Version Description
2022-11-30=
Changed simple replacement by DOMDocument parsing & replacing only on string.
Download this release
Release Info
| Developer | iworks |
| Plugin |  Orphans |
| Version | 3.1.0 |
| Comparing to | |
| See all releases | |
Code changes from version 3.0.4 to 3.1.0
- includes/iworks/class-iworks-orphan.php +140 -53
- languages/sierotki.pot +2 -2
- readme.txt +6 -3
- sierotki.php +5 -1
- vendor/index.php +2 -0
- vendor/simple_html_dom.php +2353 -0
includes/iworks/class-iworks-orphan.php
CHANGED
|
@@ -48,7 +48,26 @@ class iworks_orphan {
|
|
| 48 |
*
|
| 49 |
* @since 3.0.0
|
| 50 |
*/
|
| 51 |
-
private $version = '3.0
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 52 |
|
| 53 |
public function __construct() {
|
| 54 |
/**
|
|
@@ -193,59 +212,52 @@ class iworks_orphan {
|
|
| 193 |
}
|
| 194 |
|
| 195 |
/**
|
| 196 |
-
*
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 197 |
* possible.
|
| 198 |
*
|
| 199 |
-
* @since
|
| 200 |
*
|
| 201 |
* @param string $content String to replace
|
| 202 |
*
|
| 203 |
* @return string $content
|
| 204 |
*/
|
| 205 |
-
private function
|
| 206 |
/**
|
| 207 |
* only super-base check
|
| 208 |
*/
|
| 209 |
if ( ! is_string( $content ) || empty( $content ) ) {
|
| 210 |
return $content;
|
| 211 |
}
|
| 212 |
-
/**
|
| 213 |
-
* Avoid to replace attributes
|
| 214 |
-
*/
|
| 215 |
-
$attributes = array();
|
| 216 |
-
if ( $this->options->get_option( 'attributes' ) ) {
|
| 217 |
-
preg_match_all( '/(style|class|data-[a-z\-]+)="[^"]+"/', $content, $matches );
|
| 218 |
-
if ( ! empty( $matches ) ) {
|
| 219 |
-
$salt = 'Sae9ieCheyieph3ug7si4yeiBoog0fae4yae6biexaimie0ied7quienae3yeepo';
|
| 220 |
-
foreach ( $matches[0] as $value ) {
|
| 221 |
-
if ( empty( $value ) ) {
|
| 222 |
-
continue;
|
| 223 |
-
}
|
| 224 |
-
if ( preg_match( '/ /', $value ) ) {
|
| 225 |
-
$attributes[ $value ] = md5( $salt . $value );
|
| 226 |
-
}
|
| 227 |
-
}
|
| 228 |
-
if ( ! empty( $attributes ) ) {
|
| 229 |
-
foreach ( $attributes as $part => $to_change ) {
|
| 230 |
-
$content = str_replace( $part, $to_change, $content );
|
| 231 |
-
}
|
| 232 |
-
}
|
| 233 |
-
}
|
| 234 |
-
}
|
| 235 |
-
/**
|
| 236 |
-
* Avoid to replace inside script or styles tags
|
| 237 |
-
*/
|
| 238 |
-
preg_match_all( '@(<(script|style)[^>]*>.*?(</(script|style)>))@is', $content, $matches );
|
| 239 |
-
$exceptions = array();
|
| 240 |
-
if ( ! empty( $matches ) && ! empty( $matches[0] ) ) {
|
| 241 |
-
$salt = 'kQc6T9fn5GhEzTM3Sxn7b9TWMV4PO0mOCV06Da7AQJzSJqxYR4z3qBlsW9rtFsWK';
|
| 242 |
-
foreach ( $matches[0] as $one ) {
|
| 243 |
-
$key = sprintf( '<!-- %s %s -->', $salt, md5( $one ) );
|
| 244 |
-
$exceptions[ $key ] = $one;
|
| 245 |
-
$re = sprintf( '@%s@', preg_replace( '/@/', '\@', preg_quote( $one, '/' ) ) );
|
| 246 |
-
$content = preg_replace( $re, $key, $content );
|
| 247 |
-
}
|
| 248 |
-
}
|
| 249 |
/**
|
| 250 |
* Keep numbers together - this is independed of current language
|
| 251 |
*/
|
|
@@ -257,7 +269,7 @@ class iworks_orphan {
|
|
| 257 |
foreach ( $parts as $part ) {
|
| 258 |
$to_change = $part;
|
| 259 |
while ( preg_match( '/(\d+) ([\da-z]+)/i', $to_change, $matches ) ) {
|
| 260 |
-
$to_change = preg_replace( '/(\d+) ([\da-z]+)/i', '$1
|
| 261 |
}
|
| 262 |
if ( $part != $to_change ) {
|
| 263 |
$content = str_replace( $part, $to_change, $content );
|
|
@@ -288,16 +300,16 @@ class iworks_orphan {
|
|
| 288 |
* base therms replace
|
| 289 |
*/
|
| 290 |
$re = '/^([aiouwz]|' . preg_replace( '/\./', '\.', implode( '|', $terms ) ) . ') +/i';
|
| 291 |
-
$part_to_change = preg_replace( $re, '$1$2
|
| 292 |
/**
|
| 293 |
* single letters
|
| 294 |
*/
|
| 295 |
-
$re = '/([ >\(]
|
| 296 |
/**
|
| 297 |
* double call to handle orphan after orphan after orphan
|
| 298 |
*/
|
| 299 |
-
$part_to_change = preg_replace( $re, '$1$2
|
| 300 |
-
$part_to_change = preg_replace( $re, '$1$2
|
| 301 |
}
|
| 302 |
if ( $part_source !== $part_to_change ) {
|
| 303 |
$content = str_replace( $part_source, $part_to_change, $content );
|
|
@@ -306,8 +318,8 @@ class iworks_orphan {
|
|
| 306 |
/**
|
| 307 |
* single letter after previous orphan
|
| 308 |
*/
|
| 309 |
-
$re = '/(
|
| 310 |
-
$content = preg_replace( $re, '$1$2
|
| 311 |
/**
|
| 312 |
* bring back styles & scripts
|
| 313 |
*/
|
|
@@ -330,13 +342,80 @@ class iworks_orphan {
|
|
| 330 |
*
|
| 331 |
* @since 3.0.4
|
| 332 |
*/
|
| 333 |
-
$content = preg_replace( '/(\d) r\./', '$1
|
| 334 |
/**
|
| 335 |
* return
|
| 336 |
*/
|
| 337 |
return $content;
|
| 338 |
}
|
| 339 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 340 |
/**
|
| 341 |
* Inicialize admin area
|
| 342 |
*/
|
|
@@ -464,13 +543,21 @@ class iworks_orphan {
|
|
| 464 |
* @since 3.0.2
|
| 465 |
*/
|
| 466 |
add_filter( 'vc_shortcode_output', array( $this, 'replace' ) );
|
| 467 |
-
|
| 468 |
/**
|
| 469 |
-
*
|
| 470 |
*
|
| 471 |
-
* @since 3.0
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 472 |
*/
|
| 473 |
-
|
|
|
|
|
|
|
|
|
|
| 474 |
}
|
| 475 |
|
| 476 |
/**
|
| 48 |
*
|
| 49 |
* @since 3.0.0
|
| 50 |
*/
|
| 51 |
+
private $version = '3.1.0';
|
| 52 |
+
|
| 53 |
+
/**
|
| 54 |
+
* tags to avoid replacement
|
| 55 |
+
*
|
| 56 |
+
* @since 3.1.0
|
| 57 |
+
*/
|
| 58 |
+
private $protected_tags = array(
|
| 59 |
+
'script',
|
| 60 |
+
'style',
|
| 61 |
+
'iframe',
|
| 62 |
+
'svg',
|
| 63 |
+
);
|
| 64 |
+
|
| 65 |
+
/**
|
| 66 |
+
* nbsp placehlder
|
| 67 |
+
*
|
| 68 |
+
* @since 3.1.0
|
| 69 |
+
*/
|
| 70 |
+
private $nbsp_placeholder = ' ';
|
| 71 |
|
| 72 |
public function __construct() {
|
| 73 |
/**
|
| 212 |
}
|
| 213 |
|
| 214 |
/**
|
| 215 |
+
* Parse DOMElement Object
|
| 216 |
+
*
|
| 217 |
+
* @since 3.1.0
|
| 218 |
+
*/
|
| 219 |
+
private function parse_item( $item ) {
|
| 220 |
+
/**
|
| 221 |
+
* no tags, replace
|
| 222 |
+
*/
|
| 223 |
+
if ( ! preg_match( '/</', $item->innertext ) ) {
|
| 224 |
+
$item->innertext = $this->string_replacement( $item->innertext );
|
| 225 |
+
return;
|
| 226 |
+
}
|
| 227 |
+
/**
|
| 228 |
+
* split to slices & replace!
|
| 229 |
+
*/
|
| 230 |
+
preg_match_all( '/<[^>]+>/', $item->innertext, $matches );
|
| 231 |
+
$text_array = preg_split( '/<[^>]+>/', $item->innertext );
|
| 232 |
+
$text = '';
|
| 233 |
+
$max = sizeof( $text_array );
|
| 234 |
+
for ( $i = 0;$i < $max;$i++ ) {
|
| 235 |
+
$text .= $this->string_replacement( $text_array[ $i ] );
|
| 236 |
+
if ( isset( $matches[0][ $i ] ) ) {
|
| 237 |
+
$text .= $matches[0][ $i ];
|
| 238 |
+
}
|
| 239 |
+
}
|
| 240 |
+
$item->innertext = $text;
|
| 241 |
+
return;
|
| 242 |
+
}
|
| 243 |
+
|
| 244 |
+
/**
|
| 245 |
+
* String replacement with super-base check is replacement even
|
| 246 |
* possible.
|
| 247 |
*
|
| 248 |
+
* @since 3.1.0
|
| 249 |
*
|
| 250 |
* @param string $content String to replace
|
| 251 |
*
|
| 252 |
* @return string $content
|
| 253 |
*/
|
| 254 |
+
private function string_replacement( $content ) {
|
| 255 |
/**
|
| 256 |
* only super-base check
|
| 257 |
*/
|
| 258 |
if ( ! is_string( $content ) || empty( $content ) ) {
|
| 259 |
return $content;
|
| 260 |
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 261 |
/**
|
| 262 |
* Keep numbers together - this is independed of current language
|
| 263 |
*/
|
| 269 |
foreach ( $parts as $part ) {
|
| 270 |
$to_change = $part;
|
| 271 |
while ( preg_match( '/(\d+) ([\da-z]+)/i', $to_change, $matches ) ) {
|
| 272 |
+
$to_change = preg_replace( '/(\d+) ([\da-z]+)/i', '$1' . $this->nbsp_placeholder . '$2', $to_change );
|
| 273 |
}
|
| 274 |
if ( $part != $to_change ) {
|
| 275 |
$content = str_replace( $part, $to_change, $content );
|
| 300 |
* base therms replace
|
| 301 |
*/
|
| 302 |
$re = '/^([aiouwz]|' . preg_replace( '/\./', '\.', implode( '|', $terms ) ) . ') +/i';
|
| 303 |
+
$part_to_change = preg_replace( $re, '$1$2' . $this->nbsp_placeholder, $part_to_change );
|
| 304 |
/**
|
| 305 |
* single letters
|
| 306 |
*/
|
| 307 |
+
$re = '/([ >\(]+|' . $this->nbsp_placeholder . '|„|")([aiouwz]|' . preg_replace( '/\./', '\.', implode( '|', $terms ) ) . ') +/i';
|
| 308 |
/**
|
| 309 |
* double call to handle orphan after orphan after orphan
|
| 310 |
*/
|
| 311 |
+
$part_to_change = preg_replace( $re, '$1$2' . $this->nbsp_placeholder, $part_to_change );
|
| 312 |
+
$part_to_change = preg_replace( $re, '$1$2' . $this->nbsp_placeholder, $part_to_change );
|
| 313 |
}
|
| 314 |
if ( $part_source !== $part_to_change ) {
|
| 315 |
$content = str_replace( $part_source, $part_to_change, $content );
|
| 318 |
/**
|
| 319 |
* single letter after previous orphan
|
| 320 |
*/
|
| 321 |
+
$re = '/(' . $this->nbsp_placeholder . ')([aiouwz]) +/i';
|
| 322 |
+
$content = preg_replace( $re, '$1$2' . $this->nbsp_placeholder, $content );
|
| 323 |
/**
|
| 324 |
* bring back styles & scripts
|
| 325 |
*/
|
| 342 |
*
|
| 343 |
* @since 3.0.4
|
| 344 |
*/
|
| 345 |
+
$content = preg_replace( '/(\d) r\./', '$1' . $this->nbsp_placeholder . 'r.', $content );
|
| 346 |
/**
|
| 347 |
* return
|
| 348 |
*/
|
| 349 |
return $content;
|
| 350 |
}
|
| 351 |
|
| 352 |
+
/**
|
| 353 |
+
* Unconditional replacement with super-base check is replacement even
|
| 354 |
+
* possible.
|
| 355 |
+
*
|
| 356 |
+
* @since 2.7.8
|
| 357 |
+
* @since 3.1.0 - changed into DOM parsing
|
| 358 |
+
*
|
| 359 |
+
*
|
| 360 |
+
* @param string $content String to replace
|
| 361 |
+
*
|
| 362 |
+
* @return string $content
|
| 363 |
+
*/
|
| 364 |
+
private function unconditional_replacement( $content ) {
|
| 365 |
+
if ( ! is_string( $content ) || empty( $content ) ) {
|
| 366 |
+
return $content;
|
| 367 |
+
}
|
| 368 |
+
/**
|
| 369 |
+
* string, no tags
|
| 370 |
+
*/
|
| 371 |
+
if ( strip_tags( $content ) === $content ) {
|
| 372 |
+
return $this->string_replacement( $content );
|
| 373 |
+
}
|
| 374 |
+
/**
|
| 375 |
+
* parse
|
| 376 |
+
*/
|
| 377 |
+
$doc = str_get_html( $content );
|
| 378 |
+
/**
|
| 379 |
+
* remove protected tags
|
| 380 |
+
*/
|
| 381 |
+
$protected = array();
|
| 382 |
+
foreach ( $this->protected_tags as $tag ) {
|
| 383 |
+
foreach ( $doc->find( $tag ) as $item ) {
|
| 384 |
+
$innertext = $item->innertext;
|
| 385 |
+
$attributes = $item->getAllAttributes();
|
| 386 |
+
$key = md5( $tag . $innertext . implode( $attributes ) );
|
| 387 |
+
$protected[ $key ] = array(
|
| 388 |
+
'attributes' => $attributes,
|
| 389 |
+
'innertext' => $innertext,
|
| 390 |
+
);
|
| 391 |
+
foreach ( $attributes as $name => $value ) {
|
| 392 |
+
$item->removeAttribute( $name );
|
| 393 |
+
}
|
| 394 |
+
$item->innertext = '';
|
| 395 |
+
$item->setAttribute( 'id', 'orphans-' . $key );
|
| 396 |
+
}
|
| 397 |
+
}
|
| 398 |
+
/**
|
| 399 |
+
* replace
|
| 400 |
+
*/
|
| 401 |
+
foreach ( $doc->find( '*' ) as &$item ) {
|
| 402 |
+
$this->parse_item( $item );
|
| 403 |
+
}
|
| 404 |
+
/**
|
| 405 |
+
* revert protected tags
|
| 406 |
+
*/
|
| 407 |
+
foreach ( $protected as $key => $data ) {
|
| 408 |
+
foreach ( $doc->find( '#orphans-' . $key ) as $item ) {
|
| 409 |
+
$item->innertext = $data['innertext'];
|
| 410 |
+
$item->removeAttribute( 'id' );
|
| 411 |
+
foreach ( $data['attributes'] as $name => $value ) {
|
| 412 |
+
$item->setAttribute( $name, $value );
|
| 413 |
+
}
|
| 414 |
+
}
|
| 415 |
+
}
|
| 416 |
+
return $output = $doc->save();
|
| 417 |
+
}
|
| 418 |
+
|
| 419 |
/**
|
| 420 |
* Inicialize admin area
|
| 421 |
*/
|
| 543 |
* @since 3.0.2
|
| 544 |
*/
|
| 545 |
add_filter( 'vc_shortcode_output', array( $this, 'replace' ) );
|
|
|
|
| 546 |
/**
|
| 547 |
+
* Filter allowed change protected tags.
|
| 548 |
*
|
| 549 |
+
* @since 3.1.0
|
| 550 |
+
*
|
| 551 |
+
* @param array $args {
|
| 552 |
+
* Array of protected tags - all content of this tags will be not
|
| 553 |
+
* replaced
|
| 554 |
+
*
|
| 555 |
+
* @type string HTML tag name.
|
| 556 |
*/
|
| 557 |
+
$this->protected_tags = apply_filters(
|
| 558 |
+
'iworks_orphan_protected_tags',
|
| 559 |
+
$this->protected_tags
|
| 560 |
+
);
|
| 561 |
}
|
| 562 |
|
| 563 |
/**
|
languages/sierotki.pot
CHANGED
|
@@ -4,7 +4,7 @@ msgid ""
|
|
| 4 |
msgstr ""
|
| 5 |
"Project-Id-Version: Orphans PLUGIN_VERSION\n"
|
| 6 |
"Report-Msgid-Bugs-To: https://wordpress.org/support/plugin/sierotki-dev\n"
|
| 7 |
-
"POT-Creation-Date: 2022-
|
| 8 |
"MIME-Version: 1.0\n"
|
| 9 |
"Content-Type: text/plain; charset=utf-8\n"
|
| 10 |
"Content-Transfer-Encoding: 8bit\n"
|
|
@@ -232,7 +232,7 @@ msgstr ""
|
|
| 232 |
msgid "WordPress Help Forum"
|
| 233 |
msgstr ""
|
| 234 |
|
| 235 |
-
#: includes/iworks/class-iworks-orphan.php:
|
| 236 |
#: includes/iworks/rate/rate.php:124
|
| 237 |
msgid "Settings"
|
| 238 |
msgstr ""
|
| 4 |
msgstr ""
|
| 5 |
"Project-Id-Version: Orphans PLUGIN_VERSION\n"
|
| 6 |
"Report-Msgid-Bugs-To: https://wordpress.org/support/plugin/sierotki-dev\n"
|
| 7 |
+
"POT-Creation-Date: 2022-12-14 12:21:37+00:00\n"
|
| 8 |
"MIME-Version: 1.0\n"
|
| 9 |
"Content-Type: text/plain; charset=utf-8\n"
|
| 10 |
"Content-Transfer-Encoding: 8bit\n"
|
| 232 |
msgid "WordPress Help Forum"
|
| 233 |
msgstr ""
|
| 234 |
|
| 235 |
+
#: includes/iworks/class-iworks-orphan.php:584
|
| 236 |
#: includes/iworks/rate/rate.php:124
|
| 237 |
msgid "Settings"
|
| 238 |
msgstr ""
|
readme.txt
CHANGED
|
@@ -4,7 +4,7 @@ Donate link: https://ko-fi.com/iworks?utm_source=sierotki&utm_medium=readme-dona
|
|
| 4 |
Tags: sierotka, sierotki, spójniki, twarda spacja
|
| 5 |
Requires at least: 4.6
|
| 6 |
Tested up to: 6.1
|
| 7 |
-
Stable tag: 3.0
|
| 8 |
|
| 9 |
|
| 10 |
|
|
@@ -128,9 +128,12 @@ function remove_iworks_orphan_terms( $terms ) {
|
|
| 128 |
|
| 129 |
== Changelog ==
|
| 130 |
|
| 131 |
-
= 3.0
|
|
|
|
|
|
|
|
|
|
|
|
|
| 132 |
* Handle space after year for short year format "r.". Props for [Mastafu Design](https://wordpress.org/support/users/mastafu/)
|
| 133 |
-
* Added integration with "Goodlayers Core" on `gdlr_core_escape_content` filter.
|
| 134 |
|
| 135 |
= 3.0.3 - 2022-09-02 =
|
| 136 |
* Handle ACF integration if it is network activated plugin. Props for [maczek6000](https://profiles.wordpress.org/maczek6000/).
|
| 4 |
Tags: sierotka, sierotki, spójniki, twarda spacja
|
| 5 |
Requires at least: 4.6
|
| 6 |
Tested up to: 6.1
|
| 7 |
+
Stable tag: 3.1.0
|
| 8 |
|
| 9 |
|
| 10 |
|
| 128 |
|
| 129 |
== Changelog ==
|
| 130 |
|
| 131 |
+
= 3.1.0 - 2022-11-30=
|
| 132 |
+
|
| 133 |
+
* Changed simple replacement by DOMDocument parsing & replacing only on string.
|
| 134 |
+
|
| 135 |
+
= 3.0.4 - 2022-10-25 =
|
| 136 |
* Handle space after year for short year format "r.". Props for [Mastafu Design](https://wordpress.org/support/users/mastafu/)
|
|
|
|
| 137 |
|
| 138 |
= 3.0.3 - 2022-09-02 =
|
| 139 |
* Handle ACF integration if it is network activated plugin. Props for [maczek6000](https://profiles.wordpress.org/maczek6000/).
|
sierotki.php
CHANGED
|
@@ -5,7 +5,7 @@ Plugin URI: http://iworks.pl/2011/02/16/sierotki/
|
|
| 5 |
Text Domain: sierotki
|
| 6 |
Description: Implement Polish grammar rules with orphans.
|
| 7 |
Author: Marcin Pietrzak
|
| 8 |
-
Version: 3.0
|
| 9 |
Author URI: http://iworks.pl/
|
| 10 |
*/
|
| 11 |
|
|
@@ -13,6 +13,10 @@ include_once dirname( __FILE__ ) . '/etc/options.php';
|
|
| 13 |
|
| 14 |
load_plugin_textdomain( 'sierotki', false, dirname( __FILE__ ) . '/languages' );
|
| 15 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 16 |
$includes = dirname( __FILE__ ) . '/includes';
|
| 17 |
|
| 18 |
require_once $includes . '/iworks/class-iworks-orphan.php';
|
| 5 |
Text Domain: sierotki
|
| 6 |
Description: Implement Polish grammar rules with orphans.
|
| 7 |
Author: Marcin Pietrzak
|
| 8 |
+
Version: 3.1.0
|
| 9 |
Author URI: http://iworks.pl/
|
| 10 |
*/
|
| 11 |
|
| 13 |
|
| 14 |
load_plugin_textdomain( 'sierotki', false, dirname( __FILE__ ) . '/languages' );
|
| 15 |
|
| 16 |
+
if ( ! defined( 'HDOM_TYPE_ELEMENT' ) ) {
|
| 17 |
+
require_once dirname( __FILE__ ) . '/vendor/simple_html_dom.php';
|
| 18 |
+
}
|
| 19 |
+
|
| 20 |
$includes = dirname( __FILE__ ) . '/includes';
|
| 21 |
|
| 22 |
require_once $includes . '/iworks/class-iworks-orphan.php';
|
vendor/index.php
ADDED
|
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
| 1 |
+
<?php
|
| 2 |
+
exit;
|
vendor/simple_html_dom.php
ADDED
|
@@ -0,0 +1,2353 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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 |
+
define('HDOM_TYPE_ELEMENT', 1);
|
| 25 |
+
define('HDOM_TYPE_COMMENT', 2);
|
| 26 |
+
define('HDOM_TYPE_TEXT', 3);
|
| 27 |
+
define('HDOM_TYPE_ENDTAG', 4);
|
| 28 |
+
define('HDOM_TYPE_ROOT', 5);
|
| 29 |
+
define('HDOM_TYPE_UNKNOWN', 6);
|
| 30 |
+
define('HDOM_QUOTE_DOUBLE', 0);
|
| 31 |
+
define('HDOM_QUOTE_SINGLE', 1);
|
| 32 |
+
define('HDOM_QUOTE_NO', 3);
|
| 33 |
+
define('HDOM_INFO_BEGIN', 0);
|
| 34 |
+
define('HDOM_INFO_END', 1);
|
| 35 |
+
define('HDOM_INFO_QUOTE', 2);
|
| 36 |
+
define('HDOM_INFO_SPACE', 3);
|
| 37 |
+
define('HDOM_INFO_TEXT', 4);
|
| 38 |
+
define('HDOM_INFO_INNER', 5);
|
| 39 |
+
define('HDOM_INFO_OUTER', 6);
|
| 40 |
+
define('HDOM_INFO_ENDSPACE', 7);
|
| 41 |
+
|
| 42 |
+
defined('DEFAULT_TARGET_CHARSET') || define('DEFAULT_TARGET_CHARSET', 'UTF-8');
|
| 43 |
+
defined('DEFAULT_BR_TEXT') || define('DEFAULT_BR_TEXT', "\r\n");
|
| 44 |
+
defined('DEFAULT_SPAN_TEXT') || define('DEFAULT_SPAN_TEXT', ' ');
|
| 45 |
+
defined('MAX_FILE_SIZE') || define('MAX_FILE_SIZE', 600000);
|
| 46 |
+
define('HDOM_SMARTY_AS_TEXT', 1);
|
| 47 |
+
|
| 48 |
+
function file_get_html(
|
| 49 |
+
$url,
|
| 50 |
+
$use_include_path = false,
|
| 51 |
+
$context = null,
|
| 52 |
+
$offset = 0,
|
| 53 |
+
$maxLen = -1,
|
| 54 |
+
$lowercase = true,
|
| 55 |
+
$forceTagsClosed = true,
|
| 56 |
+
$target_charset = DEFAULT_TARGET_CHARSET,
|
| 57 |
+
$stripRN = true,
|
| 58 |
+
$defaultBRText = DEFAULT_BR_TEXT,
|
| 59 |
+
$defaultSpanText = DEFAULT_SPAN_TEXT)
|
| 60 |
+
{
|
| 61 |
+
if($maxLen <= 0) { $maxLen = MAX_FILE_SIZE; }
|
| 62 |
+
|
| 63 |
+
$dom = new simple_html_dom(
|
| 64 |
+
null,
|
| 65 |
+
$lowercase,
|
| 66 |
+
$forceTagsClosed,
|
| 67 |
+
$target_charset,
|
| 68 |
+
$stripRN,
|
| 69 |
+
$defaultBRText,
|
| 70 |
+
$defaultSpanText
|
| 71 |
+
);
|
| 72 |
+
|
| 73 |
+
/**
|
| 74 |
+
* For sourceforge users: uncomment the next line and comment the
|
| 75 |
+
* retrieve_url_contents line 2 lines down if it is not already done.
|
| 76 |
+
*/
|
| 77 |
+
$contents = file_get_contents(
|
| 78 |
+
$url,
|
| 79 |
+
$use_include_path,
|
| 80 |
+
$context,
|
| 81 |
+
$offset,
|
| 82 |
+
$maxLen
|
| 83 |
+
);
|
| 84 |
+
// $contents = retrieve_url_contents($url);
|
| 85 |
+
|
| 86 |
+
if (empty($contents) || strlen($contents) > $maxLen) {
|
| 87 |
+
$dom->clear();
|
| 88 |
+
return false;
|
| 89 |
+
}
|
| 90 |
+
|
| 91 |
+
return $dom->load($contents, $lowercase, $stripRN);
|
| 92 |
+
}
|
| 93 |
+
|
| 94 |
+
function str_get_html(
|
| 95 |
+
$str,
|
| 96 |
+
$lowercase = true,
|
| 97 |
+
$forceTagsClosed = true,
|
| 98 |
+
$target_charset = DEFAULT_TARGET_CHARSET,
|
| 99 |
+
$stripRN = true,
|
| 100 |
+
$defaultBRText = DEFAULT_BR_TEXT,
|
| 101 |
+
$defaultSpanText = DEFAULT_SPAN_TEXT)
|
| 102 |
+
{
|
| 103 |
+
$dom = new simple_html_dom(
|
| 104 |
+
null,
|
| 105 |
+
$lowercase,
|
| 106 |
+
$forceTagsClosed,
|
| 107 |
+
$target_charset,
|
| 108 |
+
$stripRN,
|
| 109 |
+
$defaultBRText,
|
| 110 |
+
$defaultSpanText
|
| 111 |
+
);
|
| 112 |
+
|
| 113 |
+
if (empty($str) || strlen($str) > MAX_FILE_SIZE) {
|
| 114 |
+
$dom->clear();
|
| 115 |
+
return false;
|
| 116 |
+
}
|
| 117 |
+
|
| 118 |
+
return $dom->load($str, $lowercase, $stripRN);
|
| 119 |
+
}
|
| 120 |
+
|
| 121 |
+
function dump_html_tree($node, $show_attr = true, $deep = 0)
|
| 122 |
+
{
|
| 123 |
+
$node->dump($node);
|
| 124 |
+
}
|
| 125 |
+
|
| 126 |
+
class simple_html_dom_node
|
| 127 |
+
{
|
| 128 |
+
public $nodetype = HDOM_TYPE_TEXT;
|
| 129 |
+
public $tag = 'text';
|
| 130 |
+
public $attr = array();
|
| 131 |
+
public $children = array();
|
| 132 |
+
public $nodes = array();
|
| 133 |
+
public $parent = null;
|
| 134 |
+
public $_ = array();
|
| 135 |
+
public $tag_start = 0;
|
| 136 |
+
private $dom = null;
|
| 137 |
+
|
| 138 |
+
function __construct($dom)
|
| 139 |
+
{
|
| 140 |
+
$this->dom = $dom;
|
| 141 |
+
$dom->nodes[] = $this;
|
| 142 |
+
}
|
| 143 |
+
|
| 144 |
+
function __destruct()
|
| 145 |
+
{
|
| 146 |
+
$this->clear();
|
| 147 |
+
}
|
| 148 |
+
|
| 149 |
+
function __toString()
|
| 150 |
+
{
|
| 151 |
+
return $this->outertext();
|
| 152 |
+
}
|
| 153 |
+
|
| 154 |
+
function clear()
|
| 155 |
+
{
|
| 156 |
+
$this->dom = null;
|
| 157 |
+
$this->nodes = null;
|
| 158 |
+
$this->parent = null;
|
| 159 |
+
$this->children = null;
|
| 160 |
+
}
|
| 161 |
+
|
| 162 |
+
function dump($show_attr = true, $depth = 0)
|
| 163 |
+
{
|
| 164 |
+
echo str_repeat("\t", $depth) . $this->tag;
|
| 165 |
+
|
| 166 |
+
if ($show_attr && count($this->attr) > 0) {
|
| 167 |
+
echo '(';
|
| 168 |
+
foreach ($this->attr as $k => $v) {
|
| 169 |
+
echo "[$k]=>\"$v\", ";
|
| 170 |
+
}
|
| 171 |
+
echo ')';
|
| 172 |
+
}
|
| 173 |
+
|
| 174 |
+
echo "\n";
|
| 175 |
+
|
| 176 |
+
if ($this->nodes) {
|
| 177 |
+
foreach ($this->nodes as $node) {
|
| 178 |
+
$node->dump($show_attr, $depth + 1);
|
| 179 |
+
}
|
| 180 |
+
}
|
| 181 |
+
}
|
| 182 |
+
|
| 183 |
+
function dump_node($echo = true)
|
| 184 |
+
{
|
| 185 |
+
$string = $this->tag;
|
| 186 |
+
|
| 187 |
+
if (count($this->attr) > 0) {
|
| 188 |
+
$string .= '(';
|
| 189 |
+
foreach ($this->attr as $k => $v) {
|
| 190 |
+
$string .= "[$k]=>\"$v\", ";
|
| 191 |
+
}
|
| 192 |
+
$string .= ')';
|
| 193 |
+
}
|
| 194 |
+
|
| 195 |
+
if (count($this->_) > 0) {
|
| 196 |
+
$string .= ' $_ (';
|
| 197 |
+
foreach ($this->_ as $k => $v) {
|
| 198 |
+
if (is_array($v)) {
|
| 199 |
+
$string .= "[$k]=>(";
|
| 200 |
+
foreach ($v as $k2 => $v2) {
|
| 201 |
+
$string .= "[$k2]=>\"$v2\", ";
|
| 202 |
+
}
|
| 203 |
+
$string .= ')';
|
| 204 |
+
} else {
|
| 205 |
+
$string .= "[$k]=>\"$v\", ";
|
| 206 |
+
}
|
| 207 |
+
}
|
| 208 |
+
$string .= ')';
|
| 209 |
+
}
|
| 210 |
+
|
| 211 |
+
if (isset($this->text)) {
|
| 212 |
+
$string .= " text: ({$this->text})";
|
| 213 |
+
}
|
| 214 |
+
|
| 215 |
+
$string .= ' HDOM_INNER_INFO: ';
|
| 216 |
+
|
| 217 |
+
if (isset($node->_[HDOM_INFO_INNER])) {
|
| 218 |
+
$string .= "'" . $node->_[HDOM_INFO_INNER] . "'";
|
| 219 |
+
} else {
|
| 220 |
+
$string .= ' NULL ';
|
| 221 |
+
}
|
| 222 |
+
|
| 223 |
+
$string .= ' children: ' . count($this->children);
|
| 224 |
+
$string .= ' nodes: ' . count($this->nodes);
|
| 225 |
+
$string .= ' tag_start: ' . $this->tag_start;
|
| 226 |
+
$string .= "\n";
|
| 227 |
+
|
| 228 |
+
if ($echo) {
|
| 229 |
+
echo $string;
|
| 230 |
+
return;
|
| 231 |
+
} else {
|
| 232 |
+
return $string;
|
| 233 |
+
}
|
| 234 |
+
}
|
| 235 |
+
|
| 236 |
+
function parent($parent = null)
|
| 237 |
+
{
|
| 238 |
+
// I am SURE that this doesn't work properly.
|
| 239 |
+
// It fails to unset the current node from it's current parents nodes or
|
| 240 |
+
// children list first.
|
| 241 |
+
if ($parent !== null) {
|
| 242 |
+
$this->parent = $parent;
|
| 243 |
+
$this->parent->nodes[] = $this;
|
| 244 |
+
$this->parent->children[] = $this;
|
| 245 |
+
}
|
| 246 |
+
|
| 247 |
+
return $this->parent;
|
| 248 |
+
}
|
| 249 |
+
|
| 250 |
+
function has_child()
|
| 251 |
+
{
|
| 252 |
+
return !empty($this->children);
|
| 253 |
+
}
|
| 254 |
+
|
| 255 |
+
function children($idx = -1)
|
| 256 |
+
{
|
| 257 |
+
if ($idx === -1) {
|
| 258 |
+
return $this->children;
|
| 259 |
+
}
|
| 260 |
+
|
| 261 |
+
if (isset($this->children[$idx])) {
|
| 262 |
+
return $this->children[$idx];
|
| 263 |
+
}
|
| 264 |
+
|
| 265 |
+
return null;
|
| 266 |
+
}
|
| 267 |
+
|
| 268 |
+
function first_child()
|
| 269 |
+
{
|
| 270 |
+
if (count($this->children) > 0) {
|
| 271 |
+
return $this->children[0];
|
| 272 |
+
}
|
| 273 |
+
return null;
|
| 274 |
+
}
|
| 275 |
+
|
| 276 |
+
function last_child()
|
| 277 |
+
{
|
| 278 |
+
if (count($this->children) > 0) {
|
| 279 |
+
return end($this->children);
|
| 280 |
+
}
|
| 281 |
+
return null;
|
| 282 |
+
}
|
| 283 |
+
|
| 284 |
+
function next_sibling()
|
| 285 |
+
{
|
| 286 |
+
if ($this->parent === null) {
|
| 287 |
+
return null;
|
| 288 |
+
}
|
| 289 |
+
|
| 290 |
+
$idx = array_search($this, $this->parent->children, true);
|
| 291 |
+
|
| 292 |
+
if ($idx !== false && isset($this->parent->children[$idx + 1])) {
|
| 293 |
+
return $this->parent->children[$idx + 1];
|
| 294 |
+
}
|
| 295 |
+
|
| 296 |
+
return null;
|
| 297 |
+
}
|
| 298 |
+
|
| 299 |
+
function prev_sibling()
|
| 300 |
+
{
|
| 301 |
+
if ($this->parent === null) {
|
| 302 |
+
return null;
|
| 303 |
+
}
|
| 304 |
+
|
| 305 |
+
$idx = array_search($this, $this->parent->children, true);
|
| 306 |
+
|
| 307 |
+
if ($idx !== false && $idx > 0) {
|
| 308 |
+
return $this->parent->children[$idx - 1];
|
| 309 |
+
}
|
| 310 |
+
|
| 311 |
+
return null;
|
| 312 |
+
}
|
| 313 |
+
|
| 314 |
+
function find_ancestor_tag($tag)
|
| 315 |
+
{
|
| 316 |
+
global $debug_object;
|
| 317 |
+
if (is_object($debug_object)) { $debug_object->debug_log_entry(1); }
|
| 318 |
+
|
| 319 |
+
if ($this->parent === null) {
|
| 320 |
+
return null;
|
| 321 |
+
}
|
| 322 |
+
|
| 323 |
+
$ancestor = $this->parent;
|
| 324 |
+
|
| 325 |
+
while (!is_null($ancestor)) {
|
| 326 |
+
if (is_object($debug_object)) {
|
| 327 |
+
$debug_object->debug_log(2, 'Current tag is: ' . $ancestor->tag);
|
| 328 |
+
}
|
| 329 |
+
|
| 330 |
+
if ($ancestor->tag === $tag) {
|
| 331 |
+
break;
|
| 332 |
+
}
|
| 333 |
+
|
| 334 |
+
$ancestor = $ancestor->parent;
|
| 335 |
+
}
|
| 336 |
+
|
| 337 |
+
return $ancestor;
|
| 338 |
+
}
|
| 339 |
+
|
| 340 |
+
function innertext()
|
| 341 |
+
{
|
| 342 |
+
if (isset($this->_[HDOM_INFO_INNER])) {
|
| 343 |
+
return $this->_[HDOM_INFO_INNER];
|
| 344 |
+
}
|
| 345 |
+
|
| 346 |
+
if (isset($this->_[HDOM_INFO_TEXT])) {
|
| 347 |
+
return $this->dom->restore_noise($this->_[HDOM_INFO_TEXT]);
|
| 348 |
+
}
|
| 349 |
+
|
| 350 |
+
$ret = '';
|
| 351 |
+
|
| 352 |
+
foreach ($this->nodes as $n) {
|
| 353 |
+
$ret .= $n->outertext();
|
| 354 |
+
}
|
| 355 |
+
|
| 356 |
+
return $ret;
|
| 357 |
+
}
|
| 358 |
+
|
| 359 |
+
function outertext()
|
| 360 |
+
{
|
| 361 |
+
global $debug_object;
|
| 362 |
+
|
| 363 |
+
if (is_object($debug_object)) {
|
| 364 |
+
$text = '';
|
| 365 |
+
|
| 366 |
+
if ($this->tag === 'text') {
|
| 367 |
+
if (!empty($this->text)) {
|
| 368 |
+
$text = ' with text: ' . $this->text;
|
| 369 |
+
}
|
| 370 |
+
}
|
| 371 |
+
|
| 372 |
+
$debug_object->debug_log(1, 'Innertext of tag: ' . $this->tag . $text);
|
| 373 |
+
}
|
| 374 |
+
|
| 375 |
+
if ($this->tag === 'root') {
|
| 376 |
+
return $this->innertext();
|
| 377 |
+
}
|
| 378 |
+
|
| 379 |
+
// todo: What is the use of this callback? Remove?
|
| 380 |
+
if ($this->dom && $this->dom->callback !== null) {
|
| 381 |
+
call_user_func_array($this->dom->callback, array($this));
|
| 382 |
+
}
|
| 383 |
+
|
| 384 |
+
if (isset($this->_[HDOM_INFO_OUTER])) {
|
| 385 |
+
return $this->_[HDOM_INFO_OUTER];
|
| 386 |
+
}
|
| 387 |
+
|
| 388 |
+
if (isset($this->_[HDOM_INFO_TEXT])) {
|
| 389 |
+
return $this->dom->restore_noise($this->_[HDOM_INFO_TEXT]);
|
| 390 |
+
}
|
| 391 |
+
|
| 392 |
+
$ret = '';
|
| 393 |
+
|
| 394 |
+
if ($this->dom && $this->dom->nodes[$this->_[HDOM_INFO_BEGIN]]) {
|
| 395 |
+
$ret = $this->dom->nodes[$this->_[HDOM_INFO_BEGIN]]->makeup();
|
| 396 |
+
}
|
| 397 |
+
|
| 398 |
+
if (isset($this->_[HDOM_INFO_INNER])) {
|
| 399 |
+
// todo: <br> should either never have HDOM_INFO_INNER or always
|
| 400 |
+
if ($this->tag !== 'br') {
|
| 401 |
+
$ret .= $this->_[HDOM_INFO_INNER];
|
| 402 |
+
}
|
| 403 |
+
} elseif ($this->nodes) {
|
| 404 |
+
foreach ($this->nodes as $n) {
|
| 405 |
+
$ret .= $this->convert_text($n->outertext());
|
| 406 |
+
}
|
| 407 |
+
}
|
| 408 |
+
|
| 409 |
+
if (isset($this->_[HDOM_INFO_END]) && $this->_[HDOM_INFO_END] != 0) {
|
| 410 |
+
$ret .= '</' . $this->tag . '>';
|
| 411 |
+
}
|
| 412 |
+
|
| 413 |
+
return $ret;
|
| 414 |
+
}
|
| 415 |
+
|
| 416 |
+
function text()
|
| 417 |
+
{
|
| 418 |
+
if (isset($this->_[HDOM_INFO_INNER])) {
|
| 419 |
+
return $this->_[HDOM_INFO_INNER];
|
| 420 |
+
}
|
| 421 |
+
|
| 422 |
+
switch ($this->nodetype) {
|
| 423 |
+
case HDOM_TYPE_TEXT: return $this->dom->restore_noise($this->_[HDOM_INFO_TEXT]);
|
| 424 |
+
case HDOM_TYPE_COMMENT: return '';
|
| 425 |
+
case HDOM_TYPE_UNKNOWN: return '';
|
| 426 |
+
}
|
| 427 |
+
|
| 428 |
+
if (strcasecmp($this->tag, 'script') === 0) { return ''; }
|
| 429 |
+
if (strcasecmp($this->tag, 'style') === 0) { return ''; }
|
| 430 |
+
|
| 431 |
+
$ret = '';
|
| 432 |
+
|
| 433 |
+
// In rare cases, (always node type 1 or HDOM_TYPE_ELEMENT - observed
|
| 434 |
+
// for some span tags, and some p tags) $this->nodes is set to NULL.
|
| 435 |
+
// NOTE: This indicates that there is a problem where it's set to NULL
|
| 436 |
+
// without a clear happening.
|
| 437 |
+
// WHY is this happening?
|
| 438 |
+
if (!is_null($this->nodes)) {
|
| 439 |
+
foreach ($this->nodes as $n) {
|
| 440 |
+
// Start paragraph after a blank line
|
| 441 |
+
if ($n->tag === 'p') {
|
| 442 |
+
$ret = trim($ret) . "\n\n";
|
| 443 |
+
}
|
| 444 |
+
|
| 445 |
+
$ret .= $this->convert_text($n->text());
|
| 446 |
+
|
| 447 |
+
// If this node is a span... add a space at the end of it so
|
| 448 |
+
// multiple spans don't run into each other. This is plaintext
|
| 449 |
+
// after all.
|
| 450 |
+
if ($n->tag === 'span') {
|
| 451 |
+
$ret .= $this->dom->default_span_text;
|
| 452 |
+
}
|
| 453 |
+
}
|
| 454 |
+
}
|
| 455 |
+
return $ret;
|
| 456 |
+
}
|
| 457 |
+
|
| 458 |
+
function xmltext()
|
| 459 |
+
{
|
| 460 |
+
$ret = $this->innertext();
|
| 461 |
+
$ret = str_ireplace('<![CDATA[', '', $ret);
|
| 462 |
+
$ret = str_replace(']]>', '', $ret);
|
| 463 |
+
return $ret;
|
| 464 |
+
}
|
| 465 |
+
|
| 466 |
+
function makeup()
|
| 467 |
+
{
|
| 468 |
+
// text, comment, unknown
|
| 469 |
+
if (isset($this->_[HDOM_INFO_TEXT])) {
|
| 470 |
+
return $this->dom->restore_noise($this->_[HDOM_INFO_TEXT]);
|
| 471 |
+
}
|
| 472 |
+
|
| 473 |
+
$ret = '<' . $this->tag;
|
| 474 |
+
$i = -1;
|
| 475 |
+
|
| 476 |
+
foreach ($this->attr as $key => $val) {
|
| 477 |
+
++$i;
|
| 478 |
+
|
| 479 |
+
// skip removed attribute
|
| 480 |
+
if ($val === null || $val === false) { continue; }
|
| 481 |
+
|
| 482 |
+
$ret .= $this->_[HDOM_INFO_SPACE][$i][0];
|
| 483 |
+
|
| 484 |
+
//no value attr: nowrap, checked selected...
|
| 485 |
+
if ($val === true) {
|
| 486 |
+
$ret .= $key;
|
| 487 |
+
} else {
|
| 488 |
+
switch ($this->_[HDOM_INFO_QUOTE][$i])
|
| 489 |
+
{
|
| 490 |
+
case HDOM_QUOTE_DOUBLE: $quote = '"'; break;
|
| 491 |
+
case HDOM_QUOTE_SINGLE: $quote = '\''; break;
|
| 492 |
+
default: $quote = '';
|
| 493 |
+
}
|
| 494 |
+
|
| 495 |
+
$ret .= $key
|
| 496 |
+
. $this->_[HDOM_INFO_SPACE][$i][1]
|
| 497 |
+
. '='
|
| 498 |
+
. $this->_[HDOM_INFO_SPACE][$i][2]
|
| 499 |
+
. $quote
|
| 500 |
+
. $val
|
| 501 |
+
. $quote;
|
| 502 |
+
}
|
| 503 |
+
}
|
| 504 |
+
|
| 505 |
+
$ret = $this->dom->restore_noise($ret);
|
| 506 |
+
return $ret . $this->_[HDOM_INFO_ENDSPACE] . '>';
|
| 507 |
+
}
|
| 508 |
+
|
| 509 |
+
function find($selector, $idx = null, $lowercase = false)
|
| 510 |
+
{
|
| 511 |
+
$selectors = $this->parse_selector($selector);
|
| 512 |
+
if (($count = count($selectors)) === 0) { return array(); }
|
| 513 |
+
$found_keys = array();
|
| 514 |
+
|
| 515 |
+
// find each selector
|
| 516 |
+
for ($c = 0; $c < $count; ++$c) {
|
| 517 |
+
// The change on the below line was documented on the sourceforge
|
| 518 |
+
// code tracker id 2788009
|
| 519 |
+
// used to be: if (($levle=count($selectors[0]))===0) return array();
|
| 520 |
+
if (($levle = count($selectors[$c])) === 0) { return array(); }
|
| 521 |
+
if (!isset($this->_[HDOM_INFO_BEGIN])) { return array(); }
|
| 522 |
+
|
| 523 |
+
$head = array($this->_[HDOM_INFO_BEGIN] => 1);
|
| 524 |
+
$cmd = ' '; // Combinator
|
| 525 |
+
|
| 526 |
+
// handle descendant selectors, no recursive!
|
| 527 |
+
for ($l = 0; $l < $levle; ++$l) {
|
| 528 |
+
$ret = array();
|
| 529 |
+
|
| 530 |
+
foreach ($head as $k => $v) {
|
| 531 |
+
$n = ($k === -1) ? $this->dom->root : $this->dom->nodes[$k];
|
| 532 |
+
//PaperG - Pass this optional parameter on to the seek function.
|
| 533 |
+
$n->seek($selectors[$c][$l], $ret, $cmd, $lowercase);
|
| 534 |
+
}
|
| 535 |
+
|
| 536 |
+
$head = $ret;
|
| 537 |
+
$cmd = $selectors[$c][$l][4]; // Next Combinator
|
| 538 |
+
}
|
| 539 |
+
|
| 540 |
+
foreach ($head as $k => $v) {
|
| 541 |
+
if (!isset($found_keys[$k])) {
|
| 542 |
+
$found_keys[$k] = 1;
|
| 543 |
+
}
|
| 544 |
+
}
|
| 545 |
+
}
|
| 546 |
+
|
| 547 |
+
// sort keys
|
| 548 |
+
ksort($found_keys);
|
| 549 |
+
|
| 550 |
+
$found = array();
|
| 551 |
+
foreach ($found_keys as $k => $v) {
|
| 552 |
+
$found[] = $this->dom->nodes[$k];
|
| 553 |
+
}
|
| 554 |
+
|
| 555 |
+
// return nth-element or array
|
| 556 |
+
if (is_null($idx)) { return $found; }
|
| 557 |
+
elseif ($idx < 0) { $idx = count($found) + $idx; }
|
| 558 |
+
return (isset($found[$idx])) ? $found[$idx] : null;
|
| 559 |
+
}
|
| 560 |
+
|
| 561 |
+
protected function seek($selector, &$ret, $parent_cmd, $lowercase = false)
|
| 562 |
+
{
|
| 563 |
+
global $debug_object;
|
| 564 |
+
if (is_object($debug_object)) { $debug_object->debug_log_entry(1); }
|
| 565 |
+
|
| 566 |
+
list($tag, $id, $class, $attributes, $cmb) = $selector;
|
| 567 |
+
$nodes = array();
|
| 568 |
+
|
| 569 |
+
if ($parent_cmd === ' ') { // Descendant Combinator
|
| 570 |
+
// Find parent closing tag if the current element doesn't have a closing
|
| 571 |
+
// tag (i.e. void element)
|
| 572 |
+
$end = (!empty($this->_[HDOM_INFO_END])) ? $this->_[HDOM_INFO_END] : 0;
|
| 573 |
+
if ($end == 0) {
|
| 574 |
+
$parent = $this->parent;
|
| 575 |
+
while (!isset($parent->_[HDOM_INFO_END]) && $parent !== null) {
|
| 576 |
+
$end -= 1;
|
| 577 |
+
$parent = $parent->parent;
|
| 578 |
+
}
|
| 579 |
+
$end += $parent->_[HDOM_INFO_END];
|
| 580 |
+
}
|
| 581 |
+
|
| 582 |
+
// Get list of target nodes
|
| 583 |
+
$nodes_start = $this->_[HDOM_INFO_BEGIN] + 1;
|
| 584 |
+
$nodes_count = $end - $nodes_start;
|
| 585 |
+
$nodes = array_slice($this->dom->nodes, $nodes_start, $nodes_count, true);
|
| 586 |
+
} elseif ($parent_cmd === '>') { // Child Combinator
|
| 587 |
+
$nodes = $this->children;
|
| 588 |
+
} elseif ($parent_cmd === '+'
|
| 589 |
+
&& $this->parent
|
| 590 |
+
&& in_array($this, $this->parent->children)) { // Next-Sibling Combinator
|
| 591 |
+
$index = array_search($this, $this->parent->children, true) + 1;
|
| 592 |
+
if ($index < count($this->parent->children))
|
| 593 |
+
$nodes[] = $this->parent->children[$index];
|
| 594 |
+
} elseif ($parent_cmd === '~'
|
| 595 |
+
&& $this->parent
|
| 596 |
+
&& in_array($this, $this->parent->children)) { // Subsequent Sibling Combinator
|
| 597 |
+
$index = array_search($this, $this->parent->children, true);
|
| 598 |
+
$nodes = array_slice($this->parent->children, $index);
|
| 599 |
+
}
|
| 600 |
+
|
| 601 |
+
// Go throgh each element starting at this element until the end tag
|
| 602 |
+
// Note: If this element is a void tag, any previous void element is
|
| 603 |
+
// skipped.
|
| 604 |
+
foreach($nodes as $node) {
|
| 605 |
+
$pass = true;
|
| 606 |
+
|
| 607 |
+
// Skip root nodes
|
| 608 |
+
if(!$node->parent) {
|
| 609 |
+
$pass = false;
|
| 610 |
+
}
|
| 611 |
+
|
| 612 |
+
// Handle 'text' selector
|
| 613 |
+
if($pass && $tag === 'text' && $node->tag === 'text') {
|
| 614 |
+
$ret[array_search($node, $this->dom->nodes, true)] = 1;
|
| 615 |
+
unset($node);
|
| 616 |
+
continue;
|
| 617 |
+
}
|
| 618 |
+
|
| 619 |
+
// Skip if node isn't a child node (i.e. text nodes)
|
| 620 |
+
if($pass && !in_array($node, $node->parent->children, true)) {
|
| 621 |
+
$pass = false;
|
| 622 |
+
}
|
| 623 |
+
|
| 624 |
+
// Skip if tag doesn't match
|
| 625 |
+
if ($pass && $tag !== '' && $tag !== $node->tag && $tag !== '*') {
|
| 626 |
+
$pass = false;
|
| 627 |
+
}
|
| 628 |
+
|
| 629 |
+
// Skip if ID doesn't exist
|
| 630 |
+
if ($pass && $id !== '' && !isset($node->attr['id'])) {
|
| 631 |
+
$pass = false;
|
| 632 |
+
}
|
| 633 |
+
|
| 634 |
+
// Check if ID matches
|
| 635 |
+
if ($pass && $id !== '' && isset($node->attr['id'])) {
|
| 636 |
+
// Note: Only consider the first ID (as browsers do)
|
| 637 |
+
$node_id = explode(' ', trim($node->attr['id']))[0];
|
| 638 |
+
|
| 639 |
+
if($id !== $node_id) { $pass = false; }
|
| 640 |
+
}
|
| 641 |
+
|
| 642 |
+
// Check if all class(es) exist
|
| 643 |
+
if ($pass && $class !== '' && is_array($class) && !empty($class)) {
|
| 644 |
+
if (isset($node->attr['class'])) {
|
| 645 |
+
$node_classes = explode(' ', $node->attr['class']);
|
| 646 |
+
|
| 647 |
+
if ($lowercase) {
|
| 648 |
+
$node_classes = array_map('strtolower', $node_classes);
|
| 649 |
+
}
|
| 650 |
+
|
| 651 |
+
foreach($class as $c) {
|
| 652 |
+
if(!in_array($c, $node_classes)) {
|
| 653 |
+
$pass = false;
|
| 654 |
+
break;
|
| 655 |
+
}
|
| 656 |
+
}
|
| 657 |
+
} else {
|
| 658 |
+
$pass = false;
|
| 659 |
+
}
|
| 660 |
+
}
|
| 661 |
+
|
| 662 |
+
// Check attributes
|
| 663 |
+
if ($pass
|
| 664 |
+
&& $attributes !== ''
|
| 665 |
+
&& is_array($attributes)
|
| 666 |
+
&& !empty($attributes)) {
|
| 667 |
+
foreach($attributes as $a) {
|
| 668 |
+
list (
|
| 669 |
+
$att_name,
|
| 670 |
+
$att_expr,
|
| 671 |
+
$att_val,
|
| 672 |
+
$att_inv,
|
| 673 |
+
$att_case_sensitivity
|
| 674 |
+
) = $a;
|
| 675 |
+
|
| 676 |
+
// Handle indexing attributes (i.e. "[2]")
|
| 677 |
+
/**
|
| 678 |
+
* Note: This is not supported by the CSS Standard but adds
|
| 679 |
+
* the ability to select items compatible to XPath (i.e.
|
| 680 |
+
* the 3rd element within it's parent).
|
| 681 |
+
*
|
| 682 |
+
* Note: This doesn't conflict with the CSS Standard which
|
| 683 |
+
* doesn't work on numeric attributes anyway.
|
| 684 |
+
*/
|
| 685 |
+
if (is_numeric($att_name)
|
| 686 |
+
&& $att_expr === ''
|
| 687 |
+
&& $att_val === '') {
|
| 688 |
+
$count = 0;
|
| 689 |
+
|
| 690 |
+
// Find index of current element in parent
|
| 691 |
+
foreach ($node->parent->children as $c) {
|
| 692 |
+
if ($c->tag === $node->tag) ++$count;
|
| 693 |
+
if ($c === $node) break;
|
| 694 |
+
}
|
| 695 |
+
|
| 696 |
+
// If this is the correct node, continue with next
|
| 697 |
+
// attribute
|
| 698 |
+
if ($count === (int)$att_name) continue;
|
| 699 |
+
}
|
| 700 |
+
|
| 701 |
+
// Check attribute availability
|
| 702 |
+
if ($att_inv) { // Attribute should NOT be set
|
| 703 |
+
if (isset($node->attr[$att_name])) {
|
| 704 |
+
$pass = false;
|
| 705 |
+
break;
|
| 706 |
+
}
|
| 707 |
+
} else { // Attribute should be set
|
| 708 |
+
// todo: "plaintext" is not a valid CSS selector!
|
| 709 |
+
if ($att_name !== 'plaintext'
|
| 710 |
+
&& !isset($node->attr[$att_name])) {
|
| 711 |
+
$pass = false;
|
| 712 |
+
break;
|
| 713 |
+
}
|
| 714 |
+
}
|
| 715 |
+
|
| 716 |
+
// Continue with next attribute if expression isn't defined
|
| 717 |
+
if ($att_expr === '') continue;
|
| 718 |
+
|
| 719 |
+
// If they have told us that this is a "plaintext"
|
| 720 |
+
// search then we want the plaintext of the node - right?
|
| 721 |
+
// todo "plaintext" is not a valid CSS selector!
|
| 722 |
+
if ($att_name === 'plaintext') {
|
| 723 |
+
$nodeKeyValue = $node->text();
|
| 724 |
+
} else {
|
| 725 |
+
$nodeKeyValue = $node->attr[$att_name];
|
| 726 |
+
}
|
| 727 |
+
|
| 728 |
+
if (is_object($debug_object)) {
|
| 729 |
+
$debug_object->debug_log(2,
|
| 730 |
+
'testing node: '
|
| 731 |
+
. $node->tag
|
| 732 |
+
. ' for attribute: '
|
| 733 |
+
. $att_name
|
| 734 |
+
. $att_expr
|
| 735 |
+
. $att_val
|
| 736 |
+
. ' where nodes value is: '
|
| 737 |
+
. $nodeKeyValue
|
| 738 |
+
);
|
| 739 |
+
}
|
| 740 |
+
|
| 741 |
+
// If lowercase is set, do a case insensitive test of
|
| 742 |
+
// the value of the selector.
|
| 743 |
+
if ($lowercase) {
|
| 744 |
+
$check = $this->match(
|
| 745 |
+
$att_expr,
|
| 746 |
+
strtolower($att_val),
|
| 747 |
+
strtolower($nodeKeyValue),
|
| 748 |
+
$att_case_sensitivity
|
| 749 |
+
);
|
| 750 |
+
} else {
|
| 751 |
+
$check = $this->match(
|
| 752 |
+
$att_expr,
|
| 753 |
+
$att_val,
|
| 754 |
+
$nodeKeyValue,
|
| 755 |
+
$att_case_sensitivity
|
| 756 |
+
);
|
| 757 |
+
}
|
| 758 |
+
|
| 759 |
+
if (is_object($debug_object)) {
|
| 760 |
+
$debug_object->debug_log(2,
|
| 761 |
+
'after match: '
|
| 762 |
+
. ($check ? 'true' : 'false')
|
| 763 |
+
);
|
| 764 |
+
}
|
| 765 |
+
|
| 766 |
+
if (!$check) {
|
| 767 |
+
$pass = false;
|
| 768 |
+
break;
|
| 769 |
+
}
|
| 770 |
+
}
|
| 771 |
+
}
|
| 772 |
+
|
| 773 |
+
// Found a match. Add to list and clear node
|
| 774 |
+
if ($pass) $ret[$node->_[HDOM_INFO_BEGIN]] = 1;
|
| 775 |
+
unset($node);
|
| 776 |
+
}
|
| 777 |
+
// It's passed by reference so this is actually what this function returns.
|
| 778 |
+
if (is_object($debug_object)) {
|
| 779 |
+
$debug_object->debug_log(1, 'EXIT - ret: ', $ret);
|
| 780 |
+
}
|
| 781 |
+
}
|
| 782 |
+
|
| 783 |
+
protected function match($exp, $pattern, $value, $case_sensitivity)
|
| 784 |
+
{
|
| 785 |
+
global $debug_object;
|
| 786 |
+
if (is_object($debug_object)) {$debug_object->debug_log_entry(1);}
|
| 787 |
+
|
| 788 |
+
if ($case_sensitivity === 'i') {
|
| 789 |
+
$pattern = strtolower($pattern);
|
| 790 |
+
$value = strtolower($value);
|
| 791 |
+
}
|
| 792 |
+
|
| 793 |
+
switch ($exp) {
|
| 794 |
+
case '=':
|
| 795 |
+
return ($value === $pattern);
|
| 796 |
+
case '!=':
|
| 797 |
+
return ($value !== $pattern);
|
| 798 |
+
case '^=':
|
| 799 |
+
return preg_match('/^' . preg_quote($pattern, '/') . '/', $value);
|
| 800 |
+
case '$=':
|
| 801 |
+
return preg_match('/' . preg_quote($pattern, '/') . '$/', $value);
|
| 802 |
+
case '*=':
|
| 803 |
+
return preg_match('/' . preg_quote($pattern, '/') . '/', $value);
|
| 804 |
+
case '|=':
|
| 805 |
+
/**
|
| 806 |
+
* [att|=val]
|
| 807 |
+
*
|
| 808 |
+
* Represents an element with the att attribute, its value
|
| 809 |
+
* either being exactly "val" or beginning with "val"
|
| 810 |
+
* immediately followed by "-" (U+002D).
|
| 811 |
+
*/
|
| 812 |
+
return strpos($value, $pattern) === 0;
|
| 813 |
+
case '~=':
|
| 814 |
+
/**
|
| 815 |
+
* [att~=val]
|
| 816 |
+
*
|
| 817 |
+
* Represents an element with the att attribute whose value is a
|
| 818 |
+
* whitespace-separated list of words, one of which is exactly
|
| 819 |
+
* "val". If "val" contains whitespace, it will never represent
|
| 820 |
+
* anything (since the words are separated by spaces). Also if
|
| 821 |
+
* "val" is the empty string, it will never represent anything.
|
| 822 |
+
*/
|
| 823 |
+
return in_array($pattern, explode(' ', trim($value)), true);
|
| 824 |
+
}
|
| 825 |
+
return false;
|
| 826 |
+
}
|
| 827 |
+
|
| 828 |
+
protected function parse_selector($selector_string)
|
| 829 |
+
{
|
| 830 |
+
global $debug_object;
|
| 831 |
+
if (is_object($debug_object)) { $debug_object->debug_log_entry(1); }
|
| 832 |
+
|
| 833 |
+
/**
|
| 834 |
+
* Pattern of CSS selectors, modified from mootools (https://mootools.net/)
|
| 835 |
+
*
|
| 836 |
+
* Paperg: Add the colon to the attribute, so that it properly finds
|
| 837 |
+
* <tag attr:ibute="something" > like google does.
|
| 838 |
+
*
|
| 839 |
+
* Note: if you try to look at this attribute, you MUST use getAttribute
|
| 840 |
+
* since $dom->x:y will fail the php syntax check.
|
| 841 |
+
*
|
| 842 |
+
* Notice the \[ starting the attribute? and the @? following? This
|
| 843 |
+
* implies that an attribute can begin with an @ sign that is not
|
| 844 |
+
* captured. This implies that an html attribute specifier may start
|
| 845 |
+
* with an @ sign that is NOT captured by the expression. Farther study
|
| 846 |
+
* is required to determine of this should be documented or removed.
|
| 847 |
+
*
|
| 848 |
+
* Matches selectors in this order:
|
| 849 |
+
*
|
| 850 |
+
* [0] - full match
|
| 851 |
+
*
|
| 852 |
+
* [1] - tag name
|
| 853 |
+
* ([\w:\*-]*)
|
| 854 |
+
* Matches the tag name consisting of zero or more words, colons,
|
| 855 |
+
* asterisks and hyphens.
|
| 856 |
+
*
|
| 857 |
+
* [2] - id name
|
| 858 |
+
* (?:\#([\w-]+))
|
| 859 |
+
* Optionally matches a id name, consisting of an "#" followed by
|
| 860 |
+
* the id name (one or more words and hyphens).
|
| 861 |
+
*
|
| 862 |
+
* [3] - class names (including dots)
|
| 863 |
+
* (?:\.([\w\.-]+))?
|
| 864 |
+
* Optionally matches a list of classs, consisting of an "."
|
| 865 |
+
* followed by the class name (one or more words and hyphens)
|
| 866 |
+
* where multiple classes can be chained (i.e. ".foo.bar.baz")
|
| 867 |
+
*
|
| 868 |
+
* [4] - attributes
|
| 869 |
+
* ((?:\[@?(?:!?[\w:-]+)(?:(?:[!*^$|~]?=)[\"']?(?:.*?)[\"']?)?(?:\s*?(?:[iIsS])?)?\])+)?
|
| 870 |
+
* Optionally matches the attributes list
|
| 871 |
+
*
|
| 872 |
+
* [5] - separator
|
| 873 |
+
* ([\/, >+~]+)
|
| 874 |
+
* Matches the selector list separator
|
| 875 |
+
*/
|
| 876 |
+
// phpcs:ignore Generic.Files.LineLength
|
| 877 |
+
$pattern = "/([\w:\*-]*)(?:\#([\w-]+))?(?:|\.([\w\.-]+))?((?:\[@?(?:!?[\w:-]+)(?:(?:[!*^$|~]?=)[\"']?(?:.*?)[\"']?)?(?:\s*?(?:[iIsS])?)?\])+)?([\/, >+~]+)/is";
|
| 878 |
+
|
| 879 |
+
preg_match_all(
|
| 880 |
+
$pattern,
|
| 881 |
+
trim($selector_string) . ' ', // Add final ' ' as pseudo separator
|
| 882 |
+
$matches,
|
| 883 |
+
PREG_SET_ORDER
|
| 884 |
+
);
|
| 885 |
+
|
| 886 |
+
if (is_object($debug_object)) {
|
| 887 |
+
$debug_object->debug_log(2, 'Matches Array: ', $matches);
|
| 888 |
+
}
|
| 889 |
+
|
| 890 |
+
$selectors = array();
|
| 891 |
+
$result = array();
|
| 892 |
+
|
| 893 |
+
foreach ($matches as $m) {
|
| 894 |
+
$m[0] = trim($m[0]);
|
| 895 |
+
|
| 896 |
+
// Skip NoOps
|
| 897 |
+
if ($m[0] === '' || $m[0] === '/' || $m[0] === '//') { continue; }
|
| 898 |
+
|
| 899 |
+
// Convert to lowercase
|
| 900 |
+
if ($this->dom->lowercase) {
|
| 901 |
+
$m[1] = strtolower($m[1]);
|
| 902 |
+
}
|
| 903 |
+
|
| 904 |
+
// Extract classes
|
| 905 |
+
if ($m[3] !== '') { $m[3] = explode('.', $m[3]); }
|
| 906 |
+
|
| 907 |
+
/* Extract attributes (pattern based on the pattern above!)
|
| 908 |
+
|
| 909 |
+
* [0] - full match
|
| 910 |
+
* [1] - attribute name
|
| 911 |
+
* [2] - attribute expression
|
| 912 |
+
* [3] - attribute value
|
| 913 |
+
* [4] - case sensitivity
|
| 914 |
+
*
|
| 915 |
+
* Note: Attributes can be negated with a "!" prefix to their name
|
| 916 |
+
*/
|
| 917 |
+
if($m[4] !== '') {
|
| 918 |
+
preg_match_all(
|
| 919 |
+
"/\[@?(!?[\w:-]+)(?:([!*^$|~]?=)[\"']?(.*?)[\"']?)?(?:\s+?([iIsS])?)?\]/is",
|
| 920 |
+
trim($m[4]),
|
| 921 |
+
$attributes,
|
| 922 |
+
PREG_SET_ORDER
|
| 923 |
+
);
|
| 924 |
+
|
| 925 |
+
// Replace element by array
|
| 926 |
+
$m[4] = array();
|
| 927 |
+
|
| 928 |
+
foreach($attributes as $att) {
|
| 929 |
+
// Skip empty matches
|
| 930 |
+
if(trim($att[0]) === '') { continue; }
|
| 931 |
+
|
| 932 |
+
$inverted = (isset($att[1][0]) && $att[1][0] === '!');
|
| 933 |
+
$m[4][] = array(
|
| 934 |
+
$inverted ? substr($att[1], 1) : $att[1], // Name
|
| 935 |
+
(isset($att[2])) ? $att[2] : '', // Expression
|
| 936 |
+
(isset($att[3])) ? $att[3] : '', // Value
|
| 937 |
+
$inverted, // Inverted Flag
|
| 938 |
+
(isset($att[4])) ? strtolower($att[4]) : '', // Case-Sensitivity
|
| 939 |
+
);
|
| 940 |
+
}
|
| 941 |
+
}
|
| 942 |
+
|
| 943 |
+
// Sanitize Separator
|
| 944 |
+
if ($m[5] !== '' && trim($m[5]) === '') { // Descendant Separator
|
| 945 |
+
$m[5] = ' ';
|
| 946 |
+
} else { // Other Separator
|
| 947 |
+
$m[5] = trim($m[5]);
|
| 948 |
+
}
|
| 949 |
+
|
| 950 |
+
// Clear Separator if it's a Selector List
|
| 951 |
+
if ($is_list = ($m[5] === ',')) { $m[5] = ''; }
|
| 952 |
+
|
| 953 |
+
// Remove full match before adding to results
|
| 954 |
+
array_shift($m);
|
| 955 |
+
$result[] = $m;
|
| 956 |
+
|
| 957 |
+
if ($is_list) { // Selector List
|
| 958 |
+
$selectors[] = $result;
|
| 959 |
+
$result = array();
|
| 960 |
+
}
|
| 961 |
+
}
|
| 962 |
+
|
| 963 |
+
if (count($result) > 0) { $selectors[] = $result; }
|
| 964 |
+
return $selectors;
|
| 965 |
+
}
|
| 966 |
+
|
| 967 |
+
function __get($name)
|
| 968 |
+
{
|
| 969 |
+
if (isset($this->attr[$name])) {
|
| 970 |
+
return $this->convert_text($this->attr[$name]);
|
| 971 |
+
}
|
| 972 |
+
switch ($name) {
|
| 973 |
+
case 'outertext': return $this->outertext();
|
| 974 |
+
case 'innertext': return $this->innertext();
|
| 975 |
+
case 'plaintext': return $this->text();
|
| 976 |
+
case 'xmltext': return $this->xmltext();
|
| 977 |
+
default: return array_key_exists($name, $this->attr);
|
| 978 |
+
}
|
| 979 |
+
}
|
| 980 |
+
|
| 981 |
+
function __set($name, $value)
|
| 982 |
+
{
|
| 983 |
+
global $debug_object;
|
| 984 |
+
if (is_object($debug_object)) { $debug_object->debug_log_entry(1); }
|
| 985 |
+
|
| 986 |
+
switch ($name) {
|
| 987 |
+
case 'outertext': return $this->_[HDOM_INFO_OUTER] = $value;
|
| 988 |
+
case 'innertext':
|
| 989 |
+
if (isset($this->_[HDOM_INFO_TEXT])) {
|
| 990 |
+
return $this->_[HDOM_INFO_TEXT] = $value;
|
| 991 |
+
}
|
| 992 |
+
return $this->_[HDOM_INFO_INNER] = $value;
|
| 993 |
+
}
|
| 994 |
+
|
| 995 |
+
if (!isset($this->attr[$name])) {
|
| 996 |
+
$this->_[HDOM_INFO_SPACE][] = array(' ', '', '');
|
| 997 |
+
$this->_[HDOM_INFO_QUOTE][] = HDOM_QUOTE_DOUBLE;
|
| 998 |
+
}
|
| 999 |
+
|
| 1000 |
+
$this->attr[$name] = $value;
|
| 1001 |
+
}
|
| 1002 |
+
|
| 1003 |
+
function __isset($name)
|
| 1004 |
+
{
|
| 1005 |
+
switch ($name) {
|
| 1006 |
+
case 'outertext': return true;
|
| 1007 |
+
case 'innertext': return true;
|
| 1008 |
+
case 'plaintext': return true;
|
| 1009 |
+
}
|
| 1010 |
+
//no value attr: nowrap, checked selected...
|
| 1011 |
+
return (array_key_exists($name, $this->attr)) ? true : isset($this->attr[$name]);
|
| 1012 |
+
}
|
| 1013 |
+
|
| 1014 |
+
function __unset($name)
|
| 1015 |
+
{
|
| 1016 |
+
if (isset($this->attr[$name])) { unset($this->attr[$name]); }
|
| 1017 |
+
}
|
| 1018 |
+
|
| 1019 |
+
function convert_text($text)
|
| 1020 |
+
{
|
| 1021 |
+
global $debug_object;
|
| 1022 |
+
if (is_object($debug_object)) { $debug_object->debug_log_entry(1); }
|
| 1023 |
+
|
| 1024 |
+
$converted_text = $text;
|
| 1025 |
+
|
| 1026 |
+
$sourceCharset = '';
|
| 1027 |
+
$targetCharset = '';
|
| 1028 |
+
|
| 1029 |
+
if ($this->dom) {
|
| 1030 |
+
$sourceCharset = strtoupper($this->dom->_charset);
|
| 1031 |
+
$targetCharset = strtoupper($this->dom->_target_charset);
|
| 1032 |
+
}
|
| 1033 |
+
|
| 1034 |
+
if (is_object($debug_object)) {
|
| 1035 |
+
$debug_object->debug_log(3,
|
| 1036 |
+
'source charset: '
|
| 1037 |
+
. $sourceCharset
|
| 1038 |
+
. ' target charaset: '
|
| 1039 |
+
. $targetCharset
|
| 1040 |
+
);
|
| 1041 |
+
}
|
| 1042 |
+
|
| 1043 |
+
if (!empty($sourceCharset)
|
| 1044 |
+
&& !empty($targetCharset)
|
| 1045 |
+
&& (strcasecmp($sourceCharset, $targetCharset) != 0)) {
|
| 1046 |
+
// Check if the reported encoding could have been incorrect and the text is actually already UTF-8
|
| 1047 |
+
if ((strcasecmp($targetCharset, 'UTF-8') == 0)
|
| 1048 |
+
&& ($this->is_utf8($text))) {
|
| 1049 |
+
$converted_text = $text;
|
| 1050 |
+
} else {
|
| 1051 |
+
$converted_text = iconv($sourceCharset, $targetCharset, $text);
|
| 1052 |
+
}
|
| 1053 |
+
}
|
| 1054 |
+
|
| 1055 |
+
// Lets make sure that we don't have that silly BOM issue with any of the utf-8 text we output.
|
| 1056 |
+
if ($targetCharset === 'UTF-8') {
|
| 1057 |
+
if (substr($converted_text, 0, 3) === "\xef\xbb\xbf") {
|
| 1058 |
+
$converted_text = substr($converted_text, 3);
|
| 1059 |
+
}
|
| 1060 |
+
|
| 1061 |
+
if (substr($converted_text, -3) === "\xef\xbb\xbf") {
|
| 1062 |
+
$converted_text = substr($converted_text, 0, -3);
|
| 1063 |
+
}
|
| 1064 |
+
}
|
| 1065 |
+
|
| 1066 |
+
return $converted_text;
|
| 1067 |
+
}
|
| 1068 |
+
|
| 1069 |
+
static function is_utf8($str)
|
| 1070 |
+
{
|
| 1071 |
+
$c = 0; $b = 0;
|
| 1072 |
+
$bits = 0;
|
| 1073 |
+
$len = strlen($str);
|
| 1074 |
+
for($i = 0; $i < $len; $i++) {
|
| 1075 |
+
$c = ord($str[$i]);
|
| 1076 |
+
if($c > 128) {
|
| 1077 |
+
if(($c >= 254)) { return false; }
|
| 1078 |
+
elseif($c >= 252) { $bits = 6; }
|
| 1079 |
+
elseif($c >= 248) { $bits = 5; }
|
| 1080 |
+
elseif($c >= 240) { $bits = 4; }
|
| 1081 |
+
elseif($c >= 224) { $bits = 3; }
|
| 1082 |
+
elseif($c >= 192) { $bits = 2; }
|
| 1083 |
+
else { return false; }
|
| 1084 |
+
if(($i + $bits) > $len) { return false; }
|
| 1085 |
+
while($bits > 1) {
|
| 1086 |
+
$i++;
|
| 1087 |
+
$b = ord($str[$i]);
|
| 1088 |
+
if($b < 128 || $b > 191) { return false; }
|
| 1089 |
+
$bits--;
|
| 1090 |
+
}
|
| 1091 |
+
}
|
| 1092 |
+
}
|
| 1093 |
+
return true;
|
| 1094 |
+
}
|
| 1095 |
+
|
| 1096 |
+
function get_display_size()
|
| 1097 |
+
{
|
| 1098 |
+
global $debug_object;
|
| 1099 |
+
|
| 1100 |
+
$width = -1;
|
| 1101 |
+
$height = -1;
|
| 1102 |
+
|
| 1103 |
+
if ($this->tag !== 'img') {
|
| 1104 |
+
return false;
|
| 1105 |
+
}
|
| 1106 |
+
|
| 1107 |
+
// See if there is aheight or width attribute in the tag itself.
|
| 1108 |
+
if (isset($this->attr['width'])) {
|
| 1109 |
+
$width = $this->attr['width'];
|
| 1110 |
+
}
|
| 1111 |
+
|
| 1112 |
+
if (isset($this->attr['height'])) {
|
| 1113 |
+
$height = $this->attr['height'];
|
| 1114 |
+
}
|
| 1115 |
+
|
| 1116 |
+
// Now look for an inline style.
|
| 1117 |
+
if (isset($this->attr['style'])) {
|
| 1118 |
+
// Thanks to user gnarf from stackoverflow for this regular expression.
|
| 1119 |
+
$attributes = array();
|
| 1120 |
+
|
| 1121 |
+
preg_match_all(
|
| 1122 |
+
'/([\w-]+)\s*:\s*([^;]+)\s*;?/',
|
| 1123 |
+
$this->attr['style'],
|
| 1124 |
+
$matches,
|
| 1125 |
+
PREG_SET_ORDER
|
| 1126 |
+
);
|
| 1127 |
+
|
| 1128 |
+
foreach ($matches as $match) {
|
| 1129 |
+
$attributes[$match[1]] = $match[2];
|
| 1130 |
+
}
|
| 1131 |
+
|
| 1132 |
+
// If there is a width in the style attributes:
|
| 1133 |
+
if (isset($attributes['width']) && $width == -1) {
|
| 1134 |
+
// check that the last two characters are px (pixels)
|
| 1135 |
+
if (strtolower(substr($attributes['width'], -2)) === 'px') {
|
| 1136 |
+
$proposed_width = substr($attributes['width'], 0, -2);
|
| 1137 |
+
// Now make sure that it's an integer and not something stupid.
|
| 1138 |
+
if (filter_var($proposed_width, FILTER_VALIDATE_INT)) {
|
| 1139 |
+
$width = $proposed_width;
|
| 1140 |
+
}
|
| 1141 |
+
}
|
| 1142 |
+
}
|
| 1143 |
+
|
| 1144 |
+
// If there is a width in the style attributes:
|
| 1145 |
+
if (isset($attributes['height']) && $height == -1) {
|
| 1146 |
+
// check that the last two characters are px (pixels)
|
| 1147 |
+
if (strtolower(substr($attributes['height'], -2)) == 'px') {
|
| 1148 |
+
$proposed_height = substr($attributes['height'], 0, -2);
|
| 1149 |
+
// Now make sure that it's an integer and not something stupid.
|
| 1150 |
+
if (filter_var($proposed_height, FILTER_VALIDATE_INT)) {
|
| 1151 |
+
$height = $proposed_height;
|
| 1152 |
+
}
|
| 1153 |
+
}
|
| 1154 |
+
}
|
| 1155 |
+
|
| 1156 |
+
}
|
| 1157 |
+
|
| 1158 |
+
// Future enhancement:
|
| 1159 |
+
// Look in the tag to see if there is a class or id specified that has
|
| 1160 |
+
// a height or width attribute to it.
|
| 1161 |
+
|
| 1162 |
+
// Far future enhancement
|
| 1163 |
+
// Look at all the parent tags of this image to see if they specify a
|
| 1164 |
+
// class or id that has an img selector that specifies a height or width
|
| 1165 |
+
// Note that in this case, the class or id will have the img subselector
|
| 1166 |
+
// for it to apply to the image.
|
| 1167 |
+
|
| 1168 |
+
// ridiculously far future development
|
| 1169 |
+
// If the class or id is specified in a SEPARATE css file thats not on
|
| 1170 |
+
// the page, go get it and do what we were just doing for the ones on
|
| 1171 |
+
// the page.
|
| 1172 |
+
|
| 1173 |
+
$result = array(
|
| 1174 |
+
'height' => $height,
|
| 1175 |
+
'width' => $width
|
| 1176 |
+
);
|
| 1177 |
+
|
| 1178 |
+
return $result;
|
| 1179 |
+
}
|
| 1180 |
+
|
| 1181 |
+
function save($filepath = '')
|
| 1182 |
+
{
|
| 1183 |
+
$ret = $this->outertext();
|
| 1184 |
+
|
| 1185 |
+
if ($filepath !== '') {
|
| 1186 |
+
file_put_contents($filepath, $ret, LOCK_EX);
|
| 1187 |
+
}
|
| 1188 |
+
|
| 1189 |
+
return $ret;
|
| 1190 |
+
}
|
| 1191 |
+
|
| 1192 |
+
function addClass($class)
|
| 1193 |
+
{
|
| 1194 |
+
if (is_string($class)) {
|
| 1195 |
+
$class = explode(' ', $class);
|
| 1196 |
+
}
|
| 1197 |
+
|
| 1198 |
+
if (is_array($class)) {
|
| 1199 |
+
foreach($class as $c) {
|
| 1200 |
+
if (isset($this->class)) {
|
| 1201 |
+
if ($this->hasClass($c)) {
|
| 1202 |
+
continue;
|
| 1203 |
+
} else {
|
| 1204 |
+
$this->class .= ' ' . $c;
|
| 1205 |
+
}
|
| 1206 |
+
} else {
|
| 1207 |
+
$this->class = $c;
|
| 1208 |
+
}
|
| 1209 |
+
}
|
| 1210 |
+
} else {
|
| 1211 |
+
if (is_object($debug_object)) {
|
| 1212 |
+
$debug_object->debug_log(2, 'Invalid type: ', gettype($class));
|
| 1213 |
+
}
|
| 1214 |
+
}
|
| 1215 |
+
}
|
| 1216 |
+
|
| 1217 |
+
function hasClass($class)
|
| 1218 |
+
{
|
| 1219 |
+
if (is_string($class)) {
|
| 1220 |
+
if (isset($this->class)) {
|
| 1221 |
+
return in_array($class, explode(' ', $this->class), true);
|
| 1222 |
+
}
|
| 1223 |
+
} else {
|
| 1224 |
+
if (is_object($debug_object)) {
|
| 1225 |
+
$debug_object->debug_log(2, 'Invalid type: ', gettype($class));
|
| 1226 |
+
}
|
| 1227 |
+
}
|
| 1228 |
+
|
| 1229 |
+
return false;
|
| 1230 |
+
}
|
| 1231 |
+
|
| 1232 |
+
function removeClass($class = null)
|
| 1233 |
+
{
|
| 1234 |
+
if (!isset($this->class)) {
|
| 1235 |
+
return;
|
| 1236 |
+
}
|
| 1237 |
+
|
| 1238 |
+
if (is_null($class)) {
|
| 1239 |
+
$this->removeAttribute('class');
|
| 1240 |
+
return;
|
| 1241 |
+
}
|
| 1242 |
+
|
| 1243 |
+
if (is_string($class)) {
|
| 1244 |
+
$class = explode(' ', $class);
|
| 1245 |
+
}
|
| 1246 |
+
|
| 1247 |
+
if (is_array($class)) {
|
| 1248 |
+
$class = array_diff(explode(' ', $this->class), $class);
|
| 1249 |
+
if (empty($class)) {
|
| 1250 |
+
$this->removeAttribute('class');
|
| 1251 |
+
} else {
|
| 1252 |
+
$this->class = implode(' ', $class);
|
| 1253 |
+
}
|
| 1254 |
+
}
|
| 1255 |
+
}
|
| 1256 |
+
|
| 1257 |
+
function getAllAttributes()
|
| 1258 |
+
{
|
| 1259 |
+
return $this->attr;
|
| 1260 |
+
}
|
| 1261 |
+
|
| 1262 |
+
function getAttribute($name)
|
| 1263 |
+
{
|
| 1264 |
+
return $this->__get($name);
|
| 1265 |
+
}
|
| 1266 |
+
|
| 1267 |
+
function setAttribute($name, $value)
|
| 1268 |
+
{
|
| 1269 |
+
$this->__set($name, $value);
|
| 1270 |
+
}
|
| 1271 |
+
|
| 1272 |
+
function hasAttribute($name)
|
| 1273 |
+
{
|
| 1274 |
+
return $this->__isset($name);
|
| 1275 |
+
}
|
| 1276 |
+
|
| 1277 |
+
function removeAttribute($name)
|
| 1278 |
+
{
|
| 1279 |
+
$this->__set($name, null);
|
| 1280 |
+
}
|
| 1281 |
+
|
| 1282 |
+
function remove()
|
| 1283 |
+
{
|
| 1284 |
+
if ($this->parent) {
|
| 1285 |
+
$this->parent->removeChild($this);
|
| 1286 |
+
}
|
| 1287 |
+
}
|
| 1288 |
+
|
| 1289 |
+
function removeChild($node)
|
| 1290 |
+
{
|
| 1291 |
+
$nidx = array_search($node, $this->nodes, true);
|
| 1292 |
+
$cidx = array_search($node, $this->children, true);
|
| 1293 |
+
$didx = array_search($node, $this->dom->nodes, true);
|
| 1294 |
+
|
| 1295 |
+
if ($nidx !== false && $cidx !== false && $didx !== false) {
|
| 1296 |
+
|
| 1297 |
+
foreach($node->children as $child) {
|
| 1298 |
+
$node->removeChild($child);
|
| 1299 |
+
}
|
| 1300 |
+
|
| 1301 |
+
foreach($node->nodes as $entity) {
|
| 1302 |
+
$enidx = array_search($entity, $node->nodes, true);
|
| 1303 |
+
$edidx = array_search($entity, $node->dom->nodes, true);
|
| 1304 |
+
|
| 1305 |
+
if ($enidx !== false && $edidx !== false) {
|
| 1306 |
+
unset($node->nodes[$enidx]);
|
| 1307 |
+
unset($node->dom->nodes[$edidx]);
|
| 1308 |
+
}
|
| 1309 |
+
}
|
| 1310 |
+
|
| 1311 |
+
unset($this->nodes[$nidx]);
|
| 1312 |
+
unset($this->children[$cidx]);
|
| 1313 |
+
unset($this->dom->nodes[$didx]);
|
| 1314 |
+
|
| 1315 |
+
$node->clear();
|
| 1316 |
+
|
| 1317 |
+
}
|
| 1318 |
+
}
|
| 1319 |
+
|
| 1320 |
+
function getElementById($id)
|
| 1321 |
+
{
|
| 1322 |
+
return $this->find("#$id", 0);
|
| 1323 |
+
}
|
| 1324 |
+
|
| 1325 |
+
function getElementsById($id, $idx = null)
|
| 1326 |
+
{
|
| 1327 |
+
return $this->find("#$id", $idx);
|
| 1328 |
+
}
|
| 1329 |
+
|
| 1330 |
+
function getElementByTagName($name)
|
| 1331 |
+
{
|
| 1332 |
+
return $this->find($name, 0);
|
| 1333 |
+
}
|
| 1334 |
+
|
| 1335 |
+
function getElementsByTagName($name, $idx = null)
|
| 1336 |
+
{
|
| 1337 |
+
return $this->find($name, $idx);
|
| 1338 |
+
}
|
| 1339 |
+
|
| 1340 |
+
function parentNode()
|
| 1341 |
+
{
|
| 1342 |
+
return $this->parent();
|
| 1343 |
+
}
|
| 1344 |
+
|
| 1345 |
+
function childNodes($idx = -1)
|
| 1346 |
+
{
|
| 1347 |
+
return $this->children($idx);
|
| 1348 |
+
}
|
| 1349 |
+
|
| 1350 |
+
function firstChild()
|
| 1351 |
+
{
|
| 1352 |
+
return $this->first_child();
|
| 1353 |
+
}
|
| 1354 |
+
|
| 1355 |
+
function lastChild()
|
| 1356 |
+
{
|
| 1357 |
+
return $this->last_child();
|
| 1358 |
+
}
|
| 1359 |
+
|
| 1360 |
+
function nextSibling()
|
| 1361 |
+
{
|
| 1362 |
+
return $this->next_sibling();
|
| 1363 |
+
}
|
| 1364 |
+
|
| 1365 |
+
function previousSibling()
|
| 1366 |
+
{
|
| 1367 |
+
return $this->prev_sibling();
|
| 1368 |
+
}
|
| 1369 |
+
|
| 1370 |
+
function hasChildNodes()
|
| 1371 |
+
{
|
| 1372 |
+
return $this->has_child();
|
| 1373 |
+
}
|
| 1374 |
+
|
| 1375 |
+
function nodeName()
|
| 1376 |
+
{
|
| 1377 |
+
return $this->tag;
|
| 1378 |
+
}
|
| 1379 |
+
|
| 1380 |
+
function appendChild($node)
|
| 1381 |
+
{
|
| 1382 |
+
$node->parent($this);
|
| 1383 |
+
return $node;
|
| 1384 |
+
}
|
| 1385 |
+
|
| 1386 |
+
}
|
| 1387 |
+
|
| 1388 |
+
class simple_html_dom
|
| 1389 |
+
{
|
| 1390 |
+
public $root = null;
|
| 1391 |
+
public $nodes = array();
|
| 1392 |
+
public $callback = null;
|
| 1393 |
+
public $lowercase = false;
|
| 1394 |
+
public $original_size;
|
| 1395 |
+
public $size;
|
| 1396 |
+
|
| 1397 |
+
protected $pos;
|
| 1398 |
+
protected $doc;
|
| 1399 |
+
protected $char;
|
| 1400 |
+
|
| 1401 |
+
protected $cursor;
|
| 1402 |
+
protected $parent;
|
| 1403 |
+
protected $noise = array();
|
| 1404 |
+
protected $token_blank = " \t\r\n";
|
| 1405 |
+
protected $token_equal = ' =/>';
|
| 1406 |
+
protected $token_slash = " />\r\n\t";
|
| 1407 |
+
protected $token_attr = ' >';
|
| 1408 |
+
|
| 1409 |
+
public $_charset = '';
|
| 1410 |
+
public $_target_charset = '';
|
| 1411 |
+
|
| 1412 |
+
protected $default_br_text = '';
|
| 1413 |
+
|
| 1414 |
+
public $default_span_text = '';
|
| 1415 |
+
|
| 1416 |
+
protected $self_closing_tags = array(
|
| 1417 |
+
'area' => 1,
|
| 1418 |
+
'base' => 1,
|
| 1419 |
+
'br' => 1,
|
| 1420 |
+
'col' => 1,
|
| 1421 |
+
'embed' => 1,
|
| 1422 |
+
'hr' => 1,
|
| 1423 |
+
'img' => 1,
|
| 1424 |
+
'input' => 1,
|
| 1425 |
+
'link' => 1,
|
| 1426 |
+
'meta' => 1,
|
| 1427 |
+
'param' => 1,
|
| 1428 |
+
'source' => 1,
|
| 1429 |
+
'track' => 1,
|
| 1430 |
+
'wbr' => 1
|
| 1431 |
+
);
|
| 1432 |
+
protected $block_tags = array(
|
| 1433 |
+
'body' => 1,
|
| 1434 |
+
'div' => 1,
|
| 1435 |
+
'form' => 1,
|
| 1436 |
+
'root' => 1,
|
| 1437 |
+
'span' => 1,
|
| 1438 |
+
'table' => 1
|
| 1439 |
+
);
|
| 1440 |
+
protected $optional_closing_tags = array(
|
| 1441 |
+
// Not optional, see
|
| 1442 |
+
// https://www.w3.org/TR/html/textlevel-semantics.html#the-b-element
|
| 1443 |
+
'b' => array('b' => 1),
|
| 1444 |
+
'dd' => array('dd' => 1, 'dt' => 1),
|
| 1445 |
+
// Not optional, see
|
| 1446 |
+
// https://www.w3.org/TR/html/grouping-content.html#the-dl-element
|
| 1447 |
+
'dl' => array('dd' => 1, 'dt' => 1),
|
| 1448 |
+
'dt' => array('dd' => 1, 'dt' => 1),
|
| 1449 |
+
'li' => array('li' => 1),
|
| 1450 |
+
'optgroup' => array('optgroup' => 1, 'option' => 1),
|
| 1451 |
+
'option' => array('optgroup' => 1, 'option' => 1),
|
| 1452 |
+
'p' => array('p' => 1),
|
| 1453 |
+
'rp' => array('rp' => 1, 'rt' => 1),
|
| 1454 |
+
'rt' => array('rp' => 1, 'rt' => 1),
|
| 1455 |
+
'td' => array('td' => 1, 'th' => 1),
|
| 1456 |
+
'th' => array('td' => 1, 'th' => 1),
|
| 1457 |
+
'tr' => array('td' => 1, 'th' => 1, 'tr' => 1),
|
| 1458 |
+
);
|
| 1459 |
+
|
| 1460 |
+
function __construct(
|
| 1461 |
+
$str = null,
|
| 1462 |
+
$lowercase = true,
|
| 1463 |
+
$forceTagsClosed = true,
|
| 1464 |
+
$target_charset = DEFAULT_TARGET_CHARSET,
|
| 1465 |
+
$stripRN = true,
|
| 1466 |
+
$defaultBRText = DEFAULT_BR_TEXT,
|
| 1467 |
+
$defaultSpanText = DEFAULT_SPAN_TEXT,
|
| 1468 |
+
$options = 0)
|
| 1469 |
+
{
|
| 1470 |
+
if ($str) {
|
| 1471 |
+
if (preg_match('/^http:\/\//i', $str) || is_file($str)) {
|
| 1472 |
+
$this->load_file($str);
|
| 1473 |
+
} else {
|
| 1474 |
+
$this->load(
|
| 1475 |
+
$str,
|
| 1476 |
+
$lowercase,
|
| 1477 |
+
$stripRN,
|
| 1478 |
+
$defaultBRText,
|
| 1479 |
+
$defaultSpanText,
|
| 1480 |
+
$options
|
| 1481 |
+
);
|
| 1482 |
+
}
|
| 1483 |
+
}
|
| 1484 |
+
// Forcing tags to be closed implies that we don't trust the html, but
|
| 1485 |
+
// it can lead to parsing errors if we SHOULD trust the html.
|
| 1486 |
+
if (!$forceTagsClosed) {
|
| 1487 |
+
$this->optional_closing_array = array();
|
| 1488 |
+
}
|
| 1489 |
+
|
| 1490 |
+
$this->_target_charset = $target_charset;
|
| 1491 |
+
}
|
| 1492 |
+
|
| 1493 |
+
function __destruct()
|
| 1494 |
+
{
|
| 1495 |
+
$this->clear();
|
| 1496 |
+
}
|
| 1497 |
+
|
| 1498 |
+
function load(
|
| 1499 |
+
$str,
|
| 1500 |
+
$lowercase = true,
|
| 1501 |
+
$stripRN = true,
|
| 1502 |
+
$defaultBRText = DEFAULT_BR_TEXT,
|
| 1503 |
+
$defaultSpanText = DEFAULT_SPAN_TEXT,
|
| 1504 |
+
$options = 0)
|
| 1505 |
+
{
|
| 1506 |
+
global $debug_object;
|
| 1507 |
+
|
| 1508 |
+
// prepare
|
| 1509 |
+
$this->prepare($str, $lowercase, $defaultBRText, $defaultSpanText);
|
| 1510 |
+
|
| 1511 |
+
// Per sourceforge http://sourceforge.net/tracker/?func=detail&aid=2949097&group_id=218559&atid=1044037
|
| 1512 |
+
// Script tags removal now preceeds style tag removal.
|
| 1513 |
+
// strip out <script> tags
|
| 1514 |
+
$this->remove_noise("'<\s*script[^>]*[^/]>(.*?)<\s*/\s*script\s*>'is");
|
| 1515 |
+
$this->remove_noise("'<\s*script\s*>(.*?)<\s*/\s*script\s*>'is");
|
| 1516 |
+
|
| 1517 |
+
// strip out the \r \n's if we are told to.
|
| 1518 |
+
if ($stripRN) {
|
| 1519 |
+
$this->doc = str_replace("\r", ' ', $this->doc);
|
| 1520 |
+
$this->doc = str_replace("\n", ' ', $this->doc);
|
| 1521 |
+
|
| 1522 |
+
// set the length of content since we have changed it.
|
| 1523 |
+
$this->size = strlen($this->doc);
|
| 1524 |
+
}
|
| 1525 |
+
|
| 1526 |
+
// strip out cdata
|
| 1527 |
+
$this->remove_noise("'<!\[CDATA\[(.*?)\]\]>'is", true);
|
| 1528 |
+
// strip out comments
|
| 1529 |
+
$this->remove_noise("'<!--(.*?)-->'is");
|
| 1530 |
+
// strip out <style> tags
|
| 1531 |
+
$this->remove_noise("'<\s*style[^>]*[^/]>(.*?)<\s*/\s*style\s*>'is");
|
| 1532 |
+
$this->remove_noise("'<\s*style\s*>(.*?)<\s*/\s*style\s*>'is");
|
| 1533 |
+
// strip out preformatted tags
|
| 1534 |
+
$this->remove_noise("'<\s*(?:code)[^>]*>(.*?)<\s*/\s*(?:code)\s*>'is");
|
| 1535 |
+
// strip out server side scripts
|
| 1536 |
+
$this->remove_noise("'(<\?)(.*?)(\?>)'s", true);
|
| 1537 |
+
|
| 1538 |
+
if($options & HDOM_SMARTY_AS_TEXT) { // Strip Smarty scripts
|
| 1539 |
+
$this->remove_noise("'(\{\w)(.*?)(\})'s", true);
|
| 1540 |
+
}
|
| 1541 |
+
|
| 1542 |
+
// parsing
|
| 1543 |
+
$this->parse();
|
| 1544 |
+
// end
|
| 1545 |
+
$this->root->_[HDOM_INFO_END] = $this->cursor;
|
| 1546 |
+
$this->parse_charset();
|
| 1547 |
+
|
| 1548 |
+
// make load function chainable
|
| 1549 |
+
return $this;
|
| 1550 |
+
}
|
| 1551 |
+
|
| 1552 |
+
function load_file()
|
| 1553 |
+
{
|
| 1554 |
+
$args = func_get_args();
|
| 1555 |
+
|
| 1556 |
+
if(($doc = call_user_func_array('file_get_contents', $args)) !== false) {
|
| 1557 |
+
$this->load($doc, true);
|
| 1558 |
+
} else {
|
| 1559 |
+
return false;
|
| 1560 |
+
}
|
| 1561 |
+
}
|
| 1562 |
+
|
| 1563 |
+
function set_callback($function_name)
|
| 1564 |
+
{
|
| 1565 |
+
$this->callback = $function_name;
|
| 1566 |
+
}
|
| 1567 |
+
|
| 1568 |
+
function remove_callback()
|
| 1569 |
+
{
|
| 1570 |
+
$this->callback = null;
|
| 1571 |
+
}
|
| 1572 |
+
|
| 1573 |
+
function save($filepath = '')
|
| 1574 |
+
{
|
| 1575 |
+
$ret = $this->root->innertext();
|
| 1576 |
+
if ($filepath !== '') { file_put_contents($filepath, $ret, LOCK_EX); }
|
| 1577 |
+
return $ret;
|
| 1578 |
+
}
|
| 1579 |
+
|
| 1580 |
+
function find($selector, $idx = null, $lowercase = false)
|
| 1581 |
+
{
|
| 1582 |
+
return $this->root->find($selector, $idx, $lowercase);
|
| 1583 |
+
}
|
| 1584 |
+
|
| 1585 |
+
function clear()
|
| 1586 |
+
{
|
| 1587 |
+
if (isset($this->nodes)) {
|
| 1588 |
+
foreach ($this->nodes as $n) {
|
| 1589 |
+
$n->clear();
|
| 1590 |
+
$n = null;
|
| 1591 |
+
}
|
| 1592 |
+
}
|
| 1593 |
+
|
| 1594 |
+
// This add next line is documented in the sourceforge repository.
|
| 1595 |
+
// 2977248 as a fix for ongoing memory leaks that occur even with the
|
| 1596 |
+
// use of clear.
|
| 1597 |
+
if (isset($this->children)) {
|
| 1598 |
+
foreach ($this->children as $n) {
|
| 1599 |
+
$n->clear();
|
| 1600 |
+
$n = null;
|
| 1601 |
+
}
|
| 1602 |
+
}
|
| 1603 |
+
|
| 1604 |
+
if (isset($this->parent)) {
|
| 1605 |
+
$this->parent->clear();
|
| 1606 |
+
unset($this->parent);
|
| 1607 |
+
}
|
| 1608 |
+
|
| 1609 |
+
if (isset($this->root)) {
|
| 1610 |
+
$this->root->clear();
|
| 1611 |
+
unset($this->root);
|
| 1612 |
+
}
|
| 1613 |
+
|
| 1614 |
+
unset($this->doc);
|
| 1615 |
+
unset($this->noise);
|
| 1616 |
+
}
|
| 1617 |
+
|
| 1618 |
+
function dump($show_attr = true)
|
| 1619 |
+
{
|
| 1620 |
+
$this->root->dump($show_attr);
|
| 1621 |
+
}
|
| 1622 |
+
|
| 1623 |
+
protected function prepare(
|
| 1624 |
+
$str, $lowercase = true,
|
| 1625 |
+
$defaultBRText = DEFAULT_BR_TEXT,
|
| 1626 |
+
$defaultSpanText = DEFAULT_SPAN_TEXT)
|
| 1627 |
+
{
|
| 1628 |
+
$this->clear();
|
| 1629 |
+
|
| 1630 |
+
$this->doc = trim($str);
|
| 1631 |
+
$this->size = strlen($this->doc);
|
| 1632 |
+
$this->original_size = $this->size; // original size of the html
|
| 1633 |
+
$this->pos = 0;
|
| 1634 |
+
$this->cursor = 1;
|
| 1635 |
+
$this->noise = array();
|
| 1636 |
+
$this->nodes = array();
|
| 1637 |
+
$this->lowercase = $lowercase;
|
| 1638 |
+
$this->default_br_text = $defaultBRText;
|
| 1639 |
+
$this->default_span_text = $defaultSpanText;
|
| 1640 |
+
$this->root = new simple_html_dom_node($this);
|
| 1641 |
+
$this->root->tag = 'root';
|
| 1642 |
+
$this->root->_[HDOM_INFO_BEGIN] = -1;
|
| 1643 |
+
$this->root->nodetype = HDOM_TYPE_ROOT;
|
| 1644 |
+
$this->parent = $this->root;
|
| 1645 |
+
if ($this->size > 0) { $this->char = $this->doc[0]; }
|
| 1646 |
+
}
|
| 1647 |
+
|
| 1648 |
+
protected function parse()
|
| 1649 |
+
{
|
| 1650 |
+
while (true) {
|
| 1651 |
+
// Read next tag if there is no text between current position and the
|
| 1652 |
+
// next opening tag.
|
| 1653 |
+
if (($s = $this->copy_until_char('<')) === '') {
|
| 1654 |
+
if($this->read_tag()) {
|
| 1655 |
+
continue;
|
| 1656 |
+
} else {
|
| 1657 |
+
return true;
|
| 1658 |
+
}
|
| 1659 |
+
}
|
| 1660 |
+
|
| 1661 |
+
// Add a text node for text between tags
|
| 1662 |
+
$node = new simple_html_dom_node($this);
|
| 1663 |
+
++$this->cursor;
|
| 1664 |
+
$node->_[HDOM_INFO_TEXT] = $s;
|
| 1665 |
+
$this->link_nodes($node, false);
|
| 1666 |
+
}
|
| 1667 |
+
}
|
| 1668 |
+
|
| 1669 |
+
protected function parse_charset()
|
| 1670 |
+
{
|
| 1671 |
+
global $debug_object;
|
| 1672 |
+
|
| 1673 |
+
$charset = null;
|
| 1674 |
+
|
| 1675 |
+
if (function_exists('get_last_retrieve_url_contents_content_type')) {
|
| 1676 |
+
$contentTypeHeader = get_last_retrieve_url_contents_content_type();
|
| 1677 |
+
$success = preg_match('/charset=(.+)/', $contentTypeHeader, $matches);
|
| 1678 |
+
if ($success) {
|
| 1679 |
+
$charset = $matches[1];
|
| 1680 |
+
if (is_object($debug_object)) {
|
| 1681 |
+
$debug_object->debug_log(2,
|
| 1682 |
+
'header content-type found charset of: '
|
| 1683 |
+
. $charset
|
| 1684 |
+
);
|
| 1685 |
+
}
|
| 1686 |
+
}
|
| 1687 |
+
}
|
| 1688 |
+
|
| 1689 |
+
if (empty($charset)) {
|
| 1690 |
+
// https://www.w3.org/TR/html/document-metadata.html#statedef-http-equiv-content-type
|
| 1691 |
+
$el = $this->root->find('meta[http-equiv=Content-Type]', 0, true);
|
| 1692 |
+
|
| 1693 |
+
if (!empty($el)) {
|
| 1694 |
+
$fullvalue = $el->content;
|
| 1695 |
+
if (is_object($debug_object)) {
|
| 1696 |
+
$debug_object->debug_log(2,
|
| 1697 |
+
'meta content-type tag found'
|
| 1698 |
+
. $fullvalue
|
| 1699 |
+
);
|
| 1700 |
+
}
|
| 1701 |
+
|
| 1702 |
+
if (!empty($fullvalue)) {
|
| 1703 |
+
$success = preg_match(
|
| 1704 |
+
'/charset=(.+)/i',
|
| 1705 |
+
$fullvalue,
|
| 1706 |
+
$matches
|
| 1707 |
+
);
|
| 1708 |
+
|
| 1709 |
+
if ($success) {
|
| 1710 |
+
$charset = $matches[1];
|
| 1711 |
+
} else {
|
| 1712 |
+
// If there is a meta tag, and they don't specify the
|
| 1713 |
+
// character set, research says that it's typically
|
| 1714 |
+
// ISO-8859-1
|
| 1715 |
+
if (is_object($debug_object)) {
|
| 1716 |
+
$debug_object->debug_log(2,
|
| 1717 |
+
'meta content-type tag couldn\'t be parsed. using iso-8859 default.'
|
| 1718 |
+
);
|
| 1719 |
+
}
|
| 1720 |
+
|
| 1721 |
+
$charset = 'ISO-8859-1';
|
| 1722 |
+
}
|
| 1723 |
+
}
|
| 1724 |
+
}
|
| 1725 |
+
}
|
| 1726 |
+
|
| 1727 |
+
if (empty($charset)) {
|
| 1728 |
+
// https://www.w3.org/TR/html/document-metadata.html#character-encoding-declaration
|
| 1729 |
+
if ($meta = $this->root->find('meta[charset]', 0)) {
|
| 1730 |
+
$charset = $meta->charset;
|
| 1731 |
+
if (is_object($debug_object)) {
|
| 1732 |
+
$debug_object->debug_log(2, 'meta charset: ' . $charset);
|
| 1733 |
+
}
|
| 1734 |
+
}
|
| 1735 |
+
}
|
| 1736 |
+
|
| 1737 |
+
if (empty($charset)) {
|
| 1738 |
+
// Try to guess the charset based on the content
|
| 1739 |
+
// Requires Multibyte String (mbstring) support (optional)
|
| 1740 |
+
if (function_exists('mb_detect_encoding')) {
|
| 1741 |
+
/**
|
| 1742 |
+
* mb_detect_encoding() is not intended to distinguish between
|
| 1743 |
+
* charsets, especially single-byte charsets. Its primary
|
| 1744 |
+
* purpose is to detect which multibyte encoding is in use,
|
| 1745 |
+
* i.e. UTF-8, UTF-16, shift-JIS, etc.
|
| 1746 |
+
*
|
| 1747 |
+
* -- https://bugs.php.net/bug.php?id=38138
|
| 1748 |
+
*
|
| 1749 |
+
* Adding both CP1251/ISO-8859-5 and CP1252/ISO-8859-1 will
|
| 1750 |
+
* always result in CP1251/ISO-8859-5 and vice versa.
|
| 1751 |
+
*
|
| 1752 |
+
* Thus, only detect if it's either UTF-8 or CP1252/ISO-8859-1
|
| 1753 |
+
* to stay compatible.
|
| 1754 |
+
*/
|
| 1755 |
+
$encoding = mb_detect_encoding(
|
| 1756 |
+
$this->doc,
|
| 1757 |
+
array( 'UTF-8', 'CP1252', 'ISO-8859-1' )
|
| 1758 |
+
);
|
| 1759 |
+
|
| 1760 |
+
if ($encoding === 'CP1252' || $encoding === 'ISO-8859-1') {
|
| 1761 |
+
// Due to a limitation of mb_detect_encoding
|
| 1762 |
+
// 'CP1251'/'ISO-8859-5' will be detected as
|
| 1763 |
+
// 'CP1252'/'ISO-8859-1'. This will cause iconv to fail, in
|
| 1764 |
+
// which case we can simply assume it is the other charset.
|
| 1765 |
+
if (!@iconv('CP1252', 'UTF-8', $this->doc)) {
|
| 1766 |
+
$encoding = 'CP1251';
|
| 1767 |
+
}
|
| 1768 |
+
}
|
| 1769 |
+
|
| 1770 |
+
if ($encoding !== false) {
|
| 1771 |
+
$charset = $encoding;
|
| 1772 |
+
if (is_object($debug_object)) {
|
| 1773 |
+
$debug_object->debug_log(2, 'mb_detect: ' . $charset);
|
| 1774 |
+
}
|
| 1775 |
+
}
|
| 1776 |
+
}
|
| 1777 |
+
}
|
| 1778 |
+
|
| 1779 |
+
if (empty($charset)) {
|
| 1780 |
+
// Assume it's UTF-8 as it is the most likely charset to be used
|
| 1781 |
+
$charset = 'UTF-8';
|
| 1782 |
+
if (is_object($debug_object)) {
|
| 1783 |
+
$debug_object->debug_log(2, 'No match found, assume ' . $charset);
|
| 1784 |
+
}
|
| 1785 |
+
}
|
| 1786 |
+
|
| 1787 |
+
// Since CP1252 is a superset, if we get one of it's subsets, we want
|
| 1788 |
+
// it instead.
|
| 1789 |
+
if ((strtolower($charset) == 'iso-8859-1')
|
| 1790 |
+
|| (strtolower($charset) == 'latin1')
|
| 1791 |
+
|| (strtolower($charset) == 'latin-1')) {
|
| 1792 |
+
$charset = 'CP1252';
|
| 1793 |
+
if (is_object($debug_object)) {
|
| 1794 |
+
$debug_object->debug_log(2,
|
| 1795 |
+
'replacing ' . $charset . ' with CP1252 as its a superset'
|
| 1796 |
+
);
|
| 1797 |
+
}
|
| 1798 |
+
}
|
| 1799 |
+
|
| 1800 |
+
if (is_object($debug_object)) {
|
| 1801 |
+
$debug_object->debug_log(1, 'EXIT - ' . $charset);
|
| 1802 |
+
}
|
| 1803 |
+
|
| 1804 |
+
return $this->_charset = $charset;
|
| 1805 |
+
}
|
| 1806 |
+
|
| 1807 |
+
protected function read_tag()
|
| 1808 |
+
{
|
| 1809 |
+
// Set end position if no further tags found
|
| 1810 |
+
if ($this->char !== '<') {
|
| 1811 |
+
$this->root->_[HDOM_INFO_END] = $this->cursor;
|
| 1812 |
+
return false;
|
| 1813 |
+
}
|
| 1814 |
+
|
| 1815 |
+
$begin_tag_pos = $this->pos;
|
| 1816 |
+
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next
|
| 1817 |
+
|
| 1818 |
+
// end tag
|
| 1819 |
+
if ($this->char === '/') {
|
| 1820 |
+
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next
|
| 1821 |
+
|
| 1822 |
+
// Skip whitespace in end tags (i.e. in "</ html>")
|
| 1823 |
+
$this->skip($this->token_blank);
|
| 1824 |
+
$tag = $this->copy_until_char('>');
|
| 1825 |
+
|
| 1826 |
+
// Skip attributes in end tags
|
| 1827 |
+
if (($pos = strpos($tag, ' ')) !== false) {
|
| 1828 |
+
$tag = substr($tag, 0, $pos);
|
| 1829 |
+
}
|
| 1830 |
+
|
| 1831 |
+
$parent_lower = strtolower($this->parent->tag);
|
| 1832 |
+
$tag_lower = strtolower($tag);
|
| 1833 |
+
|
| 1834 |
+
// The end tag is supposed to close the parent tag. Handle situations
|
| 1835 |
+
// when it doesn't
|
| 1836 |
+
if ($parent_lower !== $tag_lower) {
|
| 1837 |
+
// Parent tag does not have to be closed necessarily (optional closing tag)
|
| 1838 |
+
// Current tag is a block tag, so it may close an ancestor
|
| 1839 |
+
if (isset($this->optional_closing_tags[$parent_lower])
|
| 1840 |
+
&& isset($this->block_tags[$tag_lower])) {
|
| 1841 |
+
|
| 1842 |
+
$this->parent->_[HDOM_INFO_END] = 0;
|
| 1843 |
+
$org_parent = $this->parent;
|
| 1844 |
+
|
| 1845 |
+
// Traverse ancestors to find a matching opening tag
|
| 1846 |
+
// Stop at root node
|
| 1847 |
+
while (($this->parent->parent)
|
| 1848 |
+
&& strtolower($this->parent->tag) !== $tag_lower
|
| 1849 |
+
){
|
| 1850 |
+
$this->parent = $this->parent->parent;
|
| 1851 |
+
}
|
| 1852 |
+
|
| 1853 |
+
// If we don't have a match add current tag as text node
|
| 1854 |
+
if (strtolower($this->parent->tag) !== $tag_lower) {
|
| 1855 |
+
$this->parent = $org_parent; // restore origonal parent
|
| 1856 |
+
|
| 1857 |
+
if ($this->parent->parent) {
|
| 1858 |
+
$this->parent = $this->parent->parent;
|
| 1859 |
+
}
|
| 1860 |
+
|
| 1861 |
+
$this->parent->_[HDOM_INFO_END] = $this->cursor;
|
| 1862 |
+
return $this->as_text_node($tag);
|
| 1863 |
+
}
|
| 1864 |
+
} elseif (($this->parent->parent)
|
| 1865 |
+
&& isset($this->block_tags[$tag_lower])
|
| 1866 |
+
) {
|
| 1867 |
+
// Grandparent exists and current tag is a block tag, so our
|
| 1868 |
+
// parent doesn't have an end tag
|
| 1869 |
+
$this->parent->_[HDOM_INFO_END] = 0; // No end tag
|
| 1870 |
+
$org_parent = $this->parent;
|
| 1871 |
+
|
| 1872 |
+
// Traverse ancestors to find a matching opening tag
|
| 1873 |
+
// Stop at root node
|
| 1874 |
+
while (($this->parent->parent)
|
| 1875 |
+
&& strtolower($this->parent->tag) !== $tag_lower
|
| 1876 |
+
) {
|
| 1877 |
+
$this->parent = $this->parent->parent;
|
| 1878 |
+
}
|
| 1879 |
+
|
| 1880 |
+
// If we don't have a match add current tag as text node
|
| 1881 |
+
if (strtolower($this->parent->tag) !== $tag_lower) {
|
| 1882 |
+
$this->parent = $org_parent; // restore origonal parent
|
| 1883 |
+
$this->parent->_[HDOM_INFO_END] = $this->cursor;
|
| 1884 |
+
return $this->as_text_node($tag);
|
| 1885 |
+
}
|
| 1886 |
+
} elseif (($this->parent->parent)
|
| 1887 |
+
&& strtolower($this->parent->parent->tag) === $tag_lower
|
| 1888 |
+
) { // Grandparent exists and current tag closes it
|
| 1889 |
+
$this->parent->_[HDOM_INFO_END] = 0;
|
| 1890 |
+
$this->parent = $this->parent->parent;
|
| 1891 |
+
} else { // Random tag, add as text node
|
| 1892 |
+
return $this->as_text_node($tag);
|
| 1893 |
+
}
|
| 1894 |
+
}
|
| 1895 |
+
|
| 1896 |
+
// Set end position of parent tag to current cursor position
|
| 1897 |
+
$this->parent->_[HDOM_INFO_END] = $this->cursor;
|
| 1898 |
+
|
| 1899 |
+
if ($this->parent->parent) {
|
| 1900 |
+
$this->parent = $this->parent->parent;
|
| 1901 |
+
}
|
| 1902 |
+
|
| 1903 |
+
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next
|
| 1904 |
+
return true;
|
| 1905 |
+
}
|
| 1906 |
+
|
| 1907 |
+
// start tag
|
| 1908 |
+
$node = new simple_html_dom_node($this);
|
| 1909 |
+
$node->_[HDOM_INFO_BEGIN] = $this->cursor;
|
| 1910 |
+
++$this->cursor;
|
| 1911 |
+
$tag = $this->copy_until($this->token_slash); // Get tag name
|
| 1912 |
+
$node->tag_start = $begin_tag_pos;
|
| 1913 |
+
|
| 1914 |
+
// doctype, cdata & comments...
|
| 1915 |
+
// <!DOCTYPE html>
|
| 1916 |
+
// <![CDATA[ ... ]]>
|
| 1917 |
+
// <!-- Comment -->
|
| 1918 |
+
if (isset($tag[0]) && $tag[0] === '!') {
|
| 1919 |
+
$node->_[HDOM_INFO_TEXT] = '<' . $tag . $this->copy_until_char('>');
|
| 1920 |
+
|
| 1921 |
+
if (isset($tag[2]) && $tag[1] === '-' && $tag[2] === '-') { // Comment ("<!--")
|
| 1922 |
+
$node->nodetype = HDOM_TYPE_COMMENT;
|
| 1923 |
+
$node->tag = 'comment';
|
| 1924 |
+
} else { // Could be doctype or CDATA but we don't care
|
| 1925 |
+
$node->nodetype = HDOM_TYPE_UNKNOWN;
|
| 1926 |
+
$node->tag = 'unknown';
|
| 1927 |
+
}
|
| 1928 |
+
|
| 1929 |
+
if ($this->char === '>') { $node->_[HDOM_INFO_TEXT] .= '>'; }
|
| 1930 |
+
|
| 1931 |
+
$this->link_nodes($node, true);
|
| 1932 |
+
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next
|
| 1933 |
+
return true;
|
| 1934 |
+
}
|
| 1935 |
+
|
| 1936 |
+
// The start tag cannot contain another start tag, if so add as text
|
| 1937 |
+
// i.e. "<<html>"
|
| 1938 |
+
if ($pos = strpos($tag, '<') !== false) {
|
| 1939 |
+
$tag = '<' . substr($tag, 0, -1);
|
| 1940 |
+
$node->_[HDOM_INFO_TEXT] = $tag;
|
| 1941 |
+
$this->link_nodes($node, false);
|
| 1942 |
+
$this->char = $this->doc[--$this->pos]; // prev
|
| 1943 |
+
return true;
|
| 1944 |
+
}
|
| 1945 |
+
|
| 1946 |
+
// Handle invalid tag names (i.e. "<html#doc>")
|
| 1947 |
+
if (!preg_match('/^\w[\w:-]*$/', $tag)) {
|
| 1948 |
+
$node->_[HDOM_INFO_TEXT] = '<' . $tag . $this->copy_until('<>');
|
| 1949 |
+
|
| 1950 |
+
// Next char is the beginning of a new tag, don't touch it.
|
| 1951 |
+
if ($this->char === '<') {
|
| 1952 |
+
$this->link_nodes($node, false);
|
| 1953 |
+
return true;
|
| 1954 |
+
}
|
| 1955 |
+
|
| 1956 |
+
// Next char closes current tag, add and be done with it.
|
| 1957 |
+
if ($this->char === '>') { $node->_[HDOM_INFO_TEXT] .= '>'; }
|
| 1958 |
+
$this->link_nodes($node, false);
|
| 1959 |
+
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next
|
| 1960 |
+
return true;
|
| 1961 |
+
}
|
| 1962 |
+
|
| 1963 |
+
// begin tag, add new node
|
| 1964 |
+
$node->nodetype = HDOM_TYPE_ELEMENT;
|
| 1965 |
+
$tag_lower = strtolower($tag);
|
| 1966 |
+
$node->tag = ($this->lowercase) ? $tag_lower : $tag;
|
| 1967 |
+
|
| 1968 |
+
// handle optional closing tags
|
| 1969 |
+
if (isset($this->optional_closing_tags[$tag_lower])) {
|
| 1970 |
+
// Traverse ancestors to close all optional closing tags
|
| 1971 |
+
while (isset($this->optional_closing_tags[$tag_lower][strtolower($this->parent->tag)])) {
|
| 1972 |
+
$this->parent->_[HDOM_INFO_END] = 0;
|
| 1973 |
+
$this->parent = $this->parent->parent;
|
| 1974 |
+
}
|
| 1975 |
+
$node->parent = $this->parent;
|
| 1976 |
+
}
|
| 1977 |
+
|
| 1978 |
+
$guard = 0; // prevent infinity loop
|
| 1979 |
+
|
| 1980 |
+
// [0] Space between tag and first attribute
|
| 1981 |
+
$space = array($this->copy_skip($this->token_blank), '', '');
|
| 1982 |
+
|
| 1983 |
+
// attributes
|
| 1984 |
+
do {
|
| 1985 |
+
// Everything until the first equal sign should be the attribute name
|
| 1986 |
+
$name = $this->copy_until($this->token_equal);
|
| 1987 |
+
|
| 1988 |
+
if ($name === '' && $this->char !== null && $space[0] === '') {
|
| 1989 |
+
break;
|
| 1990 |
+
}
|
| 1991 |
+
|
| 1992 |
+
if ($guard === $this->pos) { // Escape infinite loop
|
| 1993 |
+
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next
|
| 1994 |
+
continue;
|
| 1995 |
+
}
|
| 1996 |
+
|
| 1997 |
+
$guard = $this->pos;
|
| 1998 |
+
|
| 1999 |
+
// handle endless '<'
|
| 2000 |
+
// Out of bounds before the tag ended
|
| 2001 |
+
if ($this->pos >= $this->size - 1 && $this->char !== '>') {
|
| 2002 |
+
$node->nodetype = HDOM_TYPE_TEXT;
|
| 2003 |
+
$node->_[HDOM_INFO_END] = 0;
|
| 2004 |
+
$node->_[HDOM_INFO_TEXT] = '<' . $tag . $space[0] . $name;
|
| 2005 |
+
$node->tag = 'text';
|
| 2006 |
+
$this->link_nodes($node, false);
|
| 2007 |
+
return true;
|
| 2008 |
+
}
|
| 2009 |
+
|
| 2010 |
+
// handle mismatch '<'
|
| 2011 |
+
// Attributes cannot start after opening tag
|
| 2012 |
+
if ($this->doc[$this->pos - 1] == '<') {
|
| 2013 |
+
$node->nodetype = HDOM_TYPE_TEXT;
|
| 2014 |
+
$node->tag = 'text';
|
| 2015 |
+
$node->attr = array();
|
| 2016 |
+
$node->_[HDOM_INFO_END] = 0;
|
| 2017 |
+
$node->_[HDOM_INFO_TEXT] = substr(
|
| 2018 |
+
$this->doc,
|
| 2019 |
+
$begin_tag_pos,
|
| 2020 |
+
$this->pos - $begin_tag_pos - 1
|
| 2021 |
+
);
|
| 2022 |
+
$this->pos -= 2;
|
| 2023 |
+
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next
|
| 2024 |
+
$this->link_nodes($node, false);
|
| 2025 |
+
return true;
|
| 2026 |
+
}
|
| 2027 |
+
|
| 2028 |
+
if ($name !== '/' && $name !== '') { // this is a attribute name
|
| 2029 |
+
// [1] Whitespace after attribute name
|
| 2030 |
+
$space[1] = $this->copy_skip($this->token_blank);
|
| 2031 |
+
|
| 2032 |
+
$name = $this->restore_noise($name); // might be a noisy name
|
| 2033 |
+
|
| 2034 |
+
if ($this->lowercase) { $name = strtolower($name); }
|
| 2035 |
+
|
| 2036 |
+
if ($this->char === '=') { // attribute with value
|
| 2037 |
+
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next
|
| 2038 |
+
$this->parse_attr($node, $name, $space); // get attribute value
|
| 2039 |
+
} else {
|
| 2040 |
+
//no value attr: nowrap, checked selected...
|
| 2041 |
+
$node->_[HDOM_INFO_QUOTE][] = HDOM_QUOTE_NO;
|
| 2042 |
+
$node->attr[$name] = true;
|
| 2043 |
+
if ($this->char != '>') { $this->char = $this->doc[--$this->pos]; } // prev
|
| 2044 |
+
}
|
| 2045 |
+
|
| 2046 |
+
$node->_[HDOM_INFO_SPACE][] = $space;
|
| 2047 |
+
|
| 2048 |
+
// prepare for next attribute
|
| 2049 |
+
$space = array(
|
| 2050 |
+
$this->copy_skip($this->token_blank),
|
| 2051 |
+
'',
|
| 2052 |
+
''
|
| 2053 |
+
);
|
| 2054 |
+
} else { // no more attributes
|
| 2055 |
+
break;
|
| 2056 |
+
}
|
| 2057 |
+
} while ($this->char !== '>' && $this->char !== '/'); // go until the tag ended
|
| 2058 |
+
|
| 2059 |
+
$this->link_nodes($node, true);
|
| 2060 |
+
$node->_[HDOM_INFO_ENDSPACE] = $space[0];
|
| 2061 |
+
|
| 2062 |
+
// handle empty tags (i.e. "<div/>")
|
| 2063 |
+
if ($this->copy_until_char('>') === '/') {
|
| 2064 |
+
$node->_[HDOM_INFO_ENDSPACE] .= '/';
|
| 2065 |
+
$node->_[HDOM_INFO_END] = 0;
|
| 2066 |
+
} else {
|
| 2067 |
+
// reset parent
|
| 2068 |
+
if (!isset($this->self_closing_tags[strtolower($node->tag)])) {
|
| 2069 |
+
$this->parent = $node;
|
| 2070 |
+
}
|
| 2071 |
+
}
|
| 2072 |
+
|
| 2073 |
+
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next
|
| 2074 |
+
|
| 2075 |
+
// If it's a BR tag, we need to set it's text to the default text.
|
| 2076 |
+
// This way when we see it in plaintext, we can generate formatting that the user wants.
|
| 2077 |
+
// since a br tag never has sub nodes, this works well.
|
| 2078 |
+
if ($node->tag === 'br') {
|
| 2079 |
+
$node->_[HDOM_INFO_INNER] = $this->default_br_text;
|
| 2080 |
+
}
|
| 2081 |
+
|
| 2082 |
+
return true;
|
| 2083 |
+
}
|
| 2084 |
+
|
| 2085 |
+
protected function parse_attr($node, $name, &$space)
|
| 2086 |
+
{
|
| 2087 |
+
$is_duplicate = isset($node->attr[$name]);
|
| 2088 |
+
|
| 2089 |
+
if (!$is_duplicate) // Copy whitespace between "=" and value
|
| 2090 |
+
$space[2] = $this->copy_skip($this->token_blank);
|
| 2091 |
+
|
| 2092 |
+
switch ($this->char) {
|
| 2093 |
+
case '"':
|
| 2094 |
+
$quote_type = HDOM_QUOTE_DOUBLE;
|
| 2095 |
+
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next
|
| 2096 |
+
$value = $this->copy_until_char('"');
|
| 2097 |
+
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next
|
| 2098 |
+
break;
|
| 2099 |
+
case '\'':
|
| 2100 |
+
$quote_type = HDOM_QUOTE_SINGLE;
|
| 2101 |
+
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next
|
| 2102 |
+
$value = $this->copy_until_char('\'');
|
| 2103 |
+
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next
|
| 2104 |
+
break;
|
| 2105 |
+
default:
|
| 2106 |
+
$quote_type = HDOM_QUOTE_NO;
|
| 2107 |
+
$value = $this->copy_until($this->token_attr);
|
| 2108 |
+
}
|
| 2109 |
+
|
| 2110 |
+
$value = $this->restore_noise($value);
|
| 2111 |
+
|
| 2112 |
+
// PaperG: Attributes should not have \r or \n in them, that counts as
|
| 2113 |
+
// html whitespace.
|
| 2114 |
+
$value = str_replace("\r", '', $value);
|
| 2115 |
+
$value = str_replace("\n", '', $value);
|
| 2116 |
+
|
| 2117 |
+
// PaperG: If this is a "class" selector, lets get rid of the preceeding
|
| 2118 |
+
// and trailing space since some people leave it in the multi class case.
|
| 2119 |
+
if ($name === 'class') {
|
| 2120 |
+
$value = trim($value);
|
| 2121 |
+
}
|
| 2122 |
+
|
| 2123 |
+
if (!$is_duplicate) {
|
| 2124 |
+
$node->_[HDOM_INFO_QUOTE][] = $quote_type;
|
| 2125 |
+
$node->attr[$name] = $value;
|
| 2126 |
+
}
|
| 2127 |
+
}
|
| 2128 |
+
|
| 2129 |
+
protected function link_nodes(&$node, $is_child)
|
| 2130 |
+
{
|
| 2131 |
+
$node->parent = $this->parent;
|
| 2132 |
+
$this->parent->nodes[] = $node;
|
| 2133 |
+
if ($is_child) {
|
| 2134 |
+
$this->parent->children[] = $node;
|
| 2135 |
+
}
|
| 2136 |
+
}
|
| 2137 |
+
|
| 2138 |
+
protected function as_text_node($tag)
|
| 2139 |
+
{
|
| 2140 |
+
$node = new simple_html_dom_node($this);
|
| 2141 |
+
++$this->cursor;
|
| 2142 |
+
$node->_[HDOM_INFO_TEXT] = '</' . $tag . '>';
|
| 2143 |
+
$this->link_nodes($node, false);
|
| 2144 |
+
$this->char = (++$this->pos < $this->size) ? $this->doc[$this->pos] : null; // next
|
| 2145 |
+
return true;
|
| 2146 |
+
}
|
| 2147 |
+
|
| 2148 |
+
protected function skip($chars)
|
| 2149 |
+
{
|
| 2150 |
+
$this->pos += strspn($this->doc, $chars, $this->pos);
|
| 2151 |
+
$this->char = ($this->pos < $this->size) ? $this->doc[$this->pos] : null; // next
|
| 2152 |
+
}
|
| 2153 |
+
|
| 2154 |
+
protected function copy_skip($chars)
|
| 2155 |
+
{
|
| 2156 |
+
$pos = $this->pos;
|
| 2157 |
+
$len = strspn($this->doc, $chars, $pos);
|
| 2158 |
+
$this->pos += $len;
|
| 2159 |
+
$this->char = ($this->pos < $this->size) ? $this->doc[$this->pos] : null; // next
|
| 2160 |
+
if ($len === 0) { return ''; }
|
| 2161 |
+
return substr($this->doc, $pos, $len);
|
| 2162 |
+
}
|
| 2163 |
+
|
| 2164 |
+
protected function copy_until($chars)
|
| 2165 |
+
{
|
| 2166 |
+
$pos = $this->pos;
|
| 2167 |
+
$len = strcspn($this->doc, $chars, $pos);
|
| 2168 |
+
$this->pos += $len;
|
| 2169 |
+
$this->char = ($this->pos < $this->size) ? $this->doc[$this->pos] : null; // next
|
| 2170 |
+
return substr($this->doc, $pos, $len);
|
| 2171 |
+
}
|
| 2172 |
+
|
| 2173 |
+
protected function copy_until_char($char)
|
| 2174 |
+
{
|
| 2175 |
+
if ($this->char === null) { return ''; }
|
| 2176 |
+
|
| 2177 |
+
if (($pos = strpos($this->doc, $char, $this->pos)) === false) {
|
| 2178 |
+
$ret = substr($this->doc, $this->pos, $this->size - $this->pos);
|
| 2179 |
+
$this->char = null;
|
| 2180 |
+
$this->pos = $this->size;
|
| 2181 |
+
return $ret;
|
| 2182 |
+
}
|
| 2183 |
+
|
| 2184 |
+
if ($pos === $this->pos) { return ''; }
|
| 2185 |
+
|
| 2186 |
+
$pos_old = $this->pos;
|
| 2187 |
+
$this->char = $this->doc[$pos];
|
| 2188 |
+
$this->pos = $pos;
|
| 2189 |
+
return substr($this->doc, $pos_old, $pos - $pos_old);
|
| 2190 |
+
}
|
| 2191 |
+
|
| 2192 |
+
protected function remove_noise($pattern, $remove_tag = false)
|
| 2193 |
+
{
|
| 2194 |
+
global $debug_object;
|
| 2195 |
+
if (is_object($debug_object)) { $debug_object->debug_log_entry(1); }
|
| 2196 |
+
|
| 2197 |
+
$count = preg_match_all(
|
| 2198 |
+
$pattern,
|
| 2199 |
+
$this->doc,
|
| 2200 |
+
$matches,
|
| 2201 |
+
PREG_SET_ORDER | PREG_OFFSET_CAPTURE
|
| 2202 |
+
);
|
| 2203 |
+
|
| 2204 |
+
for ($i = $count - 1; $i > -1; --$i) {
|
| 2205 |
+
$key = '___noise___' . sprintf('% 5d', count($this->noise) + 1000);
|
| 2206 |
+
|
| 2207 |
+
if (is_object($debug_object)) {
|
| 2208 |
+
$debug_object->debug_log(2, 'key is: ' . $key);
|
| 2209 |
+
}
|
| 2210 |
+
|
| 2211 |
+
$idx = ($remove_tag) ? 0 : 1; // 0 = entire match, 1 = submatch
|
| 2212 |
+
$this->noise[$key] = $matches[$i][$idx][0];
|
| 2213 |
+
$this->doc = substr_replace($this->doc, $key, $matches[$i][$idx][1], strlen($matches[$i][$idx][0]));
|
| 2214 |
+
}
|
| 2215 |
+
|
| 2216 |
+
// reset the length of content
|
| 2217 |
+
$this->size = strlen($this->doc);
|
| 2218 |
+
|
| 2219 |
+
if ($this->size > 0) {
|
| 2220 |
+
$this->char = $this->doc[0];
|
| 2221 |
+
}
|
| 2222 |
+
}
|
| 2223 |
+
|
| 2224 |
+
function restore_noise($text)
|
| 2225 |
+
{
|
| 2226 |
+
global $debug_object;
|
| 2227 |
+
if (is_object($debug_object)) { $debug_object->debug_log_entry(1); }
|
| 2228 |
+
|
| 2229 |
+
while (($pos = strpos($text, '___noise___')) !== false) {
|
| 2230 |
+
// Sometimes there is a broken piece of markup, and we don't GET the
|
| 2231 |
+
// pos+11 etc... token which indicates a problem outside of us...
|
| 2232 |
+
|
| 2233 |
+
// todo: "___noise___1000" (or any number with four or more digits)
|
| 2234 |
+
// in the DOM causes an infinite loop which could be utilized by
|
| 2235 |
+
// malicious software
|
| 2236 |
+
if (strlen($text) > $pos + 15) {
|
| 2237 |
+
$key = '___noise___'
|
| 2238 |
+
. $text[$pos + 11]
|
| 2239 |
+
. $text[$pos + 12]
|
| 2240 |
+
. $text[$pos + 13]
|
| 2241 |
+
. $text[$pos + 14]
|
| 2242 |
+
. $text[$pos + 15];
|
| 2243 |
+
|
| 2244 |
+
if (is_object($debug_object)) {
|
| 2245 |
+
$debug_object->debug_log(2, 'located key of: ' . $key);
|
| 2246 |
+
}
|
| 2247 |
+
|
| 2248 |
+
if (isset($this->noise[$key])) {
|
| 2249 |
+
$text = substr($text, 0, $pos)
|
| 2250 |
+
. $this->noise[$key]
|
| 2251 |
+
. substr($text, $pos + 16);
|
| 2252 |
+
} else {
|
| 2253 |
+
// do this to prevent an infinite loop.
|
| 2254 |
+
$text = substr($text, 0, $pos)
|
| 2255 |
+
. 'UNDEFINED NOISE FOR KEY: '
|
| 2256 |
+
. $key
|
| 2257 |
+
. substr($text, $pos + 16);
|
| 2258 |
+
}
|
| 2259 |
+
} else {
|
| 2260 |
+
// There is no valid key being given back to us... We must get
|
| 2261 |
+
// rid of the ___noise___ or we will have a problem.
|
| 2262 |
+
$text = substr($text, 0, $pos)
|
| 2263 |
+
. 'NO NUMERIC NOISE KEY'
|
| 2264 |
+
. substr($text, $pos + 11);
|
| 2265 |
+
}
|
| 2266 |
+
}
|
| 2267 |
+
return $text;
|
| 2268 |
+
}
|
| 2269 |
+
|
| 2270 |
+
function search_noise($text)
|
| 2271 |
+
{
|
| 2272 |
+
global $debug_object;
|
| 2273 |
+
if (is_object($debug_object)) { $debug_object->debug_log_entry(1); }
|
| 2274 |
+
|
| 2275 |
+
foreach($this->noise as $noiseElement) {
|
| 2276 |
+
if (strpos($noiseElement, $text) !== false) {
|
| 2277 |
+
return $noiseElement;
|
| 2278 |
+
}
|
| 2279 |
+
}
|
| 2280 |
+
}
|
| 2281 |
+
|
| 2282 |
+
function __toString()
|
| 2283 |
+
{
|
| 2284 |
+
return $this->root->innertext();
|
| 2285 |
+
}
|
| 2286 |
+
|
| 2287 |
+
function __get($name)
|
| 2288 |
+
{
|
| 2289 |
+
switch ($name) {
|
| 2290 |
+
case 'outertext':
|
| 2291 |
+
return $this->root->innertext();
|
| 2292 |
+
case 'innertext':
|
| 2293 |
+
return $this->root->innertext();
|
| 2294 |
+
case 'plaintext':
|
| 2295 |
+
return $this->root->text();
|
| 2296 |
+
case 'charset':
|
| 2297 |
+
return $this->_charset;
|
| 2298 |
+
case 'target_charset':
|
| 2299 |
+
return $this->_target_charset;
|
| 2300 |
+
}
|
| 2301 |
+
}
|
| 2302 |
+
|
| 2303 |
+
function childNodes($idx = -1)
|
| 2304 |
+
{
|
| 2305 |
+
return $this->root->childNodes($idx);
|
| 2306 |
+
}
|
| 2307 |
+
|
| 2308 |
+
function firstChild()
|
| 2309 |
+
{
|
| 2310 |
+
return $this->root->first_child();
|
| 2311 |
+
}
|
| 2312 |
+
|
| 2313 |
+
function lastChild()
|
| 2314 |
+
{
|
| 2315 |
+
return $this->root->last_child();
|
| 2316 |
+
}
|
| 2317 |
+
|
| 2318 |
+
function createElement($name, $value = null)
|
| 2319 |
+
{
|
| 2320 |
+
return @str_get_html("<$name>$value</$name>")->firstChild();
|
| 2321 |
+
}
|
| 2322 |
+
|
| 2323 |
+
function createTextNode($value)
|
| 2324 |
+
{
|
| 2325 |
+
return @end(str_get_html($value)->nodes);
|
| 2326 |
+
}
|
| 2327 |
+
|
| 2328 |
+
function getElementById($id)
|
| 2329 |
+
{
|
| 2330 |
+
return $this->find("#$id", 0);
|
| 2331 |
+
}
|
| 2332 |
+
|
| 2333 |
+
function getElementsById($id, $idx = null)
|
| 2334 |
+
{
|
| 2335 |
+
return $this->find("#$id", $idx);
|
| 2336 |
+
}
|
| 2337 |
+
|
| 2338 |
+
function getElementByTagName($name)
|
| 2339 |
+
{
|
| 2340 |
+
return $this->find($name, 0);
|
| 2341 |
+
}
|
| 2342 |
+
|
| 2343 |
+
function getElementsByTagName($name, $idx = -1)
|
| 2344 |
+
{
|
| 2345 |
+
return $this->find($name, $idx);
|
| 2346 |
+
}
|
| 2347 |
+
|
| 2348 |
+
function loadFile()
|
| 2349 |
+
{
|
| 2350 |
+
$args = func_get_args();
|
| 2351 |
+
$this->load_file($args);
|
| 2352 |
+
}
|
| 2353 |
+
}
|
