WPDB: If a site is using the utf8 charset, and their version of MySQL supports utf8mb4, auto-upgrade them to utf8mb4.

This patch also resizes some indexes, to allow for the 767 byte index size limit in standard MySQL installs.

See #21212



git-svn-id: https://develop.svn.wordpress.org/trunk@31349 602fd350-edb4-49c9-b593-d223f7449a82

src/wp-admin/includes/schema.php

@@ -44,6 +44,13 @@ function wp_get_db_schema( $scope = 'all', $blog_id = null ) {
 	// Engage multisite if in the middle of turning it on from network.php.
 	$is_multisite = is_multisite() || ( defined( 'WP_INSTALLING_NETWORK' ) && WP_INSTALLING_NETWORK );
 
+	/*
+	 * Indexes have a maximum size of 767 bytes. Historically, we haven't need to be concerned about that.
+	 * As of 4.2, however, we moved to utf8mb4, which uses 4 bytes per character. This means that an index which
+	 * used to have room for floor(767/3) = 255 characters, now only has room for floor(767/4) = 191 characters.
+	 */
+	$max_index_length = 191;
+
 	// Blog specific tables.
 	$blog_tables = "CREATE TABLE $wpdb->terms (
  term_id bigint(20) unsigned NOT NULL auto_increment,
@@ -51,8 +58,8 @@ function wp_get_db_schema( $scope = 'all', $blog_id = null ) {
  slug varchar(200) NOT NULL default '',
  term_group bigint(10) NOT NULL default 0,
  PRIMARY KEY  (term_id),
- KEY slug (slug),
+ KEY slug (slug($max_index_length)),
- KEY name (name)
+ KEY name (name($max_index_length))
 ) $charset_collate;
 CREATE TABLE $wpdb->term_taxonomy (
  term_taxonomy_id bigint(20) unsigned NOT NULL auto_increment,
@@ -79,7 +86,7 @@ CREATE TABLE $wpdb->commentmeta (
   meta_value longtext,
   PRIMARY KEY  (meta_id),
   KEY comment_id (comment_id),
-  KEY meta_key (meta_key)
+  KEY meta_key (meta_key($max_index_length))
 ) $charset_collate;
 CREATE TABLE $wpdb->comments (
   comment_ID bigint(20) unsigned NOT NULL auto_increment,
@@ -136,7 +143,7 @@ CREATE TABLE $wpdb->postmeta (
   meta_value longtext,
   PRIMARY KEY  (meta_id),
   KEY post_id (post_id),
-  KEY meta_key (meta_key)
+  KEY meta_key (meta_key($max_index_length))
 ) $charset_collate;
 CREATE TABLE $wpdb->posts (
   ID bigint(20) unsigned NOT NULL auto_increment,
@@ -163,7 +170,7 @@ CREATE TABLE $wpdb->posts (
   post_mime_type varchar(100) NOT NULL default '',
   comment_count bigint(20) NOT NULL default '0',
   PRIMARY KEY  (ID),
-  KEY post_name (post_name),
+  KEY post_name (post_name($max_index_length)),
   KEY type_status_date (post_type,post_status,post_date,ID),
   KEY post_parent (post_parent),
   KEY post_author (post_author)
@@ -213,7 +220,7 @@ CREATE TABLE $wpdb->posts (
   meta_value longtext,
   PRIMARY KEY  (umeta_id),
   KEY user_id (user_id),
-  KEY meta_key (meta_key)
+  KEY meta_key (meta_key($max_index_length))
 ) $charset_collate;\n";
 
 	// Global tables
@@ -261,7 +268,7 @@ CREATE TABLE $wpdb->site (
   domain varchar(200) NOT NULL default '',
   path varchar(100) NOT NULL default '',
   PRIMARY KEY  (id),
-  KEY domain (domain,path)
+  KEY domain (domain(140),path(51))
 ) $charset_collate;
 CREATE TABLE $wpdb->sitemeta (
   meta_id bigint(20) NOT NULL auto_increment,
@@ -269,7 +276,7 @@ CREATE TABLE $wpdb->sitemeta (
   meta_key varchar(255) default NULL,
   meta_value longtext,
   PRIMARY KEY  (meta_id),
-  KEY meta_key (meta_key),
+  KEY meta_key (meta_key($max_index_length)),
   KEY site_id (site_id)
 ) $charset_collate;
 CREATE TABLE $wpdb->signups (
@@ -288,7 +295,7 @@ CREATE TABLE $wpdb->signups (
   KEY activation_key (activation_key),
   KEY user_email (user_email),
   KEY user_login_email (user_login,user_email),
-  KEY domain_path (domain,path)
+  KEY domain_path (domain(140),path(51))
 ) $charset_collate;";
 
 	switch ( $scope ) {

src/wp-admin/includes/upgrade.php

@@ -519,6 +519,9 @@ function upgrade_all() {
 	if ( $wp_current_db_version < 29630 )
 		upgrade_400();
 
+	if ( $wp_current_db_version < 31349 )
+		upgrade_420();
+
 	maybe_disable_link_manager();
 
 	maybe_disable_automattic_widgets();
@@ -1406,6 +1409,27 @@ function upgrade_400() {
 	}
 }
 
+/**
+ * Execute changes made in WordPress 4.2.0.
+ *
+ * @since 4.2.0
+ */
+function upgrade_420() {
+	global $wp_current_db_version, $wpdb;
+
+	if ( $wp_current_db_version < 31349 && $wpdb->charset === 'utf8mb4' ) {
+		if ( is_multisite() ) {
+			$tables = $wpdb->tables( 'blog' );
+		} else {
+			$tables = $wpdb->tables( 'all' );
+		}
+
+		foreach ( $tables as $table ) {
+			maybe_convert_table_to_utf8mb4( $table );
+		}
+	}
+}
+
 /**
  * Executes network-level upgrade routines.
  *
@@ -1502,6 +1526,21 @@ function upgrade_network() {
 			update_site_option( 'illegal_names', $illegal_names );
 		}
 	}
+
+	// 4.2
+	if ( $wp_current_db_version < 31349 && $wpdb->charset === 'utf8mb4' ) {
+		if ( ! ( defined( 'DO_NOT_UPGRADE_GLOBAL_TABLES' ) && DO_NOT_UPGRADE_GLOBAL_TABLES ) ) {
+			$wpdb->query( "ALTER TABLE $wpdb->site DROP INDEX domain, ADD INDEX domain(domain(140),path(51))" );
+			$wpdb->query( "ALTER TABLE $wpdb->sitemeta DROP INDEX meta_key, ADD INDEX meta_key(meta_key(191))" );
+			$wpdb->query( "ALTER TABLE $wpdb->signups DROP INDEX domain, ADD INDEX domain(domain(140),path(51))" );
+
+			$tables = $wpdb->tables( 'global' );
+
+			foreach ( $tables as $table ) {
+				maybe_convert_table_to_utf8mb4( $table );
+			}
+		}
+	}
 }
 
 //
@@ -1607,6 +1646,42 @@ function maybe_add_column($table_name, $column_name, $create_ddl) {
 	return false;
 }
 
+/**
+ * If a table only contains utf8 or utf8mb4 columns, convert it to utf8mb4.
+ *
+ * @since 4.2.0
+ *
+ * @param string $table The table to convert.
+ * @return bool true if the table was converted, false if it wasn't.
+ */
+function maybe_convert_table_to_utf8mb4( $table ) {
+	global $wpdb;
+
+	$results = $wpdb->get_results( "SHOW FULL COLUMNS FROM `$table`" );
+	if ( ! $results ) {
+		return false;
+	}
+
+	$has_utf8 = false;
+	foreach ( $results as $column ) {
+		if ( $column->Collation ) {
+			if ( 'utf8' === $column->Collation ) {
+				$has_utf8 = true;
+			} elseif ( 'utf8mb4' !== $column->Collation ) {
+				// Don't upgrade tables that have non-utf8 columns.
+				return false;
+			}
+		}
+	}
+
+	if ( ! $has_utf8 ) {
+		// Don't bother upgrading tables that don't have utf8 columns.
+		return false;
+	}
+
+	return $wpdb->query( "ALTER TABLE $table CONVERT TO CHARACTER SET utf8mb4 COLLATE utf8mb4_unicode_ci" );
+}
+
 /**
  * Retrieve all options as it was for 1.2.
  *
@@ -2284,6 +2359,17 @@ function pre_schema_upgrade() {
 		// dbDelta() can recreate but can't drop the index.
 		$wpdb->query( "ALTER TABLE $wpdb->terms DROP INDEX slug" );
 	}
+
+	// Upgrade versions prior to 4.2.
+	if ( $wp_current_db_version < 31349 ) {
+		// So that we can change tables to utf8mb4, we need to shorten the index lengths to less than 767 bytes
+		$wpdb->query( "ALTER TABLE $wpdb->usermeta DROP INDEX meta_key, ADD INDEX meta_key(meta_key(191))" );
+		$wpdb->query( "ALTER TABLE $wpdb->terms DROP INDEX slug, ADD INDEX slug(slug(191))" );
+		$wpdb->query( "ALTER TABLE $wpdb->terms DROP INDEX name, ADD INDEX name(name(191))" );
+		$wpdb->query( "ALTER TABLE $wpdb->commentmeta DROP INDEX meta_key, ADD INDEX meta_key(meta_key(191))" );
+		$wpdb->query( "ALTER TABLE $wpdb->postmeta DROP INDEX meta_key, ADD INDEX meta_key(meta_key(191))" );
+		$wpdb->query( "ALTER TABLE $wpdb->posts DROP INDEX post_name, ADD INDEX post_name(post_name(191))" );
+	}
 }
 
 /**

src/wp-admin/setup-config.php

@@ -280,6 +280,11 @@ switch($step) {
 			case 'DB_HOST'     :
 				$config_file[ $line_num ] = "define('" . $constant . "'," . $padding . "'" . addcslashes( constant( $constant ), "\\'" ) . "');\r\n";
 				break;
+			case 'DB_CHARSET'  :
+				if ( 'utf8mb4' === $wpdb->charset || ( ! $wpdb->charset && $wpdb->has_cap( 'utf8mb4' ) ) ) {
+					$config_file[ $line_num ] = "define('" . $constant . "'," . $padding . "'utf8mb4');\r\n";
+				}
+				break;
 			case 'AUTH_KEY'         :
 			case 'SECURE_AUTH_KEY'  :
 			case 'LOGGED_IN_KEY'    :

src/wp-includes/version.php

@@ -11,7 +11,7 @@ $wp_version = '4.2-alpha-31007-src';
  *
  * @global int $wp_db_version
  */
-$wp_db_version = 30133;
+$wp_db_version = 31349;
 
 /**
  * Holds the TinyMCE version

src/wp-includes/wp-db.php

@@ -624,8 +624,6 @@ class wpdb {
 			}
 		}
 
-		$this->init_charset();
-
 		$this->dbuser = $dbuser;
 		$this->dbpassword = $dbpassword;
 		$this->dbname = $dbname;
@@ -717,16 +715,31 @@ class wpdb {
 	public function init_charset() {
 		if ( function_exists('is_multisite') && is_multisite() ) {
 			$this->charset = 'utf8';
-			if ( defined( 'DB_COLLATE' ) && DB_COLLATE )
+			if ( defined( 'DB_COLLATE' ) && DB_COLLATE ) {
 				$this->collate = DB_COLLATE;
-			else
+			} else {
 				$this->collate = 'utf8_general_ci';
+			}
 		} elseif ( defined( 'DB_COLLATE' ) ) {
 			$this->collate = DB_COLLATE;
 		}
 
-		if ( defined( 'DB_CHARSET' ) )
+		if ( defined( 'DB_CHARSET' ) ) {
 			$this->charset = DB_CHARSET;
+		}
+
+		if ( ( $this->use_mysqli && ! ( $this->dbh instanceof mysqli ) )
+		  || ( empty( $this->dbh ) || ! ( $this->dbh instanceof mysqli ) ) ) {
+			return;
+		}
+
+		if ( 'utf8' === $this->charset && $this->has_cap( 'utf8mb4' ) ) {
+			$this->charset = 'utf8mb4';
+		}
+
+		if ( 'utf8mb4' === $this->charset && ( ! $this->collate || stripos( $this->collate, 'utf8_' ) === 0 ) ) {
+			$this->collate = 'utf8mb4_unicode_ci';
+		}
 	}
 
 	/**
@@ -1476,8 +1489,14 @@ class wpdb {
 
 			return false;
 		} elseif ( $this->dbh ) {
+			if ( ! $this->has_connected ) {
+				$this->init_charset();
+			}
+
 			$this->has_connected = true;
+
 			$this->set_charset( $this->dbh );
+
 			$this->ready = true;
 			$this->set_sql_mode();
 			$this->select( $this->dbname, $this->dbh );
@@ -2249,14 +2268,14 @@ class wpdb {
 	 * Retrieves the character set for the given column.
 	 *
 	 * @since 4.2.0
-	 * @access protected
+	 * @access public
 	 *
 	 * @param string $table  Table name.
 	 * @param string $column Column name.
 	 * @return mixed Column character set as a string. False if the column has no
 	 *               character set. {@see WP_Error} object if there was an error.
 	 */
-	protected function get_col_charset( $table, $column ) {
+	public function get_col_charset( $table, $column ) {
 		$tablekey = strtolower( $table );
 		$columnkey = strtolower( $column );
 
@@ -2356,7 +2375,6 @@ class wpdb {
 			'gb2312'  => 'EUC-CN',
 			'ujis'    => 'EUC-JP',
 			'utf32'   => 'UTF-32',
-			'utf8mb4' => 'UTF-8',
 		);
 
 		$supported_charsets = array();
@@ -2391,8 +2409,8 @@ class wpdb {
 				}
 			}
 
-			// utf8(mb3) can be handled by regex, which is a bunch faster than a DB lookup.
+			// utf8 can be handled by regex, which is a bunch faster than a DB lookup.
-			if ( 'utf8' === $charset || 'utf8mb3' === $charset ) {
+			if ( 'utf8' === $charset || 'utf8mb3' === $charset || 'utf8mb4' === $charset ) {
 				$regex = '/
 					(
 						(?: [\x00-\x7F]                  # single-byte sequences   0xxxxxxx
@@ -2400,8 +2418,17 @@ class wpdb {
 						|   \xE0[\xA0-\xBF][\x80-\xBF]   # triple-byte sequences   1110xxxx 10xxxxxx * 2
 						|   [\xE1-\xEC][\x80-\xBF]{2}
 						|   \xED[\x80-\x9F][\x80-\xBF]
-						|   [\xEE-\xEF][\x80-\xBF]{2}
+						|   [\xEE-\xEF][\x80-\xBF]{2}';
-						){1,50}                          # ...one or more times
+
+				if ( 'utf8mb4' === $charset) {
+					$regex .= '
+						|    \xF0[\x90-\xBF][\x80-\xBF]{2} # four-byte sequences   11110xxx 10xxxxxx * 3
+						|    [\xF1-\xF3][\x80-\xBF]{3}
+						|    \xF4[\x80-\x8F][\x80-\xBF]{2}
+					';
+				}
+
+				$regex .= '){1,50}                          # ...one or more times
 					)
 					| .                                  # anything else
 					/x';

tests/phpunit/tests/db/charset.php

@@ -130,11 +130,12 @@ class Tests_DB_Charset extends WP_UnitTestCase {
 	}
 
 	/**
-	 * @ ticket 21212
+	 * @ticket 21212
 	 */
 	function test_process_fields_failure() {
 		global $wpdb;
-		$data = array( 'post_content' => "H€llo\xf0\x9f\x98\x88World¢" );
+		// \xf0\xff\xff\xff is invalid in utf8 and utf8mb4.
+		$data = array( 'post_content' => "H€llo\xf0\xff\xff\xffWorld¢" );
 		$this->assertFalse( self::$_wpdb->process_fields( $wpdb->posts, $data, null ) );
 	}
 
@@ -436,6 +437,6 @@ class Tests_DB_Charset extends WP_UnitTestCase {
 	 */
 	function test_invalid_characters_in_query() {
 		global $wpdb;
-		$this->assertFalse( $wpdb->query( "INSERT INTO {$wpdb->posts} (post_content) VALUES ('foo\xf0\x9f\x98\x88bar')" ) );
+		$this->assertFalse( $wpdb->query( "INSERT INTO {$wpdb->posts} (post_content) VALUES ('foo\xf0\xff\xff\xffbar')" ) );
 	}
 }