MySQL - Convert latin1 characters on a UTF8 table into UTF8

Question

Only today I realized that I was missing this in my PHP scripts:

mysql_set_charset('utf8');

All my tables are InnoDB, collation "utf8_unicode_ci", and all my VARCHAR columns are "utf8_unicode_ci" as well. I have mb_internal_encoding('UTF-8'); on my PHP scripts, and all my PHP files are encoded as UTF-8.

So, until now, every time I "INSERT" something with diacritics, example:

mysql_query('INSERT INTO `table` SET `name`="Jáuò Iñe"');

The 'name' contents would be, in this case: Jáuò Iñe.

Since I fixed the charset between PHP and MySQL, new INSERTs are now storing correctly. However, I want to fix all the older rows that are "messed" at the moment. I tried many things already, but it always breaks the strings on the first "illegal" character. Here is my current code:

$m = mysql_real_escape_string('¿<?php echo "¬<b>\'PHP &aacute; (á)ţăriîş </b>"; ?> ă-ţi abcdd;//;ñç´พดแทฝใจคçăâξβψδπλξξςαยนñ ;');
mysql_set_charset('utf8');
mysql_query('INSERT INTO `table` SET `name`="'.$m.'"');
mysql_set_charset('latin1');
mysql_query('INSERT INTO `table` SET `name`="'.$m.'"');
mysql_set_charset('utf8');

$result = mysql_iquery('SELECT * FROM `table`');
while ($row = mysql_fetch_assoc($result)) {
    $message = $row['name'];
    $message = mb_convert_encoding($message, 'ISO-8859-15', 'UTF-8');
    //$message = iconv("UTF-8", "ISO-8859-1//IGNORE", $message);
    mysql_iquery('UPDATE `table` SET `name`="'.mysql_real_escape_string($message).'" WHERE `a1`="'.$row['a1'].'"');
}

It "UPDATE"s with the expected characters, except that the string gets truncated after the character "ă". I mean, that character and following chars are not included on the string.

Also, testing with the "iconv()" (that is commented on the code) does the same, even with //IGNORE and //TRANSLIT

I also tested several charsets, between ISO-8859-1 and ISO-8859-15.

I really need help here! Thank you.

Answer 1

From what you describe, it seems you have UTF-8 data that was originally stored as Latin-1 and then not converted correctly to UTF-8. The data is recoverable; you'll need a MySQL function like

convert(cast(convert(name using  latin1) as binary) using utf8)

It's possible that you may need to omit the inner conversion, depending on how the data was altered during the encoding conversion.

Answer 2

After i searched about an hour or two for this answer. I needed to migrate a old tt_news db from typo into a new typo3 version. I already tried convert the charset in the export file and import it back, but didn't get it working.

Then i tried the answer above from ABS and startet a update on the table:

UPDATE tt_news SET 
    title=convert(cast(convert(title using  latin1) as binary) using utf8), 
    short=convert(cast(convert(short using  latin1) as binary) using utf8), 
    bodytext=convert(cast(convert(bodytext using  latin1) as binary) using utf8)
WHERE 1

You can also convert imagecaption, imagealttext, imagetitletext and keywords if needed. Hope this will help somebody migrating tt_news to new typo3 version.

Answer 3

the way is better way use connection tow you database normal

then use this code to make what you need you must make your page encoding utf-8 by meta in header cod html (dont forget this)

then use this code

    $result = mysql_query('SELECT * FROM shops');
    while ($row = mysql_fetch_assoc($ 
    $name= iconv("windows-1256", "UTF-8", $row['name']);

   mysql_query("SET NAMES 'utf8'"); 
   mysql_query("update   `shops` SET `name`='".$name."'  where ID='$row[ID]'  ");
    }