dimaslanjaka /
universal-framework
| Total Complexity | 322 |
| Total Lines | 1500 |
| Duplicated Lines | 0 % |
| Changes | 0 | ||
Complex classes like HtmlNode often do a lot of different things. To break such a class down, we need to identify a cohesive component within that class. A common approach to find such a component is to look for fields/methods that share the same prefixes, or suffixes.
Once you have determined the fields that belong together, you can apply the Extract Class refactoring. If the component makes sense as a sub-class, Extract Subclass is also a candidate, and is often faster.
While breaking up the class, it is a good idea to analyze how other classes use HtmlNode, and based on these observations, apply Extract Interface, too.
| 1 | <?php |
||
| 38 | class HtmlNode |
||
| 39 | { |
||
| 40 | const HDOM_TYPE_ELEMENT = 1; |
||
| 41 | const HDOM_TYPE_COMMENT = 2; |
||
| 42 | const HDOM_TYPE_TEXT = 3; |
||
| 43 | const HDOM_TYPE_ROOT = 5; |
||
| 44 | const HDOM_TYPE_UNKNOWN = 6; |
||
| 45 | const HDOM_TYPE_CDATA = 7; |
||
| 46 | |||
| 47 | const HDOM_QUOTE_DOUBLE = 0; |
||
| 48 | const HDOM_QUOTE_SINGLE = 1; |
||
| 49 | const HDOM_QUOTE_NO = 3; |
||
| 50 | |||
| 51 | const HDOM_INFO_BEGIN = 0; |
||
| 52 | const HDOM_INFO_END = 1; |
||
| 53 | const HDOM_INFO_QUOTE = 2; |
||
| 54 | const HDOM_INFO_SPACE = 3; |
||
| 55 | const HDOM_INFO_TEXT = 4; |
||
| 56 | const HDOM_INFO_INNER = 5; |
||
| 57 | const HDOM_INFO_OUTER = 6; |
||
| 58 | const HDOM_INFO_ENDSPACE = 7; |
||
| 59 | |||
| 60 | public $nodetype = self::HDOM_TYPE_TEXT; |
||
| 61 | public $tag = 'text'; |
||
| 62 | public $attr = []; |
||
| 63 | public $children = []; |
||
| 64 | public $nodes = []; |
||
| 65 | public $parent = null; |
||
| 66 | public $_ = []; |
||
| 67 | private $dom = null; |
||
| 68 | |||
| 69 | public function __call($func, $args) |
||
| 70 | { |
||
| 71 | // Allow users to call methods with lower_case syntax |
||
| 72 | switch ($func) { |
||
| 73 | case 'children': |
||
| 74 | $actual_function = 'childNodes'; |
||
| 75 | break; |
||
| 76 | case 'first_child': |
||
| 77 | $actual_function = 'firstChild'; |
||
| 78 | break; |
||
| 79 | case 'has_child': |
||
| 80 | $actual_function = 'hasChildNodes'; |
||
| 81 | break; |
||
| 82 | case 'last_child': |
||
| 83 | $actual_function = 'lastChild'; |
||
| 84 | break; |
||
| 85 | case 'next_sibling': |
||
| 86 | $actual_function = 'nextSibling'; |
||
| 87 | break; |
||
| 88 | case 'prev_sibling': |
||
| 89 | $actual_function = 'previousSibling'; |
||
| 90 | break; |
||
| 91 | default: |
||
| 92 | trigger_error( |
||
| 93 | 'Call to undefined method ' . __CLASS__ . '::' . $func . '()', |
||
| 94 | E_USER_ERROR |
||
| 95 | ); |
||
| 96 | } |
||
| 97 | |||
| 98 | // phpcs:ignore Generic.Files.LineLength |
||
| 99 | Debug::log(__CLASS__ . '->' . $func . '() has been deprecated and will be removed in the next major version of simplehtmldom. Use ' . __CLASS__ . '->' . $actual_function . '() instead.'); |
||
|
Comprehensibility
Best Practice
introduced
by
|
|||
| 100 | |||
| 101 | return call_user_func_array([$this, $actual_function], $args); |
||
| 102 | } |
||
| 103 | |||
| 104 | public function __construct($dom) |
||
| 105 | { |
||
| 106 | if (null === $dom) { |
||
| 107 | return $this; |
||
| 108 | } |
||
| 109 | |||
| 110 | $this->dom = $dom; |
||
| 111 | $dom->nodes[] = $this; |
||
| 112 | } |
||
| 113 | |||
| 114 | public function __debugInfo() |
||
| 115 | { |
||
| 116 | // Translate node type to human-readable form |
||
| 117 | switch ($this->nodetype) { |
||
| 118 | case self::HDOM_TYPE_ELEMENT: |
||
| 119 | $nodetype = "HDOM_TYPE_ELEMENT ($this->nodetype)"; |
||
| 120 | break; |
||
| 121 | case self::HDOM_TYPE_COMMENT: |
||
| 122 | $nodetype = "HDOM_TYPE_COMMENT ($this->nodetype)"; |
||
| 123 | break; |
||
| 124 | case self::HDOM_TYPE_TEXT: |
||
| 125 | $nodetype = "HDOM_TYPE_TEXT ($this->nodetype)"; |
||
| 126 | break; |
||
| 127 | case self::HDOM_TYPE_ROOT: |
||
| 128 | $nodetype = "HDOM_TYPE_ROOT ($this->nodetype)"; |
||
| 129 | break; |
||
| 130 | case self::HDOM_TYPE_CDATA: |
||
| 131 | $nodetype = "HDOM_TYPE_CDATA ($this->nodetype)"; |
||
| 132 | break; |
||
| 133 | case self::HDOM_TYPE_UNKNOWN: |
||
| 134 | default: |
||
| 135 | $nodetype = "HDOM_TYPE_UNKNOWN ($this->nodetype)"; |
||
| 136 | } |
||
| 137 | |||
| 138 | return [ |
||
| 139 | 'nodetype' => $nodetype, |
||
| 140 | 'tag' => $this->tag, |
||
| 141 | 'attributes' => empty($this->attr) ? 'none' : $this->attr, |
||
| 142 | 'nodes' => empty($this->nodes) ? 'none' : $this->nodes, |
||
| 143 | ]; |
||
| 144 | } |
||
| 145 | |||
| 146 | public function __toString() |
||
| 147 | { |
||
| 148 | return $this->outertext(); |
||
| 149 | } |
||
| 150 | |||
| 151 | public function clear() |
||
| 152 | { |
||
| 153 | unset($this->dom, $this->parent); // Break link to origin |
||
| 154 | // Break link to branch |
||
| 155 | } |
||
| 156 | |||
| 157 | /** @codeCoverageIgnore */ |
||
| 158 | public function dump($show_attr = true, $depth = 0) |
||
| 159 | { |
||
| 160 | echo str_repeat("\t", $depth) . $this->tag; |
||
| 161 | |||
| 162 | if ($show_attr && count($this->attr) > 0) { |
||
| 163 | echo '('; |
||
| 164 | foreach ($this->attr as $k => $v) { |
||
| 165 | echo "[$k]=>\"$v\", "; |
||
| 166 | } |
||
| 167 | echo ')'; |
||
| 168 | } |
||
| 169 | |||
| 170 | echo "\n"; |
||
| 171 | |||
| 172 | if ($this->nodes) { |
||
| 173 | foreach ($this->nodes as $node) { |
||
| 174 | $node->dump($show_attr, $depth + 1); |
||
| 175 | } |
||
| 176 | } |
||
| 177 | } |
||
| 178 | |||
| 179 | /** @codeCoverageIgnore */ |
||
| 180 | public function dump_node($echo = true) |
||
| 181 | { |
||
| 182 | $string = $this->tag; |
||
| 183 | |||
| 184 | if (count($this->attr) > 0) { |
||
| 185 | $string .= '('; |
||
| 186 | foreach ($this->attr as $k => $v) { |
||
| 187 | $string .= "[$k]=>\"$v\", "; |
||
| 188 | } |
||
| 189 | $string .= ')'; |
||
| 190 | } |
||
| 191 | |||
| 192 | if (count($this->_) > 0) { |
||
| 193 | $string .= ' $_ ('; |
||
| 194 | foreach ($this->_ as $k => $v) { |
||
| 195 | if (is_array($v)) { |
||
| 196 | $string .= "[$k]=>("; |
||
| 197 | foreach ($v as $k2 => $v2) { |
||
| 198 | $string .= "[$k2]=>\"$v2\", "; |
||
| 199 | } |
||
| 200 | $string .= ')'; |
||
| 201 | } else { |
||
| 202 | $string .= "[$k]=>\"$v\", "; |
||
| 203 | } |
||
| 204 | } |
||
| 205 | $string .= ')'; |
||
| 206 | } |
||
| 207 | |||
| 208 | if (isset($this->text)) { |
||
| 209 | $string .= " text: ({$this->text})"; |
||
| 210 | } |
||
| 211 | |||
| 212 | $string .= ' HDOM_INNER_INFO: '; |
||
| 213 | /** |
||
| 214 | * @var mixed |
||
| 215 | */ |
||
| 216 | if (isset($node)) { |
||
| 217 | if (isset($node->_[self::HDOM_INFO_INNER])) { |
||
| 218 | $string .= "'" . $node->_[self::HDOM_INFO_INNER] . "'"; |
||
| 219 | } else { |
||
| 220 | $string .= ' NULL '; |
||
| 221 | } |
||
| 222 | } |
||
| 223 | |||
| 224 | $string .= ' children: ' . count($this->children); |
||
| 225 | $string .= ' nodes: ' . count($this->nodes); |
||
| 226 | $string .= "\n"; |
||
| 227 | |||
| 228 | if ($echo) { |
||
| 229 | echo $string; |
||
| 230 | |||
| 231 | return; |
||
| 232 | } else { |
||
| 233 | return $string; |
||
| 234 | } |
||
| 235 | } |
||
| 236 | |||
| 237 | public function parent($parent = null) |
||
| 238 | { |
||
| 239 | // I am SURE that this doesn't work properly. |
||
| 240 | // It fails to unset the current node from it's current parents nodes or |
||
| 241 | // children list first. |
||
| 242 | if (null !== $parent) { |
||
| 243 | $this->parent = $parent; |
||
| 244 | $this->parent->nodes[] = $this; |
||
| 245 | $this->parent->children[] = $this; |
||
| 246 | } |
||
| 247 | |||
| 248 | return $this->parent; |
||
| 249 | } |
||
| 250 | |||
| 251 | public function find_ancestor_tag($tag) |
||
| 252 | { |
||
| 253 | if (null === $this->parent) { |
||
| 254 | return null; |
||
| 255 | } |
||
| 256 | |||
| 257 | $ancestor = $this->parent; |
||
| 258 | |||
| 259 | while (!is_null($ancestor)) { |
||
| 260 | if ($ancestor->tag === $tag) { |
||
| 261 | break; |
||
| 262 | } |
||
| 263 | |||
| 264 | $ancestor = $ancestor->parent; |
||
| 265 | } |
||
| 266 | |||
| 267 | return $ancestor; |
||
| 268 | } |
||
| 269 | |||
| 270 | public function innertext() |
||
| 271 | { |
||
| 272 | if (isset($this->_[self::HDOM_INFO_INNER])) { |
||
| 273 | $ret = $this->_[self::HDOM_INFO_INNER]; |
||
| 274 | } elseif (isset($this->_[self::HDOM_INFO_TEXT])) { |
||
| 275 | $ret = $this->_[self::HDOM_INFO_TEXT]; |
||
| 276 | } else { |
||
| 277 | $ret = ''; |
||
| 278 | } |
||
| 279 | |||
| 280 | foreach ($this->nodes as $n) { |
||
| 281 | $ret .= $n->outertext(); |
||
| 282 | } |
||
| 283 | |||
| 284 | return $this->convert_text($ret); |
||
| 285 | } |
||
| 286 | |||
| 287 | public function outertext() |
||
| 288 | { |
||
| 289 | if ('root' === $this->tag) { |
||
| 290 | return $this->innertext(); |
||
| 291 | } |
||
| 292 | |||
| 293 | // todo: What is the use of this callback? Remove? |
||
| 294 | if ($this->dom && null !== $this->dom->callback) { |
||
| 295 | call_user_func_array($this->dom->callback, [$this]); |
||
| 296 | } |
||
| 297 | |||
| 298 | if (isset($this->_[self::HDOM_INFO_OUTER])) { |
||
| 299 | return $this->convert_text($this->_[self::HDOM_INFO_OUTER]); |
||
| 300 | } |
||
| 301 | |||
| 302 | if (isset($this->_[self::HDOM_INFO_TEXT])) { |
||
| 303 | return $this->convert_text($this->_[self::HDOM_INFO_TEXT]); |
||
| 304 | } |
||
| 305 | |||
| 306 | $ret = ''; |
||
| 307 | |||
| 308 | if (isset($this->_[self::HDOM_INFO_BEGIN])) { |
||
| 309 | $ret = $this->makeup(); |
||
| 310 | } |
||
| 311 | |||
| 312 | if (isset($this->_[self::HDOM_INFO_INNER])) { |
||
| 313 | // todo: <br> should either never have self::HDOM_INFO_INNER or always |
||
| 314 | if ('br' !== $this->tag) { |
||
| 315 | $ret .= $this->_[self::HDOM_INFO_INNER]; |
||
| 316 | } |
||
| 317 | } |
||
| 318 | |||
| 319 | if ($this->nodes) { |
||
| 320 | foreach ($this->nodes as $n) { |
||
| 321 | $ret .= $n->outertext(); |
||
| 322 | } |
||
| 323 | } |
||
| 324 | |||
| 325 | if (isset($this->_[self::HDOM_INFO_END]) && 0 != $this->_[self::HDOM_INFO_END]) { |
||
| 326 | $ret .= '</' . $this->tag . '>'; |
||
| 327 | } |
||
| 328 | |||
| 329 | return $this->convert_text($ret); |
||
| 330 | } |
||
| 331 | |||
| 332 | /** |
||
| 333 | * Returns true if the provided element is a block level element. |
||
| 334 | * |
||
| 335 | * @see https://www.w3resource.com/html/HTML-block-level-and-inline-elements.php |
||
| 336 | */ |
||
| 337 | protected function is_block_element($node) |
||
| 338 | { |
||
| 339 | // todo: When we have the utility class this should be moved there |
||
| 340 | return in_array(strtolower($node->tag), [ |
||
| 341 | 'p', |
||
| 342 | 'h1', 'h2', 'h3', 'h4', 'h5', 'h6', |
||
| 343 | 'ol', 'ul', |
||
| 344 | 'pre', |
||
| 345 | 'address', |
||
| 346 | 'blockquote', |
||
| 347 | 'dl', |
||
| 348 | 'div', |
||
| 349 | 'fieldset', |
||
| 350 | 'form', |
||
| 351 | 'hr', |
||
| 352 | 'noscript', |
||
| 353 | 'table', |
||
| 354 | ]); |
||
| 355 | } |
||
| 356 | |||
| 357 | /** |
||
| 358 | * Returns true if the provided element is an inline level element. |
||
| 359 | * |
||
| 360 | * @see https://www.w3resource.com/html/HTML-block-level-and-inline-elements.php |
||
| 361 | */ |
||
| 362 | protected function is_inline_element($node) |
||
| 363 | { |
||
| 364 | // todo: When we have the utility class this should be moved there |
||
| 365 | return in_array(strtolower($node->tag), [ |
||
| 366 | 'b', 'big', 'i', 'small', 'tt', |
||
| 367 | 'abbr', 'acronym', 'cite', 'code', 'dfn', 'em', 'kbd', 'strong', 'samp', 'var', |
||
| 368 | 'a', 'bdo', 'br', 'img', 'map', 'object', 'q', 'script', 'span', 'sub', 'sup', |
||
| 369 | 'button', 'input', 'label', 'select', 'textarea', |
||
| 370 | ]); |
||
| 371 | } |
||
| 372 | |||
| 373 | public function text($trim = true) |
||
| 374 | { |
||
| 375 | $ret = ''; |
||
| 376 | |||
| 377 | if ('script' === strtolower($this->tag)) { |
||
| 378 | $ret = ''; |
||
| 379 | } elseif ('style' === strtolower($this->tag)) { |
||
| 380 | $ret = ''; |
||
| 381 | } elseif (self::HDOM_TYPE_COMMENT === $this->nodetype) { |
||
| 382 | $ret = ''; |
||
| 383 | } elseif (self::HDOM_TYPE_CDATA === $this->nodetype) { |
||
| 384 | $ret = $this->_[self::HDOM_INFO_INNER]; |
||
| 385 | } elseif (self::HDOM_TYPE_UNKNOWN === $this->nodetype) { |
||
| 386 | $ret = ''; |
||
| 387 | } elseif (isset($this->_[self::HDOM_INFO_INNER])) { |
||
| 388 | $ret = $this->_[self::HDOM_INFO_INNER]; |
||
| 389 | } elseif (self::HDOM_TYPE_TEXT === $this->nodetype) { |
||
| 390 | $ret = $this->_[self::HDOM_INFO_TEXT]; |
||
| 391 | } |
||
| 392 | |||
| 393 | if (is_null($this->nodes)) { |
||
| 394 | return ''; |
||
| 395 | } |
||
| 396 | |||
| 397 | foreach ($this->nodes as $n) { |
||
| 398 | if ($this->is_block_element($n)) { |
||
| 399 | $block = ltrim($this->convert_text($n->text(false))); |
||
| 400 | |||
| 401 | if (empty($block)) { |
||
| 402 | continue; |
||
| 403 | } |
||
| 404 | |||
| 405 | $ret = rtrim($ret) . "\n\n" . $block; |
||
| 406 | } elseif ($this->is_inline_element($n)) { |
||
| 407 | // todo: <br> introduces code smell because no space but \n |
||
| 408 | if ('br' === strtolower($n->tag)) { |
||
| 409 | $ret .= $this->dom->default_br_text ?: DEFAULT_BR_TEXT; |
||
| 410 | } else { |
||
| 411 | $inline = ltrim($this->convert_text($n->text(false))); |
||
| 412 | |||
| 413 | if (empty($inline)) { |
||
| 414 | continue; |
||
| 415 | } |
||
| 416 | |||
| 417 | $ret = $ret . $this->convert_text($n->text(false)); |
||
| 418 | } |
||
| 419 | } else { |
||
| 420 | $ret .= $this->convert_text($n->text(false)); |
||
| 421 | } |
||
| 422 | } |
||
| 423 | |||
| 424 | // Reduce whitespace at start/end to a single (or none) space |
||
| 425 | $ret = preg_replace('/[ \t\n\r\0\x0B\xC2\xA0]+$/u', $trim ? '' : ' ', $ret); |
||
| 426 | $ret = preg_replace('/^[ \t\n\r\0\x0B\xC2\xA0]+/u', $trim ? '' : ' ', $ret); |
||
| 427 | |||
| 428 | return $ret; |
||
| 429 | } |
||
| 430 | |||
| 431 | public function xmltext() |
||
| 432 | { |
||
| 433 | $ret = $this->innertext(); |
||
| 434 | $ret = str_ireplace('<![CDATA[', '', $ret); |
||
| 435 | $ret = str_replace(']]>', '', $ret); |
||
| 436 | |||
| 437 | return $ret; |
||
| 438 | } |
||
| 439 | |||
| 440 | public function makeup() |
||
| 441 | { |
||
| 442 | // text, comment, unknown |
||
| 443 | if (isset($this->_[self::HDOM_INFO_TEXT])) { |
||
| 444 | return $this->_[self::HDOM_INFO_TEXT]; |
||
| 445 | } |
||
| 446 | |||
| 447 | $ret = '<' . $this->tag; |
||
| 448 | |||
| 449 | foreach ($this->attr as $key => $val) { |
||
| 450 | // skip removed attribute |
||
| 451 | if (null === $val || false === $val) { |
||
| 452 | continue; |
||
| 453 | } |
||
| 454 | |||
| 455 | if (isset($this->_[self::HDOM_INFO_SPACE][$key])) { |
||
| 456 | $ret .= $this->_[self::HDOM_INFO_SPACE][$key][0]; |
||
| 457 | } else { |
||
| 458 | $ret .= ' '; |
||
| 459 | } |
||
| 460 | |||
| 461 | //no value attr: nowrap, checked selected... |
||
| 462 | if (true === $val) { |
||
| 463 | $ret .= $key; |
||
| 464 | } else { |
||
| 465 | if (isset($this->_[self::HDOM_INFO_QUOTE][$key])) { |
||
| 466 | $quote_type = $this->_[self::HDOM_INFO_QUOTE][$key]; |
||
| 467 | } else { |
||
| 468 | $quote_type = self::HDOM_QUOTE_DOUBLE; |
||
| 469 | } |
||
| 470 | |||
| 471 | switch ($quote_type) { |
||
| 472 | case self::HDOM_QUOTE_SINGLE: |
||
| 473 | $quote = '\''; |
||
| 474 | $val = htmlentities($val, ENT_QUOTES, $this->dom->target_charset); |
||
| 475 | break; |
||
| 476 | case self::HDOM_QUOTE_NO: |
||
| 477 | $quote = ''; |
||
| 478 | break; |
||
| 479 | case self::HDOM_QUOTE_DOUBLE: |
||
| 480 | default: |
||
| 481 | $quote = '"'; |
||
| 482 | $val = htmlentities($val, ENT_COMPAT, $this->dom->target_charset); |
||
| 483 | } |
||
| 484 | |||
| 485 | $ret .= $key |
||
| 486 | . (isset($this->_[self::HDOM_INFO_SPACE][$key]) ? $this->_[self::HDOM_INFO_SPACE][$key][1] : '') |
||
| 487 | . '=' |
||
| 488 | . (isset($this->_[self::HDOM_INFO_SPACE][$key]) ? $this->_[self::HDOM_INFO_SPACE][$key][2] : '') |
||
| 489 | . $quote |
||
| 490 | . $val |
||
| 491 | . $quote; |
||
| 492 | } |
||
| 493 | } |
||
| 494 | |||
| 495 | if (isset($this->_[self::HDOM_INFO_ENDSPACE])) { |
||
| 496 | $ret .= $this->_[self::HDOM_INFO_ENDSPACE]; |
||
| 497 | } |
||
| 498 | |||
| 499 | return $ret . '>'; |
||
| 500 | } |
||
| 501 | |||
| 502 | /** |
||
| 503 | * Element selector |
||
| 504 | * |
||
| 505 | * @param string $selector |
||
| 506 | * @param int $idx |
||
| 507 | * @param boolean $lowercase |
||
| 508 | * @return HtmlNode |
||
| 509 | */ |
||
| 510 | public function find($selector, $idx = null, $lowercase = false) |
||
| 511 | { |
||
| 512 | $selectors = $this->parse_selector($selector); |
||
| 513 | if (0 === ($count = count($selectors))) { |
||
| 514 | return []; |
||
| 515 | } |
||
| 516 | $found_keys = []; |
||
| 517 | |||
| 518 | // find each selector |
||
| 519 | for ($c = 0; $c < $count; ++$c) { |
||
| 520 | // The change on the below line was documented on the sourceforge |
||
| 521 | // code tracker id 2788009 |
||
| 522 | // used to be: if (($levle=count($selectors[0]))===0) return array(); |
||
| 523 | if (0 === ($levle = count($selectors[$c]))) { |
||
| 524 | Debug::log_once('Empty selector (' . $selector . ') matches nothing.'); |
||
| 525 | |||
| 526 | return []; |
||
| 527 | } |
||
| 528 | |||
| 529 | if (!isset($this->_[self::HDOM_INFO_BEGIN])) { |
||
| 530 | Debug::log_once('Invalid operation. The current node has no start tag.'); |
||
| 531 | |||
| 532 | return []; |
||
| 533 | } |
||
| 534 | |||
| 535 | $head = [$this->_[self::HDOM_INFO_BEGIN] => 1]; |
||
| 536 | $cmd = ' '; // Combinator |
||
| 537 | |||
| 538 | // handle descendant selectors, no recursive! |
||
| 539 | for ($l = 0; $l < $levle; ++$l) { |
||
| 540 | $ret = []; |
||
| 541 | |||
| 542 | foreach ($head as $k => $v) { |
||
| 543 | $n = (-1 === $k) ? $this->dom->root : $this->dom->nodes[$k]; |
||
| 544 | //PaperG - Pass this optional parameter on to the seek function. |
||
| 545 | $n->seek($selectors[$c][$l], $ret, $cmd, $lowercase); |
||
| 546 | } |
||
| 547 | |||
| 548 | $head = $ret; |
||
| 549 | $cmd = $selectors[$c][$l][6]; // Next Combinator |
||
| 550 | } |
||
| 551 | |||
| 552 | foreach ($head as $k => $v) { |
||
| 553 | if (!isset($found_keys[$k])) { |
||
| 554 | $found_keys[$k] = 1; |
||
| 555 | } |
||
| 556 | } |
||
| 557 | } |
||
| 558 | |||
| 559 | // sort keys |
||
| 560 | ksort($found_keys); |
||
| 561 | |||
| 562 | $found = []; |
||
| 563 | foreach ($found_keys as $k => $v) { |
||
| 564 | $found[] = $this->dom->nodes[$k]; |
||
| 565 | } |
||
| 566 | |||
| 567 | // return nth-element or array |
||
| 568 | if (is_null($idx)) { |
||
| 569 | return $found; |
||
| 570 | } elseif ($idx < 0) { |
||
| 571 | $idx = count($found) + $idx; |
||
| 572 | } |
||
| 573 | |||
| 574 | return (isset($found[$idx])) ? $found[$idx] : null; |
||
| 575 | } |
||
| 576 | |||
| 577 | public function expect($selector, $idx = null, $lowercase = false) |
||
| 578 | { |
||
| 579 | return $this->find($selector, $idx, $lowercase) ?: null; |
||
| 580 | } |
||
| 581 | |||
| 582 | protected function seek($selector, &$ret, $parent_cmd, $lowercase = false) |
||
| 583 | { |
||
| 584 | list($ps_selector, $tag, $ps_element, $id, $class, $attributes, $cmb) = $selector; |
||
| 585 | $nodes = []; |
||
| 586 | |||
| 587 | if (' ' === $parent_cmd) { // Descendant Combinator |
||
| 588 | // Find parent closing tag if the current element doesn't have a closing |
||
| 589 | // tag (i.e. void element) |
||
| 590 | $end = (!empty($this->_[self::HDOM_INFO_END])) ? $this->_[self::HDOM_INFO_END] : 0; |
||
| 591 | if (0 == $end && $this->parent) { |
||
| 592 | $parent = $this->parent; |
||
| 593 | while (null !== $parent && !isset($parent->_[self::HDOM_INFO_END])) { |
||
| 594 | --$end; |
||
| 595 | $parent = $parent->parent; |
||
| 596 | } |
||
| 597 | $end += $parent->_[self::HDOM_INFO_END]; |
||
| 598 | } |
||
| 599 | |||
| 600 | if (0 === $end) { |
||
| 601 | $end = count($this->dom->nodes); |
||
| 602 | } |
||
| 603 | |||
| 604 | // Get list of target nodes |
||
| 605 | $nodes_start = $this->_[self::HDOM_INFO_BEGIN] + 1; |
||
| 606 | |||
| 607 | // remove() makes $this->dom->nodes non-contiguous; use what is left. |
||
| 608 | $nodes = array_intersect_key( |
||
| 609 | $this->dom->nodes, |
||
| 610 | array_flip(range($nodes_start, $end)) |
||
| 611 | ); |
||
| 612 | } elseif ('>' === $parent_cmd) { // Child Combinator |
||
| 613 | $nodes = $this->children; |
||
| 614 | } elseif ( |
||
| 615 | '+' === $parent_cmd |
||
| 616 | && $this->parent |
||
| 617 | && in_array($this, $this->parent->children) |
||
| 618 | ) { // Next-Sibling Combinator |
||
| 619 | $index = array_search($this, $this->parent->children, true) + 1; |
||
| 620 | if ($index < count($this->parent->children)) { |
||
| 621 | $nodes[] = $this->parent->children[$index]; |
||
| 622 | } |
||
| 623 | } elseif ( |
||
| 624 | '~' === $parent_cmd |
||
| 625 | && $this->parent |
||
| 626 | && in_array($this, $this->parent->children) |
||
| 627 | ) { // Subsequent Sibling Combinator |
||
| 628 | $index = array_search($this, $this->parent->children, true); |
||
| 629 | $nodes = array_slice($this->parent->children, $index); |
||
| 630 | } |
||
| 631 | |||
| 632 | // Go throgh each element starting at this element until the end tag |
||
| 633 | // Note: If this element is a void tag, any previous void element is |
||
| 634 | // skipped. |
||
| 635 | foreach ($nodes as $node) { |
||
| 636 | $pass = true; |
||
| 637 | |||
| 638 | // Skip root nodes |
||
| 639 | if (!$node->parent) { |
||
| 640 | unset($node); |
||
| 641 | continue; |
||
| 642 | } |
||
| 643 | |||
| 644 | // Handle 'text' selector |
||
| 645 | if ($pass && 'text' === $tag) { |
||
| 646 | if ('text' === $node->tag) { |
||
| 647 | $ret[array_search($node, $this->dom->nodes, true)] = 1; |
||
| 648 | } |
||
| 649 | |||
| 650 | if (isset($node->_[self::HDOM_INFO_INNER])) { |
||
| 651 | $ret[$node->_[self::HDOM_INFO_BEGIN]] = 1; |
||
| 652 | } |
||
| 653 | |||
| 654 | unset($node); |
||
| 655 | continue; |
||
| 656 | } |
||
| 657 | |||
| 658 | // Handle 'cdata' selector |
||
| 659 | if ($pass && 'cdata' === $tag) { |
||
| 660 | if ('cdata' === $node->tag) { |
||
| 661 | $ret[$node->_[self::HDOM_INFO_BEGIN]] = 1; |
||
| 662 | } |
||
| 663 | |||
| 664 | unset($node); |
||
| 665 | continue; |
||
| 666 | } |
||
| 667 | |||
| 668 | // Handle 'comment' |
||
| 669 | if ($pass && 'comment' === $tag && 'comment' === $node->tag) { |
||
| 670 | $ret[$node->_[self::HDOM_INFO_BEGIN]] = 1; |
||
| 671 | unset($node); |
||
| 672 | continue; |
||
| 673 | } |
||
| 674 | |||
| 675 | // Skip if node isn't a child node (i.e. text nodes) |
||
| 676 | if ($pass && !in_array($node, $node->parent->children, true)) { |
||
| 677 | unset($node); |
||
| 678 | continue; |
||
| 679 | } |
||
| 680 | |||
| 681 | // Skip if tag doesn't match |
||
| 682 | if ($pass && '' !== $tag && $tag !== $node->tag && '*' !== $tag) { |
||
| 683 | $pass = false; |
||
| 684 | } |
||
| 685 | |||
| 686 | // Skip if ID doesn't exist |
||
| 687 | if ($pass && '' !== $id && !isset($node->attr['id'])) { |
||
| 688 | $pass = false; |
||
| 689 | } |
||
| 690 | |||
| 691 | // Check if ID matches |
||
| 692 | if ($pass && '' !== $id && isset($node->attr['id'])) { |
||
| 693 | // Note: Only consider the first ID (as browsers do) |
||
| 694 | $node_id = explode(' ', trim($node->attr['id']))[0]; |
||
| 695 | |||
| 696 | if ($id !== $node_id) { |
||
| 697 | $pass = false; |
||
| 698 | } |
||
| 699 | } |
||
| 700 | |||
| 701 | // Check if all class(es) exist |
||
| 702 | if ($pass && '' !== $class && is_array($class) && !empty($class)) { |
||
| 703 | if (isset($node->attr['class'])) { |
||
| 704 | // Apply the same rules for the pattern and attribute value |
||
| 705 | // Attribute values must not contain control characters other than space |
||
| 706 | // https://www.w3.org/TR/html/dom.html#text-content |
||
| 707 | // https://www.w3.org/TR/html/syntax.html#attribute-values |
||
| 708 | // https://www.w3.org/TR/xml/#AVNormalize |
||
| 709 | $node_classes = preg_replace("/[\r\n\t\s]+/u", ' ', $node->attr['class']); |
||
| 710 | $node_classes = trim($node_classes); |
||
| 711 | $node_classes = explode(' ', $node_classes); |
||
| 712 | |||
| 713 | if ($lowercase) { |
||
| 714 | $node_classes = array_map('strtolower', $node_classes); |
||
| 715 | } |
||
| 716 | |||
| 717 | foreach ($class as $c) { |
||
| 718 | if (!in_array($c, $node_classes)) { |
||
| 719 | $pass = false; |
||
| 720 | break; |
||
| 721 | } |
||
| 722 | } |
||
| 723 | } else { |
||
| 724 | $pass = false; |
||
| 725 | } |
||
| 726 | } |
||
| 727 | |||
| 728 | // Check attributes |
||
| 729 | if ( |
||
| 730 | $pass |
||
| 731 | && '' !== $attributes |
||
| 732 | && is_array($attributes) |
||
| 733 | && !empty($attributes) |
||
| 734 | ) { |
||
| 735 | foreach ($attributes as $a) { |
||
| 736 | list( |
||
| 737 | $att_name, |
||
| 738 | $att_expr, |
||
| 739 | $att_val, |
||
| 740 | $att_inv, |
||
| 741 | $att_case_sensitivity |
||
| 742 | ) = $a; |
||
| 743 | |||
| 744 | // Handle indexing attributes (i.e. "[2]") |
||
| 745 | /* |
||
| 746 | * Note: This is not supported by the CSS Standard but adds |
||
| 747 | * the ability to select items compatible to XPath (i.e. |
||
| 748 | * the 3rd element within it's parent). |
||
| 749 | * |
||
| 750 | * Note: This doesn't conflict with the CSS Standard which |
||
| 751 | * doesn't work on numeric attributes anyway. |
||
| 752 | */ |
||
| 753 | if ( |
||
| 754 | is_numeric($att_name) |
||
| 755 | && '' === $att_expr |
||
| 756 | && '' === $att_val |
||
| 757 | ) { |
||
| 758 | $count = 0; |
||
| 759 | |||
| 760 | // Find index of current element in parent |
||
| 761 | foreach ($node->parent->children as $c) { |
||
| 762 | if ($c->tag === $node->tag) { |
||
| 763 | ++$count; |
||
| 764 | } |
||
| 765 | if ($c === $node) { |
||
| 766 | break; |
||
| 767 | } |
||
| 768 | } |
||
| 769 | |||
| 770 | // If this is the correct node, continue with next |
||
| 771 | // attribute |
||
| 772 | if ($count === (int) $att_name) { |
||
| 773 | continue; |
||
| 774 | } |
||
| 775 | } |
||
| 776 | |||
| 777 | // Check attribute availability |
||
| 778 | if ($att_inv) { // Attribute should NOT be set |
||
| 779 | if (isset($node->attr[$att_name])) { |
||
| 780 | $pass = false; |
||
| 781 | break; |
||
| 782 | } |
||
| 783 | } else { // Attribute should be set |
||
| 784 | // todo: "plaintext" is not a valid CSS selector! |
||
| 785 | if ( |
||
| 786 | 'plaintext' !== $att_name |
||
| 787 | && !isset($node->attr[$att_name]) |
||
| 788 | ) { |
||
| 789 | $pass = false; |
||
| 790 | break; |
||
| 791 | } |
||
| 792 | } |
||
| 793 | |||
| 794 | // Continue with next attribute if expression isn't defined |
||
| 795 | if ('' === $att_expr) { |
||
| 796 | continue; |
||
| 797 | } |
||
| 798 | |||
| 799 | // If they have told us that this is a "plaintext" |
||
| 800 | // search then we want the plaintext of the node - right? |
||
| 801 | // todo "plaintext" is not a valid CSS selector! |
||
| 802 | if ('plaintext' === $att_name) { |
||
| 803 | $nodeKeyValue = $node->text(); |
||
| 804 | } else { |
||
| 805 | $nodeKeyValue = $node->attr[$att_name]; |
||
| 806 | } |
||
| 807 | |||
| 808 | // If lowercase is set, do a case insensitive test of |
||
| 809 | // the value of the selector. |
||
| 810 | if ($lowercase) { |
||
| 811 | $check = $this->match( |
||
| 812 | $att_expr, |
||
| 813 | strtolower($att_val), |
||
| 814 | strtolower($nodeKeyValue), |
||
| 815 | $att_case_sensitivity |
||
| 816 | ); |
||
| 817 | } else { |
||
| 818 | $check = $this->match( |
||
| 819 | $att_expr, |
||
| 820 | $att_val, |
||
| 821 | $nodeKeyValue, |
||
| 822 | $att_case_sensitivity |
||
| 823 | ); |
||
| 824 | } |
||
| 825 | |||
| 826 | $check = 'not' === $ps_element ? !$check : $check; |
||
| 827 | |||
| 828 | if (!$check) { |
||
| 829 | $pass = false; |
||
| 830 | break; |
||
| 831 | } |
||
| 832 | } |
||
| 833 | } |
||
| 834 | |||
| 835 | // Found a match. Add to list and clear node |
||
| 836 | $pass = 'not' === $ps_selector ? !$pass : $pass; |
||
| 837 | if ($pass) { |
||
| 838 | $ret[$node->_[self::HDOM_INFO_BEGIN]] = 1; |
||
| 839 | } |
||
| 840 | unset($node); |
||
| 841 | } |
||
| 842 | } |
||
| 843 | |||
| 844 | protected function match($exp, $pattern, $value, $case_sensitivity) |
||
| 845 | { |
||
| 846 | if ('i' === $case_sensitivity) { |
||
| 847 | $pattern = strtolower($pattern); |
||
| 848 | $value = strtolower($value); |
||
| 849 | } |
||
| 850 | |||
| 851 | // Apply the same rules for the pattern and attribute value |
||
| 852 | // Attribute values must not contain control characters other than space |
||
| 853 | // https://www.w3.org/TR/html/dom.html#text-content |
||
| 854 | // https://www.w3.org/TR/html/syntax.html#attribute-values |
||
| 855 | // https://www.w3.org/TR/xml/#AVNormalize |
||
| 856 | $pattern = preg_replace("/[\r\n\t\s]+/u", ' ', $pattern); |
||
| 857 | $pattern = trim($pattern); |
||
| 858 | |||
| 859 | $value = preg_replace("/[\r\n\t\s]+/u", ' ', $value); |
||
| 860 | $value = trim($value); |
||
| 861 | |||
| 862 | switch ($exp) { |
||
| 863 | case '=': |
||
| 864 | return $value === $pattern; |
||
| 865 | case '!=': |
||
| 866 | return $value !== $pattern; |
||
| 867 | case '^=': |
||
| 868 | return preg_match('/^' . preg_quote($pattern, '/') . '/', $value); |
||
| 869 | case '$=': |
||
| 870 | return preg_match('/' . preg_quote($pattern, '/') . '$/', $value); |
||
| 871 | case '*=': |
||
| 872 | return preg_match('/' . preg_quote($pattern, '/') . '/', $value); |
||
| 873 | case '|=': |
||
| 874 | /* |
||
| 875 | * [att|=val] |
||
| 876 | * |
||
| 877 | * Represents an element with the att attribute, its value |
||
| 878 | * either being exactly "val" or beginning with "val" |
||
| 879 | * immediately followed by "-" (U+002D). |
||
| 880 | */ |
||
| 881 | return 0 === strpos($value, $pattern); |
||
| 882 | case '~=': |
||
| 883 | /* |
||
| 884 | * [att~=val] |
||
| 885 | * |
||
| 886 | * Represents an element with the att attribute whose value is a |
||
| 887 | * whitespace-separated list of words, one of which is exactly |
||
| 888 | * "val". If "val" contains whitespace, it will never represent |
||
| 889 | * anything (since the words are separated by spaces). Also if |
||
| 890 | * "val" is the empty string, it will never represent anything. |
||
| 891 | */ |
||
| 892 | return in_array($pattern, explode(' ', trim($value)), true); |
||
| 893 | } |
||
| 894 | |||
| 895 | Debug::log('Unhandled attribute selector: ' . $exp . '!'); |
||
| 896 | |||
| 897 | return false; |
||
| 898 | } |
||
| 899 | |||
| 900 | protected function parse_selector($selector_string) |
||
| 901 | { |
||
| 902 | /** |
||
| 903 | * Pattern of CSS selectors, modified from mootools (https://mootools.net/). |
||
| 904 | * |
||
| 905 | * Paperg: Add the colon to the attribute, so that it properly finds |
||
| 906 | * <tag attr:ibute="something" > like google does. |
||
| 907 | * |
||
| 908 | * Note: if you try to look at this attribute, you MUST use getAttribute |
||
| 909 | * since $dom->x:y will fail the php syntax check. |
||
| 910 | * |
||
| 911 | * Notice the \[ starting the attribute? and the @? following? This |
||
| 912 | * implies that an attribute can begin with an @ sign that is not |
||
| 913 | * captured. This implies that an html attribute specifier may start |
||
| 914 | * with an @ sign that is NOT captured by the expression. Farther study |
||
| 915 | * is required to determine of this should be documented or removed. |
||
| 916 | * |
||
| 917 | * Matches selectors in this order: |
||
| 918 | * |
||
| 919 | * [0] - full match |
||
| 920 | * |
||
| 921 | * [1] - pseudo selector |
||
| 922 | * (?:\:(\w+)\()? |
||
| 923 | * Matches the pseudo selector (optional) |
||
| 924 | * |
||
| 925 | * [2] - tag name |
||
| 926 | * ([\w:\*-]*) |
||
| 927 | * Matches the tag name consisting of zero or more words, colons, |
||
| 928 | * asterisks and hyphens. |
||
| 929 | * |
||
| 930 | * [3] - pseudo selector |
||
| 931 | * (?:\:(\w+)\()? |
||
| 932 | * Matches the pseudo selector (optional) |
||
| 933 | * |
||
| 934 | * [4] - id name |
||
| 935 | * (?:\#([\w-]+)) |
||
| 936 | * Optionally matches a id name, consisting of an "#" followed by |
||
| 937 | * the id name (one or more words and hyphens). |
||
| 938 | * |
||
| 939 | * [5] - class names (including dots) |
||
| 940 | * (?:\.([\w\.-]+))? |
||
| 941 | * Optionally matches a list of classs, consisting of an "." |
||
| 942 | * followed by the class name (one or more words and hyphens) |
||
| 943 | * where multiple classes can be chained (i.e. ".foo.bar.baz") |
||
| 944 | * |
||
| 945 | * [6] - attributes |
||
| 946 | * ((?:\[@?(?:!?[\w:-]+)(?:(?:[!*^$|~]?=)[\"']?(?:.*?)[\"']?)?(?:\s*?(?:[iIsS])?)?\])+)? |
||
| 947 | * Optionally matches the attributes list |
||
| 948 | * |
||
| 949 | * [7] - separator |
||
| 950 | * ([\/, >+~]+) |
||
| 951 | * Matches the selector list separator |
||
| 952 | */ |
||
| 953 | // phpcs:ignore Generic.Files.LineLength |
||
| 954 | $pattern = "/(?:\:(\w+)\()?([\w:\*-]*)(?:\:(\w+)\()?(?:\#([\w-]+))?(?:|\.([\w\.-]+))?((?:\[@?(?:!?[\w:-]+)(?:(?:[!*^$|~]?=)[\"']?(?:.*?)[\"']?)?(?:\s*?(?:[iIsS])?)?\])+)?(?:\))?(?:\))?([\/, >+~]+)/is"; |
||
| 955 | |||
| 956 | preg_match_all( |
||
| 957 | $pattern, |
||
| 958 | trim($selector_string) . ' ', // Add final ' ' as pseudo separator |
||
| 959 | $matches, |
||
| 960 | PREG_SET_ORDER |
||
| 961 | ); |
||
| 962 | |||
| 963 | $selectors = []; |
||
| 964 | $result = []; |
||
| 965 | |||
| 966 | foreach ($matches as $m) { |
||
| 967 | $m[0] = trim($m[0]); |
||
| 968 | |||
| 969 | // Skip NoOps |
||
| 970 | if ('' === $m[0] || '/' === $m[0] || '//' === $m[0]) { |
||
| 971 | continue; |
||
| 972 | } |
||
| 973 | |||
| 974 | array_shift($m); |
||
| 975 | |||
| 976 | // Convert to lowercase |
||
| 977 | if ($this->dom->lowercase) { |
||
| 978 | $m[1] = strtolower($m[1]); |
||
| 979 | } |
||
| 980 | |||
| 981 | // Extract classes |
||
| 982 | if ('' !== $m[4]) { |
||
| 983 | $m[4] = explode('.', $m[4]); |
||
| 984 | } |
||
| 985 | |||
| 986 | /* Extract attributes (pattern based on the pattern above!) |
||
| 987 | |||
| 988 | * [0] - full match |
||
| 989 | * [1] - attribute name |
||
| 990 | * [2] - attribute expression |
||
| 991 | * [3] - attribute value |
||
| 992 | * [4] - case sensitivity |
||
| 993 | * |
||
| 994 | * Note: Attributes can be negated with a "!" prefix to their name |
||
| 995 | */ |
||
| 996 | if ('' !== $m[5]) { |
||
| 997 | preg_match_all( |
||
| 998 | "/\[@?(!?[\w:-]+)(?:([!*^$|~]?=)[\"']?(.*?)[\"']?)?(?:\s+?([iIsS])?)?\]/is", |
||
| 999 | trim($m[5]), |
||
| 1000 | $attributes, |
||
| 1001 | PREG_SET_ORDER |
||
| 1002 | ); |
||
| 1003 | |||
| 1004 | // Replace element by array |
||
| 1005 | $m[5] = []; |
||
| 1006 | |||
| 1007 | foreach ($attributes as $att) { |
||
| 1008 | // Skip empty matches |
||
| 1009 | if ('' === trim($att[0])) { |
||
| 1010 | continue; |
||
| 1011 | } |
||
| 1012 | |||
| 1013 | $inverted = (isset($att[1][0]) && '!' === $att[1][0]); |
||
| 1014 | $m[5][] = [ |
||
| 1015 | $inverted ? substr($att[1], 1) : $att[1], // Name |
||
| 1016 | (isset($att[2])) ? $att[2] : '', // Expression |
||
| 1017 | (isset($att[3])) ? $att[3] : '', // Value |
||
| 1018 | $inverted, // Inverted Flag |
||
| 1019 | (isset($att[4])) ? strtolower($att[4]) : '', // Case-Sensitivity |
||
| 1020 | ]; |
||
| 1021 | } |
||
| 1022 | } |
||
| 1023 | |||
| 1024 | // Sanitize Separator |
||
| 1025 | if ('' !== $m[6] && '' === trim($m[6])) { // Descendant Separator |
||
| 1026 | $m[6] = ' '; |
||
| 1027 | } else { // Other Separator |
||
| 1028 | $m[6] = trim($m[6]); |
||
| 1029 | } |
||
| 1030 | |||
| 1031 | // Clear Separator if it's a Selector List |
||
| 1032 | if ($is_list = (',' === $m[6])) { |
||
| 1033 | $m[6] = ''; |
||
| 1034 | } |
||
| 1035 | |||
| 1036 | $result[] = $m; |
||
| 1037 | |||
| 1038 | if ($is_list) { // Selector List |
||
| 1039 | $selectors[] = $result; |
||
| 1040 | $result = []; |
||
| 1041 | } |
||
| 1042 | } |
||
| 1043 | |||
| 1044 | if (count($result) > 0) { |
||
| 1045 | $selectors[] = $result; |
||
| 1046 | } |
||
| 1047 | |||
| 1048 | return $selectors; |
||
| 1049 | } |
||
| 1050 | |||
| 1051 | public function __get($name) |
||
| 1052 | { |
||
| 1053 | if (isset($this->attr[$name])) { |
||
| 1054 | return $this->convert_text($this->attr[$name]); |
||
| 1055 | } |
||
| 1056 | |||
| 1057 | switch ($name) { |
||
| 1058 | case 'outertext': |
||
| 1059 | return $this->outertext(); |
||
| 1060 | case 'innertext': |
||
| 1061 | return $this->innertext(); |
||
| 1062 | case 'plaintext': |
||
| 1063 | return $this->text(); |
||
| 1064 | case 'xmltext': |
||
| 1065 | return $this->xmltext(); |
||
| 1066 | } |
||
| 1067 | |||
| 1068 | return false; |
||
| 1069 | } |
||
| 1070 | |||
| 1071 | public function __set($name, $value) |
||
| 1072 | { |
||
| 1073 | switch ($name) { |
||
| 1074 | case 'outertext': |
||
| 1075 | $this->_[self::HDOM_INFO_OUTER] = $value; |
||
| 1076 | break; |
||
| 1077 | case 'innertext': |
||
| 1078 | if (isset($this->_[self::HDOM_INFO_TEXT])) { |
||
| 1079 | $this->_[self::HDOM_INFO_TEXT] = ''; |
||
| 1080 | } |
||
| 1081 | $this->_[self::HDOM_INFO_INNER] = $value; |
||
| 1082 | break; |
||
| 1083 | default: |
||
| 1084 | $this->attr[$name] = $value; |
||
| 1085 | } |
||
| 1086 | } |
||
| 1087 | |||
| 1088 | public function __isset($name) |
||
| 1089 | { |
||
| 1090 | switch ($name) { |
||
| 1091 | case 'outertext': |
||
| 1092 | return true; |
||
| 1093 | case 'innertext': |
||
| 1094 | return true; |
||
| 1095 | case 'plaintext': |
||
| 1096 | return true; |
||
| 1097 | } |
||
| 1098 | |||
| 1099 | return isset($this->attr[$name]); |
||
| 1100 | } |
||
| 1101 | |||
| 1102 | public function __unset($name) |
||
| 1103 | { |
||
| 1104 | if (isset($this->attr[$name])) { |
||
| 1105 | unset($this->attr[$name]); |
||
| 1106 | } |
||
| 1107 | } |
||
| 1108 | |||
| 1109 | public function convert_text($text) |
||
| 1110 | { |
||
| 1111 | $converted_text = $text; |
||
| 1112 | |||
| 1113 | $sourceCharset = ''; |
||
| 1114 | $targetCharset = ''; |
||
| 1115 | |||
| 1116 | if ($this->dom) { |
||
| 1117 | $sourceCharset = strtoupper($this->dom->_charset); |
||
| 1118 | $targetCharset = strtoupper($this->dom->_target_charset); |
||
| 1119 | } |
||
| 1120 | |||
| 1121 | if (!empty($sourceCharset) && !empty($targetCharset)) { |
||
| 1122 | if (strtoupper($sourceCharset) === strtoupper($targetCharset)) { |
||
| 1123 | $converted_text = $text; |
||
| 1124 | } elseif (('UTF-8' === strtoupper($targetCharset)) && (self::is_utf8($text))) { |
||
| 1125 | Debug::log_once('The source charset was incorrectly detected as ' . $sourceCharset . ' but should have been UTF-8'); |
||
| 1126 | $converted_text = $text; |
||
| 1127 | } else { |
||
| 1128 | $converted_text = iconv($sourceCharset, $targetCharset, $text); |
||
| 1129 | } |
||
| 1130 | } |
||
| 1131 | |||
| 1132 | // Lets make sure that we don't have that silly BOM issue with any of the utf-8 text we output. |
||
| 1133 | if ('UTF-8' === $targetCharset) { |
||
| 1134 | if ("\xef\xbb\xbf" === substr($converted_text, 0, 3)) { |
||
| 1135 | $converted_text = substr($converted_text, 3); |
||
| 1136 | } |
||
| 1137 | |||
| 1138 | if ("\xef\xbb\xbf" === substr($converted_text, -3)) { |
||
| 1139 | $converted_text = substr($converted_text, 0, -3); |
||
| 1140 | } |
||
| 1141 | } |
||
| 1142 | |||
| 1143 | return $converted_text; |
||
| 1144 | } |
||
| 1145 | |||
| 1146 | public static function is_utf8($str) |
||
| 1147 | { |
||
| 1148 | $c = 0; |
||
| 1149 | $b = 0; |
||
| 1150 | $bits = 0; |
||
| 1151 | $len = strlen($str); |
||
| 1152 | for ($i = 0; $i < $len; ++$i) { |
||
| 1153 | $c = ord($str[$i]); |
||
| 1154 | if ($c > 128) { |
||
| 1155 | if (($c >= 254)) { |
||
| 1156 | return false; |
||
| 1157 | } elseif ($c >= 252) { |
||
| 1158 | $bits = 6; |
||
| 1159 | } elseif ($c >= 248) { |
||
| 1160 | $bits = 5; |
||
| 1161 | } elseif ($c >= 240) { |
||
| 1162 | $bits = 4; |
||
| 1163 | } elseif ($c >= 224) { |
||
| 1164 | $bits = 3; |
||
| 1165 | } elseif ($c >= 192) { |
||
| 1166 | $bits = 2; |
||
| 1167 | } else { |
||
| 1168 | return false; |
||
| 1169 | } |
||
| 1170 | if (($i + $bits) > $len) { |
||
| 1171 | return false; |
||
| 1172 | } |
||
| 1173 | while ($bits > 1) { |
||
| 1174 | ++$i; |
||
| 1175 | $b = ord($str[$i]); |
||
| 1176 | if ($b < 128 || $b > 191) { |
||
| 1177 | return false; |
||
| 1178 | } |
||
| 1179 | --$bits; |
||
| 1180 | } |
||
| 1181 | } |
||
| 1182 | } |
||
| 1183 | |||
| 1184 | return true; |
||
| 1185 | } |
||
| 1186 | |||
| 1187 | public function get_display_size() |
||
| 1188 | { |
||
| 1189 | $width = -1; |
||
| 1190 | $height = -1; |
||
| 1191 | |||
| 1192 | if ('img' !== $this->tag) { |
||
| 1193 | return false; |
||
| 1194 | } |
||
| 1195 | |||
| 1196 | // See if there is aheight or width attribute in the tag itself. |
||
| 1197 | if (isset($this->attr['width'])) { |
||
| 1198 | $width = $this->attr['width']; |
||
| 1199 | } |
||
| 1200 | |||
| 1201 | if (isset($this->attr['height'])) { |
||
| 1202 | $height = $this->attr['height']; |
||
| 1203 | } |
||
| 1204 | |||
| 1205 | // Now look for an inline style. |
||
| 1206 | if (isset($this->attr['style'])) { |
||
| 1207 | // Thanks to user gnarf from stackoverflow for this regular expression. |
||
| 1208 | $attributes = []; |
||
| 1209 | |||
| 1210 | preg_match_all( |
||
| 1211 | '/([\w-]+)\s*:\s*([^;]+)\s*;?/', |
||
| 1212 | $this->attr['style'], |
||
| 1213 | $matches, |
||
| 1214 | PREG_SET_ORDER |
||
| 1215 | ); |
||
| 1216 | |||
| 1217 | foreach ($matches as $match) { |
||
| 1218 | $attributes[$match[1]] = $match[2]; |
||
| 1219 | } |
||
| 1220 | |||
| 1221 | // If there is a width in the style attributes: |
||
| 1222 | if (isset($attributes['width']) && -1 == $width) { |
||
| 1223 | // check that the last two characters are px (pixels) |
||
| 1224 | if ('px' === strtolower(substr($attributes['width'], -2))) { |
||
| 1225 | $proposed_width = substr($attributes['width'], 0, -2); |
||
| 1226 | // Now make sure that it's an integer and not something stupid. |
||
| 1227 | if (filter_var($proposed_width, FILTER_VALIDATE_INT)) { |
||
| 1228 | $width = $proposed_width; |
||
| 1229 | } |
||
| 1230 | } |
||
| 1231 | } |
||
| 1232 | |||
| 1233 | // If there is a width in the style attributes: |
||
| 1234 | if (isset($attributes['height']) && -1 == $height) { |
||
| 1235 | // check that the last two characters are px (pixels) |
||
| 1236 | if ('px' == strtolower(substr($attributes['height'], -2))) { |
||
| 1237 | $proposed_height = substr($attributes['height'], 0, -2); |
||
| 1238 | // Now make sure that it's an integer and not something stupid. |
||
| 1239 | if (filter_var($proposed_height, FILTER_VALIDATE_INT)) { |
||
| 1240 | $height = $proposed_height; |
||
| 1241 | } |
||
| 1242 | } |
||
| 1243 | } |
||
| 1244 | } |
||
| 1245 | |||
| 1246 | // Future enhancement: |
||
| 1247 | // Look in the tag to see if there is a class or id specified that has |
||
| 1248 | // a height or width attribute to it. |
||
| 1249 | |||
| 1250 | // Far future enhancement |
||
| 1251 | // Look at all the parent tags of this image to see if they specify a |
||
| 1252 | // class or id that has an img selector that specifies a height or width |
||
| 1253 | // Note that in this case, the class or id will have the img subselector |
||
| 1254 | // for it to apply to the image. |
||
| 1255 | |||
| 1256 | // ridiculously far future development |
||
| 1257 | // If the class or id is specified in a SEPARATE css file thats not on |
||
| 1258 | // the page, go get it and do what we were just doing for the ones on |
||
| 1259 | // the page. |
||
| 1260 | |||
| 1261 | $result = [ |
||
| 1262 | 'height' => $height, |
||
| 1263 | 'width' => $width, |
||
| 1264 | ]; |
||
| 1265 | |||
| 1266 | return $result; |
||
| 1267 | } |
||
| 1268 | |||
| 1269 | public function save($filepath = '') |
||
| 1270 | { |
||
| 1271 | $ret = $this->outertext(); |
||
| 1272 | |||
| 1273 | if ('' !== $filepath) { |
||
| 1274 | file_put_contents($filepath, $ret, LOCK_EX); |
||
| 1275 | } |
||
| 1276 | |||
| 1277 | return $ret; |
||
| 1278 | } |
||
| 1279 | |||
| 1280 | public function addClass($class) |
||
| 1281 | { |
||
| 1282 | if (is_string($class)) { |
||
| 1283 | $class = explode(' ', $class); |
||
| 1284 | } |
||
| 1285 | |||
| 1286 | if (is_array($class)) { |
||
| 1287 | foreach ($class as $c) { |
||
| 1288 | if (isset($this->class)) { |
||
| 1289 | if ($this->hasClass($c)) { |
||
| 1290 | continue; |
||
| 1291 | } else { |
||
| 1292 | $this->class .= ' ' . $c; |
||
| 1293 | } |
||
| 1294 | } else { |
||
| 1295 | $this->class = $c; |
||
| 1296 | } |
||
| 1297 | } |
||
| 1298 | } |
||
| 1299 | } |
||
| 1300 | |||
| 1301 | public function hasClass($class) |
||
| 1302 | { |
||
| 1303 | if (is_string($class)) { |
||
| 1304 | if (isset($this->class)) { |
||
| 1305 | return in_array($class, explode(' ', $this->class), true); |
||
| 1306 | } |
||
| 1307 | } |
||
| 1308 | |||
| 1309 | return false; |
||
| 1310 | } |
||
| 1311 | |||
| 1312 | public function removeClass($class = null) |
||
| 1313 | { |
||
| 1314 | if (!isset($this->class)) { |
||
| 1315 | return; |
||
| 1316 | } |
||
| 1317 | |||
| 1318 | if (is_null($class)) { |
||
| 1319 | $this->removeAttribute('class'); |
||
| 1320 | |||
| 1321 | return; |
||
| 1322 | } |
||
| 1323 | |||
| 1324 | if (is_string($class)) { |
||
| 1325 | $class = explode(' ', $class); |
||
| 1326 | } |
||
| 1327 | |||
| 1328 | if (is_array($class)) { |
||
| 1329 | $class = array_diff(explode(' ', $this->class), $class); |
||
| 1330 | if (empty($class)) { |
||
| 1331 | $this->removeAttribute('class'); |
||
| 1332 | } else { |
||
| 1333 | $this->class = implode(' ', $class); |
||
| 1334 | } |
||
| 1335 | } |
||
| 1336 | } |
||
| 1337 | |||
| 1338 | public function getAllAttributes() |
||
| 1339 | { |
||
| 1340 | return $this->attr; |
||
| 1341 | } |
||
| 1342 | |||
| 1343 | public function getAttribute($name) |
||
| 1344 | { |
||
| 1345 | return $this->$name; |
||
| 1346 | } |
||
| 1347 | |||
| 1348 | public function setAttribute($name, $value) |
||
| 1349 | { |
||
| 1350 | $this->$name = $value; |
||
| 1351 | } |
||
| 1352 | |||
| 1353 | public function hasAttribute($name) |
||
| 1354 | { |
||
| 1355 | return isset($this->$name); |
||
| 1356 | } |
||
| 1357 | |||
| 1358 | public function removeAttribute($name) |
||
| 1359 | { |
||
| 1360 | unset($this->$name); |
||
| 1361 | } |
||
| 1362 | |||
| 1363 | public function remove() |
||
| 1364 | { |
||
| 1365 | if ($this->parent) { |
||
| 1366 | $this->parent->removeChild($this); |
||
| 1367 | } |
||
| 1368 | } |
||
| 1369 | |||
| 1370 | public function removeChild($node) |
||
| 1371 | { |
||
| 1372 | foreach ($node->children as $child) { |
||
| 1373 | $node->removeChild($child); |
||
| 1374 | } |
||
| 1375 | |||
| 1376 | // No need to re-index node->children because it is about to be removed! |
||
| 1377 | |||
| 1378 | foreach ($node->nodes as $entity) { |
||
| 1379 | $enidx = array_search($entity, $node->nodes, true); |
||
| 1380 | $edidx = array_search($entity, $node->dom->nodes, true); |
||
| 1381 | |||
| 1382 | if (false !== $enidx) { |
||
| 1383 | unset($node->nodes[$enidx]); |
||
| 1384 | } |
||
| 1385 | |||
| 1386 | if (false !== $edidx) { |
||
| 1387 | unset($node->dom->nodes[$edidx]); |
||
| 1388 | } |
||
| 1389 | } |
||
| 1390 | |||
| 1391 | // No need to re-index node->nodes because it is about to be removed! |
||
| 1392 | |||
| 1393 | $nidx = array_search($node, $this->nodes, true); |
||
| 1394 | $cidx = array_search($node, $this->children, true); |
||
| 1395 | $didx = array_search($node, $this->dom->nodes, true); |
||
| 1396 | |||
| 1397 | if (false !== $nidx) { |
||
| 1398 | unset($this->nodes[$nidx]); |
||
| 1399 | } |
||
| 1400 | |||
| 1401 | $this->nodes = array_values($this->nodes); |
||
| 1402 | |||
| 1403 | if (false !== $cidx) { |
||
| 1404 | unset($this->children[$cidx]); |
||
| 1405 | } |
||
| 1406 | |||
| 1407 | $this->children = array_values($this->children); |
||
| 1408 | |||
| 1409 | if (false !== $didx) { |
||
| 1410 | unset($this->dom->nodes[$didx]); |
||
| 1411 | } |
||
| 1412 | |||
| 1413 | // Do not re-index dom->nodes because nodes point to other nodes in the |
||
| 1414 | // array explicitly! |
||
| 1415 | |||
| 1416 | $node->clear(); |
||
| 1417 | } |
||
| 1418 | |||
| 1419 | public function getElementById($id) |
||
| 1420 | { |
||
| 1421 | return $this->find("#$id", 0); |
||
| 1422 | } |
||
| 1423 | |||
| 1424 | public function getElementsById($id, $idx = null) |
||
| 1425 | { |
||
| 1426 | return $this->find("#$id", $idx); |
||
| 1427 | } |
||
| 1428 | |||
| 1429 | public function getElementByTagName($name) |
||
| 1430 | { |
||
| 1431 | return $this->find($name, 0); |
||
| 1432 | } |
||
| 1433 | |||
| 1434 | public function getElementsByTagName($name, $idx = null) |
||
| 1435 | { |
||
| 1436 | return $this->find($name, $idx); |
||
| 1437 | } |
||
| 1438 | |||
| 1439 | public function parentNode() |
||
| 1440 | { |
||
| 1441 | return $this->parent(); |
||
| 1442 | } |
||
| 1443 | |||
| 1444 | public function childNodes($idx = -1) |
||
| 1445 | { |
||
| 1446 | if (-1 === $idx) { |
||
| 1447 | return $this->children; |
||
| 1448 | } |
||
| 1449 | |||
| 1450 | if (isset($this->children[$idx])) { |
||
| 1451 | return $this->children[$idx]; |
||
| 1452 | } |
||
| 1453 | |||
| 1454 | return null; |
||
| 1455 | } |
||
| 1456 | |||
| 1457 | public function firstChild() |
||
| 1458 | { |
||
| 1459 | if (count($this->children) > 0) { |
||
| 1460 | return $this->children[0]; |
||
| 1461 | } |
||
| 1462 | |||
| 1463 | return null; |
||
| 1464 | } |
||
| 1465 | |||
| 1466 | public function lastChild() |
||
| 1467 | { |
||
| 1468 | if (count($this->children) > 0) { |
||
| 1469 | return end($this->children); |
||
| 1470 | } |
||
| 1471 | |||
| 1472 | return null; |
||
| 1473 | } |
||
| 1474 | |||
| 1475 | public function nextSibling() |
||
| 1488 | } |
||
| 1489 | |||
| 1490 | public function previousSibling() |
||
| 1491 | { |
||
| 1492 | if (null === $this->parent) { |
||
| 1493 | return null; |
||
| 1494 | } |
||
| 1495 | |||
| 1496 | $idx = array_search($this, $this->parent->children, true); |
||
| 1497 | |||
| 1498 | if (false !== $idx && $idx > 0) { |
||
| 1499 | return $this->parent->children[$idx - 1]; |
||
| 1500 | } |
||
| 1501 | |||
| 1502 | return null; |
||
| 1503 | } |
||
| 1504 | |||
| 1505 | public function hasChildNodes() |
||
| 1508 | } |
||
| 1509 | |||
| 1510 | public function nodeName() |
||
| 1513 | } |
||
| 1514 | |||
| 1515 | public function appendChild($node) |
||
| 1516 | { |
||
| 1517 | $node->parent = $this; |
||
| 1518 | $this->nodes[] = $node; |
||
| 1538 | } |
||
| 1539 | } |
||
| 1540 |
Loading history...