netgescon-day0/app/Services/Catalog/ProductAssetIngestionService.php

<?php
namespace App\Services\Catalog;

use App\Models\Fornitore;
use App\Models\Product;
use App\Models\ProductMedia;
use App\Support\ArchivioPaths;
use Illuminate\Support\Facades\Http;
use Illuminate\Support\Facades\Storage;
use Illuminate\Support\Str;

class ProductAssetIngestionService
{
    /**
     * @return array<string,int>
     */
    public function ingestProductAssets(Product $product, Fornitore $fornitore, bool $dryRun = false): array
    {
        $stats = [
            'images'              => 0,
            'documents'           => 0,
            'skipped_existing'    => 0,
            'skipped_unsupported' => 0,
            'failed'              => 0,
        ];

        $catalog = is_array($product->meta['catalog'] ?? null) ? $product->meta['catalog'] : [];
        $links   = is_array($catalog['private_source_links'] ?? null) ? $catalog['private_source_links'] : [];

        foreach ($this->normalizeLinkList($links['image_links'] ?? $links['image_link'] ?? null) as $index => $url) {
            $imageResult = $this->ingestSingleAsset($product, $fornitore, $url, 'image', $dryRun, $index);
            $stats[$imageResult['bucket']]++;
        }

        foreach ($this->normalizeLinkList($links['document_links'] ?? $links['document_link'] ?? null) as $index => $url) {
            $documentResult = $this->ingestSingleAsset($product, $fornitore, $url, 'document', $dryRun, $index);
            $stats[$documentResult['bucket']]++;
        }

        return $stats;
    }

    /**
     * @return array<string,int>
     */
    public function ingestForFornitore(Fornitore $fornitore, int $limit = 50, bool $dryRun = false): array
    {
        $stats = [
            'products'            => 0,
            'images'              => 0,
            'documents'           => 0,
            'skipped_existing'    => 0,
            'skipped_unsupported' => 0,
            'failed'              => 0,
        ];

        $products = Product::query()
            ->where('default_fornitore_id', (int) $fornitore->id)
            ->orderBy('id')
            ->limit(max(1, $limit))
            ->get(['id', 'internal_code', 'meta']);

        foreach ($products as $product) {
            $stats['products']++;
            $result = $this->ingestProductAssets($product, $fornitore, $dryRun);

            foreach ($result as $bucket => $count) {
                $stats[$bucket] += $count;
            }
        }

        return $stats;
    }

    /**
     * @return array{bucket:string}
     */
    private function ingestSingleAsset(Product $product, Fornitore $fornitore, string $url, string $mediaType, bool $dryRun, int $sortOrder = 0): array
    {
        $url = trim($url);
        if ($url === '' || ! str_starts_with($url, 'http')) {
            return ['bucket' => 'skipped_unsupported'];
        }

        $extension = $this->guessExtension($url, '', $mediaType);
        $folder    = $this->buildMediaFolder($fornitore, $product, $mediaType);
        $filename  = Str::slug((string) ($product->internal_code ?? ('product-' . $product->id))) . '-' . substr(sha1($url), 0, 12) . '.' . $extension;
        $path      = $folder . '/' . $filename;

        $existing = ProductMedia::query()
            ->where('product_id', (int) $product->id)
            ->where('media_type', $mediaType)
            ->whereIn('disk', ['local', 'public'])
            ->where('path', $path)
            ->exists();

        if ($existing) {
            return ['bucket' => 'skipped_existing'];
        }

        if ($dryRun) {
            return ['bucket' => $mediaType === 'image' ? 'images' : 'documents'];
        }

        try {
            $response = Http::timeout(20)
                ->withHeaders(['User-Agent' => 'NetgesconCatalogBot/1.0'])
                ->get($url);
        } catch (\Throwable) {
            return ['bucket' => 'failed'];
        }

        if (! $response->successful()) {
            return ['bucket' => 'failed'];
        }

        $mime = strtolower(trim((string) $response->header('Content-Type')));
        if ($mime !== '' && str_contains($mime, ';')) {
            $mime = trim((string) Str::before($mime, ';'));
        }

        if ($mediaType === 'image' && ! str_starts_with($mime, 'image/')) {
            return ['bucket' => 'skipped_unsupported'];
        }

        if ($mediaType === 'document' && ! str_contains($mime, 'pdf') && ! str_ends_with(Str::lower(parse_url($url, PHP_URL_PATH) ?: ''), '.pdf')) {
            return ['bucket' => 'skipped_unsupported'];
        }

        $extension = $this->guessExtension($url, $mime, $mediaType);
        $filename  = Str::slug((string) ($product->internal_code ?? ('product-' . $product->id))) . '-' . substr(sha1($url), 0, 12) . '.' . $extension;
        $path      = $folder . '/' . $filename;

        Storage::disk('public')->put($path, $response->body());

        ProductMedia::query()->create([
            'product_id' => (int) $product->id,
            'media_type' => $mediaType,
            'disk'       => 'public',
            'path'       => $path,
            'title'      => $mediaType === 'image' ? 'Immagine catalogo interna' : 'Scheda prodotto interna',
            'mime_type'  => $mime !== '' ? $mime : null,
            'size_bytes' => strlen($response->body()),
            'sort_order' => $sortOrder,
            'meta'       => [
                'source_url'    => $url,
                'source_system' => 'private_catalog_link',
                'internalized'  => true,
            ],
        ]);

        return ['bucket' => $mediaType === 'image' ? 'images' : 'documents'];
    }

    private function buildMediaFolder(Fornitore $fornitore, Product $product, string $mediaType): string
    {
        $base = ArchivioPaths::fornitoreBase($fornitore);
        if (! is_string($base) || $base === '') {
            $base = 'catalogo/fornitori/' . ($fornitore->codice_univoco ?: $fornitore->id);
        }

        return trim('catalogo-interno/' . trim($base, '/') . '/prodotti/' . ($product->internal_code ?: $product->id) . '/' . $mediaType, '/');
    }

    private function guessExtension(string $url, string $mime, string $mediaType): string
    {
        $path = Str::lower((string) (parse_url($url, PHP_URL_PATH) ?: ''));
        $ext  = pathinfo($path, PATHINFO_EXTENSION);
        if (is_string($ext) && $ext !== '') {
            return $ext;
        }

        return match (true) {
            str_contains($mime, 'png')  => 'png',
            str_contains($mime, 'webp') => 'webp',
            str_contains($mime, 'jpeg'), str_contains($mime, 'jpg') => 'jpg',
            $mediaType === 'document'   => 'pdf',
            default                     => 'bin',
        };
    }

    /**
     * @return array<int,string>
     */
    private function normalizeLinkList(mixed $links): array
    {
        if (is_array($links)) {
            return array_values(array_filter(array_map(
                static fn(mixed $value): string => trim((string) $value),
                $links,
            ), static fn(string $value): bool => $value !== ''));
        }

        $link = trim((string) $links);

        return $link !== '' ? [$link] : [];
    }
}