Formatting: Add strip_html_newlines() function

src/wp-includes/formatting.php

@@ -868,6 +868,60 @@ function shortcode_unautop( $text ) {
 	return preg_replace( $pattern, '$1', $text );
 }
 
+/**
+ * Strips unnecessary newlines from HTML content.
+ *
+ * @since 7.1.0
+ *
+ * @param string $text The HTML content to strip newlines from.
+ * @return string The HTML content with unnecessary newlines removed.
+ */
+function strip_html_newlines( $text ) {
+	if ( ! str_contains( $text, "\n" ) && ! str_contains( $text, "\r" ) ) {
+		return $text;
+	}
+
+	$preserve_newlines_elements = array( 'pre', 'code', 'kbd', 'script', 'style' );
+
+	$textarr = wp_html_split( $text );
+	$changed = false;
+	$skip    = false;
+
+	foreach ( $textarr as $i => $chunk ) {
+		if ( '' === $chunk ) {
+			continue;
+		}
+
+		if ( 0 !== $i % 2 ) {
+			foreach ( $preserve_newlines_elements as $element ) {
+				if ( stripos( $chunk, '<' . $element ) === 0 ) {
+					$skip = true;
+					break;
+				}
+				if ( stripos( $chunk, '</' . $element . '>' ) === 0 ) {
+					$skip = false;
+					break;
+				}
+			}
+			continue;
+		}
+
+		if ( ! $skip ) {
+			$stripped = preg_replace( '/[\n\r]+/', ' ', $chunk );
+			if ( $stripped !== $chunk ) {
+				$textarr[ $i ] = $stripped;
+				$changed       = true;
+			}
+		}
+	}
+
+	if ( $changed ) {
+		return implode( '', $textarr );
+	}
+
+	return $text;
+}
+
 /**
  * Checks to see if a string is utf8 encoded.
  *

tests/phpunit/tests/formatting/stripHtmlNewlines.php

@@ -0,0 +1,54 @@
+<?php
+
+/**
+ * Tests for strip_html_newlines().
+ *
+ * @group formatting
+ *
+ * @covers ::strip_html_newlines
+ */
+class Tests_Formatting_StripHtmlNewlines extends WP_UnitTestCase {
+
+	/**
+	 * Verifies that newlines and carriage returns in text nodes are replaced
+	 * with spaces, including across inline elements like anchors.
+	 *
+	 * @ticket 5678
+	 */
+	public function test_strips_newlines_from_text_nodes() {
+		$this->assertSame( '', strip_html_newlines( '' ), 'Empty string should be returned as-is.' );
+		$this->assertSame( '<p>No newlines here.</p>', strip_html_newlines( '<p>No newlines here.</p>' ), 'Text without newlines should be returned as-is.' );
+		$this->assertSame( '<p>Line one Line two Line three</p>', strip_html_newlines( "<p>Line one\n\nLine two\r\nLine three</p>" ), 'Multiple newlines and carriage returns should be collapsed to a single space.' );
+		$this->assertSame(
+			'<p>This is a paragraph in which the wpautop() <a href="#elsewhere">wrapping will happen in the middle</a> of an anchor, which is an inline element.</p>',
+			strip_html_newlines( "<p>This is a paragraph in which the\nwpautop() <a href=\"#elsewhere\">wrapping will\nhappen in the middle</a> of an\nanchor, which is an inline element.</p>" ),
+			'Newlines within and around inline elements should be stripped.'
+		);
+	}
+
+	/**
+	 *
+	 * @ticket 5678
+	 */
+	public function test_preserves_newlines_in_preformatted_elements() {
+		$input  = "<p>Normal\ntext</p>\n<pre>\nPreformatted\nlines\n</pre>\n<p>More\ntext</p>";
+		$result = strip_html_newlines( $input );
+
+		$this->assertStringContainsString( 'Normal text', $result, 'Newlines in normal text should be stripped.' );
+		$this->assertStringContainsString( 'More text', $result, 'Newlines in trailing paragraph should be stripped.' );
+		$this->assertStringContainsString( "\nPreformatted\nlines\n", $result, 'Newlines inside <pre> should be preserved.' );
+
+		$preserved_cases = array(
+			'code'   => "<p>A\nB</p><code>x\ny</code>",
+			'kbd'    => "<p>A\nB</p><kbd>x\ny</kbd>",
+			'script' => "<p>A\nB</p><script>x\ny</script>",
+			'style'  => "<p>A\nB</p><style>x\ny</style>",
+		);
+
+		foreach ( $preserved_cases as $tag => $html ) {
+			$out = strip_html_newlines( $html );
+			$this->assertStringContainsString( 'A B', $out, "Text node newline should be stripped around <{$tag}>." );
+			$this->assertStringContainsString( "x\ny", $out, "Newline inside <{$tag}> should be preserved." );
+		}
+	}
+}