SplitStreetService.php

<?php  // php 7; declare(strict_types=1);
/**
 * For Dutch consignments the street should be divided into name, number and addition.
 *
 * If you want to improve this regex, please check the test.
 * And also adapt these adjustments to our GitHub:
 * https://github.com/myparcelnl/sdk
 *
 * @author      Reindert Vetter <[email protected]>
 * @copyright   MyParcel
 * @license     http://creativecommons.org/licenses/by-nc-nd/3.0/nl/deed.en_US  CC BY-NC-ND 3.0 NL
 * @since       Regex available since Release MyParcelNL/SDK v0.1.0
 */

namespace Shipment\Service\ExternalShipmentService;


/**
 * Class SplitStreetService
 *
 * @package Shipment\Service\ExternalShipmentService
 */
class SplitStreetService
{
    /**
     * Regular expression used to split street name from house number.
     *
     * For supported addresses go to:
     * @link test/Unit/Shipment/Service/SplitStreetServiceTest.php
     */
    const SPLIT_STREET_REGEX = '~(?P<street>.*?)'.     // The rest belongs to the street
                               '\s?'.                  // Separator between street and number
                               '(?P<number>\d{1,4})'.  // Number can contain a maximum of 4 numbers
                               '[/\s-]{0,2}'.          // Separators between number and addition
                               '(?P<number_suffix>'.
                                   '[\D]{0,4}|'.       // Number suffix has up to 4 letters or
                                   '\w{1}\d{1,3}|'.    // starts with a letter followed by numbers or
                                   '\d{2}\w{1,2}'.     // starts with 2 numbers followed by letters
                               ')$~';

    /**
     * @param string $country
     * @param string $fullStreet
     *
     * @return array
     * @throws SystemException
     */
    public function getStreetData(string $country, string $fullStreet) // php 7; : array
    {
        if ($country === 'NL') {
            return $this->getDutchStreetData($fullStreet);
        }

        return [
            'street' => $fullStreet,
            'number' => '',
            'number_suffix' => '',
        ];
    }

    /**
     * Splits street data into separate parts for street name, house number and extension.
     * Only for Dutch addresses
     *
     * @param string $fullStreet The full street name including all parts
     *
     * @return array
     * @throws SystemException
     */
    public function getDutchStreetData(string $fullStreet) // php 7; : array
    {
        $result = preg_match(self::SPLIT_STREET_REGEX, $fullStreet, $matches);

        if (!$result || !is_array($matches)) {
            throw new \Exception('This address can not be split. Please check address: ' . $fullStreet);
        }

        if ($fullStreet != $matches[0]) {
            // Characters are gone by preg_match
            throw new \Exception(
                "Something went wrong with splitting up address: $fullStreet. " .
                "See split_street.log for more information"
            );
        }

        return [
            'street' => $matches['street'],
            'number' => $matches['number'],
            'number_suffix' => trim($matches['number_suffix']),
        ];
    }
}

SplitStreetServiceTest.php

<?php  // php 7; declare(strict_types=1);
/**
 * For Dutch consignments the street should be divided into name, number and addition. This code tests whether the
 * street is split properly.
 *
 * If you want to add improvements, please create a fork in our GitHub:
 * https://github.com/myparcelnl/sdk
 *
 * @author      Reindert Vetter <[email protected]>
 * @copyright   MyParcel
 * @license     http://creativecommons.org/licenses/by-nc-nd/3.0/nl/deed.en_US  CC BY-NC-ND 3.0 NL
 * @since       File available since Release MyParcelNL/SDK v0.1.0
 */

use Test\TestCase;


use Shipment\Service\ExternalShipmentService\SplitStreetService;

/**
 * Class SplitStreetServiceTest
 */
class SplitStreetServiceTest extends TestCase
{
    /**
     * @dataProvider additionProvider()
     * @runInSeparateProcess
     */
    public function testGetStreetData($fullStreet, $street, $number, $numberSuffix, $country)
    {
        $consignment = (new SplitStreetService())->getStreetData($country, $fullStreet);

        self::assertSame($street, $consignment['street'], 'Street: ' . $fullStreet);
        self::assertSame($number, $consignment['number'], 'Number from: ' . $fullStreet);
        self::assertSame($numberSuffix, $consignment['number_suffix'], 'Number suffix from: ' . $fullStreet);
    }

    /**
     * Data for the test
     *
     * @return array
     */
    public function additionProvider() // php 7; : array
    {
        return [
            [
                'full_street'   => '3b',
                'street'        => '',
                'number'        => '3',
                'number_suffix' => 'b',
                'country'       => 'NL',
            ],
            [
                'full_street'   => 'Plein 1940-45 3b',
                'street'        => 'Plein 1940-45',
                'number'        => '3',
                'number_suffix' => 'b',
                'country'       => 'NL',
            ],
            [
                'full_street'   => '300 laan 3',
                'street'        => '300 laan',
                'number'        => '3',
                'number_suffix' => '',
                'country'       => 'NL',
            ],
            [
                'full_street'   => 'A.B.C. street 12',
                'street'        => 'A.B.C. street',
                'number'        => '12',
                'number_suffix' => '',
                'country'       => 'NL',
            ],
            [
                'full_street'   => 'street street 269-133',
                'street'        => 'street street',
                'number'        => '269',
                'number_suffix' => '133',
                'country'       => 'NL',
            ],
            [
                'full_street'   => 'Abeelstreet H10',
                'street'        => 'Abeelstreet H',
                'number'        => '10',
                'number_suffix' => '',
                'country'       => 'NL',
            ],
            [
                'full_street'   => 'street street 269-1001',
                'street'        => 'street street',
                'number'        => '269',
                'number_suffix' => '1001',
                'country'       => 'NL',
            ],
            [
                'full_street'   => 'Meijhorst 50e 26',
                'street'        => 'Meijhorst 50e',
                'number'        => '26',
                'number_suffix' => '',
                'country'       => 'NL',
            ],
            [
                'full_street'   => 'street street 12 ZW',
                'street'        => 'street street',
                'number'        => '12',
                'number_suffix' => 'ZW',
                'country'       => 'NL',
            ],
            [
                'full_street'   => 'street 12',
                'street'        => 'street',
                'number'        => '12',
                'number_suffix' => '',
                'country'       => 'NL',
            ],
            [
                'full_street'   => 'Biltstreet 113 A BS',
                'street'        => 'Biltstreet',
                'number'        => '113',
                'number_suffix' => 'A BS',
                'country'       => 'NL',
            ],
            [
                'full_street'   => 'Zonegge 23 12',
                'street'        => 'Zonegge',
                'number'        => '23',
                'number_suffix' => '12',
                'country'       => 'NL',
            ],
            [
                'full_street'   => 'Markerkant 10 142',
                'street'        => 'Markerkant',
                'number'        => '10',
                'number_suffix' => '142',
                'country'       => 'NL',
            ],
            [
                'full_street'   => 'Markerkant 10 11e',
                'street'        => 'Markerkant',
                'number'        => '10',
                'number_suffix' => '11e',
                'country'       => 'NL',
            ],
            [
                'full_street'   => 'Sir Winston Churchillln 283 F008',
                'street'        => 'Sir Winston Churchillln',
                'number'        => '283',
                'number_suffix' => 'F008',
                'country'       => 'NL',
            ],
            [
                'full_street'   => 'Sir Winston Churchilllaan 283 59',
                'street'        => 'Sir Winston Churchilllaan',
                'number'        => '283',
                'number_suffix' => '59',
                'country'       => 'NL',
            ],
            [
                'full_street'   => 'Insulindestreet 69 B03',
                'street'        => 'Insulindestreet',
                'number'        => '69',
                'number_suffix' => 'B03',
                'country'       => 'NL',
            ],
            [
                'full_street'   => 'Insulindestreet 69 B3',
                'street'        => 'Insulindestreet',
                'number'        => '69',
                'number_suffix' => 'B3',
                'country'       => 'NL',
            ],
            [
                'full_street'   => 'Ir. Mr. Dr. van Waterschoot van der Grachtstraat in Heerlen 6',
                'street'        => 'Ir. Mr. Dr. van Waterschoot van der Grachtstraat in Heerlen',
                'number'        => '6',
                'number_suffix' => '',
                'country'       => 'NL',
            ],
            [
                'full_street'   => 'Scheepvaartlaan 34/302',
                'street'        => 'Scheepvaartlaan',
                'number'        => '34',
                'number_suffix' => '302',
                'country'       => 'NL',
            ],
            [
                'full_street'   => 'oan e dijk 48',
                'street'        => 'oan e dijk',
                'number'        => '48',
                'number_suffix' => '',
                'country'       => 'NL',
            ],
            [
                'full_street'   => 'Vlinderveen137',
                'street'        => 'Vlinderveen',
                'number'        => '137',
                'number_suffix' => '',
                'country'       => 'NL',
            ],
            [
                'full_street'   => 'street 39-1hg',
                'street'        => 'street 39-',
                'number'        => '1',
                'number_suffix' => 'hg',
                'country'       => 'NL',
            ],
            [
                'full_street'   => 'Nicolaas Ruyschstraat 8 02L',
                'street'        => 'Nicolaas Ruyschstraat',
                'number'        => '8',
                'number_suffix' => '02L',
                'country'       => 'NL',
            ],
            [
                'full_street'   => 'Landsdijk 49 -A',
                'street'        => 'Landsdijk',
                'number'        => '49',
                'number_suffix' => 'A',
                'country'       => 'NL',
            ],
            [
                'full_street'   => 'No. 7 street',
                'street'        => 'No. 7 street',
                'number'        => '',
                'number_suffix' => '',
                'country'       => 'FR',
            ],
        ];
    }
}