Backup #1818

Dateiinhalt

<?php

declare(strict_types=1);

namespace UseCases\Taxonomy;

// @responsibility: Taxonomie-Übersicht mit aggregierten Statistiken

use Domain\Constants;
use Domain\Repository\ChunkTaxonomyRepositoryInterface;
use Domain\Repository\EntityTaxonomyRepositoryInterface;
use Domain\Repository\TaxonomyRepositoryInterface;

final class GetTaxonomyOverviewUseCase
{
    public function __construct(
        private ChunkTaxonomyRepositoryInterface $chunkTaxonomyRepository,
        private EntityTaxonomyRepositoryInterface $entityTaxonomyRepository,
        private TaxonomyRepositoryInterface $taxonomyRepository
    ) {
    }

    /**
     * Get complete taxonomy overview with statistics.
     *
     * @return array{
     *     taxonomy_stats: array{total_terms: int, root_terms: int, max_depth: int, tagged_chunks: int},
     *     chunk_coverage: array{total: int, mapped: int, percentage: float},
     *     entity_validation: array{total: int, validated: int, pending: int},
     *     term_usage: array<int, int>
     * }
     */
    public function execute(): array
    {
        $taxonomyStats = $this->taxonomyRepository->getStats();
        $chunkCoverage = $this->calculateChunkCoverage();
        $entityValidation = $this->entityTaxonomyRepository->getValidationStats();
        $termUsage = $this->chunkTaxonomyRepository->countByTaxonomyTerm();

        return [
            'taxonomy_stats' => $taxonomyStats,
            'chunk_coverage' => $chunkCoverage,
            'entity_validation' => $entityValidation,
            'term_usage' => $termUsage,
        ];
    }

    /**
     * Get chunks that have no taxonomy mapping.
     *
     * @return array<array{id: int, document_id: int, content: string}>
     */
    public function getUnmappedChunks(int $limit = 100): array
    {
        return $this->chunkTaxonomyRepository->getUnmappedChunks($limit);
    }

    /**
     * Get entity mappings that need validation.
     *
     * @return array<\Domain\Entity\EntityTaxonomyMapping>
     */
    public function getUnvalidatedEntityMappings(int $limit = 100): array
    {
        return $this->entityTaxonomyRepository->getUnvalidatedMappings($limit);
    }

    /**
     * Get usage statistics for a specific taxonomy term.
     *
     * @return array{
     *     term: array|null,
     *     chunk_count: int,
     *     entity_count: int,
     *     chunks: array<array{chunk_id: int, confidence: float, source: string}>,
     *     entities: array<array{entity_id: int, relevance: float, validated: bool}>
     * }
     */
    public function getTaxonomyTermUsage(int $termId): array
    {
        $term = $this->taxonomyRepository->find($termId);
        $chunks = $this->chunkTaxonomyRepository->findByTaxonomyTermId($termId);
        $entities = $this->entityTaxonomyRepository->findByTaxonomyTermId($termId);

        return [
            'term' => $term,
            'chunk_count' => count($chunks),
            'entity_count' => count($entities),
            'chunks' => $chunks,
            'entities' => $entities,
        ];
    }

    /**
     * Get all taxonomy terms with their usage counts.
     *
     * @return array<array{id: int, name: string, path: string, chunk_count: int, entity_count: int}>
     */
    public function getTermsWithUsage(): array
    {
        $terms = $this->taxonomyRepository->findAll();
        $chunkCounts = $this->chunkTaxonomyRepository->countByTaxonomyTerm();

        $result = [];
        foreach ($terms as $term) {
            $termId = $term['id'];
            $entityMappings = $this->entityTaxonomyRepository->findByTaxonomyTermId($termId);

            $result[] = [
                'id' => $termId,
                'name' => $term['name'],
                'path' => $term['path'],
                'chunk_count' => $chunkCounts[$termId] ?? 0,
                'entity_count' => count($entityMappings),
            ];
        }

        return $result;
    }

    /**
     * Calculate chunk coverage statistics.
     *
     * @return array{total: int, mapped: int, percentage: float}
     */
    private function calculateChunkCoverage(): array
    {
        $taxonomyStats = $this->taxonomyRepository->getStats();
        $mappedChunks = $taxonomyStats['tagged_chunks'];

        // Get total chunks from unmapped query (limit 0 would give us count)
        // For efficiency, we estimate based on the stats
        $unmappedSample = $this->chunkTaxonomyRepository->getUnmappedChunks(1);
        $hasUnmapped = count($unmappedSample) > 0;

        // Use a heuristic: if there are unmapped chunks, estimate total
        // In production, this would be a dedicated count query
        $total = $hasUnmapped ? max($mappedChunks + 1, $mappedChunks) : $mappedChunks;

        $percentage = $total > 0 ? round(($mappedChunks / $total) * 100, 2) : 100.0;

        return [
            'total' => $total,
            'mapped' => $mappedChunks,
            'percentage' => $percentage,
        ];
    }
}

Vollständig herunterladen

Aktionen

Andere Versionen dieser Datei

← Zurück zur Übersicht