I needed to access the database from within one particular webhosting service. Pages are UTF-8 encoded and data received by forms should be inserted into database without changing the encoding. The database is also in UTF-8.
Neither SET character set 'utf8' or SET names 'utf8' worked properly here, so this workaround made sure all variables are set to utf-8.
<?php
// ... (creating a connection to mysql) ...
mysql_query("SET character_set_results = 'utf8', character_set_client = 'utf8', character_set_connection = 'utf8', character_set_database = 'utf8', character_set_server = 'utf8'", $conn);
$re = mysql_query('SHOW VARIABLES LIKE "%character_set%";')or die(mysql_error());
while ($r = mysql_fetch_assoc($re)) {var_dump ($r); echo "<br />";} exit;
?>
All important variables are now utf-8 and we can safely use INSERTs or SELECTs with mysql_escape_string($var) without any encoding functions.
mysql_set_charset
(PHP 5 >= 5.2.3)
mysql_set_charset — Sets the client character set
Warning
이 확장은 PHP 5.5.0부터 배제되었고, 차후 제거될 것입니다. 대신, MySQLi나 PDO_MySQL 확장을 사용하십시오. MySQL: API 선택하기 가이드와 관련 FAQ를 참조하십시오. 이 함수를 대체하여:
- mysqli_set_charset()
- PDO: Add charset to the connection string, such as charset=utf8
설명
bool mysql_set_charset
( string
$charset
[, resource $link_identifier = NULL
] )Sets the default character set for the current connection.
인수
-
charset -
A valid character set name.
-
link_identifier -
MySQL 연결. 지정하지 않으면 mysql_connect()로 연 마지막 연결을 사용합니다. 연결이 없으면, 인수 없이 mysql_connect()를 호출하여 연결을 만듭니다. 연결이 성립되지 않으면
E_WARNING등급의 오류를 생성합니다.
반환값
성공 시 TRUE를, 실패 시 FALSE를 반환합니다.
주의
Note:
This function requires MySQL 5.0.7 or later.
Note:
This is the preferred way to change the charset. Using mysql_query() to set it (such as SET NAMES utf8) is not recommended. See the MySQL character set concepts section for more information.
add a note
User Contributed Notes 8 notes
Ashus ¶
16 years ago
nabeelmoidu at gmail dot com ¶
16 years ago
Here's an example of how to use this feature :
I'm using PHP 5.2.5 from http://oss.oracle.com/projects/php/
I wanted to store arabic characters as UTF8 in the database and as suggested in many of the google results, I tried using
mysql_query("SET NAMES 'utf8'");
and
mysql_query("SET CHARACTER SET utf8 ");
but it did not work.
Using the following, it worked flawlessly
$link = mysql_connect('localhost', 'user', 'password');
mysql_set_charset('utf8',$link);
Once this is set we need not manually encode the text into utf using utf8_encode() or other functions. The arabic ( or any UTF8 supported ) text can be passed directly to the database and it is automatically converted by PHP.
For eg.
<?php
$link = mysql_connect('localhost', 'user', 'password');
mysql_set_charset('utf8',$link);
$db_selected = mysql_select_db('emp_feedback', $link);
if (!$db_selected) { die ('Database access error : ' . mysql_error());}
$query = "INSERT INTO feedback ( EmpName, Message ) VALUES ('$_empName','$_message')";
mysql_query($query) or die('Error, Feedback insert into database failed');
?>
Note that here $_empName is stored in English while $_message is in Arabic.
Janez R. ¶
17 years ago
I assume that this is an equivalent in previous versions of php (add some parameter validation and default values though!):
<?
if (!function_exists('mysql_set_charset')) {
function mysql_set_charset($charset,$dbh)
{
return mysql_query("set names $charset",$dbh);
}
}
?>
nag QWE svgfr RTY org ¶
17 years ago
Here's an improved version of Janez R.'s function:
<?php
if (function_exists('mysql_set_charset') === false) {
/**
* Sets the client character set.
*
* Note: This function requires MySQL 5.0.7 or later.
*
* @see http://www.php.net/mysql-set-charset
* @param string $charset A valid character set name
* @param resource $link_identifier The MySQL connection
* @return TRUE on success or FALSE on failure
*/
function mysql_set_charset($charset, $link_identifier = null)
{
if ($link_identifier == null) {
return mysql_query('SET CHARACTER SET "'.$charset.'"');
} else {
return mysql_query('SET CHARACTER SET "'.$charset.'"', $link_identifier);
}
}
}
?>
Anonymous ¶
16 years ago
Actually, this function is available in client libraries in MySQL 4.1.13 and newer, too. So the real version requirement is MySQL >= 5.0.7 OR, if you're using MySQL 4, then >= 4.1.13.
emeszenes at gmail dot com ¶
10 years ago
Massage for nabeelmoidu at gmail dot com:
For me works following code:
$mysqli = mysqli_connect( ... );
mysqli_query( $mysqli, 'SET NAMES "utf8" COLLATE "utf8_general_ci"' );
or just:
mysqli_set_charset( $mysqli, 'utf8' );
Regards, good luck!
Ronald ¶
16 years ago
I need to revoke most of my post below. What I found out afterwards is this:
1. if you do not use mysql_set_char mysql will NOT do any translations and thus store a utf8-character-byte as is. If you then retrieve this byte from the db and output it in a utf8 page it will show just fine BUT if other apps query this byte (expecting to find a latin1 byte) they will go wrong.
2. the 'bug' mentioned before only occurs if you use a ucase or lcase function in your statement (like: latin1_col = ucase('utf8 string')
Hope this helps, Ronald
Ronald ¶
16 years ago
I just hope that the text below will help someone who is struggling with charset encoding, specially when php-charset is different from the mysql-charset. Let me add that I really think that the php man-pages on the mysql-functions are lacking a lot of details on this important issues. Could someone add some useful text here?
Here is my situation. PHP5.2.4, MySql 4.1.15. A php web-application fully utf-8 encoded and a mysql database in latin1 charset.
To make this work I had to:
1. create and store all code files (php, html, inc, js, etc) in the utf-8 charset. Your editor should have an option for this, if not dump it.
2. check that your editor does not add a BOM (http://en.wikipedia.org/wiki/Byte-order_mark) at the beginning of the file. Use a hex-editor to detect them if needed.
3. Set your apache environment to utf-8 by adding 'AddDefaultCharset utf-8' to your .htaccess. If you do not use apache add 'default_charset utf-8' to your php.ini. You have to do either of them (not both), php will use the apache setting where needed.
4. Additionally add this meta-tag to your html-header: '<meta http-equiv="Content-type" content="text/html; charset=UTF-8" />'. This will help silly browsers (Oeps, IE again?) that ignore the utf-response-header send to them.
5. Check that the above line are listened to by check the 'page info' of your pages in firefox. It should show 2 (!!) utf-8 entries.
======== all of the above sofar has nothing to do with mysql ;-) ======
6. Do *NOT* (repeat NOT!) set the 'names' (set names *) or _ANY_ 'character set' (set character set *) (opposed to what they tell you on these pages).
7. Check the previous item by listing the results of the mysql query 'SHOW session VARIABLES'. All char_sets here should say 'latin1', except for the system one which is always 'utf8'. All collations should say 'latin1_*'. Furthermore the php function mysql_client_encoding() should also return latin1 (though I don't understand why; what does this value mean, I would think if php (being the client) is utf8 encoded this would be utf8?)
8 Finally test the above by storing this string in your db and output it in your webpage: 'Iñtërnâtiônàlizætiøn and €'.
Now what was interesting during testing and debugging of the above findings was:
1. If I would run 'mysql_set_charset('utf8')' _OR_ 'mysql_query("SET NAMES 'utf8'");' and then run a query in which I would have 'where char_column = 'abc''it would die with 'Illegal mix of collations'
2. If I would run 'mysql_query("SET character_set_client = 'utf8';"); mysql_query("SET character_set_result = 'utf8';")' the query would work BUT the non-ascii-characters would show scrambled in the browser.
3. BUT these 2 points above work just fine on my local dev-machine (php 5.2.3 & mysql 5.0.45)!!!!!!!!
This draws me to these 3 conclusions:
1. The Php-mysql-function library (5.2.+) does a fine job translating utf-8 queries & results to/from latin1! It's better to let php handle this for you then to have mysql do this.
2. Mysql (4.0.+) has 1 or more bugs (well, let's say unfinished features) that involve the charset-translations that are solved in 5.0.+.
3. It is not well enough documented! (Otherwise I would have to write this)
One last remark: clearly characters that exist in utf8 and not in latin1 (and vv.) will get lost during utf8-latin1-utf8 translation.
If any of the above is not correct or not complete feel free to correct this! (Or better yet, add a chapter to the php manual :-)
Cheers, Ronald
