wikimedia/mediawiki-extensions-TextExtracts
1
0
Fork 0
mirror of https://gerrit.wikimedia.org/r/mediawiki/extensions/TextExtracts synced 2024-11-23 15:56:52 +00:00
Code Issues Packages Projects Releases Wiki Activity

ExtractFormatter: Rescue headings from being removed

Browse source 
Bug: T363445
Change-Id: I662fe3dd06d6b010108c6f0ef891a8c6113b9a45
This commit is contained in:
Bartosz Dziewoński 2024-06-20 22:34:38 +02:00 committed by Krinkle
parent e623c552d2
commit 0fafa44a20
2 changed files with 25 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

15
includes/ExtractFormatter.php
View file

@ -4,6 +4,7 @@ namespace TextExtracts;
use DOMElement;
use HtmlFormatter\HtmlFormatter;
use Wikimedia\Parsoid\Utils\DOMCompat;
/**
* Provides text-only or limited-HTML extracts of page HTML
@ -80,9 +81,21 @@ class ExtractFormatter extends HtmlFormatter {
* @return array Array of removed DOMElements
*/
public function filterContent(): array {
$doc = $this->getDoc();
// Headings in a DIV wrapper may get removed by $wgExtractsRemoveClasses,
// move it outside the header to rescue it (T363445)
// https://www.mediawiki.org/wiki/Heading_HTML_changes
$headings = DOMCompat::querySelectorAll( $doc->documentElement, 'h1, h2, h3, h4, h5, h6' );
foreach ( $headings as $heading ) {
// @phan-suppress-next-line PhanTypeMismatchArgumentSuperType
if ( DOMCompat::getClassList( $heading->parentNode )->contains( 'mw-heading' ) ) {
$heading->parentNode->parentNode->insertBefore( $heading, $heading->parentNode );
}
}
$removed = parent::filterContent();
$doc = $this->getDoc();
$spans = $doc->getElementsByTagName( 'span' );
/** @var DOMElement $span */

11
tests/phpunit/ExtractFormatterTest.php
View file

@ -67,6 +67,17 @@ class ExtractFormatterTest extends MediaWikiIntegrationTestCase {
'<span class="foo"><span style="bar: baz;" lang="qux">quux</span></span>',
false,
],
[
// New heading markup from T13555 (T363445)
"<p>hello</p><h2 id=\"heading\">heading</h2><p>bye</p>",
'<p>hello</p>' .
'<div class="mw-heading mw-heading2">' .
'<h2 id="heading">heading</h2>' .
'<span class="mw-editsection">xxx</span>' .
'</div>' .
'<p>bye</p>',
false,
],
[
// Verify that TOC is properly removed (HTML mode)
"Lead\n<h1>Section</h1>\n<p>Section text</p>",