Introduce a binary-safe wrapper for strlen() and use it in seems_utf8(), utf8_uri_encode(), and wp_read_image_metadata().

Use binary-safe POMO_Reader::strlen() in MO::export_to_file_handle().

fixes #28162.
Built from https://develop.svn.wordpress.org/trunk@28806


git-svn-id: http://core.svn.wordpress.org/trunk@28615 1a063a9b-81f0-0310-95a4-ce76da25c4cd

wp-admin/includes/image.php

@@ -290,7 +290,7 @@ function wp_read_image_metadata( $file ) {
 				$caption = trim( $iptc['2#120'][0] );
 				if ( empty( $meta['title'] ) ) {
 					// Assume the title is stored in 2:120 if it's short.
-					if ( strlen( $caption ) < 80 )
+					if ( mbstring_binary_safe_strlen( $caption ) < 80 )
 						$meta['title'] = $caption;
 					else
 						$meta['caption'] = $caption;
@@ -327,7 +327,7 @@ function wp_read_image_metadata( $file ) {
 		}
 
 		if ( ! empty( $exif['ImageDescription'] ) ) {
-			if ( empty( $meta['title'] ) && strlen( $exif['ImageDescription'] ) < 80 ) {
+			if ( empty( $meta['title'] ) && mbstring_binary_safe_strlen( $exif['ImageDescription'] ) < 80 ) {
 				// Assume the title is stored in ImageDescription
 				$meta['title'] = trim( $exif['ImageDescription'] );
 				if ( empty( $meta['caption'] ) && ! empty( $exif['COMPUTED']['UserComment'] ) && trim( $exif['COMPUTED']['UserComment'] ) != $meta['title'] ) {

wp-includes/formatting.php

@@ -481,7 +481,7 @@ function shortcode_unautop( $pee ) {
  * @return bool True if $str fits a UTF-8 model, false otherwise.
  */
 function seems_utf8($str) {
-	$length = strlen($str);
+	$length = mbstring_binary_safe_strlen($str);
 	for ($i=0; $i < $length; $i++) {
 		$c = ord($str[$i]);
 		if ($c < 0x80) $n = 0; # 0bbbbbbb
@@ -705,7 +705,7 @@ function utf8_uri_encode( $utf8_string, $length = 0 ) {
 	$num_octets = 1;
 	$unicode_length = 0;
 
-	$string_length = strlen( $utf8_string );
+	$string_length = mbstring_binary_safe_strlen( $utf8_string );
 	for ($i = 0; $i < $string_length; $i++ ) {
 
 		$value = ord( $utf8_string[ $i ] );

wp-includes/functions.php

@@ -4404,6 +4404,24 @@ function reset_mbstring_encoding() {
 	mbstring_binary_safe_encoding( true );
 }
 
+/**
+ * Uses a binary-safe encoding to get the length of a string in bytes if func_overload is enabled.
+ *
+ * @see mbstring_binary_safe_encoding()
+ *
+ * @since 4.0.0
+ *
+ * @param string $string The string to get the length of.
+ * @return int The length of the string in bytes.
+ */
+function mbstring_binary_safe_strlen( $string ) {
+	mbstring_binary_safe_encoding();
+	$length = strlen( $string );
+	reset_mbstring_encoding();
+
+	return $length;
+}
+
 /**
  * Alternative to filter_var( $var, FILTER_VALIDATE_BOOLEAN )
  *

wp-includes/pomo/mo.php

@@ -75,21 +75,23 @@ class MO extends Gettext_Translations {
 		$current_addr++;
 		$originals_table = chr(0);
 
+		$reader = new POMO_Reader();
+
 		foreach($entries as $entry) {
 			$originals_table .= $this->export_original($entry) . chr(0);
-			$length = strlen($this->export_original($entry));
+			$length = $reader->strlen($this->export_original($entry));
 			fwrite($fh, pack('VV', $length, $current_addr));
 			$current_addr += $length + 1; // account for the NULL byte after
 		}
 
 		$exported_headers = $this->export_headers();
-		fwrite($fh, pack('VV', strlen($exported_headers), $current_addr));
+		fwrite($fh, pack('VV', $reader->strlen($exported_headers), $current_addr));
 		$current_addr += strlen($exported_headers) + 1;
 		$translations_table = $exported_headers . chr(0);
 
 		foreach($entries as $entry) {
 			$translations_table .= $this->export_translations($entry) . chr(0);
-			$length = strlen($this->export_translations($entry));
+			$length = $reader->strlen($this->export_translations($entry));
 			fwrite($fh, pack('VV', $length, $current_addr));
 			$current_addr += $length + 1;
 		}