Add HtmlHelper::modifyElements() for small HTML modifications

Adds HtmlHelper, which is intended to be a static utility class
for doing simple modifications to HTML strings in a safe way,
without exposing the caller to the complex Remex interface.
Currently only has one method, which is mainly useful for
modifying element attributes.

The code is largely based on the Wikibase FormatEntities class.

Bug: T217850
Change-Id: I45db9e61e47eb69df32a167d9d1dd146a8719676

1 files changed, 44 insertions, 0 deletions

diff --git a/tests/phpunit/unit/includes/HtmlHelperTest.php b/tests/phpunit/unit/includes/HtmlHelperTest.php
new file mode 100644
index 000000000000..bf432a7d6fa2
--- /dev/null
+++ b/tests/phpunit/unit/includes/HtmlHelperTest.php
@@ -0,0 +1,44 @@
+<?php
+
+namespace MediaWiki\Tests\Unit;
+
+use MediaWiki\HtmlHelper;
+use MediaWikiUnitTestCase;
+use Wikimedia\RemexHtml\HTMLData;
+use Wikimedia\RemexHtml\Serializer\SerializerNode;
+use Wikimedia\RemexHtml\Tokenizer\PlainAttributes;
+
+class HtmlHelperTest extends MediaWikiUnitTestCase {
+
+	/**
+	 * @covers \MediaWiki\HtmlHelper::modifyElements
+	 */
+	public function testModifyElements() {
+		$shouldModifyCallback = static function ( SerializerNode $node ) {
+			return $node->namespace === HTMLData::NS_HTML
+				&& $node->name === 'a'
+				&& isset( $node->attrs['href'] );
+		};
+		$modifyCallbackInPlace = static function ( SerializerNode $node ) {
+			$node->attrs['href'] = 'https://tracker.org/' . $node->attrs['href'];
+			return $node;
+		};
+		$input = '<p>Foo <a href="https://example.org/">bar</a> baz</p>';
+		$expectedOutput = '<p>Foo <a href="https://tracker.org/https://example.org/">bar</a> baz</p>';
+
+		$output = HtmlHelper::modifyElements( $input, $shouldModifyCallback, $modifyCallbackInPlace );
+		$this->assertSame( $expectedOutput, $output );
+
+		$modifyCallbackNew = static function ( SerializerNode $node ) {
+			$href = 'https://tracker.org/' . $node->attrs['href'];
+			$newNode = new SerializerNode( $node->id, $node->parentId, $node->namespace, $node->name,
+				new PlainAttributes( [ 'href' => $href ] ), $node->void );
+			$node->attrs['href'] = 'https://tracker.org/' . $node->attrs['href'];
+			return $newNode;
+		};
+
+		$output = HtmlHelper::modifyElements( $input, $shouldModifyCallback, $modifyCallbackNew );
+		$this->assertSame( $expectedOutput, $output );
+	}
+
+}