/PgAsync

Asynchronous Reactive Postgres Library in Pure PHP

Primary LanguagePHP

Build Status

PgAsync

Asynchronous Reactive Postgres Library for PHP (Non-blocking)

What it is

This is an asynchronous Postgres library for PHP. Observables are returned by the query methods allowing asynchronous row-by-row data handling (and other Rx operators on the data) See Rx.PHP. Network and event processing is handled by ReactPHP.

This is a pure PHP implementation (you don't need Postgres extensions to use it).

Example - Simple Query

$client = new PgAsync\Client([
    "host" => "127.0.0.1",
    "port" => "5432",
    "user"     => "matt",
    "database" => "matt"
]);

$client->query('SELECT * FROM channel')->subscribe(
    function ($row) {
        var_dump($row);
    },
    function ($e) {
        echo "Failed.\n";
    },
    function () {
        echo "Complete.\n";
    }
);

Example - parameterized query

$client = new PgAsync\Client([
     "host" => "127.0.0.1",
     "port" => "5432",
     "user"     => "matt",
     "database" => "matt",
     "auto_disconnect" => true //This option will force the client to disconnect as soon as it completes.  The connection will not be returned to the connection pool.

]);

$client->executeStatement('SELECT * FROM channel WHERE id = $1', ['5'])
    ->subscribe(
        function ($row) {
            var_dump($row);
        },
        function ($e) {
            echo "Failed.\n";
        },
        function () {
            echo "Complete.\n";
        }
    );

Example - LISTEN/NOTIFY

$client = new PgAsync\Client([
     "host" => "127.0.0.1",
     "port" => "5432",
     "user"     => "matt",
     "database" => "matt"
]);

$client->listen('some_channel')
    ->subscribe(function (\PgAsync\Message\NotificationResponse $message) {
        echo $message->getChannelName() . ': ' . $message->getPayload() . "\n";
    });
    
$client->query("NOTIFY some_channel, 'Hello World'")->subscribe();

Install

With composer install into you project with:

Install pgasync: composer require voryx/pgasync

What it can do

  • Run queries (CREATE, UPDATE, INSERT, SELECT, DELETE)
  • Queue commands
  • Return results asynchronously (using Observables - you get data one row at a time as it comes from the db server)
  • Prepared statements (as parameterized queries)
  • Connection pooling (basic pooling)

What it can't quite do yet

  • Transactions (Actually though, just grab a connection and you can run your transaction on that single connection)

What's next

  • Add more testing
  • Transactions
  • Take over the world

Keep in mind

This is an asynchronous library. If you begin 3 queries (subscribe to their observable):

$client->query("SELECT * FROM table1")->subscribe(...);
$client->query("SELECT * FROM table2")->subscribe(...);
$client->query("SELECT * FROM table3")->subscribe(...);

It will start all of them almost simultaneously (and you will begin receiving rows on all 3 before any of them have completed). This can be great if you want to run 3 queries at the same time, but if you have some queries that need information that was modified by other statements, this can cause a race condition:

$client->query("INSERT INTO invoices(inv_no, customer_id, amount) VALUES('1234A', 1, 35.75)")->subscribe(...);
$client->query("SELECT SUM(amount) AS balance FROM invoices WHERE customer_id = 1")->subscribe(...);

In the above situation, your balance may or may not include the invoice inserted on the first line.

You can avoid this by using the Rx concat* operator to only start up the second observable after the first has completed:

$insert = $client->query("INSERT INTO invoices(inv_no, customer_id, amount) VALUES('1234A', 1, 35.75)");
$select = $client->query("SELECT SUM(amount) AS balance FROM invoices WHERE customer_id = 1");

$insert
    ->concat($select)
    ->subscribe(...);

Testing

We use docker to run a postgresql instance for testing. To run locally, just install docker and run the following command from the project root:

docker-compose -f docker/docker-compose.yml up -d

If you need to reset the database, just stop the docker instance and delete the docker/database directory. Restart the docker with the above command and it will initialize the database again.

The tests do not change the ending structure of the database, so you should not normally need to do this.