How to change collation of all rows from latin1_swedish_ci to utf8_unicode_ci?

Question

I ignorantly used the default latin1_swedish_ci character encoding for all of the varchar rows in my database during development and I've determined that this is the root of the character encoding problems I've been having. In addition to that, it seems like most people these days are recommending that utf8_unicode_ci be used.

I'd like to convert the character encoding for all rows in my database from latin1_swedish_ci to utf8_unicode_ci, but the only way I know how to do is is change it row-by-row in phpMyAdmin, which is really time consuming.

Is there a faster way, such as a query that can be run that changes the collation of all varchar/text rows from latin1_swedish_ci to utf8_unicode_ci?

Answer 1

If the columns are using the default table character set then it's just one query per table to convert:

ALTER TABLE t CONVERT TO CHARACTER SET utf8 COLLATE utf8_unicode_ci;

If the character set is set individually on each column, AFAIK there is no way to do that on all columns of all tables in the database directly in MySql, but you could write a tiny program in your language of choice that does so.

Your program would query the INFORMATION_SCHEMA.COLUMNS table and look at the CHARACTER_SET_NAME column:

SELECT * FROM `INFORMATION_SCHEMA.COLUMNS`
WHERE TABLE_SCHEMA = 'dbname' AND CHARACTER_SET_NAME = 'latin1'

For each result row it's trivial to synthesize and execute an ALTER TABLE query on the spot that changes the character set and collation appropriately:

ALTER TABLE t MODIFY col TEXT CHARACTER SET utf8 COLLATE utf8_unicode_ci;

In the above query t, col and TEXT would be the values of the TABLE_NAME, COLUMN_NAME and DATA_TYPE columns from the INFORMATION_SCHEMA.COLUMNS result set.