README

• Support multiple database connections
• Support multiple fixtures formats loading (for each connection)

Getting started

Installation

Run the following commands to install this package

composer require --dev hotel-quickly/nette-phpunit

Usage

In order to use Database test case, you must extend your base class from HQ\Test\AbstractDbTestCase and implement these 3 methods

1. getContainer - This method will return Nette\Di\Container instance.
2. getConnections - Return an array of HQ\Test\Connection\AbstractConnection(connectionName, baseSchema, ...) connections object.

• connectionName - A database connection name, which will be used as a prefix of fixture file.
• baseSchema - Database schema which includes CREATE TABLE ... sql statements.

1. getBaseFixtureDir - Return a base directory for searching base fixture file (see

use HQ\Test\AbstractDbTestCase;
use HQ\Test\Connection\NetteConnection;

class YourBaseDbTestCase extends AbstractDbTestCase
{
    protected function getContainer()
    {
        return require dirname(__DIR__) . '/app/bootstrap.php';
    }

    protected function getConnections()
    {
        return [
            new NetteConnection('default', __DIR__ . '/../tests/skeleton.sql', $this->container->getService('database.default.context'))
        ];
    }

    public function getBaseFixtureDir()
    {
        return __DIR__;
    }
}

Working with fixtures

Loading fixtures

By default, fixtures will be automatically loaded from 3 places, in the following orders

1. base fixtures
2. class fixtures
3. instance fixtures

1. base fixtures

This is a primary fixtures that will be loaded for all test cases (think of it as initial data for the whole apps)

test
  +- YourBaseDbTestCase.php
  +- <connection-name>-fixtures.{yaml, json, php} e.g. default-fixtures.json 

2. class fixtures

This will be loaded each time test case was invoked.

test
  +- User
    +- UserTest.php
    +- <connection-name>-fixtures.{yaml, json, php} e.g. default-fixtures.json

3. instance fixtures

This is pretty much the same as class fixtures, except that we define the fixture inside TestCase itself (so only php array is supported.

class MyTest extends MyBaseDbTestCase
{
    public function getFixtures()
    {
        return [
            '<connection-name>' => [
                ...
            ]
        ]
    }
}

####Fixture formats + structure This library currently supports 3 fixture formats e.g. json, yaml and php. All formats has the same fixture structure i.e.

[
  "<table-name>": [
     [
        "<field-name1>": "<value1>",
        "<field-name2>": "<value2>"
     ],
     ...
   ]
]

json example

{
    "user": [
        {"username": "john"},
        {"username": "jane"}
    ]
}

yaml example

user:
  -
    username: john
  -
    username: jane

php example - NOTE - Don't forget to put return statement

return [
    'user' => [
        ['username' => 'john'],
        ['username' => 'jane'],
    ]
]

Improving testing speed

Due to the overhead of creating database schema at the beginning of the tests (only once). This might slow down your test suites speed. You can avoid this problem by exporting UNITTEST_CREATE_SCHEMA=false env and then run unit test as usual. This will make it skips database schema create overhead.

Common problems

• InvalidArgumentException: There is already a table named account with different table definition.

This error indicates that your fixtures for same dataset (table e.g. user) has different column structure. For example, you have default-fixtures.yaml that contains user table and it has the following fields

user:
  -
    username: john

But somewhere in other fixtures, you have same user table with different fields e.g.

user:
  -
    email: john

This will cause the above error. To fix this, just make sure you define the same fields on each fixtures.

• ErrorException: ini_set(): Cannot change zlib.output_compression - headers already sent

This happens when Nette has php ini zlib.output_compression enabled in the container. Removing this config during testing will fix this error