cbr-1.0.0/src/Plugin/Field/FieldType/CBRFieldHelper.php

<?php

namespace Drupal\cbr\Plugin\Field\FieldType;

use Drupal\Core\Cache\Cache;
use Drupal\Core\Form\FormStateInterface;
use Drupal\field\Entity\FieldConfig;

class CBRFieldHelper
{
    /**
     * Add a form for setting the fields weight.
     * @param array $form The form array.
     * @param FormStateInterface $form_state The form state.
     * @return array The form array.
     */
    public static function cbrFieldSettingsForm(array $form, FormStateInterface $form_state): array
    {
        //unset default values, added later by drupal anyway
        unset($form['label']);
        unset($form['description']);
        unset($form['required']);

        /** @var FieldConfig $field_config */
        $field_config = $form_state->getFormObject()->getEntity();
        $form['cbr_settings'] = [
            '#type' => 'details',
            '#title' => t('Case Based Reasoning'),
            '#open' => true,
            '#weight' => 99
        ];
        $form['cbr_settings']['weight'] = [
            '#type' => 'number',
            '#step' => '.001',
            '#title' => t('Attribute Weight'),
            '#default_value' => $field_config->getThirdPartySetting('cbr', 'weight', 1),
            '#required' => true,
            '#description' => t('Attribute weight for the calculation of the similarity. Can be changed, but you need to recalculate all saved similarity factors!'),
        ];
        $form['#entity_builders'][] = [static::class, 'saveCBRFieldSettings'];
        return $form;
    }

    /**
     * Save the weight to the field config.
     * @param $entity_type The entity type.
     * @param FieldConfig $field_config The field config.
     * @param $form The form array.
     * @param FormStateInterface $form_state The form state.
     */
    public static function saveCBRFieldSettings($entity_type, FieldConfig $fieldConfig, &$form, FormStateInterface $form_state)
    {
        $fieldConfig->setThirdPartySetting('cbr', 'weight', $form_state->getValue(['settings', 'cbr_settings', 'weight']));
    }

    public static function stringFieldSettingsForm(array $form, FormStateInterface $form_state): array
    {
        /** @var FieldConfig $field_config */
        $field_config = $form_state->getFormObject()->getEntity();
        $form['cbr_settings']['similarity_function'] = [
            '#type' => 'select',
            '#title' => t('Similarity function'),
            '#description' => t('Select the similarity function to use for this field.'),
            '#options' => [
                'cosine' => t('Cosine'),
                'jaccard' => t('Jaccard')
            ],
            '#default_value' => $field_config->getThirdPartySetting('cbr', 'similarity_function', 'cosine'),
            '#required' => true
        ];
        $form['#entity_builders'][] = [static::class, 'saveCBRFieldStringSettings'];
        return $form;
    }

    /**
     * Save the selected string similarity method to the field config.
     * @param $entity_type The entity type.
     * @param FieldConfig $field_config The field config.
     * @param $form The form array.
     * @param FormStateInterface $form_state The form state.
     */
    public static function saveCBRFieldStringSettings($entity_type, FieldConfig $fieldConfig, &$form, FormStateInterface $form_state)
    {
        $fieldConfig->setThirdPartySetting('cbr', 'similarity_function',  $form_state->getValue(['settings', 'cbr_settings', 'similarity_function']));
    }

    public static function calculateSimilarityBetweenNumericValues(float $value1, float $value2, ?float $min, ?float $max): float
    {
        if (is_numeric($min) && is_numeric($max) && ($max - $min) > 0) {
            return 1 - (abs($value1 - $value2) / ($max - $min));
        } else {
            return 1 / (abs($value1 - $value2) + 1);
        }
    }

    public static function calculateSimilarityBetweenStringValues(string $value1, string $value2, string $similarity_function): float
    {
        switch ($similarity_function) {
            case 'cosine':
                return self::calculateCosineSimilarity($value1, $value2);
            case 'jaccard':
                return self::calculateJaccardSimilarity($value1, $value2);
            default:
                return 0;
        }
    }

    private static function calculateCosineSimilarity($value1, $value2): float
    {
        //Calculate the cosine similarity between two strings.
        $words1 = self::stringToArray($value1);
        $words2 = self::stringToArray($value2);
        $union = array_unique(array_merge($words1, $words2));

        $number_of_words1 = array_count_values($words1);
        $number_of_words2 = array_count_values($words2);

        $vector1 = [];
        $vector2 = [];
        foreach ($union as $word) {
            $vector1[] = $number_of_words1[$word] ?? 0;
            $vector2[] = $number_of_words2[$word] ?? 0;
        }
        return self::dotProduct($vector1, $vector2) / (self::magnitude($vector1) * self::magnitude($vector2));
    }

    private static function dotProduct(array $vector1, array $vector2): float
    {
        return array_sum(array_map(function ($a, $b) {
            return $a * $b;
        }, $vector1, $vector2));
    }

    private static function magnitude(array $vector): float
    {
        return sqrt(array_sum(array_map(function ($a) {
            return $a ** 2;
        }, $vector)));
    }

    private static function calculateJaccardSimilarity($value1, $value2): float
    {
        //Calculate the jaccard similarity between two strings.
        $words1 = self::stringToArray($value1);
        $words2 = self::stringToArray($value2);
        $words1 = array_unique($words1);
        $words2 = array_unique($words2);
        $intersection = array_intersect($words1, $words2);
        $union = array_unique(array_merge($words1, $words2));
        $size = count($union);
        if ($size == 0) {
            return 0;
        }
        return count($intersection) / $size;
    }

    private static function stringToArray(string $string): array
    {
        $string = strip_tags($string);
        $string = strtolower($string);
        $string = preg_replace('/[^a-zA-Z0-9äöüÄÖÜß\s]/', '', $string);
        $string = preg_replace('/\s+/', ' ', $string);
        return explode(' ', $string);
    }

    public static function summerizeNumericValues(array $values): float
    {
        //use median for numeric values
        if ($values) {
            $count = count($values);
            sort($values);
            $mid = floor(($count - 1) / 2);
            return ($values[$mid] + $values[$mid + 1 - $count % 2]) / 2;
        }
        return 0;
    }

    public static function summarizeStringValues(array $fields): string
    {
        return join('|', $fields);
    }
}