Skip to content

Async, Netty based, database drivers for PostgreSQL and MySQL written in Scala

License

Notifications You must be signed in to change notification settings

nmarasoiu/postgresql-async

 
 

Repository files navigation

Build Status postgresql-async & mysql-async - async, Netty based, database drivers for MySQL and PostgreSQL written in Scala 2.10 and 2.11

The main goal for this project is to implement simple, async, performant and reliable database drivers for PostgreSQL and MySQL in Scala. This is not supposed to be a JDBC replacement, these drivers aim to cover the common process of send a statement, get a response that you usually see in applications out there. So it's unlikely there will be support for updating result sets live or stuff like that.

This project always returns JodaTime when dealing with date types and not the java.util.Date class.

If you want information specific to the drivers, check the PostgreSQL README and the MySQL README.

You can view the project's CHANGELOG here.

Abstractions and integrations

  • Activate Framework - full ORM solution for persisting objects using a software transactional memory (STM) layer;
  • ScalikeJDBC-Async - provides an abstraction layer on top of the driver allowing you to write less SQL and make use of a nice high level database access API;
  • mod-mysql-postgresql - vert.x module that integrates the driver into a vert.x application;
  • dbmapper - enables SQL queries with automatic mapping from the database table to the Scala class and a mechanism to create a Table Date Gateway model with very little boiler plate code;
  • Quill - A compile-time language integrated query library for Scala.

Include them as dependencies

And if you're in a hurry, you can include them in your build like this, if you're using PostgreSQL:

"com.github.mauricio" %% "postgresql-async" % "0.2.20"

Or Maven:

<dependency>
  <groupId>com.github.mauricio</groupId>
  <artifactId>postgresql-async_2.11</artifactId>
  <version>0.2.20</version>
</dependency>

And if you're into MySQL:

"com.github.mauricio" %% "mysql-async" % "0.2.20"

Or Maven:

<dependency>
  <groupId>com.github.mauricio</groupId>
  <artifactId>mysql-async_2.11</artifactId>
  <version>0.2.20</version>
</dependency>

Database connections and encodings

READ THIS NOW

Both clients will let you set the database encoding for something else. Unless you are 1000% sure of what you are doing, DO NOT change the default encoding (currently, UTF-8). Some people assume the connection encoding is the database or columns encoding but IT IS NOT, this is just the connection encoding that is used between client and servers doing communication.

When you change the encoding of the connection you are not affecting your database's encoding and your columns WILL NOT be stored with the connection encoding. If the connection and database/column encoding is different, your database will automatically translate from the connection encoding to the correct encoding and all your data will be safely stored at your database/column encoding.

This is as long as you are using the correct string types, BLOB columns will not be translated since they're supposed to hold a stream of bytes.

So, just don't touch it, create your tables and columns with the correct encoding and be happy.

Prepared statements gotcha

If you have used JDBC before, you might have heard that prepared statements are the best thing on earth when talking to databases. This isn't exactly true all the time (as you can see on this presentation by @tenderlove) and there is a memory cost in keeping prepared statements.

Prepared statements are tied to a connection, they are not database-wide, so, if you generate your queries dynamically all the time you might eventually blow up your connection memory and your database memory.

Why?

Because when you create a prepared statement, locally, the connection keeps the prepared statement description in memory. This can be the returned columns information, input parameters information, query text, query identifier that will be used to execute the query and other flags. This also causes a data structure to be created at your server for every connection.

So, prepared statements are awesome, but are not free. Use them judiciously.

What are the design goals?

  • fast, fast and faster
  • small memory footprint
  • avoid copying data as much as possible (we're always trying to use wrap and slice on buffers)
  • easy to use, call a method, get a future or a callback and be happy
  • never, ever, block
  • all features covered by tests
  • next to zero dependencies (it currently depends on Netty, JodaTime and SFL4J only)

What is missing?

  • more authentication mechanisms
  • benchmarks
  • more tests (run the jacoco:cover sbt task and see where you can improve)
  • timeout handler for initial handshare and queries

How can you help?

  • checkout the source code
  • find bugs, find places where performance can be improved
  • check the What is missing piece
  • check the issues page for bugs or new features
  • send a pull request with specs

Main public interface

Connection

Represents a connection to the database. This is the root object you will be using in your application. You will find three classes that implement this trait, PostgreSQLConnection, MySQLConnection and ConnectionPool. The difference between them is that ConnectionPool is, as the name implies, a pool of connections and you need to give it an connection factory so it can create connections and manage them.

To create both you will need a Configuration object with your database details. You can create one manually or create one from a JDBC or Heroku database URL using the URLParser object.

QueryResult

It's the output of running a statement against the database (either using sendQuery or sendPreparedStatement). This object will contain the amount of rows, status message and the possible ResultSet (Option[ResultSet]) if the query returns any rows.

ResultSet

It's an IndexedSeq[Array[Any]], every item is a row returned by the database. The database types are returned as Scala objects that fit the original type, so smallint becomes a Short, numeric becomes BigDecimal, varchar becomes String and so on. You can find the whole default transformation list at the project specific documentation.

Prepared statements

Databases support prepared or precompiled statements. These statements allow the database to precompile the query on the first execution and reuse this compiled representation on future executions, this makes it faster and also allows for safer data escaping when dealing with user provided data.

To execute a prepared statement you grab a connection and:

val connection : Connection = ...
val future = connection.sendPreparedStatement( "SELECT * FROM products WHERE products.name = ?", Array( "Dominion" ) )

The ? (question mark) in the query is a parameter placeholder, it allows you to set a value in that place in the query without having to escape stuff yourself. The driver itself will make sure this parameter is delivered to the database in a safe way so you don't have to worry about SQL injection attacks.

The basic numbers, Joda Time date, time, timestamp objects, strings and arrays of these objects are all valid values as prepared statement parameters and they will be encoded to their respective database types. Remember that not all databases are created equal, so not every type will work or might work in unexpected ways. For instance, MySQL doesn't have array types, so, if you send an array or collection to MySQL it won't work.

Remember that parameters are positional the order they show up at query should be the same as the one in the array or sequence given to the method call.

Transactions

Both drivers support transactions at the database level, the isolation level is the default for your database/connection, to change the isolation level just call your database's command to set the isolation level for what you want.

Here's an example of how transactions work:

  val future = connection.inTransaction {
    c =>
    c.sendPreparedStatement(this.insert)
     .flatMap( r => c.sendPreparedStatement(this.insert))
  }

The inTransaction method allows you to execute a collection of statements in a single transactions, just use the connection object you will receive in your block and send your statements to it. Given each statement causes a new future to be returned, you need to flatMap the calls to be able to get a Future[T] instead of Future[Future[...]] back.

If all futures succeed, the transaction is committed normally, if any of them fail, a rollback is issued to the database. You should not reuse a database connection that has rolled back a transaction, just close it and create a new connection to continue using it.

Example usage (for PostgreSQL, but it looks almost the same on MySQL)

You can find a small Play 2 app using it here and a blog post about it here.

In short, what you would usually do is:

import com.github.mauricio.async.db.postgresql.PostgreSQLConnection
import com.github.mauricio.async.db.postgresql.util.URLParser
import com.github.mauricio.async.db.util.ExecutorServiceUtils.CachedExecutionContext
import com.github.mauricio.async.db.{RowData, QueryResult, Connection}
import scala.concurrent.duration._
import scala.concurrent.{Await, Future}

object BasicExample {

  def main(args: Array[String]) {

    val configuration = URLParser.parse("jdbc:postgresql://localhost:5233/my_database?user=postgres&password=somepassword")
    val connection: Connection = new PostgreSQLConnection(configuration)

    Await.result(connection.connect, 5 seconds)

    val future: Future[QueryResult] = connection.sendQuery("SELECT 0")

    val mapResult: Future[Any] = future.map(queryResult => queryResult.rows match {
      case Some(resultSet) => {
        val row : RowData = resultSet.head
        row(0)
      }
      case None => -1
    }
    )

    val result = Await.result( mapResult, 5 seconds )

    println(result)

    connection.disconnect

  }

}

First, create a PostgreSQLConnection, connect it to the database, execute queries (this object is not thread safe, so you must execute only one query at a time) and work with the futures it returns. Once you are done, call disconnect and the connection is closed.

You can also use the ConnectionPool provided by the driver to simplify working with database connections in your app. Check the blog post above for more details and the project's ScalaDocs.

LISTEN/NOTIFY support (PostgreSQL only)

LISTEN/NOTIFY is a PostgreSQL-specific feature for database-wide publish-subscribe scenarios. You can listen to database notifications as such:

  val connection: Connection = ...

  connection.sendQuery("LISTEN my_channel")
  connection.registerNotifyListener {
    message =>
    println(s"channel: ${message.channel}, payload: ${message.payload}")
  }

Contributing

Contributing to the project is simple, fork it on Github, hack on what you're insterested in seeing done or at the bug you want to fix and send a pull request back. If you thing the change is too big or requires architectural changes please create an issue before you start working on it so we can discuss what you're trying to do.

You should be easily able to build this project in your favorite IDE since it's built by SBT using a plugin that generates your IDE's project files. You can use sbt-idea for IntelliJ Idea and sbteclipse for Eclipse integration.

Check our list of contributors!

Licence

This project is freely available under the Apache 2 licence, fork, fix and send back :)

About

Async, Netty based, database drivers for PostgreSQL and MySQL written in Scala

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages

  • Scala 99.1%
  • Other 0.9%