| 1 |
<?php namespace simplehtmldom; |
| 2 |
|
| 3 |
/** |
| 4 |
* Website: http://sourceforge.net/projects/simplehtmldom/ |
| 5 |
* Acknowledge: Jose Solorzano (https://sourceforge.net/projects/php-html/) |
| 6 |
* |
| 7 |
* Licensed under The MIT License |
| 8 |
* See the LICENSE file in the project root for more information. |
| 9 |
* |
| 10 |
* Authors: |
| 11 |
* S.C. Chen |
| 12 |
* John Schlick |
| 13 |
* Rus Carroll |
| 14 |
* logmanoriginal |
| 15 |
* |
| 16 |
* Contributors: |
| 17 |
* Yousuke Kumakura |
| 18 |
* Vadim Voituk |
| 19 |
* Antcs |
| 20 |
* |
| 21 |
* Version $Rev$ |
| 22 |
*/ |
| 23 |
|
| 24 |
include_once 'constants.php'; |
| 25 |
include_once 'Debug.php'; |
| 26 |
|
| 27 |
class HtmlNode |
| 28 |
{ |
| 29 |
const HDOM_TYPE_ELEMENT = 1; |
| 30 |
const HDOM_TYPE_COMMENT = 2; |
| 31 |
const HDOM_TYPE_TEXT = 3; |
| 32 |
const HDOM_TYPE_ROOT = 5; |
| 33 |
const HDOM_TYPE_UNKNOWN = 6; |
| 34 |
const HDOM_TYPE_CDATA = 7; |
| 35 |
|
| 36 |
const HDOM_QUOTE_DOUBLE = 0; |
| 37 |
const HDOM_QUOTE_SINGLE = 1; |
| 38 |
const HDOM_QUOTE_NO = 3; |
| 39 |
|
| 40 |
const HDOM_INFO_BEGIN = 0; |
| 41 |
const HDOM_INFO_END = 1; |
| 42 |
const HDOM_INFO_QUOTE = 2; |
| 43 |
const HDOM_INFO_SPACE = 3; |
| 44 |
const HDOM_INFO_TEXT = 4; |
| 45 |
const HDOM_INFO_INNER = 5; |
| 46 |
const HDOM_INFO_OUTER = 6; |
| 47 |
const HDOM_INFO_ENDSPACE = 7; |
| 48 |
|
| 49 |
public $nodetype = self::HDOM_TYPE_TEXT; |
| 50 |
public $tag = 'text'; |
| 51 |
public $attr = array(); |
| 52 |
public $children = array(); |
| 53 |
public $nodes = array(); |
| 54 |
public $parent = null; |
| 55 |
public $_ = array(); |
| 56 |
private $dom = null; |
| 57 |
|
| 58 |
function __call($func, $args) |
| 59 |
{ |
| 60 |
// Allow users to call methods with lower_case syntax |
| 61 |
switch($func) |
| 62 |
{ |
| 63 |
case 'children': |
| 64 |
$actual_function = 'childNodes'; break; |
| 65 |
case 'first_child': |
| 66 |
$actual_function = 'firstChild'; break; |
| 67 |
case 'has_child': |
| 68 |
$actual_function = 'hasChildNodes'; break; |
| 69 |
case 'last_child': |
| 70 |
$actual_function = 'lastChild'; break; |
| 71 |
case 'next_sibling': |
| 72 |
$actual_function = 'nextSibling'; break; |
| 73 |
case 'prev_sibling': |
| 74 |
$actual_function = 'previousSibling'; break; |
| 75 |
default: |
| 76 |
trigger_error( |
| 77 |
'Call to undefined method ' . __CLASS__ . '::' . $func . '()', |
| 78 |
E_USER_ERROR |
| 79 |
); |
| 80 |
} |
| 81 |
|
| 82 |
// phpcs:ignore Generic.Files.LineLength |
| 83 |
Debug::log(__CLASS__ . '->' . $func . '() has been deprecated and will be removed in the next major version of simplehtmldom. Use ' . __CLASS__ . '->' . $actual_function . '() instead.'); |
| 84 |
|
| 85 |
return call_user_func_array(array($this, $actual_function), $args); |
| 86 |
} |
| 87 |
|
| 88 |
function __construct($dom) |
| 89 |
{ |
| 90 |
if ($dom === null) return $this; |
| 91 |
|
| 92 |
$this->dom = $dom; |
| 93 |
$dom->nodes[] = $this; |
| 94 |
} |
| 95 |
|
| 96 |
function __debugInfo() |
| 97 |
{ |
| 98 |
// Translate node type to human-readable form |
| 99 |
switch($this->nodetype) |
| 100 |
{ |
| 101 |
case self::HDOM_TYPE_ELEMENT: |
| 102 |
$nodetype = "HDOM_TYPE_ELEMENT ($this->nodetype)"; |
| 103 |
break; |
| 104 |
case self::HDOM_TYPE_COMMENT: |
| 105 |
$nodetype = "HDOM_TYPE_COMMENT ($this->nodetype)"; |
| 106 |
break; |
| 107 |
case self::HDOM_TYPE_TEXT: |
| 108 |
$nodetype = "HDOM_TYPE_TEXT ($this->nodetype)"; |
| 109 |
break; |
| 110 |
case self::HDOM_TYPE_ROOT: |
| 111 |
$nodetype = "HDOM_TYPE_ROOT ($this->nodetype)"; |
| 112 |
break; |
| 113 |
case self::HDOM_TYPE_CDATA: |
| 114 |
$nodetype = "HDOM_TYPE_CDATA ($this->nodetype)"; |
| 115 |
break; |
| 116 |
case self::HDOM_TYPE_UNKNOWN: |
| 117 |
default: |
| 118 |
$nodetype = "HDOM_TYPE_UNKNOWN ($this->nodetype)"; |
| 119 |
} |
| 120 |
|
| 121 |
return array( |
| 122 |
'nodetype' => $nodetype, |
| 123 |
'tag' => $this->tag, |
| 124 |
'attributes' => empty($this->attr) ? 'none' : $this->attr, |
| 125 |
'nodes' => empty($this->nodes) ? 'none' : $this->nodes |
| 126 |
); |
| 127 |
} |
| 128 |
|
| 129 |
function __toString() |
| 130 |
{ |
| 131 |
return $this->outertext(); |
| 132 |
} |
| 133 |
|
| 134 |
function clear() |
| 135 |
{ |
| 136 |
unset($this->dom); // Break link to origin |
| 137 |
unset($this->parent); // Break link to branch |
| 138 |
} |
| 139 |
|
| 140 |
/** @codeCoverageIgnore */ |
| 141 |
function dump($show_attr = true, $depth = 0) |
| 142 |
{ |
| 143 |
echo str_repeat("\t", $depth) . $this->tag; |
| 144 |
|
| 145 |
if ($show_attr && count($this->attr) > 0) { |
| 146 |
echo '('; |
| 147 |
foreach ($this->attr as $k => $v) { |
| 148 |
echo "[$k]=>\"$v\", "; |
| 149 |
} |
| 150 |
echo ')'; |
| 151 |
} |
| 152 |
|
| 153 |
echo "\n"; |
| 154 |
|
| 155 |
if ($this->nodes) { |
| 156 |
foreach ($this->nodes as $node) { |
| 157 |
$node->dump($show_attr, $depth + 1); |
| 158 |
} |
| 159 |
} |
| 160 |
} |
| 161 |
|
| 162 |
/** @codeCoverageIgnore */ |
| 163 |
function dump_node($echo = true) |
| 164 |
{ |
| 165 |
$string = $this->tag; |
| 166 |
|
| 167 |
if (count($this->attr) > 0) { |
| 168 |
$string .= '('; |
| 169 |
foreach ($this->attr as $k => $v) { |
| 170 |
$string .= "[$k]=>\"$v\", "; |
| 171 |
} |
| 172 |
$string .= ')'; |
| 173 |
} |
| 174 |
|
| 175 |
if (count($this->_) > 0) { |
| 176 |
$string .= ' $_ ('; |
| 177 |
foreach ($this->_ as $k => $v) { |
| 178 |
if (is_array($v)) { |
| 179 |
$string .= "[$k]=>("; |
| 180 |
foreach ($v as $k2 => $v2) { |
| 181 |
$string .= "[$k2]=>\"$v2\", "; |
| 182 |
} |
| 183 |
$string .= ')'; |
| 184 |
} else { |
| 185 |
$string .= "[$k]=>\"$v\", "; |
| 186 |
} |
| 187 |
} |
| 188 |
$string .= ')'; |
| 189 |
} |
| 190 |
|
| 191 |
if (isset($this->text)) { |
| 192 |
$string .= " text: ({$this->text})"; |
| 193 |
} |
| 194 |
|
| 195 |
$string .= ' HDOM_INNER_INFO: '; |
| 196 |
|
| 197 |
if (isset($node->_[self::HDOM_INFO_INNER])) { |
| 198 |
$string .= "'" . $node->_[self::HDOM_INFO_INNER] . "'"; |
| 199 |
} else { |
| 200 |
$string .= ' NULL '; |
| 201 |
} |
| 202 |
|
| 203 |
$string .= ' children: ' . count($this->children); |
| 204 |
$string .= ' nodes: ' . count($this->nodes); |
| 205 |
$string .= "\n"; |
| 206 |
|
| 207 |
if ($echo) { |
| 208 |
echo $string; |
| 209 |
return; |
| 210 |
} else { |
| 211 |
return $string; |
| 212 |
} |
| 213 |
} |
| 214 |
|
| 215 |
function parent($parent = null) |
| 216 |
{ |
| 217 |
// I am SURE that this doesn't work properly. |
| 218 |
// It fails to unset the current node from it's current parents nodes or |
| 219 |
// children list first. |
| 220 |
if ($parent !== null) { |
| 221 |
$this->parent = $parent; |
| 222 |
$this->parent->nodes[] = $this; |
| 223 |
$this->parent->children[] = $this; |
| 224 |
} |
| 225 |
|
| 226 |
return $this->parent; |
| 227 |
} |
| 228 |
|
| 229 |
function find_ancestor_tag($tag) |
| 230 |
{ |
| 231 |
if ($this->parent === null) return null; |
| 232 |
|
| 233 |
$ancestor = $this->parent; |
| 234 |
|
| 235 |
while (!is_null($ancestor)) { |
| 236 |
if ($ancestor->tag === $tag) { |
| 237 |
break; |
| 238 |
} |
| 239 |
|
| 240 |
$ancestor = $ancestor->parent; |
| 241 |
} |
| 242 |
|
| 243 |
return $ancestor; |
| 244 |
} |
| 245 |
|
| 246 |
function innertext() |
| 247 |
{ |
| 248 |
if (isset($this->_[self::HDOM_INFO_INNER])) { |
| 249 |
$ret = $this->_[self::HDOM_INFO_INNER]; |
| 250 |
} elseif (isset($this->_[self::HDOM_INFO_TEXT])) { |
| 251 |
$ret = $this->_[self::HDOM_INFO_TEXT]; |
| 252 |
} else { |
| 253 |
$ret = ''; |
| 254 |
} |
| 255 |
|
| 256 |
foreach ($this->nodes as $n) { |
| 257 |
$ret .= $n->outertext(); |
| 258 |
} |
| 259 |
|
| 260 |
return $this->convert_text($ret); |
| 261 |
} |
| 262 |
|
| 263 |
function outertext() |
| 264 |
{ |
| 265 |
if ($this->tag === 'root') { |
| 266 |
return $this->innertext(); |
| 267 |
} |
| 268 |
|
| 269 |
// todo: What is the use of this callback? Remove? |
| 270 |
if ($this->dom && $this->dom->callback !== null) { |
| 271 |
call_user_func_array($this->dom->callback, array($this)); |
| 272 |
} |
| 273 |
|
| 274 |
if (isset($this->_[self::HDOM_INFO_OUTER])) { |
| 275 |
return $this->convert_text($this->_[self::HDOM_INFO_OUTER]); |
| 276 |
} |
| 277 |
|
| 278 |
if (isset($this->_[self::HDOM_INFO_TEXT])) { |
| 279 |
return $this->convert_text($this->_[self::HDOM_INFO_TEXT]); |
| 280 |
} |
| 281 |
|
| 282 |
$ret = ''; |
| 283 |
|
| 284 |
if (isset($this->_[self::HDOM_INFO_BEGIN])) { |
| 285 |
$ret = $this->makeup(); |
| 286 |
} |
| 287 |
|
| 288 |
if (isset($this->_[self::HDOM_INFO_INNER])) { |
| 289 |
// todo: <br> should either never have self::HDOM_INFO_INNER or always |
| 290 |
if ($this->tag !== 'br') { |
| 291 |
$ret .= $this->_[self::HDOM_INFO_INNER]; |
| 292 |
} |
| 293 |
} |
| 294 |
|
| 295 |
if ($this->nodes) { |
| 296 |
foreach ($this->nodes as $n) { |
| 297 |
$ret .= $n->outertext(); |
| 298 |
} |
| 299 |
} |
| 300 |
|
| 301 |
if (isset($this->_[self::HDOM_INFO_END]) && $this->_[self::HDOM_INFO_END] != 0) { |
| 302 |
$ret .= '</' . $this->tag . '>'; |
| 303 |
} |
| 304 |
|
| 305 |
return $this->convert_text($ret); |
| 306 |
} |
| 307 |
|
| 308 |
/** |
| 309 |
* Returns true if the provided element is a block level element |
| 310 |
* @link https://www.w3resource.com/html/HTML-block-level-and-inline-elements.php |
| 311 |
*/ |
| 312 |
protected function is_block_element($node) |
| 313 |
{ |
| 314 |
// todo: When we have the utility class this should be moved there |
| 315 |
return in_array(strtolower($node->tag), array( |
| 316 |
'p', |
| 317 |
'h1', 'h2', 'h3', 'h4', 'h5', 'h6', |
| 318 |
'ol', 'ul', |
| 319 |
'pre', |
| 320 |
'address', |
| 321 |
'blockquote', |
| 322 |
'dl', |
| 323 |
'div', |
| 324 |
'fieldset', |
| 325 |
'form', |
| 326 |
'hr', |
| 327 |
'noscript', |
| 328 |
'table' |
| 329 |
)); |
| 330 |
} |
| 331 |
|
| 332 |
/** |
| 333 |
* Returns true if the provided element is an inline level element |
| 334 |
* @link https://www.w3resource.com/html/HTML-block-level-and-inline-elements.php |
| 335 |
*/ |
| 336 |
protected function is_inline_element($node) |
| 337 |
{ |
| 338 |
// todo: When we have the utility class this should be moved there |
| 339 |
return in_array(strtolower($node->tag), array( |
| 340 |
'b', 'big', 'i', 'small', 'tt', |
| 341 |
'abbr', 'acronym', 'cite', 'code', 'dfn', 'em', 'kbd', 'strong', 'samp', 'var', |
| 342 |
'a', 'bdo', 'br', 'img', 'map', 'object', 'q', 'script', 'span', 'sub', 'sup', |
| 343 |
'button', 'input', 'label', 'select', 'textarea' |
| 344 |
)); |
| 345 |
} |
| 346 |
|
| 347 |
function text($trim = true) |
| 348 |
{ |
| 349 |
$ret = ''; |
| 350 |
|
| 351 |
if (strtolower($this->tag) === 'script') { |
| 352 |
$ret = ''; |
| 353 |
} elseif (strtolower($this->tag) === 'style') { |
| 354 |
$ret = ''; |
| 355 |
} elseif ($this->nodetype === self::HDOM_TYPE_COMMENT) { |
| 356 |
$ret = ''; |
| 357 |
} elseif ($this->nodetype === self::HDOM_TYPE_CDATA) { |
| 358 |
$ret = $this->_[self::HDOM_INFO_INNER]; |
| 359 |
} elseif ($this->nodetype === self::HDOM_TYPE_UNKNOWN) { |
| 360 |
$ret = ''; |
| 361 |
} elseif (isset($this->_[self::HDOM_INFO_INNER])) { |
| 362 |
$ret = $this->_[self::HDOM_INFO_INNER]; |
| 363 |
} elseif ($this->nodetype === self::HDOM_TYPE_TEXT) { |
| 364 |
$ret = $this->_[self::HDOM_INFO_TEXT]; |
| 365 |
} |
| 366 |
|
| 367 |
if (is_null($this->nodes)) { |
| 368 |
return ''; |
| 369 |
} |
| 370 |
|
| 371 |
foreach ($this->nodes as $n) { |
| 372 |
if ($this->is_block_element($n)) { |
| 373 |
|
| 374 |
$block = ltrim($this->convert_text($n->text(false))); |
| 375 |
|
| 376 |
if (empty($block)) |
| 377 |
continue; |
| 378 |
|
| 379 |
$ret = rtrim($ret) . "\n\n" . $block; |
| 380 |
|
| 381 |
} elseif ($this->is_inline_element($n)) { |
| 382 |
// todo: <br> introduces code smell because no space but \n |
| 383 |
if (strtolower($n->tag) === 'br') { |
| 384 |
$ret .= $this->dom->default_br_text ?: DEFAULT_BR_TEXT; |
| 385 |
} else { |
| 386 |
$inline = ltrim($this->convert_text($n->text(false))); |
| 387 |
|
| 388 |
if (empty($inline)) |
| 389 |
continue; |
| 390 |
|
| 391 |
$ret = $ret . $this->convert_text($n->text(false)); |
| 392 |
} |
| 393 |
} else { |
| 394 |
$ret .= $this->convert_text($n->text(false)); |
| 395 |
} |
| 396 |
} |
| 397 |
|
| 398 |
// Reduce whitespace at start/end to a single (or none) space |
| 399 |
$ret = preg_replace('/[ \t\n\r\0\x0B\xC2\xA0]+$/u', $trim ? '' : ' ', $ret); |
| 400 |
$ret = preg_replace('/^[ \t\n\r\0\x0B\xC2\xA0]+/u', $trim ? '' : ' ', $ret); |
| 401 |
|
| 402 |
return $ret; |
| 403 |
} |
| 404 |
|
| 405 |
function xmltext() |
| 406 |
{ |
| 407 |
$ret = $this->innertext(); |
| 408 |
$ret = str_ireplace('<![CDATA[', '', $ret); |
| 409 |
$ret = str_replace(']]>', '', $ret); |
| 410 |
return $ret; |
| 411 |
} |
| 412 |
|
| 413 |
function makeup() |
| 414 |
{ |
| 415 |
// text, comment, unknown |
| 416 |
if (isset($this->_[self::HDOM_INFO_TEXT])) { |
| 417 |
return $this->_[self::HDOM_INFO_TEXT]; |
| 418 |
} |
| 419 |
|
| 420 |
$ret = '<' . $this->tag; |
| 421 |
|
| 422 |
foreach ($this->attr as $key => $val) { |
| 423 |
|
| 424 |
// skip removed attribute |
| 425 |
if ($val === null || $val === false) { continue; } |
| 426 |
|
| 427 |
if (isset($this->_[self::HDOM_INFO_SPACE][$key])) { |
| 428 |
$ret .= $this->_[self::HDOM_INFO_SPACE][$key][0]; |
| 429 |
} else { |
| 430 |
$ret .= ' '; |
| 431 |
} |
| 432 |
|
| 433 |
//no value attr: nowrap, checked selected... |
| 434 |
if ($val === true) { |
| 435 |
$ret .= $key; |
| 436 |
} else { |
| 437 |
if (isset($this->_[self::HDOM_INFO_QUOTE][$key])) { |
| 438 |
$quote_type = $this->_[self::HDOM_INFO_QUOTE][$key]; |
| 439 |
} else { |
| 440 |
$quote_type = self::HDOM_QUOTE_DOUBLE; |
| 441 |
} |
| 442 |
|
| 443 |
switch ($quote_type) |
| 444 |
{ |
| 445 |
case self::HDOM_QUOTE_SINGLE: |
| 446 |
$quote = '\''; |
| 447 |
$val = htmlentities($val, ENT_QUOTES, $this->dom->target_charset); |
| 448 |
break; |
| 449 |
case self::HDOM_QUOTE_NO: |
| 450 |
$quote = ''; |
| 451 |
break; |
| 452 |
case self::HDOM_QUOTE_DOUBLE: |
| 453 |
default: |
| 454 |
$quote = '"'; |
| 455 |
$val = htmlentities($val, ENT_COMPAT, $this->dom->target_charset); |
| 456 |
} |
| 457 |
|
| 458 |
$ret .= $key |
| 459 |
. (isset($this->_[self::HDOM_INFO_SPACE][$key]) ? $this->_[self::HDOM_INFO_SPACE][$key][1] : '') |
| 460 |
. '=' |
| 461 |
. (isset($this->_[self::HDOM_INFO_SPACE][$key]) ? $this->_[self::HDOM_INFO_SPACE][$key][2] : '') |
| 462 |
. $quote |
| 463 |
. $val |
| 464 |
. $quote; |
| 465 |
} |
| 466 |
} |
| 467 |
|
| 468 |
if(isset($this->_[self::HDOM_INFO_ENDSPACE])) { |
| 469 |
$ret .= $this->_[self::HDOM_INFO_ENDSPACE]; |
| 470 |
} |
| 471 |
|
| 472 |
return $ret . '>'; |
| 473 |
} |
| 474 |
|
| 475 |
function find($selector, $idx = null, $lowercase = false) |
| 476 |
{ |
| 477 |
$selectors = $this->parse_selector($selector); |
| 478 |
if (($count = count($selectors)) === 0) { return array(); } |
| 479 |
$found_keys = array(); |
| 480 |
|
| 481 |
// find each selector |
| 482 |
for ($c = 0; $c < $count; ++$c) { |
| 483 |
// The change on the below line was documented on the sourceforge |
| 484 |
// code tracker id 2788009 |
| 485 |
// used to be: if (($levle=count($selectors[0]))===0) return array(); |
| 486 |
if (($levle = count($selectors[$c])) === 0) { |
| 487 |
Debug::log_once('Empty selector (' . $selector . ') matches nothing.'); |
| 488 |
return array(); |
| 489 |
} |
| 490 |
|
| 491 |
if (!isset($this->_[self::HDOM_INFO_BEGIN])) { |
| 492 |
Debug::log_once('Invalid operation. The current node has no start tag.'); |
| 493 |
return array(); |
| 494 |
} |
| 495 |
|
| 496 |
$head = array($this->_[self::HDOM_INFO_BEGIN] => 1); |
| 497 |
$cmd = ' '; // Combinator |
| 498 |
|
| 499 |
// handle descendant selectors, no recursive! |
| 500 |
for ($l = 0; $l < $levle; ++$l) { |
| 501 |
$ret = array(); |
| 502 |
|
| 503 |
foreach ($head as $k => $v) { |
| 504 |
$n = ($k === -1) ? $this->dom->root : $this->dom->nodes[$k]; |
| 505 |
//PaperG - Pass this optional parameter on to the seek function. |
| 506 |
$n->seek($selectors[$c][$l], $ret, $cmd, $lowercase); |
| 507 |
} |
| 508 |
|
| 509 |
$head = $ret; |
| 510 |
$cmd = $selectors[$c][$l][6]; // Next Combinator |
| 511 |
} |
| 512 |
|
| 513 |
foreach ($head as $k => $v) { |
| 514 |
if (!isset($found_keys[$k])) { |
| 515 |
$found_keys[$k] = 1; |
| 516 |
} |
| 517 |
} |
| 518 |
} |
| 519 |
|
| 520 |
// sort keys |
| 521 |
ksort($found_keys); |
| 522 |
|
| 523 |
$found = array(); |
| 524 |
foreach ($found_keys as $k => $v) { |
| 525 |
$found[] = $this->dom->nodes[$k]; |
| 526 |
} |
| 527 |
|
| 528 |
// return nth-element or array |
| 529 |
if (is_null($idx)) { return $found; } |
| 530 |
elseif ($idx < 0) { $idx = count($found) + $idx; } |
| 531 |
return (isset($found[$idx])) ? $found[$idx] : null; |
| 532 |
} |
| 533 |
|
| 534 |
function expect($selector, $idx = null, $lowercase = false) |
| 535 |
{ |
| 536 |
return $this->find($selector, $idx, $lowercase) ?: null; |
| 537 |
} |
| 538 |
|
| 539 |
protected function seek($selector, &$ret, $parent_cmd, $lowercase = false) |
| 540 |
{ |
| 541 |
list($ps_selector, $tag, $ps_element, $id, $class, $attributes, $cmb) = $selector; |
| 542 |
$nodes = array(); |
| 543 |
|
| 544 |
if ($parent_cmd === ' ') { // Descendant Combinator |
| 545 |
// Find parent closing tag if the current element doesn't have a closing |
| 546 |
// tag (i.e. void element) |
| 547 |
$end = (!empty($this->_[self::HDOM_INFO_END])) ? $this->_[self::HDOM_INFO_END] : 0; |
| 548 |
if ($end == 0 && $this->parent) { |
| 549 |
$parent = $this->parent; |
| 550 |
while ($parent !== null && !isset($parent->_[self::HDOM_INFO_END])) { |
| 551 |
$end -= 1; |
| 552 |
$parent = $parent->parent; |
| 553 |
} |
| 554 |
$end += $parent->_[self::HDOM_INFO_END]; |
| 555 |
} |
| 556 |
|
| 557 |
if ($end === 0) { |
| 558 |
$end = count($this->dom->nodes); |
| 559 |
} |
| 560 |
|
| 561 |
// Get list of target nodes |
| 562 |
$nodes_start = $this->_[self::HDOM_INFO_BEGIN] + 1; |
| 563 |
|
| 564 |
// remove() makes $this->dom->nodes non-contiguous; use what is left. |
| 565 |
$nodes = array_intersect_key( |
| 566 |
$this->dom->nodes, |
| 567 |
array_flip(range($nodes_start, $end)) |
| 568 |
); |
| 569 |
} elseif ($parent_cmd === '>') { // Child Combinator |
| 570 |
$nodes = $this->children; |
| 571 |
} elseif ($parent_cmd === '+' |
| 572 |
&& $this->parent |
| 573 |
&& in_array($this, $this->parent->children)) { // Next-Sibling Combinator |
| 574 |
$index = array_search($this, $this->parent->children, true) + 1; |
| 575 |
if ($index < count($this->parent->children)) |
| 576 |
$nodes[] = $this->parent->children[$index]; |
| 577 |
} elseif ($parent_cmd === '~' |
| 578 |
&& $this->parent |
| 579 |
&& in_array($this, $this->parent->children)) { // Subsequent Sibling Combinator |
| 580 |
$index = array_search($this, $this->parent->children, true); |
| 581 |
$nodes = array_slice($this->parent->children, $index); |
| 582 |
} |
| 583 |
|
| 584 |
// Go throgh each element starting at this element until the end tag |
| 585 |
// Note: If this element is a void tag, any previous void element is |
| 586 |
// skipped. |
| 587 |
foreach($nodes as $node) { |
| 588 |
$pass = true; |
| 589 |
|
| 590 |
// Skip root nodes |
| 591 |
if(!$node->parent) { |
| 592 |
unset($node); |
| 593 |
continue; |
| 594 |
} |
| 595 |
|
| 596 |
// Handle 'text' selector |
| 597 |
if($pass && $tag === 'text') { |
| 598 |
|
| 599 |
if($node->tag === 'text') { |
| 600 |
$ret[array_search($node, $this->dom->nodes, true)] = 1; |
| 601 |
} |
| 602 |
|
| 603 |
if(isset($node->_[self::HDOM_INFO_INNER])) { |
| 604 |
$ret[$node->_[self::HDOM_INFO_BEGIN]] = 1; |
| 605 |
} |
| 606 |
|
| 607 |
unset($node); |
| 608 |
continue; |
| 609 |
|
| 610 |
} |
| 611 |
|
| 612 |
// Handle 'cdata' selector |
| 613 |
if($pass && $tag === 'cdata') { |
| 614 |
|
| 615 |
if($node->tag === 'cdata') { |
| 616 |
$ret[$node->_[self::HDOM_INFO_BEGIN]] = 1; |
| 617 |
} |
| 618 |
|
| 619 |
unset($node); |
| 620 |
continue; |
| 621 |
|
| 622 |
} |
| 623 |
|
| 624 |
// Handle 'comment' |
| 625 |
if($pass && $tag === 'comment' && $node->tag === 'comment') { |
| 626 |
$ret[$node->_[self::HDOM_INFO_BEGIN]] = 1; |
| 627 |
unset($node); |
| 628 |
continue; |
| 629 |
} |
| 630 |
|
| 631 |
// Skip if node isn't a child node (i.e. text nodes) |
| 632 |
if($pass && !in_array($node, $node->parent->children, true)) { |
| 633 |
unset($node); |
| 634 |
continue; |
| 635 |
} |
| 636 |
|
| 637 |
// Skip if tag doesn't match |
| 638 |
if ($pass && $tag !== '' && $tag !== $node->tag && $tag !== '*') { |
| 639 |
$pass = false; |
| 640 |
} |
| 641 |
|
| 642 |
// Skip if ID doesn't exist |
| 643 |
if ($pass && $id !== '' && !isset($node->attr['id'])) { |
| 644 |
$pass = false; |
| 645 |
} |
| 646 |
|
| 647 |
// Check if ID matches |
| 648 |
if ($pass && $id !== '' && isset($node->attr['id'])) { |
| 649 |
// Note: Only consider the first ID (as browsers do) |
| 650 |
$node_id = explode(' ', trim($node->attr['id']))[0]; |
| 651 |
|
| 652 |
if($id !== $node_id) { $pass = false; } |
| 653 |
} |
| 654 |
|
| 655 |
// Check if all class(es) exist |
| 656 |
if ($pass && $class !== '' && is_array($class) && !empty($class)) { |
| 657 |
if (isset($node->attr['class'])) { |
| 658 |
// Apply the same rules for the pattern and attribute value |
| 659 |
// Attribute values must not contain control characters other than space |
| 660 |
// https://www.w3.org/TR/html/dom.html#text-content |
| 661 |
// https://www.w3.org/TR/html/syntax.html#attribute-values |
| 662 |
// https://www.w3.org/TR/xml/#AVNormalize |
| 663 |
$node_classes = preg_replace("/[\r\n\t\s]+/u", ' ', $node->attr['class']); |
| 664 |
$node_classes = trim($node_classes); |
| 665 |
$node_classes = explode(' ', $node_classes); |
| 666 |
|
| 667 |
if ($lowercase) { |
| 668 |
$node_classes = array_map('strtolower', $node_classes); |
| 669 |
} |
| 670 |
|
| 671 |
foreach($class as $c) { |
| 672 |
if(!in_array($c, $node_classes)) { |
| 673 |
$pass = false; |
| 674 |
break; |
| 675 |
} |
| 676 |
} |
| 677 |
} else { |
| 678 |
$pass = false; |
| 679 |
} |
| 680 |
} |
| 681 |
|
| 682 |
// Check attributes |
| 683 |
if ($pass |
| 684 |
&& $attributes !== '' |
| 685 |
&& is_array($attributes) |
| 686 |
&& !empty($attributes)) { |
| 687 |
foreach($attributes as $a) { |
| 688 |
list ( |
| 689 |
$att_name, |
| 690 |
$att_expr, |
| 691 |
$att_val, |
| 692 |
$att_inv, |
| 693 |
$att_case_sensitivity |
| 694 |
) = $a; |
| 695 |
|
| 696 |
// Handle indexing attributes (i.e. "[2]") |
| 697 |
/** |
| 698 |
* Note: This is not supported by the CSS Standard but adds |
| 699 |
* the ability to select items compatible to XPath (i.e. |
| 700 |
* the 3rd element within it's parent). |
| 701 |
* |
| 702 |
* Note: This doesn't conflict with the CSS Standard which |
| 703 |
* doesn't work on numeric attributes anyway. |
| 704 |
*/ |
| 705 |
if (is_numeric($att_name) |
| 706 |
&& $att_expr === '' |
| 707 |
&& $att_val === '') { |
| 708 |
$count = 0; |
| 709 |
|
| 710 |
// Find index of current element in parent |
| 711 |
foreach ($node->parent->children as $c) { |
| 712 |
if ($c->tag === $node->tag) ++$count; |
| 713 |
if ($c === $node) break; |
| 714 |
} |
| 715 |
|
| 716 |
// If this is the correct node, continue with next |
| 717 |
// attribute |
| 718 |
if ($count === (int)$att_name) continue; |
| 719 |
} |
| 720 |
|
| 721 |
// Check attribute availability |
| 722 |
if ($att_inv) { // Attribute should NOT be set |
| 723 |
if (isset($node->attr[$att_name])) { |
| 724 |
$pass = false; |
| 725 |
break; |
| 726 |
} |
| 727 |
} else { // Attribute should be set |
| 728 |
// todo: "plaintext" is not a valid CSS selector! |
| 729 |
if ($att_name !== 'plaintext' |
| 730 |
&& !isset($node->attr[$att_name])) { |
| 731 |
$pass = false; |
| 732 |
break; |
| 733 |
} |
| 734 |
} |
| 735 |
|
| 736 |
// Continue with next attribute if expression isn't defined |
| 737 |
if ($att_expr === '') continue; |
| 738 |
|
| 739 |
// If they have told us that this is a "plaintext" |
| 740 |
// search then we want the plaintext of the node - right? |
| 741 |
// todo "plaintext" is not a valid CSS selector! |
| 742 |
if ($att_name === 'plaintext') { |
| 743 |
$nodeKeyValue = $node->text(); |
| 744 |
} else { |
| 745 |
$nodeKeyValue = $node->attr[$att_name]; |
| 746 |
} |
| 747 |
|
| 748 |
// If lowercase is set, do a case insensitive test of |
| 749 |
// the value of the selector. |
| 750 |
if ($lowercase) { |
| 751 |
$check = $this->match( |
| 752 |
$att_expr, |
| 753 |
strtolower($att_val), |
| 754 |
strtolower($nodeKeyValue), |
| 755 |
$att_case_sensitivity |
| 756 |
); |
| 757 |
} else { |
| 758 |
$check = $this->match( |
| 759 |
$att_expr, |
| 760 |
$att_val, |
| 761 |
$nodeKeyValue, |
| 762 |
$att_case_sensitivity |
| 763 |
); |
| 764 |
} |
| 765 |
|
| 766 |
$check = $ps_element === 'not' ? !$check : $check; |
| 767 |
|
| 768 |
if (!$check) { |
| 769 |
$pass = false; |
| 770 |
break; |
| 771 |
} |
| 772 |
} |
| 773 |
} |
| 774 |
|
| 775 |
// Found a match. Add to list and clear node |
| 776 |
$pass = $ps_selector === 'not' ? !$pass : $pass; |
| 777 |
if ($pass) $ret[$node->_[self::HDOM_INFO_BEGIN]] = 1; |
| 778 |
unset($node); |
| 779 |
} |
| 780 |
} |
| 781 |
|
| 782 |
protected function match($exp, $pattern, $value, $case_sensitivity) |
| 783 |
{ |
| 784 |
if ($case_sensitivity === 'i') { |
| 785 |
$pattern = strtolower($pattern); |
| 786 |
$value = strtolower($value); |
| 787 |
} |
| 788 |
|
| 789 |
// Apply the same rules for the pattern and attribute value |
| 790 |
// Attribute values must not contain control characters other than space |
| 791 |
// https://www.w3.org/TR/html/dom.html#text-content |
| 792 |
// https://www.w3.org/TR/html/syntax.html#attribute-values |
| 793 |
// https://www.w3.org/TR/xml/#AVNormalize |
| 794 |
$pattern = preg_replace("/[\r\n\t\s]+/u", ' ', $pattern); |
| 795 |
$pattern = trim($pattern); |
| 796 |
|
| 797 |
$value = preg_replace("/[\r\n\t\s]+/u", ' ', $value); |
| 798 |
$value = trim($value); |
| 799 |
|
| 800 |
switch ($exp) { |
| 801 |
case '=': |
| 802 |
return ($value === $pattern); |
| 803 |
case '!=': |
| 804 |
return ($value !== $pattern); |
| 805 |
case '^=': |
| 806 |
return preg_match('/^' . preg_quote($pattern, '/') . '/', $value); |
| 807 |
case '$=': |
| 808 |
return preg_match('/' . preg_quote($pattern, '/') . '$/', $value); |
| 809 |
case '*=': |
| 810 |
return preg_match('/' . preg_quote($pattern, '/') . '/', $value); |
| 811 |
case '|=': |
| 812 |
/** |
| 813 |
* [att|=val] |
| 814 |
* |
| 815 |
* Represents an element with the att attribute, its value |
| 816 |
* either being exactly "val" or beginning with "val" |
| 817 |
* immediately followed by "-" (U+002D). |
| 818 |
*/ |
| 819 |
return strpos($value, $pattern) === 0; |
| 820 |
case '~=': |
| 821 |
/** |
| 822 |
* [att~=val] |
| 823 |
* |
| 824 |
* Represents an element with the att attribute whose value is a |
| 825 |
* whitespace-separated list of words, one of which is exactly |
| 826 |
* "val". If "val" contains whitespace, it will never represent |
| 827 |
* anything (since the words are separated by spaces). Also if |
| 828 |
* "val" is the empty string, it will never represent anything. |
| 829 |
*/ |
| 830 |
return in_array($pattern, explode(' ', trim($value)), true); |
| 831 |
} |
| 832 |
|
| 833 |
Debug::log('Unhandled attribute selector: ' . $exp . '!'); |
| 834 |
return false; |
| 835 |
} |
| 836 |
|
| 837 |
protected function parse_selector($selector_string) |
| 838 |
{ |
| 839 |
/** |
| 840 |
* Pattern of CSS selectors, modified from mootools (https://mootools.net/) |
| 841 |
* |
| 842 |
* Paperg: Add the colon to the attribute, so that it properly finds |
| 843 |
* <tag attr:ibute="something" > like google does. |
| 844 |
* |
| 845 |
* Note: if you try to look at this attribute, you MUST use getAttribute |
| 846 |
* since $dom->x:y will fail the php syntax check. |
| 847 |
* |
| 848 |
* Notice the \[ starting the attribute? and the @? following? This |
| 849 |
* implies that an attribute can begin with an @ sign that is not |
| 850 |
* captured. This implies that an html attribute specifier may start |
| 851 |
* with an @ sign that is NOT captured by the expression. Farther study |
| 852 |
* is required to determine of this should be documented or removed. |
| 853 |
* |
| 854 |
* Matches selectors in this order: |
| 855 |
* |
| 856 |
* [0] - full match |
| 857 |
* |
| 858 |
* [1] - pseudo selector |
| 859 |
* (?:\:(\w+)\()? |
| 860 |
* Matches the pseudo selector (optional) |
| 861 |
* |
| 862 |
* [2] - tag name |
| 863 |
* ([\w:\*-]*) |
| 864 |
* Matches the tag name consisting of zero or more words, colons, |
| 865 |
* asterisks and hyphens. |
| 866 |
* |
| 867 |
* [3] - pseudo selector |
| 868 |
* (?:\:(\w+)\()? |
| 869 |
* Matches the pseudo selector (optional) |
| 870 |
* |
| 871 |
* [4] - id name |
| 872 |
* (?:\#([\w-]+)) |
| 873 |
* Optionally matches a id name, consisting of an "#" followed by |
| 874 |
* the id name (one or more words and hyphens). |
| 875 |
* |
| 876 |
* [5] - class names (including dots) |
| 877 |
* (?:\.([\w\.-]+))? |
| 878 |
* Optionally matches a list of classs, consisting of an "." |
| 879 |
* followed by the class name (one or more words and hyphens) |
| 880 |
* where multiple classes can be chained (i.e. ".foo.bar.baz") |
| 881 |
* |
| 882 |
* [6] - attributes |
| 883 |
* ((?:\[@?(?:!?[\w:-]+)(?:(?:[!*^$|~]?=)[\"']?(?:.*?)[\"']?)?(?:\s*?(?:[iIsS])?)?\])+)? |
| 884 |
* Optionally matches the attributes list |
| 885 |
* |
| 886 |
* [7] - separator |
| 887 |
* ([\/, >+~]+) |
| 888 |
* Matches the selector list separator |
| 889 |
*/ |
| 890 |
// phpcs:ignore Generic.Files.LineLength |
| 891 |
$pattern = "/(?:\:(\w+)\()?([\w:\*-]*)(?:\:(\w+)\()?(?:\#([\w-]+))?(?:|\.([\w\.-]+))?((?:\[@?(?:!?[\w:-]+)(?:(?:[!*^$|~]?=)[\"']?(?:.*?)[\"']?)?(?:\s*?(?:[iIsS])?)?\])+)?(?:\))?(?:\))?([\/, >+~]+)/is"; |
| 892 |
|
| 893 |
preg_match_all( |
| 894 |
$pattern, |
| 895 |
trim($selector_string) . ' ', // Add final ' ' as pseudo separator |
| 896 |
$matches, |
| 897 |
PREG_SET_ORDER |
| 898 |
); |
| 899 |
|
| 900 |
$selectors = array(); |
| 901 |
$result = array(); |
| 902 |
|
| 903 |
foreach ($matches as $m) { |
| 904 |
$m[0] = trim($m[0]); |
| 905 |
|
| 906 |
// Skip NoOps |
| 907 |
if ($m[0] === '' || $m[0] === '/' || $m[0] === '//') { continue; } |
| 908 |
|
| 909 |
array_shift($m); |
| 910 |
|
| 911 |
// Convert to lowercase |
| 912 |
if ($this->dom->lowercase) { |
| 913 |
$m[1] = strtolower($m[1]); |
| 914 |
} |
| 915 |
|
| 916 |
// Extract classes |
| 917 |
if ($m[4] !== '') { $m[4] = explode('.', $m[4]); } |
| 918 |
|
| 919 |
/* Extract attributes (pattern based on the pattern above!) |
| 920 |
|
| 921 |
* [0] - full match |
| 922 |
* [1] - attribute name |
| 923 |
* [2] - attribute expression |
| 924 |
* [3] - attribute value |
| 925 |
* [4] - case sensitivity |
| 926 |
* |
| 927 |
* Note: Attributes can be negated with a "!" prefix to their name |
| 928 |
*/ |
| 929 |
if($m[5] !== '') { |
| 930 |
preg_match_all( |
| 931 |
"/\[@?(!?[\w:-]+)(?:([!*^$|~]?=)[\"']?(.*?)[\"']?)?(?:\s+?([iIsS])?)?\]/is", |
| 932 |
trim($m[5]), |
| 933 |
$attributes, |
| 934 |
PREG_SET_ORDER |
| 935 |
); |
| 936 |
|
| 937 |
// Replace element by array |
| 938 |
$m[5] = array(); |
| 939 |
|
| 940 |
foreach($attributes as $att) { |
| 941 |
// Skip empty matches |
| 942 |
if(trim($att[0]) === '') { continue; } |
| 943 |
|
| 944 |
$inverted = (isset($att[1][0]) && $att[1][0] === '!'); |
| 945 |
$m[5][] = array( |
| 946 |
$inverted ? substr($att[1], 1) : $att[1], // Name |
| 947 |
(isset($att[2])) ? $att[2] : '', // Expression |
| 948 |
(isset($att[3])) ? $att[3] : '', // Value |
| 949 |
$inverted, // Inverted Flag |
| 950 |
(isset($att[4])) ? strtolower($att[4]) : '', // Case-Sensitivity |
| 951 |
); |
| 952 |
} |
| 953 |
} |
| 954 |
|
| 955 |
// Sanitize Separator |
| 956 |
if ($m[6] !== '' && trim($m[6]) === '') { // Descendant Separator |
| 957 |
$m[6] = ' '; |
| 958 |
} else { // Other Separator |
| 959 |
$m[6] = trim($m[6]); |
| 960 |
} |
| 961 |
|
| 962 |
// Clear Separator if it's a Selector List |
| 963 |
if ($is_list = ($m[6] === ',')) { $m[6] = ''; } |
| 964 |
|
| 965 |
$result[] = $m; |
| 966 |
|
| 967 |
if ($is_list) { // Selector List |
| 968 |
$selectors[] = $result; |
| 969 |
$result = array(); |
| 970 |
} |
| 971 |
} |
| 972 |
|
| 973 |
if (count($result) > 0) { $selectors[] = $result; } |
| 974 |
return $selectors; |
| 975 |
} |
| 976 |
|
| 977 |
function __get($name) |
| 978 |
{ |
| 979 |
if (isset($this->attr[$name])) { |
| 980 |
return $this->convert_text($this->attr[$name]); |
| 981 |
} |
| 982 |
|
| 983 |
switch ($name) { |
| 984 |
case 'outertext': return $this->outertext(); |
| 985 |
case 'innertext': return $this->innertext(); |
| 986 |
case 'plaintext': return $this->text(); |
| 987 |
case 'xmltext': return $this->xmltext(); |
| 988 |
} |
| 989 |
|
| 990 |
return false; |
| 991 |
} |
| 992 |
|
| 993 |
function __set($name, $value) |
| 994 |
{ |
| 995 |
switch ($name) { |
| 996 |
case 'outertext': |
| 997 |
$this->_[self::HDOM_INFO_OUTER] = $value; |
| 998 |
break; |
| 999 |
case 'innertext': |
| 1000 |
if (isset($this->_[self::HDOM_INFO_TEXT])) { |
| 1001 |
$this->_[self::HDOM_INFO_TEXT] = ''; |
| 1002 |
} |
| 1003 |
$this->_[self::HDOM_INFO_INNER] = $value; |
| 1004 |
break; |
| 1005 |
default: $this->attr[$name] = $value; |
| 1006 |
} |
| 1007 |
} |
| 1008 |
|
| 1009 |
function __isset($name) |
| 1010 |
{ |
| 1011 |
switch ($name) { |
| 1012 |
case 'outertext': return true; |
| 1013 |
case 'innertext': return true; |
| 1014 |
case 'plaintext': return true; |
| 1015 |
} |
| 1016 |
|
| 1017 |
return isset($this->attr[$name]); |
| 1018 |
} |
| 1019 |
|
| 1020 |
function __unset($name) |
| 1021 |
{ |
| 1022 |
if (isset($this->attr[$name])) { unset($this->attr[$name]); } |
| 1023 |
} |
| 1024 |
|
| 1025 |
function convert_text($text) |
| 1026 |
{ |
| 1027 |
$converted_text = $text; |
| 1028 |
|
| 1029 |
$sourceCharset = ''; |
| 1030 |
$targetCharset = ''; |
| 1031 |
|
| 1032 |
if ($this->dom) { |
| 1033 |
$sourceCharset = strtoupper($this->dom->_charset); |
| 1034 |
$targetCharset = strtoupper($this->dom->_target_charset); |
| 1035 |
} |
| 1036 |
|
| 1037 |
if (!empty($sourceCharset) && !empty($targetCharset)) { |
| 1038 |
if (strtoupper($sourceCharset) === strtoupper($targetCharset)) { |
| 1039 |
$converted_text = $text; |
| 1040 |
} elseif ((strtoupper($targetCharset) === 'UTF-8') && (self::is_utf8($text))) { |
| 1041 |
Debug::log_once('The source charset was incorrectly detected as ' . $sourceCharset . ' but should have been UTF-8'); |
| 1042 |
$converted_text = $text; |
| 1043 |
} else { |
| 1044 |
$converted_text = iconv($sourceCharset, $targetCharset, $text); |
| 1045 |
} |
| 1046 |
} |
| 1047 |
|
| 1048 |
// Lets make sure that we don't have that silly BOM issue with any of the utf-8 text we output. |
| 1049 |
if ($targetCharset === 'UTF-8') { |
| 1050 |
if (substr($converted_text, 0, 3) === "\xef\xbb\xbf") { |
| 1051 |
$converted_text = substr($converted_text, 3); |
| 1052 |
} |
| 1053 |
|
| 1054 |
if (substr($converted_text, -3) === "\xef\xbb\xbf") { |
| 1055 |
$converted_text = substr($converted_text, 0, -3); |
| 1056 |
} |
| 1057 |
} |
| 1058 |
|
| 1059 |
return $converted_text; |
| 1060 |
} |
| 1061 |
|
| 1062 |
static function is_utf8($str) |
| 1063 |
{ |
| 1064 |
$c = 0; $b = 0; |
| 1065 |
$bits = 0; |
| 1066 |
$len = strlen($str); |
| 1067 |
for($i = 0; $i < $len; $i++) { |
| 1068 |
$c = ord($str[$i]); |
| 1069 |
if($c > 128) { |
| 1070 |
if(($c >= 254)) { return false; } |
| 1071 |
elseif($c >= 252) { $bits = 6; } |
| 1072 |
elseif($c >= 248) { $bits = 5; } |
| 1073 |
elseif($c >= 240) { $bits = 4; } |
| 1074 |
elseif($c >= 224) { $bits = 3; } |
| 1075 |
elseif($c >= 192) { $bits = 2; } |
| 1076 |
else { return false; } |
| 1077 |
if(($i + $bits) > $len) { return false; } |
| 1078 |
while($bits > 1) { |
| 1079 |
$i++; |
| 1080 |
$b = ord($str[$i]); |
| 1081 |
if($b < 128 || $b > 191) { return false; } |
| 1082 |
$bits--; |
| 1083 |
} |
| 1084 |
} |
| 1085 |
} |
| 1086 |
return true; |
| 1087 |
} |
| 1088 |
|
| 1089 |
function get_display_size() |
| 1090 |
{ |
| 1091 |
$width = -1; |
| 1092 |
$height = -1; |
| 1093 |
|
| 1094 |
if ($this->tag !== 'img') { |
| 1095 |
return false; |
| 1096 |
} |
| 1097 |
|
| 1098 |
// See if there is aheight or width attribute in the tag itself. |
| 1099 |
if (isset($this->attr['width'])) { |
| 1100 |
$width = $this->attr['width']; |
| 1101 |
} |
| 1102 |
|
| 1103 |
if (isset($this->attr['height'])) { |
| 1104 |
$height = $this->attr['height']; |
| 1105 |
} |
| 1106 |
|
| 1107 |
// Now look for an inline style. |
| 1108 |
if (isset($this->attr['style'])) { |
| 1109 |
// Thanks to user gnarf from stackoverflow for this regular expression. |
| 1110 |
$attributes = array(); |
| 1111 |
|
| 1112 |
preg_match_all( |
| 1113 |
'/([\w-]+)\s*:\s*([^;]+)\s*;?/', |
| 1114 |
$this->attr['style'], |
| 1115 |
$matches, |
| 1116 |
PREG_SET_ORDER |
| 1117 |
); |
| 1118 |
|
| 1119 |
foreach ($matches as $match) { |
| 1120 |
$attributes[$match[1]] = $match[2]; |
| 1121 |
} |
| 1122 |
|
| 1123 |
// If there is a width in the style attributes: |
| 1124 |
if (isset($attributes['width']) && $width == -1) { |
| 1125 |
// check that the last two characters are px (pixels) |
| 1126 |
if (strtolower(substr($attributes['width'], -2)) === 'px') { |
| 1127 |
$proposed_width = substr($attributes['width'], 0, -2); |
| 1128 |
// Now make sure that it's an integer and not something stupid. |
| 1129 |
if (filter_var($proposed_width, FILTER_VALIDATE_INT)) { |
| 1130 |
$width = $proposed_width; |
| 1131 |
} |
| 1132 |
} |
| 1133 |
} |
| 1134 |
|
| 1135 |
// If there is a width in the style attributes: |
| 1136 |
if (isset($attributes['height']) && $height == -1) { |
| 1137 |
// check that the last two characters are px (pixels) |
| 1138 |
if (strtolower(substr($attributes['height'], -2)) == 'px') { |
| 1139 |
$proposed_height = substr($attributes['height'], 0, -2); |
| 1140 |
// Now make sure that it's an integer and not something stupid. |
| 1141 |
if (filter_var($proposed_height, FILTER_VALIDATE_INT)) { |
| 1142 |
$height = $proposed_height; |
| 1143 |
} |
| 1144 |
} |
| 1145 |
} |
| 1146 |
|
| 1147 |
} |
| 1148 |
|
| 1149 |
// Future enhancement: |
| 1150 |
// Look in the tag to see if there is a class or id specified that has |
| 1151 |
// a height or width attribute to it. |
| 1152 |
|
| 1153 |
// Far future enhancement |
| 1154 |
// Look at all the parent tags of this image to see if they specify a |
| 1155 |
// class or id that has an img selector that specifies a height or width |
| 1156 |
// Note that in this case, the class or id will have the img subselector |
| 1157 |
// for it to apply to the image. |
| 1158 |
|
| 1159 |
// ridiculously far future development |
| 1160 |
// If the class or id is specified in a SEPARATE css file thats not on |
| 1161 |
// the page, go get it and do what we were just doing for the ones on |
| 1162 |
// the page. |
| 1163 |
|
| 1164 |
$result = array( |
| 1165 |
'height' => $height, |
| 1166 |
'width' => $width |
| 1167 |
); |
| 1168 |
|
| 1169 |
return $result; |
| 1170 |
} |
| 1171 |
|
| 1172 |
function save($filepath = '') |
| 1173 |
{ |
| 1174 |
$ret = $this->outertext(); |
| 1175 |
|
| 1176 |
if ($filepath !== '') { |
| 1177 |
file_put_contents($filepath, $ret, LOCK_EX); |
| 1178 |
} |
| 1179 |
|
| 1180 |
return $ret; |
| 1181 |
} |
| 1182 |
|
| 1183 |
function addClass($class) |
| 1184 |
{ |
| 1185 |
if (is_string($class)) { |
| 1186 |
$class = explode(' ', $class); |
| 1187 |
} |
| 1188 |
|
| 1189 |
if (is_array($class)) { |
| 1190 |
foreach($class as $c) { |
| 1191 |
if (isset($this->class)) { |
| 1192 |
if ($this->hasClass($c)) { |
| 1193 |
continue; |
| 1194 |
} else { |
| 1195 |
$this->class .= ' ' . $c; |
| 1196 |
} |
| 1197 |
} else { |
| 1198 |
$this->class = $c; |
| 1199 |
} |
| 1200 |
} |
| 1201 |
} |
| 1202 |
} |
| 1203 |
|
| 1204 |
function hasClass($class) |
| 1205 |
{ |
| 1206 |
if (is_string($class)) { |
| 1207 |
if (isset($this->class)) { |
| 1208 |
return in_array($class, explode(' ', $this->class), true); |
| 1209 |
} |
| 1210 |
} |
| 1211 |
|
| 1212 |
return false; |
| 1213 |
} |
| 1214 |
|
| 1215 |
function removeClass($class = null) |
| 1216 |
{ |
| 1217 |
if (!isset($this->class)) { |
| 1218 |
return; |
| 1219 |
} |
| 1220 |
|
| 1221 |
if (is_null($class)) { |
| 1222 |
$this->removeAttribute('class'); |
| 1223 |
return; |
| 1224 |
} |
| 1225 |
|
| 1226 |
if (is_string($class)) { |
| 1227 |
$class = explode(' ', $class); |
| 1228 |
} |
| 1229 |
|
| 1230 |
if (is_array($class)) { |
| 1231 |
$class = array_diff(explode(' ', $this->class), $class); |
| 1232 |
if (empty($class)) { |
| 1233 |
$this->removeAttribute('class'); |
| 1234 |
} else { |
| 1235 |
$this->class = implode(' ', $class); |
| 1236 |
} |
| 1237 |
} |
| 1238 |
} |
| 1239 |
|
| 1240 |
function getAllAttributes() |
| 1241 |
{ |
| 1242 |
return $this->attr; |
| 1243 |
} |
| 1244 |
|
| 1245 |
function getAttribute($name) |
| 1246 |
{ |
| 1247 |
return $this->$name; |
| 1248 |
} |
| 1249 |
|
| 1250 |
function setAttribute($name, $value) |
| 1251 |
{ |
| 1252 |
$this->$name = $value; |
| 1253 |
} |
| 1254 |
|
| 1255 |
function hasAttribute($name) |
| 1256 |
{ |
| 1257 |
return isset($this->$name); |
| 1258 |
} |
| 1259 |
|
| 1260 |
function removeAttribute($name) |
| 1261 |
{ |
| 1262 |
unset($this->$name); |
| 1263 |
} |
| 1264 |
|
| 1265 |
function remove() |
| 1266 |
{ |
| 1267 |
if ($this->parent) { |
| 1268 |
$this->parent->removeChild($this); |
| 1269 |
} |
| 1270 |
} |
| 1271 |
|
| 1272 |
function removeChild($node) |
| 1273 |
{ |
| 1274 |
foreach($node->children as $child) { |
| 1275 |
$node->removeChild($child); |
| 1276 |
} |
| 1277 |
|
| 1278 |
// No need to re-index node->children because it is about to be removed! |
| 1279 |
|
| 1280 |
foreach($node->nodes as $entity) { |
| 1281 |
$enidx = array_search($entity, $node->nodes, true); |
| 1282 |
$edidx = array_search($entity, $node->dom->nodes, true); |
| 1283 |
|
| 1284 |
if ($enidx !== false) { |
| 1285 |
unset($node->nodes[$enidx]); |
| 1286 |
} |
| 1287 |
|
| 1288 |
if ($edidx !== false) { |
| 1289 |
unset($node->dom->nodes[$edidx]); |
| 1290 |
} |
| 1291 |
} |
| 1292 |
|
| 1293 |
// No need to re-index node->nodes because it is about to be removed! |
| 1294 |
|
| 1295 |
$nidx = array_search($node, $this->nodes, true); |
| 1296 |
$cidx = array_search($node, $this->children, true); |
| 1297 |
$didx = array_search($node, $this->dom->nodes, true); |
| 1298 |
|
| 1299 |
if ($nidx !== false) { |
| 1300 |
unset($this->nodes[$nidx]); |
| 1301 |
} |
| 1302 |
|
| 1303 |
$this->nodes = array_values($this->nodes); |
| 1304 |
|
| 1305 |
if ($cidx !== false) { |
| 1306 |
unset($this->children[$cidx]); |
| 1307 |
} |
| 1308 |
|
| 1309 |
$this->children = array_values($this->children); |
| 1310 |
|
| 1311 |
if ($didx !== false) { |
| 1312 |
unset($this->dom->nodes[$didx]); |
| 1313 |
} |
| 1314 |
|
| 1315 |
// Do not re-index dom->nodes because nodes point to other nodes in the |
| 1316 |
// array explicitly! |
| 1317 |
|
| 1318 |
$node->clear(); |
| 1319 |
} |
| 1320 |
|
| 1321 |
function getElementById($id) |
| 1322 |
{ |
| 1323 |
return $this->find("#$id", 0); |
| 1324 |
} |
| 1325 |
|
| 1326 |
function getElementsById($id, $idx = null) |
| 1327 |
{ |
| 1328 |
return $this->find("#$id", $idx); |
| 1329 |
} |
| 1330 |
|
| 1331 |
function getElementByTagName($name) |
| 1332 |
{ |
| 1333 |
return $this->find($name, 0); |
| 1334 |
} |
| 1335 |
|
| 1336 |
function getElementsByTagName($name, $idx = null) |
| 1337 |
{ |
| 1338 |
return $this->find($name, $idx); |
| 1339 |
} |
| 1340 |
|
| 1341 |
function parentNode() |
| 1342 |
{ |
| 1343 |
return $this->parent(); |
| 1344 |
} |
| 1345 |
|
| 1346 |
function childNodes($idx = -1) |
| 1347 |
{ |
| 1348 |
if ($idx === -1) { |
| 1349 |
return $this->children; |
| 1350 |
} |
| 1351 |
|
| 1352 |
if (isset($this->children[$idx])) { |
| 1353 |
return $this->children[$idx]; |
| 1354 |
} |
| 1355 |
|
| 1356 |
return null; |
| 1357 |
} |
| 1358 |
|
| 1359 |
function firstChild() |
| 1360 |
{ |
| 1361 |
if (count($this->children) > 0) { |
| 1362 |
return $this->children[0]; |
| 1363 |
} |
| 1364 |
return null; |
| 1365 |
} |
| 1366 |
|
| 1367 |
function lastChild() |
| 1368 |
{ |
| 1369 |
if (count($this->children) > 0) { |
| 1370 |
return end($this->children); |
| 1371 |
} |
| 1372 |
return null; |
| 1373 |
} |
| 1374 |
|
| 1375 |
function nextSibling() |
| 1376 |
{ |
| 1377 |
if ($this->parent === null) { |
| 1378 |
return null; |
| 1379 |
} |
| 1380 |
|
| 1381 |
$idx = array_search($this, $this->parent->children, true); |
| 1382 |
|
| 1383 |
if ($idx !== false && isset($this->parent->children[$idx + 1])) { |
| 1384 |
return $this->parent->children[$idx + 1]; |
| 1385 |
} |
| 1386 |
|
| 1387 |
return null; |
| 1388 |
} |
| 1389 |
|
| 1390 |
function previousSibling() |
| 1391 |
{ |
| 1392 |
if ($this->parent === null) { |
| 1393 |
return null; |
| 1394 |
} |
| 1395 |
|
| 1396 |
$idx = array_search($this, $this->parent->children, true); |
| 1397 |
|
| 1398 |
if ($idx !== false && $idx > 0) { |
| 1399 |
return $this->parent->children[$idx - 1]; |
| 1400 |
} |
| 1401 |
|
| 1402 |
return null; |
| 1403 |
|
| 1404 |
} |
| 1405 |
|
| 1406 |
function hasChildNodes() |
| 1407 |
{ |
| 1408 |
return !empty($this->children); |
| 1409 |
} |
| 1410 |
|
| 1411 |
function nodeName() |
| 1412 |
{ |
| 1413 |
return $this->tag; |
| 1414 |
} |
| 1415 |
|
| 1416 |
function appendChild($node) |
| 1417 |
{ |
| 1418 |
$node->parent = $this; |
| 1419 |
$this->nodes[] = $node; |
| 1420 |
$this->children[] = $node; |
| 1421 |
|
| 1422 |
if ($this->dom) { // Attach current node to DOM (recursively) |
| 1423 |
$children = array($node); |
| 1424 |
|
| 1425 |
while($children) { |
| 1426 |
$child = array_pop($children); |
| 1427 |
$children = array_merge($children, $child->children); |
| 1428 |
|
| 1429 |
$this->dom->nodes[] = $child; |
| 1430 |
$child->dom = $this->dom; |
| 1431 |
$child->_[self::HDOM_INFO_BEGIN] = count($this->dom->nodes) - 1; |
| 1432 |
$child->_[self::HDOM_INFO_END] = $child->_[self::HDOM_INFO_BEGIN]; |
| 1433 |
} |
| 1434 |
|
| 1435 |
$this->dom->root->_[self::HDOM_INFO_END] = count($this->dom->nodes) - 1; |
| 1436 |
} |
| 1437 |
|
| 1438 |
return $this; |
| 1439 |
} |
| 1440 |
|
| 1441 |
} |
| 1442 |
|
