errorpart . '"></script> |
|
| 123 | 122 | |
| 124 | 123 | <noscript> |
| 125 | 124 | <iframe src="'. $server . '/noscript?k=' . $pubkey . $errorpart . '" height="300" width="500" frameborder="0"></iframe><br/> |
@@ -135,20 +134,20 @@ discard block |
||
| 135 | 134 | * A ReCaptchaResponse is returned from recaptcha_check_answer() |
| 136 | 135 | */ |
| 137 | 136 | class ReCaptchaResponse { |
| 138 | - var $is_valid; |
|
| 139 | - var $error; |
|
| 137 | + var $is_valid; |
|
| 138 | + var $error; |
|
| 140 | 139 | } |
| 141 | 140 | |
| 142 | 141 | |
| 143 | 142 | /** |
| 144 | - * Calls an HTTP POST function to verify if the user's guess was correct |
|
| 145 | - * @param string $privkey |
|
| 146 | - * @param string $remoteip |
|
| 147 | - * @param string $challenge |
|
| 148 | - * @param string $response |
|
| 149 | - * @param array $extra_params an array of extra variables to post to the server |
|
| 150 | - * @return ReCaptchaResponse |
|
| 151 | - */ |
|
| 143 | + * Calls an HTTP POST function to verify if the user's guess was correct |
|
| 144 | + * @param string $privkey |
|
| 145 | + * @param string $remoteip |
|
| 146 | + * @param string $challenge |
|
| 147 | + * @param string $response |
|
| 148 | + * @param array $extra_params an array of extra variables to post to the server |
|
| 149 | + * @return ReCaptchaResponse |
|
| 150 | + */ |
|
| 152 | 151 | function recaptcha_check_answer ($privkey, $remoteip, $challenge, $response, $extra_params = array()) |
| 153 | 152 | { |
| 154 | 153 | if ($privkey == null || $privkey == '') { |
@@ -161,34 +160,34 @@ discard block |
||
| 161 | 160 | |
| 162 | 161 | |
| 163 | 162 | |
| 164 | - //discard spam submissions |
|
| 165 | - if ($challenge == null || strlen($challenge) == 0 || $response == null || strlen($response) == 0) { |
|
| 166 | - $recaptcha_response = new ReCaptchaResponse(); |
|
| 167 | - $recaptcha_response->is_valid = false; |
|
| 168 | - $recaptcha_response->error = 'incorrect-captcha-sol'; |
|
| 169 | - return $recaptcha_response; |
|
| 170 | - } |
|
| 171 | - |
|
| 172 | - $response = _recaptcha_http_post (RECAPTCHA_VERIFY_SERVER, "/recaptcha/api/verify", |
|
| 173 | - array ( |
|
| 174 | - 'privatekey' => $privkey, |
|
| 175 | - 'remoteip' => $remoteip, |
|
| 176 | - 'challenge' => $challenge, |
|
| 177 | - 'response' => $response |
|
| 178 | - ) + $extra_params |
|
| 179 | - ); |
|
| 180 | - |
|
| 181 | - $answers = explode ("\n", $response [1]); |
|
| 182 | - $recaptcha_response = new ReCaptchaResponse(); |
|
| 183 | - |
|
| 184 | - if (trim ($answers [0]) == 'true') { |
|
| 185 | - $recaptcha_response->is_valid = true; |
|
| 186 | - } |
|
| 187 | - else { |
|
| 188 | - $recaptcha_response->is_valid = false; |
|
| 189 | - $recaptcha_response->error = $answers [1]; |
|
| 190 | - } |
|
| 191 | - return $recaptcha_response; |
|
| 163 | + //discard spam submissions |
|
| 164 | + if ($challenge == null || strlen($challenge) == 0 || $response == null || strlen($response) == 0) { |
|
| 165 | + $recaptcha_response = new ReCaptchaResponse(); |
|
| 166 | + $recaptcha_response->is_valid = false; |
|
| 167 | + $recaptcha_response->error = 'incorrect-captcha-sol'; |
|
| 168 | + return $recaptcha_response; |
|
| 169 | + } |
|
| 170 | + |
|
| 171 | + $response = _recaptcha_http_post (RECAPTCHA_VERIFY_SERVER, "/recaptcha/api/verify", |
|
| 172 | + array ( |
|
| 173 | + 'privatekey' => $privkey, |
|
| 174 | + 'remoteip' => $remoteip, |
|
| 175 | + 'challenge' => $challenge, |
|
| 176 | + 'response' => $response |
|
| 177 | + ) + $extra_params |
|
| 178 | + ); |
|
| 179 | + |
|
| 180 | + $answers = explode ("\n", $response [1]); |
|
| 181 | + $recaptcha_response = new ReCaptchaResponse(); |
|
| 182 | + |
|
| 183 | + if (trim ($answers [0]) == 'true') { |
|
| 184 | + $recaptcha_response->is_valid = true; |
|
| 185 | + } |
|
| 186 | + else { |
|
| 187 | + $recaptcha_response->is_valid = false; |
|
| 188 | + $recaptcha_response->error = $answers [1]; |
|
| 189 | + } |
|
| 190 | + return $recaptcha_response; |
|
| 192 | 191 | |
| 193 | 192 | } |
| 194 | 193 | |
@@ -230,7 +229,7 @@ discard block |
||
| 230 | 229 | function recaptcha_mailhide_url($pubkey, $privkey, $email) { |
| 231 | 230 | if ($pubkey == '' || $pubkey == null || $privkey == "" || $privkey == null) { |
| 232 | 231 | die ("To use reCAPTCHA Mailhide, you have to sign up for a public and private key, " . |
| 233 | - "you can do so at <a href='http://www.google.com/recaptcha/mailhide/apikey'>http://www.google.com/recaptcha/mailhide/apikey</a>"); |
|
| 232 | + "you can do so at <a href='http://www.google.com/recaptcha/mailhide/apikey'>http://www.google.com/recaptcha/mailhide/apikey</a>"); |
|
| 234 | 233 | } |
| 235 | 234 | |
| 236 | 235 | |
@@ -39,114 +39,114 @@ discard block |
||
| 39 | 39 | * @abstract |
| 40 | 40 | */ |
| 41 | 41 | class TSax3_StateParser { |
| 42 | - /** |
|
| 43 | - * Instance of user front end class to be passed to callbacks |
|
| 44 | - * @var TSax3 |
|
| 45 | - * @access private |
|
| 46 | - */ |
|
| 47 | - public $htmlsax; |
|
| 48 | - /** |
|
| 49 | - * User defined object for handling elements |
|
| 50 | - * @var object |
|
| 51 | - * @access private |
|
| 52 | - */ |
|
| 53 | - public $handler_object_element; |
|
| 54 | - /** |
|
| 55 | - * User defined open tag handler method |
|
| 56 | - * @var string |
|
| 57 | - * @access private |
|
| 58 | - */ |
|
| 59 | - public $handler_method_opening; |
|
| 60 | - /** |
|
| 61 | - * User defined close tag handler method |
|
| 62 | - * @var string |
|
| 63 | - * @access private |
|
| 64 | - */ |
|
| 65 | - public $handler_method_closing; |
|
| 66 | - /** |
|
| 67 | - * User defined object for handling data in elements |
|
| 68 | - * @var object |
|
| 69 | - * @access private |
|
| 70 | - */ |
|
| 71 | - public $handler_object_data; |
|
| 72 | - /** |
|
| 73 | - * User defined data handler method |
|
| 74 | - * @var string |
|
| 75 | - * @access private |
|
| 76 | - */ |
|
| 77 | - public $handler_method_data; |
|
| 78 | - /** |
|
| 79 | - * User defined object for handling processing instructions |
|
| 80 | - * @var object |
|
| 81 | - * @access private |
|
| 82 | - */ |
|
| 83 | - public $handler_object_pi; |
|
| 84 | - /** |
|
| 85 | - * User defined processing instruction handler method |
|
| 86 | - * @var string |
|
| 87 | - * @access private |
|
| 88 | - */ |
|
| 89 | - public $handler_method_pi; |
|
| 90 | - /** |
|
| 91 | - * User defined object for handling JSP/ASP tags |
|
| 92 | - * @var object |
|
| 93 | - * @access private |
|
| 94 | - */ |
|
| 95 | - public $handler_object_jasp; |
|
| 96 | - /** |
|
| 97 | - * User defined JSP/ASP handler method |
|
| 98 | - * @var string |
|
| 99 | - * @access private |
|
| 100 | - */ |
|
| 101 | - public $handler_method_jasp; |
|
| 102 | - /** |
|
| 103 | - * User defined object for handling XML escapes |
|
| 104 | - * @var object |
|
| 105 | - * @access private |
|
| 106 | - */ |
|
| 107 | - public $handler_object_escape; |
|
| 108 | - /** |
|
| 109 | - * User defined XML escape handler method |
|
| 110 | - * @var string |
|
| 111 | - * @access private |
|
| 112 | - */ |
|
| 113 | - public $handler_method_escape; |
|
| 114 | - /** |
|
| 115 | - * User defined handler object or NullHandler |
|
| 116 | - * @var object |
|
| 117 | - * @access private |
|
| 118 | - */ |
|
| 119 | - public $handler_default; |
|
| 120 | - /** |
|
| 121 | - * Parser options determining parsing behavior |
|
| 122 | - * @var array |
|
| 123 | - * @access private |
|
| 124 | - */ |
|
| 125 | - protected $parser_options = array(); |
|
| 126 | - /** |
|
| 127 | - * XML document being parsed |
|
| 128 | - * @var string |
|
| 129 | - * @access private |
|
| 130 | - */ |
|
| 131 | - protected $rawtext; |
|
| 132 | - /** |
|
| 133 | - * Position in XML document relative to start (0) |
|
| 134 | - * @var int |
|
| 135 | - * @access private |
|
| 136 | - */ |
|
| 137 | - protected $position; |
|
| 138 | - /** |
|
| 139 | - * Length of the XML document in characters |
|
| 140 | - * @var int |
|
| 141 | - * @access private |
|
| 142 | - */ |
|
| 143 | - protected $length; |
|
| 144 | - /** |
|
| 145 | - * Array of state objects |
|
| 146 | - * @var array |
|
| 147 | - * @access private |
|
| 148 | - */ |
|
| 149 | - protected $State = array(); |
|
| 42 | + /** |
|
| 43 | + * Instance of user front end class to be passed to callbacks |
|
| 44 | + * @var TSax3 |
|
| 45 | + * @access private |
|
| 46 | + */ |
|
| 47 | + public $htmlsax; |
|
| 48 | + /** |
|
| 49 | + * User defined object for handling elements |
|
| 50 | + * @var object |
|
| 51 | + * @access private |
|
| 52 | + */ |
|
| 53 | + public $handler_object_element; |
|
| 54 | + /** |
|
| 55 | + * User defined open tag handler method |
|
| 56 | + * @var string |
|
| 57 | + * @access private |
|
| 58 | + */ |
|
| 59 | + public $handler_method_opening; |
|
| 60 | + /** |
|
| 61 | + * User defined close tag handler method |
|
| 62 | + * @var string |
|
| 63 | + * @access private |
|
| 64 | + */ |
|
| 65 | + public $handler_method_closing; |
|
| 66 | + /** |
|
| 67 | + * User defined object for handling data in elements |
|
| 68 | + * @var object |
|
| 69 | + * @access private |
|
| 70 | + */ |
|
| 71 | + public $handler_object_data; |
|
| 72 | + /** |
|
| 73 | + * User defined data handler method |
|
| 74 | + * @var string |
|
| 75 | + * @access private |
|
| 76 | + */ |
|
| 77 | + public $handler_method_data; |
|
| 78 | + /** |
|
| 79 | + * User defined object for handling processing instructions |
|
| 80 | + * @var object |
|
| 81 | + * @access private |
|
| 82 | + */ |
|
| 83 | + public $handler_object_pi; |
|
| 84 | + /** |
|
| 85 | + * User defined processing instruction handler method |
|
| 86 | + * @var string |
|
| 87 | + * @access private |
|
| 88 | + */ |
|
| 89 | + public $handler_method_pi; |
|
| 90 | + /** |
|
| 91 | + * User defined object for handling JSP/ASP tags |
|
| 92 | + * @var object |
|
| 93 | + * @access private |
|
| 94 | + */ |
|
| 95 | + public $handler_object_jasp; |
|
| 96 | + /** |
|
| 97 | + * User defined JSP/ASP handler method |
|
| 98 | + * @var string |
|
| 99 | + * @access private |
|
| 100 | + */ |
|
| 101 | + public $handler_method_jasp; |
|
| 102 | + /** |
|
| 103 | + * User defined object for handling XML escapes |
|
| 104 | + * @var object |
|
| 105 | + * @access private |
|
| 106 | + */ |
|
| 107 | + public $handler_object_escape; |
|
| 108 | + /** |
|
| 109 | + * User defined XML escape handler method |
|
| 110 | + * @var string |
|
| 111 | + * @access private |
|
| 112 | + */ |
|
| 113 | + public $handler_method_escape; |
|
| 114 | + /** |
|
| 115 | + * User defined handler object or NullHandler |
|
| 116 | + * @var object |
|
| 117 | + * @access private |
|
| 118 | + */ |
|
| 119 | + public $handler_default; |
|
| 120 | + /** |
|
| 121 | + * Parser options determining parsing behavior |
|
| 122 | + * @var array |
|
| 123 | + * @access private |
|
| 124 | + */ |
|
| 125 | + protected $parser_options = array(); |
|
| 126 | + /** |
|
| 127 | + * XML document being parsed |
|
| 128 | + * @var string |
|
| 129 | + * @access private |
|
| 130 | + */ |
|
| 131 | + protected $rawtext; |
|
| 132 | + /** |
|
| 133 | + * Position in XML document relative to start (0) |
|
| 134 | + * @var int |
|
| 135 | + * @access private |
|
| 136 | + */ |
|
| 137 | + protected $position; |
|
| 138 | + /** |
|
| 139 | + * Length of the XML document in characters |
|
| 140 | + * @var int |
|
| 141 | + * @access private |
|
| 142 | + */ |
|
| 143 | + protected $length; |
|
| 144 | + /** |
|
| 145 | + * Array of state objects |
|
| 146 | + * @var array |
|
| 147 | + * @access private |
|
| 148 | + */ |
|
| 149 | + protected $State = array(); |
|
| 150 | 150 | |
| 151 | 151 | const TSAX3_STATE_STOP = 0; |
| 152 | 152 | const TSAX3_STATE_START = 1; |
@@ -157,166 +157,166 @@ discard block |
||
| 157 | 157 | const TSAX3_STATE_JASP = 7; |
| 158 | 158 | const TSAX3_STATE_PI = 8; |
| 159 | 159 | |
| 160 | - /** |
|
| 161 | - * Constructs TSax3_StateParser setting up states |
|
| 162 | - * @var TSax3 instance of user front end class |
|
| 163 | - * @access protected |
|
| 164 | - */ |
|
| 165 | - protected function __construct($htmlsax) { |
|
| 166 | - $this->htmlsax = $htmlsax; |
|
| 167 | - $this->State[self::TSAX3_STATE_START] = new TSax3_StartingState(); |
|
| 168 | - |
|
| 169 | - $this->State[self::TSAX3_STATE_CLOSING_TAG] = new TSax3_ClosingTagState(); |
|
| 170 | - $this->State[self::TSAX3_STATE_TAG] = new TSax3_TagState(); |
|
| 171 | - $this->State[self::TSAX3_STATE_OPENING_TAG] = new TSax3_OpeningTagState(); |
|
| 172 | - |
|
| 173 | - $this->State[self::TSAX3_STATE_PI] = new TSax3_PiState(); |
|
| 174 | - $this->State[self::TSAX3_STATE_JASP] = new TSax3_JaspState(); |
|
| 175 | - $this->State[self::TSAX3_STATE_ESCAPE] = new TSax3_EscapeState(); |
|
| 176 | - } |
|
| 177 | - |
|
| 178 | - /** |
|
| 179 | - * Moves the position back one character |
|
| 180 | - * @access protected |
|
| 181 | - * @return void |
|
| 182 | - */ |
|
| 183 | - function unscanCharacter() { |
|
| 184 | - $this->position -= 1; |
|
| 185 | - } |
|
| 186 | - |
|
| 187 | - /** |
|
| 188 | - * Moves the position forward one character |
|
| 189 | - * @access protected |
|
| 190 | - * @return void |
|
| 191 | - */ |
|
| 192 | - function ignoreCharacter() { |
|
| 193 | - $this->position += 1; |
|
| 194 | - } |
|
| 195 | - |
|
| 196 | - /** |
|
| 197 | - * Returns the next character from the XML document or void if at end |
|
| 198 | - * @access protected |
|
| 199 | - * @return mixed |
|
| 200 | - */ |
|
| 201 | - function scanCharacter() { |
|
| 202 | - if ($this->position < $this->length) { |
|
| 203 | - return $this->rawtext{$this->position++}; |
|
| 204 | - } |
|
| 205 | - } |
|
| 206 | - |
|
| 207 | - /** |
|
| 208 | - * Returns a string from the current position to the next occurance |
|
| 209 | - * of the supplied string |
|
| 210 | - * @param string string to search until |
|
| 211 | - * @access protected |
|
| 212 | - * @return string |
|
| 213 | - */ |
|
| 214 | - function scanUntilString($string) { |
|
| 215 | - $start = $this->position; |
|
| 216 | - $this->position = strpos($this->rawtext, $string, $start); |
|
| 217 | - if ($this->position === FALSE) { |
|
| 218 | - $this->position = $this->length; |
|
| 219 | - } |
|
| 220 | - return substr($this->rawtext, $start, $this->position - $start); |
|
| 221 | - } |
|
| 222 | - |
|
| 223 | - /** |
|
| 224 | - * Returns a string from the current position until the first instance of |
|
| 225 | - * one of the characters in the supplied string argument |
|
| 226 | - * @param string string to search until |
|
| 227 | - * @access protected |
|
| 228 | - * @return string |
|
| 229 | - * @abstract |
|
| 230 | - */ |
|
| 231 | - function scanUntilCharacters($string) {} |
|
| 232 | - |
|
| 233 | - /** |
|
| 234 | - * Moves the position forward past any whitespace characters |
|
| 235 | - * @access protected |
|
| 236 | - * @return void |
|
| 237 | - * @abstract |
|
| 238 | - */ |
|
| 239 | - function ignoreWhitespace() {} |
|
| 240 | - |
|
| 241 | - /** |
|
| 242 | - * Begins the parsing operation, setting up any decorators, depending on |
|
| 243 | - * parse options invoking _parse() to execute parsing |
|
| 244 | - * @param string XML document to parse |
|
| 245 | - * @access protected |
|
| 246 | - * @return void |
|
| 247 | - */ |
|
| 248 | - function parse($data) { |
|
| 249 | - if ($this->parser_options['XML_OPTION_TRIM_DATA_NODES']==1) { |
|
| 250 | - $decorator = new TSax3_Trim( |
|
| 251 | - $this->handler_object_data, |
|
| 252 | - $this->handler_method_data); |
|
| 253 | - $this->handler_object_data =& $decorator; |
|
| 254 | - $this->handler_method_data = 'trimData'; |
|
| 255 | - } |
|
| 256 | - if ($this->parser_options['XML_OPTION_CASE_FOLDING']==1) { |
|
| 257 | - $open_decor = new TSax3_CaseFolding( |
|
| 258 | - $this->handler_object_element, |
|
| 259 | - $this->handler_method_opening, |
|
| 260 | - $this->handler_method_closing); |
|
| 261 | - $this->handler_object_element =& $open_decor; |
|
| 262 | - $this->handler_method_opening ='foldOpen'; |
|
| 263 | - $this->handler_method_closing ='foldClose'; |
|
| 264 | - } |
|
| 265 | - if ($this->parser_options['XML_OPTION_LINEFEED_BREAK']==1) { |
|
| 266 | - $decorator = new TSax3_Linefeed( |
|
| 267 | - $this->handler_object_data, |
|
| 268 | - $this->handler_method_data); |
|
| 269 | - $this->handler_object_data =& $decorator; |
|
| 270 | - $this->handler_method_data = 'breakData'; |
|
| 271 | - } |
|
| 272 | - if ($this->parser_options['XML_OPTION_TAB_BREAK']==1) { |
|
| 273 | - $decorator = new TSax3_Tab( |
|
| 274 | - $this->handler_object_data, |
|
| 275 | - $this->handler_method_data); |
|
| 276 | - $this->handler_object_data =& $decorator; |
|
| 277 | - $this->handler_method_data = 'breakData'; |
|
| 278 | - } |
|
| 279 | - if ($this->parser_options['XML_OPTION_ENTITIES_UNPARSED']==1) { |
|
| 280 | - $decorator = new TSax3_Entities_Unparsed( |
|
| 281 | - $this->handler_object_data, |
|
| 282 | - $this->handler_method_data); |
|
| 283 | - $this->handler_object_data =& $decorator; |
|
| 284 | - $this->handler_method_data = 'breakData'; |
|
| 285 | - } |
|
| 286 | - if ($this->parser_options['XML_OPTION_ENTITIES_PARSED']==1) { |
|
| 287 | - $decorator = new TSax3_Entities_Parsed( |
|
| 288 | - $this->handler_object_data, |
|
| 289 | - $this->handler_method_data); |
|
| 290 | - $this->handler_object_data =& $decorator; |
|
| 291 | - $this->handler_method_data = 'breakData'; |
|
| 292 | - } |
|
| 293 | - // Note switched on by default |
|
| 294 | - if ($this->parser_options['XML_OPTION_STRIP_ESCAPES']==1) { |
|
| 295 | - $decorator = new TSax3_Escape_Stripper( |
|
| 296 | - $this->handler_object_escape, |
|
| 297 | - $this->handler_method_escape); |
|
| 298 | - $this->handler_object_escape =& $decorator; |
|
| 299 | - $this->handler_method_escape = 'strip'; |
|
| 300 | - } |
|
| 301 | - $this->rawtext = $data; |
|
| 302 | - $this->length = strlen($data); |
|
| 303 | - $this->position = 0; |
|
| 304 | - $this->_parse(); |
|
| 305 | - } |
|
| 306 | - |
|
| 307 | - /** |
|
| 308 | - * Performs the parsing itself, delegating calls to a specific parser |
|
| 309 | - * state |
|
| 310 | - * @param constant state object to parse with |
|
| 311 | - * @access protected |
|
| 312 | - * @return void |
|
| 313 | - */ |
|
| 314 | - function _parse($state = self::TSAX3_STATE_START) { |
|
| 315 | - do { |
|
| 316 | - $state = $this->State[$state]->parse($this); |
|
| 317 | - } while ($state != self::TSAX3_STATE_STOP && |
|
| 318 | - $this->position < $this->length); |
|
| 319 | - } |
|
| 160 | + /** |
|
| 161 | + * Constructs TSax3_StateParser setting up states |
|
| 162 | + * @var TSax3 instance of user front end class |
|
| 163 | + * @access protected |
|
| 164 | + */ |
|
| 165 | + protected function __construct($htmlsax) { |
|
| 166 | + $this->htmlsax = $htmlsax; |
|
| 167 | + $this->State[self::TSAX3_STATE_START] = new TSax3_StartingState(); |
|
| 168 | + |
|
| 169 | + $this->State[self::TSAX3_STATE_CLOSING_TAG] = new TSax3_ClosingTagState(); |
|
| 170 | + $this->State[self::TSAX3_STATE_TAG] = new TSax3_TagState(); |
|
| 171 | + $this->State[self::TSAX3_STATE_OPENING_TAG] = new TSax3_OpeningTagState(); |
|
| 172 | + |
|
| 173 | + $this->State[self::TSAX3_STATE_PI] = new TSax3_PiState(); |
|
| 174 | + $this->State[self::TSAX3_STATE_JASP] = new TSax3_JaspState(); |
|
| 175 | + $this->State[self::TSAX3_STATE_ESCAPE] = new TSax3_EscapeState(); |
|
| 176 | + } |
|
| 177 | + |
|
| 178 | + /** |
|
| 179 | + * Moves the position back one character |
|
| 180 | + * @access protected |
|
| 181 | + * @return void |
|
| 182 | + */ |
|
| 183 | + function unscanCharacter() { |
|
| 184 | + $this->position -= 1; |
|
| 185 | + } |
|
| 186 | + |
|
| 187 | + /** |
|
| 188 | + * Moves the position forward one character |
|
| 189 | + * @access protected |
|
| 190 | + * @return void |
|
| 191 | + */ |
|
| 192 | + function ignoreCharacter() { |
|
| 193 | + $this->position += 1; |
|
| 194 | + } |
|
| 195 | + |
|
| 196 | + /** |
|
| 197 | + * Returns the next character from the XML document or void if at end |
|
| 198 | + * @access protected |
|
| 199 | + * @return mixed |
|
| 200 | + */ |
|
| 201 | + function scanCharacter() { |
|
| 202 | + if ($this->position < $this->length) { |
|
| 203 | + return $this->rawtext{$this->position++}; |
|
| 204 | + } |
|
| 205 | + } |
|
| 206 | + |
|
| 207 | + /** |
|
| 208 | + * Returns a string from the current position to the next occurance |
|
| 209 | + * of the supplied string |
|
| 210 | + * @param string string to search until |
|
| 211 | + * @access protected |
|
| 212 | + * @return string |
|
| 213 | + */ |
|
| 214 | + function scanUntilString($string) { |
|
| 215 | + $start = $this->position; |
|
| 216 | + $this->position = strpos($this->rawtext, $string, $start); |
|
| 217 | + if ($this->position === FALSE) { |
|
| 218 | + $this->position = $this->length; |
|
| 219 | + } |
|
| 220 | + return substr($this->rawtext, $start, $this->position - $start); |
|
| 221 | + } |
|
| 222 | + |
|
| 223 | + /** |
|
| 224 | + * Returns a string from the current position until the first instance of |
|
| 225 | + * one of the characters in the supplied string argument |
|
| 226 | + * @param string string to search until |
|
| 227 | + * @access protected |
|
| 228 | + * @return string |
|
| 229 | + * @abstract |
|
| 230 | + */ |
|
| 231 | + function scanUntilCharacters($string) {} |
|
| 232 | + |
|
| 233 | + /** |
|
| 234 | + * Moves the position forward past any whitespace characters |
|
| 235 | + * @access protected |
|
| 236 | + * @return void |
|
| 237 | + * @abstract |
|
| 238 | + */ |
|
| 239 | + function ignoreWhitespace() {} |
|
| 240 | + |
|
| 241 | + /** |
|
| 242 | + * Begins the parsing operation, setting up any decorators, depending on |
|
| 243 | + * parse options invoking _parse() to execute parsing |
|
| 244 | + * @param string XML document to parse |
|
| 245 | + * @access protected |
|
| 246 | + * @return void |
|
| 247 | + */ |
|
| 248 | + function parse($data) { |
|
| 249 | + if ($this->parser_options['XML_OPTION_TRIM_DATA_NODES']==1) { |
|
| 250 | + $decorator = new TSax3_Trim( |
|
| 251 | + $this->handler_object_data, |
|
| 252 | + $this->handler_method_data); |
|
| 253 | + $this->handler_object_data =& $decorator; |
|
| 254 | + $this->handler_method_data = 'trimData'; |
|
| 255 | + } |
|
| 256 | + if ($this->parser_options['XML_OPTION_CASE_FOLDING']==1) { |
|
| 257 | + $open_decor = new TSax3_CaseFolding( |
|
| 258 | + $this->handler_object_element, |
|
| 259 | + $this->handler_method_opening, |
|
| 260 | + $this->handler_method_closing); |
|
| 261 | + $this->handler_object_element =& $open_decor; |
|
| 262 | + $this->handler_method_opening ='foldOpen'; |
|
| 263 | + $this->handler_method_closing ='foldClose'; |
|
| 264 | + } |
|
| 265 | + if ($this->parser_options['XML_OPTION_LINEFEED_BREAK']==1) { |
|
| 266 | + $decorator = new TSax3_Linefeed( |
|
| 267 | + $this->handler_object_data, |
|
| 268 | + $this->handler_method_data); |
|
| 269 | + $this->handler_object_data =& $decorator; |
|
| 270 | + $this->handler_method_data = 'breakData'; |
|
| 271 | + } |
|
| 272 | + if ($this->parser_options['XML_OPTION_TAB_BREAK']==1) { |
|
| 273 | + $decorator = new TSax3_Tab( |
|
| 274 | + $this->handler_object_data, |
|
| 275 | + $this->handler_method_data); |
|
| 276 | + $this->handler_object_data =& $decorator; |
|
| 277 | + $this->handler_method_data = 'breakData'; |
|
| 278 | + } |
|
| 279 | + if ($this->parser_options['XML_OPTION_ENTITIES_UNPARSED']==1) { |
|
| 280 | + $decorator = new TSax3_Entities_Unparsed( |
|
| 281 | + $this->handler_object_data, |
|
| 282 | + $this->handler_method_data); |
|
| 283 | + $this->handler_object_data =& $decorator; |
|
| 284 | + $this->handler_method_data = 'breakData'; |
|
| 285 | + } |
|
| 286 | + if ($this->parser_options['XML_OPTION_ENTITIES_PARSED']==1) { |
|
| 287 | + $decorator = new TSax3_Entities_Parsed( |
|
| 288 | + $this->handler_object_data, |
|
| 289 | + $this->handler_method_data); |
|
| 290 | + $this->handler_object_data =& $decorator; |
|
| 291 | + $this->handler_method_data = 'breakData'; |
|
| 292 | + } |
|
| 293 | + // Note switched on by default |
|
| 294 | + if ($this->parser_options['XML_OPTION_STRIP_ESCAPES']==1) { |
|
| 295 | + $decorator = new TSax3_Escape_Stripper( |
|
| 296 | + $this->handler_object_escape, |
|
| 297 | + $this->handler_method_escape); |
|
| 298 | + $this->handler_object_escape =& $decorator; |
|
| 299 | + $this->handler_method_escape = 'strip'; |
|
| 300 | + } |
|
| 301 | + $this->rawtext = $data; |
|
| 302 | + $this->length = strlen($data); |
|
| 303 | + $this->position = 0; |
|
| 304 | + $this->_parse(); |
|
| 305 | + } |
|
| 306 | + |
|
| 307 | + /** |
|
| 308 | + * Performs the parsing itself, delegating calls to a specific parser |
|
| 309 | + * state |
|
| 310 | + * @param constant state object to parse with |
|
| 311 | + * @access protected |
|
| 312 | + * @return void |
|
| 313 | + */ |
|
| 314 | + function _parse($state = self::TSAX3_STATE_START) { |
|
| 315 | + do { |
|
| 316 | + $state = $this->State[$state]->parse($this); |
|
| 317 | + } while ($state != self::TSAX3_STATE_STOP && |
|
| 318 | + $this->position < $this->length); |
|
| 319 | + } |
|
| 320 | 320 | } |
| 321 | 321 | |
| 322 | 322 | /** |
@@ -327,61 +327,61 @@ discard block |
||
| 327 | 327 | * @see TSax3_StateParser_Gtet430 |
| 328 | 328 | */ |
| 329 | 329 | class TSax3_StateParser_Lt430 extends TSax3_StateParser { |
| 330 | - /** |
|
| 331 | - * Constructs TSax3_StateParser_Lt430 defining available |
|
| 332 | - * parser options |
|
| 333 | - * @var TSax3 instance of user front end class |
|
| 334 | - * @access protected |
|
| 335 | - */ |
|
| 336 | - function __construct(& $htmlsax) { |
|
| 337 | - parent::__construct($htmlsax); |
|
| 338 | - $this->parser_options['XML_OPTION_TRIM_DATA_NODES'] = 0; |
|
| 339 | - $this->parser_options['XML_OPTION_CASE_FOLDING'] = 0; |
|
| 340 | - $this->parser_options['XML_OPTION_LINEFEED_BREAK'] = 0; |
|
| 341 | - $this->parser_options['XML_OPTION_TAB_BREAK'] = 0; |
|
| 342 | - $this->parser_options['XML_OPTION_ENTITIES_PARSED'] = 0; |
|
| 343 | - $this->parser_options['XML_OPTION_ENTITIES_UNPARSED'] = 0; |
|
| 344 | - $this->parser_options['XML_OPTION_STRIP_ESCAPES'] = 0; |
|
| 330 | + /** |
|
| 331 | + * Constructs TSax3_StateParser_Lt430 defining available |
|
| 332 | + * parser options |
|
| 333 | + * @var TSax3 instance of user front end class |
|
| 334 | + * @access protected |
|
| 335 | + */ |
|
| 336 | + function __construct(& $htmlsax) { |
|
| 337 | + parent::__construct($htmlsax); |
|
| 338 | + $this->parser_options['XML_OPTION_TRIM_DATA_NODES'] = 0; |
|
| 339 | + $this->parser_options['XML_OPTION_CASE_FOLDING'] = 0; |
|
| 340 | + $this->parser_options['XML_OPTION_LINEFEED_BREAK'] = 0; |
|
| 341 | + $this->parser_options['XML_OPTION_TAB_BREAK'] = 0; |
|
| 342 | + $this->parser_options['XML_OPTION_ENTITIES_PARSED'] = 0; |
|
| 343 | + $this->parser_options['XML_OPTION_ENTITIES_UNPARSED'] = 0; |
|
| 344 | + $this->parser_options['XML_OPTION_STRIP_ESCAPES'] = 0; |
|
| 345 | 345 | //var_dump($this->parser_options); |
| 346 | - } |
|
| 347 | - |
|
| 348 | - /** |
|
| 349 | - * Returns a string from the current position until the first instance of |
|
| 350 | - * one of the characters in the supplied string argument |
|
| 351 | - * @param string string to search until |
|
| 352 | - * @access protected |
|
| 353 | - * @return string |
|
| 354 | - */ |
|
| 355 | - function scanUntilCharacters($string) { |
|
| 356 | - $startpos = $this->position; |
|
| 357 | - while ($this->position < $this->length && strpos($string, $this->rawtext{$this->position}) === FALSE) { |
|
| 358 | - $this->position++; |
|
| 359 | - } |
|
| 360 | - return substr($this->rawtext, $startpos, $this->position - $startpos); |
|
| 361 | - } |
|
| 362 | - |
|
| 363 | - /** |
|
| 364 | - * Moves the position forward past any whitespace characters |
|
| 365 | - * @access protected |
|
| 366 | - * @return void |
|
| 367 | - */ |
|
| 368 | - function ignoreWhitespace() { |
|
| 369 | - while ($this->position < $this->length && |
|
| 370 | - strpos(" \n\r\t", $this->rawtext{$this->position}) !== FALSE) { |
|
| 371 | - $this->position++; |
|
| 372 | - } |
|
| 373 | - } |
|
| 374 | - |
|
| 375 | - /** |
|
| 376 | - * Begins the parsing operation, setting up the unparsed XML entities |
|
| 377 | - * decorator if necessary then delegating further work to parent |
|
| 378 | - * @param string XML document to parse |
|
| 379 | - * @access protected |
|
| 380 | - * @return void |
|
| 381 | - */ |
|
| 382 | - function parse($data) { |
|
| 383 | - parent::parse($data); |
|
| 384 | - } |
|
| 346 | + } |
|
| 347 | + |
|
| 348 | + /** |
|
| 349 | + * Returns a string from the current position until the first instance of |
|
| 350 | + * one of the characters in the supplied string argument |
|
| 351 | + * @param string string to search until |
|
| 352 | + * @access protected |
|
| 353 | + * @return string |
|
| 354 | + */ |
|
| 355 | + function scanUntilCharacters($string) { |
|
| 356 | + $startpos = $this->position; |
|
| 357 | + while ($this->position < $this->length && strpos($string, $this->rawtext{$this->position}) === FALSE) { |
|
| 358 | + $this->position++; |
|
| 359 | + } |
|
| 360 | + return substr($this->rawtext, $startpos, $this->position - $startpos); |
|
| 361 | + } |
|
| 362 | + |
|
| 363 | + /** |
|
| 364 | + * Moves the position forward past any whitespace characters |
|
| 365 | + * @access protected |
|
| 366 | + * @return void |
|
| 367 | + */ |
|
| 368 | + function ignoreWhitespace() { |
|
| 369 | + while ($this->position < $this->length && |
|
| 370 | + strpos(" \n\r\t", $this->rawtext{$this->position}) !== FALSE) { |
|
| 371 | + $this->position++; |
|
| 372 | + } |
|
| 373 | + } |
|
| 374 | + |
|
| 375 | + /** |
|
| 376 | + * Begins the parsing operation, setting up the unparsed XML entities |
|
| 377 | + * decorator if necessary then delegating further work to parent |
|
| 378 | + * @param string XML document to parse |
|
| 379 | + * @access protected |
|
| 380 | + * @return void |
|
| 381 | + */ |
|
| 382 | + function parse($data) { |
|
| 383 | + parent::parse($data); |
|
| 384 | + } |
|
| 385 | 385 | } |
| 386 | 386 | |
| 387 | 387 | /** |
@@ -392,56 +392,56 @@ discard block |
||
| 392 | 392 | * @see TSax3_StateParser_Lt430 |
| 393 | 393 | */ |
| 394 | 394 | class TSax3_StateParser_Gtet430 extends TSax3_StateParser { |
| 395 | - /** |
|
| 396 | - * Constructs TSax3_StateParser_Gtet430 defining available |
|
| 397 | - * parser options |
|
| 398 | - * @var TSax3 instance of user front end class |
|
| 399 | - * @access protected |
|
| 400 | - */ |
|
| 401 | - function __construct(& $htmlsax) { |
|
| 402 | - parent::__construct($htmlsax); |
|
| 403 | - $this->parser_options['XML_OPTION_TRIM_DATA_NODES'] = 0; |
|
| 404 | - $this->parser_options['XML_OPTION_CASE_FOLDING'] = 0; |
|
| 405 | - $this->parser_options['XML_OPTION_LINEFEED_BREAK'] = 0; |
|
| 406 | - $this->parser_options['XML_OPTION_TAB_BREAK'] = 0; |
|
| 407 | - $this->parser_options['XML_OPTION_ENTITIES_PARSED'] = 0; |
|
| 408 | - $this->parser_options['XML_OPTION_ENTITIES_UNPARSED'] = 0; |
|
| 409 | - $this->parser_options['XML_OPTION_STRIP_ESCAPES'] = 0; |
|
| 410 | - } |
|
| 411 | - /** |
|
| 412 | - * Returns a string from the current position until the first instance of |
|
| 413 | - * one of the characters in the supplied string argument. |
|
| 414 | - * @param string string to search until |
|
| 415 | - * @access protected |
|
| 416 | - * @return string |
|
| 417 | - */ |
|
| 418 | - function scanUntilCharacters($string) { |
|
| 419 | - $startpos = $this->position; |
|
| 420 | - $length = strcspn($this->rawtext, $string, $startpos); |
|
| 421 | - $this->position += $length; |
|
| 422 | - return substr($this->rawtext, $startpos, $length); |
|
| 423 | - } |
|
| 424 | - |
|
| 425 | - /** |
|
| 426 | - * Moves the position forward past any whitespace characters |
|
| 427 | - * @access protected |
|
| 428 | - * @return void |
|
| 429 | - */ |
|
| 430 | - function ignoreWhitespace() { |
|
| 431 | - $this->position += strspn($this->rawtext, " \n\r\t", $this->position); |
|
| 432 | - } |
|
| 433 | - |
|
| 434 | - /** |
|
| 435 | - * Begins the parsing operation, setting up the parsed and unparsed |
|
| 436 | - * XML entity decorators if necessary then delegating further work |
|
| 437 | - * to parent |
|
| 438 | - * @param string XML document to parse |
|
| 439 | - * @access protected |
|
| 440 | - * @return void |
|
| 441 | - */ |
|
| 442 | - function parse($data) { |
|
| 443 | - parent::parse($data); |
|
| 444 | - } |
|
| 395 | + /** |
|
| 396 | + * Constructs TSax3_StateParser_Gtet430 defining available |
|
| 397 | + * parser options |
|
| 398 | + * @var TSax3 instance of user front end class |
|
| 399 | + * @access protected |
|
| 400 | + */ |
|
| 401 | + function __construct(& $htmlsax) { |
|
| 402 | + parent::__construct($htmlsax); |
|
| 403 | + $this->parser_options['XML_OPTION_TRIM_DATA_NODES'] = 0; |
|
| 404 | + $this->parser_options['XML_OPTION_CASE_FOLDING'] = 0; |
|
| 405 | + $this->parser_options['XML_OPTION_LINEFEED_BREAK'] = 0; |
|
| 406 | + $this->parser_options['XML_OPTION_TAB_BREAK'] = 0; |
|
| 407 | + $this->parser_options['XML_OPTION_ENTITIES_PARSED'] = 0; |
|
| 408 | + $this->parser_options['XML_OPTION_ENTITIES_UNPARSED'] = 0; |
|
| 409 | + $this->parser_options['XML_OPTION_STRIP_ESCAPES'] = 0; |
|
| 410 | + } |
|
| 411 | + /** |
|
| 412 | + * Returns a string from the current position until the first instance of |
|
| 413 | + * one of the characters in the supplied string argument. |
|
| 414 | + * @param string string to search until |
|
| 415 | + * @access protected |
|
| 416 | + * @return string |
|
| 417 | + */ |
|
| 418 | + function scanUntilCharacters($string) { |
|
| 419 | + $startpos = $this->position; |
|
| 420 | + $length = strcspn($this->rawtext, $string, $startpos); |
|
| 421 | + $this->position += $length; |
|
| 422 | + return substr($this->rawtext, $startpos, $length); |
|
| 423 | + } |
|
| 424 | + |
|
| 425 | + /** |
|
| 426 | + * Moves the position forward past any whitespace characters |
|
| 427 | + * @access protected |
|
| 428 | + * @return void |
|
| 429 | + */ |
|
| 430 | + function ignoreWhitespace() { |
|
| 431 | + $this->position += strspn($this->rawtext, " \n\r\t", $this->position); |
|
| 432 | + } |
|
| 433 | + |
|
| 434 | + /** |
|
| 435 | + * Begins the parsing operation, setting up the parsed and unparsed |
|
| 436 | + * XML entity decorators if necessary then delegating further work |
|
| 437 | + * to parent |
|
| 438 | + * @param string XML document to parse |
|
| 439 | + * @access protected |
|
| 440 | + * @return void |
|
| 441 | + */ |
|
| 442 | + function parse($data) { |
|
| 443 | + parent::parse($data); |
|
| 444 | + } |
|
| 445 | 445 | } |
| 446 | 446 | |
| 447 | 447 | /** |
@@ -450,13 +450,13 @@ discard block |
||
| 450 | 450 | * @access protected |
| 451 | 451 | */ |
| 452 | 452 | class TSax3_NullHandler { |
| 453 | - /** |
|
| 454 | - * Generic handler method which does nothing |
|
| 455 | - * @access protected |
|
| 456 | - * @return void |
|
| 457 | - */ |
|
| 458 | - function DoNothing() { |
|
| 459 | - } |
|
| 453 | + /** |
|
| 454 | + * Generic handler method which does nothing |
|
| 455 | + * @access protected |
|
| 456 | + * @return void |
|
| 457 | + */ |
|
| 458 | + function DoNothing() { |
|
| 459 | + } |
|
| 460 | 460 | } |
| 461 | 461 | |
| 462 | 462 | /** |
@@ -465,229 +465,229 @@ discard block |
||
| 465 | 465 | * @access public |
| 466 | 466 | */ |
| 467 | 467 | class TSax3 { |
| 468 | - /** |
|
| 469 | - * Instance of concrete subclass of TSax3_StateParser |
|
| 470 | - * @var TSax3_StateParser |
|
| 471 | - * @access private |
|
| 472 | - */ |
|
| 473 | - private $state_parser; |
|
| 474 | - |
|
| 475 | - /** |
|
| 476 | - * Constructs TSax3 selecting concrete StateParser subclass |
|
| 477 | - * depending on PHP version being used as well as setting the default |
|
| 478 | - * NullHandler for all callbacks<br /> |
|
| 479 | - * <b>Example:</b> |
|
| 480 | - * <pre> |
|
| 481 | - * $myHandler = & new MyHandler(); |
|
| 482 | - * $parser = new TSax3(); |
|
| 483 | - * $parser->set_object($myHandler); |
|
| 484 | - * $parser->set_option('XML_OPTION_CASE_FOLDING'); |
|
| 485 | - * $parser->set_element_handler('myOpenHandler','myCloseHandler'); |
|
| 486 | - * $parser->set_data_handler('myDataHandler'); |
|
| 487 | - * $parser->parser($xml); |
|
| 488 | - * </pre> |
|
| 489 | - * @access public |
|
| 490 | - */ |
|
| 491 | - function __construct() { |
|
| 492 | - if (version_compare(phpversion(), '4.3', 'ge')) { |
|
| 493 | - $this->state_parser = new TSax3_StateParser_Gtet430($this); |
|
| 494 | - } else { |
|
| 495 | - $this->state_parser = new TSax3_StateParser_Lt430($this); |
|
| 496 | - } |
|
| 497 | - $nullhandler = new TSax3_NullHandler(); |
|
| 498 | - $this->set_object($nullhandler); |
|
| 499 | - $this->set_element_handler('DoNothing', 'DoNothing'); |
|
| 500 | - $this->set_data_handler('DoNothing'); |
|
| 501 | - $this->set_pi_handler('DoNothing'); |
|
| 502 | - $this->set_jasp_handler('DoNothing'); |
|
| 503 | - $this->set_escape_handler('DoNothing'); |
|
| 504 | - } |
|
| 505 | - |
|
| 506 | - /** |
|
| 507 | - * Sets the user defined handler object. Returns a PEAR Error |
|
| 508 | - * if supplied argument is not an object. |
|
| 509 | - * @param object handler object containing SAX callback methods |
|
| 510 | - * @access public |
|
| 511 | - * @return mixed |
|
| 512 | - */ |
|
| 513 | - function set_object(&$object) { |
|
| 514 | - if ( is_object($object) ) { |
|
| 515 | - $this->state_parser->handler_default =& $object; |
|
| 516 | - return true; |
|
| 517 | - } else { |
|
| 518 | - require_once('PEAR.php'); |
|
| 519 | - PEAR::raiseError('TSax3::set_object requires '. |
|
| 520 | - 'an object instance'); |
|
| 521 | - } |
|
| 522 | - } |
|
| 523 | - |
|
| 524 | - /** |
|
| 525 | - * Sets a parser option. By default all options are switched off. |
|
| 526 | - * Returns a PEAR Error if option is invalid<br /> |
|
| 527 | - * <b>Available options:</b> |
|
| 528 | - * <ul> |
|
| 529 | - * <li>XML_OPTION_TRIM_DATA_NODES: trim whitespace off the beginning |
|
| 530 | - * and end of data passed to the data handler</li> |
|
| 531 | - * <li>XML_OPTION_LINEFEED_BREAK: linefeeds result in additional data |
|
| 532 | - * handler calls</li> |
|
| 533 | - * <li>XML_OPTION_TAB_BREAK: tabs result in additional data handler |
|
| 534 | - * calls</li> |
|
| 535 | - * <li>XML_OPTION_ENTITIES_UNPARSED: XML entities are returned as |
|
| 536 | - * seperate data handler calls in unparsed form</li> |
|
| 537 | - * <li>XML_OPTION_ENTITIES_PARSED: (PHP 4.3.0+ only) XML entities are |
|
| 538 | - * returned as seperate data handler calls and are parsed with |
|
| 539 | - * PHP's html_entity_decode() function</li> |
|
| 540 | - * <li>XML_OPTION_STRIP_ESCAPES: strips out the -- -- comment markers |
|
| 541 | - * or CDATA markup inside an XML escape, if found.</li> |
|
| 542 | - * </ul> |
|
| 543 | - * To get HTMLSax to behave in the same way as the native PHP SAX parser, |
|
| 544 | - * using it's default state, you need to switch on XML_OPTION_LINEFEED_BREAK, |
|
| 545 | - * XML_OPTION_ENTITIES_PARSED and XML_OPTION_CASE_FOLDING |
|
| 546 | - * @param string name of parser option |
|
| 547 | - * @param int (optional) 1 to switch on, 0 for off |
|
| 548 | - * @access public |
|
| 549 | - * @return boolean |
|
| 550 | - */ |
|
| 551 | - function set_option($name, $value=1) { |
|
| 552 | - if ( array_key_exists($name,$this->state_parser->parser_options) ) { |
|
| 553 | - $this->state_parser->parser_options[$name] = $value; |
|
| 554 | - return true; |
|
| 555 | - } else { |
|
| 556 | - require_once('PEAR.php'); |
|
| 557 | - PEAR::raiseError('TSax3::set_option('.$name.') illegal'); |
|
| 558 | - } |
|
| 559 | - } |
|
| 560 | - |
|
| 561 | - /** |
|
| 562 | - * Sets the data handler method which deals with the contents of XML |
|
| 563 | - * elements.<br /> |
|
| 564 | - * The handler method must accept two arguments, the first being an |
|
| 565 | - * instance of TSax3 and the second being the contents of an |
|
| 566 | - * XML element e.g. |
|
| 567 | - * <pre> |
|
| 568 | - * function myDataHander(& $parser,$data){} |
|
| 569 | - * </pre> |
|
| 570 | - * @param string name of method |
|
| 571 | - * @access public |
|
| 572 | - * @return void |
|
| 573 | - * @see set_object |
|
| 574 | - */ |
|
| 575 | - function set_data_handler($data_method) { |
|
| 576 | - $this->state_parser->handler_object_data =& $this->state_parser->handler_default; |
|
| 577 | - $this->state_parser->handler_method_data = $data_method; |
|
| 578 | - } |
|
| 579 | - |
|
| 580 | - /** |
|
| 581 | - * Sets the open and close tag handlers |
|
| 582 | - * <br />The open handler method must accept three arguments; the parser, |
|
| 583 | - * the tag name and an array of attributes e.g. |
|
| 584 | - * <pre> |
|
| 585 | - * function myOpenHander(& $parser,$tagname,$attrs=array()){} |
|
| 586 | - * </pre> |
|
| 587 | - * The close handler method must accept two arguments; the parser and |
|
| 588 | - * the tag name e.g. |
|
| 589 | - * <pre> |
|
| 590 | - * function myCloseHander(& $parser,$tagname){} |
|
| 591 | - * </pre> |
|
| 592 | - * @param string name of open method |
|
| 593 | - * @param string name of close method |
|
| 594 | - * @access public |
|
| 595 | - * @return void |
|
| 596 | - * @see set_object |
|
| 597 | - */ |
|
| 598 | - function set_element_handler($opening_method, $closing_method) { |
|
| 599 | - $this->state_parser->handler_object_element =& $this->state_parser->handler_default; |
|
| 600 | - $this->state_parser->handler_method_opening = $opening_method; |
|
| 601 | - $this->state_parser->handler_method_closing = $closing_method; |
|
| 602 | - } |
|
| 603 | - |
|
| 604 | - /** |
|
| 605 | - * Sets the processing instruction handler method e.g. for PHP open |
|
| 606 | - * and close tags<br /> |
|
| 607 | - * The handler method must accept three arguments; the parser, the |
|
| 608 | - * PI target and data inside the PI |
|
| 609 | - * <pre> |
|
| 610 | - * function myPIHander(& $parser,$target, $data){} |
|
| 611 | - * </pre> |
|
| 612 | - * @param string name of method |
|
| 613 | - * @access public |
|
| 614 | - * @return void |
|
| 615 | - * @see set_object |
|
| 616 | - */ |
|
| 617 | - function set_pi_handler($pi_method) { |
|
| 618 | - $this->state_parser->handler_object_pi =& $this->state_parser->handler_default; |
|
| 619 | - $this->state_parser->handler_method_pi = $pi_method; |
|
| 620 | - } |
|
| 621 | - |
|
| 622 | - /** |
|
| 623 | - * Sets the XML escape handler method e.g. for comments and doctype |
|
| 624 | - * declarations<br /> |
|
| 625 | - * The handler method must accept two arguments; the parser and the |
|
| 626 | - * contents of the escaped section |
|
| 627 | - * <pre> |
|
| 628 | - * function myEscapeHander(& $parser, $data){} |
|
| 629 | - * </pre> |
|
| 630 | - * @param string name of method |
|
| 631 | - * @access public |
|
| 632 | - * @return void |
|
| 633 | - * @see set_object |
|
| 634 | - */ |
|
| 635 | - function set_escape_handler($escape_method) { |
|
| 636 | - $this->state_parser->handler_object_escape =& $this->state_parser->handler_default; |
|
| 637 | - $this->state_parser->handler_method_escape = $escape_method; |
|
| 638 | - } |
|
| 639 | - |
|
| 640 | - /** |
|
| 641 | - * Sets the JSP/ASP markup handler<br /> |
|
| 642 | - * The handler method must accept two arguments; the parser and |
|
| 643 | - * body of the JASP tag |
|
| 644 | - * <pre> |
|
| 645 | - * function myJaspHander(& $parser, $data){} |
|
| 646 | - * </pre> |
|
| 647 | - * @param string name of method |
|
| 648 | - * @access public |
|
| 649 | - * @return void |
|
| 650 | - * @see set_object |
|
| 651 | - */ |
|
| 652 | - function set_jasp_handler ($jasp_method) { |
|
| 653 | - $this->state_parser->handler_object_jasp =& $this->state_parser->handler_default; |
|
| 654 | - $this->state_parser->handler_method_jasp = $jasp_method; |
|
| 655 | - } |
|
| 656 | - |
|
| 657 | - /** |
|
| 658 | - * Returns the current string position of the "cursor" inside the XML |
|
| 659 | - * document |
|
| 660 | - * <br />Intended for use from within a user defined handler called |
|
| 661 | - * via the $parser reference e.g. |
|
| 662 | - * <pre> |
|
| 663 | - * function myDataHandler(& $parser,$data) { |
|
| 664 | - * echo( 'Current position: '.$parser->get_current_position() ); |
|
| 665 | - * } |
|
| 666 | - * </pre> |
|
| 667 | - * @access public |
|
| 668 | - * @return int |
|
| 669 | - * @see get_length |
|
| 670 | - */ |
|
| 671 | - function get_current_position() { |
|
| 672 | - return $this->state_parser->position; |
|
| 673 | - } |
|
| 674 | - |
|
| 675 | - /** |
|
| 676 | - * Returns the string length of the XML document being parsed |
|
| 677 | - * @access public |
|
| 678 | - * @return int |
|
| 679 | - */ |
|
| 680 | - function get_length() { |
|
| 681 | - return $this->state_parser->length; |
|
| 682 | - } |
|
| 683 | - |
|
| 684 | - /** |
|
| 685 | - * Start parsing some XML |
|
| 686 | - * @param string XML document |
|
| 687 | - * @access public |
|
| 688 | - * @return void |
|
| 689 | - */ |
|
| 690 | - function parse($data) { |
|
| 691 | - $this->state_parser->parse($data); |
|
| 692 | - } |
|
| 468 | + /** |
|
| 469 | + * Instance of concrete subclass of TSax3_StateParser |
|
| 470 | + * @var TSax3_StateParser |
|
| 471 | + * @access private |
|
| 472 | + */ |
|
| 473 | + private $state_parser; |
|
| 474 | + |
|
| 475 | + /** |
|
| 476 | + * Constructs TSax3 selecting concrete StateParser subclass |
|
| 477 | + * depending on PHP version being used as well as setting the default |
|
| 478 | + * NullHandler for all callbacks<br /> |
|
| 479 | + * <b>Example:</b> |
|
| 480 | + * <pre> |
|
| 481 | + * $myHandler = & new MyHandler(); |
|
| 482 | + * $parser = new TSax3(); |
|
| 483 | + * $parser->set_object($myHandler); |
|
| 484 | + * $parser->set_option('XML_OPTION_CASE_FOLDING'); |
|
| 485 | + * $parser->set_element_handler('myOpenHandler','myCloseHandler'); |
|
| 486 | + * $parser->set_data_handler('myDataHandler'); |
|
| 487 | + * $parser->parser($xml); |
|
| 488 | + * </pre> |
|
| 489 | + * @access public |
|
| 490 | + */ |
|
| 491 | + function __construct() { |
|
| 492 | + if (version_compare(phpversion(), '4.3', 'ge')) { |
|
| 493 | + $this->state_parser = new TSax3_StateParser_Gtet430($this); |
|
| 494 | + } else { |
|
| 495 | + $this->state_parser = new TSax3_StateParser_Lt430($this); |
|
| 496 | + } |
|
| 497 | + $nullhandler = new TSax3_NullHandler(); |
|
| 498 | + $this->set_object($nullhandler); |
|
| 499 | + $this->set_element_handler('DoNothing', 'DoNothing'); |
|
| 500 | + $this->set_data_handler('DoNothing'); |
|
| 501 | + $this->set_pi_handler('DoNothing'); |
|
| 502 | + $this->set_jasp_handler('DoNothing'); |
|
| 503 | + $this->set_escape_handler('DoNothing'); |
|
| 504 | + } |
|
| 505 | + |
|
| 506 | + /** |
|
| 507 | + * Sets the user defined handler object. Returns a PEAR Error |
|
| 508 | + * if supplied argument is not an object. |
|
| 509 | + * @param object handler object containing SAX callback methods |
|
| 510 | + * @access public |
|
| 511 | + * @return mixed |
|
| 512 | + */ |
|
| 513 | + function set_object(&$object) { |
|
| 514 | + if ( is_object($object) ) { |
|
| 515 | + $this->state_parser->handler_default =& $object; |
|
| 516 | + return true; |
|
| 517 | + } else { |
|
| 518 | + require_once('PEAR.php'); |
|
| 519 | + PEAR::raiseError('TSax3::set_object requires '. |
|
| 520 | + 'an object instance'); |
|
| 521 | + } |
|
| 522 | + } |
|
| 523 | + |
|
| 524 | + /** |
|
| 525 | + * Sets a parser option. By default all options are switched off. |
|
| 526 | + * Returns a PEAR Error if option is invalid<br /> |
|
| 527 | + * <b>Available options:</b> |
|
| 528 | + * <ul> |
|
| 529 | + * <li>XML_OPTION_TRIM_DATA_NODES: trim whitespace off the beginning |
|
| 530 | + * and end of data passed to the data handler</li> |
|
| 531 | + * <li>XML_OPTION_LINEFEED_BREAK: linefeeds result in additional data |
|
| 532 | + * handler calls</li> |
|
| 533 | + * <li>XML_OPTION_TAB_BREAK: tabs result in additional data handler |
|
| 534 | + * calls</li> |
|
| 535 | + * <li>XML_OPTION_ENTITIES_UNPARSED: XML entities are returned as |
|
| 536 | + * seperate data handler calls in unparsed form</li> |
|
| 537 | + * <li>XML_OPTION_ENTITIES_PARSED: (PHP 4.3.0+ only) XML entities are |
|
| 538 | + * returned as seperate data handler calls and are parsed with |
|
| 539 | + * PHP's html_entity_decode() function</li> |
|
| 540 | + * <li>XML_OPTION_STRIP_ESCAPES: strips out the -- -- comment markers |
|
| 541 | + * or CDATA markup inside an XML escape, if found.</li> |
|
| 542 | + * </ul> |
|
| 543 | + * To get HTMLSax to behave in the same way as the native PHP SAX parser, |
|
| 544 | + * using it's default state, you need to switch on XML_OPTION_LINEFEED_BREAK, |
|
| 545 | + * XML_OPTION_ENTITIES_PARSED and XML_OPTION_CASE_FOLDING |
|
| 546 | + * @param string name of parser option |
|
| 547 | + * @param int (optional) 1 to switch on, 0 for off |
|
| 548 | + * @access public |
|
| 549 | + * @return boolean |
|
| 550 | + */ |
|
| 551 | + function set_option($name, $value=1) { |
|
| 552 | + if ( array_key_exists($name,$this->state_parser->parser_options) ) { |
|
| 553 | + $this->state_parser->parser_options[$name] = $value; |
|
| 554 | + return true; |
|
| 555 | + } else { |
|
| 556 | + require_once('PEAR.php'); |
|
| 557 | + PEAR::raiseError('TSax3::set_option('.$name.') illegal'); |
|
| 558 | + } |
|
| 559 | + } |
|
| 560 | + |
|
| 561 | + /** |
|
| 562 | + * Sets the data handler method which deals with the contents of XML |
|
| 563 | + * elements.<br /> |
|
| 564 | + * The handler method must accept two arguments, the first being an |
|
| 565 | + * instance of TSax3 and the second being the contents of an |
|
| 566 | + * XML element e.g. |
|
| 567 | + * <pre> |
|
| 568 | + * function myDataHander(& $parser,$data){} |
|
| 569 | + * </pre> |
|
| 570 | + * @param string name of method |
|
| 571 | + * @access public |
|
| 572 | + * @return void |
|
| 573 | + * @see set_object |
|
| 574 | + */ |
|
| 575 | + function set_data_handler($data_method) { |
|
| 576 | + $this->state_parser->handler_object_data =& $this->state_parser->handler_default; |
|
| 577 | + $this->state_parser->handler_method_data = $data_method; |
|
| 578 | + } |
|
| 579 | + |
|
| 580 | + /** |
|
| 581 | + * Sets the open and close tag handlers |
|
| 582 | + * <br />The open handler method must accept three arguments; the parser, |
|
| 583 | + * the tag name and an array of attributes e.g. |
|
| 584 | + * <pre> |
|
| 585 | + * function myOpenHander(& $parser,$tagname,$attrs=array()){} |
|
| 586 | + * </pre> |
|
| 587 | + * The close handler method must accept two arguments; the parser and |
|
| 588 | + * the tag name e.g. |
|
| 589 | + * <pre> |
|
| 590 | + * function myCloseHander(& $parser,$tagname){} |
|
| 591 | + * </pre> |
|
| 592 | + * @param string name of open method |
|
| 593 | + * @param string name of close method |
|
| 594 | + * @access public |
|
| 595 | + * @return void |
|
| 596 | + * @see set_object |
|
| 597 | + */ |
|
| 598 | + function set_element_handler($opening_method, $closing_method) { |
|
| 599 | + $this->state_parser->handler_object_element =& $this->state_parser->handler_default; |
|
| 600 | + $this->state_parser->handler_method_opening = $opening_method; |
|
| 601 | + $this->state_parser->handler_method_closing = $closing_method; |
|
| 602 | + } |
|
| 603 | + |
|
| 604 | + /** |
|
| 605 | + * Sets the processing instruction handler method e.g. for PHP open |
|
| 606 | + * and close tags<br /> |
|
| 607 | + * The handler method must accept three arguments; the parser, the |
|
| 608 | + * PI target and data inside the PI |
|
| 609 | + * <pre> |
|
| 610 | + * function myPIHander(& $parser,$target, $data){} |
|
| 611 | + * </pre> |
|
| 612 | + * @param string name of method |
|
| 613 | + * @access public |
|
| 614 | + * @return void |
|
| 615 | + * @see set_object |
|
| 616 | + */ |
|
| 617 | + function set_pi_handler($pi_method) { |
|
| 618 | + $this->state_parser->handler_object_pi =& $this->state_parser->handler_default; |
|
| 619 | + $this->state_parser->handler_method_pi = $pi_method; |
|
| 620 | + } |
|
| 621 | + |
|
| 622 | + /** |
|
| 623 | + * Sets the XML escape handler method e.g. for comments and doctype |
|
| 624 | + * declarations<br /> |
|
| 625 | + * The handler method must accept two arguments; the parser and the |
|
| 626 | + * contents of the escaped section |
|
| 627 | + * <pre> |
|
| 628 | + * function myEscapeHander(& $parser, $data){} |
|
| 629 | + * </pre> |
|
| 630 | + * @param string name of method |
|
| 631 | + * @access public |
|
| 632 | + * @return void |
|
| 633 | + * @see set_object |
|
| 634 | + */ |
|
| 635 | + function set_escape_handler($escape_method) { |
|
| 636 | + $this->state_parser->handler_object_escape =& $this->state_parser->handler_default; |
|
| 637 | + $this->state_parser->handler_method_escape = $escape_method; |
|
| 638 | + } |
|
| 639 | + |
|
| 640 | + /** |
|
| 641 | + * Sets the JSP/ASP markup handler<br /> |
|
| 642 | + * The handler method must accept two arguments; the parser and |
|
| 643 | + * body of the JASP tag |
|
| 644 | + * <pre> |
|
| 645 | + * function myJaspHander(& $parser, $data){} |
|
| 646 | + * </pre> |
|
| 647 | + * @param string name of method |
|
| 648 | + * @access public |
|
| 649 | + * @return void |
|
| 650 | + * @see set_object |
|
| 651 | + */ |
|
| 652 | + function set_jasp_handler ($jasp_method) { |
|
| 653 | + $this->state_parser->handler_object_jasp =& $this->state_parser->handler_default; |
|
| 654 | + $this->state_parser->handler_method_jasp = $jasp_method; |
|
| 655 | + } |
|
| 656 | + |
|
| 657 | + /** |
|
| 658 | + * Returns the current string position of the "cursor" inside the XML |
|
| 659 | + * document |
|
| 660 | + * <br />Intended for use from within a user defined handler called |
|
| 661 | + * via the $parser reference e.g. |
|
| 662 | + * <pre> |
|
| 663 | + * function myDataHandler(& $parser,$data) { |
|
| 664 | + * echo( 'Current position: '.$parser->get_current_position() ); |
|
| 665 | + * } |
|
| 666 | + * </pre> |
|
| 667 | + * @access public |
|
| 668 | + * @return int |
|
| 669 | + * @see get_length |
|
| 670 | + */ |
|
| 671 | + function get_current_position() { |
|
| 672 | + return $this->state_parser->position; |
|
| 673 | + } |
|
| 674 | + |
|
| 675 | + /** |
|
| 676 | + * Returns the string length of the XML document being parsed |
|
| 677 | + * @access public |
|
| 678 | + * @return int |
|
| 679 | + */ |
|
| 680 | + function get_length() { |
|
| 681 | + return $this->state_parser->length; |
|
| 682 | + } |
|
| 683 | + |
|
| 684 | + /** |
|
| 685 | + * Start parsing some XML |
|
| 686 | + * @param string XML document |
|
| 687 | + * @access public |
|
| 688 | + * @return void |
|
| 689 | + */ |
|
| 690 | + function parse($data) { |
|
| 691 | + $this->state_parser->parse($data); |
|
| 692 | + } |
|
| 693 | 693 | } |
@@ -57,622 +57,622 @@ |
||
| 57 | 57 | */ |
| 58 | 58 | class TSafeHtmlParser |
| 59 | 59 | { |
| 60 | - /** |
|
| 61 | - * Storage for resulting HTML output |
|
| 62 | - * |
|
| 63 | - * @var string |
|
| 64 | - * @access private |
|
| 65 | - */ |
|
| 66 | - private $_xhtml = ''; |
|
| 67 | - |
|
| 68 | - /** |
|
| 69 | - * Array of counters for each tag |
|
| 70 | - * |
|
| 71 | - * @var array |
|
| 72 | - * @access private |
|
| 73 | - */ |
|
| 74 | - private $_counter = array(); |
|
| 75 | - |
|
| 76 | - /** |
|
| 77 | - * Stack of unclosed tags |
|
| 78 | - * |
|
| 79 | - * @var array |
|
| 80 | - * @access private |
|
| 81 | - */ |
|
| 82 | - private $_stack = array(); |
|
| 83 | - |
|
| 84 | - /** |
|
| 85 | - * Array of counters for tags that must be deleted with all content |
|
| 86 | - * |
|
| 87 | - * @var array |
|
| 88 | - * @access private |
|
| 89 | - */ |
|
| 90 | - private $_dcCounter = array(); |
|
| 91 | - |
|
| 92 | - /** |
|
| 93 | - * Stack of unclosed tags that must be deleted with all content |
|
| 94 | - * |
|
| 95 | - * @var array |
|
| 96 | - * @access private |
|
| 97 | - */ |
|
| 98 | - private $_dcStack = array(); |
|
| 99 | - |
|
| 100 | - /** |
|
| 101 | - * Stores level of list (ol/ul) nesting |
|
| 102 | - * |
|
| 103 | - * @var int |
|
| 104 | - * @access private |
|
| 105 | - */ |
|
| 106 | - private $_listScope = 0; |
|
| 107 | - |
|
| 108 | - /** |
|
| 109 | - * Stack of unclosed list tags |
|
| 110 | - * |
|
| 111 | - * @var array |
|
| 112 | - * @access private |
|
| 113 | - */ |
|
| 114 | - private $_liStack = array(); |
|
| 115 | - |
|
| 116 | - /** |
|
| 117 | - * Array of prepared regular expressions for protocols (schemas) matching |
|
| 118 | - * |
|
| 119 | - * @var array |
|
| 120 | - * @access private |
|
| 121 | - */ |
|
| 122 | - private $_protoRegexps = array(); |
|
| 123 | - |
|
| 124 | - /** |
|
| 125 | - * Array of prepared regular expressions for CSS matching |
|
| 126 | - * |
|
| 127 | - * @var array |
|
| 128 | - * @access private |
|
| 129 | - */ |
|
| 130 | - private $_cssRegexps = array(); |
|
| 131 | - |
|
| 132 | - /** |
|
| 133 | - * List of single tags ("<tag />") |
|
| 134 | - * |
|
| 135 | - * @var array |
|
| 136 | - * @access public |
|
| 137 | - */ |
|
| 138 | - public $singleTags = array('area', 'br', 'img', 'input', 'hr', 'wbr', ); |
|
| 139 | - |
|
| 140 | - /** |
|
| 141 | - * List of dangerous tags (such tags will be deleted) |
|
| 142 | - * |
|
| 143 | - * @var array |
|
| 144 | - * @access public |
|
| 145 | - */ |
|
| 146 | - public $deleteTags = array( |
|
| 147 | - 'applet', 'base', 'basefont', 'bgsound', 'blink', 'body', |
|
| 148 | - 'embed', 'frame', 'frameset', 'head', 'html', 'ilayer', |
|
| 149 | - 'iframe', 'layer', 'link', 'meta', 'object', 'style', |
|
| 150 | - 'title', 'script', |
|
| 151 | - ); |
|
| 152 | - |
|
| 153 | - /** |
|
| 154 | - * List of dangerous tags (such tags will be deleted, and all content |
|
| 155 | - * inside this tags will be also removed) |
|
| 156 | - * |
|
| 157 | - * @var array |
|
| 158 | - * @access public |
|
| 159 | - */ |
|
| 160 | - public $deleteTagsContent = array('script', 'style', 'title', 'xml', ); |
|
| 161 | - |
|
| 162 | - /** |
|
| 163 | - * Type of protocols filtering ('white' or 'black') |
|
| 164 | - * |
|
| 165 | - * @var string |
|
| 166 | - * @access public |
|
| 167 | - */ |
|
| 168 | - public $protocolFiltering = 'white'; |
|
| 169 | - |
|
| 170 | - /** |
|
| 171 | - * List of "dangerous" protocols (used for blacklist-filtering) |
|
| 172 | - * |
|
| 173 | - * @var array |
|
| 174 | - * @access public |
|
| 175 | - */ |
|
| 176 | - public $blackProtocols = array( |
|
| 177 | - 'about', 'chrome', 'data', 'disk', 'hcp', |
|
| 178 | - 'help', 'javascript', 'livescript', 'lynxcgi', 'lynxexec', |
|
| 179 | - 'ms-help', 'ms-its', 'mhtml', 'mocha', 'opera', |
|
| 180 | - 'res', 'resource', 'shell', 'vbscript', 'view-source', |
|
| 181 | - 'vnd.ms.radio', 'wysiwyg', |
|
| 182 | - ); |
|
| 183 | - |
|
| 184 | - /** |
|
| 185 | - * List of "safe" protocols (used for whitelist-filtering) |
|
| 186 | - * |
|
| 187 | - * @var array |
|
| 188 | - * @access public |
|
| 189 | - */ |
|
| 190 | - public $whiteProtocols = array( |
|
| 191 | - 'ed2k', 'file', 'ftp', 'gopher', 'http', 'https', |
|
| 192 | - 'irc', 'mailto', 'news', 'nntp', 'telnet', 'webcal', |
|
| 193 | - 'xmpp', 'callto', |
|
| 194 | - ); |
|
| 195 | - |
|
| 196 | - /** |
|
| 197 | - * List of attributes that can contain protocols |
|
| 198 | - * |
|
| 199 | - * @var array |
|
| 200 | - * @access public |
|
| 201 | - */ |
|
| 202 | - public $protocolAttributes = array( |
|
| 203 | - 'action', 'background', 'codebase', 'dynsrc', 'href', 'lowsrc', 'src', |
|
| 204 | - ); |
|
| 205 | - |
|
| 206 | - /** |
|
| 207 | - * List of dangerous CSS keywords |
|
| 208 | - * |
|
| 209 | - * Whole style="" attribute will be removed, if parser will find one of |
|
| 210 | - * these keywords |
|
| 211 | - * |
|
| 212 | - * @var array |
|
| 213 | - * @access public |
|
| 214 | - */ |
|
| 215 | - public $cssKeywords = array( |
|
| 216 | - 'absolute', 'behavior', 'behaviour', 'content', 'expression', |
|
| 217 | - 'fixed', 'include-source', 'moz-binding', |
|
| 218 | - ); |
|
| 219 | - |
|
| 220 | - /** |
|
| 221 | - * List of tags that can have no "closing tag" |
|
| 222 | - * |
|
| 223 | - * @var array |
|
| 224 | - * @access public |
|
| 225 | - * @deprecated XHTML does not allow such tags |
|
| 226 | - */ |
|
| 227 | - public $noClose = array(); |
|
| 228 | - |
|
| 229 | - /** |
|
| 230 | - * List of block-level tags that terminates paragraph |
|
| 231 | - * |
|
| 232 | - * Paragraph will be closed when this tags opened |
|
| 233 | - * |
|
| 234 | - * @var array |
|
| 235 | - * @access public |
|
| 236 | - */ |
|
| 237 | - public $closeParagraph = array( |
|
| 238 | - 'address', 'blockquote', 'center', 'dd', 'dir', 'div', |
|
| 239 | - 'dl', 'dt', 'h1', 'h2', 'h3', 'h4', |
|
| 240 | - 'h5', 'h6', 'hr', 'isindex', 'listing', 'marquee', |
|
| 241 | - 'menu', 'multicol', 'ol', 'p', 'plaintext', 'pre', |
|
| 242 | - 'table', 'ul', 'xmp', |
|
| 243 | - ); |
|
| 244 | - |
|
| 245 | - /** |
|
| 246 | - * List of table tags, all table tags outside a table will be removed |
|
| 247 | - * |
|
| 248 | - * @var array |
|
| 249 | - * @access public |
|
| 250 | - */ |
|
| 251 | - public $tableTags = array( |
|
| 252 | - 'caption', 'col', 'colgroup', 'tbody', 'td', 'tfoot', 'th', |
|
| 253 | - 'thead', 'tr', |
|
| 254 | - ); |
|
| 255 | - |
|
| 256 | - /** |
|
| 257 | - * List of list tags |
|
| 258 | - * |
|
| 259 | - * @var array |
|
| 260 | - * @access public |
|
| 261 | - */ |
|
| 262 | - public $listTags = array('dir', 'menu', 'ol', 'ul', 'dl', ); |
|
| 263 | - |
|
| 264 | - /** |
|
| 265 | - * List of dangerous attributes |
|
| 266 | - * |
|
| 267 | - * @var array |
|
| 268 | - * @access public |
|
| 269 | - */ |
|
| 270 | - public $attributes = array('dynsrc'); |
|
| 271 | - //public $attributes = array('dynsrc', 'id', 'name', ); //id and name are dangerous? |
|
| 272 | - |
|
| 273 | - /** |
|
| 274 | - * List of allowed "namespaced" attributes |
|
| 275 | - * |
|
| 276 | - * @var array |
|
| 277 | - * @access public |
|
| 278 | - */ |
|
| 279 | - public $attributesNS = array('xml:lang', ); |
|
| 280 | - |
|
| 281 | - /** |
|
| 282 | - * Constructs class |
|
| 283 | - * |
|
| 284 | - * @access public |
|
| 285 | - */ |
|
| 286 | - public function __construct() |
|
| 287 | - { |
|
| 288 | - //making regular expressions based on Proto & CSS arrays |
|
| 289 | - foreach ($this->blackProtocols as $proto) { |
|
| 290 | - $preg = "/[\s\x01-\x1F]*"; |
|
| 291 | - for ($i=0; $i<strlen($proto); $i++) { |
|
| 292 | - $preg .= $proto{$i} . "[\s\x01-\x1F]*"; |
|
| 293 | - } |
|
| 294 | - $preg .= ":/i"; |
|
| 295 | - $this->_protoRegexps[] = $preg; |
|
| 296 | - } |
|
| 297 | - |
|
| 298 | - foreach ($this->cssKeywords as $css) { |
|
| 299 | - $this->_cssRegexps[] = '/' . $css . '/i'; |
|
| 300 | - } |
|
| 301 | - return true; |
|
| 302 | - } |
|
| 303 | - |
|
| 304 | - /** |
|
| 305 | - * Handles the writing of attributes - called from $this->_openHandler() |
|
| 306 | - * |
|
| 307 | - * @param array $attrs array of attributes $name => $value |
|
| 308 | - * @return boolean |
|
| 309 | - * @access private |
|
| 310 | - */ |
|
| 311 | - private function _writeAttrs ($attrs) |
|
| 312 | - { |
|
| 313 | - if (is_array($attrs)) { |
|
| 314 | - foreach ($attrs as $name => $value) { |
|
| 315 | - |
|
| 316 | - $name = strtolower($name); |
|
| 317 | - |
|
| 318 | - if (strpos($name, 'on') === 0) { |
|
| 319 | - continue; |
|
| 320 | - } |
|
| 321 | - if (strpos($name, 'data') === 0) { |
|
| 322 | - continue; |
|
| 323 | - } |
|
| 324 | - if (in_array($name, $this->attributes)) { |
|
| 325 | - continue; |
|
| 326 | - } |
|
| 327 | - if (!preg_match("/^[a-z0-9]+$/i", $name)) { |
|
| 328 | - if (!in_array($name, $this->attributesNS)) |
|
| 329 | - { |
|
| 330 | - continue; |
|
| 331 | - } |
|
| 332 | - } |
|
| 333 | - |
|
| 334 | - if (($value === TRUE) || (is_null($value))) { |
|
| 335 | - $value = $name; |
|
| 336 | - } |
|
| 337 | - |
|
| 338 | - if ($name == 'style') { |
|
| 339 | - |
|
| 340 | - // removes insignificant backslahes |
|
| 341 | - $value = str_replace("\\", '', $value); |
|
| 342 | - |
|
| 343 | - // removes CSS comments |
|
| 344 | - while (1) |
|
| 345 | - { |
|
| 346 | - $_value = preg_replace("!/\*.*?\*/!s", '', $value); |
|
| 347 | - if ($_value == $value) break; |
|
| 348 | - $value = $_value; |
|
| 349 | - } |
|
| 350 | - |
|
| 351 | - // replace all & to & |
|
| 352 | - $value = str_replace('&', '&', $value); |
|
| 353 | - $value = str_replace('&', '&', $value); |
|
| 354 | - |
|
| 355 | - foreach ($this->_cssRegexps as $css) { |
|
| 356 | - if (preg_match($css, $value)) { |
|
| 357 | - continue 2; |
|
| 358 | - } |
|
| 359 | - } |
|
| 360 | - foreach ($this->_protoRegexps as $proto) { |
|
| 361 | - if (preg_match($proto, $value)) { |
|
| 362 | - continue 2; |
|
| 363 | - } |
|
| 364 | - } |
|
| 365 | - } |
|
| 366 | - |
|
| 367 | - $tempval = preg_replace_callback( |
|
| 368 | - '/&#(\d+);?/m', |
|
| 369 | - function ($matches) { |
|
| 370 | - return chr($matches[0]); |
|
| 371 | - }, |
|
| 372 | - $value |
|
| 373 | - ); //"' |
|
| 374 | - |
|
| 375 | - $tempval = preg_replace_callback( |
|
| 376 | - '/&#x([0-9a-f]+);?/mi', |
|
| 377 | - function ($matches) { |
|
| 378 | - return chr(hexdec($matches[0])); |
|
| 379 | - }, |
|
| 380 | - $tempval |
|
| 381 | - ); |
|
| 382 | - |
|
| 383 | - if ((in_array($name, $this->protocolAttributes)) && |
|
| 384 | - (strpos($tempval, ':') !== false)) |
|
| 385 | - { |
|
| 386 | - if ($this->protocolFiltering == 'black') { |
|
| 387 | - foreach ($this->_protoRegexps as $proto) { |
|
| 388 | - if (preg_match($proto, $tempval)) continue 2; |
|
| 389 | - } |
|
| 390 | - } else { |
|
| 391 | - $_tempval = explode(':', $tempval); |
|
| 392 | - $proto = $_tempval[0]; |
|
| 393 | - if (!in_array($proto, $this->whiteProtocols)) { |
|
| 394 | - continue; |
|
| 395 | - } |
|
| 396 | - } |
|
| 397 | - } |
|
| 398 | - |
|
| 399 | - $value = str_replace("\"", """, $value); |
|
| 400 | - $this->_xhtml .= ' ' . $name . '="' . $value . '"'; |
|
| 401 | - } |
|
| 402 | - } |
|
| 403 | - return true; |
|
| 404 | - } |
|
| 405 | - |
|
| 406 | - /** |
|
| 407 | - * Opening tag handler - called from HTMLSax |
|
| 408 | - * |
|
| 409 | - * @param object $parser HTML Parser |
|
| 410 | - * @param string $name tag name |
|
| 411 | - * @param array $attrs tag attributes |
|
| 412 | - * @return boolean |
|
| 413 | - * @access private |
|
| 414 | - */ |
|
| 415 | - public function _openHandler(&$parser, $name, $attrs) |
|
| 416 | - { |
|
| 417 | - $name = strtolower($name); |
|
| 418 | - |
|
| 419 | - if (in_array($name, $this->deleteTagsContent)) { |
|
| 420 | - array_push($this->_dcStack, $name); |
|
| 421 | - $this->_dcCounter[$name] = isset($this->_dcCounter[$name]) ? $this->_dcCounter[$name]+1 : 1; |
|
| 422 | - } |
|
| 423 | - if (count($this->_dcStack) != 0) { |
|
| 424 | - return true; |
|
| 425 | - } |
|
| 426 | - |
|
| 427 | - if (in_array($name, $this->deleteTags)) { |
|
| 428 | - return true; |
|
| 429 | - } |
|
| 430 | - |
|
| 431 | - if (!preg_match("/^[a-z0-9]+$/i", $name)) { |
|
| 432 | - if (preg_match("!(?:\@|://)!i", $name)) { |
|
| 433 | - $this->_xhtml .= '<' . $name . '>'; |
|
| 434 | - } |
|
| 435 | - return true; |
|
| 436 | - } |
|
| 437 | - |
|
| 438 | - if (in_array($name, $this->singleTags)) { |
|
| 439 | - $this->_xhtml .= '<' . $name; |
|
| 440 | - $this->_writeAttrs($attrs); |
|
| 441 | - $this->_xhtml .= ' />'; |
|
| 442 | - return true; |
|
| 443 | - } |
|
| 444 | - |
|
| 445 | - // TABLES: cannot open table elements when we are not inside table |
|
| 446 | - if ((isset($this->_counter['table'])) && ($this->_counter['table'] <= 0) |
|
| 447 | - && (in_array($name, $this->tableTags))) |
|
| 448 | - { |
|
| 449 | - return true; |
|
| 450 | - } |
|
| 451 | - |
|
| 452 | - // PARAGRAPHS: close paragraph when closeParagraph tags opening |
|
| 453 | - if ((in_array($name, $this->closeParagraph)) && (in_array('p', $this->_stack))) { |
|
| 454 | - $this->_closeHandler($parser, 'p'); |
|
| 455 | - } |
|
| 456 | - |
|
| 457 | - // LISTS: we should close <li> if <li> of the same level opening |
|
| 458 | - if ($name == 'li' && count($this->_liStack) && |
|
| 459 | - $this->_listScope == $this->_liStack[count($this->_liStack)-1]) |
|
| 460 | - { |
|
| 461 | - $this->_closeHandler($parser, 'li'); |
|
| 462 | - } |
|
| 463 | - |
|
| 464 | - // LISTS: we want to know on what nesting level of lists we are |
|
| 465 | - if (in_array($name, $this->listTags)) { |
|
| 466 | - $this->_listScope++; |
|
| 467 | - } |
|
| 468 | - if ($name == 'li') { |
|
| 469 | - array_push($this->_liStack, $this->_listScope); |
|
| 470 | - } |
|
| 471 | - |
|
| 472 | - $this->_xhtml .= '<' . $name; |
|
| 473 | - $this->_writeAttrs($attrs); |
|
| 474 | - $this->_xhtml .= '>'; |
|
| 475 | - array_push($this->_stack,$name); |
|
| 476 | - $this->_counter[$name] = isset($this->_counter[$name]) ? $this->_counter[$name]+1 : 1; |
|
| 477 | - return true; |
|
| 478 | - } |
|
| 479 | - |
|
| 480 | - /** |
|
| 481 | - * Closing tag handler - called from HTMLSax |
|
| 482 | - * |
|
| 483 | - * @param object $parsers HTML parser |
|
| 484 | - * @param string $name tag name |
|
| 485 | - * @return boolean |
|
| 486 | - * @access private |
|
| 487 | - */ |
|
| 488 | - public function _closeHandler(&$parser, $name) |
|
| 489 | - { |
|
| 490 | - |
|
| 491 | - $name = strtolower($name); |
|
| 492 | - |
|
| 493 | - if (isset($this->_dcCounter[$name]) && ($this->_dcCounter[$name] > 0) && |
|
| 494 | - (in_array($name, $this->deleteTagsContent))) |
|
| 495 | - { |
|
| 496 | - while ($name != ($tag = array_pop($this->_dcStack))) { |
|
| 497 | - $this->_dcCounter[$tag]--; |
|
| 498 | - } |
|
| 499 | - |
|
| 500 | - $this->_dcCounter[$name]--; |
|
| 501 | - } |
|
| 502 | - |
|
| 503 | - if (count($this->_dcStack) != 0) { |
|
| 504 | - return true; |
|
| 505 | - } |
|
| 506 | - |
|
| 507 | - if ((isset($this->_counter[$name])) && ($this->_counter[$name] > 0)) { |
|
| 508 | - while ($name != ($tag = array_pop($this->_stack))) { |
|
| 509 | - $this->_closeTag($tag); |
|
| 510 | - } |
|
| 511 | - |
|
| 512 | - $this->_closeTag($name); |
|
| 513 | - } |
|
| 514 | - return true; |
|
| 515 | - } |
|
| 516 | - |
|
| 517 | - /** |
|
| 518 | - * Closes tag |
|
| 519 | - * |
|
| 520 | - * @param string $tag tag name |
|
| 521 | - * @return boolean |
|
| 522 | - * @access private |
|
| 523 | - */ |
|
| 524 | - public function _closeTag($tag) |
|
| 525 | - { |
|
| 526 | - if (!in_array($tag, $this->noClose)) { |
|
| 527 | - $this->_xhtml .= '</' . $tag . '>'; |
|
| 528 | - } |
|
| 529 | - |
|
| 530 | - $this->_counter[$tag]--; |
|
| 531 | - |
|
| 532 | - if (in_array($tag, $this->listTags)) { |
|
| 533 | - $this->_listScope--; |
|
| 534 | - } |
|
| 535 | - |
|
| 536 | - if ($tag == 'li') { |
|
| 537 | - array_pop($this->_liStack); |
|
| 538 | - } |
|
| 539 | - return true; |
|
| 540 | - } |
|
| 541 | - |
|
| 542 | - /** |
|
| 543 | - * Character data handler - called from HTMLSax |
|
| 544 | - * |
|
| 545 | - * @param object $parser HTML parser |
|
| 546 | - * @param string $data textual data |
|
| 547 | - * @return boolean |
|
| 548 | - * @access private |
|
| 549 | - */ |
|
| 550 | - public function _dataHandler(&$parser, $data) |
|
| 551 | - { |
|
| 552 | - if (count($this->_dcStack) == 0) { |
|
| 553 | - $this->_xhtml .= $data; |
|
| 554 | - } |
|
| 555 | - return true; |
|
| 556 | - } |
|
| 557 | - |
|
| 558 | - /** |
|
| 559 | - * Escape handler - called from HTMLSax |
|
| 560 | - * |
|
| 561 | - * @param object $parser HTML parser |
|
| 562 | - * @param string $data comments or other type of data |
|
| 563 | - * @return boolean |
|
| 564 | - * @access private |
|
| 565 | - */ |
|
| 566 | - public function _escapeHandler(&$parser, $data) |
|
| 567 | - { |
|
| 568 | - return true; |
|
| 569 | - } |
|
| 570 | - |
|
| 571 | - /** |
|
| 572 | - * Returns the XHTML document |
|
| 573 | - * |
|
| 574 | - * @return string Processed (X)HTML document |
|
| 575 | - * @access public |
|
| 576 | - */ |
|
| 577 | - public function getXHTML () |
|
| 578 | - { |
|
| 579 | - while ($tag = array_pop($this->_stack)) { |
|
| 580 | - $this->_closeTag($tag); |
|
| 581 | - } |
|
| 582 | - |
|
| 583 | - return $this->_xhtml; |
|
| 584 | - } |
|
| 585 | - |
|
| 586 | - /** |
|
| 587 | - * Clears current document data |
|
| 588 | - * |
|
| 589 | - * @return boolean |
|
| 590 | - * @access public |
|
| 591 | - */ |
|
| 592 | - public function clear() |
|
| 593 | - { |
|
| 594 | - $this->_xhtml = ''; |
|
| 595 | - return true; |
|
| 596 | - } |
|
| 597 | - |
|
| 598 | - /** |
|
| 599 | - * Main parsing fuction |
|
| 600 | - * |
|
| 601 | - * @param string $doc HTML document for processing |
|
| 602 | - * @return string Processed (X)HTML document |
|
| 603 | - * @access public |
|
| 604 | - */ |
|
| 605 | - public function parse($doc, $isUTF7=false) |
|
| 606 | - { |
|
| 60 | + /** |
|
| 61 | + * Storage for resulting HTML output |
|
| 62 | + * |
|
| 63 | + * @var string |
|
| 64 | + * @access private |
|
| 65 | + */ |
|
| 66 | + private $_xhtml = ''; |
|
| 67 | + |
|
| 68 | + /** |
|
| 69 | + * Array of counters for each tag |
|
| 70 | + * |
|
| 71 | + * @var array |
|
| 72 | + * @access private |
|
| 73 | + */ |
|
| 74 | + private $_counter = array(); |
|
| 75 | + |
|
| 76 | + /** |
|
| 77 | + * Stack of unclosed tags |
|
| 78 | + * |
|
| 79 | + * @var array |
|
| 80 | + * @access private |
|
| 81 | + */ |
|
| 82 | + private $_stack = array(); |
|
| 83 | + |
|
| 84 | + /** |
|
| 85 | + * Array of counters for tags that must be deleted with all content |
|
| 86 | + * |
|
| 87 | + * @var array |
|
| 88 | + * @access private |
|
| 89 | + */ |
|
| 90 | + private $_dcCounter = array(); |
|
| 91 | + |
|
| 92 | + /** |
|
| 93 | + * Stack of unclosed tags that must be deleted with all content |
|
| 94 | + * |
|
| 95 | + * @var array |
|
| 96 | + * @access private |
|
| 97 | + */ |
|
| 98 | + private $_dcStack = array(); |
|
| 99 | + |
|
| 100 | + /** |
|
| 101 | + * Stores level of list (ol/ul) nesting |
|
| 102 | + * |
|
| 103 | + * @var int |
|
| 104 | + * @access private |
|
| 105 | + */ |
|
| 106 | + private $_listScope = 0; |
|
| 107 | + |
|
| 108 | + /** |
|
| 109 | + * Stack of unclosed list tags |
|
| 110 | + * |
|
| 111 | + * @var array |
|
| 112 | + * @access private |
|
| 113 | + */ |
|
| 114 | + private $_liStack = array(); |
|
| 115 | + |
|
| 116 | + /** |
|
| 117 | + * Array of prepared regular expressions for protocols (schemas) matching |
|
| 118 | + * |
|
| 119 | + * @var array |
|
| 120 | + * @access private |
|
| 121 | + */ |
|
| 122 | + private $_protoRegexps = array(); |
|
| 123 | + |
|
| 124 | + /** |
|
| 125 | + * Array of prepared regular expressions for CSS matching |
|
| 126 | + * |
|
| 127 | + * @var array |
|
| 128 | + * @access private |
|
| 129 | + */ |
|
| 130 | + private $_cssRegexps = array(); |
|
| 131 | + |
|
| 132 | + /** |
|
| 133 | + * List of single tags ("<tag />") |
|
| 134 | + * |
|
| 135 | + * @var array |
|
| 136 | + * @access public |
|
| 137 | + */ |
|
| 138 | + public $singleTags = array('area', 'br', 'img', 'input', 'hr', 'wbr', ); |
|
| 139 | + |
|
| 140 | + /** |
|
| 141 | + * List of dangerous tags (such tags will be deleted) |
|
| 142 | + * |
|
| 143 | + * @var array |
|
| 144 | + * @access public |
|
| 145 | + */ |
|
| 146 | + public $deleteTags = array( |
|
| 147 | + 'applet', 'base', 'basefont', 'bgsound', 'blink', 'body', |
|
| 148 | + 'embed', 'frame', 'frameset', 'head', 'html', 'ilayer', |
|
| 149 | + 'iframe', 'layer', 'link', 'meta', 'object', 'style', |
|
| 150 | + 'title', 'script', |
|
| 151 | + ); |
|
| 152 | + |
|
| 153 | + /** |
|
| 154 | + * List of dangerous tags (such tags will be deleted, and all content |
|
| 155 | + * inside this tags will be also removed) |
|
| 156 | + * |
|
| 157 | + * @var array |
|
| 158 | + * @access public |
|
| 159 | + */ |
|
| 160 | + public $deleteTagsContent = array('script', 'style', 'title', 'xml', ); |
|
| 161 | + |
|
| 162 | + /** |
|
| 163 | + * Type of protocols filtering ('white' or 'black') |
|
| 164 | + * |
|
| 165 | + * @var string |
|
| 166 | + * @access public |
|
| 167 | + */ |
|
| 168 | + public $protocolFiltering = 'white'; |
|
| 169 | + |
|
| 170 | + /** |
|
| 171 | + * List of "dangerous" protocols (used for blacklist-filtering) |
|
| 172 | + * |
|
| 173 | + * @var array |
|
| 174 | + * @access public |
|
| 175 | + */ |
|
| 176 | + public $blackProtocols = array( |
|
| 177 | + 'about', 'chrome', 'data', 'disk', 'hcp', |
|
| 178 | + 'help', 'javascript', 'livescript', 'lynxcgi', 'lynxexec', |
|
| 179 | + 'ms-help', 'ms-its', 'mhtml', 'mocha', 'opera', |
|
| 180 | + 'res', 'resource', 'shell', 'vbscript', 'view-source', |
|
| 181 | + 'vnd.ms.radio', 'wysiwyg', |
|
| 182 | + ); |
|
| 183 | + |
|
| 184 | + /** |
|
| 185 | + * List of "safe" protocols (used for whitelist-filtering) |
|
| 186 | + * |
|
| 187 | + * @var array |
|
| 188 | + * @access public |
|
| 189 | + */ |
|
| 190 | + public $whiteProtocols = array( |
|
| 191 | + 'ed2k', 'file', 'ftp', 'gopher', 'http', 'https', |
|
| 192 | + 'irc', 'mailto', 'news', 'nntp', 'telnet', 'webcal', |
|
| 193 | + 'xmpp', 'callto', |
|
| 194 | + ); |
|
| 195 | + |
|
| 196 | + /** |
|
| 197 | + * List of attributes that can contain protocols |
|
| 198 | + * |
|
| 199 | + * @var array |
|
| 200 | + * @access public |
|
| 201 | + */ |
|
| 202 | + public $protocolAttributes = array( |
|
| 203 | + 'action', 'background', 'codebase', 'dynsrc', 'href', 'lowsrc', 'src', |
|
| 204 | + ); |
|
| 205 | + |
|
| 206 | + /** |
|
| 207 | + * List of dangerous CSS keywords |
|
| 208 | + * |
|
| 209 | + * Whole style="" attribute will be removed, if parser will find one of |
|
| 210 | + * these keywords |
|
| 211 | + * |
|
| 212 | + * @var array |
|
| 213 | + * @access public |
|
| 214 | + */ |
|
| 215 | + public $cssKeywords = array( |
|
| 216 | + 'absolute', 'behavior', 'behaviour', 'content', 'expression', |
|
| 217 | + 'fixed', 'include-source', 'moz-binding', |
|
| 218 | + ); |
|
| 219 | + |
|
| 220 | + /** |
|
| 221 | + * List of tags that can have no "closing tag" |
|
| 222 | + * |
|
| 223 | + * @var array |
|
| 224 | + * @access public |
|
| 225 | + * @deprecated XHTML does not allow such tags |
|
| 226 | + */ |
|
| 227 | + public $noClose = array(); |
|
| 228 | + |
|
| 229 | + /** |
|
| 230 | + * List of block-level tags that terminates paragraph |
|
| 231 | + * |
|
| 232 | + * Paragraph will be closed when this tags opened |
|
| 233 | + * |
|
| 234 | + * @var array |
|
| 235 | + * @access public |
|
| 236 | + */ |
|
| 237 | + public $closeParagraph = array( |
|
| 238 | + 'address', 'blockquote', 'center', 'dd', 'dir', 'div', |
|
| 239 | + 'dl', 'dt', 'h1', 'h2', 'h3', 'h4', |
|
| 240 | + 'h5', 'h6', 'hr', 'isindex', 'listing', 'marquee', |
|
| 241 | + 'menu', 'multicol', 'ol', 'p', 'plaintext', 'pre', |
|
| 242 | + 'table', 'ul', 'xmp', |
|
| 243 | + ); |
|
| 244 | + |
|
| 245 | + /** |
|
| 246 | + * List of table tags, all table tags outside a table will be removed |
|
| 247 | + * |
|
| 248 | + * @var array |
|
| 249 | + * @access public |
|
| 250 | + */ |
|
| 251 | + public $tableTags = array( |
|
| 252 | + 'caption', 'col', 'colgroup', 'tbody', 'td', 'tfoot', 'th', |
|
| 253 | + 'thead', 'tr', |
|
| 254 | + ); |
|
| 255 | + |
|
| 256 | + /** |
|
| 257 | + * List of list tags |
|
| 258 | + * |
|
| 259 | + * @var array |
|
| 260 | + * @access public |
|
| 261 | + */ |
|
| 262 | + public $listTags = array('dir', 'menu', 'ol', 'ul', 'dl', ); |
|
| 263 | + |
|
| 264 | + /** |
|
| 265 | + * List of dangerous attributes |
|
| 266 | + * |
|
| 267 | + * @var array |
|
| 268 | + * @access public |
|
| 269 | + */ |
|
| 270 | + public $attributes = array('dynsrc'); |
|
| 271 | + //public $attributes = array('dynsrc', 'id', 'name', ); //id and name are dangerous? |
|
| 272 | + |
|
| 273 | + /** |
|
| 274 | + * List of allowed "namespaced" attributes |
|
| 275 | + * |
|
| 276 | + * @var array |
|
| 277 | + * @access public |
|
| 278 | + */ |
|
| 279 | + public $attributesNS = array('xml:lang', ); |
|
| 280 | + |
|
| 281 | + /** |
|
| 282 | + * Constructs class |
|
| 283 | + * |
|
| 284 | + * @access public |
|
| 285 | + */ |
|
| 286 | + public function __construct() |
|
| 287 | + { |
|
| 288 | + //making regular expressions based on Proto & CSS arrays |
|
| 289 | + foreach ($this->blackProtocols as $proto) { |
|
| 290 | + $preg = "/[\s\x01-\x1F]*"; |
|
| 291 | + for ($i=0; $i<strlen($proto); $i++) { |
|
| 292 | + $preg .= $proto{$i} . "[\s\x01-\x1F]*"; |
|
| 293 | + } |
|
| 294 | + $preg .= ":/i"; |
|
| 295 | + $this->_protoRegexps[] = $preg; |
|
| 296 | + } |
|
| 297 | + |
|
| 298 | + foreach ($this->cssKeywords as $css) { |
|
| 299 | + $this->_cssRegexps[] = '/' . $css . '/i'; |
|
| 300 | + } |
|
| 301 | + return true; |
|
| 302 | + } |
|
| 303 | + |
|
| 304 | + /** |
|
| 305 | + * Handles the writing of attributes - called from $this->_openHandler() |
|
| 306 | + * |
|
| 307 | + * @param array $attrs array of attributes $name => $value |
|
| 308 | + * @return boolean |
|
| 309 | + * @access private |
|
| 310 | + */ |
|
| 311 | + private function _writeAttrs ($attrs) |
|
| 312 | + { |
|
| 313 | + if (is_array($attrs)) { |
|
| 314 | + foreach ($attrs as $name => $value) { |
|
| 315 | + |
|
| 316 | + $name = strtolower($name); |
|
| 317 | + |
|
| 318 | + if (strpos($name, 'on') === 0) { |
|
| 319 | + continue; |
|
| 320 | + } |
|
| 321 | + if (strpos($name, 'data') === 0) { |
|
| 322 | + continue; |
|
| 323 | + } |
|
| 324 | + if (in_array($name, $this->attributes)) { |
|
| 325 | + continue; |
|
| 326 | + } |
|
| 327 | + if (!preg_match("/^[a-z0-9]+$/i", $name)) { |
|
| 328 | + if (!in_array($name, $this->attributesNS)) |
|
| 329 | + { |
|
| 330 | + continue; |
|
| 331 | + } |
|
| 332 | + } |
|
| 333 | + |
|
| 334 | + if (($value === TRUE) || (is_null($value))) { |
|
| 335 | + $value = $name; |
|
| 336 | + } |
|
| 337 | + |
|
| 338 | + if ($name == 'style') { |
|
| 339 | + |
|
| 340 | + // removes insignificant backslahes |
|
| 341 | + $value = str_replace("\\", '', $value); |
|
| 342 | + |
|
| 343 | + // removes CSS comments |
|
| 344 | + while (1) |
|
| 345 | + { |
|
| 346 | + $_value = preg_replace("!/\*.*?\*/!s", '', $value); |
|
| 347 | + if ($_value == $value) break; |
|
| 348 | + $value = $_value; |
|
| 349 | + } |
|
| 350 | + |
|
| 351 | + // replace all & to & |
|
| 352 | + $value = str_replace('&', '&', $value); |
|
| 353 | + $value = str_replace('&', '&', $value); |
|
| 354 | + |
|
| 355 | + foreach ($this->_cssRegexps as $css) { |
|
| 356 | + if (preg_match($css, $value)) { |
|
| 357 | + continue 2; |
|
| 358 | + } |
|
| 359 | + } |
|
| 360 | + foreach ($this->_protoRegexps as $proto) { |
|
| 361 | + if (preg_match($proto, $value)) { |
|
| 362 | + continue 2; |
|
| 363 | + } |
|
| 364 | + } |
|
| 365 | + } |
|
| 366 | + |
|
| 367 | + $tempval = preg_replace_callback( |
|
| 368 | + '/&#(\d+);?/m', |
|
| 369 | + function ($matches) { |
|
| 370 | + return chr($matches[0]); |
|
| 371 | + }, |
|
| 372 | + $value |
|
| 373 | + ); //"' |
|
| 374 | + |
|
| 375 | + $tempval = preg_replace_callback( |
|
| 376 | + '/&#x([0-9a-f]+);?/mi', |
|
| 377 | + function ($matches) { |
|
| 378 | + return chr(hexdec($matches[0])); |
|
| 379 | + }, |
|
| 380 | + $tempval |
|
| 381 | + ); |
|
| 382 | + |
|
| 383 | + if ((in_array($name, $this->protocolAttributes)) && |
|
| 384 | + (strpos($tempval, ':') !== false)) |
|
| 385 | + { |
|
| 386 | + if ($this->protocolFiltering == 'black') { |
|
| 387 | + foreach ($this->_protoRegexps as $proto) { |
|
| 388 | + if (preg_match($proto, $tempval)) continue 2; |
|
| 389 | + } |
|
| 390 | + } else { |
|
| 391 | + $_tempval = explode(':', $tempval); |
|
| 392 | + $proto = $_tempval[0]; |
|
| 393 | + if (!in_array($proto, $this->whiteProtocols)) { |
|
| 394 | + continue; |
|
| 395 | + } |
|
| 396 | + } |
|
| 397 | + } |
|
| 398 | + |
|
| 399 | + $value = str_replace("\"", """, $value); |
|
| 400 | + $this->_xhtml .= ' ' . $name . '="' . $value . '"'; |
|
| 401 | + } |
|
| 402 | + } |
|
| 403 | + return true; |
|
| 404 | + } |
|
| 405 | + |
|
| 406 | + /** |
|
| 407 | + * Opening tag handler - called from HTMLSax |
|
| 408 | + * |
|
| 409 | + * @param object $parser HTML Parser |
|
| 410 | + * @param string $name tag name |
|
| 411 | + * @param array $attrs tag attributes |
|
| 412 | + * @return boolean |
|
| 413 | + * @access private |
|
| 414 | + */ |
|
| 415 | + public function _openHandler(&$parser, $name, $attrs) |
|
| 416 | + { |
|
| 417 | + $name = strtolower($name); |
|
| 418 | + |
|
| 419 | + if (in_array($name, $this->deleteTagsContent)) { |
|
| 420 | + array_push($this->_dcStack, $name); |
|
| 421 | + $this->_dcCounter[$name] = isset($this->_dcCounter[$name]) ? $this->_dcCounter[$name]+1 : 1; |
|
| 422 | + } |
|
| 423 | + if (count($this->_dcStack) != 0) { |
|
| 424 | + return true; |
|
| 425 | + } |
|
| 426 | + |
|
| 427 | + if (in_array($name, $this->deleteTags)) { |
|
| 428 | + return true; |
|
| 429 | + } |
|
| 430 | + |
|
| 431 | + if (!preg_match("/^[a-z0-9]+$/i", $name)) { |
|
| 432 | + if (preg_match("!(?:\@|://)!i", $name)) { |
|
| 433 | + $this->_xhtml .= '<' . $name . '>'; |
|
| 434 | + } |
|
| 435 | + return true; |
|
| 436 | + } |
|
| 437 | + |
|
| 438 | + if (in_array($name, $this->singleTags)) { |
|
| 439 | + $this->_xhtml .= '<' . $name; |
|
| 440 | + $this->_writeAttrs($attrs); |
|
| 441 | + $this->_xhtml .= ' />'; |
|
| 442 | + return true; |
|
| 443 | + } |
|
| 444 | + |
|
| 445 | + // TABLES: cannot open table elements when we are not inside table |
|
| 446 | + if ((isset($this->_counter['table'])) && ($this->_counter['table'] <= 0) |
|
| 447 | + && (in_array($name, $this->tableTags))) |
|
| 448 | + { |
|
| 449 | + return true; |
|
| 450 | + } |
|
| 451 | + |
|
| 452 | + // PARAGRAPHS: close paragraph when closeParagraph tags opening |
|
| 453 | + if ((in_array($name, $this->closeParagraph)) && (in_array('p', $this->_stack))) { |
|
| 454 | + $this->_closeHandler($parser, 'p'); |
|
| 455 | + } |
|
| 456 | + |
|
| 457 | + // LISTS: we should close <li> if <li> of the same level opening |
|
| 458 | + if ($name == 'li' && count($this->_liStack) && |
|
| 459 | + $this->_listScope == $this->_liStack[count($this->_liStack)-1]) |
|
| 460 | + { |
|
| 461 | + $this->_closeHandler($parser, 'li'); |
|
| 462 | + } |
|
| 463 | + |
|
| 464 | + // LISTS: we want to know on what nesting level of lists we are |
|
| 465 | + if (in_array($name, $this->listTags)) { |
|
| 466 | + $this->_listScope++; |
|
| 467 | + } |
|
| 468 | + if ($name == 'li') { |
|
| 469 | + array_push($this->_liStack, $this->_listScope); |
|
| 470 | + } |
|
| 471 | + |
|
| 472 | + $this->_xhtml .= '<' . $name; |
|
| 473 | + $this->_writeAttrs($attrs); |
|
| 474 | + $this->_xhtml .= '>'; |
|
| 475 | + array_push($this->_stack,$name); |
|
| 476 | + $this->_counter[$name] = isset($this->_counter[$name]) ? $this->_counter[$name]+1 : 1; |
|
| 477 | + return true; |
|
| 478 | + } |
|
| 479 | + |
|
| 480 | + /** |
|
| 481 | + * Closing tag handler - called from HTMLSax |
|
| 482 | + * |
|
| 483 | + * @param object $parsers HTML parser |
|
| 484 | + * @param string $name tag name |
|
| 485 | + * @return boolean |
|
| 486 | + * @access private |
|
| 487 | + */ |
|
| 488 | + public function _closeHandler(&$parser, $name) |
|
| 489 | + { |
|
| 490 | + |
|
| 491 | + $name = strtolower($name); |
|
| 492 | + |
|
| 493 | + if (isset($this->_dcCounter[$name]) && ($this->_dcCounter[$name] > 0) && |
|
| 494 | + (in_array($name, $this->deleteTagsContent))) |
|
| 495 | + { |
|
| 496 | + while ($name != ($tag = array_pop($this->_dcStack))) { |
|
| 497 | + $this->_dcCounter[$tag]--; |
|
| 498 | + } |
|
| 499 | + |
|
| 500 | + $this->_dcCounter[$name]--; |
|
| 501 | + } |
|
| 502 | + |
|
| 503 | + if (count($this->_dcStack) != 0) { |
|
| 504 | + return true; |
|
| 505 | + } |
|
| 506 | + |
|
| 507 | + if ((isset($this->_counter[$name])) && ($this->_counter[$name] > 0)) { |
|
| 508 | + while ($name != ($tag = array_pop($this->_stack))) { |
|
| 509 | + $this->_closeTag($tag); |
|
| 510 | + } |
|
| 511 | + |
|
| 512 | + $this->_closeTag($name); |
|
| 513 | + } |
|
| 514 | + return true; |
|
| 515 | + } |
|
| 516 | + |
|
| 517 | + /** |
|
| 518 | + * Closes tag |
|
| 519 | + * |
|
| 520 | + * @param string $tag tag name |
|
| 521 | + * @return boolean |
|
| 522 | + * @access private |
|
| 523 | + */ |
|
| 524 | + public function _closeTag($tag) |
|
| 525 | + { |
|
| 526 | + if (!in_array($tag, $this->noClose)) { |
|
| 527 | + $this->_xhtml .= '</' . $tag . '>'; |
|
| 528 | + } |
|
| 529 | + |
|
| 530 | + $this->_counter[$tag]--; |
|
| 531 | + |
|
| 532 | + if (in_array($tag, $this->listTags)) { |
|
| 533 | + $this->_listScope--; |
|
| 534 | + } |
|
| 535 | + |
|
| 536 | + if ($tag == 'li') { |
|
| 537 | + array_pop($this->_liStack); |
|
| 538 | + } |
|
| 539 | + return true; |
|
| 540 | + } |
|
| 541 | + |
|
| 542 | + /** |
|
| 543 | + * Character data handler - called from HTMLSax |
|
| 544 | + * |
|
| 545 | + * @param object $parser HTML parser |
|
| 546 | + * @param string $data textual data |
|
| 547 | + * @return boolean |
|
| 548 | + * @access private |
|
| 549 | + */ |
|
| 550 | + public function _dataHandler(&$parser, $data) |
|
| 551 | + { |
|
| 552 | + if (count($this->_dcStack) == 0) { |
|
| 553 | + $this->_xhtml .= $data; |
|
| 554 | + } |
|
| 555 | + return true; |
|
| 556 | + } |
|
| 557 | + |
|
| 558 | + /** |
|
| 559 | + * Escape handler - called from HTMLSax |
|
| 560 | + * |
|
| 561 | + * @param object $parser HTML parser |
|
| 562 | + * @param string $data comments or other type of data |
|
| 563 | + * @return boolean |
|
| 564 | + * @access private |
|
| 565 | + */ |
|
| 566 | + public function _escapeHandler(&$parser, $data) |
|
| 567 | + { |
|
| 568 | + return true; |
|
| 569 | + } |
|
| 570 | + |
|
| 571 | + /** |
|
| 572 | + * Returns the XHTML document |
|
| 573 | + * |
|
| 574 | + * @return string Processed (X)HTML document |
|
| 575 | + * @access public |
|
| 576 | + */ |
|
| 577 | + public function getXHTML () |
|
| 578 | + { |
|
| 579 | + while ($tag = array_pop($this->_stack)) { |
|
| 580 | + $this->_closeTag($tag); |
|
| 581 | + } |
|
| 582 | + |
|
| 583 | + return $this->_xhtml; |
|
| 584 | + } |
|
| 585 | + |
|
| 586 | + /** |
|
| 587 | + * Clears current document data |
|
| 588 | + * |
|
| 589 | + * @return boolean |
|
| 590 | + * @access public |
|
| 591 | + */ |
|
| 592 | + public function clear() |
|
| 593 | + { |
|
| 594 | + $this->_xhtml = ''; |
|
| 595 | + return true; |
|
| 596 | + } |
|
| 597 | + |
|
| 598 | + /** |
|
| 599 | + * Main parsing fuction |
|
| 600 | + * |
|
| 601 | + * @param string $doc HTML document for processing |
|
| 602 | + * @return string Processed (X)HTML document |
|
| 603 | + * @access public |
|
| 604 | + */ |
|
| 605 | + public function parse($doc, $isUTF7=false) |
|
| 606 | + { |
|
| 607 | 607 | $this->clear(); |
| 608 | 608 | |
| 609 | - // Save all '<' symbols |
|
| 610 | - $doc = preg_replace("/<(?=[^a-zA-Z\/\!\?\%])/", '<', (string)$doc); |
|
| 611 | - |
|
| 612 | - // Web documents shouldn't contains \x00 symbol |
|
| 613 | - $doc = str_replace("\x00", '', $doc); |
|
| 614 | - |
|
| 615 | - // Opera6 bug workaround |
|
| 616 | - $doc = str_replace("\xC0\xBC", '<', $doc); |
|
| 617 | - |
|
| 618 | - // UTF-7 encoding ASCII decode |
|
| 619 | - if($isUTF7) |
|
| 620 | - $doc = $this->repackUTF7($doc); |
|
| 621 | - |
|
| 622 | - // Instantiate the parser |
|
| 623 | - $parser= new TSax3(); |
|
| 624 | - |
|
| 625 | - // Set up the parser |
|
| 626 | - $parser->set_object($this); |
|
| 627 | - |
|
| 628 | - $parser->set_element_handler('_openHandler','_closeHandler'); |
|
| 629 | - $parser->set_data_handler('_dataHandler'); |
|
| 630 | - $parser->set_escape_handler('_escapeHandler'); |
|
| 631 | - |
|
| 632 | - $parser->parse($doc); |
|
| 633 | - |
|
| 634 | - return $this->getXHTML(); |
|
| 635 | - |
|
| 636 | - } |
|
| 637 | - |
|
| 638 | - |
|
| 639 | - /** |
|
| 640 | - * UTF-7 decoding fuction |
|
| 641 | - * |
|
| 642 | - * @param string $str HTML document for recode ASCII part of UTF-7 back to ASCII |
|
| 643 | - * @return string Decoded document |
|
| 644 | - * @access private |
|
| 645 | - */ |
|
| 646 | - private function repackUTF7($str) |
|
| 647 | - { |
|
| 648 | - return preg_replace_callback('!\+([0-9a-zA-Z/]+)\-!', array($this, 'repackUTF7Callback'), $str); |
|
| 649 | - } |
|
| 650 | - |
|
| 651 | - /** |
|
| 652 | - * Additional UTF-7 decoding fuction |
|
| 653 | - * |
|
| 654 | - * @param string $str String for recode ASCII part of UTF-7 back to ASCII |
|
| 655 | - * @return string Recoded string |
|
| 656 | - * @access private |
|
| 657 | - */ |
|
| 658 | - private function repackUTF7Callback($str) |
|
| 659 | - { |
|
| 660 | - $str = base64_decode($str[1]); |
|
| 661 | - $str = preg_replace_callback('/^((?:\x00.)*)((?:[^\x00].)+)/', array($this, 'repackUTF7Back'), $str); |
|
| 662 | - return preg_replace('/\x00(.)/', '$1', $str); |
|
| 663 | - } |
|
| 664 | - |
|
| 665 | - /** |
|
| 666 | - * Additional UTF-7 encoding fuction |
|
| 667 | - * |
|
| 668 | - * @param string $str String for recode ASCII part of UTF-7 back to ASCII |
|
| 669 | - * @return string Recoded string |
|
| 670 | - * @access private |
|
| 671 | - */ |
|
| 672 | - private function repackUTF7Back($str) |
|
| 673 | - { |
|
| 674 | - return $str[1].'+'.rtrim(base64_encode($str[2]), '=').'-'; |
|
| 675 | - } |
|
| 609 | + // Save all '<' symbols |
|
| 610 | + $doc = preg_replace("/<(?=[^a-zA-Z\/\!\?\%])/", '<', (string)$doc); |
|
| 611 | + |
|
| 612 | + // Web documents shouldn't contains \x00 symbol |
|
| 613 | + $doc = str_replace("\x00", '', $doc); |
|
| 614 | + |
|
| 615 | + // Opera6 bug workaround |
|
| 616 | + $doc = str_replace("\xC0\xBC", '<', $doc); |
|
| 617 | + |
|
| 618 | + // UTF-7 encoding ASCII decode |
|
| 619 | + if($isUTF7) |
|
| 620 | + $doc = $this->repackUTF7($doc); |
|
| 621 | + |
|
| 622 | + // Instantiate the parser |
|
| 623 | + $parser= new TSax3(); |
|
| 624 | + |
|
| 625 | + // Set up the parser |
|
| 626 | + $parser->set_object($this); |
|
| 627 | + |
|
| 628 | + $parser->set_element_handler('_openHandler','_closeHandler'); |
|
| 629 | + $parser->set_data_handler('_dataHandler'); |
|
| 630 | + $parser->set_escape_handler('_escapeHandler'); |
|
| 631 | + |
|
| 632 | + $parser->parse($doc); |
|
| 633 | + |
|
| 634 | + return $this->getXHTML(); |
|
| 635 | + |
|
| 636 | + } |
|
| 637 | + |
|
| 638 | + |
|
| 639 | + /** |
|
| 640 | + * UTF-7 decoding fuction |
|
| 641 | + * |
|
| 642 | + * @param string $str HTML document for recode ASCII part of UTF-7 back to ASCII |
|
| 643 | + * @return string Decoded document |
|
| 644 | + * @access private |
|
| 645 | + */ |
|
| 646 | + private function repackUTF7($str) |
|
| 647 | + { |
|
| 648 | + return preg_replace_callback('!\+([0-9a-zA-Z/]+)\-!', array($this, 'repackUTF7Callback'), $str); |
|
| 649 | + } |
|
| 650 | + |
|
| 651 | + /** |
|
| 652 | + * Additional UTF-7 decoding fuction |
|
| 653 | + * |
|
| 654 | + * @param string $str String for recode ASCII part of UTF-7 back to ASCII |
|
| 655 | + * @return string Recoded string |
|
| 656 | + * @access private |
|
| 657 | + */ |
|
| 658 | + private function repackUTF7Callback($str) |
|
| 659 | + { |
|
| 660 | + $str = base64_decode($str[1]); |
|
| 661 | + $str = preg_replace_callback('/^((?:\x00.)*)((?:[^\x00].)+)/', array($this, 'repackUTF7Back'), $str); |
|
| 662 | + return preg_replace('/\x00(.)/', '$1', $str); |
|
| 663 | + } |
|
| 664 | + |
|
| 665 | + /** |
|
| 666 | + * Additional UTF-7 encoding fuction |
|
| 667 | + * |
|
| 668 | + * @param string $str String for recode ASCII part of UTF-7 back to ASCII |
|
| 669 | + * @return string Recoded string |
|
| 670 | + * @access private |
|
| 671 | + */ |
|
| 672 | + private function repackUTF7Back($str) |
|
| 673 | + { |
|
| 674 | + return $str[1].'+'.rtrim(base64_encode($str[2]), '=').'-'; |
|
| 675 | + } |
|
| 676 | 676 | } |
| 677 | 677 | |
| 678 | 678 | /* |
@@ -32,40 +32,40 @@ discard block |
||
| 32 | 32 | * @access protected |
| 33 | 33 | */ |
| 34 | 34 | class TSax3_Trim { |
| 35 | - /** |
|
| 36 | - * Original handler object |
|
| 37 | - * @var object |
|
| 38 | - * @access private |
|
| 39 | - */ |
|
| 40 | - private $orig_obj; |
|
| 41 | - /** |
|
| 42 | - * Original handler method |
|
| 43 | - * @var string |
|
| 44 | - * @access private |
|
| 45 | - */ |
|
| 46 | - private $orig_method; |
|
| 47 | - /** |
|
| 48 | - * Constructs TSax3_Trim |
|
| 49 | - * @param object handler object being decorated |
|
| 50 | - * @param string original handler method |
|
| 51 | - * @access protected |
|
| 52 | - */ |
|
| 53 | - function __construct(&$orig_obj, $orig_method) { |
|
| 54 | - $this->orig_obj =& $orig_obj; |
|
| 55 | - $this->orig_method = $orig_method; |
|
| 56 | - } |
|
| 57 | - /** |
|
| 58 | - * Trims the data |
|
| 59 | - * @param TSax3 |
|
| 60 | - * @param string element data |
|
| 61 | - * @access protected |
|
| 62 | - */ |
|
| 63 | - function trimData(&$parser, $data) { |
|
| 64 | - $data = trim($data); |
|
| 65 | - if ($data != '') { |
|
| 66 | - $this->orig_obj->{$this->orig_method}($parser, $data); |
|
| 67 | - } |
|
| 68 | - } |
|
| 35 | + /** |
|
| 36 | + * Original handler object |
|
| 37 | + * @var object |
|
| 38 | + * @access private |
|
| 39 | + */ |
|
| 40 | + private $orig_obj; |
|
| 41 | + /** |
|
| 42 | + * Original handler method |
|
| 43 | + * @var string |
|
| 44 | + * @access private |
|
| 45 | + */ |
|
| 46 | + private $orig_method; |
|
| 47 | + /** |
|
| 48 | + * Constructs TSax3_Trim |
|
| 49 | + * @param object handler object being decorated |
|
| 50 | + * @param string original handler method |
|
| 51 | + * @access protected |
|
| 52 | + */ |
|
| 53 | + function __construct(&$orig_obj, $orig_method) { |
|
| 54 | + $this->orig_obj =& $orig_obj; |
|
| 55 | + $this->orig_method = $orig_method; |
|
| 56 | + } |
|
| 57 | + /** |
|
| 58 | + * Trims the data |
|
| 59 | + * @param TSax3 |
|
| 60 | + * @param string element data |
|
| 61 | + * @access protected |
|
| 62 | + */ |
|
| 63 | + function trimData(&$parser, $data) { |
|
| 64 | + $data = trim($data); |
|
| 65 | + if ($data != '') { |
|
| 66 | + $this->orig_obj->{$this->orig_method}($parser, $data); |
|
| 67 | + } |
|
| 68 | + } |
|
| 69 | 69 | } |
| 70 | 70 | /** |
| 71 | 71 | * Coverts tag names to upper case |
@@ -73,55 +73,55 @@ discard block |
||
| 73 | 73 | * @access protected |
| 74 | 74 | */ |
| 75 | 75 | class TSax3_CaseFolding { |
| 76 | - /** |
|
| 77 | - * Original handler object |
|
| 78 | - * @var object |
|
| 79 | - * @access private |
|
| 80 | - */ |
|
| 81 | - private $orig_obj; |
|
| 82 | - /** |
|
| 83 | - * Original open handler method |
|
| 84 | - * @var string |
|
| 85 | - * @access private |
|
| 86 | - */ |
|
| 87 | - private $orig_open_method; |
|
| 88 | - /** |
|
| 89 | - * Original close handler method |
|
| 90 | - * @var string |
|
| 91 | - * @access private |
|
| 92 | - */ |
|
| 93 | - private $orig_close_method; |
|
| 94 | - /** |
|
| 95 | - * Constructs TSax3_CaseFolding |
|
| 96 | - * @param object handler object being decorated |
|
| 97 | - * @param string original open handler method |
|
| 98 | - * @param string original close handler method |
|
| 99 | - * @access protected |
|
| 100 | - */ |
|
| 101 | - function __construct(&$orig_obj, $orig_open_method, $orig_close_method) { |
|
| 102 | - $this->orig_obj =& $orig_obj; |
|
| 103 | - $this->orig_open_method = $orig_open_method; |
|
| 104 | - $this->orig_close_method = $orig_close_method; |
|
| 105 | - } |
|
| 106 | - /** |
|
| 107 | - * Folds up open tag callbacks |
|
| 108 | - * @param TSax3 |
|
| 109 | - * @param string tag name |
|
| 110 | - * @param array tag attributes |
|
| 111 | - * @access protected |
|
| 112 | - */ |
|
| 113 | - function foldOpen(&$parser, $tag, $attrs=array(), $empty = FALSE) { |
|
| 114 | - $this->orig_obj->{$this->orig_open_method}($parser, strtoupper($tag), $attrs, $empty); |
|
| 115 | - } |
|
| 116 | - /** |
|
| 117 | - * Folds up close tag callbacks |
|
| 118 | - * @param TSax3 |
|
| 119 | - * @param string tag name |
|
| 120 | - * @access protected |
|
| 121 | - */ |
|
| 122 | - function foldClose(&$parser, $tag, $empty = FALSE) { |
|
| 123 | - $this->orig_obj->{$this->orig_close_method}($parser, strtoupper($tag), $empty); |
|
| 124 | - } |
|
| 76 | + /** |
|
| 77 | + * Original handler object |
|
| 78 | + * @var object |
|
| 79 | + * @access private |
|
| 80 | + */ |
|
| 81 | + private $orig_obj; |
|
| 82 | + /** |
|
| 83 | + * Original open handler method |
|
| 84 | + * @var string |
|
| 85 | + * @access private |
|
| 86 | + */ |
|
| 87 | + private $orig_open_method; |
|
| 88 | + /** |
|
| 89 | + * Original close handler method |
|
| 90 | + * @var string |
|
| 91 | + * @access private |
|
| 92 | + */ |
|
| 93 | + private $orig_close_method; |
|
| 94 | + /** |
|
| 95 | + * Constructs TSax3_CaseFolding |
|
| 96 | + * @param object handler object being decorated |
|
| 97 | + * @param string original open handler method |
|
| 98 | + * @param string original close handler method |
|
| 99 | + * @access protected |
|
| 100 | + */ |
|
| 101 | + function __construct(&$orig_obj, $orig_open_method, $orig_close_method) { |
|
| 102 | + $this->orig_obj =& $orig_obj; |
|
| 103 | + $this->orig_open_method = $orig_open_method; |
|
| 104 | + $this->orig_close_method = $orig_close_method; |
|
| 105 | + } |
|
| 106 | + /** |
|
| 107 | + * Folds up open tag callbacks |
|
| 108 | + * @param TSax3 |
|
| 109 | + * @param string tag name |
|
| 110 | + * @param array tag attributes |
|
| 111 | + * @access protected |
|
| 112 | + */ |
|
| 113 | + function foldOpen(&$parser, $tag, $attrs=array(), $empty = FALSE) { |
|
| 114 | + $this->orig_obj->{$this->orig_open_method}($parser, strtoupper($tag), $attrs, $empty); |
|
| 115 | + } |
|
| 116 | + /** |
|
| 117 | + * Folds up close tag callbacks |
|
| 118 | + * @param TSax3 |
|
| 119 | + * @param string tag name |
|
| 120 | + * @access protected |
|
| 121 | + */ |
|
| 122 | + function foldClose(&$parser, $tag, $empty = FALSE) { |
|
| 123 | + $this->orig_obj->{$this->orig_close_method}($parser, strtoupper($tag), $empty); |
|
| 124 | + } |
|
| 125 | 125 | } |
| 126 | 126 | /** |
| 127 | 127 | * Breaks up data by linefeed characters, resulting in additional |
@@ -130,40 +130,40 @@ discard block |
||
| 130 | 130 | * @access protected |
| 131 | 131 | */ |
| 132 | 132 | class TSax3_Linefeed { |
| 133 | - /** |
|
| 134 | - * Original handler object |
|
| 135 | - * @var object |
|
| 136 | - * @access private |
|
| 137 | - */ |
|
| 138 | - private $orig_obj; |
|
| 139 | - /** |
|
| 140 | - * Original handler method |
|
| 141 | - * @var string |
|
| 142 | - * @access private |
|
| 143 | - */ |
|
| 144 | - private $orig_method; |
|
| 145 | - /** |
|
| 146 | - * Constructs TSax3_LineFeed |
|
| 147 | - * @param object handler object being decorated |
|
| 148 | - * @param string original handler method |
|
| 149 | - * @access protected |
|
| 150 | - */ |
|
| 151 | - function __construct(&$orig_obj, $orig_method) { |
|
| 152 | - $this->orig_obj =& $orig_obj; |
|
| 153 | - $this->orig_method = $orig_method; |
|
| 154 | - } |
|
| 155 | - /** |
|
| 156 | - * Breaks the data up by linefeeds |
|
| 157 | - * @param TSax3 |
|
| 158 | - * @param string element data |
|
| 159 | - * @access protected |
|
| 160 | - */ |
|
| 161 | - function breakData(&$parser, $data) { |
|
| 162 | - $data = explode("\n",$data); |
|
| 163 | - foreach ( $data as $chunk ) { |
|
| 164 | - $this->orig_obj->{$this->orig_method}($parser, $chunk); |
|
| 165 | - } |
|
| 166 | - } |
|
| 133 | + /** |
|
| 134 | + * Original handler object |
|
| 135 | + * @var object |
|
| 136 | + * @access private |
|
| 137 | + */ |
|
| 138 | + private $orig_obj; |
|
| 139 | + /** |
|
| 140 | + * Original handler method |
|
| 141 | + * @var string |
|
| 142 | + * @access private |
|
| 143 | + */ |
|
| 144 | + private $orig_method; |
|
| 145 | + /** |
|
| 146 | + * Constructs TSax3_LineFeed |
|
| 147 | + * @param object handler object being decorated |
|
| 148 | + * @param string original handler method |
|
| 149 | + * @access protected |
|
| 150 | + */ |
|
| 151 | + function __construct(&$orig_obj, $orig_method) { |
|
| 152 | + $this->orig_obj =& $orig_obj; |
|
| 153 | + $this->orig_method = $orig_method; |
|
| 154 | + } |
|
| 155 | + /** |
|
| 156 | + * Breaks the data up by linefeeds |
|
| 157 | + * @param TSax3 |
|
| 158 | + * @param string element data |
|
| 159 | + * @access protected |
|
| 160 | + */ |
|
| 161 | + function breakData(&$parser, $data) { |
|
| 162 | + $data = explode("\n",$data); |
|
| 163 | + foreach ( $data as $chunk ) { |
|
| 164 | + $this->orig_obj->{$this->orig_method}($parser, $chunk); |
|
| 165 | + } |
|
| 166 | + } |
|
| 167 | 167 | } |
| 168 | 168 | /** |
| 169 | 169 | * Breaks up data by tab characters, resulting in additional |
@@ -172,40 +172,40 @@ discard block |
||
| 172 | 172 | * @access protected |
| 173 | 173 | */ |
| 174 | 174 | class TSax3_Tab { |
| 175 | - /** |
|
| 176 | - * Original handler object |
|
| 177 | - * @var object |
|
| 178 | - * @access private |
|
| 179 | - */ |
|
| 180 | - private $orig_obj; |
|
| 181 | - /** |
|
| 182 | - * Original handler method |
|
| 183 | - * @var string |
|
| 184 | - * @access private |
|
| 185 | - */ |
|
| 186 | - private $orig_method; |
|
| 187 | - /** |
|
| 188 | - * Constructs TSax3_Tab |
|
| 189 | - * @param object handler object being decorated |
|
| 190 | - * @param string original handler method |
|
| 191 | - * @access protected |
|
| 192 | - */ |
|
| 193 | - function __construct(&$orig_obj, $orig_method) { |
|
| 194 | - $this->orig_obj =& $orig_obj; |
|
| 195 | - $this->orig_method = $orig_method; |
|
| 196 | - } |
|
| 197 | - /** |
|
| 198 | - * Breaks the data up by linefeeds |
|
| 199 | - * @param TSax3 |
|
| 200 | - * @param string element data |
|
| 201 | - * @access protected |
|
| 202 | - */ |
|
| 203 | - function breakData(&$parser, $data) { |
|
| 204 | - $data = explode("\t",$data); |
|
| 205 | - foreach ( $data as $chunk ) { |
|
| 206 | - $this->orig_obj->{$this->orig_method}($this, $chunk); |
|
| 207 | - } |
|
| 208 | - } |
|
| 175 | + /** |
|
| 176 | + * Original handler object |
|
| 177 | + * @var object |
|
| 178 | + * @access private |
|
| 179 | + */ |
|
| 180 | + private $orig_obj; |
|
| 181 | + /** |
|
| 182 | + * Original handler method |
|
| 183 | + * @var string |
|
| 184 | + * @access private |
|
| 185 | + */ |
|
| 186 | + private $orig_method; |
|
| 187 | + /** |
|
| 188 | + * Constructs TSax3_Tab |
|
| 189 | + * @param object handler object being decorated |
|
| 190 | + * @param string original handler method |
|
| 191 | + * @access protected |
|
| 192 | + */ |
|
| 193 | + function __construct(&$orig_obj, $orig_method) { |
|
| 194 | + $this->orig_obj =& $orig_obj; |
|
| 195 | + $this->orig_method = $orig_method; |
|
| 196 | + } |
|
| 197 | + /** |
|
| 198 | + * Breaks the data up by linefeeds |
|
| 199 | + * @param TSax3 |
|
| 200 | + * @param string element data |
|
| 201 | + * @access protected |
|
| 202 | + */ |
|
| 203 | + function breakData(&$parser, $data) { |
|
| 204 | + $data = explode("\t",$data); |
|
| 205 | + foreach ( $data as $chunk ) { |
|
| 206 | + $this->orig_obj->{$this->orig_method}($this, $chunk); |
|
| 207 | + } |
|
| 208 | + } |
|
| 209 | 209 | } |
| 210 | 210 | /** |
| 211 | 211 | * Breaks up data by XML entities and parses them with html_entity_decode(), |
@@ -215,50 +215,50 @@ discard block |
||
| 215 | 215 | * @access protected |
| 216 | 216 | */ |
| 217 | 217 | class TSax3_Entities_Parsed { |
| 218 | - /** |
|
| 219 | - * Original handler object |
|
| 220 | - * @var object |
|
| 221 | - * @access private |
|
| 222 | - */ |
|
| 223 | - private $orig_obj; |
|
| 224 | - /** |
|
| 225 | - * Original handler method |
|
| 226 | - * @var string |
|
| 227 | - * @access private |
|
| 228 | - */ |
|
| 229 | - private $orig_method; |
|
| 230 | - /** |
|
| 231 | - * Constructs TSax3_Entities_Parsed |
|
| 232 | - * @param object handler object being decorated |
|
| 233 | - * @param string original handler method |
|
| 234 | - * @access protected |
|
| 235 | - */ |
|
| 236 | - function __construct(&$orig_obj, $orig_method) { |
|
| 237 | - $this->orig_obj =& $orig_obj; |
|
| 238 | - $this->orig_method = $orig_method; |
|
| 239 | - } |
|
| 240 | - /** |
|
| 241 | - * Breaks the data up by XML entities |
|
| 242 | - * @param TSax3 |
|
| 243 | - * @param string element data |
|
| 244 | - * @access protected |
|
| 245 | - */ |
|
| 246 | - function breakData(&$parser, $data) { |
|
| 247 | - $data = preg_split('/(&.+?;)/',$data,-1,PREG_SPLIT_DELIM_CAPTURE | PREG_SPLIT_NO_EMPTY); |
|
| 248 | - foreach ( $data as $chunk ) { |
|
| 249 | - $chunk = html_entity_decode($chunk,ENT_NOQUOTES); |
|
| 250 | - $this->orig_obj->{$this->orig_method}($this, $chunk); |
|
| 251 | - } |
|
| 252 | - } |
|
| 218 | + /** |
|
| 219 | + * Original handler object |
|
| 220 | + * @var object |
|
| 221 | + * @access private |
|
| 222 | + */ |
|
| 223 | + private $orig_obj; |
|
| 224 | + /** |
|
| 225 | + * Original handler method |
|
| 226 | + * @var string |
|
| 227 | + * @access private |
|
| 228 | + */ |
|
| 229 | + private $orig_method; |
|
| 230 | + /** |
|
| 231 | + * Constructs TSax3_Entities_Parsed |
|
| 232 | + * @param object handler object being decorated |
|
| 233 | + * @param string original handler method |
|
| 234 | + * @access protected |
|
| 235 | + */ |
|
| 236 | + function __construct(&$orig_obj, $orig_method) { |
|
| 237 | + $this->orig_obj =& $orig_obj; |
|
| 238 | + $this->orig_method = $orig_method; |
|
| 239 | + } |
|
| 240 | + /** |
|
| 241 | + * Breaks the data up by XML entities |
|
| 242 | + * @param TSax3 |
|
| 243 | + * @param string element data |
|
| 244 | + * @access protected |
|
| 245 | + */ |
|
| 246 | + function breakData(&$parser, $data) { |
|
| 247 | + $data = preg_split('/(&.+?;)/',$data,-1,PREG_SPLIT_DELIM_CAPTURE | PREG_SPLIT_NO_EMPTY); |
|
| 248 | + foreach ( $data as $chunk ) { |
|
| 249 | + $chunk = html_entity_decode($chunk,ENT_NOQUOTES); |
|
| 250 | + $this->orig_obj->{$this->orig_method}($this, $chunk); |
|
| 251 | + } |
|
| 252 | + } |
|
| 253 | 253 | } |
| 254 | 254 | /** |
| 255 | 255 | * Compatibility with older PHP versions |
| 256 | 256 | */ |
| 257 | 257 | if (version_compare(phpversion(), '4.3', '<') && !function_exists('html_entity_decode') ) { |
| 258 | - function html_entity_decode($str, $style=ENT_NOQUOTES) { |
|
| 259 | - return strtr($str, |
|
| 260 | - array_flip(get_html_translation_table(HTML_ENTITIES,$style))); |
|
| 261 | - } |
|
| 258 | + function html_entity_decode($str, $style=ENT_NOQUOTES) { |
|
| 259 | + return strtr($str, |
|
| 260 | + array_flip(get_html_translation_table(HTML_ENTITIES,$style))); |
|
| 261 | + } |
|
| 262 | 262 | } |
| 263 | 263 | /** |
| 264 | 264 | * Breaks up data by XML entities but leaves them unparsed, |
@@ -267,40 +267,40 @@ discard block |
||
| 267 | 267 | * @access protected |
| 268 | 268 | */ |
| 269 | 269 | class TSax3_Entities_Unparsed { |
| 270 | - /** |
|
| 271 | - * Original handler object |
|
| 272 | - * @var object |
|
| 273 | - * @access private |
|
| 274 | - */ |
|
| 275 | - private $orig_obj; |
|
| 276 | - /** |
|
| 277 | - * Original handler method |
|
| 278 | - * @var string |
|
| 279 | - * @access private |
|
| 280 | - */ |
|
| 281 | - private $orig_method; |
|
| 282 | - /** |
|
| 283 | - * Constructs TSax3_Entities_Unparsed |
|
| 284 | - * @param object handler object being decorated |
|
| 285 | - * @param string original handler method |
|
| 286 | - * @access protected |
|
| 287 | - */ |
|
| 288 | - function __construct(&$orig_obj, $orig_method) { |
|
| 289 | - $this->orig_obj =& $orig_obj; |
|
| 290 | - $this->orig_method = $orig_method; |
|
| 291 | - } |
|
| 292 | - /** |
|
| 293 | - * Breaks the data up by XML entities |
|
| 294 | - * @param TSax3 |
|
| 295 | - * @param string element data |
|
| 296 | - * @access protected |
|
| 297 | - */ |
|
| 298 | - function breakData(&$parser, $data) { |
|
| 299 | - $data = preg_split('/(&.+?;)/',$data,-1,PREG_SPLIT_DELIM_CAPTURE | PREG_SPLIT_NO_EMPTY); |
|
| 300 | - foreach ( $data as $chunk ) { |
|
| 301 | - $this->orig_obj->{$this->orig_method}($this, $chunk); |
|
| 302 | - } |
|
| 303 | - } |
|
| 270 | + /** |
|
| 271 | + * Original handler object |
|
| 272 | + * @var object |
|
| 273 | + * @access private |
|
| 274 | + */ |
|
| 275 | + private $orig_obj; |
|
| 276 | + /** |
|
| 277 | + * Original handler method |
|
| 278 | + * @var string |
|
| 279 | + * @access private |
|
| 280 | + */ |
|
| 281 | + private $orig_method; |
|
| 282 | + /** |
|
| 283 | + * Constructs TSax3_Entities_Unparsed |
|
| 284 | + * @param object handler object being decorated |
|
| 285 | + * @param string original handler method |
|
| 286 | + * @access protected |
|
| 287 | + */ |
|
| 288 | + function __construct(&$orig_obj, $orig_method) { |
|
| 289 | + $this->orig_obj =& $orig_obj; |
|
| 290 | + $this->orig_method = $orig_method; |
|
| 291 | + } |
|
| 292 | + /** |
|
| 293 | + * Breaks the data up by XML entities |
|
| 294 | + * @param TSax3 |
|
| 295 | + * @param string element data |
|
| 296 | + * @access protected |
|
| 297 | + */ |
|
| 298 | + function breakData(&$parser, $data) { |
|
| 299 | + $data = preg_split('/(&.+?;)/',$data,-1,PREG_SPLIT_DELIM_CAPTURE | PREG_SPLIT_NO_EMPTY); |
|
| 300 | + foreach ( $data as $chunk ) { |
|
| 301 | + $this->orig_obj->{$this->orig_method}($this, $chunk); |
|
| 302 | + } |
|
| 303 | + } |
|
| 304 | 304 | } |
| 305 | 305 | |
| 306 | 306 | /** |
@@ -310,52 +310,52 @@ discard block |
||
| 310 | 310 | * @access protected |
| 311 | 311 | */ |
| 312 | 312 | class TSax3_Escape_Stripper { |
| 313 | - /** |
|
| 314 | - * Original handler object |
|
| 315 | - * @var object |
|
| 316 | - * @access private |
|
| 317 | - */ |
|
| 318 | - private $orig_obj; |
|
| 319 | - /** |
|
| 320 | - * Original handler method |
|
| 321 | - * @var string |
|
| 322 | - * @access private |
|
| 323 | - */ |
|
| 324 | - private $orig_method; |
|
| 325 | - /** |
|
| 326 | - * Constructs TSax3_Entities_Unparsed |
|
| 327 | - * @param object handler object being decorated |
|
| 328 | - * @param string original handler method |
|
| 329 | - * @access protected |
|
| 330 | - */ |
|
| 331 | - function __construct(&$orig_obj, $orig_method) { |
|
| 332 | - $this->orig_obj =& $orig_obj; |
|
| 333 | - $this->orig_method = $orig_method; |
|
| 334 | - } |
|
| 335 | - /** |
|
| 336 | - * Breaks the data up by XML entities |
|
| 337 | - * @param TSax3 |
|
| 338 | - * @param string element data |
|
| 339 | - * @access protected |
|
| 340 | - */ |
|
| 341 | - function strip(&$parser, $data) { |
|
| 342 | - // Check for HTML comments first |
|
| 343 | - if ( substr($data,0,2) == '--' ) { |
|
| 344 | - $patterns = array( |
|
| 345 | - '/^\-\-/', // Opening comment: -- |
|
| 346 | - '/\-\-$/', // Closing comment: -- |
|
| 347 | - ); |
|
| 348 | - $data = preg_replace($patterns,'',$data); |
|
| 313 | + /** |
|
| 314 | + * Original handler object |
|
| 315 | + * @var object |
|
| 316 | + * @access private |
|
| 317 | + */ |
|
| 318 | + private $orig_obj; |
|
| 319 | + /** |
|
| 320 | + * Original handler method |
|
| 321 | + * @var string |
|
| 322 | + * @access private |
|
| 323 | + */ |
|
| 324 | + private $orig_method; |
|
| 325 | + /** |
|
| 326 | + * Constructs TSax3_Entities_Unparsed |
|
| 327 | + * @param object handler object being decorated |
|
| 328 | + * @param string original handler method |
|
| 329 | + * @access protected |
|
| 330 | + */ |
|
| 331 | + function __construct(&$orig_obj, $orig_method) { |
|
| 332 | + $this->orig_obj =& $orig_obj; |
|
| 333 | + $this->orig_method = $orig_method; |
|
| 334 | + } |
|
| 335 | + /** |
|
| 336 | + * Breaks the data up by XML entities |
|
| 337 | + * @param TSax3 |
|
| 338 | + * @param string element data |
|
| 339 | + * @access protected |
|
| 340 | + */ |
|
| 341 | + function strip(&$parser, $data) { |
|
| 342 | + // Check for HTML comments first |
|
| 343 | + if ( substr($data,0,2) == '--' ) { |
|
| 344 | + $patterns = array( |
|
| 345 | + '/^\-\-/', // Opening comment: -- |
|
| 346 | + '/\-\-$/', // Closing comment: -- |
|
| 347 | + ); |
|
| 348 | + $data = preg_replace($patterns,'',$data); |
|
| 349 | 349 | |
| 350 | - // Check for XML CDATA sections (note: don't do both!) |
|
| 351 | - } else if ( substr($data,0,1) == '[' ) { |
|
| 352 | - $patterns = array( |
|
| 353 | - '/^\[.*CDATA.*\[/s', // Opening CDATA |
|
| 354 | - '/\].*\]$/s', // Closing CDATA |
|
| 355 | - ); |
|
| 356 | - $data = preg_replace($patterns,'',$data); |
|
| 357 | - } |
|
| 350 | + // Check for XML CDATA sections (note: don't do both!) |
|
| 351 | + } else if ( substr($data,0,1) == '[' ) { |
|
| 352 | + $patterns = array( |
|
| 353 | + '/^\[.*CDATA.*\[/s', // Opening CDATA |
|
| 354 | + '/\].*\]$/s', // Closing CDATA |
|
| 355 | + ); |
|
| 356 | + $data = preg_replace($patterns,'',$data); |
|
| 357 | + } |
|
| 358 | 358 | |
| 359 | - $this->orig_obj->{$this->orig_method}($this, $data); |
|
| 360 | - } |
|
| 359 | + $this->orig_obj->{$this->orig_method}($this, $data); |
|
| 360 | + } |
|
| 361 | 361 | } |