wikimedia/mediawiki-extensions-TextExtracts
1
0
Fork 0
mirror of https://gerrit.wikimedia.org/r/mediawiki/extensions/TextExtracts synced 2024-11-15 03:35:20 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge "Tidy is no longer configurable in MW 1.35"

Browse source
This commit is contained in:
jenkins-bot 2020-06-15 12:31:10 +00:00 committed by Gerrit Code Review
parent 5cba86206a c1397847c0
commit 47f07178ee
3 changed files with 20 additions and 29 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

7
includes/ApiQueryExtracts.php
View file

@ -10,7 +10,6 @@ use Config;
use FauxRequest;
use MediaWiki\Logger\LoggerFactory;
use MediaWiki\MediaWikiServices;
use MWTidy;
use ParserOptions;
use Title;
use User;
@ -304,10 +303,10 @@ class ApiQueryExtracts extends ApiQueryBase {
* @return string
*/
private function truncate( $text ) {
if ( !$this->params['plaintext'] && MWTidy::isEnabled() ) {
$truncator = new TextTruncator( MWTidy::singleton() );
if ( !$this->params['plaintext'] ) {
$truncator = new TextTruncator( true );
} else {
$truncator = new TextTruncator();
$truncator = new TextTruncator( false );
}
if ( $this->params['chars'] ) {

24
includes/TextTruncator.php
View file

@ -2,27 +2,26 @@
namespace TextExtracts;
use MediaWiki\Tidy\TidyDriverBase;
use MWTidy;
/**
* This class needs to understand HTML as well as plain text. It tries to not break HTML tags, but
* might break pairs of tags, leaving unclosed tags behind. A Tidy instance must be provided to fix
* might break pairs of tags, leaving unclosed tags behind. We can tidy the output to fix
* this.
*
* @license GPL-2.0-or-later
*/
class TextTruncator {
/**
* @var bool Whether to tidy the output
*/
private $useTidy;
/**
* @var TidyDriverBase|null
* @param bool $useTidy
*/
private $tidyDriver;
/**
* @param TidyDriverBase|null $tidy
*/
public function __construct( TidyDriverBase $tidy = null ) {
$this->tidyDriver = $tidy;
public function __construct( bool $useTidy ) {
$this->useTidy = $useTidy;
}
/**
@ -99,9 +98,8 @@ class TextTruncator {
* @return string
*/
private function tidy( $text ) {
if ( $this->tidyDriver ) {
// Fix possibly unclosed HTML tags.
$text = $this->tidyDriver->tidy( $text );
if ( $this->useTidy ) {
$text = MWTidy::tidy( $text );
}
return trim( $text );

18
tests/phpunit/TextTruncatorTest.php
View file

@ -2,7 +2,6 @@
namespace TextExtracts\Test;
use MediaWiki\Tidy\TidyDriverBase;
use TextExtracts\TextTruncator;
/**
@ -19,7 +18,7 @@ class TextTruncatorTest extends \PHPUnit\Framework\TestCase {
* @param string $expected
*/
public function testGetFirstSentences( $text, $sentences, $expected ) {
$truncator = new TextTruncator();
$truncator = new TextTruncator( false );
$this->assertSame( $expected, $truncator->getFirstSentences( $text, $sentences ) );
}
@ -136,7 +135,7 @@ class TextTruncatorTest extends \PHPUnit\Framework\TestCase {
* @param string $expected
*/
public function testGetFirstChars( $text, $chars, $expected ) {
$truncator = new TextTruncator();
$truncator = new TextTruncator( false );
$this->assertSame( $expected, $truncator->getFirstChars( $text, $chars ) );
}
@ -169,16 +168,11 @@ class TextTruncatorTest extends \PHPUnit\Framework\TestCase {
}
public function testTidyIntegration() {
$tidy = $this->createMock( TidyDriverBase::class );
$tidy->method( 'tidy' )
->willReturnCallback( function ( $text ) {
return "<tidy>$text</tidy>";
} );
$truncator = new TextTruncator( $tidy );
$truncator = new TextTruncator( true );
$text = 'Aa. Bb.';
$this->assertSame( '<tidy>Aa.</tidy>', $truncator->getFirstSentences( $text, 1 ) );
$this->assertSame( '<tidy>Aa</tidy>', $truncator->getFirstChars( $text, 1 ) );
$text = '<b>Aa. Bb.</b>';
$this->assertSame( '<p><b>Aa.</b></p>', $truncator->getFirstSentences( $text, 1 ) );
$this->assertSame( '<p><b>Aa</b></p>', $truncator->getFirstChars( $text, 4 ) );
}
}