Mojo::mysql - Mojolicious and Async MySQL
use Mojo::mysql;
# Create a table
my $mysql = Mojo::mysql->new('mysql://username@/test');
$mysql->db->query(
'create table names (id integer auto_increment primary key, name text)');
# Insert a few rows
my $db = $mysql->db;
$db->query('insert into names (name) values (?)', 'Sara');
$db->query('insert into names (name) values (?)', 'Stefan');
# Insert more rows in a transaction
eval {
my $tx = $db->begin;
$db->query('insert into names (name) values (?)', 'Baerbel');
$db->query('insert into names (name) values (?)', 'Wolfgang');
$tx->commit;
};
say $@ if $@;
# Insert another row and return the generated id
say $db->query('insert into names (name) values (?)', 'Daniel')
->last_insert_id;
# Select one row at a time
my $results = $db->query('select * from names');
while (my $next = $results->hash) {
say $next->{name};
}
# Select all rows blocking
$db->query('select * from names')
->hashes->map(sub { $_->{name} })->join("\n")->say;
# Select all rows non-blocking
Mojo::IOLoop->delay(
sub {
my $delay = shift;
$db->query('select * from names' => $delay->begin);
},
sub {
my ($delay, $err, $results) = @_;
$results->hashes->map(sub { $_->{name} })->join("\n")->say;
}
)->wait;
# Send and receive notifications non-blocking
$mysql->pubsub->listen(foo => sub {
my ($pubsub, $payload) = @_;
say "foo: $payload";
$pubsub->notify(bar => $payload);
});
$mysql->pubsub->listen(bar => sub {
my ($pubsub, $payload) = @_;
say "bar: $payload";
});
$mysql->pubsub->notify(foo => 'MySQL rocks!');
Mojo::IOLoop->start unless Mojo::IOLoop->is_running;
Mojo::mysql is a tiny wrapper around DBD::mysql that makes MySQL a lot of fun to use with the Mojolicious real-time web framework.
Database and handles are cached automatically, so they can be reused transparently to increase performance. And you can handle connection timeouts gracefully by holding on to them only for short amounts of time.
use Mojolicious::Lite;
use Mojo::mysql;
helper mysql =>
sub { state $pg = Mojo::mysql->new('mysql://sri:s3cret@localhost/db') };
get '/' => sub {
my $c = shift;
my $db = $c->mysql->db;
$c->render(json => $db->query('select now() as time')->hash);
};
app->start;
While all I/O operations are performed blocking, you can wait for long running queries asynchronously, allowing the Mojo::IOLoop event loop to perform other tasks in the meantime. Since database connections usually have a very low latency, this often results in very good performance.
Every database connection can only handle one active query at a time, this includes asynchronous ones. So if you start more than one, they will be put on a waiting list and performed sequentially. To perform multiple queries concurrently, you have to use multiple connections.
# Performed sequentially (10 seconds)
my $db = $mysql->db;
$db->query('select sleep(5)' => sub {...});
$db->query('select sleep(5)' => sub {...});
# Performed concurrently (5 seconds)
$mysql->db->query('select sleep(5)' => sub {...});
$mysql->db->query('select sleep(5)' => sub {...});
All cached database handles will be reset automatically if a new process has been forked, this allows multiple processes to share the same Mojo::mysql object safely.
Note that this whole distribution is EXPERIMENTAL and will change without warning!
Mojo::mysql inherits all events from Mojo::EventEmitter and can emit the following new ones.
$mysql->on(connection => sub {
my ($mysql, $dbh) = @_;
...
});
Emitted when a new database connection has been established.
Mojo::mysql implements the following attributes.
my $dsn = $mysql->dsn;
$mysql = $mysql->dsn('dbi:mysql:dbname=foo');
Data Source Name, defaults to dbi:mysql:dbname=test
.
my $max = $mysql->max_connections;
$mysql = $mysql->max_connections(3);
Maximum number of idle database handles to cache for future use, defaults to 5
.
my $migrations = $mysql->migrations;
$mysql = $mysql->migrations(Mojo::mysql::Migrations->new);
Mojo::mysql::Migrations object you can use to change your database schema more easily.
# Load migrations from file and migrate to latest version
$mysql->migrations->from_file('/Users/sri/migrations.sql')->migrate;
MySQL does not support nested transactions and DDL transactions. DDL statements cause implicit COMMIT
. ROLLBACK
will be called if any step of migration script fails, but only DML statements after the last implicit or explicit COMMIT
can be reverted. Not all MySQL storage engines (like MYISAM
) support transactions.
This means database will most likely be left in unknown state if migration script fails. Use this feature with caution and remember to always backup your database.
my $options = $mysql->options;
$mysql = $mysql->options({mysql_use_result => 1});
Options for database handles, defaults to activating mysql_enable_utf8
, AutoCommit
, AutoInactiveDestroy
as well as RaiseError
and deactivating PrintError
. Note that AutoCommit
and RaiseError
are considered mandatory, so deactivating them would be very dangerous.
mysql_auto_reconnect
is never enabled, Mojo::mysql takes care of dead connections.
AutoCommit
cannot not be disabled, use $db->begin to manage transactions.
RaiseError
is enabled for blocking and disabled in event loop for non-blocking queries.
my $password = $mysql->password;
$mysql = $mysql->password('s3cret');
Database password, defaults to an empty string.
my $pubsub = $mysql->pubsub;
$mysql = $mysql->pubsub(Mojo::mysql::PubSub->new);
Mojo::mysql::PubSub object you can use to send and receive notifications very efficiently, by sharing a single database connection with many consumers.
# Subscribe to a channel
$mysql->pubsub->listen(news => sub {
my ($pubsub, $payload) = @_;
say "Received: $payload";
});
# Notify a channel
$mysql->pubsub->notify(news => 'MySQL rocks!');
my $username = $mysql->username;
$mysql = $mysql->username('batman');
Database username, defaults to an empty string.
Mojo::mysql inherits all methods from Mojo::EventEmitter and implements the following new ones.
my $db = $mysql->db;
Get Mojo::mysql::Database object for a cached or newly created database handle. The database handle will be automatically cached again when that object is destroyed, so you can handle connection timeouts gracefully by holding on to it only for short amounts of time.
$mysql = $mysql->from_string('mysql://user@/test');
Parse configuration from connection string.
# Just a database
$mysql->from_string('mysql:///db1');
# Username and database
$mysql->from_string('mysql://batman@/db2');
# Username, password, host and database
$mysql->from_string('mysql://batman:s3cret@localhost/db3');
# Username, domain socket and database
$mysql->from_string('mysql://batman@%2ftmp%2fmysql.sock/db4');
# Username, database and additional options
$mysql->from_string('mysql://batman@/db5?PrintError=1&RaiseError=0');
my $mysql = Mojo::mysql->new;
my $mysql = Mojo::mysql->new('mysql://user@/test');
Construct a new Mojo::mysql object and parse connection string with "from_string" if necessary.
This is the class hierarchy of the Mojo::mysql distribution.
Curt Hochwender, [email protected]
.
Jan Henning Thorsen, [email protected]
.
This code is mostly a rip-off from Sebastian Riedel's Mojo::Pg.
Copyright (C) 2014-2015, Jan Henning Thorsen.
This program is free software, you can redistribute it and/or modify it under the terms of the Artistic License version 2.0.
https://github.com/jhthorsen/mojo-mysql,
Mojo::Pg Async Connector for PostgreSQL using DBD::Pg, https://github.com/kraih/mojo-pg,
Mojo::MySQL5 Pure-Perl non-blocking I/O MySQL Connector, https://github.com/harry-bix/mojo-mysql5,