myep/src/BusProNet/XmlParser/InsuranceParser.php

<?php

declare(strict_types=1);

namespace App\BusProNet\XmlParser;

use App\BusProNet\Model\Insurance;
use Symfony\Component\DomCrawler\Crawler;

/**
 * Parses insurance XML data into Insurance model objects.
 *
 * Handles both individual insurance products (versicherung) and insurance packages
 * (versicherungspaket) with attribute-based parsing, type conversion, and filtering logic.
 */
class InsuranceParser extends AbstractParser
{
    /**
     * Parses insurance XML nodes into Insurance objects.
     *
     * @param Crawler $xmlContent The XML crawler containing insurance data
     *
     * @return array<string|int, Insurance> Array of Insurance objects indexed by id
     */
    public function parse(Crawler $xmlContent): array
    {
        $insurances = [];

        // First pass: collect referenced insurance IDs from packages
        $referencedIds = $this->collectReferencedInsuranceIds($xmlContent);

        // Second pass: Parse individual insurances with conditional filtering
        $individualInsurances = [];
        $xmlContent->filterXPath('//versicherungen/versicherung')
            ->each(function (Crawler $node) use (&$individualInsurances, &$insurances) {
                $isComplementary = $this->getBoolAttributeValue($node->attr('zusatzversicherung'));

                // Parse all individual insurances for reference lookup
                $insurance = $this->parseInsuranceNode($node, false);
                if (null !== $insurance && null !== $insurance->id) {
                    // Set complementary flag from XML attribute
                    $insurance->complementary = $isComplementary;

                    $individualInsurances[$insurance->id] = $insurance;

                    // Include ALL insurances in the result array
                    // Complementary insurances will be filtered out at the form field level
                    $insurances[$insurance->id] = $insurance;
                }
            });

        // Third pass: Parse packages with family detection based on contained insurances
        $xmlContent->filterXPath('//versicherungspakete/versicherungspaket')
            ->each(function (Crawler $node) use (&$insurances, $individualInsurances) {
                $insurance = $this->parseInsuranceNode($node, true, $individualInsurances);
                if (null !== $insurance && null !== $insurance->id) {
                    // Parse contained insurance IDs
                    $insurance->containedInsuranceIds = $this->parseContainedInsuranceIds($node);

                    // Populate containedInsurances with actual Insurance objects
                    $insurance->containedInsurances = [];
                    foreach ($insurance->containedInsuranceIds as $containedId) {
                        if (isset($individualInsurances[$containedId])) {
                            $insurance->containedInsurances[] = $individualInsurances[$containedId];
                        }
                    }

                    $insurances[$insurance->id] = $insurance;
                }
            });

        return $insurances;
    }

    /**
     * Determines if a package contains any family insurances.
     *
     * @param Crawler               $packageNode          The package XML node
     * @param array<int, Insurance> $individualInsurances Parsed individual insurances for reference
     *
     * @return bool True if any contained insurance is a family insurance
     */
    private function isPackageFamilyInsurance(Crawler $packageNode, array $individualInsurances): bool
    {
        // Check if any contained insurance is a family insurance
        $containedInsuranceIds = $this->parseContainedInsuranceIds($packageNode);
        foreach ($containedInsuranceIds as $containedId) {
            if (isset($individualInsurances[$containedId]) && true === $individualInsurances[$containedId]->familyInsurance) {
                return true;
            }
        }

        return false;
    }

    /**
     * Parses a single insurance or package node.
     *
     * @param Crawler               $node                 The XML node to parse
     * @param bool                  $isPackage            Whether this is a package node
     * @param array<int, Insurance> $individualInsurances Individual insurances for package family detection
     *
     * @return Insurance|null The parsed insurance object
     */
    private function parseInsuranceNode(Crawler $node, bool $isPackage, array $individualInsurances = []): ?Insurance
    {
        $insurance = new Insurance();
        $insurance->package = $isPackage;

        // Basic identifiers - packages have string IDs with 'P' prefix, individual insurances have int IDs
        $idValue = $node->attr('idbuspro');
        $insurance->id = $idValue;
        $insurance->code = $node->attr('code');
        $insurance->label = $this->normalizeInsuranceLabel($node->attr('bezeichnung'));
        $insurance->subType = $node->attr('unterart');
        $insurance->price = $node->attr('preis') ?
            $this->stringToFloat($node->attr('preis')) : null;
        $insurance->familyInsurance = $this->stringToBool($node->attr('familienversicherung'));

        // For packages, check if any contained insurances are family insurances
        if ($isPackage && !$insurance->familyInsurance && !empty($individualInsurances)) {
            $insurance->familyInsurance = $this->isPackageFamilyInsurance($node, $individualInsurances);
        }

        // Date constraints
        $insurance->travelDateFrom = $node->attr('reisedatumvon') ?
            $this->stringToDate($node->attr('reisedatumvon')) : null;
        $insurance->travelDateTo = $node->attr('reisedatumbis') ?
            $this->stringToDate($node->attr('reisedatumbis')) : null;
        $insurance->bookingDateFrom = $node->attr('buchungdatumvon') ?
            $this->stringToDate($node->attr('buchungdatumvon')) : null;
        $insurance->bookingDateTo = $node->attr('buchungdatumbis') ?
            $this->stringToDate($node->attr('buchungdatumbis')) : null;

        // Age constraints
        $insurance->ageFrom = $node->attr('altervon') ? (int) $node->attr('altervon') : null;
        $insurance->ageTo = $node->attr('alterbis') ? (int) $node->attr('alterbis') : null;

        // Price constraints
        $insurance->travelPriceFrom = $node->attr('reisepreisvon') ?
            $this->stringToFloat($node->attr('reisepreisvon')) : null;
        $insurance->travelPriceTo = $node->attr('reisepreisbis') ?
            $this->stringToFloat($node->attr('reisepreisbis')) : null;

        // Duration constraints
        $insurance->travelDurationFrom = $node->attr('reisedauervon') ? (int) $node->attr('reisedauervon') : null;
        $insurance->travelDurationTo = $node->attr('reisedauerbis') ? (int) $node->attr('reisedauerbis') : null;

        // Information URLs
        $insurance->urlInfo = $node->attr('urlinfo');
        $insurance->urlProductInfo = $node->attr('urlproduktinfo');
        $insurance->urlTerms = $node->attr('urlagb');

        return $insurance;
    }

    /**
     * Collects all insurance IDs referenced by packages.
     *
     * @param Crawler $xmlContent The XML content to scan
     *
     * @return array<int> Array of referenced insurance IDs
     */
    private function collectReferencedInsuranceIds(Crawler $xmlContent): array
    {
        $referencedIds = [];
        $xmlContent->filterXPath('//versicherungspakete/versicherungspaket')
            ->each(function (Crawler $packageNode) use (&$referencedIds) {
                $packageNode->filterXPath('.//enthalteneversicherung')
                    ->each(function (Crawler $refNode) use (&$referencedIds) {
                        $referencedIds[] = $refNode->attr('idbuspro');
                    });
            });

        return array_unique($referencedIds);
    }

    /**
     * Parses contained insurance IDs from a package node.
     *
     * @param Crawler $node The package XML node
     *
     * @return array<int> Array of contained insurance IDs (always int, as they reference individual insurances)
     */
    private function parseContainedInsuranceIds(Crawler $node): array
    {
        $ids = [];
        $node->filterXPath('.//enthalteneversicherungen/enthalteneversicherung')
            ->each(function (Crawler $insuranceNode) use (&$ids) {
                $ids[] = $insuranceNode->attr('idbuspro'); // Individual insurance IDs are always int
            });

        return $ids;
    }

    /**
     * Normalizes insurance labels by removing redundant phrases and replacing abbreviations.
     *
     * Transformations:
     * - Removes "Auto/Bahn/Bus (Europa)" - redundant travel mode specification
     * - Replaces "FAM" with "Familie" - clearer German terminology
     *
     * @param string|null $label The original insurance label from XML
     *
     * @return string|null The normalized label, or null if input was null
     */
    private function normalizeInsuranceLabel(?string $label): ?string
    {
        if (null === $label) {
            return null;
        }

        // Remove redundant travel mode specification
        $label = str_replace('Auto/Bahn/Bus (Europa)', '', $label);

        // Replace abbreviation with full German word
        $label = str_replace('FAM', 'Familie', $label);

        // Clean up extra spaces that may result from removals
        $label = preg_replace('/\s+/', ' ', $label);
        $label = trim($label);

        return $label;
    }

    private function getBoolAttributeValue(?string $value): bool
    {
        return !empty($value) && $this->stringToBool($value);
    }
}