<?php

declare(strict_types=1);

namespace TFGraphics\SmartImport\Search;

use TFGraphics\SmartImport\Domain\DTO\{ProductDTO, AttributeDTO, CategoryDTO, SeoDTO};

/** Local, repeatable normalization. Never searches or changes physical identity. */
final class ProductEvidenceNormalizer
{
    public static function isHardware(string $text): bool
    {
        return preg_match('/\b(?:(?:dvd|blu[ -]?ray|cd)[ -]*(?:players?|spelers?|lecteurs?)|(?:lettor[ei]|lecteurs?)[ -]+(?:dvd|blu[ -]?ray|cd)|toner|telecomand[oi]|remote control|afstandsbediening|game consoles?|masterizzator[ei]|blank (?:dvds?|cds?|discs?|blu[ -]?ray(?: discs?)?)|(?:dvd|cd)[ -]?(?:r|rw)|(?:cd|dvd) vergini|custodi[ae]|disc cases?)\b/iu', $text) === 1;
    }

    /** Exact structured category only: never infer a film from a loose DVD mention. */
    public static function hasVideoCategory(ProductDTO $product): bool
    {
        $values = array_map(static fn ($c): string => $c->getName(), $product->getCategories());
        foreach ($product->getAttributes() as $attribute) {
            if (in_array(strtolower($attribute->getName()), ['category', 'source_category', 'google_category'], true)) {
                $values[] = $attribute->getValue();
            }
        }
        foreach ($values as $value) {
            if (preg_match('/^(?:dvd|dvd video|blu[ -]?ray(?: 4k(?: uhd)?)?|vhs|super ?8|movies|films?|film e dvd|movies & tv)$/iu', trim($value))) { return true; }
        }
        return false;
    }

    private static function hasMusicCategory(ProductDTO $product): bool
    {
        $values = array_map(static fn ($c): string => $c->getName(), $product->getCategories());
        foreach ($product->getAttributes() as $a) {
            if (in_array(strtolower($a->getName()), ['category', 'source_category', 'google_category'], true)) { $values[] = $a->getValue(); }
        }
        foreach ($values as $value) {
            if (preg_match('/^(?:cd|cd audio|audio cd|vinile|vinyl|audio cassette|audiocassetta)$/iu', trim($value))) { return true; }
        }
        return false;
    }

    public function normalize(ProductDTO $p): ProductDTO
    {
        // An exact identity-only source must not acquire fields from title heuristics.
        if (\TFGraphics\SmartImport\Providers\Foundation\RuntimeProviderCatalog::hasCapability((string) $p->getProviderInfo()?->getName(), 'conservative_identity_only')) { return $p; }
        $a = [];
        foreach ($p->getAttributes() as $attribute) {
            $key = strtolower($attribute->getName());
            if (!isset($a[$key])) { $a[$key] = $attribute->getValue(); }
        }
        $sourceAttributes = $a;
        $supplement = SupplementContentPolicy::editable($p)
            && (SupplementContentPolicy::applies($p) || ($a['supplement_merge_evidence'] ?? '') === 'same_barcode_compatible_records');
        $title = $p->getName();
        $description = $p->getDescription();
        $short = $p->getShortDescription();
        $seo = $p->getSeo();
        if (strtolower((string) $p->getProviderInfo()?->getName()) === 'braveweb') {
            $title = trim(explode('|', $title)[0]);
            $title = trim((string) preg_replace('/\s+(?:kopen|acquista|buy)\??\s*$/iu', '', $title));
            foreach (['description', 'short'] as $field) {
                $$field = trim((string) preg_replace('/\s*[-–|]\s*(?:buyer protection|fast shipping|spedizione gratuita).*$/iu', '', (string) $$field));
            }
            if ($title !== $p->getName() || $description !== $p->getDescription()) {
                $a['source_title'] = $a['source_title'] ?? $p->getName();
                $seo = new SeoDTO($title, $description ?: $title, $seo?->getKeywords());
            }
        }
        $categories = $p->getCategories();
        $categoryText = implode(' ', array_map(static fn ($c): string => $c->getName(), $categories));
        $categoryText .= ' ' . ($a['category'] ?? '') . ' ' . ($a['google_category'] ?? '');
        $evidence = $title . ' ' . $categoryText;
        $domain = $a['media_type'] ?? $a['resolved_media_type'] ?? 'generic';
        if (!empty($a['aic']) || !empty($a['aic_package'])) { $domain = 'medicine'; }
        elseif (!empty($a['isbn']) || !empty($a['isbn_13'])) { $domain = 'book'; }
        $authoritativeDomain = in_array($a['classification_authority'] ?? '', ['operator_explicit_domain', 'operator_category_decision', 'identifier_semantics'], true) ? $domain : null;
        if ($supplement && $authoritativeDomain === null) {
            if ($categories === [] && empty($a['category'])) { $categories = [new CategoryDTO('Integratori alimentari')]; }
            if (in_array($domain, ['auto', 'generic', 'unknown'], true)) { $domain = 'food'; }
        }
        // A retailer taxonomy describes a shelf, not necessarily this product.
        // Preserve an already recognised film when its own title confirms the
        // physical video edition. Hardware in the title still vetoes it.
        $titleHardware = self::isHardware($title);
        $explicitVideoTitle = preg_match('/(?:\[(?:dvd|blu[ -]?ray(?: 4k(?: uhd)?)?)\]|\((?:[1-9]\d*\s+)?(?:dvd|blu[ -]?ray|4k(?: uhd)?)\)|\bdvd(?:\s*\[region\s*[0-9]+\])?\s*$|\bblu[ -]?ray(?:\s+(?:4k|uhd))*\s*$)/iu', $title) === 1;
        $videoEdition = !$titleHardware && (
            ($domain === 'movie' && preg_match('/\b(?:dvd|blu[ -]?ray|vhs|videocassetta|super ?8)\b/iu', $title) === 1)
            || (in_array($domain, ['auto', 'generic', 'unknown'], true) && $explicitVideoTitle)
        );
        if ($videoEdition && $authoritativeDomain === null) { $domain = 'movie'; }
        if ($videoEdition && $authoritativeDomain === null) {
            $rejected = [];
            $categories = array_values(array_filter($categories, static function ($category) use (&$rejected): bool {
                if (!self::isHardware($category->getName())) { return true; }
                $rejected[] = $category->getName();
                return false;
            }));
            foreach (['category', 'google_category', 'source_category'] as $key) {
                if (isset($a[$key]) && self::isHardware($a[$key])) {
                    $rejected[] = $a[$key];
                    unset($a[$key]);
                }
            }
            if ($rejected !== []) {
                $a['source_rejected_hardware_categories'] = implode(' | ', array_unique($rejected));
                $a['category_conflict_reason'] = 'video_edition_with_hardware_taxonomy';
                if ($seo !== null && self::isHardware((string) $seo->getKeywords())) {
                    $seo = new SeoDTO($seo->getTitle(), $seo->getDescription(), null);
                }
            }
        }
        $hardware = $titleHardware || (!$videoEdition && self::isHardware($categoryText));
        $stationery = preg_match('/\b(?:highlighter|evidenziatore|textmarker|stationery)\b/iu', $evidence) === 1
            || (strtolower(trim($a['brand'] ?? '')) === 'stabilo' && preg_match('/\bboss\b/i', $title));
        if (($hardware || $stationery) && $authoritativeDomain === null) {
            $domain = 'generic';
            unset($a['sale_format'], $a['sale_format_source']);
            if ($stationery) {
                foreach (['nutriscore', 'ingredients', 'allergens', 'labels', 'energy_kj', 'energy_kcal', 'fat', 'saturated_fat', 'carbohydrates', 'sugars', 'protein', 'salt'] as $field) { unset($a[$field]); }
                $categories = [new CategoryDTO('Cancelleria')];
            }
        } elseif (preg_match('/\b(?:videogames?|video games?|playstation|ps[345]|xbox|nintendo)\b/iu', $evidence)) {
            $domain = 'game';
        } elseif ((in_array($domain, ['auto', 'generic', 'unknown'], true) && self::hasMusicCategory($p)) || preg_match('/\b(?:music cds?|music & sound recordings|audio cd|compact disc)\b/iu', $evidence)) {
            $domain = 'music';
            if (preg_match('/\bcd\b/i', $title)) { $a['format'] = $a['format'] ?? 'CD'; }
        } elseif (preg_match('/\b(?:sapone|body soap|tělová mýdla|cosmetics|cosmetica)\b/iu', $evidence . ' ' . $description)) {
            $domain = 'beauty';
        } elseif (in_array($domain, ['auto', 'generic', 'unknown'], true)
            && (preg_match('/(?:\((?:[1-9]\d*\s+)?(?:dvd|blu[ -]?ray|4k(?: uhd)?)\)|\bDVD)\s*$/i', $title)
                || self::hasVideoCategory($p))) {
            $domain = 'movie';
        }
        if ($authoritativeDomain !== null) { $domain = $authoritativeDomain; }
        if ($domain !== ($a['media_type'] ?? 'generic')) {
            $a['resolved_media_type'] = $domain;
            $a['domain_evidence_source'] = 'product_title_and_category';
        }
        $a['media_type'] = $domain;
        if ($domain === 'movie' && ! $hardware) {
            // The semantic result wins over broad retailer shelves such as
            // "Software" or a stale jewelry support restored by the UI.
            $rejectedCategories = [];
            $normalizedCategories = [];
            foreach ($categories as $category) {
                $name = trim($category->getName());
                if (self::isIncompatibleVideoCategory($name)) {
                    $rejectedCategories[] = $name;
                    continue;
                }
                if (preg_match('/^(?:movies?|films?|dvd video|blu[ -]?ray)$/iu', $name) === 1) {
                    $name = 'Film';
                }
                if ($name !== '') {
                    $normalizedCategories[strtolower($name)] = new CategoryDTO($name);
                }
            }
            $normalizedCategories['film'] = new CategoryDTO('Film');
            $categories = array_values($normalizedCategories);

            foreach (['category', 'google_category', 'source_category'] as $categoryField) {
                if (! isset($a[$categoryField]) || ! self::isIncompatibleVideoCategory($a[$categoryField])) {
                    continue;
                }
                $rejectedCategories[] = $a[$categoryField];
                $a['source_' . $categoryField] = $a[$categoryField];
                unset($a[$categoryField]);
            }
            if ($rejectedCategories !== []) {
                $a['source_rejected_video_categories'] = implode(' | ', array_values(array_unique($rejectedCategories)));
                $a['category_conflict_reason'] = 'canonical_movie_overrides_retailer_shelf';
            }

            $a['macro_category'] = 'film_home_video';
            foreach (['jewelry_kind', 'metal_purity', 'stone'] as $jewelryField) {
                unset($a[$jewelryField]);
            }

            $formatEvidence = implode(' ', array_filter([
                $title,
                $a['physical_format'] ?? '',
                $a['format'] ?? '',
            ]));
            $saleFormat = self::videoSaleFormat($formatEvidence);
            if ($saleFormat !== null) {
                if (empty($a['sale_format'])) { $a['sale_format'] = $saleFormat; }
                // POST roundtrips do not confer authority through provenance
                // strings. Re-establish the source only when the physical
                // text independently confirms the preserved format.
                if ($a['sale_format'] === $saleFormat && empty($a['sale_format_source'])) {
                    $a['sale_format_source'] = 'physical_identifier_evidence';
                }
            }

            // Technical edition labels remain available as attributes, but
            // must never become plot, excerpt, SEO copy or subtitle.
            if (self::isTechnicalPhysicalText($description)) {
                $a['source_physical_description'] ??= (string) $description;
                $a['description_source_type'] = 'physical_edition_metadata';
                $description = null;
            }
            if (self::isTechnicalPhysicalText($short)) {
                $a['source_physical_short_description'] = (string) $short;
                $short = null;
            }
        }
        if ($domain === 'music') {
            foreach (['catalog_number'=>'catno', 'genres'=>'genre', 'release_date'=>'year'] as $target=>$source) {
                if (empty($a[$target])) {
                    $values = [];
                    foreach ($p->getAttributes() as $field) { if ($field->getName() === $source && trim($field->getValue()) !== '') { $values[] = $field->getValue(); } }
                    if ($values !== []) { $a[$target] = implode(', ', array_unique($values)); }
                }
            }
            if (empty($a['sale_format']) && preg_match('/\b45\s*(?:RPM|giri)\b/i', $a['format'] ?? '')) { $a['sale_format'] = 'vinyl_45'; }
            elseif (empty($a['sale_format']) && preg_match('/\b33(?:[.,]1?3?| 1\/3)?\s*(?:RPM|giri)\b/i', $a['format'] ?? '')) { $a['sale_format'] = 'vinyl_33'; }
        }
        if ($categories === [] && trim($a['category'] ?? '') !== '') {
            $categories = [new CategoryDTO($a['category'])];
        }
        $categoryMap = [
            'Apparel & Accessories > Shoes' => 'Abbigliamento e accessori > Scarpe',
            'Media > Music & Sound Recordings > Music CDs' => 'Musica > CD',
            'Zdraví a krása > Osobní péče > Kosmetika > Koupele a péče o tělo > Tělová mýdla' => 'Bellezza e cura personale > Saponi per il corpo',
        ];
        $categories = array_map(static fn ($c) => isset($categoryMap[$c->getName()]) ? new CategoryDTO($categoryMap[$c->getName()]) : $c, $categories);
        if (preg_match('/^org-/i', $a['manufacturer'] ?? '')) { unset($a['manufacturer']); }
        if (trim($a['package_quantity'] ?? '') === '0') { unset($a['package_quantity']); }
        if (strtolower($a['packaging'] ?? '') === 'bocal') { $a['packaging'] = 'Vasetto'; }
        if (isset($a['labels'])) {
            $labels = preg_split('/\s*[,;]\s*/u', $a['labels']) ?: [];
            $map = ['vegetarian' => 'Vegetariano', 'végétarien' => 'Vegetariano', 'no-gluten' => 'Senza glutine', 'gluten-free' => 'Senza glutine'];
            $a['labels'] = implode(', ', array_unique(array_map(static fn ($v) => $map[function_exists('mb_strtolower') ? mb_strtolower($v) : strtolower($v)] ?? $v, $labels)));
        }
        if ($supplement) {
            if (preg_match('/^(?:unknown|n\/a|n\/d|null|undefined)$/i', trim($a['nutriscore'] ?? ''))) {
                $a['source_nutriscore'] ??= $a['nutriscore'];
                unset($a['nutriscore']);
            }
            $minsan = trim($a['minsan'] ?? '');
            if ($minsan === '' && preg_match('/\bMINSAN\s*[:#-]?\s*(\d{9})\b/i', (string) $description, $m)) { $minsan = $m[1]; }
            if (preg_match('/^\d{9}$/', $minsan)) {
                $a['minsan'] = $minsan;
                foreach (['model', 'mpn'] as $field) {
                    if (trim($a[$field] ?? '') === $minsan) { $a['source_' . $field] ??= $a[$field]; unset($a[$field]); }
                }
            }
            $notes = [];
            if ($minsan === '' && preg_match('/^\d{9}$/', trim($a['mpn'] ?? $a['model'] ?? ''))) {
                $notes[] = 'Codice a nove cifre da verificare: la fonte non ne dichiara il tipo.';
                if (isset($a['model'], $a['mpn']) && $a['model'] === $a['mpn']) {
                    $a['source_model'] ??= $a['model'];
                    unset($a['model']);
                }
            }
            if (SupplementContentPolicy::truncated($description)) { $notes[] = 'Descrizione incompleta nella fonte: verificare il testo integrale.'; }
            if (SupplementContentPolicy::medicineClaim($description) || SupplementContentPolicy::medicineClaim($short)) {
                $notes[] = 'Conflitto: una descrizione definisce medicinale un prodotto identificato come integratore.';
                if (SupplementContentPolicy::medicineClaim($description)) { $a['source_conflicting_description'] = (string) $description; $description = null; }
                if (SupplementContentPolicy::medicineClaim($short)) { $a['source_conflicting_short_description'] = (string) $short; $short = null; }
            }
            if (isset($a['source_supplement_candidates']) && SupplementContentPolicy::medicineClaim($a['source_supplement_candidates'])) {
                $notes[] = 'Una fonte contiene una descrizione medicinale incompatibile, esclusa dalla selezione automatica.';
            }
            foreach ($notes as $note) {
                if (!str_contains($a['content_review_notes'] ?? '', $note)) {
                    $a['content_review_notes'] = trim(($a['content_review_notes'] ?? '') . ' ' . $note);
                }
            }
        }
        if (!$supplement && empty($a['model']) && !empty($a['mpn'])) { $a['model'] = $a['mpn']; }
        if (empty($a['model']) && preg_match('/\b[A-Z]{1,5}\d{1,4}-[A-Z0-9]{3,}\b/', $title, $m)) { $a['model'] = $m[0]; }
        if (preg_match('/\btoner\b/i', $title)) {
            if (empty($a['mpn']) && preg_match('/\((\d{6,12})\)/', $title, $m)) { $a['mpn'] = $m[1]; }
            if (empty($a['color']) && preg_match('/\b(cyan|magenta|yellow|black)\b/i', $title, $m)) { $a['color'] = ucfirst(strtolower($m[1])); }
            if (empty($a['specifications']) && preg_match('/\b([\d.]+)\s+(?:Seiten|pages|pagine)\b/i', $title, $m)) { $a['specifications'] = 'Resa dichiarata: ' . $m[1] . ' pagine'; }
        }
        if (empty($a['size']) && preg_match('/\b\d{2}(?:[.,]\d)?\s*EU(?:\s*\/\s*\d+(?:[.,]\d)?\s*[A-Z]?\s*US)?\b/i', $title, $m)) { $a['size'] = $m[0]; }
        if (!$supplement && empty($a['model']) && !empty($a['mpn'])) { $a['model'] = $a['mpn']; }
        if (empty($a['brand']) && !empty($a['model']) && preg_match('/^([A-Z][a-zA-Z]+)\s+' . preg_quote($a['model'], '/') . '\b/', $title, $m)) { $a['brand'] = $m[1]; }
        if (str_contains(strtolower($categoryText), 'shoes') && empty($a['color']) && preg_match('/,\s*(Coffee|Black|Brown|White)\s*,/i', $title, $m)) { $a['color'] = $m[1]; }
        if (strtolower((string) $p->getProviderInfo()?->getName()) === 'braveweb' && !empty($a['brand'])) {
            $brandPattern = preg_quote((string) $a['brand'], '/');
            foreach (['description', 'short'] as $field) {
                $original = (string) $$field;
                $$field = preg_replace('/^(' . $brandPattern . ')\s+' . $brandPattern . '(?=\s)/iu', '$1', $original);
                if ($$field !== $original) { $a['source_' . $field] ??= $original; }
            }
        }
        if ($domain === 'jewelry') {
            $notes = [];
            if (preg_match('/^\s*collana\b/iu', $title)) {
                $a['jewelry_kind'] = 'necklace';
                if ($categories === []) { $categories[] = new CategoryDTO('Collane'); }
            }
            if (preg_match('/\bacciaio\b/iu', $title) && preg_match('/^metallo$/iu', $a['material'] ?? '')) {
                $notes[] = 'Materiale da verificare: il titolo indica acciaio, il campo materiale indica metallo.';
            }
            if (preg_match('/smalto\s+bianco/iu', (string) $description) && preg_match('/turchese/iu', $a['color'] ?? '')) {
                $notes[] = 'Colore da verificare: la descrizione indica smalto bianco, il campo colore indica turchese.';
            }
            if (preg_match('/Certificazione:\s*([^,|]*\b\d+\s*(?:mesi|months)\b)/iu', $a['specifications'] ?? '', $m)) {
                $a['source_certification'] = $m[1];
                $a['specifications'] = str_replace($m[0], 'Durata dichiarata da verificare: ' . $m[1], $a['specifications']);
                $notes[] = 'La durata in mesi non prova una certificazione: verificare se si tratta della garanzia.';
            }
            if ($notes !== []) { $a['content_review_notes'] = implode(' ', $notes); }
        }
        $cleaner = new ProductDescriptionCleaner();
        $cleanDescription = $cleaner->clean($description, $title);
        $cleanShort = $cleaner->clean($short, $title);
        if ($cleanDescription !== $description || $cleanShort !== $short) {
            $a['description_cleanup_status'] = 'cleaned';
        }
        $description = $cleanDescription;
        $short = $cleanShort;
        $seo = new SeoDTO(
            $cleaner->clean($supplement ? $title : ($seo?->getTitle() ?? $title)) ?: $title,
            $cleaner->clean($supplement
                ? (SupplementContentPolicy::completeExcerpt($description) ?: SupplementContentPolicy::completeExcerpt($short) ?: $title)
                : ($description ?: ($short ?: $title))) ?: $title,
            $supplement ? implode(', ', array_filter(preg_split('/[,;|]+/u', (string) $seo?->getKeywords()) ?: [],
                static fn (string $v): bool => trim($v) !== '' && !preg_match('/^(?:unknown|n\/a|n\/d|null|undefined)$/i', trim($v)))) : $seo?->getKeywords()
        );
        if ($supplement && $p->getImages() !== [] && !isset($a['selected_image_urls'])
            && !isset($a['featured_image']) && !isset($a['image_selection_confirmed'])) {
            // Initial acquisition only: match the cover to the featured-first
            // image order produced by the merger. Existing selections win.
            $a['cover_url'] = $p->getImages()[0]->getUrl();
        }
        $changedSources = [];
        foreach ($sourceAttributes as $key => $value) {
            if ($key !== 'source_normalization_evidence' && (!isset($a[$key]) || $a[$key] !== $value)) { $changedSources[$key] = $value; }
        }
        if ($changedSources !== [] && !isset($a['source_normalization_evidence'])) {
            $a['source_normalization_evidence'] = (string) json_encode($changedSources, JSON_UNESCAPED_UNICODE | JSON_UNESCAPED_SLASHES);
        }
        $attributes = [];
        foreach ($a as $key => $value) { $attributes[] = new AttributeDTO($key, (string) $value); }
        // Preserve multi-valued provider fields (artists, labels, categories).
        $seen = [];
        foreach ($p->getAttributes() as $attribute) {
            $key = strtolower($attribute->getName());
            if (isset($seen[$key]) && isset($a[$key]) && !in_array($key, ['media_type', 'resolved_media_type', 'labels'], true)) { $attributes[] = $attribute; }
            $seen[$key] = true;
        }
        return (new ProductImageCanonicalizer())->normalize(new ProductDTO($p->getId(), $title, $description, $short, $p->getPrice(), $p->getInventory(), $categories, $p->getImages(), $attributes, $seo, $p->getProviderInfo(), $p->getSku(), $p->getBarcode()));
    }

    private static function isIncompatibleVideoCategory(string $category): bool
    {
        $category = trim($category);
        if ($category === '') { return false; }

        return self::isHardware($category)
            || preg_match('/^(?:software|jewelry|jewellery|gioielli|abbigliamento|clothing|apparel|electronics?|elettronica)$/iu', $category) === 1;
    }

    private static function videoSaleFormat(string $evidence): ?string
    {
        if (preg_match('/\b(?:4k\s*(?:uhd)?|uhd\s*4k|ultra\s*hd)\b/iu', $evidence) === 1) { return 'uhd_4k'; }
        if (preg_match('/\bblu[ -]?ray\b/iu', $evidence) === 1) { return 'blu_ray'; }
        if (preg_match('/\bdvd(?:-standard)?\b/iu', $evidence) === 1) { return 'dvd'; }
        if (preg_match('/\bvhs\b/iu', $evidence) === 1) { return 'vhs'; }
        if (preg_match('/\bsuper[ -]?8\b/iu', $evidence) === 1) { return 'super8'; }
        return null;
    }

    private static function isTechnicalPhysicalText(?string $value): bool
    {
        $value = trim((string) $value);
        if ($value === '') { return false; }
        $labels = preg_match_all('/\b(?:Label|Format|Region|Release Date|Video Format|No\.? of Discs?|UPC)\s*:/iu', $value);

        return $labels >= 2
            || preg_match('/^ATTENTION:\s*REQUIRES\s+(?:NTSC|PAL)\s+REGION\b/iu', $value) === 1;
    }
}