1: <?php
  2: /**
  3:  * @filesource Kotchasan/DOMParser.php
  4:  *
  5:  * @copyright 2016 Goragod.com
  6:  * @license http://www.kotchasan.com/license/
  7:  *
  8:  * @see http://www.kotchasan.com/
  9:  */
 10: 
 11: namespace Kotchasan;
 12: 
 13: /**
 14:  * คลาสสำหรับ การ Parse DOM.
 15:  *
 16:  * @author Goragod Wiriya <admin@goragod.com>
 17:  *
 18:  * @since 1.0
 19:  */
 20: class DOMParser
 21: {
 22:     /**
 23:      * รายการโหนด.
 24:      *
 25:      * @var array
 26:      */
 27:     protected $doms = array();
 28: 
 29:     /**
 30:      * class constructor.
 31:      *
 32:      * @param string $html    HTML code
 33:      * @param string $charset encoding default utf-8
 34:      */
 35:     public function __construct($html, $charset = 'utf-8')
 36:     {
 37:         $patt = array(
 38:             '/^(.*)<body[^>]{0,}>(.*)<\/body>(.*)$/is' => '\\2',
 39:             '#<(style|script)(.*?)>(.*?)</\\1>#is' => '',
 40:             '@<!--.*-->@is' => '',
 41:             '/<(\!|link|meta)[^>]+\>/i' => '',
 42:             '@>[\s\t]{0,}[\r\n]+[\s\t]{0,}<@' => '><',
 43:             '@[\s\t\r\n]{0,}<(\/?(br|hr|figure|figcaption|p|div|footer|article|section|blockquote|code|aside|navy|table|tr|td|th|thead|tbody|tfoot|caption|ul|ol|li|dl|dt|dd|h[1-6])[^>]{0,})>[\s\t\r\n]{0,}@i' => '<\\1>',
 44:             '@[\s\t]{0,}[\r\n]+[\s\t]{0,}@' => ' ',
 45:         );
 46:         $html = preg_replace(array_keys($patt), array_values($patt), $html);
 47:         if (strtolower($charset) != 'utf-8') {
 48:             $html = iconv('utf-8', $charset, $html);
 49:         }
 50:         $node = null;
 51:         foreach (preg_split('/<(.*)>/U', $html, -1, PREG_SPLIT_DELIM_CAPTURE) as $i => $e) {
 52:             if ($e != '') {
 53:                 if ($i % 2 == 0) {
 54:                     // text node
 55:                     if ($e != '') {
 56:                         if ($node) {
 57:                             $node->childNodes[] = new DOMNode('', $node, array(), $e);
 58:                         } else {
 59:                             $this->doms[] = new DOMNode('', $node, array(), $e);
 60:                         }
 61:                     }
 62:                 } elseif ($e[0] == '/') {
 63:                     // close tag
 64:                     $node = $node->parentNode;
 65:                 } elseif (preg_match('/^([a-zA-Z0-9]+)([\s]{0,}(.*))?$/', $e, $a2)) {
 66:                     // open tag
 67:                     $tag = strtoupper($a2[1]);
 68:                     // attributes
 69:                     $attributes = array();
 70:                     if (!empty($a2[3]) && preg_match_all('/(\\w+)\s*=\\s*("[^"]*"|\'[^\']*\'|[^"\'\\s>]*)/', $a2[3], $matches, PREG_SET_ORDER)) {
 71:                         foreach ($matches as $match) {
 72:                             if (($match[2][0] == '"' || $match[2][0] == "'") && $match[2][0] == $match[2][strlen($match[2]) - 1]) {
 73:                                 $match[2] = substr($match[2], 1, -1);
 74:                             }
 75:                             $attributes[strtoupper($match[1])] = $match[2];
 76:                         }
 77:                     }
 78:                     if ($tag == 'BR' || $tag == 'IMG' || $tag == 'INPUT' || $tag == 'HR') {
 79:                         if ($node) {
 80:                             $node->childNodes[] = new DOMNode($tag, $node, $attributes);
 81:                         } else {
 82:                             $this->doms[] = new DOMNode($tag, $node, $attributes);
 83:                         }
 84:                     } else {
 85:                         $temp = $node;
 86:                         $node = new DOMNode($tag, $temp, $attributes);
 87:                         if ($temp) {
 88:                             $temp->childNodes[] = $node;
 89:                         } else {
 90:                             $this->doms[] = $node;
 91:                         }
 92:                     }
 93:                 }
 94:             }
 95:         }
 96:         // ตรวจสอบ previousSibling และ nextSibling
 97:         $currentNode = null;
 98:         foreach ($this->doms as $node) {
 99:             $node->previousSibling = $currentNode;
100:             $this->populate($node);
101:             if ($node->previousSibling) {
102:                 $node->previousSibling->nextSibling = $node;
103:             }
104:             $currentNode = $node;
105:         }
106:     }
107: 
108:     /**
109:      * parse HTML จาก URL.
110:      *
111:      * @param string $url URL ที่ต้องการ parse
112:      *
113:      * @return \static
114:      */
115:     public static function load($url)
116:     {
117:         $obj = new static(file_get_contents($url));
118: 
119:         return $obj;
120:     }
121: 
122:     /**
123:      * คืนค่า node ทั้งหมด.
124:      *
125:      * @return array
126:      */
127:     public function nodes()
128:     {
129:         return $this->doms;
130:     }
131: 
132:     /**
133:      * ส่งออกเป็นโค้ด HTML จากที่ parse แล้ว.
134:      *
135:      * @return string
136:      */
137:     public function toHTML()
138:     {
139:         $html = '';
140:         foreach ($this->doms as $node) {
141:             $html .= $this->drawNode($node);
142:         }
143: 
144:         return $html;
145:     }
146: 
147:     /**
148:      * @param DOMNode $node
149:      *
150:      * @return string
151:      */
152:     private function drawNode($node)
153:     {
154:         $html = '';
155:         if ($node->nodeName == '') {
156:             $html .= $node->nodeValue;
157:         } else {
158:             $prop = array();
159:             foreach ($node->attributes as $k => $v) {
160:                 $prop[] = $k.'="'.$v.'"';
161:             }
162:             if ($node->nodeName == 'BR' || $node->nodeName == 'IMG') {
163:                 $html .= '<'.$node->nodeName.' '.implode(' ', $prop).'>';
164:             } else {
165:                 $prop = empty($prop) ? '' : ' '.implode(' ', $prop);
166:                 $html .= '<'.$node->nodeName.$prop.'>';
167:                 if (empty($node->childNodes)) {
168:                     $html .= $node->nodeValue;
169:                 } else {
170:                     foreach ($node->childNodes as $child) {
171:                         $html .= $this->drawNode($child);
172:                     }
173:                 }
174:                 if ($node->previousSibling) {
175:                     $html .= '('.$node->previousSibling->nodeName.')';
176:                 }
177:                 $html .= '</'.$node->nodeName.'>';
178:             }
179:         }
180: 
181:         return $html;
182:     }
183: 
184:     /**
185:      * ตรวจสอบ previousSibling และ nextSibling.
186:      *
187:      * @param DOMNode $node
188:      */
189:     private function populate($node)
190:     {
191:         $currentNode = null;
192:         foreach ($node->childNodes as $item) {
193:             $item->previousSibling = $currentNode;
194:             foreach ($item->childNodes as $child) {
195:                 $this->populate($child);
196:                 if ($node->previousSibling) {
197:                     $node->previousSibling->nextSibling = $node;
198:                 }
199:             }
200:             $currentNode = $item;
201:         }
202:     }
203: }
204: