| 1 |
<?php |
| 2 |
|
| 3 |
namespace simplehtmldom; |
| 4 |
|
| 5 |
if (! defined('ABSPATH')) exit; // Exit if accessed directly |
| 6 |
/** |
| 7 |
* Website: http://sourceforge.net/projects/simplehtmldom/ |
| 8 |
* Acknowledge: Jose Solorzano (https://sourceforge.net/projects/php-html/) |
| 9 |
* |
| 10 |
* Licensed under The MIT License |
| 11 |
* See the LICENSE file in the project root for more information. |
| 12 |
* |
| 13 |
* Authors: |
| 14 |
* S.C. Chen |
| 15 |
* John Schlick |
| 16 |
* Rus Carroll |
| 17 |
* logmanoriginal |
| 18 |
* |
| 19 |
* Contributors: |
| 20 |
* Yousuke Kumakura |
| 21 |
* Vadim Voituk |
| 22 |
* Antcs |
| 23 |
* |
| 24 |
* Version Rev. 2.0-RC2 (415) |
| 25 |
*/ |
| 26 |
|
| 27 |
include_once 'constants.php'; |
| 28 |
include_once 'HtmlNode.php'; |
| 29 |
include_once 'Debug.php'; |
| 30 |
|
| 31 |
class HtmlDocument |
| 32 |
{ |
| 33 |
public $root = null; |
| 34 |
public $nodes = array(); |
| 35 |
public $callback = null; |
| 36 |
public $lowercase = false; |
| 37 |
public $original_size; |
| 38 |
public $size; |
| 39 |
|
| 40 |
protected $pos; |
| 41 |
protected $doc; |
| 42 |
protected $char; |
| 43 |
|
| 44 |
protected $cursor; |
| 45 |
protected $parent; |
| 46 |
protected $noise = array(); |
| 47 |
protected $token_blank = " \t\r\n"; |
| 48 |
protected $token_equal = ' =/>'; |
| 49 |
protected $token_slash = " />\r\n\t"; |
| 50 |
protected $token_attr = ' >'; |
| 51 |
|
| 52 |
public $_charset = ''; |
| 53 |
public $_target_charset = ''; |
| 54 |
|
| 55 |
public $default_br_text = ''; |
| 56 |
public $default_span_text = ''; |
| 57 |
|
| 58 |
protected $self_closing_tags = array( |
| 59 |
'area' => 1, |
| 60 |
'base' => 1, |
| 61 |
'br' => 1, |
| 62 |
'col' => 1, |
| 63 |
'embed' => 1, |
| 64 |
'hr' => 1, |
| 65 |
'img' => 1, |
| 66 |
'input' => 1, |
| 67 |
'link' => 1, |
| 68 |
'meta' => 1, |
| 69 |
'param' => 1, |
| 70 |
'source' => 1, |
| 71 |
'track' => 1, |
| 72 |
'wbr' => 1 |
| 73 |
); |
| 74 |
protected $block_tags = array( |
| 75 |
'body' => 1, |
| 76 |
'div' => 1, |
| 77 |
'form' => 1, |
| 78 |
'root' => 1, |
| 79 |
'span' => 1, |
| 80 |
'table' => 1 |
| 81 |
); |
| 82 |
protected $optional_closing_tags = array( |
| 83 |
// Not optional, see |
| 84 |
// https://www.w3.org/TR/html/textlevel-semantics.html#the-b-element |
| 85 |
'b' => array('b' => 1), |
| 86 |
'dd' => array('dd' => 1, 'dt' => 1), |
| 87 |
// Not optional, see |
| 88 |
// https://www.w3.org/TR/html/grouping-content.html#the-dl-element |
| 89 |
'dl' => array('dd' => 1, 'dt' => 1), |
| 90 |
'dt' => array('dd' => 1, 'dt' => 1), |
| 91 |
'li' => array('li' => 1), |
| 92 |
'optgroup' => array('optgroup' => 1, 'option' => 1), |
| 93 |
'option' => array('optgroup' => 1, 'option' => 1), |
| 94 |
'p' => array('p' => 1), |
| 95 |
'rp' => array('rp' => 1, 'rt' => 1), |
| 96 |
'rt' => array('rp' => 1, 'rt' => 1), |
| 97 |
'td' => array('td' => 1, 'th' => 1), |
| 98 |
'th' => array('td' => 1, 'th' => 1), |
| 99 |
'tr' => array('td' => 1, 'th' => 1, 'tr' => 1), |
| 100 |
); |
| 101 |
|
| 102 |
function __call($func, $args) |
| 103 |
{ |
| 104 |
// Allow users to call methods with lower_case syntax |
| 105 |
switch ($func) { |
| 106 |
case 'load_file': |
| 107 |
$actual_function = 'loadFile'; |
| 108 |
break; |
| 109 |
case 'clear': |
| 110 |
return; /* no-op */ |
| 111 |
default: |
| 112 |
// trigger_error( |
| 113 |
// 'Call to undefined method ' . __CLASS__ . '::' . $func . '()', //phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped |
| 114 |
// E_USER_ERROR |
| 115 |
// ); |
| 116 |
} |
| 117 |
|
| 118 |
// phpcs:ignore Generic.Files.LineLength |
| 119 |
Debug::log(__CLASS__ . '->' . $func . '() has been deprecated and will be removed in the next major version of simplehtmldom. Use ' . __CLASS__ . '->' . $actual_function . '() instead.'); |
| 120 |
|
| 121 |
return call_user_func_array(array($this, $actual_function), $args); |
| 122 |
} |
| 123 |
|
| 124 |
function __construct( |
| 125 |
$str = null, |
| 126 |
$lowercase = true, |
| 127 |
$forceTagsClosed = true, |
| 128 |
$target_charset = DEFAULT_TARGET_CHARSET, |
| 129 |
$stripRN = true, |
| 130 |
$defaultBRText = DEFAULT_BR_TEXT, |
| 131 |
$defaultSpanText = DEFAULT_SPAN_TEXT, |
| 132 |
$options = 0 |
| 133 |
) { |
| 134 |
if ($str) { |
| 135 |
if (preg_match('/^http:\/\//i', $str) || is_file($str)) { |
| 136 |
$this->load_file($str); |
| 137 |
} else { |
| 138 |
$this->load( |
| 139 |
$str, |
| 140 |
$lowercase, |
| 141 |
$stripRN, |
| 142 |
$defaultBRText, |
| 143 |
$defaultSpanText, |
| 144 |
$options |
| 145 |
); |
| 146 |
} |
| 147 |
} else { |
| 148 |
$this->prepare($str, $lowercase, $defaultBRText, $defaultSpanText); |
| 149 |
} |
| 150 |
// Forcing tags to be closed implies that we don't trust the html, but |
| 151 |
// it can lead to parsing errors if we SHOULD trust the html. |
| 152 |
if (!$forceTagsClosed) { |
| 153 |
$this->optional_closing_array = array(); |
| 154 |
} |
| 155 |
|
| 156 |
$this->_target_charset = $target_charset; |
| 157 |
} |
| 158 |
|
| 159 |
function __debugInfo() |
| 160 |
{ |
| 161 |
return array( |
| 162 |
'root' => $this->root, |
| 163 |
'noise' => empty($this->noise) ? 'none' : $this->noise, |
| 164 |
'charset' => $this->_charset, |
| 165 |
'target charset' => $this->_target_charset, |
| 166 |
'original size' => $this->original_size |
| 167 |
); |
| 168 |
} |
| 169 |
|
| 170 |
function __destruct() |
| 171 |
{ |
| 172 |
if (isset($this->nodes)) { |
| 173 |
foreach ($this->nodes as $n) { |
| 174 |
$n->clear(); |
| 175 |
} |
| 176 |
} |
| 177 |
} |
| 178 |
|
| 179 |
function load( |
| 180 |
$str, |
| 181 |
$lowercase = true, |
| 182 |
$stripRN = true, |
| 183 |
$defaultBRText = DEFAULT_BR_TEXT, |
| 184 |
$defaultSpanText = DEFAULT_SPAN_TEXT, |
| 185 |
$options = 0 |
| 186 |
) { |
| 187 |
// prepare |
| 188 |
$this->prepare($str, $lowercase, $defaultBRText, $defaultSpanText); |
| 189 |
|
| 190 |
if ($stripRN) { |
| 191 |
// Temporarily remove any element that shouldn't loose whitespace |
| 192 |
$this->remove_noise("'<\s*script[^>]*>(.*?)<\s*/\s*script\s*>'is"); |
| 193 |
$this->remove_noise("'<!\[CDATA\[(.*?)\]\]>'is"); |
| 194 |
$this->remove_noise("'<!--(.*?)-->'is"); |
| 195 |
$this->remove_noise("'<\s*style[^>]*>(.*?)<\s*/\s*style\s*>'is"); |
| 196 |
$this->remove_noise("'<\s*(?:code)[^>]*>(.*?)<\s*/\s*(?:code)\s*>'is"); |
| 197 |
|
| 198 |
// Remove whitespace and newlines between tags |
| 199 |
$this->doc = preg_replace('/\>([\t\s]*[\r\n]^[\t\s]*)\</m', '><', $this->doc); |
| 200 |
|
| 201 |
// Remove whitespace and newlines in text |
| 202 |
$this->doc = preg_replace('/([\t\s]*[\r\n]^[\t\s]*)/m', ' ', $this->doc); |
| 203 |
|
| 204 |
// Restore temporarily removed elements and calculate new size |
| 205 |
$this->doc = $this->restore_noise($this->doc); |
| 206 |
$this->size = strlen($this->doc); |
| 207 |
} |
| 208 |
|
| 209 |
$this->remove_noise("'(<\?)(.*?)(\?>)'s", true); // server-side script |
| 210 |
if (count($this->noise)) { |
| 211 |
// phpcs:ignore Generic.Files.LineLength |
| 212 |
Debug::log('Support for server-side scripts has been deprecated and will be removed in the next major version of simplehtmldom.'); |
| 213 |
} |
| 214 |
|
| 215 |
if ($options & HDOM_SMARTY_AS_TEXT) { // Strip Smarty scripts |
| 216 |
$this->remove_noise("'(\{\w)(.*?)(\})'s", true); |
| 217 |
// phpcs:ignore Generic.Files.LineLength |
| 218 |
Debug::log('Support for Smarty scripts has been deprecated and will be removed in the next major version of simplehtmldom.'); |
| 219 |
} |
| 220 |
|
| 221 |
// parsing |
| 222 |
$this->parse($stripRN); |
| 223 |
// end |
| 224 |
$this->root->_[HtmlNode::HDOM_INFO_END] = $this->cursor; |
| 225 |
$this->parse_charset(); |
| 226 |
$this->decode(); |
| 227 |
unset($this->doc); |
| 228 |
|
| 229 |
// make load function chainable |
| 230 |
return $this; |
| 231 |
} |
| 232 |
|
| 233 |
function set_callback($function_name) |
| 234 |
{ |
| 235 |
$this->callback = $function_name; |
| 236 |
} |
| 237 |
|
| 238 |
function remove_callback() |
| 239 |
{ |
| 240 |
$this->callback = null; |
| 241 |
} |
| 242 |
|
| 243 |
function save($filepath = '') |
| 244 |
{ |
| 245 |
$ret = $this->root->innertext(); |
| 246 |
if ($filepath !== '') { |
| 247 |
file_put_contents($filepath, $ret, LOCK_EX); |
| 248 |
} |
| 249 |
return $ret; |
| 250 |
} |
| 251 |
|
| 252 |
function find($selector, $idx = null, $lowercase = false) |
| 253 |
{ |
| 254 |
return $this->root->find($selector, $idx, $lowercase); |
| 255 |
} |
| 256 |
|
| 257 |
function expect($selector, $idx = null, $lowercase = false) |
| 258 |
{ |
| 259 |
return $this->root->expect($selector, $idx, $lowercase); |
| 260 |
} |
| 261 |
|
| 262 |
/** @codeCoverageIgnore */ |
| 263 |
function dump($show_attr = true) |
| 264 |
{ |
| 265 |
$this->root->dump($show_attr); |
| 266 |
} |
| 267 |
|
| 268 |
protected function prepare( |
| 269 |
$str, |
| 270 |
$lowercase = true, |
| 271 |
$defaultBRText = DEFAULT_BR_TEXT, |
| 272 |
$defaultSpanText = DEFAULT_SPAN_TEXT |
| 273 |
) { |
| 274 |
$this->clear(); |
| 275 |
|
| 276 |
$this->doc = trim((string) $str); |
| 277 |
$this->size = strlen($this->doc); |
| 278 |
$this->original_size = $this->size; // original size of the html |
| 279 |
$this->pos = 0; |
| 280 |
$this->cursor = 1; |
| 281 |
$this->noise = array(); |
| 282 |
$this->nodes = array(); |
| 283 |
$this->lowercase = $lowercase; |
| 284 |
$this->default_br_text = $defaultBRText; |
| 285 |
$this->default_span_text = $defaultSpanText; |
| 286 |
$this->root = new HtmlNode($this); |
| 287 |
$this->root->tag = 'root'; |
| 288 |
$this->root->_[HtmlNode::HDOM_INFO_BEGIN] = -1; |
| 289 |
$this->root->nodetype = HtmlNode::HDOM_TYPE_ROOT; |
| 290 |
$this->parent = $this->root; |
| 291 |
if ($this->size > 0) { |
| 292 |
$this->char = $this->doc[0]; |
| 293 |
} |
| 294 |
} |
| 295 |
|
| 296 |
protected function decode() |
| 297 |
{ |
| 298 |
foreach ($this->nodes as $node) { |
| 299 |
if (isset($node->_[HtmlNode::HDOM_INFO_TEXT])) { |
| 300 |
$node->_[HtmlNode::HDOM_INFO_TEXT] = html_entity_decode( |
| 301 |
$this->restore_noise($node->_[HtmlNode::HDOM_INFO_TEXT]), |
| 302 |
ENT_QUOTES | ENT_HTML5, |
| 303 |
$this->_target_charset |
| 304 |
); |
| 305 |
} |
| 306 |
if (isset($node->_[HtmlNode::HDOM_INFO_INNER])) { |
| 307 |
$node->_[HtmlNode::HDOM_INFO_INNER] = html_entity_decode( |
| 308 |
$this->restore_noise($node->_[HtmlNode::HDOM_INFO_INNER]), |
| 309 |
ENT_QUOTES | ENT_HTML5, |
| 310 |
$this->_target_charset |
| 311 |
); |
| 312 |
} |
| 313 |
if (isset($node->attr) && is_array($node->attr)) { |
| 314 |
foreach ($node->attr as $a => $v) { |
| 315 |
if ($v === true) continue; |
| 316 |
$node->attr[$a] = html_entity_decode( |
| 317 |
$v, |
| 318 |
ENT_QUOTES | ENT_HTML5, |
| 319 |
$this->_target_charset |
| 320 |
); |
| 321 |
} |
| 322 |
} |
| 323 |
} |
| 324 |
} |
| 325 |
|
| 326 |
protected function parse($trim = false) |
| 327 |
{ |
| 328 |
while (true) { |
| 329 |
|
| 330 |
if ($this->char !== '<') { |
| 331 |
$content = $this->copy_until_char('<'); |
| 332 |
|
| 333 |
if ($content !== '') { |
| 334 |
|
| 335 |
// Skip whitespace between tags? (</a> <b>) |
| 336 |
if ($trim && trim($content) === '') { |
| 337 |
continue; |
| 338 |
} |
| 339 |
|
| 340 |
$node = new HtmlNode($this); |
| 341 |
++$this->cursor; |
| 342 |
$node->_[HtmlNode::HDOM_INFO_TEXT] = $content; |
| 343 |
$this->link_nodes($node, false); |
| 344 |
} |
| 345 |
} |
| 346 |
|
| 347 |
if ($this->read_tag($trim) === false) { |
| 348 |
break; |
| 349 |
} |
| 350 |
} |
| 351 |
} |
| 352 |
|
| 353 |
protected function parse_charset() |
| 354 |
{ |
| 355 |
$charset = null; |
| 356 |
|
| 357 |
if (function_exists('get_last_retrieve_url_contents_content_type')) { |
| 358 |
$contentTypeHeader = get_last_retrieve_url_contents_content_type(); |
| 359 |
$success = preg_match('/charset=(.+)/', $contentTypeHeader, $matches); |
| 360 |
if ($success) { |
| 361 |
$charset = $matches[1]; |
| 362 |
} |
| 363 |
|
| 364 |
// phpcs:ignore Generic.Files.LineLength |
| 365 |
Debug::log('Determining charset using get_last_retrieve_url_contents_content_type() ' . ($success ? 'successful' : 'failed')); |
| 366 |
} |
| 367 |
|
| 368 |
if (empty($charset)) { |
| 369 |
// https://www.w3.org/TR/html/document-metadata.html#statedef-http-equiv-content-type |
| 370 |
$el = $this->root->find('meta[http-equiv=Content-Type]', 0, true); |
| 371 |
|
| 372 |
if (!empty($el)) { |
| 373 |
$fullvalue = $el->content; |
| 374 |
|
| 375 |
if (!empty($fullvalue)) { |
| 376 |
$success = preg_match( |
| 377 |
'/charset=(.+)/i', |
| 378 |
$fullvalue, |
| 379 |
$matches |
| 380 |
); |
| 381 |
|
| 382 |
if ($success) { |
| 383 |
$charset = $matches[1]; |
| 384 |
} |
| 385 |
} |
| 386 |
} |
| 387 |
} |
| 388 |
|
| 389 |
if (empty($charset)) { |
| 390 |
// https://www.w3.org/TR/html/document-metadata.html#character-encoding-declaration |
| 391 |
if ($meta = $this->root->find('meta[charset]', 0)) { |
| 392 |
$charset = $meta->charset; |
| 393 |
} |
| 394 |
} |
| 395 |
|
| 396 |
if (empty($charset)) { |
| 397 |
// Try to guess the charset based on the content |
| 398 |
// Requires Multibyte String (mbstring) support (optional) |
| 399 |
if (function_exists('mb_detect_encoding')) { |
| 400 |
/** |
| 401 |
* mb_detect_encoding() is not intended to distinguish between |
| 402 |
* charsets, especially single-byte charsets. Its primary |
| 403 |
* purpose is to detect which multibyte encoding is in use, |
| 404 |
* i.e. UTF-8, UTF-16, shift-JIS, etc. |
| 405 |
* |
| 406 |
* -- https://bugs.php.net/bug.php?id=38138 |
| 407 |
* |
| 408 |
* Adding both CP1251/ISO-8859-5 and CP1252/ISO-8859-1 will |
| 409 |
* always result in CP1251/ISO-8859-5 and vice versa. |
| 410 |
* |
| 411 |
* Thus, only detect if it's either UTF-8 or CP1252/ISO-8859-1 |
| 412 |
* to stay compatible. |
| 413 |
*/ |
| 414 |
$encoding = mb_detect_encoding( |
| 415 |
$this->doc, |
| 416 |
array('UTF-8', 'CP1252', 'ISO-8859-1') |
| 417 |
); |
| 418 |
|
| 419 |
if ($encoding === 'CP1252' || $encoding === 'ISO-8859-1') { |
| 420 |
// Due to a limitation of mb_detect_encoding |
| 421 |
// 'CP1251'/'ISO-8859-5' will be detected as |
| 422 |
// 'CP1252'/'ISO-8859-1'. This will cause iconv to fail, in |
| 423 |
// which case we can simply assume it is the other charset. |
| 424 |
if (!@iconv('CP1252', 'UTF-8', $this->doc)) { |
| 425 |
$encoding = 'CP1251'; |
| 426 |
} |
| 427 |
} |
| 428 |
|
| 429 |
if ($encoding !== false) { |
| 430 |
$charset = $encoding; |
| 431 |
} |
| 432 |
} |
| 433 |
} |
| 434 |
|
| 435 |
if (empty($charset)) { |
| 436 |
Debug::log('Unable to determine charset from source document. Assuming UTF-8'); |
| 437 |
$charset = 'UTF-8'; |
| 438 |
} |
| 439 |
|
| 440 |
// Since CP1252 is a superset, if we get one of it's subsets, we want |
| 441 |
// it instead. |
| 442 |
if ((strtolower($charset) == 'iso-8859-1') |
| 443 |
|| (strtolower($charset) == 'latin1') |
| 444 |
|| (strtolower($charset) == 'latin-1') |
| 445 |
) { |
| 446 |
$charset = 'CP1252'; |
| 447 |
} |
| 448 |
|
| 449 |
return $this->_charset = $charset; |
| 450 |
} |
| 451 |
|
| 452 |
protected function read_tag($trim) |
| 453 |
{ |
| 454 |
if ($this->char !== '<') { // End Of File |
| 455 |
$this->root->_[HtmlNode::HDOM_INFO_END] = $this->cursor; |
| 456 |
|
| 457 |
// We might be in a nest of unclosed elements for which the end tags |
| 458 |
// can be omitted. Close them for faster seek operations. |
| 459 |
do { |
| 460 |
if (isset($this->optional_closing_tags[strtolower($this->parent->tag)])) { |
| 461 |
$this->parent->_[HtmlNode::HDOM_INFO_END] = $this->cursor; |
| 462 |
} |
| 463 |
} while ($this->parent = $this->parent->parent); |
| 464 |
|
| 465 |
return false; |
| 466 |
} |
| 467 |
|
| 468 |
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 469 |
|
| 470 |
if ($trim) { // "< /html>" |
| 471 |
$this->skip($this->token_blank); |
| 472 |
} |
| 473 |
|
| 474 |
// End tag: https://dev.w3.org/html5/pf-summary/syntax.html#end-tags |
| 475 |
if ($this->char === '/') { |
| 476 |
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 477 |
|
| 478 |
$tag = $this->copy_until_char('>'); |
| 479 |
$tag = $trim ? ltrim($tag, $this->token_blank) : $tag; |
| 480 |
|
| 481 |
// Skip attributes and whitespace in end tags |
| 482 |
if ($trim && ($pos = strpos($tag, ' ')) !== false) { |
| 483 |
// phpcs:ignore Generic.Files.LineLength |
| 484 |
Debug::log_once('Source document contains superfluous whitespace in end tags (</html >).'); |
| 485 |
$tag = substr($tag, 0, $pos); |
| 486 |
} |
| 487 |
|
| 488 |
if (strcasecmp($this->parent->tag, $tag)) { // Parent is not start tag |
| 489 |
$parent_lower = strtolower($this->parent->tag); |
| 490 |
$tag_lower = strtolower($tag); |
| 491 |
if (isset($this->optional_closing_tags[$parent_lower]) && isset($this->block_tags[$tag_lower])) { |
| 492 |
$org_parent = $this->parent; |
| 493 |
|
| 494 |
// Look for the start tag |
| 495 |
while (($this->parent->parent) && strtolower($this->parent->tag) !== $tag_lower) { |
| 496 |
// Close any unclosed element with optional end tags |
| 497 |
if (isset($this->optional_closing_tags[strtolower($this->parent->tag)])) |
| 498 |
$this->parent->_[HtmlNode::HDOM_INFO_END] = $this->cursor; |
| 499 |
$this->parent = $this->parent->parent; |
| 500 |
} |
| 501 |
|
| 502 |
// No start tag, close grandparent |
| 503 |
if (strtolower($this->parent->tag) !== $tag_lower) { |
| 504 |
$this->parent = $org_parent; |
| 505 |
|
| 506 |
if ($this->parent->parent) { |
| 507 |
$this->parent = $this->parent->parent; |
| 508 |
} |
| 509 |
|
| 510 |
$this->parent->_[HtmlNode::HDOM_INFO_END] = $this->cursor; |
| 511 |
return $this->as_text_node($tag); |
| 512 |
} |
| 513 |
} elseif (($this->parent->parent) && isset($this->block_tags[$tag_lower])) { |
| 514 |
// grandparent exists + current is block tag |
| 515 |
// Parent has no end tag |
| 516 |
$this->parent->_[HtmlNode::HDOM_INFO_END] = 0; |
| 517 |
$org_parent = $this->parent; |
| 518 |
|
| 519 |
// Find start tag |
| 520 |
while (($this->parent->parent) && strtolower($this->parent->tag) !== $tag_lower) { |
| 521 |
$this->parent = $this->parent->parent; |
| 522 |
} |
| 523 |
|
| 524 |
// No start tag, close parent |
| 525 |
if (strtolower($this->parent->tag) !== $tag_lower) { |
| 526 |
$this->parent = $org_parent; // restore origonal parent |
| 527 |
$this->parent->_[HtmlNode::HDOM_INFO_END] = $this->cursor; |
| 528 |
return $this->as_text_node($tag); |
| 529 |
} |
| 530 |
} elseif (($this->parent->parent) && strtolower($this->parent->parent->tag) === $tag_lower) { |
| 531 |
// Grandparent exists and current tag closes it |
| 532 |
$this->parent->_[HtmlNode::HDOM_INFO_END] = 0; |
| 533 |
$this->parent = $this->parent->parent; |
| 534 |
} else { // Random tag, add as text node |
| 535 |
return $this->as_text_node($tag); |
| 536 |
} |
| 537 |
} |
| 538 |
|
| 539 |
// Link with start tag |
| 540 |
$this->parent->_[HtmlNode::HDOM_INFO_END] = $this->cursor; |
| 541 |
|
| 542 |
if ($this->parent->parent) { |
| 543 |
$this->parent = $this->parent->parent; |
| 544 |
} |
| 545 |
|
| 546 |
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 547 |
return true; |
| 548 |
} |
| 549 |
|
| 550 |
// Start tag: https://dev.w3.org/html5/pf-summary/syntax.html#start-tags |
| 551 |
$node = new HtmlNode($this); |
| 552 |
$node->_[HtmlNode::HDOM_INFO_BEGIN] = $this->cursor++; |
| 553 |
|
| 554 |
// Tag name |
| 555 |
$tag = $this->copy_until($this->token_slash); |
| 556 |
|
| 557 |
if (isset($tag[0]) && $tag[0] === '!') { // Doctype, CData, Comment |
| 558 |
if (isset($tag[2]) && $tag[1] === '-' && $tag[2] === '-') { // Comment ("<!--") |
| 559 |
|
| 560 |
// Go back until $tag only contains start of comment "!--". |
| 561 |
while (strlen($tag) > 3) { |
| 562 |
$this->char = $this->doc[--$this->pos]; // previous |
| 563 |
$tag = substr($tag, 0, strlen($tag) - 1); |
| 564 |
} |
| 565 |
|
| 566 |
$node->nodetype = HtmlNode::HDOM_TYPE_COMMENT; |
| 567 |
$node->tag = 'comment'; |
| 568 |
|
| 569 |
$data = ''; |
| 570 |
|
| 571 |
// There is a rare chance of empty comment: "<!---->" |
| 572 |
// In which case the current char is the first "-" of the end tag |
| 573 |
// But the comment could also just be a dash: "<!----->" |
| 574 |
while (true) { |
| 575 |
// Copy until first char of end tag |
| 576 |
$data .= $this->copy_until_char('-'); |
| 577 |
|
| 578 |
// Look ahead in the document, maybe we are at the end |
| 579 |
if (($this->pos + 3) > $this->size) { // End of document |
| 580 |
Debug::log('Source document ended unexpectedly!'); |
| 581 |
break; |
| 582 |
} elseif (substr($this->doc, $this->pos, 3) === '-->') { // end |
| 583 |
$data .= $this->copy_until_char('>'); |
| 584 |
break; |
| 585 |
} |
| 586 |
|
| 587 |
$data .= $this->char; |
| 588 |
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 589 |
} |
| 590 |
|
| 591 |
$tag .= $data; |
| 592 |
$tag = $this->restore_noise($tag); |
| 593 |
|
| 594 |
// Comment starts after "!--" and ends before "--" (5 chars total) |
| 595 |
$node->_[HtmlNode::HDOM_INFO_INNER] = substr($tag, 3, strlen($tag) - 5); |
| 596 |
} elseif (substr($tag, 1, 7) === '[CDATA[') { |
| 597 |
|
| 598 |
// Go back until $tag only contains start of cdata "![CDATA[". |
| 599 |
while (strlen($tag) > 8) { |
| 600 |
$this->char = $this->doc[--$this->pos]; // previous |
| 601 |
$tag = substr($tag, 0, strlen($tag) - 1); |
| 602 |
} |
| 603 |
|
| 604 |
// CDATA can contain HTML stuff, need to find closing tags first |
| 605 |
$node->nodetype = HtmlNode::HDOM_TYPE_CDATA; |
| 606 |
$node->tag = 'cdata'; |
| 607 |
|
| 608 |
$data = ''; |
| 609 |
|
| 610 |
// There is a rare chance of empty CDATA: "<[CDATA[]]>" |
| 611 |
// In which case the current char is the first "[" of the end tag |
| 612 |
// But the CDATA could also just be a bracket: "<[CDATA[]]]>" |
| 613 |
while (true) { |
| 614 |
// Copy until first char of end tag |
| 615 |
$data .= $this->copy_until_char(']'); |
| 616 |
|
| 617 |
// Look ahead in the document, maybe we are at the end |
| 618 |
if (($this->pos + 3) > $this->size) { // End of document |
| 619 |
Debug::log('Source document ended unexpectedly!'); |
| 620 |
break; |
| 621 |
} elseif (substr($this->doc, $this->pos, 3) === ']]>') { // end |
| 622 |
$data .= $this->copy_until_char('>'); |
| 623 |
break; |
| 624 |
} |
| 625 |
|
| 626 |
$data .= $this->char; |
| 627 |
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 628 |
} |
| 629 |
|
| 630 |
$tag .= $data; |
| 631 |
$tag = $this->restore_noise($tag); |
| 632 |
|
| 633 |
// CDATA starts after "![CDATA[" and ends before "]]" (10 chars total) |
| 634 |
$node->_[HtmlNode::HDOM_INFO_INNER] = substr($tag, 8, strlen($tag) - 10); |
| 635 |
} else { // Unknown |
| 636 |
Debug::log('Source document contains unknown declaration: <' . $tag); |
| 637 |
$node->nodetype = HtmlNode::HDOM_TYPE_UNKNOWN; |
| 638 |
$node->tag = 'unknown'; |
| 639 |
} |
| 640 |
|
| 641 |
$node->_[HtmlNode::HDOM_INFO_TEXT] = '<' . $tag . $this->copy_until_char('>'); |
| 642 |
|
| 643 |
if ($this->char === '>') { |
| 644 |
$node->_[HtmlNode::HDOM_INFO_TEXT] .= '>'; |
| 645 |
} |
| 646 |
|
| 647 |
$this->link_nodes($node, true); |
| 648 |
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 649 |
return true; |
| 650 |
} |
| 651 |
|
| 652 |
if (!preg_match('/^\w[\w:-]*$/', $tag)) { // Invalid tag name |
| 653 |
$node->_[HtmlNode::HDOM_INFO_TEXT] = '<' . $tag . $this->copy_until('<>'); |
| 654 |
|
| 655 |
if ($this->char === '>') { // End tag |
| 656 |
$node->_[HtmlNode::HDOM_INFO_TEXT] .= '>'; |
| 657 |
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 658 |
} |
| 659 |
|
| 660 |
$this->link_nodes($node, false); |
| 661 |
Debug::log('Source document contains invalid tag name: ' . $node->_[HtmlNode::HDOM_INFO_TEXT]); |
| 662 |
return true; |
| 663 |
} |
| 664 |
|
| 665 |
// Valid tag name |
| 666 |
$node->nodetype = HtmlNode::HDOM_TYPE_ELEMENT; |
| 667 |
$tag_lower = strtolower($tag); |
| 668 |
$node->tag = ($this->lowercase) ? $tag_lower : $tag; |
| 669 |
|
| 670 |
if (isset($this->optional_closing_tags[$tag_lower])) { // Optional closing tag |
| 671 |
while (isset($this->optional_closing_tags[$tag_lower][strtolower($this->parent->tag)])) { |
| 672 |
// Previous element was the last element of ancestor |
| 673 |
$this->parent->_[HtmlNode::HDOM_INFO_END] = $node->_[HtmlNode::HDOM_INFO_BEGIN] - 1; |
| 674 |
$this->parent = $this->parent->parent; |
| 675 |
} |
| 676 |
$node->parent = $this->parent; |
| 677 |
} |
| 678 |
|
| 679 |
$guard = 0; // prevent infinity loop |
| 680 |
|
| 681 |
// [0] Space between tag and first attribute |
| 682 |
$space = array($this->copy_skip($this->token_blank), '', ''); |
| 683 |
|
| 684 |
do { // Parse attributes |
| 685 |
$name = $this->copy_until($this->token_equal); |
| 686 |
|
| 687 |
if ($name === '' && $this->char !== null && $space[0] === '') { |
| 688 |
break; |
| 689 |
} |
| 690 |
|
| 691 |
if ($guard === $this->pos) { // Escape infinite loop |
| 692 |
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 693 |
continue; |
| 694 |
} |
| 695 |
|
| 696 |
$guard = $this->pos; |
| 697 |
|
| 698 |
if ($this->pos >= $this->size - 1 && $this->char !== '>') { // End Of File |
| 699 |
Debug::log('Source document ended unexpectedly!'); |
| 700 |
$node->nodetype = HtmlNode::HDOM_TYPE_TEXT; |
| 701 |
$node->_[HtmlNode::HDOM_INFO_END] = 0; |
| 702 |
$node->_[HtmlNode::HDOM_INFO_TEXT] = '<' . $tag . $space[0] . $name; |
| 703 |
$node->tag = 'text'; |
| 704 |
$this->link_nodes($node, false); |
| 705 |
return true; |
| 706 |
} |
| 707 |
|
| 708 |
if ($name === '/' || $name === '') { // No more attributes |
| 709 |
break; |
| 710 |
} |
| 711 |
|
| 712 |
// [1] Whitespace after attribute name |
| 713 |
$space[1] = (strpos($this->token_blank, $this->char) === false) ? '' : $this->copy_skip($this->token_blank); |
| 714 |
|
| 715 |
$name = $this->restore_noise($name); // might be a noisy name |
| 716 |
|
| 717 |
if ($this->lowercase) { |
| 718 |
$name = strtolower($name); |
| 719 |
} |
| 720 |
|
| 721 |
if ($this->char === '=') { // Attribute with value |
| 722 |
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 723 |
$this->parse_attr($node, $name, $space, $trim); // get attribute value |
| 724 |
} else { // Attribute without value |
| 725 |
$node->_[HtmlNode::HDOM_INFO_QUOTE][$name] = HtmlNode::HDOM_QUOTE_NO; |
| 726 |
$node->attr[$name] = true; |
| 727 |
if ($this->char !== '>') { |
| 728 |
$this->char = $this->doc[--$this->pos]; |
| 729 |
} // prev |
| 730 |
} |
| 731 |
|
| 732 |
// Space before attribute and around equal sign |
| 733 |
if (!$trim && $space !== array(' ', '', '')) { |
| 734 |
// phpcs:ignore Generic.Files.LineLength |
| 735 |
Debug::log_once('Source document contains superfluous whitespace in attributes (<e attribute = "value">). Enable trimming or fix attribute spacing for best performance.'); |
| 736 |
$node->_[HtmlNode::HDOM_INFO_SPACE][$name] = $space; |
| 737 |
} |
| 738 |
|
| 739 |
// prepare for next attribute |
| 740 |
$space = array( |
| 741 |
((strpos($this->token_blank, $this->char) === false) ? '' : $this->copy_skip($this->token_blank)), |
| 742 |
'', |
| 743 |
'' |
| 744 |
); |
| 745 |
} while ($this->char !== '>' && $this->char !== '/'); |
| 746 |
|
| 747 |
$this->link_nodes($node, true); |
| 748 |
|
| 749 |
// Space after last attribute before closing the tag |
| 750 |
if (!$trim && $space[0] !== '') { |
| 751 |
// phpcs:ignore Generic.Files.LineLength |
| 752 |
Debug::log_once('Source document contains superfluous whitespace before the closing braket (<e attribute="value" >). Enable trimming or remove spaces before closing brackets for best performance.'); |
| 753 |
$node->_[HtmlNode::HDOM_INFO_ENDSPACE] = $space[0]; |
| 754 |
} |
| 755 |
|
| 756 |
$rest = ($this->char === '>') ? '' : $this->copy_until_char('>'); |
| 757 |
$rest = ($trim) ? trim($rest) : $rest; // <html / > |
| 758 |
|
| 759 |
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 760 |
|
| 761 |
if (trim($rest) === '/') { // Void element |
| 762 |
if ($rest !== '') { |
| 763 |
if (isset($node->_[HtmlNode::HDOM_INFO_ENDSPACE])) { |
| 764 |
$node->_[HtmlNode::HDOM_INFO_ENDSPACE] .= $rest; |
| 765 |
} else { |
| 766 |
$node->_[HtmlNode::HDOM_INFO_ENDSPACE] = $rest; |
| 767 |
} |
| 768 |
} |
| 769 |
$node->_[HtmlNode::HDOM_INFO_END] = 0; |
| 770 |
} elseif (!isset($this->self_closing_tags[strtolower($node->tag)])) { |
| 771 |
$innertext = $this->copy_until_char('<'); |
| 772 |
if ($innertext !== '') { |
| 773 |
$node->_[HtmlNode::HDOM_INFO_INNER] = $innertext; |
| 774 |
} |
| 775 |
$this->parent = $node; |
| 776 |
} |
| 777 |
|
| 778 |
if ($node->tag === 'br') { |
| 779 |
$node->_[HtmlNode::HDOM_INFO_INNER] = $this->default_br_text; |
| 780 |
} elseif ($node->tag === 'script') { |
| 781 |
$data = ''; |
| 782 |
|
| 783 |
// There is a rare chance of empty script: "<script></script>" |
| 784 |
// In which case the current char is the start of the end tag |
| 785 |
// But the script could also just contain tags: "<script><div></script>" |
| 786 |
while (true) { |
| 787 |
// Copy until first char of end tag |
| 788 |
$data .= $this->copy_until_char('<'); |
| 789 |
|
| 790 |
// Look ahead in the document, maybe we are at the end |
| 791 |
if (($this->pos + 9) > $this->size) { // End of document |
| 792 |
Debug::log('Source document ended unexpectedly!'); |
| 793 |
break; |
| 794 |
} elseif (substr($this->doc, $this->pos, 8) === '</script') { // end |
| 795 |
$this->skip('>'); // don't include the end tag |
| 796 |
break; |
| 797 |
} |
| 798 |
|
| 799 |
// Note: A script tag may contain any other tag except </script> |
| 800 |
// which needs to be escaped as <\/script> |
| 801 |
|
| 802 |
$data .= $this->char; |
| 803 |
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 804 |
} |
| 805 |
|
| 806 |
$node = new HtmlNode($this); |
| 807 |
++$this->cursor; |
| 808 |
$node->_[HtmlNode::HDOM_INFO_TEXT] = $data; |
| 809 |
$this->link_nodes($node, false); |
| 810 |
} |
| 811 |
|
| 812 |
return true; |
| 813 |
} |
| 814 |
|
| 815 |
protected function parse_attr($node, $name, &$space, $trim) |
| 816 |
{ |
| 817 |
$is_duplicate = isset($node->attr[$name]); |
| 818 |
|
| 819 |
if (!$is_duplicate) // Copy whitespace between "=" and value |
| 820 |
$space[2] = (strpos($this->token_blank, $this->char) === false) ? '' : $this->copy_skip($this->token_blank); |
| 821 |
|
| 822 |
switch ($this->char) { |
| 823 |
case '"': |
| 824 |
$quote_type = HtmlNode::HDOM_QUOTE_DOUBLE; |
| 825 |
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 826 |
$value = $this->copy_until_char('"'); |
| 827 |
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 828 |
break; |
| 829 |
case '\'': |
| 830 |
// phpcs:ignore Generic.Files.LineLength |
| 831 |
Debug::log_once('Source document contains attribute values with single quotes (<e attribute=\'value\'>). Use double quotes for best performance.'); |
| 832 |
$quote_type = HtmlNode::HDOM_QUOTE_SINGLE; |
| 833 |
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 834 |
$value = $this->copy_until_char('\''); |
| 835 |
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 836 |
break; |
| 837 |
default: |
| 838 |
// phpcs:ignore Generic.Files.LineLength |
| 839 |
Debug::log_once('Source document contains attribute values without quotes (<e attribute=value>). Use double quotes for best performance'); |
| 840 |
$quote_type = HtmlNode::HDOM_QUOTE_NO; |
| 841 |
$value = $this->copy_until($this->token_attr); |
| 842 |
} |
| 843 |
|
| 844 |
$value = $this->restore_noise($value); |
| 845 |
|
| 846 |
if ($trim) { |
| 847 |
// Attribute values must not contain control characters other than space |
| 848 |
// https://www.w3.org/TR/html/dom.html#text-content |
| 849 |
// https://www.w3.org/TR/html/syntax.html#attribute-values |
| 850 |
// https://www.w3.org/TR/xml/#AVNormalize |
| 851 |
$value = preg_replace("/[\r\n\t\s]+/u", ' ', $value); |
| 852 |
$value = trim($value); |
| 853 |
} |
| 854 |
|
| 855 |
if (!$is_duplicate) { |
| 856 |
if ($quote_type !== HtmlNode::HDOM_QUOTE_DOUBLE) { |
| 857 |
$node->_[HtmlNode::HDOM_INFO_QUOTE][$name] = $quote_type; |
| 858 |
} |
| 859 |
$node->attr[$name] = $value; |
| 860 |
} |
| 861 |
} |
| 862 |
|
| 863 |
protected function link_nodes(&$node, $is_child) |
| 864 |
{ |
| 865 |
$node->parent = $this->parent; |
| 866 |
$this->parent->nodes[] = $node; |
| 867 |
if ($is_child) { |
| 868 |
$this->parent->children[] = $node; |
| 869 |
} |
| 870 |
} |
| 871 |
|
| 872 |
protected function as_text_node($tag) |
| 873 |
{ |
| 874 |
$node = new HtmlNode($this); |
| 875 |
++$this->cursor; |
| 876 |
$node->_[HtmlNode::HDOM_INFO_TEXT] = '</' . $tag . '>'; |
| 877 |
$this->link_nodes($node, false); |
| 878 |
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 879 |
return true; |
| 880 |
} |
| 881 |
|
| 882 |
protected function skip($chars) |
| 883 |
{ |
| 884 |
$this->pos += strspn($this->doc, $chars, $this->pos); |
| 885 |
$this->char = ($this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 886 |
} |
| 887 |
|
| 888 |
protected function copy_skip($chars) |
| 889 |
{ |
| 890 |
$pos = $this->pos; |
| 891 |
$len = strspn($this->doc, $chars, $pos); |
| 892 |
if ($len === 0) { |
| 893 |
return ''; |
| 894 |
} |
| 895 |
$this->pos += $len; |
| 896 |
$this->char = ($this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 897 |
return substr($this->doc, $pos, $len); |
| 898 |
} |
| 899 |
|
| 900 |
protected function copy_until($chars) |
| 901 |
{ |
| 902 |
$pos = $this->pos; |
| 903 |
$len = strcspn($this->doc, $chars, $pos); |
| 904 |
$this->pos += $len; |
| 905 |
$this->char = ($this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 906 |
return substr($this->doc, $pos, $len); |
| 907 |
} |
| 908 |
|
| 909 |
protected function copy_until_char($char) |
| 910 |
{ |
| 911 |
if ($this->char === null) { |
| 912 |
return ''; |
| 913 |
} |
| 914 |
|
| 915 |
if (($pos = strpos($this->doc, $char, $this->pos)) === false) { |
| 916 |
$ret = substr($this->doc, $this->pos, $this->size - $this->pos); |
| 917 |
$this->char = null; |
| 918 |
$this->pos = $this->size; |
| 919 |
return $ret; |
| 920 |
} |
| 921 |
|
| 922 |
if ($pos === $this->pos) { |
| 923 |
return ''; |
| 924 |
} |
| 925 |
|
| 926 |
$pos_old = $this->pos; |
| 927 |
$this->char = $this->doc[$pos]; |
| 928 |
$this->pos = $pos; |
| 929 |
return substr($this->doc, $pos_old, $pos - $pos_old); |
| 930 |
} |
| 931 |
|
| 932 |
protected function remove_noise($pattern, $remove_tag = false) |
| 933 |
{ |
| 934 |
$count = preg_match_all( |
| 935 |
$pattern, |
| 936 |
$this->doc, |
| 937 |
$matches, |
| 938 |
PREG_SET_ORDER | PREG_OFFSET_CAPTURE |
| 939 |
); |
| 940 |
|
| 941 |
for ($i = $count - 1; $i > -1; --$i) { |
| 942 |
$key = '___noise___' . sprintf('% 5d', count($this->noise) + 1000); |
| 943 |
|
| 944 |
$idx = ($remove_tag) ? 0 : 1; // 0 = entire match, 1 = submatch |
| 945 |
$this->noise[$key] = $matches[$i][$idx][0]; |
| 946 |
$this->doc = substr_replace($this->doc, $key, $matches[$i][$idx][1], strlen($matches[$i][$idx][0])); |
| 947 |
} |
| 948 |
|
| 949 |
// reset the length of content |
| 950 |
$this->size = strlen($this->doc); |
| 951 |
|
| 952 |
if ($this->size > 0) { |
| 953 |
$this->char = $this->doc[0]; |
| 954 |
} |
| 955 |
} |
| 956 |
|
| 957 |
function restore_noise($text) |
| 958 |
{ |
| 959 |
if (empty($this->noise)) return $text; // nothing to restore |
| 960 |
$pos = 0; |
| 961 |
while (($pos = strpos($text, '___noise___', $pos)) !== false) { |
| 962 |
// Sometimes there is a broken piece of markup, and we don't GET the |
| 963 |
// pos+11 etc... token which indicates a problem outside of us... |
| 964 |
|
| 965 |
// todo: "___noise___1000" (or any number with four or more digits) |
| 966 |
// in the DOM causes an infinite loop which could be utilized by |
| 967 |
// malicious software |
| 968 |
if (strlen($text) > $pos + 15) { |
| 969 |
$key = '___noise___' |
| 970 |
. $text[$pos + 11] |
| 971 |
. $text[$pos + 12] |
| 972 |
. $text[$pos + 13] |
| 973 |
. $text[$pos + 14] |
| 974 |
. $text[$pos + 15]; |
| 975 |
|
| 976 |
if (isset($this->noise[$key])) { |
| 977 |
$text = substr($text, 0, $pos) |
| 978 |
. $this->noise[$key] |
| 979 |
. substr($text, $pos + 16); |
| 980 |
|
| 981 |
unset($this->noise[$key]); |
| 982 |
} else { |
| 983 |
Debug::log_once('Noise restoration failed. DOM has been corrupted!'); |
| 984 |
// do this to prevent an infinite loop. |
| 985 |
// FIXME: THis causes an infinite loop because the keyword ___NOISE___ is included in the key! |
| 986 |
$text = substr($text, 0, $pos) |
| 987 |
. 'UNDEFINED NOISE FOR KEY: ' |
| 988 |
. $key |
| 989 |
. substr($text, $pos + 16); |
| 990 |
} |
| 991 |
} else { |
| 992 |
// There is no valid key being given back to us... We must get |
| 993 |
// rid of the ___noise___ or we will have a problem. |
| 994 |
Debug::log_once('Noise restoration failed. The provided key is incomplete: ' . $text); |
| 995 |
$text = substr($text, 0, $pos) |
| 996 |
. 'NO NUMERIC NOISE KEY' |
| 997 |
. substr($text, $pos + 11); |
| 998 |
} |
| 999 |
} |
| 1000 |
return $text; |
| 1001 |
} |
| 1002 |
|
| 1003 |
function search_noise($text) |
| 1004 |
{ |
| 1005 |
foreach ($this->noise as $noiseElement) { |
| 1006 |
if (strpos($noiseElement, $text) !== false) { |
| 1007 |
return $noiseElement; |
| 1008 |
} |
| 1009 |
} |
| 1010 |
} |
| 1011 |
|
| 1012 |
function __toString() |
| 1013 |
{ |
| 1014 |
return $this->root->innertext(); |
| 1015 |
} |
| 1016 |
|
| 1017 |
function __get($name) |
| 1018 |
{ |
| 1019 |
switch ($name) { |
| 1020 |
case 'outertext': |
| 1021 |
return $this->root->innertext(); |
| 1022 |
case 'innertext': |
| 1023 |
return $this->root->innertext(); |
| 1024 |
case 'plaintext': |
| 1025 |
return $this->root->text(); |
| 1026 |
case 'charset': |
| 1027 |
return $this->_charset; |
| 1028 |
case 'target_charset': |
| 1029 |
return $this->_target_charset; |
| 1030 |
} |
| 1031 |
} |
| 1032 |
|
| 1033 |
function childNodes($idx = -1) |
| 1034 |
{ |
| 1035 |
return $this->root->childNodes($idx); |
| 1036 |
} |
| 1037 |
|
| 1038 |
function firstChild() |
| 1039 |
{ |
| 1040 |
return $this->root->firstChild(); |
| 1041 |
} |
| 1042 |
|
| 1043 |
function lastChild() |
| 1044 |
{ |
| 1045 |
return $this->root->lastChild(); |
| 1046 |
} |
| 1047 |
|
| 1048 |
function createElement($name, $value = null) |
| 1049 |
{ |
| 1050 |
$node = new HtmlNode(null); |
| 1051 |
$node->nodetype = HtmlNode::HDOM_TYPE_ELEMENT; |
| 1052 |
$node->_[HtmlNode::HDOM_INFO_BEGIN] = 1; |
| 1053 |
$node->_[HtmlNode::HDOM_INFO_END] = 1; |
| 1054 |
|
| 1055 |
if ($value !== null) { |
| 1056 |
$node->_[HtmlNode::HDOM_INFO_INNER] = $value; |
| 1057 |
} |
| 1058 |
|
| 1059 |
$node->tag = $name; |
| 1060 |
|
| 1061 |
return $node; |
| 1062 |
} |
| 1063 |
|
| 1064 |
function createTextNode($value) |
| 1065 |
{ |
| 1066 |
$node = new HtmlNode($this); |
| 1067 |
$node->nodetype = HtmlNode::HDOM_TYPE_TEXT; |
| 1068 |
|
| 1069 |
if ($value !== null) { |
| 1070 |
$node->_[HtmlNode::HDOM_INFO_TEXT] = $value; |
| 1071 |
} |
| 1072 |
|
| 1073 |
return $node; |
| 1074 |
} |
| 1075 |
|
| 1076 |
function getElementById($id) |
| 1077 |
{ |
| 1078 |
return $this->find("#$id", 0); |
| 1079 |
} |
| 1080 |
|
| 1081 |
function getElementsById($id, $idx = null) |
| 1082 |
{ |
| 1083 |
return $this->find("#$id", $idx); |
| 1084 |
} |
| 1085 |
|
| 1086 |
function getElementByTagName($name) |
| 1087 |
{ |
| 1088 |
return $this->find($name, 0); |
| 1089 |
} |
| 1090 |
|
| 1091 |
function getElementsByTagName($name, $idx = null) |
| 1092 |
{ |
| 1093 |
return $this->find($name, $idx); |
| 1094 |
} |
| 1095 |
|
| 1096 |
function loadFile($file) |
| 1097 |
{ |
| 1098 |
$args = func_get_args(); |
| 1099 |
|
| 1100 |
if (($doc = call_user_func_array('file_get_contents', $args)) !== false) { |
| 1101 |
$this->load($doc, true); |
| 1102 |
} else { |
| 1103 |
return false; |
| 1104 |
} |
| 1105 |
} |
| 1106 |
} |
| 1107 |
|
