wasinger /
htmlpagedom
Complex classes like HtmlPageCrawler 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. You can also have a look at the cohesion graph to spot any un-connected, or weakly-connected components.
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 HtmlPageCrawler, and based on these observations, apply Extract Interface, too.
| 1 | <?php |
||
| 15 | class HtmlPageCrawler extends Crawler |
||
| 16 | { |
||
| 17 | /** |
||
| 18 | * the (internal) root element name used when importing html fragments |
||
| 19 | * */ |
||
| 20 | const FRAGMENT_ROOT_TAGNAME = '_root'; |
||
| 21 | |||
| 22 | /** |
||
| 23 | * Get an HtmlPageCrawler object from a HTML string, DOMNode, DOMNodeList or HtmlPageCrawler |
||
| 24 | * |
||
| 25 | * This is the equivalent to jQuery's $() function when used for wrapping DOMNodes or creating DOMElements from HTML code. |
||
| 26 | * |
||
| 27 | * @param string|HtmlPageCrawler|\DOMNode|\DOMNodeList|array $content |
||
| 28 | * @return HtmlPageCrawler |
||
| 29 | * @api |
||
| 30 | */ |
||
| 31 | 19 | public static function create($content) |
|
| 39 | |||
| 40 | /** |
||
| 41 | * Adds the specified class(es) to each element in the set of matched elements. |
||
| 42 | * |
||
| 43 | * @param string $name One or more space-separated classes to be added to the class attribute of each matched element. |
||
| 44 | * @return HtmlPageCrawler $this for chaining |
||
| 45 | * @api |
||
| 46 | */ |
||
| 47 | 1 | public function addClass($name) |
|
| 68 | |||
| 69 | /** |
||
| 70 | * Insert content, specified by the parameter, after each element in the set of matched elements. |
||
| 71 | * |
||
| 72 | * @param string|HtmlPageCrawler|\DOMNode|\DOMNodeList $content |
||
| 73 | * @return HtmlPageCrawler $this for chaining |
||
| 74 | * @api |
||
| 75 | */ |
||
| 76 | 3 | public function after($content) |
|
| 98 | |||
| 99 | /** |
||
| 100 | * Insert HTML content as child nodes of each element after existing children |
||
| 101 | * |
||
| 102 | * @param string|HtmlPageCrawler|\DOMNode|\DOMNodeList $content HTML code fragment or DOMNode to append |
||
| 103 | * @return HtmlPageCrawler $this for chaining |
||
| 104 | * @api |
||
| 105 | */ |
||
| 106 | 2 | public function append($content) |
|
| 130 | |||
| 131 | /** |
||
| 132 | * Insert every element in the set of matched elements to the end of the target. |
||
| 133 | * |
||
| 134 | * @param string|HtmlPageCrawler|\DOMNode|\DOMNodeList $element |
||
| 135 | * @return \Wa72\HtmlPageDom\HtmlPageCrawler A new Crawler object containing all elements appended to the target elements |
||
| 136 | * @api |
||
| 137 | */ |
||
| 138 | 2 | public function appendTo($element) |
|
| 155 | |||
| 156 | /** |
||
| 157 | * Returns the attribute value of the first node of the list, or sets an attribute on each element |
||
| 158 | * |
||
| 159 | * @see HtmlPageCrawler::getAttribute() |
||
| 160 | * @see HtmlPageCrawler::setAttribute |
||
| 161 | * |
||
| 162 | * @param string $name |
||
| 163 | * @param null|string $value |
||
| 164 | * @return null|string|HtmlPageCrawler |
||
| 165 | * @api |
||
| 166 | */ |
||
| 167 | 2 | public function attr($name, $value = null) |
|
| 168 | { |
||
| 169 | 2 | if ($value === null) { |
|
| 170 | 2 | return $this->getAttribute($name); |
|
| 171 | } else { |
||
| 172 | 1 | @trigger_error('It will not be possible any more to use method attr($name, $value) as setter function in version 2.0. Use setAttribute($name, $value) instead.', E_USER_DEPRECATED); |
|
|
|
|||
| 173 | 1 | return $this->setAttribute($name, $value); |
|
| 174 | } |
||
| 175 | } |
||
| 176 | |||
| 177 | /** |
||
| 178 | * Sets an attribute on each element |
||
| 179 | * |
||
| 180 | * @param string $name |
||
| 181 | * @param string $value |
||
| 182 | * @return HtmlPageCrawler $this for chaining |
||
| 183 | */ |
||
| 184 | 3 | public function setAttribute($name, $value) |
|
| 185 | { |
||
| 186 | 3 | foreach ($this as $node) { |
|
| 187 | 3 | if ($node instanceof \DOMElement) { |
|
| 188 | /** @var \DOMElement $node */ |
||
| 189 | 3 | $node->setAttribute($name, $value); |
|
| 190 | } |
||
| 191 | } |
||
| 192 | 3 | return $this; |
|
| 193 | } |
||
| 194 | |||
| 195 | /** |
||
| 196 | * Returns the attribute value of the first node of the list. |
||
| 197 | * |
||
| 198 | * @param string $name The attribute name |
||
| 199 | * @return string|null The attribute value or null if the attribute does not exist |
||
| 200 | * @throws \InvalidArgumentException When current node is empty |
||
| 201 | * |
||
| 202 | */ |
||
| 203 | 2 | public function getAttribute($name) |
|
| 204 | { |
||
| 205 | 2 | if (!count($this)) { |
|
| 206 | 1 | throw new \InvalidArgumentException('The current node list is empty.'); |
|
| 207 | } |
||
| 208 | 1 | $node = $this->getNode(0); |
|
| 209 | 1 | return $node->hasAttribute($name) ? $node->getAttribute($name) : null; |
|
| 210 | } |
||
| 211 | |||
| 212 | /** |
||
| 213 | * Insert content, specified by the parameter, before each element in the set of matched elements. |
||
| 214 | * |
||
| 215 | * @param string|HtmlPageCrawler|\DOMNode|\DOMNodeList $content |
||
| 216 | * @return HtmlPageCrawler $this for chaining |
||
| 217 | * @api |
||
| 218 | */ |
||
| 219 | 2 | public function before($content) |
|
| 220 | { |
||
| 221 | 2 | $content = self::create($content); |
|
| 222 | 2 | $newnodes = array(); |
|
| 223 | 2 | foreach ($this as $i => $node) { |
|
| 224 | /** @var \DOMNode $node */ |
||
| 225 | 2 | foreach ($content as $newnode) { |
|
| 226 | /** @var \DOMNode $newnode */ |
||
| 227 | 2 | if ($node !== $newnode) { |
|
| 228 | 2 | $newnode = static::importNewnode($newnode, $node, $i); |
|
| 229 | 2 | $node->parentNode->insertBefore($newnode, $node); |
|
| 230 | 2 | $newnodes[] = $newnode; |
|
| 231 | } |
||
| 232 | } |
||
| 233 | } |
||
| 234 | 2 | $content->clear(); |
|
| 235 | 2 | $content->add($newnodes); |
|
| 236 | 2 | return $this; |
|
| 237 | } |
||
| 238 | |||
| 239 | /** |
||
| 240 | * Create a deep copy of the set of matched elements. |
||
| 241 | * |
||
| 242 | * Equivalent to clone() in jQuery (clone is not a valid PHP function name) |
||
| 243 | * |
||
| 244 | * @return HtmlPageCrawler |
||
| 245 | * @api |
||
| 246 | */ |
||
| 247 | 1 | public function makeClone() |
|
| 248 | { |
||
| 249 | 1 | return clone $this; |
|
| 250 | } |
||
| 251 | |||
| 252 | 1 | public function __clone() |
|
| 253 | { |
||
| 254 | 1 | $newnodes = array(); |
|
| 255 | 1 | foreach ($this as $node) { |
|
| 256 | /** @var \DOMNode $node */ |
||
| 257 | 1 | $newnodes[] = $node->cloneNode(true); |
|
| 258 | } |
||
| 259 | 1 | $this->clear(); |
|
| 260 | 1 | $this->add($newnodes); |
|
| 261 | 1 | } |
|
| 262 | |||
| 263 | /** |
||
| 264 | * Get one CSS style property of the first element or set it for all elements in the list |
||
| 265 | * |
||
| 266 | * Function is here for compatibility with jQuery; it is the same as getStyle() and setStyle() |
||
| 267 | * |
||
| 268 | * @see HtmlPageCrawler::getStyle() |
||
| 269 | * @see HtmlPageCrawler::setStyle() |
||
| 270 | * |
||
| 271 | * @param string $key The name of the style property |
||
| 272 | * @param null|string $value The CSS value to set, or NULL to get the current value |
||
| 273 | * @return HtmlPageCrawler|string If no param is provided, returns the CSS styles of the first element |
||
| 274 | * @api |
||
| 275 | */ |
||
| 276 | 1 | public function css($key, $value = null) |
|
| 277 | { |
||
| 278 | 1 | if (null === $value) { |
|
| 279 | 1 | return $this->getStyle($key); |
|
| 280 | } else { |
||
| 281 | 1 | return $this->setStyle($key, $value); |
|
| 282 | } |
||
| 283 | } |
||
| 284 | |||
| 285 | /** |
||
| 286 | * get one CSS style property of the first element |
||
| 287 | * |
||
| 288 | * @param string $key name of the property |
||
| 289 | * @return string|null value of the property |
||
| 290 | */ |
||
| 291 | 1 | public function getStyle($key) |
|
| 292 | { |
||
| 293 | 1 | $styles = Helpers::cssStringToArray($this->getAttribute('style')); |
|
| 294 | 1 | return (isset($styles[$key]) ? $styles[$key] : null); |
|
| 295 | } |
||
| 296 | |||
| 297 | /** |
||
| 298 | * set one CSS style property for all elements in the list |
||
| 299 | * |
||
| 300 | * @param string $key name of the property |
||
| 301 | * @param string $value value of the property |
||
| 302 | * @return HtmlPageCrawler $this for chaining |
||
| 303 | */ |
||
| 304 | 1 | public function setStyle($key, $value) |
|
| 305 | { |
||
| 306 | 1 | foreach ($this as $node) { |
|
| 307 | 1 | if ($node instanceof \DOMElement) { |
|
| 308 | /** @var \DOMElement $node */ |
||
| 309 | 1 | $styles = Helpers::cssStringToArray($node->getAttribute('style')); |
|
| 310 | 1 | if ($value != '') { |
|
| 311 | 1 | $styles[$key] = $value; |
|
| 312 | 1 | } elseif (isset($styles[$key])) { |
|
| 313 | 1 | unset($styles[$key]); |
|
| 314 | } |
||
| 315 | 1 | $node->setAttribute('style', Helpers::cssArrayToString($styles)); |
|
| 316 | } |
||
| 317 | } |
||
| 318 | 1 | return $this; |
|
| 319 | } |
||
| 320 | |||
| 321 | /** |
||
| 322 | * Removes all child nodes and text from all nodes in set |
||
| 323 | * |
||
| 324 | * Equivalent to jQuery's empty() function which is not a valid function name in PHP |
||
| 325 | * @return HtmlPageCrawler $this |
||
| 326 | * @api |
||
| 327 | */ |
||
| 328 | 1 | public function makeEmpty() |
|
| 329 | { |
||
| 330 | 1 | foreach ($this as $node) { |
|
| 331 | 1 | $node->nodeValue = ''; |
|
| 332 | } |
||
| 333 | 1 | return $this; |
|
| 334 | } |
||
| 335 | |||
| 336 | /** |
||
| 337 | * Determine whether any of the matched elements are assigned the given class. |
||
| 338 | * |
||
| 339 | * @param string $name |
||
| 340 | * @return bool |
||
| 341 | * @api |
||
| 342 | */ |
||
| 343 | 2 | public function hasClass($name) |
|
| 344 | { |
||
| 345 | 2 | foreach ($this as $node) { |
|
| 346 | 2 | if ($node instanceof \DOMElement && $class = $node->getAttribute('class')) { |
|
| 347 | 2 | $classes = preg_split('/\s+/s', $class); |
|
| 348 | 2 | if (in_array($name, $classes)) { |
|
| 349 | 2 | return true; |
|
| 350 | } |
||
| 351 | } |
||
| 352 | } |
||
| 353 | 2 | return false; |
|
| 354 | } |
||
| 355 | |||
| 356 | /** |
||
| 357 | * Get the HTML contents of the first element in the set of matched elements |
||
| 358 | * or set the HTML contents of every matched element. |
||
| 359 | * |
||
| 360 | * Deprecation warning: It will not be possible any more to use method html($html) as setter function in version 2.0. Use setInnerHtml($html) instead. |
||
| 361 | * |
||
| 362 | * @see HtmlPageCrawler::setInnerHtml() |
||
| 363 | * |
||
| 364 | * @param string|HtmlPageCrawler|\DOMNode|\DOMNodeList|null $html The HTML content to set, or NULL to get the current content |
||
| 365 | * |
||
| 366 | * @return HtmlPageCrawler|string If no param is provided, returns the HTML content of the first element |
||
| 367 | * @api |
||
| 368 | */ |
||
| 369 | 1 | public function html($html = null) |
|
| 370 | { |
||
| 371 | 1 | if (null === $html) { |
|
| 372 | 1 | return $this->getInnerHtml(); |
|
| 373 | } else { |
||
| 374 | 1 | @trigger_error('It will not be possible any more to use method html($html) as setter function in version 2.0. Use setInnerHtml($html) instead.', E_USER_DEPRECATED); |
|
| 375 | 1 | $this->setInnerHtml($html); |
|
| 376 | 1 | return $this; |
|
| 377 | } |
||
| 378 | } |
||
| 379 | |||
| 380 | /** |
||
| 381 | * Get the innerHTML contents of the first element |
||
| 382 | * |
||
| 383 | * @return string HTML code fragment |
||
| 384 | * @see html() |
||
| 385 | */ |
||
| 386 | 4 | public function getInnerHtml() |
|
| 387 | { |
||
| 388 | 4 | $node = $this->getNode(0); |
|
| 389 | 4 | if ($node instanceof \DOMNode) { |
|
| 390 | 3 | $doc = new \DOMDocument('1.0', 'UTF-8'); |
|
| 391 | 3 | $doc->appendChild($doc->importNode($node, true)); |
|
| 392 | 3 | $html = trim($doc->saveHTML()); |
|
| 393 | 3 | $tag = $node->nodeName; |
|
| 394 | 3 | return preg_replace('@^<' . $tag . '[^>]*>|</' . $tag . '>$@', '', $html); |
|
| 395 | } else { |
||
| 396 | 1 | return ''; |
|
| 397 | } |
||
| 398 | } |
||
| 399 | |||
| 400 | /** |
||
| 401 | * Set the HTML contents of each element |
||
| 402 | * |
||
| 403 | * @param string|HtmlPageCrawler|\DOMNode|\DOMNodeList $content HTML code fragment |
||
| 404 | * @return HtmlPageCrawler $this for chaining |
||
| 405 | */ |
||
| 406 | 3 | public function setInnerHtml($content) |
|
| 407 | { |
||
| 408 | 3 | $content = self::create($content); |
|
| 409 | 3 | foreach ($this as $node) { |
|
| 410 | 3 | $node->nodeValue = ''; |
|
| 411 | 3 | foreach ($content as $newnode) { |
|
| 412 | /** @var \DOMNode $node */ |
||
| 413 | /** @var \DOMNode $newnode */ |
||
| 414 | 3 | $newnode = static::importNewnode($newnode, $node); |
|
| 415 | 3 | $node->appendChild($newnode); |
|
| 416 | } |
||
| 417 | } |
||
| 418 | 3 | return $this; |
|
| 419 | } |
||
| 420 | |||
| 421 | /** |
||
| 422 | * Insert every element in the set of matched elements after the target. |
||
| 423 | * |
||
| 424 | * @param string|HtmlPageCrawler|\DOMNode|\DOMNodeList $element |
||
| 425 | * @return \Wa72\HtmlPageDom\HtmlPageCrawler A new Crawler object containing all elements appended to the target elements |
||
| 426 | * @api |
||
| 427 | */ |
||
| 428 | 2 | public function insertAfter($element) |
|
| 429 | { |
||
| 430 | 2 | $e = self::create($element); |
|
| 431 | 2 | $newnodes = array(); |
|
| 432 | 2 | foreach ($e as $i => $node) { |
|
| 433 | /** @var \DOMNode $node */ |
||
| 434 | 2 | $refnode = $node->nextSibling; |
|
| 435 | 2 | foreach ($this as $newnode) { |
|
| 436 | /** @var \DOMNode $newnode */ |
||
| 437 | 2 | $newnode = static::importNewnode($newnode, $node, $i); |
|
| 438 | 2 | if ($refnode === null) { |
|
| 439 | 2 | $node->parentNode->appendChild($newnode); |
|
| 440 | } else { |
||
| 441 | 1 | $node->parentNode->insertBefore($newnode, $refnode); |
|
| 442 | } |
||
| 443 | 2 | $newnodes[] = $newnode; |
|
| 444 | } |
||
| 445 | } |
||
| 446 | 2 | return self::create($newnodes); |
|
| 447 | } |
||
| 448 | |||
| 449 | /** |
||
| 450 | * Insert every element in the set of matched elements before the target. |
||
| 451 | * |
||
| 452 | * @param string|HtmlPageCrawler|\DOMNode|\DOMNodeList $element |
||
| 453 | * @return \Wa72\HtmlPageDom\HtmlPageCrawler A new Crawler object containing all elements appended to the target elements |
||
| 454 | * @api |
||
| 455 | */ |
||
| 456 | 2 | public function insertBefore($element) |
|
| 457 | { |
||
| 458 | 2 | $e = self::create($element); |
|
| 459 | 2 | $newnodes = array(); |
|
| 460 | 2 | foreach ($e as $i => $node) { |
|
| 461 | /** @var \DOMNode $node */ |
||
| 462 | 2 | foreach ($this as $newnode) { |
|
| 463 | /** @var \DOMNode $newnode */ |
||
| 464 | 2 | $newnode = static::importNewnode($newnode, $node, $i); |
|
| 465 | 2 | if ($newnode !== $node) { |
|
| 466 | 2 | $node->parentNode->insertBefore($newnode, $node); |
|
| 467 | } |
||
| 468 | 2 | $newnodes[] = $newnode; |
|
| 469 | } |
||
| 470 | } |
||
| 471 | 2 | return self::create($newnodes); |
|
| 472 | } |
||
| 473 | |||
| 474 | /** |
||
| 475 | * Insert content, specified by the parameter, to the beginning of each element in the set of matched elements. |
||
| 476 | * |
||
| 477 | * @param string|HtmlPageCrawler|\DOMNode|\DOMNodeList $content HTML code fragment |
||
| 478 | * @return HtmlPageCrawler $this for chaining |
||
| 479 | * @api |
||
| 480 | */ |
||
| 481 | 2 | public function prepend($content) |
|
| 482 | { |
||
| 483 | 2 | $content = self::create($content); |
|
| 484 | 2 | $newnodes = array(); |
|
| 485 | 2 | foreach ($this as $i => $node) { |
|
| 486 | 2 | $refnode = $node->firstChild; |
|
| 487 | /** @var \DOMNode $node */ |
||
| 488 | 2 | foreach ($content as $newnode) { |
|
| 489 | /** @var \DOMNode $newnode */ |
||
| 490 | 2 | $newnode = static::importNewnode($newnode, $node, $i); |
|
| 491 | 2 | if ($refnode === null) { |
|
| 492 | 1 | $node->appendChild($newnode); |
|
| 493 | 2 | } else if ($refnode !== $newnode) { |
|
| 494 | 2 | $node->insertBefore($newnode, $refnode); |
|
| 495 | } |
||
| 496 | 2 | $newnodes[] = $newnode; |
|
| 497 | } |
||
| 498 | } |
||
| 499 | 2 | $content->clear(); |
|
| 500 | 2 | $content->add($newnodes); |
|
| 501 | 2 | return $this; |
|
| 502 | } |
||
| 503 | |||
| 504 | /** |
||
| 505 | * Insert every element in the set of matched elements to the beginning of the target. |
||
| 506 | * |
||
| 507 | * @param string|HtmlPageCrawler|\DOMNode|\DOMNodeList $element |
||
| 508 | * @return \Wa72\HtmlPageDom\HtmlPageCrawler A new Crawler object containing all elements prepended to the target elements |
||
| 509 | * @api |
||
| 510 | */ |
||
| 511 | 1 | public function prependTo($element) |
|
| 512 | { |
||
| 513 | 1 | $e = self::create($element); |
|
| 514 | 1 | $newnodes = array(); |
|
| 515 | 1 | foreach ($e as $i => $node) { |
|
| 516 | 1 | $refnode = $node->firstChild; |
|
| 517 | /** @var \DOMNode $node */ |
||
| 518 | 1 | foreach ($this as $newnode) { |
|
| 519 | /** @var \DOMNode $newnode */ |
||
| 520 | 1 | $newnode = static::importNewnode($newnode, $node, $i); |
|
| 521 | 1 | if ($newnode !== $node) { |
|
| 522 | 1 | if ($refnode === null) { |
|
| 523 | 1 | $node->appendChild($newnode); |
|
| 524 | } else { |
||
| 525 | 1 | $node->insertBefore($newnode, $refnode); |
|
| 526 | } |
||
| 527 | } |
||
| 528 | 1 | $newnodes[] = $newnode; |
|
| 529 | } |
||
| 530 | } |
||
| 531 | 1 | return self::create($newnodes); |
|
| 532 | } |
||
| 533 | |||
| 534 | /** |
||
| 535 | * Remove the set of matched elements from the DOM. |
||
| 536 | * |
||
| 537 | * (as opposed to Crawler::clear() which detaches the nodes only from Crawler |
||
| 538 | * but leaves them in the DOM) |
||
| 539 | * |
||
| 540 | * @api |
||
| 541 | */ |
||
| 542 | 2 | public function remove() |
|
| 543 | { |
||
| 544 | 2 | foreach ($this as $node) { |
|
| 545 | /** |
||
| 546 | * @var \DOMNode $node |
||
| 547 | */ |
||
| 548 | 2 | if ($node->parentNode instanceof \DOMElement) { |
|
| 549 | 2 | $node->parentNode->removeChild($node); |
|
| 550 | } |
||
| 551 | } |
||
| 552 | 2 | $this->clear(); |
|
| 553 | 2 | } |
|
| 554 | |||
| 555 | /** |
||
| 556 | * Remove an attribute from each element in the set of matched elements. |
||
| 557 | * |
||
| 558 | * Alias for removeAttribute for compatibility with jQuery |
||
| 559 | * |
||
| 560 | * @param string $name |
||
| 561 | * @return HtmlPageCrawler |
||
| 562 | * @api |
||
| 563 | */ |
||
| 564 | 1 | public function removeAttr($name) |
|
| 565 | { |
||
| 566 | 1 | return $this->removeAttribute($name); |
|
| 567 | } |
||
| 568 | |||
| 569 | /** |
||
| 570 | * Remove an attribute from each element in the set of matched elements. |
||
| 571 | * |
||
| 572 | * @param string $name |
||
| 573 | * @return HtmlPageCrawler |
||
| 574 | */ |
||
| 575 | 1 | public function removeAttribute($name) |
|
| 576 | { |
||
| 577 | 1 | foreach ($this as $node) { |
|
| 578 | 1 | if ($node instanceof \DOMElement) { |
|
| 579 | /** @var \DOMElement $node */ |
||
| 580 | 1 | if ($node->hasAttribute($name)) { |
|
| 581 | 1 | $node->removeAttribute($name); |
|
| 582 | } |
||
| 583 | } |
||
| 584 | } |
||
| 585 | 1 | return $this; |
|
| 586 | } |
||
| 587 | |||
| 588 | /** |
||
| 589 | * Remove a class from each element in the list |
||
| 590 | * |
||
| 591 | * @param string $name |
||
| 592 | * @return HtmlPageCrawler $this for chaining |
||
| 593 | * @api |
||
| 594 | */ |
||
| 595 | 2 | public function removeClass($name) |
|
| 596 | { |
||
| 597 | 2 | foreach ($this as $node) { |
|
| 598 | 2 | if ($node instanceof \DOMElement) { |
|
| 599 | /** @var \DOMElement $node */ |
||
| 600 | 2 | $classes = preg_split('/\s+/s', $node->getAttribute('class')); |
|
| 601 | 2 | $count = count($classes); |
|
| 602 | 2 | for ($i = 0; $i < $count; $i++) { |
|
| 603 | 2 | if ($classes[$i] == $name) { |
|
| 604 | 2 | unset($classes[$i]); |
|
| 605 | } |
||
| 606 | } |
||
| 607 | 2 | $node->setAttribute('class', trim(join(' ', $classes))); |
|
| 608 | } |
||
| 609 | } |
||
| 610 | 2 | return $this; |
|
| 611 | } |
||
| 612 | |||
| 613 | /** |
||
| 614 | * Replace each target element with the set of matched elements. |
||
| 615 | * |
||
| 616 | * @param string|HtmlPageCrawler|\DOMNode|\DOMNodeList $element |
||
| 617 | * @return \Wa72\HtmlPageDom\HtmlPageCrawler A new Crawler object containing all elements appended to the target elements |
||
| 618 | * @api |
||
| 619 | */ |
||
| 620 | 2 | public function replaceAll($element) |
|
| 621 | { |
||
| 622 | 2 | $e = self::create($element); |
|
| 623 | 2 | $newnodes = array(); |
|
| 624 | 2 | foreach ($e as $i => $node) { |
|
| 625 | /** @var \DOMNode $node */ |
||
| 626 | 2 | $parent = $node->parentNode; |
|
| 627 | 2 | $refnode = $node->nextSibling; |
|
| 628 | 2 | foreach ($this as $j => $newnode) { |
|
| 629 | /** @var \DOMNode $newnode */ |
||
| 630 | 2 | $newnode = static::importNewnode($newnode, $node, $i); |
|
| 631 | 2 | if ($j == 0) { |
|
| 632 | 2 | $parent->replaceChild($newnode, $node); |
|
| 633 | } else { |
||
| 634 | 1 | $parent->insertBefore($newnode, $refnode); |
|
| 635 | } |
||
| 636 | 2 | $newnodes[] = $newnode; |
|
| 637 | } |
||
| 638 | } |
||
| 639 | 2 | return self::create($newnodes); |
|
| 640 | } |
||
| 641 | |||
| 642 | /** |
||
| 643 | * Replace each element in the set of matched elements with the provided new content and return the set of elements that was removed. |
||
| 644 | * |
||
| 645 | * @param string|HtmlPageCrawler|\DOMNode|\DOMNodeList $content |
||
| 646 | * @return \Wa72\HtmlPageDom\HtmlPageCrawler $this for chaining |
||
| 647 | * @api |
||
| 648 | */ |
||
| 649 | 2 | public function replaceWith($content) |
|
| 650 | { |
||
| 651 | 2 | $content = self::create($content); |
|
| 652 | 2 | $newnodes = array(); |
|
| 653 | 2 | foreach ($this as $i => $node) { |
|
| 654 | /** @var \DOMNode $node */ |
||
| 655 | 2 | $parent = $node->parentNode; |
|
| 656 | 2 | $refnode = $node->nextSibling; |
|
| 657 | 2 | foreach ($content as $j => $newnode) { |
|
| 658 | /** @var \DOMNode $newnode */ |
||
| 659 | 2 | $newnode = static::importNewnode($newnode, $node, $i); |
|
| 660 | 2 | if ($j == 0) { |
|
| 661 | 2 | $parent->replaceChild($newnode, $node); |
|
| 662 | } else { |
||
| 663 | 1 | $parent->insertBefore($newnode, $refnode); |
|
| 664 | } |
||
| 665 | 2 | $newnodes[] = $newnode; |
|
| 666 | } |
||
| 667 | } |
||
| 668 | 2 | $content->clear(); |
|
| 669 | 2 | $content->add($newnodes); |
|
| 670 | 2 | return $this; |
|
| 671 | } |
||
| 672 | |||
| 673 | /** |
||
| 674 | * Get the combined text contents of each element in the set of matched elements, including their descendants, |
||
| 675 | * or set the text contents of the matched elements. |
||
| 676 | * |
||
| 677 | * ATTENTION: Contrary to the parent Crawler class, which returns the text from the first element only, |
||
| 678 | * this functions returns the combined text of all elements (as jQuery does). If this is not what you need you |
||
| 679 | * must call ->first() before calling ->text(), e.g. |
||
| 680 | * |
||
| 681 | * in Symfony\DOMCrawler\Crawler: $c->filter('p')->text() returns the text of the first paragraph only |
||
| 682 | * in HtmlPageCrawler you need to call: $c->filter('p')->first()->text() |
||
| 683 | * |
||
| 684 | * DEPRECATION WARNING: |
||
| 685 | * This function will be removed from here in 2.0, so calling text() then will call the parent implementation, i.e. |
||
| 686 | * it will return the text from the first node only, and an argument passed is treated as default value for the |
||
| 687 | * getter function from Symfony 4.3 onwards. It will not be possible to use this function as setter. |
||
| 688 | * |
||
| 689 | * Use getCombinedText() for a getter with the old behavior, and use setText() for setting text content. |
||
| 690 | * |
||
| 691 | * @see setText() |
||
| 692 | * @see getCombinedText() |
||
| 693 | * |
||
| 694 | * @param null|string $text |
||
| 695 | * @return string|HtmlPageCrawler |
||
| 696 | * @api |
||
| 697 | */ |
||
| 698 | 2 | public function text($text = null) |
|
| 699 | { |
||
| 700 | 2 | if ($text === null) { |
|
| 701 | 2 | @trigger_error('In Version 2.0, Method text() will return the text from only the first element in the set. Consider using getCombinedText() instead.', E_USER_NOTICE); |
|
| 702 | 2 | return $this->getCombinedText(); |
|
| 703 | } else { |
||
| 704 | 1 | @trigger_error('It will not be possible any more to use method text($text) as setter function in version 2.0. Use setText($text) instead.', E_USER_DEPRECATED); |
|
| 705 | 1 | return $this->setText($text); |
|
| 706 | } |
||
| 707 | } |
||
| 708 | |||
| 709 | /** |
||
| 710 | * Get the combined text contents of each element in the set of matched elements, including their descendants. |
||
| 711 | * This is what the jQuery text() function does, contrary to the Crawler::text() method that returns only |
||
| 712 | * the text of the first node. |
||
| 713 | * |
||
| 714 | * @return string |
||
| 715 | * @api |
||
| 716 | * @since 1.4 |
||
| 717 | */ |
||
| 718 | 2 | public function getCombinedText() |
|
| 719 | { |
||
| 720 | 2 | $text = ''; |
|
| 721 | 2 | foreach ($this as $node) { |
|
| 722 | /** @var \DOMNode $node */ |
||
| 723 | 2 | $text .= $node->nodeValue; |
|
| 724 | } |
||
| 725 | 2 | return $text; |
|
| 726 | } |
||
| 727 | /** |
||
| 728 | * Set the text contents of the matched elements. |
||
| 729 | * |
||
| 730 | * @param string $text |
||
| 731 | * @return HtmlPageCrawler |
||
| 732 | * @api |
||
| 733 | * @since 1.4 |
||
| 734 | */ |
||
| 735 | 2 | public function setText($text) |
|
| 736 | { |
||
| 737 | 2 | $text = htmlspecialchars($text); |
|
| 738 | 2 | foreach ($this as $node) { |
|
| 739 | /** @var \DOMNode $node */ |
||
| 740 | 2 | $node->nodeValue = $text; |
|
| 741 | } |
||
| 742 | 2 | return $this; |
|
| 743 | } |
||
| 744 | |||
| 745 | |||
| 746 | /** |
||
| 747 | * Add or remove one or more classes from each element in the set of matched elements, depending the class’s presence. |
||
| 748 | * |
||
| 749 | * @param string $classname One or more classnames separated by spaces |
||
| 750 | * @return \Wa72\HtmlPageDom\HtmlPageCrawler $this for chaining |
||
| 751 | * @api |
||
| 752 | */ |
||
| 753 | 1 | public function toggleClass($classname) |
|
| 769 | |||
| 770 | /** |
||
| 771 | * Remove the parents of the set of matched elements from the DOM, leaving the matched elements in their place. |
||
| 772 | * |
||
| 773 | * @return \Wa72\HtmlPageDom\HtmlPageCrawler $this for chaining |
||
| 774 | * @api |
||
| 775 | */ |
||
| 776 | 1 | public function unwrap() |
|
| 786 | |||
| 787 | /** |
||
| 788 | * Remove the matched elements, but promote the children to take their place. |
||
| 789 | * |
||
| 790 | * @return \Wa72\HtmlPageDom\HtmlPageCrawler $this for chaining |
||
| 791 | * @api |
||
| 792 | */ |
||
| 793 | 2 | public function unwrapInner() |
|
| 809 | |||
| 810 | |||
| 811 | /** |
||
| 812 | * Wrap an HTML structure around each element in the set of matched elements |
||
| 813 | * |
||
| 814 | * The HTML structure must contain only one root node, e.g.: |
||
| 815 | * Works: <div><div></div></div> |
||
| 816 | * Does not work: <div></div><div></div> |
||
| 817 | * |
||
| 818 | * @param string|HtmlPageCrawler|\DOMNode $wrappingElement |
||
| 819 | * @return HtmlPageCrawler $this for chaining |
||
| 820 | * @api |
||
| 821 | */ |
||
| 822 | 1 | public function wrap($wrappingElement) |
|
| 859 | |||
| 860 | /** |
||
| 861 | * Wrap an HTML structure around all elements in the set of matched elements. |
||
| 862 | * |
||
| 863 | * @param string|HtmlPageCrawler|\DOMNode|\DOMNodeList $content |
||
| 864 | * @throws \LogicException |
||
| 865 | * @return \Wa72\HtmlPageDom\HtmlPageCrawler $this for chaining |
||
| 866 | * @api |
||
| 867 | */ |
||
| 868 | 1 | public function wrapAll($content) |
|
| 869 | { |
||
| 870 | 1 | $content = self::create($content); |
|
| 871 | 1 | $parent = $this->getNode(0)->parentNode; |
|
| 872 | 1 | foreach ($this as $i => $node) { |
|
| 873 | /** @var \DOMNode $node */ |
||
| 874 | 1 | if ($node->parentNode !== $parent) { |
|
| 875 | throw new \LogicException('Nodes to be wrapped with wrapAll() must all have the same parent'); |
||
| 876 | } |
||
| 877 | } |
||
| 878 | |||
| 879 | 1 | $newnode = $content->getNode(0); |
|
| 880 | /** @var \DOMNode $newnode */ |
||
| 881 | 1 | $newnode = static::importNewnode($newnode, $parent); |
|
| 882 | |||
| 883 | 1 | $newnode = $parent->insertBefore($newnode,$this->getNode(0)); |
|
| 884 | 1 | $content->clear(); |
|
| 885 | 1 | $content->add($newnode); |
|
| 886 | |||
| 887 | 1 | while ($newnode->hasChildNodes()) { |
|
| 888 | 1 | $elementFound = false; |
|
| 889 | 1 | foreach ($newnode->childNodes as $child) { |
|
| 890 | 1 | if ($child instanceof \DOMElement) { |
|
| 891 | 1 | $newnode = $child; |
|
| 892 | 1 | $elementFound = true; |
|
| 893 | 1 | break; |
|
| 894 | } |
||
| 895 | } |
||
| 896 | 1 | if (!$elementFound) { |
|
| 897 | break; |
||
| 898 | } |
||
| 899 | } |
||
| 900 | 1 | foreach ($this as $i => $node) { |
|
| 901 | /** @var \DOMNode $node */ |
||
| 902 | 1 | $newnode->appendChild($node); |
|
| 903 | } |
||
| 904 | 1 | return $this; |
|
| 905 | } |
||
| 906 | |||
| 907 | /** |
||
| 908 | * Wrap an HTML structure around the content of each element in the set of matched elements. |
||
| 909 | * |
||
| 910 | * @param string|HtmlPageCrawler|\DOMNode|\DOMNodeList $content |
||
| 911 | * @return \Wa72\HtmlPageDom\HtmlPageCrawler $this for chaining |
||
| 912 | * @api |
||
| 913 | */ |
||
| 914 | 1 | public function wrapInner($content) |
|
| 922 | |||
| 923 | /** |
||
| 924 | * Get the HTML code fragment of all elements and their contents. |
||
| 925 | * |
||
| 926 | * If the first node contains a complete HTML document return only |
||
| 927 | * the full code of this document. |
||
| 928 | * |
||
| 929 | * @return string HTML code (fragment) |
||
| 930 | * @api |
||
| 931 | */ |
||
| 932 | 8 | public function saveHTML() |
|
| 946 | |||
| 947 | 4 | public function __toString() |
|
| 951 | |||
| 952 | /** |
||
| 953 | * checks whether the first node contains a complete html document |
||
| 954 | * (as opposed to a document fragment) |
||
| 955 | * |
||
| 956 | * @return boolean |
||
| 957 | */ |
||
| 958 | 8 | public function isHtmlDocument() |
|
| 971 | |||
| 972 | /** |
||
| 973 | * get ownerDocument of the first element |
||
| 974 | * |
||
| 975 | * @return \DOMDocument|null |
||
| 976 | */ |
||
| 977 | 1 | public function getDOMDocument() |
|
| 988 | |||
| 989 | /** |
||
| 990 | * Filters the list of nodes with a CSS selector. |
||
| 991 | * |
||
| 992 | * @param string $selector |
||
| 993 | * @return HtmlPageCrawler |
||
| 994 | */ |
||
| 995 | 8 | public function filter($selector) |
|
| 999 | |||
| 1000 | /** |
||
| 1001 | * Filters the list of nodes with an XPath expression. |
||
| 1002 | * |
||
| 1003 | * @param string $xpath An XPath expression |
||
| 1004 | * |
||
| 1005 | * @return HtmlPageCrawler A new instance of Crawler with the filtered list of nodes |
||
| 1006 | * |
||
| 1007 | * @api |
||
| 1008 | */ |
||
| 1009 | 2 | public function filterXPath($xpath) |
|
| 1013 | |||
| 1014 | /** |
||
| 1015 | * Adds HTML/XML content to the HtmlPageCrawler object (but not to the DOM of an already attached node). |
||
| 1016 | * |
||
| 1017 | * Function overriden from Crawler because HTML fragments are always added as complete documents there |
||
| 1018 | * |
||
| 1019 | * |
||
| 1020 | * @param string $content A string to parse as HTML/XML |
||
| 1021 | * @param null|string $type The content type of the string |
||
| 1022 | * |
||
| 1023 | * @return null|void |
||
| 1024 | */ |
||
| 1025 | 18 | public function addContent($content, $type = null) |
|
| 1037 | |||
| 1038 | 16 | public function addHtmlFragment($content, $charset = 'UTF-8') |
|
| 1050 | |||
| 1051 | /** |
||
| 1052 | * returns the first node |
||
| 1053 | * deprecated, use getNode(0) instead |
||
| 1054 | * |
||
| 1055 | * @return \DOMNode|null |
||
| 1056 | * @deprecated |
||
| 1057 | * @see Crawler::getNode |
||
| 1058 | */ |
||
| 1059 | 1 | public function getFirstNode() |
|
| 1063 | |||
| 1064 | /** |
||
| 1065 | * @param int $position |
||
| 1066 | * |
||
| 1067 | * overridden from Crawler because it is not public in Symfony 2.3 |
||
| 1068 | * TODO: throw away as soon as we don't need to support SF 2.3 any more |
||
| 1069 | * |
||
| 1070 | * @return \DOMElement|null |
||
| 1071 | */ |
||
| 1072 | 14 | public function getNode($position) |
|
| 1076 | |||
| 1077 | /** |
||
| 1078 | * Returns the node name of the first node of the list. |
||
| 1079 | * |
||
| 1080 | * in Crawler (parent), this function will be available starting with 2.6.0, |
||
| 1081 | * therefore this method be removed from here as soon as we don't need to keep compatibility |
||
| 1082 | * with Symfony < 2.6 |
||
| 1083 | * |
||
| 1084 | * TODO: throw away as soon as we don't need to support SF 2.3 any more |
||
| 1085 | * |
||
| 1086 | * @return string The node name |
||
| 1087 | * |
||
| 1088 | * @throws \InvalidArgumentException When current node is empty |
||
| 1089 | */ |
||
| 1090 | 1 | public function nodeName() |
|
| 1097 | |||
| 1098 | /** |
||
| 1099 | * Adds a node to the current list of nodes. |
||
| 1100 | * |
||
| 1101 | * This method uses the appropriate specialized add*() method based |
||
| 1102 | * on the type of the argument. |
||
| 1103 | * |
||
| 1104 | * Overwritten from parent to allow Crawler to be added |
||
| 1105 | * |
||
| 1106 | * @param null|\DOMNodeList|array|\DOMNode|Crawler $node A node |
||
| 1107 | * |
||
| 1108 | * @api |
||
| 1109 | */ |
||
| 1110 | 30 | public function add($node) |
|
| 1120 | |||
| 1121 | /** |
||
| 1122 | * @param \DOMNode $newnode |
||
| 1123 | * @param \DOMNode $referencenode |
||
| 1124 | * @param int $clone |
||
| 1125 | * @return \DOMNode |
||
| 1126 | */ |
||
| 1127 | 6 | protected static function importNewnode(\DOMNode $newnode, \DOMNode $referencenode, $clone = 0) { |
|
| 1137 | |||
| 1138 | /** |
||
| 1139 | * Checks whether the first node in the set is disconnected (has no parent node) |
||
| 1140 | * |
||
| 1141 | * @return bool |
||
| 1142 | * @deprecated |
||
| 1143 | */ |
||
| 1144 | 1 | public function isDisconnected() |
|
| 1145 | { |
||
| 1146 | 1 | @trigger_error('Method isDisconnected() is deprecated and will be removed in 2.0.', E_USER_DEPRECATED); |
|
| 1147 | 1 | $parent = $this->getNode(0)->parentNode; |
|
| 1148 | 1 | return ($parent == null || $parent->tagName == self::FRAGMENT_ROOT_TAGNAME); |
|
| 1149 | } |
||
| 1150 | |||
| 1151 | 1 | public function __get($name) |
|
| 1152 | { |
||
| 1153 | 1 | switch ($name) { |
|
| 1154 | 1 | case 'count': |
|
| 1155 | 1 | case 'length': |
|
| 1156 | 1 | return count($this); |
|
| 1157 | } |
||
| 1158 | 1 | throw new \Exception('No such property ' . $name); |
|
| 1159 | } |
||
| 1160 | } |
||
| 1161 |
Loading history...
If you suppress an error, we recommend checking for the error condition explicitly: