| 1 |
<?php |
| 2 |
/** |
| 3 |
* Website: http://sourceforge.net/projects/simplehtmldom/ |
| 4 |
* Additional projects: http://sourceforge.net/projects/debugobject/ |
| 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. 1.9.1 (291) |
| 22 |
*/ |
| 23 |
|
| 24 |
|
| 25 |
if (!defined('ABSPATH')){ exit(); } |
| 26 |
|
| 27 |
# edited |
| 28 |
defined('HGFLMAX_FILE_SIZE') || define('HGFLMAX_FILE_SIZE', 2000000); # Process HTML up to 2 Mb |
| 29 |
defined('HGFLDEFAULT_TARGET_CHARSET') || define('HGFLDEFAULT_TARGET_CHARSET', 'UTF-8'); |
| 30 |
defined('HGFLDEFAULT_BR_TEXT') || define('HGFLDEFAULT_BR_TEXT', "\r\n"); |
| 31 |
defined('HGFLDEFAULT_SPAN_TEXT') || define('HGFLDEFAULT_SPAN_TEXT', ' '); |
| 32 |
|
| 33 |
define('HGFLHDOM_TYPE_ELEMENT', 1); |
| 34 |
define('HGFLHDOM_TYPE_COMMENT', 2); |
| 35 |
define('HGFLHDOM_TYPE_TEXT', 3); |
| 36 |
define('HGFLHDOM_TYPE_ENDTAG', 4); |
| 37 |
define('HGFLHDOM_TYPE_ROOT', 5); |
| 38 |
define('HGFLHDOM_TYPE_UNKNOWN', 6); |
| 39 |
define('HGFLHDOM_QUOTE_DOUBLE', 0); |
| 40 |
define('HGFLHDOM_QUOTE_SINGLE', 1); |
| 41 |
define('HGFLHDOM_QUOTE_NO', 3); |
| 42 |
define('HGFLHDOM_INFO_BEGIN', 0); |
| 43 |
define('HGFLHDOM_INFO_END', 1); |
| 44 |
define('HGFLHDOM_INFO_QUOTE', 2); |
| 45 |
define('HGFLHDOM_INFO_SPACE', 3); |
| 46 |
define('HGFLHDOM_INFO_TEXT', 4); |
| 47 |
define('HGFLHDOM_INFO_INNER', 5); |
| 48 |
define('HGFLHDOM_INFO_OUTER', 6); |
| 49 |
define('HGFLHDOM_INFO_ENDSPACE', 7); |
| 50 |
|
| 51 |
defined('HGFLDEFAULT_TARGET_CHARSET') || define('HGFLDEFAULT_TARGET_CHARSET', 'UTF-8'); |
| 52 |
defined('HGFLDEFAULT_BR_TEXT') || define('HGFLDEFAULT_BR_TEXT', "\r\n"); |
| 53 |
defined('HGFLDEFAULT_SPAN_TEXT') || define('HGFLDEFAULT_SPAN_TEXT', ' '); |
| 54 |
defined('HGFLMAX_FILE_SIZE') || define('HGFLMAX_FILE_SIZE', 600000); |
| 55 |
define('HGFLHDOM_SMARTY_AS_TEXT', 1); |
| 56 |
|
| 57 |
function hgfl_file_get_html( |
| 58 |
$url, |
| 59 |
$use_include_path = false, |
| 60 |
$context = null, |
| 61 |
$offset = 0, |
| 62 |
$maxLen = -1, |
| 63 |
$lowercase = true, |
| 64 |
$forceTagsClosed = true, |
| 65 |
$target_charset = HGFLDEFAULT_TARGET_CHARSET, |
| 66 |
$stripRN = true, |
| 67 |
$defaultBRText = HGFLDEFAULT_BR_TEXT, |
| 68 |
$defaultSpanText = HGFLDEFAULT_SPAN_TEXT) |
| 69 |
{ |
| 70 |
if($maxLen <= 0) { $maxLen = HGFLMAX_FILE_SIZE; } |
| 71 |
|
| 72 |
$dom = new hgfl_simple_html_dom( |
| 73 |
null, |
| 74 |
$lowercase, |
| 75 |
$forceTagsClosed, |
| 76 |
$target_charset, |
| 77 |
$stripRN, |
| 78 |
$defaultBRText, |
| 79 |
$defaultSpanText |
| 80 |
); |
| 81 |
|
| 82 |
/** |
| 83 |
* For sourceforge users: uncomment the next line and comment the |
| 84 |
* retrieve_url_contents line 2 lines down if it is not already done. |
| 85 |
*/ |
| 86 |
$contents = file_get_contents( |
| 87 |
$url, |
| 88 |
$use_include_path, |
| 89 |
$context, |
| 90 |
$offset, |
| 91 |
$maxLen |
| 92 |
); |
| 93 |
// $contents = retrieve_url_contents($url); |
| 94 |
|
| 95 |
if (empty($contents) || strlen($contents) > $maxLen) { |
| 96 |
$dom->clear(); |
| 97 |
return false; |
| 98 |
} |
| 99 |
|
| 100 |
return $dom->load($contents, $lowercase, $stripRN); |
| 101 |
} |
| 102 |
|
| 103 |
function hgfl_str_get_html( |
| 104 |
$str, |
| 105 |
$lowercase = true, |
| 106 |
$forceTagsClosed = true, |
| 107 |
$target_charset = HGFLDEFAULT_TARGET_CHARSET, |
| 108 |
$stripRN = true, |
| 109 |
$defaultBRText = HGFLDEFAULT_BR_TEXT, |
| 110 |
$defaultSpanText = HGFLDEFAULT_SPAN_TEXT) |
| 111 |
{ |
| 112 |
$dom = new hgfl_simple_html_dom( |
| 113 |
null, |
| 114 |
$lowercase, |
| 115 |
$forceTagsClosed, |
| 116 |
$target_charset, |
| 117 |
$stripRN, |
| 118 |
$defaultBRText, |
| 119 |
$defaultSpanText |
| 120 |
); |
| 121 |
|
| 122 |
if (empty($str) || strlen($str) > HGFLMAX_FILE_SIZE) { |
| 123 |
$dom->clear(); |
| 124 |
return false; |
| 125 |
} |
| 126 |
|
| 127 |
return $dom->load($str, $lowercase, $stripRN); |
| 128 |
} |
| 129 |
|
| 130 |
function hgfl_dump_html_tree($node, $show_attr = true, $deep = 0) |
| 131 |
{ |
| 132 |
$node->dump($node); |
| 133 |
} |
| 134 |
|
| 135 |
class hgfl_simple_html_dom_node |
| 136 |
{ |
| 137 |
public $nodetype = HGFLHDOM_TYPE_TEXT; |
| 138 |
public $tag = 'text'; |
| 139 |
public $attr = array(); |
| 140 |
public $children = array(); |
| 141 |
public $nodes = array(); |
| 142 |
public $parent = null; |
| 143 |
public $_ = array(); |
| 144 |
public $tag_start = 0; |
| 145 |
private $dom = null; |
| 146 |
|
| 147 |
function __construct($dom) |
| 148 |
{ |
| 149 |
$this->dom = $dom; |
| 150 |
$dom->nodes[] = $this; |
| 151 |
} |
| 152 |
|
| 153 |
function __destruct() |
| 154 |
{ |
| 155 |
$this->clear(); |
| 156 |
} |
| 157 |
|
| 158 |
function __toString() |
| 159 |
{ |
| 160 |
return $this->outertext(); |
| 161 |
} |
| 162 |
|
| 163 |
function clear() |
| 164 |
{ |
| 165 |
$this->dom = null; |
| 166 |
$this->nodes = null; |
| 167 |
$this->parent = null; |
| 168 |
$this->children = null; |
| 169 |
} |
| 170 |
|
| 171 |
function dump($show_attr = true, $depth = 0) |
| 172 |
{ |
| 173 |
echo str_repeat("\t", $depth) . $this->tag; |
| 174 |
|
| 175 |
if ($show_attr && count($this->attr) > 0) { |
| 176 |
echo '('; |
| 177 |
foreach ($this->attr as $k => $v) { |
| 178 |
echo "[$k]=>\"$v\", "; |
| 179 |
} |
| 180 |
echo ')'; |
| 181 |
} |
| 182 |
|
| 183 |
echo "\n"; |
| 184 |
|
| 185 |
if ($this->nodes) { |
| 186 |
foreach ($this->nodes as $node) { |
| 187 |
$node->dump($show_attr, $depth + 1); |
| 188 |
} |
| 189 |
} |
| 190 |
} |
| 191 |
|
| 192 |
function dump_node($echo = true) |
| 193 |
{ |
| 194 |
$string = $this->tag; |
| 195 |
|
| 196 |
if (count($this->attr) > 0) { |
| 197 |
$string .= '('; |
| 198 |
foreach ($this->attr as $k => $v) { |
| 199 |
$string .= "[$k]=>\"$v\", "; |
| 200 |
} |
| 201 |
$string .= ')'; |
| 202 |
} |
| 203 |
|
| 204 |
if (count($this->_) > 0) { |
| 205 |
$string .= ' $_ ('; |
| 206 |
foreach ($this->_ as $k => $v) { |
| 207 |
if (is_array($v)) { |
| 208 |
$string .= "[$k]=>("; |
| 209 |
foreach ($v as $k2 => $v2) { |
| 210 |
$string .= "[$k2]=>\"$v2\", "; |
| 211 |
} |
| 212 |
$string .= ')'; |
| 213 |
} else { |
| 214 |
$string .= "[$k]=>\"$v\", "; |
| 215 |
} |
| 216 |
} |
| 217 |
$string .= ')'; |
| 218 |
} |
| 219 |
|
| 220 |
if (isset($this->text)) { |
| 221 |
$string .= " text: ({$this->text})"; |
| 222 |
} |
| 223 |
|
| 224 |
$string .= ' HGFLHDOM_INNER_INFO: '; |
| 225 |
|
| 226 |
if (isset($node->_[HGFLHDOM_INFO_INNER])) { |
| 227 |
$string .= "'" . $node->_[HGFLHDOM_INFO_INNER] . "'"; |
| 228 |
} else { |
| 229 |
$string .= ' NULL '; |
| 230 |
} |
| 231 |
|
| 232 |
$string .= ' children: ' . count($this->children); |
| 233 |
$string .= ' nodes: ' . count($this->nodes); |
| 234 |
$string .= ' tag_start: ' . $this->tag_start; |
| 235 |
$string .= "\n"; |
| 236 |
|
| 237 |
if ($echo) { |
| 238 |
echo $string; |
| 239 |
return; |
| 240 |
} else { |
| 241 |
return $string; |
| 242 |
} |
| 243 |
} |
| 244 |
|
| 245 |
function parent($parent = null) |
| 246 |
{ |
| 247 |
// I am SURE that this doesn't work properly. |
| 248 |
// It fails to unset the current node from it's current parents nodes or |
| 249 |
// children list first. |
| 250 |
if ($parent !== null) { |
| 251 |
$this->parent = $parent; |
| 252 |
$this->parent->nodes[] = $this; |
| 253 |
$this->parent->children[] = $this; |
| 254 |
} |
| 255 |
|
| 256 |
return $this->parent; |
| 257 |
} |
| 258 |
|
| 259 |
function has_child() |
| 260 |
{ |
| 261 |
return !empty($this->children); |
| 262 |
} |
| 263 |
|
| 264 |
function children($idx = -1) |
| 265 |
{ |
| 266 |
if ($idx === -1) { |
| 267 |
return $this->children; |
| 268 |
} |
| 269 |
|
| 270 |
if (isset($this->children[$idx])) { |
| 271 |
return $this->children[$idx]; |
| 272 |
} |
| 273 |
|
| 274 |
return null; |
| 275 |
} |
| 276 |
|
| 277 |
function first_child() |
| 278 |
{ |
| 279 |
if (count($this->children) > 0) { |
| 280 |
return $this->children[0]; |
| 281 |
} |
| 282 |
return null; |
| 283 |
} |
| 284 |
|
| 285 |
function last_child() |
| 286 |
{ |
| 287 |
if (count($this->children) > 0) { |
| 288 |
return end($this->children); |
| 289 |
} |
| 290 |
return null; |
| 291 |
} |
| 292 |
|
| 293 |
function next_sibling() |
| 294 |
{ |
| 295 |
if ($this->parent === null) { |
| 296 |
return null; |
| 297 |
} |
| 298 |
|
| 299 |
$idx = array_search($this, $this->parent->children, true); |
| 300 |
|
| 301 |
if ($idx !== false && isset($this->parent->children[$idx + 1])) { |
| 302 |
return $this->parent->children[$idx + 1]; |
| 303 |
} |
| 304 |
|
| 305 |
return null; |
| 306 |
} |
| 307 |
|
| 308 |
function prev_sibling() |
| 309 |
{ |
| 310 |
if ($this->parent === null) { |
| 311 |
return null; |
| 312 |
} |
| 313 |
|
| 314 |
$idx = array_search($this, $this->parent->children, true); |
| 315 |
|
| 316 |
if ($idx !== false && $idx > 0) { |
| 317 |
return $this->parent->children[$idx - 1]; |
| 318 |
} |
| 319 |
|
| 320 |
return null; |
| 321 |
} |
| 322 |
|
| 323 |
function find_ancestor_tag($tag) |
| 324 |
{ |
| 325 |
global $debug_object; |
| 326 |
if (is_object($debug_object)) { $debug_object->debug_log_entry(1); } |
| 327 |
|
| 328 |
if ($this->parent === null) { |
| 329 |
return null; |
| 330 |
} |
| 331 |
|
| 332 |
$ancestor = $this->parent; |
| 333 |
|
| 334 |
while (!is_null($ancestor)) { |
| 335 |
if (is_object($debug_object)) { |
| 336 |
$debug_object->debug_log(2, 'Current tag is: ' . $ancestor->tag); |
| 337 |
} |
| 338 |
|
| 339 |
if ($ancestor->tag === $tag) { |
| 340 |
break; |
| 341 |
} |
| 342 |
|
| 343 |
$ancestor = $ancestor->parent; |
| 344 |
} |
| 345 |
|
| 346 |
return $ancestor; |
| 347 |
} |
| 348 |
|
| 349 |
function innertext() |
| 350 |
{ |
| 351 |
if (isset($this->_[HGFLHDOM_INFO_INNER])) { |
| 352 |
return $this->_[HGFLHDOM_INFO_INNER]; |
| 353 |
} |
| 354 |
|
| 355 |
if (isset($this->_[HGFLHDOM_INFO_TEXT])) { |
| 356 |
return $this->dom->restore_noise($this->_[HGFLHDOM_INFO_TEXT]); |
| 357 |
} |
| 358 |
|
| 359 |
$ret = ''; |
| 360 |
|
| 361 |
foreach ($this->nodes as $n) { |
| 362 |
$ret .= $n->outertext(); |
| 363 |
} |
| 364 |
|
| 365 |
return $ret; |
| 366 |
} |
| 367 |
|
| 368 |
function outertext() |
| 369 |
{ |
| 370 |
global $debug_object; |
| 371 |
|
| 372 |
if (is_object($debug_object)) { |
| 373 |
$text = ''; |
| 374 |
|
| 375 |
if ($this->tag === 'text') { |
| 376 |
if (!empty($this->text)) { |
| 377 |
$text = ' with text: ' . $this->text; |
| 378 |
} |
| 379 |
} |
| 380 |
|
| 381 |
$debug_object->debug_log(1, 'Innertext of tag: ' . $this->tag . $text); |
| 382 |
} |
| 383 |
|
| 384 |
if ($this->tag === 'root') { |
| 385 |
return $this->innertext(); |
| 386 |
} |
| 387 |
|
| 388 |
// todo: What is the use of this callback? Remove? |
| 389 |
if ($this->dom && $this->dom->callback !== null) { |
| 390 |
call_user_func_array($this->dom->callback, array($this)); |
| 391 |
} |
| 392 |
|
| 393 |
if (isset($this->_[HGFLHDOM_INFO_OUTER])) { |
| 394 |
return $this->_[HGFLHDOM_INFO_OUTER]; |
| 395 |
} |
| 396 |
|
| 397 |
if (isset($this->_[HGFLHDOM_INFO_TEXT])) { |
| 398 |
return $this->dom->restore_noise($this->_[HGFLHDOM_INFO_TEXT]); |
| 399 |
} |
| 400 |
|
| 401 |
$ret = ''; |
| 402 |
|
| 403 |
if ($this->dom && $this->dom->nodes[$this->_[HGFLHDOM_INFO_BEGIN]]) { |
| 404 |
$ret = $this->dom->nodes[$this->_[HGFLHDOM_INFO_BEGIN]]->makeup(); |
| 405 |
} |
| 406 |
|
| 407 |
if (isset($this->_[HGFLHDOM_INFO_INNER])) { |
| 408 |
// todo: <br> should either never have HGFLHDOM_INFO_INNER or always |
| 409 |
if ($this->tag !== 'br') { |
| 410 |
$ret .= $this->_[HGFLHDOM_INFO_INNER]; |
| 411 |
} |
| 412 |
} elseif ($this->nodes) { |
| 413 |
foreach ($this->nodes as $n) { |
| 414 |
$ret .= $this->convert_text($n->outertext()); |
| 415 |
} |
| 416 |
} |
| 417 |
|
| 418 |
if (isset($this->_[HGFLHDOM_INFO_END]) && $this->_[HGFLHDOM_INFO_END] != 0) { |
| 419 |
$ret .= '</' . $this->tag . '>'; |
| 420 |
} |
| 421 |
|
| 422 |
return $ret; |
| 423 |
} |
| 424 |
|
| 425 |
function text() |
| 426 |
{ |
| 427 |
if (isset($this->_[HGFLHDOM_INFO_INNER])) { |
| 428 |
return $this->_[HGFLHDOM_INFO_INNER]; |
| 429 |
} |
| 430 |
|
| 431 |
switch ($this->nodetype) { |
| 432 |
case HGFLHDOM_TYPE_TEXT: return $this->dom->restore_noise($this->_[HGFLHDOM_INFO_TEXT]); |
| 433 |
case HGFLHDOM_TYPE_COMMENT: return ''; |
| 434 |
case HGFLHDOM_TYPE_UNKNOWN: return ''; |
| 435 |
} |
| 436 |
|
| 437 |
if (strcasecmp($this->tag, 'script') === 0) { return ''; } |
| 438 |
if (strcasecmp($this->tag, 'style') === 0) { return ''; } |
| 439 |
|
| 440 |
$ret = ''; |
| 441 |
|
| 442 |
// In rare cases, (always node type 1 or HGFLHDOM_TYPE_ELEMENT - observed |
| 443 |
// for some span tags, and some p tags) $this->nodes is set to NULL. |
| 444 |
// NOTE: This indicates that there is a problem where it's set to NULL |
| 445 |
// without a clear happening. |
| 446 |
// WHY is this happening? |
| 447 |
if (!is_null($this->nodes)) { |
| 448 |
foreach ($this->nodes as $n) { |
| 449 |
// Start paragraph after a blank line |
| 450 |
if ($n->tag === 'p') { |
| 451 |
$ret = trim($ret) . "\n\n"; |
| 452 |
} |
| 453 |
|
| 454 |
$ret .= $this->convert_text($n->text()); |
| 455 |
|
| 456 |
// If this node is a span... add a space at the end of it so |
| 457 |
// multiple spans don't run into each other. This is plaintext |
| 458 |
// after all. |
| 459 |
if ($n->tag === 'span') { |
| 460 |
$ret .= $this->dom->HGFLDEFAULT_span_text; |
| 461 |
} |
| 462 |
} |
| 463 |
} |
| 464 |
return $ret; |
| 465 |
} |
| 466 |
|
| 467 |
function xmltext() |
| 468 |
{ |
| 469 |
$ret = $this->innertext(); |
| 470 |
$ret = str_ireplace('<![CDATA[', '', $ret); |
| 471 |
$ret = str_replace(']]>', '', $ret); |
| 472 |
return $ret; |
| 473 |
} |
| 474 |
|
| 475 |
function makeup() |
| 476 |
{ |
| 477 |
// text, comment, unknown |
| 478 |
if (isset($this->_[HGFLHDOM_INFO_TEXT])) { |
| 479 |
return $this->dom->restore_noise($this->_[HGFLHDOM_INFO_TEXT]); |
| 480 |
} |
| 481 |
|
| 482 |
$ret = '<' . $this->tag; |
| 483 |
$i = -1; |
| 484 |
|
| 485 |
foreach ($this->attr as $key => $val) { |
| 486 |
++$i; |
| 487 |
|
| 488 |
// skip removed attribute |
| 489 |
if ($val === null || $val === false) { continue; } |
| 490 |
|
| 491 |
$ret .= $this->_[HGFLHDOM_INFO_SPACE][$i][0]; |
| 492 |
|
| 493 |
//no value attr: nowrap, checked selected... |
| 494 |
if ($val === true) { |
| 495 |
$ret .= $key; |
| 496 |
} else { |
| 497 |
switch ($this->_[HGFLHDOM_INFO_QUOTE][$i]) |
| 498 |
{ |
| 499 |
case HGFLHDOM_QUOTE_DOUBLE: $quote = '"'; break; |
| 500 |
case HGFLHDOM_QUOTE_SINGLE: $quote = '\''; break; |
| 501 |
default: $quote = ''; |
| 502 |
} |
| 503 |
|
| 504 |
$ret .= $key |
| 505 |
. $this->_[HGFLHDOM_INFO_SPACE][$i][1] |
| 506 |
. '=' |
| 507 |
. $this->_[HGFLHDOM_INFO_SPACE][$i][2] |
| 508 |
. $quote |
| 509 |
. $val |
| 510 |
. $quote; |
| 511 |
} |
| 512 |
} |
| 513 |
|
| 514 |
$ret = $this->dom->restore_noise($ret); |
| 515 |
return $ret . $this->_[HGFLHDOM_INFO_ENDSPACE] . '>'; |
| 516 |
} |
| 517 |
|
| 518 |
function find($selector, $idx = null, $lowercase = false) |
| 519 |
{ |
| 520 |
$selectors = $this->parse_selector($selector); |
| 521 |
if (($count = count($selectors)) === 0) { return array(); } |
| 522 |
$found_keys = array(); |
| 523 |
|
| 524 |
// find each selector |
| 525 |
for ($c = 0; $c < $count; ++$c) { |
| 526 |
// The change on the below line was documented on the sourceforge |
| 527 |
// code tracker id 2788009 |
| 528 |
// used to be: if (($levle=count($selectors[0]))===0) return array(); |
| 529 |
if (($levle = count($selectors[$c])) === 0) { return array(); } |
| 530 |
if (!isset($this->_[HGFLHDOM_INFO_BEGIN])) { return array(); } |
| 531 |
|
| 532 |
$head = array($this->_[HGFLHDOM_INFO_BEGIN] => 1); |
| 533 |
$cmd = ' '; // Combinator |
| 534 |
|
| 535 |
// handle descendant selectors, no recursive! |
| 536 |
for ($l = 0; $l < $levle; ++$l) { |
| 537 |
$ret = array(); |
| 538 |
|
| 539 |
foreach ($head as $k => $v) { |
| 540 |
$n = ($k === -1) ? $this->dom->root : $this->dom->nodes[$k]; |
| 541 |
//PaperG - Pass this optional parameter on to the seek function. |
| 542 |
$n->seek($selectors[$c][$l], $ret, $cmd, $lowercase); |
| 543 |
} |
| 544 |
|
| 545 |
$head = $ret; |
| 546 |
$cmd = $selectors[$c][$l][4]; // Next Combinator |
| 547 |
} |
| 548 |
|
| 549 |
foreach ($head as $k => $v) { |
| 550 |
if (!isset($found_keys[$k])) { |
| 551 |
$found_keys[$k] = 1; |
| 552 |
} |
| 553 |
} |
| 554 |
} |
| 555 |
|
| 556 |
// sort keys |
| 557 |
ksort($found_keys); |
| 558 |
|
| 559 |
$found = array(); |
| 560 |
foreach ($found_keys as $k => $v) { |
| 561 |
$found[] = $this->dom->nodes[$k]; |
| 562 |
} |
| 563 |
|
| 564 |
// return nth-element or array |
| 565 |
if (is_null($idx)) { return $found; } |
| 566 |
elseif ($idx < 0) { $idx = count($found) + $idx; } |
| 567 |
return (isset($found[$idx])) ? $found[$idx] : null; |
| 568 |
} |
| 569 |
|
| 570 |
protected function seek($selector, &$ret, $parent_cmd, $lowercase = false) |
| 571 |
{ |
| 572 |
global $debug_object; |
| 573 |
if (is_object($debug_object)) { $debug_object->debug_log_entry(1); } |
| 574 |
|
| 575 |
list($tag, $id, $class, $attributes, $cmb) = $selector; |
| 576 |
$nodes = array(); |
| 577 |
|
| 578 |
if ($parent_cmd === ' ') { // Descendant Combinator |
| 579 |
// Find parent closing tag if the current element doesn't have a closing |
| 580 |
// tag (i.e. void element) |
| 581 |
$end = (!empty($this->_[HGFLHDOM_INFO_END])) ? $this->_[HGFLHDOM_INFO_END] : 0; |
| 582 |
if ($end == 0) { |
| 583 |
$parent = $this->parent; |
| 584 |
while (!isset($parent->_[HGFLHDOM_INFO_END]) && $parent !== null) { |
| 585 |
$end -= 1; |
| 586 |
$parent = $parent->parent; |
| 587 |
} |
| 588 |
$end += $parent->_[HGFLHDOM_INFO_END]; |
| 589 |
} |
| 590 |
|
| 591 |
// Get list of target nodes |
| 592 |
$nodes_start = $this->_[HGFLHDOM_INFO_BEGIN] + 1; |
| 593 |
$nodes_count = $end - $nodes_start; |
| 594 |
$nodes = array_slice($this->dom->nodes, $nodes_start, $nodes_count, true); |
| 595 |
} elseif ($parent_cmd === '>') { // Child Combinator |
| 596 |
$nodes = $this->children; |
| 597 |
} elseif ($parent_cmd === '+' |
| 598 |
&& $this->parent |
| 599 |
&& in_array($this, $this->parent->children)) { // Next-Sibling Combinator |
| 600 |
$index = array_search($this, $this->parent->children, true) + 1; |
| 601 |
if ($index < count($this->parent->children)) |
| 602 |
$nodes[] = $this->parent->children[$index]; |
| 603 |
} elseif ($parent_cmd === '~' |
| 604 |
&& $this->parent |
| 605 |
&& in_array($this, $this->parent->children)) { // Subsequent Sibling Combinator |
| 606 |
$index = array_search($this, $this->parent->children, true); |
| 607 |
$nodes = array_slice($this->parent->children, $index); |
| 608 |
} |
| 609 |
|
| 610 |
// Go throgh each element starting at this element until the end tag |
| 611 |
// Note: If this element is a void tag, any previous void element is |
| 612 |
// skipped. |
| 613 |
foreach($nodes as $node) { |
| 614 |
$pass = true; |
| 615 |
|
| 616 |
// Skip root nodes |
| 617 |
if(!$node->parent) { |
| 618 |
$pass = false; |
| 619 |
} |
| 620 |
|
| 621 |
// Handle 'text' selector |
| 622 |
if($pass && $tag === 'text' && $node->tag === 'text') { |
| 623 |
$ret[array_search($node, $this->dom->nodes, true)] = 1; |
| 624 |
unset($node); |
| 625 |
continue; |
| 626 |
} |
| 627 |
|
| 628 |
// Skip if node isn't a child node (i.e. text nodes) |
| 629 |
if($pass && !in_array($node, $node->parent->children, true)) { |
| 630 |
$pass = false; |
| 631 |
} |
| 632 |
|
| 633 |
// Skip if tag doesn't match |
| 634 |
if ($pass && $tag !== '' && $tag !== $node->tag && $tag !== '*') { |
| 635 |
$pass = false; |
| 636 |
} |
| 637 |
|
| 638 |
// Skip if ID doesn't exist |
| 639 |
if ($pass && $id !== '' && !isset($node->attr['id'])) { |
| 640 |
$pass = false; |
| 641 |
} |
| 642 |
|
| 643 |
// Check if ID matches |
| 644 |
if ($pass && $id !== '' && isset($node->attr['id'])) { |
| 645 |
// Note: Only consider the first ID (as browsers do) |
| 646 |
$node_id = explode(' ', trim($node->attr['id']))[0]; |
| 647 |
|
| 648 |
if($id !== $node_id) { $pass = false; } |
| 649 |
} |
| 650 |
|
| 651 |
// Check if all class(es) exist |
| 652 |
if ($pass && $class !== '' && is_array($class) && !empty($class)) { |
| 653 |
if (isset($node->attr['class'])) { |
| 654 |
$node_classes = explode(' ', $node->attr['class']); |
| 655 |
|
| 656 |
if ($lowercase) { |
| 657 |
$node_classes = array_map('strtolower', $node_classes); |
| 658 |
} |
| 659 |
|
| 660 |
foreach($class as $c) { |
| 661 |
if(!in_array($c, $node_classes)) { |
| 662 |
$pass = false; |
| 663 |
break; |
| 664 |
} |
| 665 |
} |
| 666 |
} else { |
| 667 |
$pass = false; |
| 668 |
} |
| 669 |
} |
| 670 |
|
| 671 |
// Check attributes |
| 672 |
if ($pass |
| 673 |
&& $attributes !== '' |
| 674 |
&& is_array($attributes) |
| 675 |
&& !empty($attributes)) { |
| 676 |
foreach($attributes as $a) { |
| 677 |
list ( |
| 678 |
$att_name, |
| 679 |
$att_expr, |
| 680 |
$att_val, |
| 681 |
$att_inv, |
| 682 |
$att_case_sensitivity |
| 683 |
) = $a; |
| 684 |
|
| 685 |
// Handle indexing attributes (i.e. "[2]") |
| 686 |
/** |
| 687 |
* Note: This is not supported by the CSS Standard but adds |
| 688 |
* the ability to select items compatible to XPath (i.e. |
| 689 |
* the 3rd element within it's parent). |
| 690 |
* |
| 691 |
* Note: This doesn't conflict with the CSS Standard which |
| 692 |
* doesn't work on numeric attributes anyway. |
| 693 |
*/ |
| 694 |
if (is_numeric($att_name) |
| 695 |
&& $att_expr === '' |
| 696 |
&& $att_val === '') { |
| 697 |
$count = 0; |
| 698 |
|
| 699 |
// Find index of current element in parent |
| 700 |
foreach ($node->parent->children as $c) { |
| 701 |
if ($c->tag === $node->tag) ++$count; |
| 702 |
if ($c === $node) break; |
| 703 |
} |
| 704 |
|
| 705 |
// If this is the correct node, continue with next |
| 706 |
// attribute |
| 707 |
if ($count === (int)$att_name) continue; |
| 708 |
} |
| 709 |
|
| 710 |
// Check attribute availability |
| 711 |
if ($att_inv) { // Attribute should NOT be set |
| 712 |
if (isset($node->attr[$att_name])) { |
| 713 |
$pass = false; |
| 714 |
break; |
| 715 |
} |
| 716 |
} else { // Attribute should be set |
| 717 |
// todo: "plaintext" is not a valid CSS selector! |
| 718 |
if ($att_name !== 'plaintext' |
| 719 |
&& !isset($node->attr[$att_name])) { |
| 720 |
$pass = false; |
| 721 |
break; |
| 722 |
} |
| 723 |
} |
| 724 |
|
| 725 |
// Continue with next attribute if expression isn't defined |
| 726 |
if ($att_expr === '') continue; |
| 727 |
|
| 728 |
// If they have told us that this is a "plaintext" |
| 729 |
// search then we want the plaintext of the node - right? |
| 730 |
// todo "plaintext" is not a valid CSS selector! |
| 731 |
if ($att_name === 'plaintext') { |
| 732 |
$nodeKeyValue = $node->text(); |
| 733 |
} else { |
| 734 |
$nodeKeyValue = $node->attr[$att_name]; |
| 735 |
} |
| 736 |
|
| 737 |
if (is_object($debug_object)) { |
| 738 |
$debug_object->debug_log(2, |
| 739 |
'testing node: ' |
| 740 |
. $node->tag |
| 741 |
. ' for attribute: ' |
| 742 |
. $att_name |
| 743 |
. $att_expr |
| 744 |
. $att_val |
| 745 |
. ' where nodes value is: ' |
| 746 |
. $nodeKeyValue |
| 747 |
); |
| 748 |
} |
| 749 |
|
| 750 |
// If lowercase is set, do a case insensitive test of |
| 751 |
// the value of the selector. |
| 752 |
if ($lowercase) { |
| 753 |
$check = $this->match( |
| 754 |
$att_expr, |
| 755 |
strtolower($att_val), |
| 756 |
strtolower($nodeKeyValue), |
| 757 |
$att_case_sensitivity |
| 758 |
); |
| 759 |
} else { |
| 760 |
$check = $this->match( |
| 761 |
$att_expr, |
| 762 |
$att_val, |
| 763 |
$nodeKeyValue, |
| 764 |
$att_case_sensitivity |
| 765 |
); |
| 766 |
} |
| 767 |
|
| 768 |
if (is_object($debug_object)) { |
| 769 |
$debug_object->debug_log(2, |
| 770 |
'after match: ' |
| 771 |
. ($check ? 'true' : 'false') |
| 772 |
); |
| 773 |
} |
| 774 |
|
| 775 |
if (!$check) { |
| 776 |
$pass = false; |
| 777 |
break; |
| 778 |
} |
| 779 |
} |
| 780 |
} |
| 781 |
|
| 782 |
// Found a match. Add to list and clear node |
| 783 |
if ($pass) $ret[$node->_[HGFLHDOM_INFO_BEGIN]] = 1; |
| 784 |
unset($node); |
| 785 |
} |
| 786 |
// It's passed by reference so this is actually what this function returns. |
| 787 |
if (is_object($debug_object)) { |
| 788 |
$debug_object->debug_log(1, 'EXIT - ret: ', $ret); |
| 789 |
} |
| 790 |
} |
| 791 |
|
| 792 |
protected function match($exp, $pattern, $value, $case_sensitivity) |
| 793 |
{ |
| 794 |
global $debug_object; |
| 795 |
if (is_object($debug_object)) {$debug_object->debug_log_entry(1);} |
| 796 |
|
| 797 |
if ($case_sensitivity === 'i') { |
| 798 |
$pattern = strtolower($pattern); |
| 799 |
$value = strtolower($value); |
| 800 |
} |
| 801 |
|
| 802 |
switch ($exp) { |
| 803 |
case '=': |
| 804 |
return ($value === $pattern); |
| 805 |
case '!=': |
| 806 |
return ($value !== $pattern); |
| 807 |
case '^=': |
| 808 |
return preg_match('/^' . preg_quote($pattern, '/') . '/', $value); |
| 809 |
case '$=': |
| 810 |
return preg_match('/' . preg_quote($pattern, '/') . '$/', $value); |
| 811 |
case '*=': |
| 812 |
return preg_match('/' . preg_quote($pattern, '/') . '/', $value); |
| 813 |
case '|=': |
| 814 |
/** |
| 815 |
* [att|=val] |
| 816 |
* |
| 817 |
* Represents an element with the att attribute, its value |
| 818 |
* either being exactly "val" or beginning with "val" |
| 819 |
* immediately followed by "-" (U+002D). |
| 820 |
*/ |
| 821 |
return strpos($value, $pattern) === 0; |
| 822 |
case '~=': |
| 823 |
/** |
| 824 |
* [att~=val] |
| 825 |
* |
| 826 |
* Represents an element with the att attribute whose value is a |
| 827 |
* whitespace-separated list of words, one of which is exactly |
| 828 |
* "val". If "val" contains whitespace, it will never represent |
| 829 |
* anything (since the words are separated by spaces). Also if |
| 830 |
* "val" is the empty string, it will never represent anything. |
| 831 |
*/ |
| 832 |
return in_array($pattern, explode(' ', trim($value)), true); |
| 833 |
} |
| 834 |
return false; |
| 835 |
} |
| 836 |
|
| 837 |
protected function parse_selector($selector_string) |
| 838 |
{ |
| 839 |
global $debug_object; |
| 840 |
if (is_object($debug_object)) { $debug_object->debug_log_entry(1); } |
| 841 |
|
| 842 |
/** |
| 843 |
* Pattern of CSS selectors, modified from mootools (https://mootools.net/) |
| 844 |
* |
| 845 |
* Paperg: Add the colon to the attribute, so that it properly finds |
| 846 |
* <tag attr:ibute="something" > like google does. |
| 847 |
* |
| 848 |
* Note: if you try to look at this attribute, you MUST use getAttribute |
| 849 |
* since $dom->x:y will fail the php syntax check. |
| 850 |
* |
| 851 |
* Notice the \[ starting the attribute? and the @? following? This |
| 852 |
* implies that an attribute can begin with an @ sign that is not |
| 853 |
* captured. This implies that an html attribute specifier may start |
| 854 |
* with an @ sign that is NOT captured by the expression. Farther study |
| 855 |
* is required to determine of this should be documented or removed. |
| 856 |
* |
| 857 |
* Matches selectors in this order: |
| 858 |
* |
| 859 |
* [0] - full match |
| 860 |
* |
| 861 |
* [1] - tag name |
| 862 |
* ([\w:\*-]*) |
| 863 |
* Matches the tag name consisting of zero or more words, colons, |
| 864 |
* asterisks and hyphens. |
| 865 |
* |
| 866 |
* [2] - id name |
| 867 |
* (?:\#([\w-]+)) |
| 868 |
* Optionally matches a id name, consisting of an "#" followed by |
| 869 |
* the id name (one or more words and hyphens). |
| 870 |
* |
| 871 |
* [3] - class names (including dots) |
| 872 |
* (?:\.([\w\.-]+))? |
| 873 |
* Optionally matches a list of classs, consisting of an "." |
| 874 |
* followed by the class name (one or more words and hyphens) |
| 875 |
* where multiple classes can be chained (i.e. ".foo.bar.baz") |
| 876 |
* |
| 877 |
* [4] - attributes |
| 878 |
* ((?:\[@?(?:!?[\w:-]+)(?:(?:[!*^$|~]?=)[\"']?(?:.*?)[\"']?)?(?:\s*?(?:[iIsS])?)?\])+)? |
| 879 |
* Optionally matches the attributes list |
| 880 |
* |
| 881 |
* [5] - separator |
| 882 |
* ([\/, >+~]+) |
| 883 |
* Matches the selector list separator |
| 884 |
*/ |
| 885 |
// phpcs:ignore Generic.Files.LineLength |
| 886 |
$pattern = "/([\w:\*-]*)(?:\#([\w-]+))?(?:|\.([\w\.-]+))?((?:\[@?(?:!?[\w:-]+)(?:(?:[!*^$|~]?=)[\"']?(?:.*?)[\"']?)?(?:\s*?(?:[iIsS])?)?\])+)?([\/, >+~]+)/is"; |
| 887 |
|
| 888 |
preg_match_all( |
| 889 |
$pattern, |
| 890 |
trim($selector_string) . ' ', // Add final ' ' as pseudo separator |
| 891 |
$matches, |
| 892 |
PREG_SET_ORDER |
| 893 |
); |
| 894 |
|
| 895 |
if (is_object($debug_object)) { |
| 896 |
$debug_object->debug_log(2, 'Matches Array: ', $matches); |
| 897 |
} |
| 898 |
|
| 899 |
$selectors = array(); |
| 900 |
$result = array(); |
| 901 |
|
| 902 |
foreach ($matches as $m) { |
| 903 |
$m[0] = trim($m[0]); |
| 904 |
|
| 905 |
// Skip NoOps |
| 906 |
if ($m[0] === '' || $m[0] === '/' || $m[0] === '//') { continue; } |
| 907 |
|
| 908 |
// Convert to lowercase |
| 909 |
if ($this->dom->lowercase) { |
| 910 |
$m[1] = strtolower($m[1]); |
| 911 |
} |
| 912 |
|
| 913 |
// Extract classes |
| 914 |
if ($m[3] !== '') { $m[3] = explode('.', $m[3]); } |
| 915 |
|
| 916 |
/* Extract attributes (pattern based on the pattern above!) |
| 917 |
|
| 918 |
* [0] - full match |
| 919 |
* [1] - attribute name |
| 920 |
* [2] - attribute expression |
| 921 |
* [3] - attribute value |
| 922 |
* [4] - case sensitivity |
| 923 |
* |
| 924 |
* Note: Attributes can be negated with a "!" prefix to their name |
| 925 |
*/ |
| 926 |
if($m[4] !== '') { |
| 927 |
preg_match_all( |
| 928 |
"/\[@?(!?[\w:-]+)(?:([!*^$|~]?=)[\"']?(.*?)[\"']?)?(?:\s+?([iIsS])?)?\]/is", |
| 929 |
trim($m[4]), |
| 930 |
$attributes, |
| 931 |
PREG_SET_ORDER |
| 932 |
); |
| 933 |
|
| 934 |
// Replace element by array |
| 935 |
$m[4] = array(); |
| 936 |
|
| 937 |
foreach($attributes as $att) { |
| 938 |
// Skip empty matches |
| 939 |
if(trim($att[0]) === '') { continue; } |
| 940 |
|
| 941 |
$inverted = (isset($att[1][0]) && $att[1][0] === '!'); |
| 942 |
$m[4][] = array( |
| 943 |
$inverted ? substr($att[1], 1) : $att[1], // Name |
| 944 |
(isset($att[2])) ? $att[2] : '', // Expression |
| 945 |
(isset($att[3])) ? $att[3] : '', // Value |
| 946 |
$inverted, // Inverted Flag |
| 947 |
(isset($att[4])) ? strtolower($att[4]) : '', // Case-Sensitivity |
| 948 |
); |
| 949 |
} |
| 950 |
} |
| 951 |
|
| 952 |
// Sanitize Separator |
| 953 |
if ($m[5] !== '' && trim($m[5]) === '') { // Descendant Separator |
| 954 |
$m[5] = ' '; |
| 955 |
} else { // Other Separator |
| 956 |
$m[5] = trim($m[5]); |
| 957 |
} |
| 958 |
|
| 959 |
// Clear Separator if it's a Selector List |
| 960 |
if ($is_list = ($m[5] === ',')) { $m[5] = ''; } |
| 961 |
|
| 962 |
// Remove full match before adding to results |
| 963 |
array_shift($m); |
| 964 |
$result[] = $m; |
| 965 |
|
| 966 |
if ($is_list) { // Selector List |
| 967 |
$selectors[] = $result; |
| 968 |
$result = array(); |
| 969 |
} |
| 970 |
} |
| 971 |
|
| 972 |
if (count($result) > 0) { $selectors[] = $result; } |
| 973 |
return $selectors; |
| 974 |
} |
| 975 |
|
| 976 |
function __get($name) |
| 977 |
{ |
| 978 |
if (isset($this->attr[$name])) { |
| 979 |
return $this->convert_text($this->attr[$name]); |
| 980 |
} |
| 981 |
switch ($name) { |
| 982 |
case 'outertext': return $this->outertext(); |
| 983 |
case 'innertext': return $this->innertext(); |
| 984 |
case 'plaintext': return $this->text(); |
| 985 |
case 'xmltext': return $this->xmltext(); |
| 986 |
default: return array_key_exists($name, $this->attr); |
| 987 |
} |
| 988 |
} |
| 989 |
|
| 990 |
function __set($name, $value) |
| 991 |
{ |
| 992 |
global $debug_object; |
| 993 |
if (is_object($debug_object)) { $debug_object->debug_log_entry(1); } |
| 994 |
|
| 995 |
switch ($name) { |
| 996 |
case 'outertext': return $this->_[HGFLHDOM_INFO_OUTER] = $value; |
| 997 |
case 'innertext': |
| 998 |
if (isset($this->_[HGFLHDOM_INFO_TEXT])) { |
| 999 |
return $this->_[HGFLHDOM_INFO_TEXT] = $value; |
| 1000 |
} |
| 1001 |
return $this->_[HGFLHDOM_INFO_INNER] = $value; |
| 1002 |
} |
| 1003 |
|
| 1004 |
if (!isset($this->attr[$name])) { |
| 1005 |
$this->_[HGFLHDOM_INFO_SPACE][] = array(' ', '', ''); |
| 1006 |
$this->_[HGFLHDOM_INFO_QUOTE][] = HGFLHDOM_QUOTE_DOUBLE; |
| 1007 |
} |
| 1008 |
|
| 1009 |
$this->attr[$name] = $value; |
| 1010 |
} |
| 1011 |
|
| 1012 |
function __isset($name) |
| 1013 |
{ |
| 1014 |
switch ($name) { |
| 1015 |
case 'outertext': return true; |
| 1016 |
case 'innertext': return true; |
| 1017 |
case 'plaintext': return true; |
| 1018 |
} |
| 1019 |
//no value attr: nowrap, checked selected... |
| 1020 |
return (array_key_exists($name, $this->attr)) ? true : isset($this->attr[$name]); |
| 1021 |
} |
| 1022 |
|
| 1023 |
function __unset($name) |
| 1024 |
{ |
| 1025 |
if (isset($this->attr[$name])) { unset($this->attr[$name]); } |
| 1026 |
} |
| 1027 |
|
| 1028 |
function convert_text($text) |
| 1029 |
{ |
| 1030 |
global $debug_object; |
| 1031 |
if (is_object($debug_object)) { $debug_object->debug_log_entry(1); } |
| 1032 |
|
| 1033 |
$converted_text = $text; |
| 1034 |
|
| 1035 |
$sourceCharset = ''; |
| 1036 |
$targetCharset = ''; |
| 1037 |
|
| 1038 |
if ($this->dom) { |
| 1039 |
$sourceCharset = strtoupper($this->dom->_charset); |
| 1040 |
$targetCharset = strtoupper($this->dom->_target_charset); |
| 1041 |
} |
| 1042 |
|
| 1043 |
if (is_object($debug_object)) { |
| 1044 |
$debug_object->debug_log(3, |
| 1045 |
'source charset: ' |
| 1046 |
. $sourceCharset |
| 1047 |
. ' target charaset: ' |
| 1048 |
. $targetCharset |
| 1049 |
); |
| 1050 |
} |
| 1051 |
|
| 1052 |
if (!empty($sourceCharset) |
| 1053 |
&& !empty($targetCharset) |
| 1054 |
&& (strcasecmp($sourceCharset, $targetCharset) != 0)) { |
| 1055 |
// Check if the reported encoding could have been incorrect and the text is actually already UTF-8 |
| 1056 |
if ((strcasecmp($targetCharset, 'UTF-8') == 0) |
| 1057 |
&& ($this->is_utf8($text))) { |
| 1058 |
$converted_text = $text; |
| 1059 |
} else { |
| 1060 |
if(function_exists('mb_substitute_character')) { mb_substitute_character("none"); } |
| 1061 |
$converted_text= mb_convert_encoding($text, $targetCharset, $sourceCharset); |
| 1062 |
} |
| 1063 |
} |
| 1064 |
|
| 1065 |
// Lets make sure that we don't have that silly BOM issue with any of the utf-8 text we output. |
| 1066 |
if ($targetCharset === 'UTF-8') { |
| 1067 |
if (substr($converted_text, 0, 3) === "\xef\xbb\xbf") { |
| 1068 |
$converted_text = substr($converted_text, 3); |
| 1069 |
} |
| 1070 |
|
| 1071 |
if (substr($converted_text, -3) === "\xef\xbb\xbf") { |
| 1072 |
$converted_text = substr($converted_text, 0, -3); |
| 1073 |
} |
| 1074 |
} |
| 1075 |
|
| 1076 |
return $converted_text; |
| 1077 |
} |
| 1078 |
|
| 1079 |
static function is_utf8($str) |
| 1080 |
{ |
| 1081 |
$c = 0; $b = 0; |
| 1082 |
$bits = 0; |
| 1083 |
$len = strlen($str); |
| 1084 |
for($i = 0; $i < $len; $i++) { |
| 1085 |
$c = ord($str[$i]); |
| 1086 |
if($c > 128) { |
| 1087 |
if(($c >= 254)) { return false; } |
| 1088 |
elseif($c >= 252) { $bits = 6; } |
| 1089 |
elseif($c >= 248) { $bits = 5; } |
| 1090 |
elseif($c >= 240) { $bits = 4; } |
| 1091 |
elseif($c >= 224) { $bits = 3; } |
| 1092 |
elseif($c >= 192) { $bits = 2; } |
| 1093 |
else { return false; } |
| 1094 |
if(($i + $bits) > $len) { return false; } |
| 1095 |
while($bits > 1) { |
| 1096 |
$i++; |
| 1097 |
$b = ord($str[$i]); |
| 1098 |
if($b < 128 || $b > 191) { return false; } |
| 1099 |
$bits--; |
| 1100 |
} |
| 1101 |
} |
| 1102 |
} |
| 1103 |
return true; |
| 1104 |
} |
| 1105 |
|
| 1106 |
function get_display_size() |
| 1107 |
{ |
| 1108 |
global $debug_object; |
| 1109 |
|
| 1110 |
$width = -1; |
| 1111 |
$height = -1; |
| 1112 |
|
| 1113 |
if ($this->tag !== 'img') { |
| 1114 |
return false; |
| 1115 |
} |
| 1116 |
|
| 1117 |
// See if there is aheight or width attribute in the tag itself. |
| 1118 |
if (isset($this->attr['width'])) { |
| 1119 |
$width = $this->attr['width']; |
| 1120 |
} |
| 1121 |
|
| 1122 |
if (isset($this->attr['height'])) { |
| 1123 |
$height = $this->attr['height']; |
| 1124 |
} |
| 1125 |
|
| 1126 |
// Now look for an inline style. |
| 1127 |
if (isset($this->attr['style'])) { |
| 1128 |
// Thanks to user gnarf from stackoverflow for this regular expression. |
| 1129 |
$attributes = array(); |
| 1130 |
|
| 1131 |
preg_match_all( |
| 1132 |
'/([\w-]+)\s*:\s*([^;]+)\s*;?/', |
| 1133 |
$this->attr['style'], |
| 1134 |
$matches, |
| 1135 |
PREG_SET_ORDER |
| 1136 |
); |
| 1137 |
|
| 1138 |
foreach ($matches as $match) { |
| 1139 |
$attributes[$match[1]] = $match[2]; |
| 1140 |
} |
| 1141 |
|
| 1142 |
// If there is a width in the style attributes: |
| 1143 |
if (isset($attributes['width']) && $width == -1) { |
| 1144 |
// check that the last two characters are px (pixels) |
| 1145 |
if (strtolower(substr($attributes['width'], -2)) === 'px') { |
| 1146 |
$proposed_width = substr($attributes['width'], 0, -2); |
| 1147 |
// Now make sure that it's an integer and not something stupid. |
| 1148 |
if (filter_var($proposed_width, FILTER_VALIDATE_INT)) { |
| 1149 |
$width = $proposed_width; |
| 1150 |
} |
| 1151 |
} |
| 1152 |
} |
| 1153 |
|
| 1154 |
// If there is a width in the style attributes: |
| 1155 |
if (isset($attributes['height']) && $height == -1) { |
| 1156 |
// check that the last two characters are px (pixels) |
| 1157 |
if (strtolower(substr($attributes['height'], -2)) == 'px') { |
| 1158 |
$proposed_height = substr($attributes['height'], 0, -2); |
| 1159 |
// Now make sure that it's an integer and not something stupid. |
| 1160 |
if (filter_var($proposed_height, FILTER_VALIDATE_INT)) { |
| 1161 |
$height = $proposed_height; |
| 1162 |
} |
| 1163 |
} |
| 1164 |
} |
| 1165 |
|
| 1166 |
} |
| 1167 |
|
| 1168 |
// Future enhancement: |
| 1169 |
// Look in the tag to see if there is a class or id specified that has |
| 1170 |
// a height or width attribute to it. |
| 1171 |
|
| 1172 |
// Far future enhancement |
| 1173 |
// Look at all the parent tags of this image to see if they specify a |
| 1174 |
// class or id that has an img selector that specifies a height or width |
| 1175 |
// Note that in this case, the class or id will have the img subselector |
| 1176 |
// for it to apply to the image. |
| 1177 |
|
| 1178 |
// ridiculously far future development |
| 1179 |
// If the class or id is specified in a SEPARATE css file thats not on |
| 1180 |
// the page, go get it and do what we were just doing for the ones on |
| 1181 |
// the page. |
| 1182 |
|
| 1183 |
$result = array( |
| 1184 |
'height' => $height, |
| 1185 |
'width' => $width |
| 1186 |
); |
| 1187 |
|
| 1188 |
return $result; |
| 1189 |
} |
| 1190 |
|
| 1191 |
function save($filepath = '') |
| 1192 |
{ |
| 1193 |
$ret = $this->outertext(); |
| 1194 |
|
| 1195 |
if ($filepath !== '') { |
| 1196 |
file_put_contents($filepath, $ret, LOCK_EX); |
| 1197 |
} |
| 1198 |
|
| 1199 |
return $ret; |
| 1200 |
} |
| 1201 |
|
| 1202 |
function addClass($class) |
| 1203 |
{ |
| 1204 |
if (is_string($class)) { |
| 1205 |
$class = explode(' ', $class); |
| 1206 |
} |
| 1207 |
|
| 1208 |
if (is_array($class)) { |
| 1209 |
foreach($class as $c) { |
| 1210 |
if (isset($this->class)) { |
| 1211 |
if ($this->hasClass($c)) { |
| 1212 |
continue; |
| 1213 |
} else { |
| 1214 |
$this->class .= ' ' . $c; |
| 1215 |
} |
| 1216 |
} else { |
| 1217 |
$this->class = $c; |
| 1218 |
} |
| 1219 |
} |
| 1220 |
} else { |
| 1221 |
if (is_object($debug_object)) { |
| 1222 |
$debug_object->debug_log(2, 'Invalid type: ', gettype($class)); |
| 1223 |
} |
| 1224 |
} |
| 1225 |
} |
| 1226 |
|
| 1227 |
function hasClass($class) |
| 1228 |
{ |
| 1229 |
if (is_string($class)) { |
| 1230 |
if (isset($this->class)) { |
| 1231 |
return in_array($class, explode(' ', $this->class), true); |
| 1232 |
} |
| 1233 |
} else { |
| 1234 |
if (is_object($debug_object)) { |
| 1235 |
$debug_object->debug_log(2, 'Invalid type: ', gettype($class)); |
| 1236 |
} |
| 1237 |
} |
| 1238 |
|
| 1239 |
return false; |
| 1240 |
} |
| 1241 |
|
| 1242 |
function removeClass($class = null) |
| 1243 |
{ |
| 1244 |
if (!isset($this->class)) { |
| 1245 |
return; |
| 1246 |
} |
| 1247 |
|
| 1248 |
if (is_null($class)) { |
| 1249 |
$this->removeAttribute('class'); |
| 1250 |
return; |
| 1251 |
} |
| 1252 |
|
| 1253 |
if (is_string($class)) { |
| 1254 |
$class = explode(' ', $class); |
| 1255 |
} |
| 1256 |
|
| 1257 |
if (is_array($class)) { |
| 1258 |
$class = array_diff(explode(' ', $this->class), $class); |
| 1259 |
if (empty($class)) { |
| 1260 |
$this->removeAttribute('class'); |
| 1261 |
} else { |
| 1262 |
$this->class = implode(' ', $class); |
| 1263 |
} |
| 1264 |
} |
| 1265 |
} |
| 1266 |
|
| 1267 |
function getAllAttributes() |
| 1268 |
{ |
| 1269 |
return $this->attr; |
| 1270 |
} |
| 1271 |
|
| 1272 |
function getAttribute($name) |
| 1273 |
{ |
| 1274 |
return $this->__get($name); |
| 1275 |
} |
| 1276 |
|
| 1277 |
function setAttribute($name, $value) |
| 1278 |
{ |
| 1279 |
$this->__set($name, $value); |
| 1280 |
} |
| 1281 |
|
| 1282 |
function hasAttribute($name) |
| 1283 |
{ |
| 1284 |
return $this->__isset($name); |
| 1285 |
} |
| 1286 |
|
| 1287 |
function removeAttribute($name) |
| 1288 |
{ |
| 1289 |
$this->__set($name, null); |
| 1290 |
} |
| 1291 |
|
| 1292 |
function remove() |
| 1293 |
{ |
| 1294 |
if ($this->parent) { |
| 1295 |
$this->parent->removeChild($this); |
| 1296 |
} |
| 1297 |
} |
| 1298 |
|
| 1299 |
function removeChild($node) |
| 1300 |
{ |
| 1301 |
$nidx = array_search($node, $this->nodes, true); |
| 1302 |
$cidx = array_search($node, $this->children, true); |
| 1303 |
$didx = array_search($node, $this->dom->nodes, true); |
| 1304 |
|
| 1305 |
if ($nidx !== false && $cidx !== false && $didx !== false) { |
| 1306 |
|
| 1307 |
foreach($node->children as $child) { |
| 1308 |
$node->removeChild($child); |
| 1309 |
} |
| 1310 |
|
| 1311 |
foreach($node->nodes as $entity) { |
| 1312 |
$enidx = array_search($entity, $node->nodes, true); |
| 1313 |
$edidx = array_search($entity, $node->dom->nodes, true); |
| 1314 |
|
| 1315 |
if ($enidx !== false && $edidx !== false) { |
| 1316 |
unset($node->nodes[$enidx]); |
| 1317 |
unset($node->dom->nodes[$edidx]); |
| 1318 |
} |
| 1319 |
} |
| 1320 |
|
| 1321 |
unset($this->nodes[$nidx]); |
| 1322 |
unset($this->children[$cidx]); |
| 1323 |
unset($this->dom->nodes[$didx]); |
| 1324 |
|
| 1325 |
$node->clear(); |
| 1326 |
|
| 1327 |
} |
| 1328 |
} |
| 1329 |
|
| 1330 |
function getElementById($id) |
| 1331 |
{ |
| 1332 |
return $this->find("#$id", 0); |
| 1333 |
} |
| 1334 |
|
| 1335 |
function getElementsById($id, $idx = null) |
| 1336 |
{ |
| 1337 |
return $this->find("#$id", $idx); |
| 1338 |
} |
| 1339 |
|
| 1340 |
function getElementByTagName($name) |
| 1341 |
{ |
| 1342 |
return $this->find($name, 0); |
| 1343 |
} |
| 1344 |
|
| 1345 |
function getElementsByTagName($name, $idx = null) |
| 1346 |
{ |
| 1347 |
return $this->find($name, $idx); |
| 1348 |
} |
| 1349 |
|
| 1350 |
function parentNode() |
| 1351 |
{ |
| 1352 |
return $this->parent(); |
| 1353 |
} |
| 1354 |
|
| 1355 |
function childNodes($idx = -1) |
| 1356 |
{ |
| 1357 |
return $this->children($idx); |
| 1358 |
} |
| 1359 |
|
| 1360 |
function firstChild() |
| 1361 |
{ |
| 1362 |
return $this->first_child(); |
| 1363 |
} |
| 1364 |
|
| 1365 |
function lastChild() |
| 1366 |
{ |
| 1367 |
return $this->last_child(); |
| 1368 |
} |
| 1369 |
|
| 1370 |
function nextSibling() |
| 1371 |
{ |
| 1372 |
return $this->next_sibling(); |
| 1373 |
} |
| 1374 |
|
| 1375 |
function previousSibling() |
| 1376 |
{ |
| 1377 |
return $this->prev_sibling(); |
| 1378 |
} |
| 1379 |
|
| 1380 |
function hasChildNodes() |
| 1381 |
{ |
| 1382 |
return $this->has_child(); |
| 1383 |
} |
| 1384 |
|
| 1385 |
function nodeName() |
| 1386 |
{ |
| 1387 |
return $this->tag; |
| 1388 |
} |
| 1389 |
|
| 1390 |
function appendChild($node) |
| 1391 |
{ |
| 1392 |
$node->parent($this); |
| 1393 |
return $node; |
| 1394 |
} |
| 1395 |
|
| 1396 |
} |
| 1397 |
|
| 1398 |
class hgfl_simple_html_dom |
| 1399 |
{ |
| 1400 |
public $root = null; |
| 1401 |
public $nodes = array(); |
| 1402 |
public $callback = null; |
| 1403 |
public $lowercase = false; |
| 1404 |
public $original_size; |
| 1405 |
public $size; |
| 1406 |
|
| 1407 |
protected $pos; |
| 1408 |
protected $doc; |
| 1409 |
protected $char; |
| 1410 |
|
| 1411 |
protected $cursor; |
| 1412 |
protected $parent; |
| 1413 |
protected $noise = array(); |
| 1414 |
protected $token_blank = " \t\r\n"; |
| 1415 |
protected $token_equal = ' =/>'; |
| 1416 |
protected $token_slash = " />\r\n\t"; |
| 1417 |
protected $token_attr = ' >'; |
| 1418 |
|
| 1419 |
public $_charset = ''; |
| 1420 |
public $_target_charset = ''; |
| 1421 |
|
| 1422 |
protected $HGFLDEFAULT_br_text = ''; |
| 1423 |
|
| 1424 |
public $HGFLDEFAULT_span_text = ''; |
| 1425 |
|
| 1426 |
protected $self_closing_tags = array( |
| 1427 |
'area' => 1, |
| 1428 |
'base' => 1, |
| 1429 |
'br' => 1, |
| 1430 |
'col' => 1, |
| 1431 |
'embed' => 1, |
| 1432 |
'hr' => 1, |
| 1433 |
'img' => 1, |
| 1434 |
'input' => 1, |
| 1435 |
'link' => 1, |
| 1436 |
'meta' => 1, |
| 1437 |
'param' => 1, |
| 1438 |
'source' => 1, |
| 1439 |
'track' => 1, |
| 1440 |
'wbr' => 1 |
| 1441 |
); |
| 1442 |
protected $block_tags = array( |
| 1443 |
'body' => 1, |
| 1444 |
'div' => 1, |
| 1445 |
'form' => 1, |
| 1446 |
'root' => 1, |
| 1447 |
'span' => 1, |
| 1448 |
'table' => 1 |
| 1449 |
); |
| 1450 |
protected $optional_closing_tags = array( |
| 1451 |
// Not optional, see |
| 1452 |
// https://www.w3.org/TR/html/textlevel-semantics.html#the-b-element |
| 1453 |
'b' => array('b' => 1), |
| 1454 |
'dd' => array('dd' => 1, 'dt' => 1), |
| 1455 |
// Not optional, see |
| 1456 |
// https://www.w3.org/TR/html/grouping-content.html#the-dl-element |
| 1457 |
'dl' => array('dd' => 1, 'dt' => 1), |
| 1458 |
'dt' => array('dd' => 1, 'dt' => 1), |
| 1459 |
'li' => array('li' => 1), |
| 1460 |
'optgroup' => array('optgroup' => 1, 'option' => 1), |
| 1461 |
'option' => array('optgroup' => 1, 'option' => 1), |
| 1462 |
'p' => array('p' => 1), |
| 1463 |
'rp' => array('rp' => 1, 'rt' => 1), |
| 1464 |
'rt' => array('rp' => 1, 'rt' => 1), |
| 1465 |
'td' => array('td' => 1, 'th' => 1), |
| 1466 |
'th' => array('td' => 1, 'th' => 1), |
| 1467 |
'tr' => array('td' => 1, 'th' => 1, 'tr' => 1), |
| 1468 |
); |
| 1469 |
|
| 1470 |
function __construct( |
| 1471 |
$str = null, |
| 1472 |
$lowercase = true, |
| 1473 |
$forceTagsClosed = true, |
| 1474 |
$target_charset = HGFLDEFAULT_TARGET_CHARSET, |
| 1475 |
$stripRN = true, |
| 1476 |
$defaultBRText = HGFLDEFAULT_BR_TEXT, |
| 1477 |
$defaultSpanText = HGFLDEFAULT_SPAN_TEXT, |
| 1478 |
$options = 0) |
| 1479 |
{ |
| 1480 |
if ($str) { |
| 1481 |
if (preg_match('/^http:\/\//i', $str) || is_file($str)) { |
| 1482 |
$this->load_file($str); |
| 1483 |
} else { |
| 1484 |
$this->load( |
| 1485 |
$str, |
| 1486 |
$lowercase, |
| 1487 |
$stripRN, |
| 1488 |
$defaultBRText, |
| 1489 |
$defaultSpanText, |
| 1490 |
$options |
| 1491 |
); |
| 1492 |
} |
| 1493 |
} |
| 1494 |
// Forcing tags to be closed implies that we don't trust the html, but |
| 1495 |
// it can lead to parsing errors if we SHOULD trust the html. |
| 1496 |
if (!$forceTagsClosed) { |
| 1497 |
$this->optional_closing_array = array(); |
| 1498 |
} |
| 1499 |
|
| 1500 |
$this->_target_charset = $target_charset; |
| 1501 |
} |
| 1502 |
|
| 1503 |
function __destruct() |
| 1504 |
{ |
| 1505 |
$this->clear(); |
| 1506 |
} |
| 1507 |
|
| 1508 |
function load( |
| 1509 |
$str, |
| 1510 |
$lowercase = true, |
| 1511 |
$stripRN = true, |
| 1512 |
$defaultBRText = HGFLDEFAULT_BR_TEXT, |
| 1513 |
$defaultSpanText = HGFLDEFAULT_SPAN_TEXT, |
| 1514 |
$options = 0) |
| 1515 |
{ |
| 1516 |
global $debug_object; |
| 1517 |
|
| 1518 |
// prepare |
| 1519 |
$this->prepare($str, $lowercase, $defaultBRText, $defaultSpanText); |
| 1520 |
|
| 1521 |
// Per sourceforge http://sourceforge.net/tracker/?func=detail&aid=2949097&group_id=218559&atid=1044037 |
| 1522 |
// Script tags removal now preceeds style tag removal. |
| 1523 |
// strip out <script> tags |
| 1524 |
$this->remove_noise("'<\s*script[^>]*[^/]>(.*?)<\s*/\s*script\s*>'is"); |
| 1525 |
$this->remove_noise("'<\s*script\s*>(.*?)<\s*/\s*script\s*>'is"); |
| 1526 |
|
| 1527 |
// strip out the \r \n's if we are told to. |
| 1528 |
if ($stripRN) { |
| 1529 |
$this->doc = str_replace("\r", ' ', $this->doc); |
| 1530 |
$this->doc = str_replace("\n", ' ', $this->doc); |
| 1531 |
|
| 1532 |
// set the length of content since we have changed it. |
| 1533 |
$this->size = strlen($this->doc); |
| 1534 |
} |
| 1535 |
|
| 1536 |
// strip out cdata |
| 1537 |
$this->remove_noise("'<!\[CDATA\[(.*?)\]\]>'is", true); |
| 1538 |
// strip out comments |
| 1539 |
$this->remove_noise("'<!--(.*?)-->'is"); |
| 1540 |
// strip out <style> tags |
| 1541 |
$this->remove_noise("'<\s*style[^>]*[^/]>(.*?)<\s*/\s*style\s*>'is"); |
| 1542 |
$this->remove_noise("'<\s*style\s*>(.*?)<\s*/\s*style\s*>'is"); |
| 1543 |
// strip out preformatted tags |
| 1544 |
$this->remove_noise("'<\s*(?:code)[^>]*>(.*?)<\s*/\s*(?:code)\s*>'is"); |
| 1545 |
// strip out server side scripts |
| 1546 |
$this->remove_noise("'(<\?)(.*?)(\?>)'s", true); |
| 1547 |
|
| 1548 |
if($options & HGFLHDOM_SMARTY_AS_TEXT) { // Strip Smarty scripts |
| 1549 |
$this->remove_noise("'(\{\w)(.*?)(\})'s", true); |
| 1550 |
} |
| 1551 |
|
| 1552 |
// parsing |
| 1553 |
$this->parse(); |
| 1554 |
// end |
| 1555 |
$this->root->_[HGFLHDOM_INFO_END] = $this->cursor; |
| 1556 |
$this->parse_charset(); |
| 1557 |
|
| 1558 |
// make load function chainable |
| 1559 |
return $this; |
| 1560 |
} |
| 1561 |
|
| 1562 |
function load_file() |
| 1563 |
{ |
| 1564 |
$args = func_get_args(); |
| 1565 |
|
| 1566 |
if(($doc = call_user_func_array('file_get_contents', $args)) !== false) { |
| 1567 |
$this->load($doc, true); |
| 1568 |
} else { |
| 1569 |
return false; |
| 1570 |
} |
| 1571 |
} |
| 1572 |
|
| 1573 |
function set_callback($function_name) |
| 1574 |
{ |
| 1575 |
$this->callback = $function_name; |
| 1576 |
} |
| 1577 |
|
| 1578 |
function remove_callback() |
| 1579 |
{ |
| 1580 |
$this->callback = null; |
| 1581 |
} |
| 1582 |
|
| 1583 |
function save($filepath = '') |
| 1584 |
{ |
| 1585 |
$ret = $this->root->innertext(); |
| 1586 |
if ($filepath !== '') { file_put_contents($filepath, $ret, LOCK_EX); } |
| 1587 |
return $ret; |
| 1588 |
} |
| 1589 |
|
| 1590 |
function find($selector, $idx = null, $lowercase = false) |
| 1591 |
{ |
| 1592 |
return $this->root->find($selector, $idx, $lowercase); |
| 1593 |
} |
| 1594 |
|
| 1595 |
function clear() |
| 1596 |
{ |
| 1597 |
if (isset($this->nodes)) { |
| 1598 |
foreach ($this->nodes as $n) { |
| 1599 |
$n->clear(); |
| 1600 |
$n = null; |
| 1601 |
} |
| 1602 |
} |
| 1603 |
|
| 1604 |
// This add next line is documented in the sourceforge repository. |
| 1605 |
// 2977248 as a fix for ongoing memory leaks that occur even with the |
| 1606 |
// use of clear. |
| 1607 |
if (isset($this->children)) { |
| 1608 |
foreach ($this->children as $n) { |
| 1609 |
$n->clear(); |
| 1610 |
$n = null; |
| 1611 |
} |
| 1612 |
} |
| 1613 |
|
| 1614 |
if (isset($this->parent)) { |
| 1615 |
$this->parent->clear(); |
| 1616 |
unset($this->parent); |
| 1617 |
} |
| 1618 |
|
| 1619 |
if (isset($this->root)) { |
| 1620 |
$this->root->clear(); |
| 1621 |
unset($this->root); |
| 1622 |
} |
| 1623 |
|
| 1624 |
unset($this->doc); |
| 1625 |
unset($this->noise); |
| 1626 |
} |
| 1627 |
|
| 1628 |
function dump($show_attr = true) |
| 1629 |
{ |
| 1630 |
$this->root->dump($show_attr); |
| 1631 |
} |
| 1632 |
|
| 1633 |
protected function prepare( |
| 1634 |
$str, $lowercase = true, |
| 1635 |
$defaultBRText = HGFLDEFAULT_BR_TEXT, |
| 1636 |
$defaultSpanText = HGFLDEFAULT_SPAN_TEXT) |
| 1637 |
{ |
| 1638 |
$this->clear(); |
| 1639 |
|
| 1640 |
$this->doc = trim($str); |
| 1641 |
$this->size = strlen($this->doc); |
| 1642 |
$this->original_size = $this->size; // original size of the html |
| 1643 |
$this->pos = 0; |
| 1644 |
$this->cursor = 1; |
| 1645 |
$this->noise = array(); |
| 1646 |
$this->nodes = array(); |
| 1647 |
$this->lowercase = $lowercase; |
| 1648 |
$this->HGFLDEFAULT_br_text = $defaultBRText; |
| 1649 |
$this->HGFLDEFAULT_span_text = $defaultSpanText; |
| 1650 |
$this->root = new hgfl_simple_html_dom_node($this); |
| 1651 |
$this->root->tag = 'root'; |
| 1652 |
$this->root->_[HGFLHDOM_INFO_BEGIN] = -1; |
| 1653 |
$this->root->nodetype = HGFLHDOM_TYPE_ROOT; |
| 1654 |
$this->parent = $this->root; |
| 1655 |
if ($this->size > 0) { $this->char = $this->doc[0]; } |
| 1656 |
} |
| 1657 |
|
| 1658 |
protected function parse() |
| 1659 |
{ |
| 1660 |
while (true) { |
| 1661 |
// Read next tag if there is no text between current position and the |
| 1662 |
// next opening tag. |
| 1663 |
if (($s = $this->copy_until_char('<')) === '') { |
| 1664 |
if($this->read_tag()) { |
| 1665 |
continue; |
| 1666 |
} else { |
| 1667 |
return true; |
| 1668 |
} |
| 1669 |
} |
| 1670 |
|
| 1671 |
// Add a text node for text between tags |
| 1672 |
$node = new hgfl_simple_html_dom_node($this); |
| 1673 |
++$this->cursor; |
| 1674 |
$node->_[HGFLHDOM_INFO_TEXT] = $s; |
| 1675 |
$this->link_nodes($node, false); |
| 1676 |
} |
| 1677 |
} |
| 1678 |
|
| 1679 |
protected function parse_charset() |
| 1680 |
{ |
| 1681 |
global $debug_object; |
| 1682 |
|
| 1683 |
$charset = null; |
| 1684 |
|
| 1685 |
if (function_exists('get_last_retrieve_url_contents_content_type')) { |
| 1686 |
$contentTypeHeader = get_last_retrieve_url_contents_content_type(); |
| 1687 |
$success = preg_match('/charset=(.+)/', $contentTypeHeader, $matches); |
| 1688 |
if ($success) { |
| 1689 |
$charset = $matches[1]; |
| 1690 |
if (is_object($debug_object)) { |
| 1691 |
$debug_object->debug_log(2, |
| 1692 |
'header content-type found charset of: ' |
| 1693 |
. $charset |
| 1694 |
); |
| 1695 |
} |
| 1696 |
} |
| 1697 |
} |
| 1698 |
|
| 1699 |
if (empty($charset)) { |
| 1700 |
// https://www.w3.org/TR/html/document-metadata.html#statedef-http-equiv-content-type |
| 1701 |
$el = $this->root->find('meta[http-equiv=Content-Type]', 0, true); |
| 1702 |
|
| 1703 |
if (!empty($el)) { |
| 1704 |
$fullvalue = $el->content; |
| 1705 |
if (is_object($debug_object)) { |
| 1706 |
$debug_object->debug_log(2, |
| 1707 |
'meta content-type tag found' |
| 1708 |
. $fullvalue |
| 1709 |
); |
| 1710 |
} |
| 1711 |
|
| 1712 |
if (!empty($fullvalue)) { |
| 1713 |
$success = preg_match( |
| 1714 |
'/charset=(.+)/i', |
| 1715 |
$fullvalue, |
| 1716 |
$matches |
| 1717 |
); |
| 1718 |
|
| 1719 |
if ($success) { |
| 1720 |
$charset = $matches[1]; |
| 1721 |
} else { |
| 1722 |
// If there is a meta tag, and they don't specify the |
| 1723 |
// character set, research says that it's typically |
| 1724 |
// ISO-8859-1 |
| 1725 |
if (is_object($debug_object)) { |
| 1726 |
$debug_object->debug_log(2, |
| 1727 |
'meta content-type tag couldn\'t be parsed. using iso-8859 default.' |
| 1728 |
); |
| 1729 |
} |
| 1730 |
|
| 1731 |
$charset = 'ISO-8859-1'; |
| 1732 |
} |
| 1733 |
} |
| 1734 |
} |
| 1735 |
} |
| 1736 |
|
| 1737 |
if (empty($charset)) { |
| 1738 |
// https://www.w3.org/TR/html/document-metadata.html#character-encoding-declaration |
| 1739 |
if ($meta = $this->root->find('meta[charset]', 0)) { |
| 1740 |
$charset = $meta->charset; |
| 1741 |
if (is_object($debug_object)) { |
| 1742 |
$debug_object->debug_log(2, 'meta charset: ' . $charset); |
| 1743 |
} |
| 1744 |
} |
| 1745 |
} |
| 1746 |
|
| 1747 |
if (empty($charset)) { |
| 1748 |
// Try to guess the charset based on the content |
| 1749 |
// Requires Multibyte String (mbstring) support (optional) |
| 1750 |
if (function_exists('mb_detect_encoding')) { |
| 1751 |
/** |
| 1752 |
* mb_detect_encoding() is not intended to distinguish between |
| 1753 |
* charsets, especially single-byte charsets. Its primary |
| 1754 |
* purpose is to detect which multibyte encoding is in use, |
| 1755 |
* i.e. UTF-8, UTF-16, shift-JIS, etc. |
| 1756 |
* |
| 1757 |
* -- https://bugs.php.net/bug.php?id=38138 |
| 1758 |
* |
| 1759 |
* Adding both CP1251/ISO-8859-5 and CP1252/ISO-8859-1 will |
| 1760 |
* always result in CP1251/ISO-8859-5 and vice versa. |
| 1761 |
* |
| 1762 |
* Thus, only detect if it's either UTF-8 or CP1252/ISO-8859-1 |
| 1763 |
* to stay compatible. |
| 1764 |
*/ |
| 1765 |
$encoding = mb_detect_encoding( |
| 1766 |
$this->doc, |
| 1767 |
array( 'UTF-8', 'CP1252', 'ISO-8859-1' ) |
| 1768 |
); |
| 1769 |
|
| 1770 |
if ($encoding === 'CP1252' || $encoding === 'ISO-8859-1') { |
| 1771 |
// Due to a limitation of mb_detect_encoding |
| 1772 |
// 'CP1251'/'ISO-8859-5' will be detected as |
| 1773 |
// 'CP1252'/'ISO-8859-1'. This will cause iconv to fail, in |
| 1774 |
// which case we can simply assume it is the other charset. |
| 1775 |
if (!@iconv('CP1252', 'UTF-8', $this->doc)) { |
| 1776 |
$encoding = 'CP1251'; |
| 1777 |
} |
| 1778 |
} |
| 1779 |
|
| 1780 |
if ($encoding !== false) { |
| 1781 |
$charset = $encoding; |
| 1782 |
if (is_object($debug_object)) { |
| 1783 |
$debug_object->debug_log(2, 'mb_detect: ' . $charset); |
| 1784 |
} |
| 1785 |
} |
| 1786 |
} |
| 1787 |
} |
| 1788 |
|
| 1789 |
if (empty($charset)) { |
| 1790 |
// Assume it's UTF-8 as it is the most likely charset to be used |
| 1791 |
$charset = 'UTF-8'; |
| 1792 |
if (is_object($debug_object)) { |
| 1793 |
$debug_object->debug_log(2, 'No match found, assume ' . $charset); |
| 1794 |
} |
| 1795 |
} |
| 1796 |
|
| 1797 |
// Since CP1252 is a superset, if we get one of it's subsets, we want |
| 1798 |
// it instead. |
| 1799 |
if ((strtolower($charset) == 'iso-8859-1') |
| 1800 |
|| (strtolower($charset) == 'latin1') |
| 1801 |
|| (strtolower($charset) == 'latin-1')) { |
| 1802 |
$charset = 'CP1252'; |
| 1803 |
if (is_object($debug_object)) { |
| 1804 |
$debug_object->debug_log(2, |
| 1805 |
'replacing ' . $charset . ' with CP1252 as its a superset' |
| 1806 |
); |
| 1807 |
} |
| 1808 |
} |
| 1809 |
|
| 1810 |
if (is_object($debug_object)) { |
| 1811 |
$debug_object->debug_log(1, 'EXIT - ' . $charset); |
| 1812 |
} |
| 1813 |
|
| 1814 |
return $this->_charset = $charset; |
| 1815 |
} |
| 1816 |
|
| 1817 |
protected function read_tag() |
| 1818 |
{ |
| 1819 |
// Set end position if no further tags found |
| 1820 |
if ($this->char !== '<') { |
| 1821 |
$this->root->_[HGFLHDOM_INFO_END] = $this->cursor; |
| 1822 |
return false; |
| 1823 |
} |
| 1824 |
|
| 1825 |
$begin_tag_pos = $this->pos; |
| 1826 |
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 1827 |
|
| 1828 |
// end tag |
| 1829 |
if ($this->char === '/') { |
| 1830 |
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 1831 |
|
| 1832 |
// Skip whitespace in end tags (i.e. in "</ html>") |
| 1833 |
$this->skip($this->token_blank); |
| 1834 |
$tag = $this->copy_until_char('>'); |
| 1835 |
|
| 1836 |
// Skip attributes in end tags |
| 1837 |
if (($pos = strpos($tag, ' ')) !== false) { |
| 1838 |
$tag = substr($tag, 0, $pos); |
| 1839 |
} |
| 1840 |
|
| 1841 |
$parent_lower = strtolower($this->parent->tag); |
| 1842 |
$tag_lower = strtolower($tag); |
| 1843 |
|
| 1844 |
// The end tag is supposed to close the parent tag. Handle situations |
| 1845 |
// when it doesn't |
| 1846 |
if ($parent_lower !== $tag_lower) { |
| 1847 |
// Parent tag does not have to be closed necessarily (optional closing tag) |
| 1848 |
// Current tag is a block tag, so it may close an ancestor |
| 1849 |
if (isset($this->optional_closing_tags[$parent_lower]) |
| 1850 |
&& isset($this->block_tags[$tag_lower])) { |
| 1851 |
|
| 1852 |
$this->parent->_[HGFLHDOM_INFO_END] = 0; |
| 1853 |
$org_parent = $this->parent; |
| 1854 |
|
| 1855 |
// Traverse ancestors to find a matching opening tag |
| 1856 |
// Stop at root node |
| 1857 |
while (($this->parent->parent) |
| 1858 |
&& strtolower($this->parent->tag) !== $tag_lower |
| 1859 |
){ |
| 1860 |
$this->parent = $this->parent->parent; |
| 1861 |
} |
| 1862 |
|
| 1863 |
// If we don't have a match add current tag as text node |
| 1864 |
if (strtolower($this->parent->tag) !== $tag_lower) { |
| 1865 |
$this->parent = $org_parent; // restore origonal parent |
| 1866 |
|
| 1867 |
if ($this->parent->parent) { |
| 1868 |
$this->parent = $this->parent->parent; |
| 1869 |
} |
| 1870 |
|
| 1871 |
$this->parent->_[HGFLHDOM_INFO_END] = $this->cursor; |
| 1872 |
return $this->as_text_node($tag); |
| 1873 |
} |
| 1874 |
} elseif (($this->parent->parent) |
| 1875 |
&& isset($this->block_tags[$tag_lower]) |
| 1876 |
) { |
| 1877 |
// Grandparent exists and current tag is a block tag, so our |
| 1878 |
// parent doesn't have an end tag |
| 1879 |
$this->parent->_[HGFLHDOM_INFO_END] = 0; // No end tag |
| 1880 |
$org_parent = $this->parent; |
| 1881 |
|
| 1882 |
// Traverse ancestors to find a matching opening tag |
| 1883 |
// Stop at root node |
| 1884 |
while (($this->parent->parent) |
| 1885 |
&& strtolower($this->parent->tag) !== $tag_lower |
| 1886 |
) { |
| 1887 |
$this->parent = $this->parent->parent; |
| 1888 |
} |
| 1889 |
|
| 1890 |
// If we don't have a match add current tag as text node |
| 1891 |
if (strtolower($this->parent->tag) !== $tag_lower) { |
| 1892 |
$this->parent = $org_parent; // restore origonal parent |
| 1893 |
$this->parent->_[HGFLHDOM_INFO_END] = $this->cursor; |
| 1894 |
return $this->as_text_node($tag); |
| 1895 |
} |
| 1896 |
} elseif (($this->parent->parent) |
| 1897 |
&& strtolower($this->parent->parent->tag) === $tag_lower |
| 1898 |
) { // Grandparent exists and current tag closes it |
| 1899 |
$this->parent->_[HGFLHDOM_INFO_END] = 0; |
| 1900 |
$this->parent = $this->parent->parent; |
| 1901 |
} else { // Random tag, add as text node |
| 1902 |
return $this->as_text_node($tag); |
| 1903 |
} |
| 1904 |
} |
| 1905 |
|
| 1906 |
// Set end position of parent tag to current cursor position |
| 1907 |
$this->parent->_[HGFLHDOM_INFO_END] = $this->cursor; |
| 1908 |
|
| 1909 |
if ($this->parent->parent) { |
| 1910 |
$this->parent = $this->parent->parent; |
| 1911 |
} |
| 1912 |
|
| 1913 |
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 1914 |
return true; |
| 1915 |
} |
| 1916 |
|
| 1917 |
// start tag |
| 1918 |
$node = new hgfl_simple_html_dom_node($this); |
| 1919 |
$node->_[HGFLHDOM_INFO_BEGIN] = $this->cursor; |
| 1920 |
++$this->cursor; |
| 1921 |
$tag = $this->copy_until($this->token_slash); // Get tag name |
| 1922 |
$node->tag_start = $begin_tag_pos; |
| 1923 |
|
| 1924 |
// doctype, cdata & comments... |
| 1925 |
// <!DOCTYPE html> |
| 1926 |
// <![CDATA[ ... ]]> |
| 1927 |
// <!-- Comment --> |
| 1928 |
if (isset($tag[0]) && $tag[0] === '!') { |
| 1929 |
$node->_[HGFLHDOM_INFO_TEXT] = '<' . $tag . $this->copy_until_char('>'); |
| 1930 |
|
| 1931 |
if (isset($tag[2]) && $tag[1] === '-' && $tag[2] === '-') { // Comment ("<!--") |
| 1932 |
$node->nodetype = HGFLHDOM_TYPE_COMMENT; |
| 1933 |
$node->tag = 'comment'; |
| 1934 |
} else { // Could be doctype or CDATA but we don't care |
| 1935 |
$node->nodetype = HGFLHDOM_TYPE_UNKNOWN; |
| 1936 |
$node->tag = 'unknown'; |
| 1937 |
} |
| 1938 |
|
| 1939 |
if ($this->char === '>') { $node->_[HGFLHDOM_INFO_TEXT] .= '>'; } |
| 1940 |
|
| 1941 |
$this->link_nodes($node, true); |
| 1942 |
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 1943 |
return true; |
| 1944 |
} |
| 1945 |
|
| 1946 |
// The start tag cannot contain another start tag, if so add as text |
| 1947 |
// i.e. "<<html>" |
| 1948 |
if ($pos = strpos($tag, '<') !== false) { |
| 1949 |
$tag = '<' . substr($tag, 0, -1); |
| 1950 |
$node->_[HGFLHDOM_INFO_TEXT] = $tag; |
| 1951 |
$this->link_nodes($node, false); |
| 1952 |
$this->char = $this->doc[--$this->pos]; // prev |
| 1953 |
return true; |
| 1954 |
} |
| 1955 |
|
| 1956 |
// Handle invalid tag names (i.e. "<html#doc>") |
| 1957 |
if (!preg_match('/^\w[\w:-]*$/', $tag)) { |
| 1958 |
$node->_[HGFLHDOM_INFO_TEXT] = '<' . $tag . $this->copy_until('<>'); |
| 1959 |
|
| 1960 |
// Next char is the beginning of a new tag, don't touch it. |
| 1961 |
if ($this->char === '<') { |
| 1962 |
$this->link_nodes($node, false); |
| 1963 |
return true; |
| 1964 |
} |
| 1965 |
|
| 1966 |
// Next char closes current tag, add and be done with it. |
| 1967 |
if ($this->char === '>') { $node->_[HGFLHDOM_INFO_TEXT] .= '>'; } |
| 1968 |
$this->link_nodes($node, false); |
| 1969 |
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 1970 |
return true; |
| 1971 |
} |
| 1972 |
|
| 1973 |
// begin tag, add new node |
| 1974 |
$node->nodetype = HGFLHDOM_TYPE_ELEMENT; |
| 1975 |
$tag_lower = strtolower($tag); |
| 1976 |
$node->tag = ($this->lowercase) ? $tag_lower : $tag; |
| 1977 |
|
| 1978 |
// handle optional closing tags |
| 1979 |
if (isset($this->optional_closing_tags[$tag_lower])) { |
| 1980 |
// Traverse ancestors to close all optional closing tags |
| 1981 |
while (isset($this->optional_closing_tags[$tag_lower][strtolower($this->parent->tag)])) { |
| 1982 |
$this->parent->_[HGFLHDOM_INFO_END] = 0; |
| 1983 |
$this->parent = $this->parent->parent; |
| 1984 |
} |
| 1985 |
$node->parent = $this->parent; |
| 1986 |
} |
| 1987 |
|
| 1988 |
$guard = 0; // prevent infinity loop |
| 1989 |
|
| 1990 |
// [0] Space between tag and first attribute |
| 1991 |
$space = array($this->copy_skip($this->token_blank), '', ''); |
| 1992 |
|
| 1993 |
// attributes |
| 1994 |
do { |
| 1995 |
// Everything until the first equal sign should be the attribute name |
| 1996 |
$name = $this->copy_until($this->token_equal); |
| 1997 |
|
| 1998 |
if ($name === '' && $this->char !== null && $space[0] === '') { |
| 1999 |
break; |
| 2000 |
} |
| 2001 |
|
| 2002 |
if ($guard === $this->pos) { // Escape infinite loop |
| 2003 |
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 2004 |
continue; |
| 2005 |
} |
| 2006 |
|
| 2007 |
$guard = $this->pos; |
| 2008 |
|
| 2009 |
// handle endless '<' |
| 2010 |
// Out of bounds before the tag ended |
| 2011 |
if ($this->pos >= $this->size - 1 && $this->char !== '>') { |
| 2012 |
$node->nodetype = HGFLHDOM_TYPE_TEXT; |
| 2013 |
$node->_[HGFLHDOM_INFO_END] = 0; |
| 2014 |
$node->_[HGFLHDOM_INFO_TEXT] = '<' . $tag . $space[0] . $name; |
| 2015 |
$node->tag = 'text'; |
| 2016 |
$this->link_nodes($node, false); |
| 2017 |
return true; |
| 2018 |
} |
| 2019 |
|
| 2020 |
// handle mismatch '<' |
| 2021 |
// Attributes cannot start after opening tag |
| 2022 |
if ($this->doc[$this->pos - 1] == '<') { |
| 2023 |
$node->nodetype = HGFLHDOM_TYPE_TEXT; |
| 2024 |
$node->tag = 'text'; |
| 2025 |
$node->attr = array(); |
| 2026 |
$node->_[HGFLHDOM_INFO_END] = 0; |
| 2027 |
$node->_[HGFLHDOM_INFO_TEXT] = substr( |
| 2028 |
$this->doc, |
| 2029 |
$begin_tag_pos, |
| 2030 |
$this->pos - $begin_tag_pos - 1 |
| 2031 |
); |
| 2032 |
$this->pos -= 2; |
| 2033 |
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 2034 |
$this->link_nodes($node, false); |
| 2035 |
return true; |
| 2036 |
} |
| 2037 |
|
| 2038 |
if ($name !== '/' && $name !== '') { // this is a attribute name |
| 2039 |
// [1] Whitespace after attribute name |
| 2040 |
$space[1] = $this->copy_skip($this->token_blank); |
| 2041 |
|
| 2042 |
$name = $this->restore_noise($name); // might be a noisy name |
| 2043 |
|
| 2044 |
if ($this->lowercase) { $name = strtolower($name); } |
| 2045 |
|
| 2046 |
if ($this->char === '=') { // attribute with value |
| 2047 |
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 2048 |
$this->parse_attr($node, $name, $space); // get attribute value |
| 2049 |
} else { |
| 2050 |
//no value attr: nowrap, checked selected... |
| 2051 |
$node->_[HGFLHDOM_INFO_QUOTE][] = HGFLHDOM_QUOTE_NO; |
| 2052 |
$node->attr[$name] = true; |
| 2053 |
if ($this->char != '>') { $this->char = $this->doc[--$this->pos]; } // prev |
| 2054 |
} |
| 2055 |
|
| 2056 |
$node->_[HGFLHDOM_INFO_SPACE][] = $space; |
| 2057 |
|
| 2058 |
// prepare for next attribute |
| 2059 |
$space = array( |
| 2060 |
$this->copy_skip($this->token_blank), |
| 2061 |
'', |
| 2062 |
'' |
| 2063 |
); |
| 2064 |
} else { // no more attributes |
| 2065 |
break; |
| 2066 |
} |
| 2067 |
} while ($this->char !== '>' && $this->char !== '/'); // go until the tag ended |
| 2068 |
|
| 2069 |
$this->link_nodes($node, true); |
| 2070 |
$node->_[HGFLHDOM_INFO_ENDSPACE] = $space[0]; |
| 2071 |
|
| 2072 |
// handle empty tags (i.e. "<div/>") |
| 2073 |
if ($this->copy_until_char('>') === '/') { |
| 2074 |
$node->_[HGFLHDOM_INFO_ENDSPACE] .= '/'; |
| 2075 |
$node->_[HGFLHDOM_INFO_END] = 0; |
| 2076 |
} else { |
| 2077 |
// reset parent |
| 2078 |
if (!isset($this->self_closing_tags[strtolower($node->tag)])) { |
| 2079 |
$this->parent = $node; |
| 2080 |
} |
| 2081 |
} |
| 2082 |
|
| 2083 |
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 2084 |
|
| 2085 |
// If it's a BR tag, we need to set it's text to the default text. |
| 2086 |
// This way when we see it in plaintext, we can generate formatting that the user wants. |
| 2087 |
// since a br tag never has sub nodes, this works well. |
| 2088 |
if ($node->tag === 'br') { |
| 2089 |
$node->_[HGFLHDOM_INFO_INNER] = $this->HGFLDEFAULT_br_text; |
| 2090 |
} |
| 2091 |
|
| 2092 |
return true; |
| 2093 |
} |
| 2094 |
|
| 2095 |
protected function parse_attr($node, $name, &$space) |
| 2096 |
{ |
| 2097 |
$is_duplicate = isset($node->attr[$name]); |
| 2098 |
|
| 2099 |
if (!$is_duplicate) // Copy whitespace between "=" and value |
| 2100 |
$space[2] = $this->copy_skip($this->token_blank); |
| 2101 |
|
| 2102 |
switch ($this->char) { |
| 2103 |
case '"': |
| 2104 |
$quote_type = HGFLHDOM_QUOTE_DOUBLE; |
| 2105 |
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 2106 |
$value = $this->copy_until_char('"'); |
| 2107 |
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 2108 |
break; |
| 2109 |
case '\'': |
| 2110 |
$quote_type = HGFLHDOM_QUOTE_SINGLE; |
| 2111 |
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 2112 |
$value = $this->copy_until_char('\''); |
| 2113 |
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 2114 |
break; |
| 2115 |
default: |
| 2116 |
$quote_type = HGFLHDOM_QUOTE_NO; |
| 2117 |
$value = $this->copy_until($this->token_attr); |
| 2118 |
} |
| 2119 |
|
| 2120 |
$value = $this->restore_noise($value); |
| 2121 |
|
| 2122 |
// PaperG: Attributes should not have \r or \n in them, that counts as |
| 2123 |
// html whitespace. |
| 2124 |
if ($name !== 'd') { |
| 2125 |
$value = str_replace("\r", '', $value); |
| 2126 |
$value = str_replace("\n", '', $value); |
| 2127 |
} |
| 2128 |
|
| 2129 |
// PaperG: If this is a "class" selector, lets get rid of the preceeding |
| 2130 |
// and trailing space since some phgflle leave it in the multi class case. |
| 2131 |
if ($name === 'class') { |
| 2132 |
$value = trim($value); |
| 2133 |
} |
| 2134 |
|
| 2135 |
if (!$is_duplicate) { |
| 2136 |
$node->_[HGFLHDOM_INFO_QUOTE][] = $quote_type; |
| 2137 |
$node->attr[$name] = $value; |
| 2138 |
} |
| 2139 |
} |
| 2140 |
|
| 2141 |
protected function link_nodes(&$node, $is_child) |
| 2142 |
{ |
| 2143 |
$node->parent = $this->parent; |
| 2144 |
$this->parent->nodes[] = $node; |
| 2145 |
if ($is_child) { |
| 2146 |
$this->parent->children[] = $node; |
| 2147 |
} |
| 2148 |
} |
| 2149 |
|
| 2150 |
protected function as_text_node($tag) |
| 2151 |
{ |
| 2152 |
$node = new hgfl_simple_html_dom_node($this); |
| 2153 |
++$this->cursor; |
| 2154 |
$node->_[HGFLHDOM_INFO_TEXT] = '</' . $tag . '>'; |
| 2155 |
$this->link_nodes($node, false); |
| 2156 |
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 2157 |
return true; |
| 2158 |
} |
| 2159 |
|
| 2160 |
protected function skip($chars) |
| 2161 |
{ |
| 2162 |
$this->pos += strspn($this->doc, $chars, $this->pos); |
| 2163 |
$this->char = ($this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 2164 |
} |
| 2165 |
|
| 2166 |
protected function copy_skip($chars) |
| 2167 |
{ |
| 2168 |
$pos = $this->pos; |
| 2169 |
$len = strspn($this->doc, $chars, $pos); |
| 2170 |
$this->pos += $len; |
| 2171 |
$this->char = ($this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 2172 |
if ($len === 0) { return ''; } |
| 2173 |
return substr($this->doc, $pos, $len); |
| 2174 |
} |
| 2175 |
|
| 2176 |
protected function copy_until($chars) |
| 2177 |
{ |
| 2178 |
$pos = $this->pos; |
| 2179 |
$len = strcspn($this->doc, $chars, $pos); |
| 2180 |
$this->pos += $len; |
| 2181 |
$this->char = ($this->pos < $this->size) ? $this->doc[$this->pos] : null; // next |
| 2182 |
return substr($this->doc, $pos, $len); |
| 2183 |
} |
| 2184 |
|
| 2185 |
protected function copy_until_char($char) |
| 2186 |
{ |
| 2187 |
if ($this->char === null) { return ''; } |
| 2188 |
|
| 2189 |
if (($pos = strpos($this->doc, $char, $this->pos)) === false) { |
| 2190 |
$ret = substr($this->doc, $this->pos, $this->size - $this->pos); |
| 2191 |
$this->char = null; |
| 2192 |
$this->pos = $this->size; |
| 2193 |
return $ret; |
| 2194 |
} |
| 2195 |
|
| 2196 |
if ($pos === $this->pos) { return ''; } |
| 2197 |
|
| 2198 |
$pos_old = $this->pos; |
| 2199 |
$this->char = $this->doc[$pos]; |
| 2200 |
$this->pos = $pos; |
| 2201 |
return substr($this->doc, $pos_old, $pos - $pos_old); |
| 2202 |
} |
| 2203 |
|
| 2204 |
protected function remove_noise($pattern, $remove_tag = false) |
| 2205 |
{ |
| 2206 |
global $debug_object; |
| 2207 |
if (is_object($debug_object)) { $debug_object->debug_log_entry(1); } |
| 2208 |
|
| 2209 |
$count = preg_match_all( |
| 2210 |
$pattern, |
| 2211 |
$this->doc, |
| 2212 |
$matches, |
| 2213 |
PREG_SET_ORDER | PREG_OFFSET_CAPTURE |
| 2214 |
); |
| 2215 |
|
| 2216 |
for ($i = $count - 1; $i > -1; --$i) { |
| 2217 |
$key = '___noise___' . sprintf('% 5d', count($this->noise) + 1000); |
| 2218 |
|
| 2219 |
if (is_object($debug_object)) { |
| 2220 |
$debug_object->debug_log(2, 'key is: ' . $key); |
| 2221 |
} |
| 2222 |
|
| 2223 |
$idx = ($remove_tag) ? 0 : 1; // 0 = entire match, 1 = submatch |
| 2224 |
$this->noise[$key] = $matches[$i][$idx][0]; |
| 2225 |
$this->doc = substr_replace($this->doc, $key, $matches[$i][$idx][1], strlen($matches[$i][$idx][0])); |
| 2226 |
} |
| 2227 |
|
| 2228 |
// reset the length of content |
| 2229 |
$this->size = strlen($this->doc); |
| 2230 |
|
| 2231 |
if ($this->size > 0) { |
| 2232 |
$this->char = $this->doc[0]; |
| 2233 |
} |
| 2234 |
} |
| 2235 |
|
| 2236 |
function restore_noise($text) |
| 2237 |
{ |
| 2238 |
global $debug_object; |
| 2239 |
if (is_object($debug_object)) { $debug_object->debug_log_entry(1); } |
| 2240 |
|
| 2241 |
while (($pos = strpos($text, '___noise___')) !== false) { |
| 2242 |
// Sometimes there is a broken piece of markup, and we don't GET the |
| 2243 |
// pos+11 etc... token which indicates a problem outside of us... |
| 2244 |
|
| 2245 |
// todo: "___noise___1000" (or any number with four or more digits) |
| 2246 |
// in the DOM causes an infinite loop which could be utilized by |
| 2247 |
// malicious software |
| 2248 |
if (strlen($text) > $pos + 15) { |
| 2249 |
$key = '___noise___' |
| 2250 |
. $text[$pos + 11] |
| 2251 |
. $text[$pos + 12] |
| 2252 |
. $text[$pos + 13] |
| 2253 |
. $text[$pos + 14] |
| 2254 |
. $text[$pos + 15]; |
| 2255 |
|
| 2256 |
if (is_object($debug_object)) { |
| 2257 |
$debug_object->debug_log(2, 'located key of: ' . $key); |
| 2258 |
} |
| 2259 |
|
| 2260 |
if (isset($this->noise[$key])) { |
| 2261 |
$text = substr($text, 0, $pos) |
| 2262 |
. $this->noise[$key] |
| 2263 |
. substr($text, $pos + 16); |
| 2264 |
} else { |
| 2265 |
// do this to prevent an infinite loop. |
| 2266 |
$text = substr($text, 0, $pos) |
| 2267 |
. 'UNDEFINED NOISE FOR KEY: ' |
| 2268 |
. $key |
| 2269 |
. substr($text, $pos + 16); |
| 2270 |
} |
| 2271 |
} else { |
| 2272 |
// There is no valid key being given back to us... We must get |
| 2273 |
// rid of the ___noise___ or we will have a problem. |
| 2274 |
$text = substr($text, 0, $pos) |
| 2275 |
. 'NO NUMERIC NOISE KEY' |
| 2276 |
. substr($text, $pos + 11); |
| 2277 |
} |
| 2278 |
} |
| 2279 |
return $text; |
| 2280 |
} |
| 2281 |
|
| 2282 |
function search_noise($text) |
| 2283 |
{ |
| 2284 |
global $debug_object; |
| 2285 |
if (is_object($debug_object)) { $debug_object->debug_log_entry(1); } |
| 2286 |
|
| 2287 |
foreach($this->noise as $noiseElement) { |
| 2288 |
if (strpos($noiseElement, $text) !== false) { |
| 2289 |
return $noiseElement; |
| 2290 |
} |
| 2291 |
} |
| 2292 |
} |
| 2293 |
|
| 2294 |
function __toString() |
| 2295 |
{ |
| 2296 |
return $this->root->innertext(); |
| 2297 |
} |
| 2298 |
|
| 2299 |
function __get($name) |
| 2300 |
{ |
| 2301 |
switch ($name) { |
| 2302 |
case 'outertext': |
| 2303 |
return $this->root->innertext(); |
| 2304 |
case 'innertext': |
| 2305 |
return $this->root->innertext(); |
| 2306 |
case 'plaintext': |
| 2307 |
return $this->root->text(); |
| 2308 |
case 'charset': |
| 2309 |
return $this->_charset; |
| 2310 |
case 'target_charset': |
| 2311 |
return $this->_target_charset; |
| 2312 |
} |
| 2313 |
} |
| 2314 |
|
| 2315 |
function childNodes($idx = -1) |
| 2316 |
{ |
| 2317 |
return $this->root->childNodes($idx); |
| 2318 |
} |
| 2319 |
|
| 2320 |
function firstChild() |
| 2321 |
{ |
| 2322 |
return $this->root->first_child(); |
| 2323 |
} |
| 2324 |
|
| 2325 |
function lastChild() |
| 2326 |
{ |
| 2327 |
return $this->root->last_child(); |
| 2328 |
} |
| 2329 |
|
| 2330 |
function createElement($name, $value = null) |
| 2331 |
{ |
| 2332 |
return @hgfl_str_get_html("<$name>$value</$name>")->firstChild(); |
| 2333 |
} |
| 2334 |
|
| 2335 |
function createTextNode($value) |
| 2336 |
{ |
| 2337 |
return @end(hgfl_str_get_html($value)->nodes); |
| 2338 |
} |
| 2339 |
|
| 2340 |
function getElementById($id) |
| 2341 |
{ |
| 2342 |
return $this->find("#$id", 0); |
| 2343 |
} |
| 2344 |
|
| 2345 |
function getElementsById($id, $idx = null) |
| 2346 |
{ |
| 2347 |
return $this->find("#$id", $idx); |
| 2348 |
} |
| 2349 |
|
| 2350 |
function getElementByTagName($name) |
| 2351 |
{ |
| 2352 |
return $this->find($name, 0); |
| 2353 |
} |
| 2354 |
|
| 2355 |
function getElementsByTagName($name, $idx = -1) |
| 2356 |
{ |
| 2357 |
return $this->find($name, $idx); |
| 2358 |
} |
| 2359 |
|
| 2360 |
function loadFile() |
| 2361 |
{ |
| 2362 |
$args = func_get_args(); |
| 2363 |
$this->load_file($args); |
| 2364 |
} |
| 2365 |
} |
