This document is part of the guide on How to Fix the Collation and Character Set of a MySQL Database. Please refer to that page for more information.

Table of Contents

• MySQL Collation Repair: Database Level Changes
• MySQL Collation Repair: Table Level Changes
• MySQL Collation Repair: Column Level Changes
• MySQL Collation Repair: Column Level Encoding Issues
• MySQL Collation Repair: Case Study - Repairing a Production Database

Before Proceeding

When to apply this fix

If you notice that your content is not displaying correctly, you should apply this fix. For example, if you have the string José that is being displayed as José, that is an example of when to apply this fix.

Repairing a column with incorrectly encoded content

You'll need to select the column converting to the previousEncoding (such as latin1) and cast it to binary; then to utf8.

Setting SQL_SAFE_UPDATES to 0 before issuing the query allows you to update each row in the table. This is potentially unsafe, and should be tested thoroughly before being applied to a production system.

SET SQL_SAFE_UPDATES=0;
UPDATE tableName SET columnName = CONVERT(CAST(CONVERT(columnName USING previousEncoding) AS BINARY) USING utf8);
SET SQL_SAFE_UPDATES=1;

Depending on the character and how it's encoded, this query may not be enough to resolve the encoding issues you're seeing. You may wish to try one of the alternative methods outlined in the "Further Reading" section below.

Further Reading:

• Percona - Jervin Real: utf8 data on latin1 tables; converting to utf8 without downtime or double encoding
• Joni Salonen - Fixing Mangled Characters
• Joni Salonen - Fixing Doubly UTF-8 Encoded Text in MySQL