A UTF8 round trip with MySQL

In ASCII, a character always maps to a single byte.
In UTF-8, a character may map to multiple bytes.
For Perl to know whether the data it receives from an external source (which could be a string, or binary data such as an image) as a string of bytes or as a UTF-8 string, it uses the internal UTF8 flag.
Nothing external to Perl (eg the console or the database) knows about this flag, so we need to transform all input/output data into a form that each program understands, which we do using Encode.
To convert an input string of bytes which represents a UTF-8 string, into Perl's internal string format, we DECODE the byes to from UTF-8, using Encode::decode() or Encode::decode_utf8()
To convert a Perl string into a string of bytes representing a UTF-8 for other programs to understand, we ENCODE the string using Encode::encode() or Encode::encode_utf8()

An example round trip

The steps below demonstrate how to accept UTF-8 strings from outside, store them in MySQL, retrieve them from the database, and re-output them


                Manual, or with              Automatic with 
                PerlIO layers                MySQL::dbd > v 4
                ---------------              ---------------- 
 ---------                       ---------                       -----
+-----
Ś  UTF-8  Ś ---decode_utf8()->> Ś  Perl   Ś ---encode_utf8()->> Ś  UTF
+-8   Ś
Ś Console Ś <<-encode_utf8()--- Ś strings Ś <<-decode_utf8()--- Ś MySQ
+L DB Ś
 ---------                       ---------                       -----
+-----
[download]

Create a table in MySQL which uses the UTF-8 character set:

This step ensures that all UTF-8 aware programs that interact with this database know to treat the stored data as UTF-8
```
    CREATE TABLE test_db.test ( string VARCHAR(50) ) CHARACTER SET utf
+8;
[download]
```

Get a UTF-8 string:

This step accepts a string of bytes representing a UTF-8 string, and converts them into Perl's internal string format.

From a UTF-8 console:


    use Encode qw( decode_utf8 );
    my $string      = <>;
    my $utf8_string = decode_utf8($string);
[download]

or, from an ISO-8859-1 console:


   use Encode qw( decode );
   my $iso_8859_string = <>;
   my $utf8_string     = decode('ISO-8859-1',$iso_8859_string);
[download]

or from within a Perl script:


    use utf8;         # Tells Perl that the script itself is written i
+n UTF-8

    my $utf8_string = "UTF-8 string with special chars: ń ć ô";
[download]

Open a UTF-8 enabled database connection:

This step connects to the database, and tells DBD::mysql to auto-convert to/from UTF-8.

IMPORTANT: This requires a version of DBD::mysql greater than version 4


    use DBI();
    my $dbh = DBI->connect ('dbi:mysql:test_db',
                            $username,
                            $password, 
                            {mysql_enable_utf8 => 1}
    );
[download]

Write to and read from the DB:


    $dbh->do('INSERT INTO test_db.test VALUES(?)', $utf8_string);

    $dbh->do('SELECT string FROM test_db.test LIMIT 1');
    my $new_string = $dbh->fetchrow_arrayref->[0];
[download]

Display the retrieved string:

The output data needs to be converted from Perl's internal format into a string of bytes that the console will understand.

on a UTF-8 console:


    use Encode qw( encode_utf8 );
    print Encode::encode_utf8($new_string);

 OR

    # Add an auto-encoding layer
    binmode (STDIN,':utf8');
    print $new_string;
[download]

or, on an ISO-8859-1 console:


    use Encode qw( encode );
    print Encode::encode('ISO-8859-1', $new_string);
[download]

For more info, see perlunitut: Unicode in Perl, perluniintro, perlunicode, perlrun, binmode, open and PerlIO.

UPDATE - Added readmore tags. Added diagram illustrating round trip

UPDATE - Corrected a type: TO utf8, not from utf8

Back to Meditations

An example round trip

Create a table in MySQL which uses the UTF-8 character set:

Get a UTF-8 string:

Open a UTF-8 enabled database connection:

Write to and read from the DB:

Display the retrieved string: