Comments: Encode emoji for utf8mb3 comment columns in wp_insert_comme…

src/wp-includes/comment.php

@@ -2162,6 +2162,8 @@ function wp_insert_comment( $commentdata ) {
 		'user_id'
 	);
 
+	$compacted = _wp_encode_emoji_in_fields( $compacted, $wpdb->comments, array( 'comment_author', 'comment_content' ) );
+
 	if ( ! $wpdb->insert( $wpdb->comments, $compacted ) ) {
 		return false;
 	}
@@ -2759,6 +2761,8 @@ function wp_update_comment( $commentarr, $wp_error = false ) {
 
 	$data = wp_array_slice_assoc( $data, $keys );
 
+	$data = _wp_encode_emoji_in_fields( $data, $wpdb->comments, array( 'comment_author', 'comment_content' ) );
+
 	$result = $wpdb->update( $wpdb->comments, $data, array( 'comment_ID' => $comment_id ) );
 
 	if ( false === $result ) {

src/wp-includes/formatting.php

@@ -6116,6 +6116,42 @@ function wp_encode_emoji( $content ) {
 	return $content;
 }
 
+/**
+ * Encodes emoji in the given fields of a data array whose corresponding
+ * database columns cannot natively store 4-byte UTF-8 characters.
+ *
+ * Used by wp_insert_post(), wp_insert_comment(), and wp_update_comment() so
+ * that emoji (and other 4-byte characters) are stored as HTML entities on
+ * `utf8`/`utf8mb3` columns instead of causing truncated or failed writes.
+ *
+ * @access private
+ * @since 7.1.0
+ *
+ * @global wpdb $wpdb WordPress database abstraction object.
+ *
+ * @param array  $data   Array of field name => value pairs to check, keyed the
+ *                       same way as the corresponding database columns.
+ * @param string $table  Name of the database table the fields belong to.
+ * @param array  $fields Names of the fields within $data that may contain emoji.
+ * @return array $data with emoji encoded in the given fields, where needed.
+ */
+function _wp_encode_emoji_in_fields( $data, $table, $fields ) {
+	global $wpdb;
+
+	foreach ( $fields as $field ) {
+		if ( isset( $data[ $field ] ) ) {
+			$charset = $wpdb->get_col_charset( $table, $field );
+
+			// The 'utf8' character set is a deprecated alias of 'utf8mb3'. See <https://dev.mysql.com/doc/refman/8.4/en/charset-unicode-utf8.html>.
+			if ( 'utf8' === $charset || 'utf8mb3' === $charset ) {
+				$data[ $field ] = wp_encode_emoji( $data[ $field ] );
+			}
+		}
+	}
+
+	return $data;
+}
+
 /**
  * Converts emoji to a static img element.
  *

src/wp-includes/post.php

@@ -4917,18 +4917,7 @@ function wp_insert_post( $postarr, $wp_error = false, $fire_after_hooks = true )
 		'guid'
 	);
 
-	$emoji_fields = array( 'post_title', 'post_content', 'post_excerpt' );
-
-	foreach ( $emoji_fields as $emoji_field ) {
-		if ( isset( $data[ $emoji_field ] ) ) {
-			$charset = $wpdb->get_col_charset( $wpdb->posts, $emoji_field );
-
-			// The 'utf8' character set is a deprecated alias of 'utf8mb3'. See <https://dev.mysql.com/doc/refman/8.4/en/charset-unicode-utf8.html>.
-			if ( 'utf8' === $charset || 'utf8mb3' === $charset ) {
-				$data[ $emoji_field ] = wp_encode_emoji( $data[ $emoji_field ] );
-			}
-		}
-	}
+	$data = _wp_encode_emoji_in_fields( $data, $wpdb->posts, array( 'post_title', 'post_content', 'post_excerpt' ) );
 
 	if ( 'attachment' === $post_type ) {
 		/**

tests/phpunit/tests/comment.php

@@ -2045,4 +2045,112 @@ public function test_get_comment_should_only_treat_numeric_values_as_comment_ids
 		$this->assertNull( get_comment( $comment_id . 'abc' ), 'Expected a malformed numeric string not to be cast to a comment ID.' );
 		$this->assertNull( get_comment( true ), 'Expected true not to be cast to comment ID 1.' ); // @phpstan-ignore argument.type (Intentionally passing an invalid value.)
 	}
+
+	/**
+	 * `wp_insert_post()` already encodes emoji as HTML entities before saving to a
+	 * utf8/utf8mb3 column (see #31242). This forces the same code path for comments
+	 * via the `pre_get_col_charset` filter, so it runs deterministically regardless
+	 * of the test database's actual charset.
+	 *
+	 * @ticket 65700
+	 *
+	 * @covers ::wp_insert_comment
+	 * @covers ::_wp_encode_emoji_in_fields
+	 */
+	public function test_wp_insert_comment_encodes_emoji_for_utf8mb3_columns(): void {
+		add_filter( 'pre_get_col_charset', array( $this, 'filter_pre_get_col_charset_utf8mb3_for_comment_fields' ), 10, 3 );
+
+		$comment_id = wp_insert_comment(
+			array(
+				'comment_post_ID' => self::$post_id,
+				'comment_author'  => "foo\xf0\x9f\x98\x88bar",
+				'comment_content' => "foo\xf0\x9f\x98\x8ebaz",
+			)
+		);
+
+		remove_filter( 'pre_get_col_charset', array( $this, 'filter_pre_get_col_charset_utf8mb3_for_comment_fields' ), 10 );
+
+		$this->assertIsInt( $comment_id, 'Expected wp_insert_comment() to return the new comment ID.' );
+
+		$comment = get_comment( $comment_id );
+
+		$this->assertSame( 'foo😈bar', $comment->comment_author );
+		$this->assertSame( 'foo😎baz', $comment->comment_content );
+	}
+
+	/**
+	 * Control for test_wp_insert_comment_encodes_emoji_for_utf8mb3_columns(): on a
+	 * utf8mb4 column, emoji should be stored as-is rather than as HTML entities.
+	 *
+	 * @ticket 65700
+	 *
+	 * @covers ::wp_insert_comment
+	 */
+	public function test_wp_insert_comment_does_not_encode_emoji_for_utf8mb4_columns(): void {
+		add_filter( 'pre_get_col_charset', array( $this, 'filter_pre_get_col_charset_utf8mb4_for_comment_fields' ), 10, 3 );
+
+		$comment_id = wp_insert_comment(
+			array(
+				'comment_post_ID' => self::$post_id,
+				'comment_author'  => "foo\xf0\x9f\x98\x88bar",
+				'comment_content' => "foo\xf0\x9f\x98\x8ebaz",
+			)
+		);
+
+		remove_filter( 'pre_get_col_charset', array( $this, 'filter_pre_get_col_charset_utf8mb4_for_comment_fields' ), 10 );
+
+		$comment = get_comment( $comment_id );
+
+		$this->assertSame( "foo\xf0\x9f\x98\x88bar", $comment->comment_author );
+		$this->assertSame( "foo\xf0\x9f\x98\x8ebaz", $comment->comment_content );
+	}
+
+	/**
+	 * @ticket 65700
+	 *
+	 * @covers ::wp_update_comment
+	 * @covers ::_wp_encode_emoji_in_fields
+	 */
+	public function test_wp_update_comment_encodes_emoji_for_utf8mb3_columns(): void {
+		$comment_id = self::factory()->comment->create( array( 'comment_post_ID' => self::$post_id ) );
+
+		add_filter( 'pre_get_col_charset', array( $this, 'filter_pre_get_col_charset_utf8mb3_for_comment_fields' ), 10, 3 );
+
+		$result = wp_update_comment(
+			array(
+				'comment_ID'      => $comment_id,
+				'comment_author'  => "foo\xf0\x9f\x98\x88bar",
+				'comment_content' => "foo\xf0\x9f\x98\x8ebaz",
+			)
+		);
+
+		remove_filter( 'pre_get_col_charset', array( $this, 'filter_pre_get_col_charset_utf8mb3_for_comment_fields' ), 10 );
+
+		$this->assertSame( 1, $result );
+
+		$comment = get_comment( $comment_id );
+
+		$this->assertSame( 'foo😈bar', $comment->comment_author );
+		$this->assertSame( 'foo😎baz', $comment->comment_content );
+	}
+
+	public function filter_pre_get_col_charset_utf8mb3_for_comment_fields( $charset, $table, $column ) {
+		global $wpdb;
+
+		if ( $wpdb->comments === $table && in_array( $column, array( 'comment_author', 'comment_content' ), true ) ) {
+			return 'utf8mb3';
+		}
+
+		return $charset;
+	}
+
+	public function filter_pre_get_col_charset_utf8mb4_for_comment_fields( $charset, $table, $column ) {
+		global $wpdb;
+
+		if ( $wpdb->comments === $table && in_array( $column, array( 'comment_author', 'comment_content' ), true ) ) {
+			return 'utf8mb4';
+		}
+
+		return $charset;
+	}
 }

tests/phpunit/tests/formatting/emoji.php

@@ -376,4 +376,95 @@ public function data_wp_staticize_emoji() {
 	public function test_wp_staticize_emoji( $emoji, $expected ) {
 		$this->assertSame( $expected, wp_staticize_emoji( $emoji ) );
 	}
+
+	/**
+	 * @ticket 65700
+	 *
+	 * @covers ::_wp_encode_emoji_in_fields
+	 */
+	public function test_wp_encode_emoji_in_fields_only_touches_requested_utf8mb3_fields(): void {
+		$data = array(
+			'legacy_field'  => '🙂',
+			'modern_field'  => '🙂',
+			'ignored_field' => '🙂',
+		);
+
+		add_filter( 'pre_get_col_charset', array( $this, 'filter_pre_get_col_charset_for_encode_emoji_in_fields_test' ), 10, 3 );
+
+		$result = _wp_encode_emoji_in_fields( $data, 'some_table', array( 'legacy_field', 'modern_field' ) );
+
+		remove_filter( 'pre_get_col_charset', array( $this, 'filter_pre_get_col_charset_for_encode_emoji_in_fields_test' ), 10 );
+
+		$this->assertSame(
+			'🙂',
+			$result['legacy_field'],
+			'Expected emoji to be encoded for a field on a utf8/utf8mb3 column.'
+		);
+		$this->assertSame(
+			'🙂',
+			$result['modern_field'],
+			'Expected emoji to be left alone for a field on a utf8mb4 column.'
+		);
+		$this->assertSame(
+			'🙂',
+			$result['ignored_field'],
+			'Expected a field not passed in $fields to be left alone entirely.'
+		);
+	}
+
+	/**
+	 * @ticket 65700
+	 *
+	 * @covers ::_wp_encode_emoji_in_fields
+	 */
+	public function test_wp_encode_emoji_in_fields_recognizes_the_utf8_alias(): void {
+		$data = array( 'legacy_field' => '🙂' );
+
+		add_filter( 'pre_get_col_charset', array( $this, 'filter_pre_get_col_charset_utf8_alias_for_encode_emoji_in_fields_test' ) );
+
+		$result = _wp_encode_emoji_in_fields( $data, 'some_table', array( 'legacy_field' ) );
+
+		remove_filter( 'pre_get_col_charset', array( $this, 'filter_pre_get_col_charset_utf8_alias_for_encode_emoji_in_fields_test' ) );
+
+		$this->assertSame(
+			'🙂',
+			$result['legacy_field'],
+			'Expected emoji to be encoded for a field reporting the deprecated utf8 charset alias.'
+		);
+	}
+
+	/**
+	 * @ticket 65700
+	 *
+	 * @covers ::_wp_encode_emoji_in_fields
+	 */
+	public function test_wp_encode_emoji_in_fields_ignores_fields_missing_from_data(): void {
+		$data = array( 'present_field' => '🙂' );
+
+		// Short-circuits wpdb::get_col_charset() so this doesn't hit a real,
+		// nonexistent 'some_table' table in the test database.
+		add_filter( 'pre_get_col_charset', array( $this, 'filter_pre_get_col_charset_utf8mb4_for_encode_emoji_in_fields_test' ) );
+
+		$result = _wp_encode_emoji_in_fields( $data, 'some_table', array( 'present_field', 'absent_field' ) );
+
+		remove_filter( 'pre_get_col_charset', array( $this, 'filter_pre_get_col_charset_utf8mb4_for_encode_emoji_in_fields_test' ) );
+
+		$this->assertArrayNotHasKey(
+			'absent_field',
+			$result,
+			'Expected a field missing from $data to not be added by _wp_encode_emoji_in_fields().'
+		);
+	}
+
+	public function filter_pre_get_col_charset_for_encode_emoji_in_fields_test( $charset, $table, $column ) {
+		return 'legacy_field' === $column ? 'utf8mb3' : 'utf8mb4';
+	}
+
+	public function filter_pre_get_col_charset_utf8_alias_for_encode_emoji_in_fields_test() {
+		return 'utf8';
+	}
+
+	public function filter_pre_get_col_charset_utf8mb4_for_encode_emoji_in_fields_test() {
+		return 'utf8mb4';
+	}
 }

tests/phpunit/tests/post.php

@@ -395,6 +395,46 @@ public function test_utf8mb3_post_saves_with_emoji() {
 		}
 	}
 
+	/**
+	 * Forces the utf8mb3 code path via the `pre_get_col_charset` filter, rather
+	 * than relying on the test database actually using that character set, so
+	 * this runs deterministically regardless of the environment's DB charset.
+	 *
+	 * @ticket 65700
+	 *
+	 * @covers ::wp_insert_post
+	 * @covers ::_wp_encode_emoji_in_fields
+	 */
+	public function test_wp_insert_post_encodes_emoji_for_utf8mb3_columns_via_filter(): void {
+		add_filter( 'pre_get_col_charset', array( $this, 'filter_pre_get_col_charset_utf8mb3_for_post_fields' ), 10, 3 );
+
+		$post_id = self::factory()->post->create(
+			array(
+				'post_title'   => "foo\xf0\x9f\x98\x88bar",
+				'post_content' => "foo\xf0\x9f\x98\x8ebaz",
+				'post_excerpt' => "foo\xf0\x9f\x98\x90bat",
+			)
+		);
+
+		remove_filter( 'pre_get_col_charset', array( $this, 'filter_pre_get_col_charset_utf8mb3_for_post_fields' ), 10 );
+
+		$post = get_post( $post_id );
+
+		$this->assertSame( 'foo😈bar', $post->post_title );
+		$this->assertSame( 'foo😎baz', $post->post_content );
+		$this->assertSame( 'foo😐bat', $post->post_excerpt );
+	}
+
+	public function filter_pre_get_col_charset_utf8mb3_for_post_fields( $charset, $table, $column ) {
+		global $wpdb;
+
+		if ( $wpdb->posts === $table && in_array( $column, array( 'post_title', 'post_content', 'post_excerpt' ), true ) ) {
+			return 'utf8mb3';
+		}
+
+		return $charset;
+	}
+
 	/**
 	 * If a sticky post is updated via `wp_update_post()` by a user
 	 * without the `publish_posts` capability, it should stay sticky.