0004-Use-HtmlHelper-modifyElements-instead-of-regexps-to-.patch
matmarex (Bartosz Dziewoński)
Actions

Authored By

	matmarex
	Apr 29 2023, 3:14 PM

Size

3 KB

Referenced Files

None

Subscribers

None

0004-Use-HtmlHelper-modifyElements-instead-of-regexps-to-.patch
View Options

	From 85bc65c547348043e7960bd2c5b4613e5f9453c3 Mon Sep 17 00:00:00 2001
	From: =?UTF-8?q?Bartosz=20Dziewo=C5=84ski?= <matma.rex@gmail.com>
	Date: Sat, 29 Apr 2023 17:07:54 +0200
	Subject: [PATCH 4/4] Use HtmlHelper::modifyElements() instead of regexps to
	modify links

	This may be slower, but the result of this method is already cached,
	and it is more correct (e.g. internal links with fragments were
	previously handled incorrectly).

	Change-Id: I323e80b777b63425989f67f36a7e4d9a34fdaf6b
	---
	includes/DoubleWiki.php \| 69 +++++++++++++++++++++++++++++------------
	1 file changed, 49 insertions(+), 20 deletions(-)

	diff --git a/includes/DoubleWiki.php b/includes/DoubleWiki.php
	index 92ccc13..d73db7c 100644
	--- a/includes/DoubleWiki.php
	+++ b/includes/DoubleWiki.php
	@@ -24,6 +24,7 @@ use Language;
	use MediaWiki\Hook\BeforePageDisplayHook;
	use MediaWiki\Hook\OutputPageBeforeHTMLHook;
	use MediaWiki\Html\Html;
	+use MediaWiki\Html\HtmlHelper;
	use MediaWiki\Http\HttpRequestFactory;
	use MediaWiki\Languages\LanguageFactory;
	use MediaWiki\Languages\LanguageNameUtils;
	@@ -31,6 +32,7 @@ use OutputPage;
	use Skin;
	use Title;
	use WANObjectCache;
	+use Wikimedia\RemexHtml\Serializer\SerializerNode;

	class DoubleWiki implements OutputPageBeforeHTMLHook, BeforePageDisplayHook {

	@@ -134,33 +136,60 @@ class DoubleWiki implements OutputPageBeforeHTMLHook, BeforePageDisplayHook {
	*/
	private function getMangledTextAndTranslation( string $text, string $translation, string $matchLangCode ): array {
	// add prefixes to internal links, in order to prevent duplicates
	- $translation = preg_replace(
	- "/<a href=\"#(.*?)\"/i",
	- "<a href=\"#l_\\1\"",
	- $translation
	+ $translation = HtmlHelper::modifyElements(
	+ $translation,
	+ static function ( SerializerNode $n ): bool {
	+ return $n->name === 'a' && isset( $n->attrs['href'] ) && str_starts_with( $n->attrs['href'], '#' );
	+ },
	+ static function ( SerializerNode $n ): SerializerNode {
	+ $n->attrs['href'] = '#l_' . substr( $n->attrs['href'], 1 );
	+ return $n;
	+ }
	);
	- $translation = preg_replace(
	- "/<li id=\"(.*?)\"/i",
	- "<li id=\"l_\\1\"",
	- $translation
	+ $translation = HtmlHelper::modifyElements(
	+ $translation,
	+ static function ( SerializerNode $n ): bool {
	+ return $n->name === 'li' && isset( $n->attrs['id'] );
	+ },
	+ static function ( SerializerNode $n ): SerializerNode {
	+ $n->attrs['id'] = 'l_' . $n->attrs['id'];
	+ return $n;
	+ }
	);

	- $text = preg_replace(
	- "/<a href=\"#(.*?)\"/i",
	- "<a href=\"#r_\\1\"",
	- $text
	+ $text = HtmlHelper::modifyElements(
	+ $text,
	+ static function ( SerializerNode $n ): bool {
	+ return $n->name === 'a' && isset( $n->attrs['href'] ) && str_starts_with( $n->attrs['href'], '#' );
	+ },
	+ static function ( SerializerNode $n ): SerializerNode {
	+ $n->attrs['href'] = '#r_' . substr( $n->attrs['href'], 1 );
	+ return $n;
	+ }
	);
	- $text = preg_replace(
	- "/<li id=\"(.*?)\"/i",
	- "<li id=\"r_\\1\"",
	- $text
	+ $text = HtmlHelper::modifyElements(
	+ $text,
	+ static function ( SerializerNode $n ): bool {
	+ return $n->name === 'li' && isset( $n->attrs['id'] );
	+ },
	+ static function ( SerializerNode $n ): SerializerNode {
	+ $n->attrs['id'] = 'r_' . $n->attrs['id'];
	+ return $n;
	+ }
	);

	// add ?match= to local links of the local wiki
	- $text = preg_replace(
	- "/<a href=\"\/([^\"\?]*)\"/i",
	- "<a href=\"/\\1?match={$matchLangCode}\"",
	- $text
	+ $text = HtmlHelper::modifyElements(
	+ $text,
	+ static function ( SerializerNode $n ): bool {
	+ return $n->name === 'a' && isset( $n->attrs['href'] )
	+ && str_starts_with( $n->attrs['href'], '/' )
	+ && !str_contains( $n->attrs['href'], '?' );
	+ },
	+ static function ( SerializerNode $n ) use ( $matchLangCode ): SerializerNode {
	+ $n->attrs['href'] = wfAppendQuery( $n->attrs['href'], [ 'match' => $matchLangCode ] );
	+ return $n;
	+ }
	);

	return [ $text, $translation ];
	--
	2.28.0.windows.1

File Metadata

Mime Type: text/x-diff
Storage Engine: blob
Storage Format: Raw Data
Storage Handle: 10808814
Default Alt Text: 0004-Use-HtmlHelper-modifyElements-instead-of-regexps-to-.patch (3 KB)

0004-Use-HtmlHelper-modifyElements-instead-of-regexps-to-.patchmatmarex (Bartosz Dziewoński)Actions

0004-Use-HtmlHelper-modifyElements-instead-of-regexps-to-.patchView Options

File Metadata

Event Timeline

0004-Use-HtmlHelper-modifyElements-instead-of-regexps-to-.patch
matmarex (Bartosz Dziewoński)
Actions

0004-Use-HtmlHelper-modifyElements-instead-of-regexps-to-.patch
View Options