Jobeet - Day 3: The Data Model

Previously on Jobeet

Those of you itching to open your text editor and lay down some PHP will be happy to know today's tutorial will get us into some development. We will define the Jobeet data model, use an ORM to interact with the database, and build the first module of our application. But as symfony does a lot of the work for us, we will have a fully functional web module without writing too much PHP code.

The Relational Model

The user stories we have written yesterday describe the main objects of our project: jobs, affiliates, and categories. Here is the corresponding entity relationship diagram:

In addition to the columns described in the stories, we have also added a created_at field to some tables. Symfony recognizes such fields and sets the value to the current system time when a record is created. That's the same for updated_at fields: Their value is set to the system time whenever the record is updated.

The Schema

To store the jobs, affiliates, and categories, we obviously need a relational database.

But as symfony is an Object-Oriented framework, we like to manipulate objects whenever we can. For example, instead of writing SQL statements to retrieve records from the database, we'd rather prefer to use objects.

The relational database information must be mapped to an object model. This can be done with an ORM tool and thankfully, symfony comes bundled with two of them: Propel and Doctrine. In this tutorial, we will use Propel.

The ORM needs a description of the tables and their relationships to create the related classes. There are two ways to create this description schema: by introspecting an existing database or by creating it by hand.

Some tools allow you to build a database graphically (for instance Fabforce's Dbdesigner) and generate directly a schema.xml (with DB Designer 4 TO Propel Schema Converter).

As the database does not exist yet and as we want to keep Jobeet database agnostic, let's create the schema file by hand by editing the empty config/schema.yml file:

# config/schema.yml
propel:
  jobeet_category:
    id:           ~
    name:         { type: varchar(255), required: true }
 
  jobeet_job:
    id:           ~
    category_id:  { type: integer, foreignTable: jobeet_category, foreignReference: id, required: true }
    type:         { type: varchar(255) }
    company:      { type: varchar(255), required: true }
    logo:         { type: varchar(255) }
    url:          { type: varchar(255) }
    position:     { type: varchar(255), required: true }
    location:     { type: varchar(255), required: true }
    description:  { type: longvarchar, required: true }
    how_to_apply: { type: longvarchar, required: true }
    token:        { type: varchar(255), required: true, index: unique }
    is_public:    { type: boolean, required: true, default: 1 }
    is_activated: { type: boolean, required: true, default: 0 }
    email:        { type: varchar(255), required: true }
    expires_at:   { type: timestamp, required: true }
    created_at:   ~
    updated_at:   ~
 
  jobeet_affiliate:
    id:           ~
    url:          { type: varchar(255), required: true }
    email:        { type: varchar(255), required: true, index: unique }
    token:        { type: varchar(255), required: true }
    is_active:    { type: boolean, required: true, default: 0 }
    created_at:   ~
 
  jobeet_category_affiliate:
    category_id:  { type: integer, foreignTable: jobeet_category, foreignReference: id, required: true, primaryKey: true, onDelete: cascade }
    affiliate_id: { type: integer, foreignTable: jobeet_affiliate, foreignReference: id, required: true, primaryKey: true, onDelete: cascade }
 

If you have decided to create the tables by writing SQL statements, you can generate the corresponding schema.yml configuration file by running the propel:build-schema task.

The schema is the direct translation of the entity relationship diagram in the YAML format.

The YAML Format

According to the official YAML website, YAML is "is a human friendly data serialization standard for all programming languages"

Put another way, YAML is a simple language to describe data (strings, integers, dates, arrays, and hashes).

In YAML, structure is shown through indentation, sequence items are denoted by a dash, and key/value pairs within a map are separated by a colon. YAML also has a shorthand syntax to describe the same structure with fewer lines, where arrays are explicitly shown with [] and hashes with {}.

If you are not yet familiar with YAML, it is time to get started as the symfony framework uses it extensively for its configuration files.

The schema.yml file contains the description of all tables and their columns. Each column is described with the following information:

• type: The column type (boolean, tinyint, smallint, integer, bigint, double, float, real, decimal, char, varchar(size), longvarchar, date, time, timestamp, blob, and clob)
• required: Set it to true if you want the column to be required
• index: Set it to true if you want to create an index for the column or to unique if you want a unique index to be created on the column.

For columns set to ~ (id, created_at, and updated_at), symfony will guess the best configuration (primary key for id and timestamp for created_at and updated_at).

The onDelete attribute defines the ON DELETE behavior of foreign keys, and Propel supports CASCADE, SETNULL, and RESTRICT. For instance, when a job record is deleted, all the jobeet_category_affiliate related records will be automatically deleted by the database or by Propel if the underlying engine does not support this functionality.

The Database

The symfony framework supports all PDO-supported databases (MySQL, PostgreSQL, SQLite, Oracle, MSSQL, ...). PDO is the database abstraction layer bundled with PHP.

Let's use MySQL for this tutorial:

$ mysqladmin -uroot -pmYsEcret create jobeet

Feel free to choose another database engine if you want. It won't be difficult to adapt the code we will write as we will use the ORM will write the SQL for us.

We need to tell symfony to use this database for the Jobeet project:

$ php symfony configure:database "mysql:host=localhost;dbname=jobeet" root mYsEcret

The configure:database task takes three arguments: the PDO DSN, the username, and the password to access the database. If you don't have any password on your development server, just omit the third argument.

The configure:database task stores the database configuration into the config/databases.yml configuration file. Instead of using the task, you can edit this file by hand.

The ORM

Thanks to the database description from the schema.yml file, we can use some Propel built-in tasks to generate the SQL statements needed to create the database tables:

$ php symfony propel:build-sql

The propel:build-sql task generates SQL statements in the data/sql directory, optimized for the database engine we have configured:

# snippet from data/sql/lib.model.schema.sql
CREATE TABLE `jobeet_category`
(
  `id` INTEGER  NOT NULL AUTO_INCREMENT,
  `name` VARCHAR(255)  NOT NULL,
  PRIMARY KEY (`id`)
)Type=InnoDB;
 

To actually create the tables in the database, you need to run the propel:insert-sql task:

$ php symfony propel:insert-sql

As the task drops the current tables before re-creating them, you are required to confirm the operation. You can also add the --no-confirmation option to bypass the question, which is useful if you want to run the task from within a non-interactive batch:

$ php symfony propel:insert-sql --no-confirmation

As for any command line tool, symfony tasks can take arguments and options. Each task comes with a built-in help message that can be displayed by running the help task:

$ php symfony help propel:insert-sql

The help message lists all the possible arguments and options, gives the default values for each of them, and provides some useful usage examples.

The ORM also generates PHP classes that map table records to objects:

$ php symfony propel:build-model

The propel:build-model task generates PHP files in the lib/model directory that can be used to interact with the database.

By browsing the generated files, you have probably noticed that Propel generates four classes per table. For the jobeet_job table:

• JobeetJob: An object of this class represents a single record of the jobeet_job table. The class is empty by default.

• BaseJobeetJob: The parent class of JobeetJob. Each time you run propel:build-model, this class is overridden, so all customizations must be done in the JobeetJob class.

• JobeetJobPeer: The class defines static methods that mostly return collections of JobeetJob objects. The class is empty by default.

• BaseJobeetJobPeer: The parent class of JobeetJobPeer. Each time you run propel:build-model, this class is overridden, so all customizations must be done in the JobeetJobPeer class.

The column values of a record can be manipulated with a model object by using some accessors (get*() methods) and mutators (set*() methods):

$job = new JobeetJob();
$job->setPosition('Web developer');
$job->save();
 
echo $job->getPosition();
 
$job->delete();
 

You can also define foreign keys directly by linking objects together:

$category = new JobeetCategory();
$category->setName('Programming');
 
$job = new JobeetJob();
$job->setCategory($category);
 

The propel:build-all task is a shortcut for the tasks we have run in this section and some more. So, run this task now to generate forms and validators for the Jobeet model classes:

$ php symfony propel:build-all

You will see validators in action at the end of the day and forms will be explained in great details on day 10.

The propel:build-all-load task is a shortcut for the propel:build-all task followed by the propel:data-load task.

As you will see later on, symfony autoloads PHP classes for you, which means that you never need to use a require in your code. It is one of the numerous things that symfony automates for the developer but there is one downside: whenever you add a new class, you need to clear the symfony cache. As the propel:build-model has created a lot of new classes, let's clear the cache:

 $ php symfony cache:clear

A symfony task is made of a namespace and a task name. Each one can be shortened as much as there is no ambiguity with other tasks. So, the following command is equivalent to cache:clear:

$ php symfony cc

The Initial Data

The tables have been created in the database but there is no data in them. For any web application, there are three types of data:

• Initial data: Initial data are needed for the application to work. For example, Jobeet needs some initial categories. If not, nobody will be able to submit a job. We also need an admin user to be able to login to the backend.

• Test data: Test data are needed for the application to be tested. As a developer, you will write tests to ensure that Jobeet behaves as described in the user stories, and the best way is to write automated tests. So, each time you run your tests, you need a clean database with some fresh data to test on.

• User data: User data are created by the users during the normal life of the application.

Each time symfony creates the tables in the database, all the data are lost. To populate the database with the initial data we could create a PHP script, or execute some SQL with the mysql program. But as the need is quite common, there is a better way with symfony: create YAML files in the data/fixtures/ directory and use the propel:data-load task to load them into the database:

# data/fixtures/010_categories.yml
JobeetCategory:
  design:        { name: Design }
  programming:   { name: Programming }
  manager:       { name: Manager }
  administrator: { name: Administrator }
 
# data/fixtures/020_jobs.yml
JobeetJob:
  job_sensio_labs:
    category_id:  programming
    type:         full-time
    company:      Sensio Labs
    logo:         /uploads/jobs/sensio_labs.png
    url:          http://www.sensiolabs.com/
    position:     Web Developer
    location:     Paris, France
    description:  |
      You've already developed websites with symfony and you want to work
      with Open-Source technologies. You have a minimum of 3 years
      experience in web development with PHP or Java and you wish to
      participate to development of Web 2.0 sites using the best
      frameworks available.
    how_to_apply: |
      Send your resume to fabien.potencier [at] sensio.com
    is_public:    true
    is_activated: true
    token:        job_sensio_labs
    email:        job@example.com
    expires_at:   2010-10-10
 
  job_extreme_sensio:
    category_id:  design
    type:         part-time
    company:      Extreme Sensio
    logo:         /uploads/jobs/extreme_sensio.png
    url:          http://www.extreme-sensio.com/
    position:     Web Designer
    location:     Paris, France
    description:  |
      Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do
      eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut
      enim ad minim veniam, quis nostrud exercitation ullamco laboris
      nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor
      in reprehenderit in.
 
      Voluptate velit esse cillum dolore eu fugiat nulla pariatur.
      Excepteur sint occaecat cupidatat non proident, sunt in culpa
      qui officia deserunt mollit anim id est laborum.
    how_to_apply: |
      Send your resume to fabien.potencier [at] sensio.com
    is_public:    true
    is_activated: true
    token:        job_extreme_sensio
    email:        job@example.com
    expires_at:   2010-10-10
 

A fixtures file is written in YAML, and defines model objects, labelled with a unique name. This label is of great use to link related objects without having to define primary keys (which are often auto-incremented and cannot be set). For instance, the job_sensio_labs job category is programming, which is the label given to the 'Programming' category.

A fixture file can contain object from one or several models.

Notice the numbers prefixing the filenames. This is a simple way to control the order of data loading. Later in the project, if we need to insert some new fixture file, it will be easy as we have some free numbers between existing ones.

In a fixture file, you don't need to define all columns values. If not, symfony will use the default value defined in the database schema. And as symfony uses Propel to load the data into the database, all the built-in behaviors (like setting the created_at or updated_at columns), or the custom behaviors you might have added to the model classes are activated.

Loading the initial data into the database is as simple as running the propel:data-load task:

$ php symfony propel:data-load

See it in Action in the Browser

We have used the command line interface a lot but that's not really exciting, especially for a web project. We now have everything we need to create Web pages that interact with the database.

Let's see how to display the list of jobs, how to edit an existing job, and how to delete a job. As explained during day 1, a symfony project is made of applications. Each application is made of modules. A module is a self-contained set of PHP code that represents a feature of the application (the API module for example), or a set of manipulations the user can do on a model object (a job module for example).

Symfony is able to automatically generate a module for a given model that provides basic manipulation features:

$ php symfony propel:generate-module --with-show --non-verbose-templates frontend job JobeetJob

The propel:generate-module generates a job module in the frontend application for the JobeetJob model. As with most symfony tasks, some files and directories have been created for you under the apps/frontend/modules/job directory:

The actions/actions.class.php file defines all the available actions for the job module:

You can now test the job module in a browser:

 http://jobeet.localhost/frontend_dev.php/job

If you try to edit a job, you will have an exception because symfony needs a text representation of a category. A PHP object representation can be defined with the PHP __toString() magic method. The text representation of a category record should be defined in the JobeetCategory model class:

// lib/model/JobeetCategory.php
class JobeetCategory extends BaseJobeetCategory
{
  public function __toString()
  {
    return $this->getName();
  }
}
 

Now each time symfony needs a text representation of a category, it calls the __toString() method which returns the category name. As we will need a text representation of all model classes at one point or another, let's define a __toString() method for every model class:

// lib/model/JobeetJob.php
class JobeetJob extends BaseJobeetJob
{
  public function __toString()
  {
    return sprintf('%s at %s (%s)', $this->getPosition(), $this->getCompany(), $this->getLocation());
  }
}
 
// lib/model/JobeetAffiliate.php
class JobeetAffiliate extends BaseJobeetAffiliate
{
  public function __toString()
  {
    return $this->getUrl();
  }
}
 

You can now create and edit jobs. Try to leave a required field blank, or try to enter an invalid date. That's right, symfony has created basic validation rules by introspecting the database schema.

See you Tomorrow

That's all for today. I have warned you in the introduction. Today, we have barely written PHP code but we have a working web module for the job model, ready to be tweaked and customized. Remember, no PHP code also means no bugs!

If you still have some energy left, feel free to read the generated code for the module and the model and try to understand how it works. It not, don't worry and sleep well, as tomorrow, we will talk about one of the most used paradigm in web frameworks, the MVC design pattern.

The code written today is available on the Jobeet SVN repository under the release_day_03 tag (http://svn.jobeet.org/tags/release_day_03/).