This tools helps you to verify the quality of your migrated data.
composer require --dev axelerant/mdtt
Basically you follow these steps:
- Specify test specifications
- Specify test definitions
- Optionally, specify transform plugins
- Run the tests
You can find the basic template for the tool usage here.
Specify the test specification inside the directory tests/mdtt
. The specification must be written inside the spec.php
. Below is a sample specification:
<?php
return [
'databases' => [
// Source database credentials
'source_db' => [
'database' => "db_drupal7",
'username' => "db",
'password' => "db",
'host' => "127.0.0.1",
'port' => "59002",
],
// Destination database credentials
"destination_db" => [
'database' => "db",
'username' => "db",
'password' => "db",
'host' => "127.0.0.1",
'port' => "59002",
],
],
'http' => [
// JSON datasource basic authentication credetials
'source' => [
'username' => 'username',
'password' => 'fubar',
// Basic authentication protocol: basic, digest, or ntlm.
// Protocol is optional. If not mentioned, basic is considered.
'protocol' => 'ntlm',
],
'destination' => [
'username' => 'username',
'password' => 'puffball',
]
]
];
The test definitions are written in yaml
format, and must be present inside the directory tests/mdtt
. Let's say that you want to validate the migrated user data. Create a file called validate_users.yml
. The name of the file doesn't matter.
# Test definition ID
id: validate_users
# Test definition description
description: Validates users
# Group to which this test belongs to
group: migration_validation
# The query used for generating the source dataset
source:
type: database
data: "select * from users"
# The source database credentials
database: source_db
# The query used for generating the destination dataset
destination:
type: database
data: "select * from users_field_data"
# The destination database credentials
database: destination_db
# Test cases
tests:
-
# Field from source datasource used in this comparison
sourceField: name
# Field from destination datasource used in this comparison
destinationField: name
-
sourceField: mail
destinationField: mail
Consider that the data is exposed by the source legacy system in the following format:
{
"jsonapi": {
"version": 1.2,
},
"results": [
{
"employee": {
"id": 3453,
"info": {
"name": "Aloy",
"email": "aloy@nora.tribe",
"gender": "female",
"status": "active"
}
}
},
{
"employee": {
"id": 3452,
"info": {
"name": "Varl",
"email": "varl@nora.tribe",
"gender": "male",
"status": "active"
}
}
},
{
"employee": {
"id": 3450,
"info": {
"name": "Rost",
"email": "rost@nora.tribe",
"gender": "male",
"status": "inactive"
}
}
}
]
}
Consider that the destination new system exposes data in the following format:
"jsonapi": {
"version": 1.2,
},
"results": [
{
"id": 3453,
"name": "Aloy",
"email": "aloy@nora.tribe",
"gender": "female",
"status": "active"
},
{
"id": 3452,
"name": "Varl",
"email": "varl@nora.tribe",
"gender": "male",
"status": "active"
},
{
"id": 3450,
"name": "Rost",
"email": "rost@nora.tribe",
"gender": "male",
"status": "active"
}
]
}
id: validate_public_apis
description: Validate public apis
group: migration_validation
# The endpoint that returns the source dataset.
source:
type: json
data: https://dev.legacy-system.com/api/v1/users
# The pointer where the list of items resides. Refer https://github.com/halaxa/json-machine#what-is-json-pointer-anyway for examples
selector: "/results"
# Datasource basic authentication credentials. Note that the key is same as mentioned in the test specification. This is optional if the endpoint is publicly accessible.
credential: source
# The endpoint that returns the destination dataset.
destination:
type: json
data: https://dev.new-system.com/api/v1/users
selector: "/results"
credential: destination
tests:
-
# `/` separated field which you want to test.
# This is applicable when the value you want to test is in nested format.
sourceField: "employee/info/name"
destinationField: name
-
sourceField: "employee/info/email"
destinationField: email
There could be a scenario where instead of directly storing the data from source, it must be transformed in some way (say, whitespaces must be stripped) before storing it in the destination database. A QA engineer can write their own plugin, to validate the business logic that does the transformation.
The test case would look like this:
tests:
-
sourceField: name
transform: trim
destinationField: name
The QA engineer must specify the plugin class inside tests/mdtt/src/Plugin/Transform
. The file name (and, class name) must be same as the plugin name mentioned in the test case with the first character in upper case, i.e. Trim
. The plugin class must implement \Mdtt\Transform\Transform
interface.
<?php
// File location: tests/mdtt/src/Plugin/Transform/Trim.php.
class Trim implements \Mdtt\Transform\Transform
{
/**
* @inheritDoc
*/
public function name(): string
{
// Specifies the plugin name.
// This must be unique, if you have multiple plugin definitions.
// The same plugin name must be mentioned in the test case.
return "trim";
}
/**
* @inheritDoc
*/
public function process(mixed $data): mixed
{
// The logic that does the transformation.
return trim($data);
}
}
./vendor/bin/mdtt run
./vendor/bin/mdtt run -vvv
You can view all available options by doing this:
./vendor/bin/mdtt run --help
./vendor/bin/mdtt run --email foo@bar.mail
By default, the log files are created inside directory called logs
. The directory will be present in the root of the test suite. The log files are named based on the datetime when the test is executed.
- Database (MySQL)
- JSON
- Database (MySQL)
- JSON
Fork this repository and do the changes.
composer test