1: <?php
  2: /**
  3:  * This file contains the html validator class.
  4:  *
  5:  * @package    Core
  6:  * @subpackage Backend
  7:  * @author     timo.hummel
  8:  * @copyright  four for business AG <www.4fb.de>
  9:  * @license    http://www.contenido.org/license/LIZENZ.txt
 10:  * @link       http://www.4fb.de
 11:  * @link       http://www.contenido.org
 12:  */
 13: 
 14: defined('CON_FRAMEWORK') || die('Illegal call: Missing framework initialization - request aborted.');
 15: 
 16: /**
 17:  * This class validates HTML.
 18:  *
 19:  * @package    Core
 20:  * @subpackage Backend
 21:  */
 22: class cHTMLValidator
 23: {
 24:     /**
 25:      * @var array
 26:      */
 27:     protected $_doubleTags = [
 28:         "form",
 29:         "head",
 30:         "body",
 31:         "html",
 32:         "td",
 33:         "tr",
 34:         "table",
 35:         "a",
 36:         "tbody",
 37:         "title",
 38:         "container",
 39:         "span",
 40:         "div",
 41:     ];
 42: 
 43:     /**
 44:      * @var array
 45:      */
 46:     public $missingNodes = [];
 47: 
 48:     /**
 49:      * @var array
 50:      */
 51:     public $missingTags = [];
 52: 
 53:     /**
 54:      * @deprecated not used anymore
 55:      * @var string
 56:      */
 57:     public $iNodeName;
 58: 
 59:     /**
 60:      * @var string
 61:      */
 62:     protected $_html;
 63: 
 64:     /**
 65:      * @var array
 66:      */
 67:     protected $_nestingLevel = [];
 68: 
 69:     /**
 70:      * @var array
 71:      */
 72:     protected $_nestingNodes = [];
 73: 
 74:     /**
 75:      * @var array
 76:      */
 77:     protected $_existingTags = [];
 78: 
 79:     /**
 80:      * @param string $html
 81:      */
 82:     public function validate($html)
 83:     {
 84:         $nestingLevel = 0;
 85: 
 86:         // Clean up HTML first from any PHP scripts, and clean up line breaks
 87:         $this->_html = $this->_cleanHTML($html);
 88: 
 89:         $htmlParser = new HtmlParser($this->_html);
 90: 
 91:         while ($htmlParser->parse()) {
 92:             $nodeName              = $htmlParser->getNodeName();
 93:             $this->_existingTags[] = $nodeName;
 94: 
 95:             // Check if we found a double tag
 96:             if (in_array($nodeName, $this->_doubleTags)) {
 97:                 if (!array_key_exists($nodeName, $this->_nestingLevel)) {
 98:                     $this->_nestingLevel[$nodeName] = 0;
 99:                 }
100: 
101:                 if (!array_key_exists($nodeName, $this->_nestingNodes)) {
102:                     $this->_nestingNodes[$nodeName][intval($this->_nestingLevel[$nodeName])] = [];
103:                 }
104: 
105:                 // Check if it's a start tag
106:                 if ($htmlParser->getNodeType() == HtmlParser::NODE_TYPE_ELEMENT) {
107:                     // Push the current element to the stack, remember ID and Name, if possible
108:                     $nestingLevel++;
109: 
110:                     $this->_nestingNodes[$nodeName][intval($this->_nestingLevel[$nodeName])]["name"]  = $htmlParser->getNodeAttributes('name');
111:                     $this->_nestingNodes[$nodeName][intval($this->_nestingLevel[$nodeName])]["id"]    = $htmlParser->getNodeAttributes('id');
112:                     $this->_nestingNodes[$nodeName][intval($this->_nestingLevel[$nodeName])]["level"] = $nestingLevel;
113:                     $this->_nestingNodes[$nodeName][intval($this->_nestingLevel[$nodeName])]["char"]  = $htmlParser->getHtmlTextIndex();
114:                     $this->_nestingLevel[$nodeName]++;
115:                 }
116: 
117:                 if ($htmlParser->getNodeType() == HtmlParser::NODE_TYPE_ENDELEMENT) {
118:                     // Check if we've an element of this type on the stack
119:                     if ($this->_nestingLevel[$nodeName] > 0) {
120:                         unset($this->_nestingNodes[$nodeName][$this->_nestingLevel[$nodeName]]);
121:                         $this->_nestingLevel[$nodeName]--;
122: 
123:                         // TODO check for the wrong nesting level
124:                         // if ($this->_nestingNodes[$nodeName][intval($this->_nestingLevel[$nodeName])]["level"] != $nestingLevel) {
125:                         // }
126: 
127:                         $nestingLevel--;
128:                     }
129:                 }
130:             }
131:         }
132: 
133:         // missingNodes should be an empty array by default
134:         $this->missingNodes = [];
135: 
136:         // Collect all missing nodes
137:         foreach ($this->_nestingLevel as $key => $value) {
138:             // One or more missing tags found
139:             if ($value > 0) {
140:                 // Step trough all missing tags
141:                 for ($i = 0; $i < $value; $i++) {
142:                     $node = $this->_nestingNodes[$key][$i];
143: 
144:                     list($line, $char) = $this->_getLineAndCharPos($node["char"]);
145:                     $this->missingNodes[] = [
146:                         "tag"  => $key,
147:                         "id"   => $node["id"],
148:                         "name" => $node["name"],
149:                         "line" => $line,
150:                         "char" => $char,
151:                     ];
152: 
153:                     $this->missingTags[$line][$char] = true;
154:                 }
155:             }
156:         }
157:     }
158: 
159:     /**
160:      * @param string $tag
161:      *
162:      * @return bool
163:      */
164:     public function tagExists($tag)
165:     {
166:         return in_array($tag, $this->_existingTags);
167:     }
168: 
169:     /**
170:      * @param string $html
171:      *
172:      * @return mixed
173:      */
174:     protected function _cleanHTML($html)
175:     {
176:         // remove all php code from layout
177:         $resultingHTML = preg_replace('/<\?(php)?((.)|(\s))*?\?>/i', '', $html);
178: 
179:         // We respect only \n, but we need to take care of windows (\n\r) and other systems (\r)
180:         $resultingHTML = str_replace("\r\n", "\n", $resultingHTML);
181:         $resultingHTML = str_replace("\r", "\n", $resultingHTML);
182: 
183:         return $resultingHTML;
184:     }
185: 
186:     /**
187:      * @deprecated not used anymore
188:      * @return string
189:      */
190:     protected function _returnErrorMap()
191:     {
192:         $html = "<pre>";
193: 
194:         $chunks = explode("\n", $this->_html);
195: 
196:         foreach ($chunks as $key => $value) {
197:             $html .= ($key + 1) . " ";
198: 
199:             for ($i = 0; $i < cString::getStringLength($value); $i++) {
200:                 $char = cString::getPartOfString($value, $i, 1);
201: 
202:                 if (is_array($this->missingTags[$key + 1])) {
203:                     if (array_key_exists($i + 2, $this->missingTags[$key + 1])) {
204:                         $html .= "<u><b>" . conHtmlSpecialChars($char) . "</b></u>";
205:                     } else {
206:                         $html .= conHtmlSpecialChars($char);
207:                     }
208:                 } else {
209:                     $html .= conHtmlSpecialChars($char);
210:                 }
211:             }
212: 
213:             $html .= "<br>";
214:         }
215: 
216:         return $html;
217:     }
218: 
219:     /**
220:      * @param int $charpos
221:      *
222:      * @return array
223:      */
224:     protected function _getLineAndCharPos($charpos)
225:     {
226:         $mangled = cString::getPartOfString($this->_html, 0, $charpos);
227:         $line    = cString::countSubstring($mangled, "\n") + 1;
228:         $char    = $charpos - cString::findLastPos($mangled, "\n");
229: 
230:         return [$line, $char];
231:     }
232: }
233: