Merge branch '4.4'

* 4.4:
  cs fix bis
  cs fix
  [DomCrawler] add a value() method, normalize whitespaces
This commit is contained in:
Nicolas Grekas 2019-09-27 19:13:30 +02:00
commit 335632152e
3 changed files with 24 additions and 4 deletions

View File

@ -14,6 +14,7 @@ CHANGELOG
* Added `Crawler::matches()` method.
* Added `Crawler::closest()` method.
* Added `Crawler::outerHtml()` method.
* Added an argument to the `Crawler::text()` method to opt-in normalizing whitespaces.
4.3.0
-----

View File

@ -569,15 +569,16 @@ class Crawler implements \Countable, \IteratorAggregate
}
/**
* Returns the node value of the first node of the list.
* Returns the text of the first node of the list.
*
* @param mixed $default When provided and the current node is empty, this value is returned and no exception is thrown
* @param mixed $default When provided and the current node is empty, this value is returned and no exception is thrown
* @param bool $normalizeWhitespace Whether whitespaces should be trimmed and normalized to single spaces
*
* @return string The node value
*
* @throws \InvalidArgumentException When current node is empty
*/
public function text($default = null)
public function text($default = null, bool $normalizeWhitespace = false)
{
if (!$this->nodes) {
if (0 < \func_num_args()) {
@ -587,7 +588,13 @@ class Crawler implements \Countable, \IteratorAggregate
throw new \InvalidArgumentException('The current node list is empty.');
}
return $this->getNode(0)->nodeValue;
$text = $this->getNode(0)->nodeValue;
if (\func_num_args() > 1 && func_get_arg(1)) {
return trim(preg_replace('/(?:\s{2,}+|[^\S ])/', ' ', $text));
}
return $text;
}
/**

View File

@ -253,6 +253,14 @@ abstract class AbstractCrawlerTest extends TestCase
$this->assertCount(0, $crawler->eq(100), '->eq() returns an empty crawler if the nth node does not exist');
}
public function testNormalizeWhiteSpace()
{
$crawler = $this->createTestCrawler()->filterXPath('//p');
$this->assertSame('Elsa <3', $crawler->text(null, true), '->text(null, true) returns the text with normalized whitespace');
$this->assertNotSame('Elsa <3', $crawler->text(null, false));
$this->assertNotSame('Elsa <3', $crawler->text());
}
public function testEach()
{
$data = $this->createTestCrawler()->filterXPath('//ul[1]/li')->each(function ($node, $i) {
@ -1235,6 +1243,10 @@ HTML;
<li>Two Bis</li>
<li>Three Bis</li>
</ul>
<p class="whitespace">
Elsa
&lt;3
</p>
<div id="parent">
<div id="child"></div>
<div id="child2" xmlns:foo="http://example.com"></div>