1: <?php
  2: /**
  3:  * This file contains the html validator class.
  4:  *
  5:  * @package    Core
  6:  * @subpackage Backend
  7:  * @version    SVN Revision $Rev:$
  8:  *
  9:  * @author     Unknown
 10:  * @copyright  four for business AG <www.4fb.de>
 11:  * @license    http://www.contenido.org/license/LIZENZ.txt
 12:  * @link       http://www.4fb.de
 13:  * @link       http://www.contenido.org
 14:  */
 15: 
 16: defined('CON_FRAMEWORK') || die('Illegal call: Missing framework initialization - request aborted.');
 17: 
 18: /**
 19:  * This class validates HTML.
 20:  *
 21:  * @package    Core
 22:  * @subpackage Backend
 23:  */
 24: class cHTMLValidator {
 25: 
 26:     protected $_doubleTags = array(
 27:         "form",
 28:         "head",
 29:         "body",
 30:         "html",
 31:         "td",
 32:         "tr",
 33:         "table",
 34:         "a",
 35:         "tbody",
 36:         "title",
 37:         "container",
 38:         "span",
 39:         "div"
 40:     );
 41: 
 42:     protected $_html;
 43: 
 44:     protected $_nestingLevel = array();
 45: 
 46:     public $missingNodes = array();
 47: 
 48:     public $iNodeName;
 49: 
 50:     protected $_nestingNodes = array();
 51: 
 52:     protected $_existingTags = array();
 53: 
 54:     protected function _cleanHTML($html) {
 55:         // remove all php code from layout
 56:         $resultingHTML = preg_replace('/<\?(php)?((.)|(\s))*?\?>/i', '', $html);
 57: 
 58:         // We respect only \n, but we need to take care of windows (\n\r) and
 59:         // other systems (\r)
 60:         $resultingHTML = str_replace("\r\n", "\n", $resultingHTML);
 61:         $resultingHTML = str_replace("\r", "\n", $resultingHTML);
 62: 
 63:         return $resultingHTML;
 64:     }
 65: 
 66:     public function validate($html) {
 67:         $nestingLevel = 0;
 68: 
 69:         // Clean up HTML first from any PHP scripts, and clean up line breaks
 70:         $this->_html = $this->_cleanHTML($html);
 71: 
 72:         $htmlParser = new HtmlParser($this->_html);
 73: 
 74:         while ($htmlParser->parse()) {
 75:             $this->_existingTags[] = $htmlParser->iNodeName;
 76:             // Check if we found a double tag
 77:             if (in_array($htmlParser->iNodeName, $this->_doubleTags)) {
 78:                 if (!array_key_exists($htmlParser->iNodeName, $this->_nestingLevel)) {
 79:                     $this->_nestingLevel[$htmlParser->iNodeName] = 0;
 80:                 }
 81: 
 82:                 if (!array_key_exists($htmlParser->iNodeName, $this->_nestingNodes)) {
 83:                     $this->_nestingNodes[$htmlParser->iNodeName][intval($this->_nestingLevel[$htmlParser->iNodeName])] = array();
 84:                 }
 85: 
 86:                 // Check if it's a start tag
 87:                 if ($htmlParser->iNodeType == HtmlParser::NODE_TYPE_ELEMENT) {
 88:                     // Push the current element to the stack, remember ID and
 89:                     // Name, if possible
 90:                     $nestingLevel++;
 91: 
 92:                     $this->_nestingNodes[$htmlParser->iNodeName][intval($this->_nestingLevel[$htmlParser->iNodeName])]["name"] = $htmlParser->iNodeAttributes["name"];
 93:                     $this->_nestingNodes[$htmlParser->iNodeName][intval($this->_nestingLevel[$htmlParser->iNodeName])]["id"] = $htmlParser->iNodeAttributes["id"];
 94:                     $this->_nestingNodes[$htmlParser->iNodeName][intval($this->_nestingLevel[$htmlParser->iNodeName])]["level"] = $nestingLevel;
 95:                     $this->_nestingNodes[$htmlParser->iNodeName][intval($this->_nestingLevel[$htmlParser->iNodeName])]["char"] = $htmlParser->iHtmlTextIndex;
 96:                     $this->_nestingLevel[$htmlParser->iNodeName]++;
 97:                 }
 98: 
 99:                 if ($htmlParser->iNodeType == HtmlParser::NODE_TYPE_ENDELEMENT) {
100:                     // Check if we've an element of this type on the stack
101:                     if ($this->_nestingLevel[$htmlParser->iNodeName] > 0) {
102:                         unset($this->_nestingNodes[$htmlParser->iNodeName][$this->_nestingLevel[$htmlParser->iNodeName]]);
103:                         $this->_nestingLevel[$htmlParser->iNodeName]--;
104: 
105:                         if ($this->_nestingNodes[$htmlParser->iNodeName][intval($this->_nestingLevel[$htmlParser->iNodeName])]["level"] != $nestingLevel) {
106:                             // Todo: Check for the wrong nesting level
107:                         }
108: 
109:                         $nestingLevel--;
110:                     }
111:                 }
112:             }
113:         }
114: 
115:         // missingNodes should be an empty array by default
116:         $this->missingNodes = array();
117: 
118:         // Collect all missing nodes
119:         foreach ($this->_nestingLevel as $key => $value) {
120:             // One or more missing tags found
121:             if ($value > 0) {
122:                 // Step trough all missing tags
123:                 for ($i = 0; $i < $value; $i++) {
124:                     $node = $this->_nestingNodes[$key][$i];
125: 
126:                     list($line, $char) = $this->_getLineAndCharPos($node["char"]);
127:                     $this->missingNodes[] = array(
128:                         "tag" => $key,
129:                         "id" => $node["id"],
130:                         "name" => $node["name"],
131:                         "line" => $line,
132:                         "char" => $char
133:                     );
134: 
135:                     $this->missingTags[$line][$char] = true;
136:                 }
137:             }
138:         }
139:     }
140: 
141:     public function tagExists($tag) {
142:         if (in_array($tag, $this->_existingTags)) {
143:             return true;
144:         } else {
145:             return false;
146:         }
147:     }
148: 
149:     protected function _returnErrorMap() {
150:         $html = "<pre>";
151: 
152:         $chunks = explode("\n", $this->_html);
153: 
154:         foreach ($chunks as $key => $value) {
155:             $html .= ($key + 1) . " ";
156: 
157:             for ($i = 0; $i < strlen($value); $i++) {
158:                 $char = substr($value, $i, 1);
159: 
160:                 if (is_array($this->missingTags[$key + 1])) {
161:                     // echo ($key+1) . " ". $i."<br>";
162:                     if (array_key_exists($i + 2, $this->missingTags[$key + 1])) {
163:                         $html .= "<u><b>" . conHtmlSpecialChars($char) . "</b></u>";
164:                     } else {
165:                         $html .= conHtmlSpecialChars($char);
166:                     }
167:                 } else {
168:                     $html .= conHtmlSpecialChars($char);
169:                 }
170:             }
171: 
172:             $html .= "<br>";
173:         }
174: 
175:         return $html;
176:     }
177: 
178:     protected function _getLineAndCharPos($charpos) {
179:         $mangled = substr($this->_html, 0, $charpos);
180: 
181:         $line = substr_count($mangled, "\n") + 1;
182:         $char = $charpos - strrpos($mangled, "\n");
183: 
184:         return array(
185:             $line,
186:             $char
187:         );
188:     }
189: 
190: }
191: