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 timo.hummel
 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:     /**
 27:      *
 28:      * @var array
 29:      */
 30:     protected $_doubleTags = array(
 31:         "form",
 32:         "head",
 33:         "body",
 34:         "html",
 35:         "td",
 36:         "tr",
 37:         "table",
 38:         "a",
 39:         "tbody",
 40:         "title",
 41:         "container",
 42:         "span",
 43:         "div"
 44:     );
 45: 
 46:     /**
 47:      *
 48:      * @var array
 49:      */
 50:     public $missingNodes = array();
 51: 
 52:     /**
 53:      *
 54:      * @deprecated
 55:      *         not used anymore
 56:      * @var string
 57:      */
 58:     public $iNodeName;
 59: 
 60:     /**
 61:      *
 62:      * @var string
 63:      */
 64:     protected $_html;
 65: 
 66:     /**
 67:      *
 68:      * @var array
 69:      */
 70:     protected $_nestingLevel = array();
 71: 
 72:     /**
 73:      *
 74:      * @var array
 75:      */
 76:     protected $_nestingNodes = array();
 77: 
 78:     /**
 79:      *
 80:      * @var array
 81:      */
 82:     protected $_existingTags = array();
 83: 
 84:     /**
 85:      *
 86:      * @param string $html
 87:      */
 88:     public function validate($html) {
 89:         $nestingLevel = 0;
 90: 
 91:         // Clean up HTML first from any PHP scripts, and clean up line breaks
 92:         $this->_html = $this->_cleanHTML($html);
 93: 
 94:         $htmlParser = new HtmlParser($this->_html);
 95: 
 96:         while ($htmlParser->parse()) {
 97:             $this->_existingTags[] = $htmlParser->iNodeName;
 98:             // Check if we found a double tag
 99:             if (in_array($htmlParser->iNodeName, $this->_doubleTags)) {
100:                 if (!array_key_exists($htmlParser->iNodeName, $this->_nestingLevel)) {
101:                     $this->_nestingLevel[$htmlParser->iNodeName] = 0;
102:                 }
103: 
104:                 if (!array_key_exists($htmlParser->iNodeName, $this->_nestingNodes)) {
105:                     $this->_nestingNodes[$htmlParser->iNodeName][intval($this->_nestingLevel[$htmlParser->iNodeName])] = array();
106:                 }
107: 
108:                 // Check if it's a start tag
109:                 if ($htmlParser->iNodeType == HtmlParser::NODE_TYPE_ELEMENT) {
110:                     // Push the current element to the stack, remember ID and
111:                     // Name, if possible
112:                     $nestingLevel++;
113: 
114:                     $this->_nestingNodes[$htmlParser->iNodeName][intval($this->_nestingLevel[$htmlParser->iNodeName])]["name"] = $htmlParser->iNodeAttributes["name"];
115:                     $this->_nestingNodes[$htmlParser->iNodeName][intval($this->_nestingLevel[$htmlParser->iNodeName])]["id"] = $htmlParser->iNodeAttributes["id"];
116:                     $this->_nestingNodes[$htmlParser->iNodeName][intval($this->_nestingLevel[$htmlParser->iNodeName])]["level"] = $nestingLevel;
117:                     $this->_nestingNodes[$htmlParser->iNodeName][intval($this->_nestingLevel[$htmlParser->iNodeName])]["char"] = $htmlParser->iHtmlTextIndex;
118:                     $this->_nestingLevel[$htmlParser->iNodeName]++;
119:                 }
120: 
121:                 if ($htmlParser->iNodeType == HtmlParser::NODE_TYPE_ENDELEMENT) {
122:                     // Check if we've an element of this type on the stack
123:                     if ($this->_nestingLevel[$htmlParser->iNodeName] > 0) {
124:                         unset($this->_nestingNodes[$htmlParser->iNodeName][$this->_nestingLevel[$htmlParser->iNodeName]]);
125:                         $this->_nestingLevel[$htmlParser->iNodeName]--;
126: 
127:                         if ($this->_nestingNodes[$htmlParser->iNodeName][intval($this->_nestingLevel[$htmlParser->iNodeName])]["level"] != $nestingLevel) {
128:                             // Todo: Check for the wrong nesting level
129:                         }
130: 
131:                         $nestingLevel--;
132:                     }
133:                 }
134:             }
135:         }
136: 
137:         // missingNodes should be an empty array by default
138:         $this->missingNodes = array();
139: 
140:         // Collect all missing nodes
141:         foreach ($this->_nestingLevel as $key => $value) {
142:             // One or more missing tags found
143:             if ($value > 0) {
144:                 // Step trough all missing tags
145:                 for ($i = 0; $i < $value; $i++) {
146:                     $node = $this->_nestingNodes[$key][$i];
147: 
148:                     list($line, $char) = $this->_getLineAndCharPos($node["char"]);
149:                     $this->missingNodes[] = array(
150:                         "tag" => $key,
151:                         "id" => $node["id"],
152:                         "name" => $node["name"],
153:                         "line" => $line,
154:                         "char" => $char
155:                     );
156: 
157:                     $this->missingTags[$line][$char] = true;
158:                 }
159:             }
160:         }
161:     }
162: 
163:     /**
164:      *
165:      * @param string $tag
166:      * @return bool
167:      */
168:     public function tagExists($tag) {
169:         if (in_array($tag, $this->_existingTags)) {
170:             return true;
171:         } else {
172:             return false;
173:         }
174:     }
175: 
176:     /**
177:      *
178:      * @param string $html
179:      * @return mixed
180:      */
181:     protected function _cleanHTML($html) {
182:         // remove all php code from layout
183:         $resultingHTML = preg_replace('/<\?(php)?((.)|(\s))*?\?>/i', '', $html);
184: 
185:         // We respect only \n, but we need to take care of windows (\n\r) and
186:         // other systems (\r)
187:         $resultingHTML = str_replace("\r\n", "\n", $resultingHTML);
188:         $resultingHTML = str_replace("\r", "\n", $resultingHTML);
189: 
190:         return $resultingHTML;
191:     }
192: 
193:     /**
194:      *
195:      * @deprecated
196:      *         not used anymore
197:      * @return string
198:      */
199:     protected function _returnErrorMap() {
200:         $html = "<pre>";
201: 
202:         $chunks = explode("\n", $this->_html);
203: 
204:         foreach ($chunks as $key => $value) {
205:             $html .= ($key + 1) . " ";
206: 
207:             for ($i = 0; $i < strlen($value); $i++) {
208:                 $char = substr($value, $i, 1);
209: 
210:                 if (is_array($this->missingTags[$key + 1])) {
211:                     // echo ($key+1) . " ". $i."<br>";
212:                     if (array_key_exists($i + 2, $this->missingTags[$key + 1])) {
213:                         $html .= "<u><b>" . conHtmlSpecialChars($char) . "</b></u>";
214:                     } else {
215:                         $html .= conHtmlSpecialChars($char);
216:                     }
217:                 } else {
218:                     $html .= conHtmlSpecialChars($char);
219:                 }
220:             }
221: 
222:             $html .= "<br>";
223:         }
224: 
225:         return $html;
226:     }
227: 
228:     /**
229:      *
230:      * @param int $charpos
231:      * @return array
232:      */
233:     protected function _getLineAndCharPos($charpos) {
234:         $mangled = substr($this->_html, 0, $charpos);
235: 
236:         $line = substr_count($mangled, "\n") + 1;
237:         $char = $charpos - strrpos($mangled, "\n");
238: 
239:         return array(
240:             $line,
241:             $char
242:         );
243:     }
244: }
245: