CSV Import in SilverStripe Duplicates and RelationCallbacks - php

Question

I need to understand the code below, specially how exactly $duplicateChecks and $relationCallbacks work but there is little explanation on the official documentation. Can somebody explain how these work or suggest some other documentation I can look at?
class PlayerCsvBulkLoader extends CsvBulkLoader {
public $columnMap = array(
'Number' => 'PlayerNumber',
'Name' => '->importFirstAndLastName',
'Birthday' => 'Birthday',
'Team' => 'Team.Title',
);
public $duplicateChecks = array(
'Number' => 'PlayerNumber'
);
public $relationCallbacks = array(
'Team.Title' => array(
'relationname' => 'Team',
'callback' => 'getTeamByTitle'
)
);
public static function importFirstAndLastName(&$obj, $val, $record) {
$parts = explode(' ', $val);
if(count($parts) != 2) return false;
$obj->FirstName = $parts[0];
$obj->LastName = $parts[1];
}
public static function getTeamByTitle(&$obj, $val, $record) {
return FootballTeam::get()->filter('Title', $val)->First();
}
}

Answer 1

$duplicateChecks is used by findExistingObject function in the CsvBulkLoader class. It is iterated over to find any object that has a column with the specified value. In that example, it checks the "PlayerNumber" column.
It can also be passed a callback like so:
public $duplicateCheck = array(
'Number' => array(
'callback' => 'checkPlayerNumberFunction'
)
);
The callback specified needs to either exist on an instance of the class specified on the property objectClass or on the CsvBulkLoader itself (which would happen if you extended it). These callbacks are used to do more complex duplicate lookups and return an existing object (if any) found.
$relationCallbacks on the other hand is used by the main processRecord function. The callback works in the same way as the $duplicateCheck callback, it needs to either exist on an instance of the class specified on the proeprty objectClass or on the CsvBulkLoader. These callbacks can return an object that will be related back to a specific object record (new or existing) as a has_one.
There is a little more to it than that though the best way to learn is by a bit of experimentation and jumping through the code of the class itself. I have linked to the various functions etc in my answer.