Day 3: The Data Model

Those of you itching to open your text editor and lay down some PHP will be happy to know today 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 the 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 saw 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 Doctrine.

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.

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/doctrine/schema.yml file:

# config/doctrine/schema.yml
JobeetCategory:
  actAs: { Timestampable: ~ }
  columns:
    name: { type: string(255), notnull: true, unique: true }
 
JobeetJob:
  actAs: { Timestampable: ~ }
  columns:
    category_id:  { type: integer, notnull: true }
    type:         { type: string(255) }
    company:      { type: string(255), notnull: true }
    logo:         { type: string(255) }
    url:          { type: string(255) }
    position:     { type: string(255), notnull: true }
    location:     { type: string(255), notnull: true }
    description:  { type: string(4000), notnull: true }
    how_to_apply: { type: string(4000), notnull: true }
    token:        { type: string(255), notnull: true, unique: true }
    is_public:    { type: boolean, notnull: true, default: 1 }
    is_activated: { type: boolean, notnull: true, default: 0 }
    email:        { type: string(255), notnull: true }
    expires_at:   { type: timestamp, notnull: true }
  relations:
    JobeetCategory: { onDelete: CASCADE, local: category_id, foreign: id, foreignAlias: JobeetJobs } 
 
JobeetAffiliate:
  actAs: { Timestampable: ~ }
  columns:
    url:       { type: string(255), notnull: true }
    email:     { type: string(255), notnull: true, unique: true }
    token:     { type: string(255), notnull: true }
    is_active: { type: boolean, notnull: true, default: 0 }
  relations:
    JobeetCategories:
      class: JobeetCategory
      refClass: JobeetCategoryAffiliate
      local: affiliate_id
      foreign: category_id
      foreignAlias: JobeetAffiliates
 
JobeetCategoryAffiliate:
  columns:
    category_id:  { type: integer, primary: true }
    affiliate_id: { type: integer, primary: true }
  relations:
    JobeetCategory:  { onDelete: CASCADE, local: category_id, foreign: id }
    JobeetAffiliate: { onDelete: CASCADE, local: affiliate_id, foreign: id }

$ php symfony doctrine:build-schema

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

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, integer, float, decimal, string, array, object, blob, clob, timestamp, time, date, enum, gzip)
• notnull: Set it to true if you want the column to be required
• unique: Set it to true if you want to create a unique index for the column.

The Database

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

Let's use MySQL for this tutorial:

$ mysqladmin -uroot -p create jobeet
Enter password: mYsEcret ## The password will echo as ********

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 need a password to access your database on the development server, just omit the third argument.

The ORM

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

First in order to generate the SQL you must build your models from your schema files.

$ php symfony doctrine:build --model

Now that your models are present you can generate and insert the SQL.

$ php symfony doctrine:build --sql

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

# snippet from data/sql/schema.sql
CREATE TABLE jobeet_category (id BIGINT AUTO_INCREMENT, name VARCHAR(255)
NOT NULL COMMENT 'test', created_at DATETIME, updated_at DATETIME, slug
VARCHAR(255), UNIQUE INDEX sluggable_idx (slug), PRIMARY KEY(id))
ENGINE = INNODB;

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

$ php symfony doctrine:insert-sql

$ php symfony help doctrine:insert-sql

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

$ php symfony doctrine:build --model

The doctrine: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 Doctrine generates three 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 doctrine:build --model, this class is overwritten, so all customizations must be done in the JobeetJob class.

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

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 doctrine: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 doctrine:build --all --no-confirmation

You will see validators in action today and forms will be explained in great details on day 10.

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 some initial data, we could create a PHP script, or execute some SQL statements 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 doctrine:data-load task to load them into the database.

First, create the following fixture files:

# data/fixtures/categories.yml
JobeetCategory:
  design:
    name: Design
  programming:
    name: Programming
  manager:
    name: Manager
  administrator:
    name: Administrator
 
# data/fixtures/jobs.yml
JobeetJob:
  job_sensio_labs:
    JobeetCategory: programming
    type:         full-time
    company:      Sensio Labs
    logo:         sensio-labs.gif
    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:
    JobeetCategory:  design
    type:         part-time
    company:      Extreme Sensio
    logo:         extreme-sensio.gif
    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 (for instance, we have defined two jobs labelled job_sensio_labs and job_extreme_sensio). 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.

Although a fixture file can contain objects from one or several models, we have decided to create one file per model for the Jobeet fixtures.

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 Doctrine to load the data into the database, all the built-in behaviors (like automatically setting the created_at or updated_at columns) and 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 doctrine:data-load task:

$ php symfony doctrine:data-load

Run the doctrine:build --all --and-load task to make sure everything is generated from your schema. This will generate your forms, filters, models, drop your database and re-create it with all the tables.

$ php symfony doctrine:build --all --and-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 the first day, a symfony project is made of applications. Each application is further divided into 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 doctrine:generate-module --with-show --non-verbose-templates frontend job JobeetJob

The doctrine: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 action for the job module:

You can now test the job module in a browser:

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

If you try to edit a job, you will notice the Category id drop down has a list of all the category names. The value of each option is gotten from the __toString() method.

Doctrine will try and provide a base __toString() method by guessing a descriptive column name like, title, name, subject, etc. If you want something custom then you will need to add your own __toString() methods like below. The JobeetCategory model is able to guess the __toString() method by using the name column of the jobeet_category table.

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

Final Thoughts

That's all. 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. If 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.