Don't apply French spacing in raw text elements

This also means we don't need to take special care for French spacing in
attributes, since it's no longer applied there.

Adds a test that captures this change.

Note that the test "Nowiki and french spacing" wonders whether this
escaping should be applied to nowiki content.

Bug: T255007
Change-Id: Ic8965e81882d7cf024bdced437f684064a30ac86

includes/parser/Parser.php

@@ -36,6 +36,7 @@ use MediaWiki\Revision\RevisionAccessException;
 use MediaWiki\Revision\RevisionRecord;
 use MediaWiki\Revision\SlotRecord;
 use MediaWiki\SpecialPage\SpecialPageFactory;
+use MediaWiki\Tidy\RemexDriver;
 use Psr\Log\LoggerInterface;
 use Psr\Log\NullLogger;
 use Wikimedia\IPUtils;
@@ -354,6 +355,9 @@ class Parser {
 	/** @var HookRunner */
 	private $hookRunner;
 
+	/** @var RemexDriver */
+	private $remexDriver;
+
 	/**
 	 * @internal For use by ServiceWiring
 	 */
@@ -378,7 +382,8 @@ class Parser {
 		'StylePath',
 		'TranscludeCacheExpiry',
 		'PreprocessorCacheThreshold',
-		'DisableLangConversion'
+		'DisableLangConversion',
+		'TidyConfig',
 	];
 
 	/**
@@ -468,6 +473,10 @@ class Parser {
 			MediaWikiServices::getInstance()->getHookContainer();
 		$this->hookRunner = new HookRunner( $this->hookContainer );
 
+		$this->remexDriver = new RemexDriver(
+			$this->svcOptions->get( 'TidyConfig' ) ?? []
+		);
+
 		// T250444: This will eventually be inlined here and the
 		// standalone method removed.
 		$this->firstCallInit();
@@ -1673,12 +1682,9 @@ class Parser {
 
 		$text = $this->mStripState->unstripGeneral( $text );
 
-		# Clean up special characters, only run once, after doBlockLevels
-		$text = Sanitizer::armorFrenchSpaces( $text );
-
 		$text = Sanitizer::normalizeCharReferences( $text );
 
-		$text = MWTidy::tidy( $text );
+		$text = $this->remexDriver->tidy( $text, [ Sanitizer::class, 'armorFrenchSpaces' ] );
 
 		if ( $isMain ) {
 			$this->hookRunner->onParserAfterTidy( $this, $text );

includes/parser/Sanitizer.php

@@ -784,9 +784,6 @@ class Sanitizer {
 			'__'   => '__',
 		] );
 
-		# Armor against French spaces detection (T5158)
-		$encValue = self::armorFrenchSpaces( $encValue, ' ' );
-
 		# Stupid hack
 		$encValue = preg_replace_callback(
 			'/((?i)' . wfUrlProtocols() . ')/',

includes/tidy/RemexCompatFormatter.php

@@ -16,18 +16,34 @@ class RemexCompatFormatter extends HtmlFormatter {
 		'tr' => true,
 	];
 
+	/* @var ?callable */
+	private $textProcessor;
+
 	public function __construct( $options = [] ) {
 		parent::__construct( $options );
 		$this->attributeEscapes["\u{00A0}"] = ' ';
 		unset( $this->attributeEscapes["&"] );
 		$this->textEscapes["\u{00A0}"] = ' ';
 		unset( $this->textEscapes["&"] );
+		$this->textProcessor = $options['textProcessor'] ?? null;
 	}
 
 	public function startDocument( $fragmentNamespace, $fragmentName ) {
 		return '';
 	}
 
+	public function characters( SerializerNode $parent, $text, $start, $length ) {
+		$text = parent::characters( $parent, $text, $start, $length );
+		if ( $parent->namespace !== HTMLData::NS_HTML
+			|| !isset( $this->rawTextElements[$parent->name] )
+		) {
+			if ( $this->textProcessor !== null ) {
+				$text = call_user_func( $this->textProcessor, $text );
+			}
+		}
+		return $text;
+	}
+
 	public function element( SerializerNode $parent, SerializerNode $node, $contents ) {
 		$data = $node->snData;
 		if ( $data && $data->isPWrapper ) {

includes/tidy/RemexDriver.php

@@ -29,12 +29,11 @@ class RemexDriver extends TidyDriverBase {
 		parent::__construct( $config );
 	}
 
-	public function tidy( $text ) {
+	public function tidy( $text, callable $textProcessor = null ) {
 		$traceCallback = static function ( $msg ) {
 			wfDebug( "RemexHtml: $msg" );
 		};
-
-		$formatter = new RemexCompatFormatter;
+		$formatter = new RemexCompatFormatter( [ 'textProcessor' => $textProcessor ] );
 		if ( $this->serializerTrace ) {
 			$serializer = new SerializerWithTracer( $formatter, null, $traceCallback );
 		} else {

tests/parser/parserTests.txt

@@ -1754,6 +1754,23 @@ Nowiki and french spacing
 <p><span typeof="mw:Nowiki">test<span typeof="mw:DisplaySpace"> </span>: 123</span></p>
 !! end
 
+!! test
+T255007: French spacing in raw text elements
+!! options
+wgRawHtml=1
+!! wikitext
+<html>
+<script>test ; 123</script>
+<style>test : 123</style>
+</html>
+!! html/php
+<p>
+<script>test ; 123</script>
+<style>test : 123</style>
+
+</p>
+!! end
+
 ###
 ### Comments
 ###
@@ -18253,7 +18270,7 @@ Punctuation: CSS ! important (T13874; with space after)
 !! wikitext
 <div style="width:50% ! important">important</div>
 !! html
-<div style="width:50%&#32;! important">important</div>
+<div style="width:50% ! important">important</div>
 !! end
 
 !! test
@@ -22820,7 +22837,7 @@ Play a bit with r67090 and T5158
 <div style="width:50% !important">&#160;</div>
 <div style="width:50%&#160;!important">&#160;</div>
 <div style="width:50%&#160;!important">&#160;</div>
-<div style="border&#32;: solid;">&#160;</div>
+<div style="border : solid;">&#160;</div>
 !! html/parsoid
 <div style="width:50% !important" data-parsoid='{"stx":"html"}'><span typeof="mw:Entity" data-parsoid='{"srcContent":" "}'> </span></div>
 <div style="width:50% !important" data-parsoid='{"stx":"html","a":{"style":"width:50% !important"},"sa":{"style":"width:50%&amp;nbsp;!important"}}'><span typeof="mw:Entity" data-parsoid='{"srcContent":" "}'> </span></div>
@@ -22860,7 +22877,7 @@ T5158: Test for French spaces in attributes
 !! wikitext
 <br style=" clear : both ; " />
 !! html/php
-<p><br style="clear&#32;: both&#32;;" />
+<p><br style="clear : both ;" />
 </p>
 !! end
 

tests/phpunit/includes/parser/ParserTest.php

@@ -8,7 +8,11 @@ class ParserTest extends MediaWikiIntegrationTestCase {
 		// Create a mock Config object that will satisfy ServiceOptions::__construct
 		$mockConfig = $this->createMock( Config::class );
 		$mockConfig->method( 'has' )->willReturn( true );
-		$mockConfig->method( 'get' )->willReturn( 'I like otters.' );
+		$mockConfig->method( 'get' )->will(
+			$this->returnCallback( function ( $arg ) {
+				return ( $arg === 'TidyConfig' ) ? null : 'I like otters.';
+			} )
+		);
 
 		// Stub out a MagicWordFactory so the Parser can initialize its
 		// function hooks when it is created.