Setting Character Sets and Collations

You are viewing an old version of this article. View the current version here.

In MariaDB, the default character set is latin1, and the default collation is latin1_swedish_ci (however this may differ in some distros, see for example Differences in MariaDB in Debian). Both character sets and collations can be specified from the server right down to the column level, as well as for client-server connections. When changing a character set and not specifying a collation, the default collation for the new character set is always used.

Character sets and collations always cascade down, so a column without a specified collation will look for the table default, the table for the database, and the database for the server. It's therefore possible to have extremely fine-grained control over all the character sets and collations used in your data.

Default collations for each character set can be viewed with the SHOW COLLATION statement, for example, to find the default collation for the latin2 character set:

SHOW COLLATION LIKE 'latin2%';
+---------------------+---------+----+---------+----------+---------+
| Collation           | Charset | Id | Default | Compiled | Sortlen |
+---------------------+---------+----+---------+----------+---------+
| latin2_czech_cs     | latin2  |  2 |         | Yes      |       4 |
| latin2_general_ci   | latin2  |  9 | Yes     | Yes      |       1 |
| latin2_hungarian_ci | latin2  | 21 |         | Yes      |       1 |
| latin2_croatian_ci  | latin2  | 27 |         | Yes      |       1 |
| latin2_bin          | latin2  | 77 |         | Yes      |       1 |
+---------------------+---------+----+---------+----------+---------+

Server Level

The character_set_server system variable can be used to change the default server character set. It can be set both on startup or dynamically, with the SET command:

SET character_set_server = 'latin2';

Similarly, the collation_server variable is used for setting the default server collation.

SET collation_server = 'latin2_czech_cs';

Database Level

The CREATE DATABASE and ALTER DATABASE statements have optional character set and collation clauses. If these are left out, the server defaults are used.

CREATE DATABASE czech_slovak_names 
  CHARACTER SET = 'keybcs2'
  COLLATE = 'keybcs2_bin';
ALTER DATABASE czech_slovak_names COLLATE = 'keybcs2_general_ci';

To determine the default character set used by a database, use:

SHOW CREATE DATABASE czech_slovak_names;
+--------------------+--------------------------------------------------------------------------------+
| Database           | Create Database                                                                |
+--------------------+--------------------------------------------------------------------------------+
| czech_slovak_names | CREATE DATABASE `czech_slovak_names` /*!40100 DEFAULT CHARACTER SET keybcs2 */ |
+--------------------+--------------------------------------------------------------------------------+

or alternatively, for the character set and collation:

SELECT * FROM INFORMATION_SCHEMA.SCHEMATA;
+--------------+--------------------+----------------------------+------------------------+----------+
| CATALOG_NAME | SCHEMA_NAME        | DEFAULT_CHARACTER_SET_NAME | DEFAULT_COLLATION_NAME | SQL_PATH |
+--------------+--------------------+----------------------------+------------------------+----------+
| def          | czech_slovak_names | keybcs2                    | keybcs2_general_ci     | NULL     |
| def          | information_schema | utf8                       | utf8_general_ci        | NULL     |
| def          | mysql              | latin1                     | latin1_swedish_ci      | NULL     |
| def          | performance_schema | utf8                       | utf8_general_ci        | NULL     |
| def          | test               | latin1                     | latin1_swedish_ci      | NULL     |
+--------------+--------------------+----------------------------+------------------------+----------+

It is also possible to specify only the collation, and, since each collation only applies to one character set, the associated character set will automatically be specified.

CREATE DATABASE danish_names COLLATE 'utf8_danish_ci';

SHOW CREATE DATABASE danish_names;
+--------------+----------------------------------------------------------------------------------------------+
| Database     | Create Database                                                                              |
+--------------+----------------------------------------------------------------------------------------------+
| danish_names | CREATE DATABASE `danish_names` /*!40100 DEFAULT CHARACTER SET utf8 COLLATE utf8_danish_ci */ |
+--------------+----------------------------------------------------------------------------------------------+

Although there are character_set_database and collation_database system variables which can be set dynamically, these are used for determining the character set and collation for the default database, and should only be set by the server.

Table Level

The CREATE TABLE and ALTER TABLE statements support optional character set and collation clauses, a MariaDB and MySQL extension to standard SQL.

CREATE TABLE english_names (id INT, name VARCHAR(40)) 
  CHARACTER SET 'utf8' 
  COLLATE 'utf8_icelandic_ci';

If neither character set nor collation is provided, the database default will be used. If only the character set is provided, the default collation for that character set will be used . If only the collation is provided, the associated character set will be used. See Supported Character Sets and Collations.

ALTER TABLE table_name
 CONVERT TO CHARACTER SET charset_name [COLLATE collation_name];

If no collation is provided, the collation will be set to the default collation for that character set. See Supported Character Sets and Collations.

For VARCHAR or TEXT columns, CONVERT TO CHARACTER SET changes the data type if needed to ensure the new column is long enough to store as many characters as the original column.

For example, an ascii TEXT column requires a single byte per character, so the column can hold up to 65,535 characters. If the column is converted to utf8, 3 bytes can be required for each character, so the column will be converted to MEDIUMTEXT to be able to hold the same number of characters.

CONVERT TO CHARACTER SET binary will convert CHAR, VARCHAR and TEXT columns to BINARY, VARBINARY and BLOB respectively, and from that point will no longer have a character set, or be affected by future CONVERT TO CHARACTER SET statements.

To avoid data type changes resulting from CONVERT TO CHARACTER SET, use MODIFY on the individual columns instead. For example:

ALTER TABLE table_name MODIFY ascii_text_column TEXT CHARACTER SET utf8;
ALTER TABLE table_name MODIFY ascii_varchar_column VARCHAR(M) CHARACTER SET utf8;

Column Level

Character sets and collations can also be specified for columns that are character types CHAR, TEXT or VARCHAR. The CREATE TABLE and ALTER TABLE statements support optional character set and collation clauses for this purpose - unlike those at the table level, the column level definitions are standard SQL.

CREATE TABLE european_names (
  croatian_names VARCHAR(40) COLLATE 'cp1250_croatian_ci',
  greek_names VARCHAR(40) CHARACTER SET 'greek');

If neither collation nor character set is provided, the table default is used. If only the character set is specified, that character set's default collation is used, while if only the collation is specified, the associated character set is used.

When using ALTER TABLE to change a column's character set, you need to ensure the character sets are compatible with your data. MariaDB will map the data as best it can, but it's possible to lose data if care is not taken.

The SHOW CREATE TABLE statement or INFORMATION SCHEMA database can be used to determine column character sets and collations.

SHOW CREATE TABLE european_names\G
*************************** 1. row ***************************
       Table: european_names
Create Table: CREATE TABLE `european_names` (
  `croatian_names` varchar(40) CHARACTER SET cp1250 COLLATE cp1250_croatian_ci DEFAULT NULL,
  `greek_names` varchar(40) CHARACTER SET greek DEFAULT NULL
) ENGINE=InnoDB DEFAULT CHARSET=utf8 COLLATE=utf8_danish_ci
SELECT * FROM INFORMATION_SCHEMA.COLUMNS WHERE TABLE_NAME LIKE 'european%'\G
*************************** 1. row ***************************
           TABLE_CATALOG: def
            TABLE_SCHEMA: danish_names
              TABLE_NAME: european_names
             COLUMN_NAME: croatian_names
        ORDINAL_POSITION: 1
          COLUMN_DEFAULT: NULL
             IS_NULLABLE: YES
               DATA_TYPE: varchar
CHARACTER_MAXIMUM_LENGTH: 40
  CHARACTER_OCTET_LENGTH: 40
       NUMERIC_PRECISION: NULL
           NUMERIC_SCALE: NULL
      DATETIME_PRECISION: NULL
      CHARACTER_SET_NAME: cp1250
          COLLATION_NAME: cp1250_croatian_ci
             COLUMN_TYPE: varchar(40)
              COLUMN_KEY: 
                   EXTRA: 
              PRIVILEGES: select,insert,update,references
          COLUMN_COMMENT: 
*************************** 2. row ***************************
           TABLE_CATALOG: def
            TABLE_SCHEMA: danish_names
              TABLE_NAME: european_names
             COLUMN_NAME: greek_names
        ORDINAL_POSITION: 2
          COLUMN_DEFAULT: NULL
             IS_NULLABLE: YES
               DATA_TYPE: varchar
CHARACTER_MAXIMUM_LENGTH: 40
  CHARACTER_OCTET_LENGTH: 40
       NUMERIC_PRECISION: NULL
           NUMERIC_SCALE: NULL
      DATETIME_PRECISION: NULL
      CHARACTER_SET_NAME: greek
          COLLATION_NAME: greek_general_ci
             COLUMN_TYPE: varchar(40)
              COLUMN_KEY: 
                   EXTRA: 
              PRIVILEGES: select,insert,update,references
          COLUMN_COMMENT: 

Filenames

Since MariaDB 5.1, the character_set_filesystem system variable has controlled interpretation of file names that are given as literal strings. This affects the following statements and functions:

Literals

By default, the character set and collation used for literals is determined by the character_set_connection and collation_connection system variables. However, they can also be specified explicitly:

[_charset_name]'string' [COLLATE collation_name]

The character set of string literals that do not have a character set introducer is determined by the character_set_connection system variable.

This query:

  SELECT CHARSET('a'), @@character_set_connection;

always returns the same character set name in both columns.

character_set_client and character_set_connection are normally (e.g. during handshake, or after a SET NAMES query) are set to equal values. However, it's possible to set to different values.

Examples

Examples when setting @@character_set_client and @@character_set_connection to different values can be useful:

Example 1:

Suppose, we have a utf8 database with this table:

CREATE TABLE t1 (a VARCHAR(10)) CHARACTER SET utf8 COLLATE utf8_general_ci;
INSERT INTO t1 VALUES ('oe'),('ö');

Now we connect to it using "mysql.exe", which uses the DOS character set (cp850 on a West European machine), and want to fetch all records that are equal to 'ö' according to the German phonebook rules.

It's possible with the following:

SET @@character_set_client=cp850, @@character_set_connection=utf8;
SELECT a FROM t1 WHERE a='ö' COLLATE utf8_german2_ci;

This will return:

+------+
| a    |
+------+
| oe   |
| ö    |
+------+

It works as follows:

  1. The client sends the query using cp850.
  2. The server, when parsing the query, creates a utf8 string literal by converting 'ö' from @@character_set_client (cp850) to @@character_set_connection (utf8)
  3. The server applies the collation "utf8_german2_ci" to this string literal.
  4. The server uses utf8_german2_ci for comparison.

Note, if we rewrite the script like this:

SET NAMES cp850;
SELECT a FROM t1 WHERE a='ö' COLLATE utf8_german2_ci;

we'll get an error:

ERROR 1253 (42000): COLLATION 'utf8_german2_ci' is not valid for CHARACTER SET 'cp850'

because:

  • on step #2, the literal is not converted to utf8 any more and is created using cp850.
  • on step #3, the server fails to apply utf8_german2_ci to an cp850 string literal.

Example 2:

Suppose we have a utf8 database and use "mysql.exe" from a West European machine again.

We can do this:

SET @@character_set_client=cp850, @@character_set_connection=utf8;
CREATE TABLE t2 AS SELECT 'ö';

It will create a table with a column of the type VARCHAR(1) CHARACTER SET utf8.

Note, if we rewrite the query like this:

SET NAMES cp850;
CREATE TABLE t2 AS SELECT 'ö';

It will create a table with a column of the type VARCHAR(1) CHARACTER SET cp850, which is probably not a good idea.

N

Also, N or n can be used as prefix to convert a literal into the National Character set (which in MariaDB is always utf8).

For example:

SELECT _latin2 'Müller';
+-----------+
| MĂźller   |
+-----------+
| MĂźller   |
+-----------+
SELECT CHARSET(N'a string');
+----------------------+
| CHARSET(N'a string') |
+----------------------+
| utf8                 |
+----------------------+
SELECT 'Mueller' = 'Müller' COLLATE 'latin1_german2_ci';
+---------------------------------------------------+
| 'Mueller' = 'Müller' COLLATE 'latin1_german2_ci'  |
+---------------------------------------------------+
|                                                 1 |
+---------------------------------------------------+

Stored Programs and Views

The literals which occur in stored programs and views, by default, use the character set and collation which was specified by the character_set_connection and collation_connection system variables when the stored program was created. These values can be seen using the SHOW CREATE statements. To change the character sets used for literals in an existing stored program, it is necessary to drop and recreate the stored program.

For stored routines parameters and return values, a character set and a collation can be specified via the CHARACTER SET and COLLATE clauses. Before 5.5, specifying a collation was not supported.

The following example shows that the character set and collation are determined at the time of creation:

SET @@local.character_set_connection='latin1';

DELIMITER ||
CREATE PROCEDURE `test`.`x`()
BEGIN
	SELECT CHARSET('x');
END;
||
Query OK, 0 rows affected (0.00 sec)

DELIMITER ;
SET @@local.character_set_connection='utf8';

CALL `test`.`x`();
+--------------+
| CHARSET('x') |
+--------------+
| latin1       |
+--------------+

The following example shows how to specify a function parameters character set and collation:

CREATE FUNCTION `test`.`y`(`str` TEXT CHARACTER SET utf8 COLLATE utf8_bin)
	RETURNS TEXT CHARACTER SET latin1 COLLATE latin1_bin
BEGIN
	SET @param_coll = COLLATION(`str`);
	RETURN `str`;
END;

-- return value's collation:
SELECT COLLATION(`test`.`y`('Hello, planet!'));
+-----------------------------------------+
| COLLATION(`test`.`y`('Hello, planet!')) |
+-----------------------------------------+
| latin1_bin                              |
+-----------------------------------------+

-- parameter's collation:
SELECT @param_coll;
+-------------+
| @param_coll |
+-------------+
| utf8_bin    |
+-------------+

Illegal Collation Mix

MariaDB 10.1.28 - 10.1.29

In MariaDB 10.1.28, you may encounter Error 1267 when performing comparison operations in views on tables that use binary constants. For instance,

CREATE TABLE test.t1 (
   a TEXT CHARACTER SET gbk 
) ENGINE=InnoDB 
CHARSET=latin1
COLLATE=latin1_general_cs;

INSERT INTO t1 VALUES ('user_a');

CREATE VIEW v1 AS
SELECT a <> 0xEE5D FROM t1;

SELECT * FROM v1;
Error 1267 (HY000): Illegal mix of collations (gbk_chinese_ci, IMPLICIT)
and (latin_swedish_ci, COERCIBLE) for operation

When the view query is written to file, MariaDB converts the binary character into a string literal, which causes it to be misinterpreted when you execute the SELECT statement. If you encounter this issue, set the character set in the view to force it to the value you want.

MariaDB throws this error due to a bug that was fixed in MariaDB 10.1.29. Later releases do not throw errors in this situation.

Example: Changing the Default Character Set To UTF-8

To change the default character set from latin1 to UTF-8, the following settings should be specified in the my.cnf configuration file.

[client]
...
default-character-set=utf8mb4
...
[mysql]
...
default-character-set=utf8mb4
...
[mysqld]
...
collation-server = utf8mb4_unicode_ci
init-connect='SET NAMES utf8mb4'
character-set-server = utf8mb4
...

Note that the default-character-set option is a client option, not a server option.

See Also

Comments

Comments loading...
Content reproduced on this site is the property of its respective owners, and this content is not reviewed in advance by MariaDB. The views, information and opinions expressed by this content do not necessarily represent those of MariaDB or any other party.