Handle <sub> and <sup> elements as neighbors. Fixes #134.

mundschenk-at · Mar 24, 2024 · 71c246c · 71c246c
1 parent 0156382
commit 71c246c
Show file tree

Hide file tree

Showing 4 changed files with 66 additions and 31 deletions.
diff --git a/src/class-dom.php b/src/class-dom.php
@@ -63,6 +63,17 @@ abstract class DOM {
 		'math',
 	];
 
+	/**
+	 * Elements that are acceptable as neighbor nodes.
+	 *
+	 * @since 7.0.0
+	 */
+	private const ACCEPTABLE_NEIGHBOR_ELEMENTS = [
+		'br'  => true,
+		'sub' => true,
+		'sup' => true,
+	];
+
 	/**
 	 * Retrieves an array of block tags.
 	 *
@@ -235,12 +246,20 @@ public static function get_next_chr( \DOMNode $node ) {
 	 * @return string The character or an empty string.
 	 */
 	private static function get_adjacent_character( \DOMNode $node, $position, $length, callable $get_textnode ) {
-		$adjacent_node = $get_textnode( [ __CLASS__, 'is_text_or_linebreak' ], $node );
+		$adjacent_node = $get_textnode( [ __CLASS__, 'is_acceptable_neighbor_node' ], $node );
 		$character     = '';
 
 		if ( null !== $adjacent_node ) {
-			if ( self::is_linebreak( $adjacent_node ) ) {
-				$character = ' ';
+			if ( $adjacent_node instanceof \DOMElement ) {
+				switch ( $adjacent_node->tagName ) {
+					case 'br':
+						$character = ' ';
+						break;
+					case 'sub':
+					case 'sup':
+					default:
+						$character = '';
+				}
 			} elseif ( $adjacent_node instanceof \DOMText ) {
 				$node_data = $adjacent_node->data;
 				$character = \preg_replace( '/\p{C}/Su', '', Strings::functions( $node_data )['substr']( $node_data, $position, $length ) );
@@ -250,19 +269,6 @@ private static function get_adjacent_character( \DOMNode $node, $position, $leng
 		return $character;
 	}
 
-	/**
-	 * Determines if the node is a <br> element.
-	 *
-	 * @since 7.0.0
-	 *
-	 * @param ?\DOMNode $node The node to test.
-	 *
-	 * @return bool
-	 */
-	private static function is_linebreak( ?\DOMNode $node ): bool {
-		return $node instanceof \DOMElement && ( $node->tagName ?? '' ) === 'br';
-	}
-
 	/**
 	 * Determines if the node is a textnode.
 	 *
@@ -277,16 +283,16 @@ private static function is_textnode( ?\DOMNode $node ): bool {
 	}
 
 	/**
-	 * Determines if the node is a textnode or <br> element.
+	 * Determines if the node is a textnode or one of the acceptable elements.
 	 *
 	 * @since 7.0.0
 	 *
 	 * @param ?\DOMNode $node The node to test.
 	 *
 	 * @return bool
 	 */
-	private static function is_text_or_linebreak( ?\DOMNode $node ): bool {
-		return $node instanceof \DOMText || ( $node instanceof \DOMElement && 'br' === $node->tagName );
+	private static function is_acceptable_neighbor_node( ?\DOMNode $node ): bool {
+		return $node instanceof \DOMText || ( $node instanceof \DOMElement && isset( self::ACCEPTABLE_NEIGHBOR_ELEMENTS[ $node->tagName ] ) );
 	}
 
 

diff --git a/tests/class-dom-test.php b/tests/class-dom-test.php
@@ -302,8 +302,7 @@ public function test_get_block_parent_name( $input_xpath, $parent_xpath, $parent
 	 * @covers ::get_previous_acceptable_node
 	 * @covers ::get_adjacent_node
 	 * @covers ::is_block_tag
-	 * @covers ::is_linebreak
-	 * @covers ::is_text_or_linebreak
+	 * @covers ::is_acceptable_neighbor_node
 	 *
 	 * @uses ::get_last_acceptable_node
 	 * @uses ::get_edge_node
@@ -330,8 +329,7 @@ public function test_get_prev_chr() {
 	 * @covers ::get_adjacent_character
 	 * @covers ::get_previous_acceptable_node
 	 * @covers ::get_adjacent_node
-	 * @covers ::is_linebreak
-	 * @covers ::is_text_or_linebreak
+	 * @covers ::is_acceptable_neighbor_node
 	 *
 	 * @uses ::is_block_tag
 	 * @uses ::get_last_acceptable_node
@@ -374,8 +372,7 @@ public function test_get_previous_textnode_null() {
 	 * @covers ::get_next_acceptable_node
 	 * @covers ::get_adjacent_node
 	 * @covers ::is_block_tag
-	 * @covers ::is_linebreak
-	 * @covers ::is_text_or_linebreak
+	 * @covers ::is_acceptable_neighbor_node
 	 *
 	 * @uses ::get_first_acceptable_node
 	 * @uses ::get_edge_node
@@ -403,8 +400,7 @@ public function test_get_next_chr() {
 	 * @covers ::get_next_acceptable_node
 	 * @covers ::get_adjacent_node
 	 * @covers ::is_block_tag
-	 * @covers ::is_linebreak
-	 * @covers ::is_text_or_linebreak
+	 * @covers ::is_acceptable_neighbor_node
 	 *
 	 * @uses ::get_first_acceptable_node
 	 * @uses ::get_edge_node
@@ -424,6 +420,34 @@ public function test_get_next_chr_with_br() {
 		$this->assertSame( ' ', $prev_char );
 	}
 
+	/**
+	 * Test get_next_chr followed by <br>.
+	 *
+	 * @covers ::get_next_chr
+	 * @covers ::get_adjacent_character
+	 * @covers ::get_next_acceptable_node
+	 * @covers ::get_adjacent_node
+	 * @covers ::is_block_tag
+	 * @covers ::is_acceptable_neighbor_node
+	 *
+	 * @uses ::get_first_acceptable_node
+	 * @uses ::get_edge_node
+	 * @uses PHP_Typography\Strings::functions
+	 */
+	public function test_get_next_chr_with_sub_sup() {
+		$html  = '<p><span id="foo">A</span><span id="bar"><sup>new</sup> hope.</span><sub>x</sub></p><p><span>The empire</span> strikes back.</p<';
+		$doc   = $this->load_html( $html );
+		$xpath = new \DOMXPath( $doc );
+
+		$textnodes = $xpath->query( "//*[@id='foo']/text()" ); // really only one.
+		$prev_char = DOM::get_next_chr( $textnodes->item( 0 ) );
+		$this->assertSame( '', $prev_char );
+
+		$textnodes = $xpath->query( "//*[@id='bar']/text()" ); // really only one.
+		$prev_char = DOM::get_next_chr( $textnodes->item( 0 ) );
+		$this->assertSame( '', $prev_char );
+	}
+
 	/**
 	 * Test get_next_textnode.
 	 *

diff --git a/tests/class-php-typography-test.php b/tests/class-php-typography-test.php
@@ -506,6 +506,7 @@ public function provide_process_data() {
 			[ 'ein-, zweimal', 'ein&#8209;, zweimal', true ],
 			[ 'В зависимости от региона, может выращиватся на зерно и силос. После колосовых может выращиватся на второй посев.', 'В зависимости от региона, может выращиватся на зерно и&nbsp;силос. После колосовых может выращиватся на второй посев.', false ],
 			[ '"Text."<br>Text after.', '<span class="pull-double">&ldquo;</span>Text.&rdquo;<br>Text after.', '&ldquo;Text.&rdquo;<br>Text after.' ],
+			[ 'à "l’âge"<sup>N112</sup>', '&agrave; <span class="push-double"></span>&#8203;<span class="pull-double">&ldquo;</span>l&rsquo;&acirc;ge&rdquo;<sup><span class="caps">N<span class="numbers">112</span></span></sup>', '&agrave; &ldquo;l&rsquo;&acirc;ge&rdquo;<sup>N112</sup>' ],
 		];
 	}
 

diff --git a/tests/fixes/node-fixes/class-smart-quotes-fix-test.php b/tests/fixes/node-fixes/class-smart-quotes-fix-test.php
@@ -119,10 +119,12 @@ public function provide_smart_quotes_special_data() {
 				'"',
 			],
 			[
-				'à "l’âge"<br>N112',
-				'&agrave; &laquo;&#8239;l&rsquo;&acirc;ge&#8239;&raquo;<br>N112',
+				'à "l’âge"',
+				'&agrave; &laquo;&#8239;l&rsquo;&acirc;ge&#8239;&raquo;',
 				Quote_Style::DOUBLE_GUILLEMETS_FRENCH,
 				Quote_Style::SINGLE_GUILLEMETS,
+				null,
+				new \DOMElement( 'br' ),
 			],
 			[
 				'à "l’âge" N112',
@@ -131,10 +133,12 @@ public function provide_smart_quotes_special_data() {
 				Quote_Style::SINGLE_GUILLEMETS,
 			],
 			[
-				'à "l’âge"<sup>N112</sup>',
-				'&agrave; &laquo;&#8239;l&rsquo;&acirc;ge&#8239;&raquo;<sup>N112</sup>',
+				'à "l’âge"',
+				'&agrave; &laquo;&#8239;l&rsquo;&acirc;ge&#8239;&raquo;',
 				Quote_Style::DOUBLE_GUILLEMETS_FRENCH,
 				Quote_Style::SINGLE_GUILLEMETS,
+				null,
+				new \DOMElement( 'sup', 'N112' ),
 			],
 		];
 	}