Import product photos from live sites; link downloads and SEO meta to products

- vitec:import-product-images: attach staged photos (fileadmin/user_upload/
  products/<slug>/, 116 files harvested from vitec.com and datapath.co.uk
  carousels) to image-less products via FAL/DataHandler; fill-up only.
  fix_image_reference_fieldname.php repairs the first run's references
  (TCA field is productimage, not the JSON key images)
- vitec:link-product-downloads: match the download records imported on
  2026-08-06 to products by title (doc-type/brand/role suffix stripping,
  alias table for live-site naming deviations, series rules for g45xx/e38xx/
  chassis docs); 55 unique-match links added, add-only, idempotent
- SEO meta: seo_meta_vitec.json + apply_seo_meta.php fill empty seotitle/
  seometa from the live vitec.com product pages (11 products, fill-up only)"
This commit is contained in:
2026-09-04 14:34:38 +02:00
parent 4a49196bcc
commit c91063d6f3
5 changed files with 601 additions and 0 deletions

View File

@@ -0,0 +1,75 @@
<?php
declare(strict_types=1);
/**
* Fill empty seotitle/seometa product fields from a scraped mapping file
* (migrations/seo_meta_vitec.json - source: the live vitec.com product
* pages, 2026-09-04). Fill-up ONLY: fields that already hold a value are
* never overwritten.
*
* Dry run: php migrations/apply_seo_meta.php
* Write: php migrations/apply_seo_meta.php --apply
*/
$apply = in_array('--apply', $argv, true);
$root = null;
$dir = __DIR__;
for ($i = 0; $i < 5; $i++) {
if (is_file($dir . '/config/system/settings.php')) {
$root = $dir;
break;
}
$dir = dirname($dir);
}
if ($root === null) {
exit("Could not locate project root above " . __DIR__ . "\n");
}
$map = json_decode((string)file_get_contents(__DIR__ . '/seo_meta_vitec.json'), true);
if (!is_array($map)) {
exit("seo_meta_vitec.json missing or invalid.\n");
}
unset($map['_quelle']);
$settings = require $root . '/config/system/settings.php';
$db = $settings['DB']['Connections']['Default'];
$pdo = new PDO(
sprintf('mysql:host=%s;port=%s;dbname=%s;charset=utf8mb4', $db['host'] ?? 'localhost', $db['port'] ?? 3306, $db['dbname'] ?? ''),
$db['user'] ?? '',
$db['password'] ?? '',
[PDO::ATTR_ERRMODE => PDO::ERRMODE_EXCEPTION, PDO::ATTR_DEFAULT_FETCH_MODE => PDO::FETCH_ASSOC]
);
$select = $pdo->prepare("SELECT uid, title, seotitle, seometa FROM tx_vitec_domain_model_product WHERE deleted = 0 AND slug = ?");
$update = $pdo->prepare("UPDATE tx_vitec_domain_model_product SET seotitle = ?, seometa = ?, tstamp = ? WHERE uid = ?");
$written = 0;
foreach ($map as $slug => $meta) {
$select->execute([$slug]);
$product = $select->fetch();
if (!$product) {
echo "FEHLT $slug - kein Produkt\n";
continue;
}
$newTitle = trim((string)$product['seotitle']) === '' ? trim((string)($meta['seotitle'] ?? '')) : (string)$product['seotitle'];
$newMeta = trim((string)$product['seometa']) === '' ? trim((string)($meta['seometa'] ?? '')) : (string)$product['seometa'];
$changes = [];
if ($newTitle !== (string)$product['seotitle']) {
$changes[] = 'seotitle';
}
if ($newMeta !== (string)$product['seometa']) {
$changes[] = 'seometa';
}
if ($changes === []) {
echo "ok {$product['title']} - beide Felder bereits gefuellt oder keine Daten\n";
continue;
}
printf("%s %s (uid %d): %s\n", $apply ? 'WRITE ' : 'plan ', $product['title'], $product['uid'], implode(' + ', $changes));
if ($apply) {
$update->execute([$newTitle, $newMeta, time(), (int)$product['uid']]);
}
$written++;
}
printf("\n%d Produkt(e) %s.%s\n", $written, $apply ? 'geschrieben' : 'geplant', $apply ? ' Cache leeren: vendor/bin/typo3 cache:flush' : ' Mit --apply schreiben.');

View File

@@ -0,0 +1,59 @@
<?php
declare(strict_types=1);
/**
* One-shot repair (2026-09-04): the first vitec:import-product-images run
* wrote its sys_file_reference rows with fieldname 'images' - the JSON key -
* but the TCA field (and everything that queries it) is 'productimage'.
* Re-files those references and refreshes the products' inline counter.
*
* Usage: php migrations/fix_image_reference_fieldname.php
*/
$root = null;
$dir = __DIR__;
for ($i = 0; $i < 5; $i++) {
if (is_file($dir . '/config/system/settings.php')) {
$root = $dir;
break;
}
$dir = dirname($dir);
}
if ($root === null) {
exit("Could not locate project root above " . __DIR__ . "\n");
}
$settings = require $root . '/config/system/settings.php';
$db = $settings['DB']['Connections']['Default'];
$pdo = new PDO(
sprintf('mysql:host=%s;port=%s;dbname=%s;charset=utf8mb4', $db['host'] ?? 'localhost', $db['port'] ?? 3306, $db['dbname'] ?? ''),
$db['user'] ?? '',
$db['password'] ?? '',
[PDO::ATTR_ERRMODE => PDO::ERRMODE_EXCEPTION]
);
$wrong = (int)$pdo->query(
"SELECT COUNT(*) FROM sys_file_reference
WHERE deleted = 0 AND tablenames = 'tx_vitec_domain_model_product' AND fieldname = 'images'"
)->fetchColumn();
echo "Misfiled references (fieldname 'images'): $wrong\n";
if ($wrong > 0) {
$pdo->exec(
"UPDATE sys_file_reference SET fieldname = 'productimage'
WHERE deleted = 0 AND tablenames = 'tx_vitec_domain_model_product' AND fieldname = 'images'"
);
echo "Re-filed to 'productimage'.\n";
}
$updated = $pdo->exec(
"UPDATE tx_vitec_domain_model_product p
SET p.productimage = (
SELECT COUNT(*) FROM sys_file_reference r
WHERE r.deleted = 0 AND r.uid_foreign = p.uid
AND r.tablenames = 'tx_vitec_domain_model_product' AND r.fieldname = 'productimage'
)
WHERE p.deleted = 0"
);
echo "Inline counters refreshed on $updated product(s). Flush the cache: vendor/bin/typo3 cache:flush\n";

View File

@@ -0,0 +1,51 @@
{
"_quelle": "vitec.com Produktseiten, Scrape 2026-09-04. Nur Produkt-Detailseiten; Kategorieseiten-Metas (ChannelLink, 45-series, PRISM, 38-Series) bewusst ausgelassen (zu generisch). 'VITEC - '-Praefix entfernt.",
"mgw-diamond-sdi-encoder": {
"seotitle": "MGW Diamond - 4K and Multi-Channel SD/HD HEVC Encoder",
"seometa": "MGW Diamond is a small, power-efficient quad channel HD or one channel 4K HEVC video encoder ideal for field-based applications. It features a powerful encoding engine with the ability to output up to eight streams simultaneously."
},
"mgw-ace-encoder": {
"seotitle": "MGW Ace Encoder Compact HEVC (H.265) Hardware Encoder",
"seometa": "MGW Ace Encoder is the world's first HEVC / H.265 hardware encoder in a professional grade portable streaming appliance. Powered by VITEC HEVC GEN2+ encoder, it delivers industry's best video quality with up to 50% bandwidth savings compared to H.264 and Ultra Low Latency streaming down to 16ms glass-to-glass."
},
"mgw-ace-decoder": {
"seotitle": "MGW Ace Decoder - Professional Portable HEVC & H.264 Decoder",
"seometa": "MGW Ace Decoder is a professional grade, high performance IP decoder supporting the bandwidth efficient HEVC/H.265 and H.264/AVC compression standards, with 4:2:2 10-bit decoding from IP or DVB-ASI, genlock support and 4K-ready 12G-SDI and HDMI outputs."
},
"mgw-pico-portable-encoder": {
"seotitle": "MGW Pico Encoder - Ultra-Compact, Low Latency H.264 Encoding & Streaming Appliance",
"seometa": "The MGW Pico Encoder is the world's smallest H.264 HD/SD portable encoding appliance. With 3G/HD/SD-SDI and Composite inputs, low power consumption and robust industrial design, the appliance is ideal for any video field-based streaming application."
},
"mgw-diamond-og-sdi-composite-blade-encoder": {
"seotitle": "MGW Diamond OG - 4K and Multi-Channel SD/HD HEVC VITEC OG Encoder Card",
"seometa": "MGW Diamond OG is a small, power-efficient quad channel HD or one channel 4K HEVC video encoder card for the openGear ecosystem, with a powerful encoding engine able to output up to eight streams simultaneously."
},
"mgw-diamond-og-sdi-blade-encoder": {
"seotitle": "MGW Diamond+ OG Encoder - Multi-codec, Broadcast-grade 4K/Multichannel HD VITEC OG Encoder Card",
"seometa": "MGW Diamond+ OG is a broadcast grade HEVC, H.264 and MPEG-2 IP encoder that is ideal for contribution or point-to-point streaming applications and compatible with the openGear ecosystem for seamless integration."
},
"mgw-diamond-hx-og": {
"seotitle": "MGW Diamond-Hx OG Encoder 4K & Multi-Channel SD/HD HDMI VITEC OG Encoder Card",
"seometa": "openGear 4K HEVC video encoder card that is ideal for IPTV distribution or Direct-to-Web applications, featuring a powerful encoding engine with the ability to deliver up to four streams simultaneously."
},
"mgw-ace-decoder-og-ultra-low-latency-blade-encoder": {
"seotitle": "MGW Ace Decoder OG - Professional HEVC & H.264 openGear Decoder Card",
"seometa": "MGW Ace Decoder OG is a professional grade, high performance IP decoder card supporting HEVC/H.265 and H.264/AVC, with 4:2:2 10-bit decoding from IP or DVB-ASI, genlock support and 4K-ready outputs for the openGear ecosystem."
},
"diamond-ip-og-ip-to-ip-encoder": {
"seotitle": "Diamond-IP OG Encoder - 4K HEVC encoder with SMPTE ST2110 capture",
"seometa": ""
},
"mges7000-high-density-4kuhdhd-hevc-h264-iptv-encoding-blade": {
"seotitle": "MGES 7000 - Market Leading High Density 4K/UHD/HD HEVC & H.264 IPTV Encoding Blade",
"seometa": "Featuring the highest density available on the market, VITEC's 4K/UHD/HD HEVC & H.264 IPTV 8-input blade offers real-time hardware encoding with secondary channel, integrated resolution and frame-rate scaling, AES 256/128-bit encryption and low latency mode."
},
"mges6000-broadcast-quality-hdsd-h264-quad-input-iptv-encoder": {
"seotitle": "MGES 6000 - Broadcast-quality HD/SD H.264 Quad-Input IPTV Encoder",
"seometa": "The MGES-6000 Blade provides real-time hardware encoding of HD and SD video for all IPTV applications, with best-in-class picture quality, a secondary low-res channel, streaming to up to seven IP destinations per port and optional AES-256/128-bit encryption."
},
"x-player-end-points-xp23-xp25-xp26": {
"seotitle": "VITEC X-Player End-Points",
"seometa": "High-performance IP video & digital signage playback devices"
}
}

View File

@@ -0,0 +1,142 @@
<?php
declare(strict_types=1);
namespace Evomedien\Vitec\Command;
use Doctrine\DBAL\ParameterType;
use Symfony\Component\Console\Attribute\AsCommand;
use Symfony\Component\Console\Command\Command;
use Symfony\Component\Console\Input\InputInterface;
use Symfony\Component\Console\Input\InputOption;
use Symfony\Component\Console\Output\OutputInterface;
use TYPO3\CMS\Core\Core\Bootstrap;
use TYPO3\CMS\Core\Database\ConnectionPool;
use TYPO3\CMS\Core\DataHandling\DataHandler;
use TYPO3\CMS\Core\Resource\StorageRepository;
use TYPO3\CMS\Core\Utility\GeneralUtility;
/**
* Attach the product photos harvested from vitec.com / datapath.co.uk
* (2026-09-04) to the product records.
*
* Files live in fileadmin/user_upload/products/<product-slug>/ - one folder
* per product, downloaded straight from the live sites. For every folder
* whose slug matches a product record WITHOUT any image (fill-up only,
* existing images are never touched), the files are indexed in FAL and
* appended to the `images` field through DataHandler (sys_file_reference),
* in alphabetical file order (the live carousels' _01.._NN order).
*
* Idempotent: products that got their images on a previous run are skipped
* on the next one.
*
* vendor/bin/typo3 vitec:import-product-images --dry-run
* vendor/bin/typo3 vitec:import-product-images
*/
#[AsCommand(
name: 'vitec:import-product-images',
description: 'Attach staged product photos (fileadmin/user_upload/products/<slug>/) to image-less products'
)]
final class ImportProductImagesCommand extends Command
{
private const TABLE = 'tx_vitec_domain_model_product';
private const STAGE = 'user_upload/products';
protected function configure(): void
{
$this->addOption('dry-run', null, InputOption::VALUE_NONE, 'Report the plan - write nothing');
}
protected function execute(InputInterface $input, OutputInterface $output): int
{
Bootstrap::initializeBackendAuthentication();
$dryRun = (bool)$input->getOption('dry-run');
$storage = GeneralUtility::makeInstance(StorageRepository::class)->getDefaultStorage();
if ($storage === null || !$storage->hasFolder(self::STAGE)) {
$output->writeln('<error>Staging folder ' . self::STAGE . ' not found in the default storage.</error>');
return Command::FAILURE;
}
$stageFolder = $storage->getFolder(self::STAGE);
$connection = GeneralUtility::makeInstance(ConnectionPool::class)->getConnectionForTable(self::TABLE);
$datamap = [];
$newIndex = 0;
$planned = 0;
$skippedExisting = 0;
foreach ($stageFolder->getSubfolders() as $folder) {
$slug = $folder->getName();
$product = $connection->fetchAssociative(
'SELECT uid, pid, title FROM ' . self::TABLE . ' WHERE deleted = 0 AND slug = ?',
[$slug]
);
if ($product === false) {
$output->writeln(sprintf('<comment>skip %-45s - no product with this slug</comment>', $slug));
continue;
}
$uid = (int)$product['uid'];
// NB: the TCA field is `productimage` - `images` is only the
// JSON key the serializers emit (lesson from the first run).
$existing = (int)$connection->fetchOne(
"SELECT COUNT(*) FROM sys_file_reference
WHERE deleted = 0 AND tablenames = ? AND fieldname = 'productimage' AND uid_foreign = ?",
[self::TABLE, $uid]
);
if ($existing > 0) {
$output->writeln(sprintf('keep %-45s - already has %d image(s)', $slug, $existing));
$skippedExisting++;
continue;
}
$files = $folder->getFiles();
usort($files, static fn($a, $b): int => strcasecmp($a->getName(), $b->getName()));
if ($files === []) {
continue;
}
$referenceIds = [];
foreach ($files as $file) {
// getFiles() delivers indexed File objects (sys_file created on
// the fly for new files), so uid is available right away.
$newId = 'NEW' . ++$newIndex;
$referenceIds[] = $newId;
$datamap['sys_file_reference'][$newId] = [
'table_local' => 'sys_file',
'uid_local' => $file->getUid(),
'uid_foreign' => $uid,
'tablenames' => self::TABLE,
'fieldname' => 'productimage',
'pid' => (int)$product['pid'],
];
}
$datamap[self::TABLE][$uid]['productimage'] = implode(',', $referenceIds);
$planned++;
$output->writeln(sprintf('%s %-45s -> uid %d "%s": %d image(s)',
$dryRun ? 'plan ' : 'WRITE', $slug, $uid, $product['title'], count($referenceIds)));
}
$output->writeln(sprintf("\n%d product(s) to fill, %d already had images.", $planned, $skippedExisting));
if ($planned === 0) {
return Command::SUCCESS;
}
if ($dryRun) {
$output->writeln('DRY RUN - nothing written.');
return Command::SUCCESS;
}
$dataHandler = GeneralUtility::makeInstance(DataHandler::class);
$dataHandler->start($datamap, []);
$dataHandler->process_datamap();
if ($dataHandler->errorLog !== []) {
foreach ($dataHandler->errorLog as $error) {
$output->writeln('<error>' . $error . '</error>');
}
return Command::FAILURE;
}
$output->writeln('Done. Flush the cache: vendor/bin/typo3 cache:flush');
return Command::SUCCESS;
}
}

View File

@@ -0,0 +1,274 @@
<?php
declare(strict_types=1);
namespace Evomedien\Vitec\Command;
use Symfony\Component\Console\Attribute\AsCommand;
use Symfony\Component\Console\Command\Command;
use Symfony\Component\Console\Input\InputInterface;
use Symfony\Component\Console\Input\InputOption;
use Symfony\Component\Console\Output\OutputInterface;
use TYPO3\CMS\Core\Core\Bootstrap;
use TYPO3\CMS\Core\Database\ConnectionPool;
use TYPO3\CMS\Core\DataHandling\DataHandler;
use TYPO3\CMS\Core\Utility\GeneralUtility;
/**
* Link the imported download records (vitec:import-downloads, 2026-08-06) to
* their products via `downloads` (tx_vitec_product_download_mm) by matching
* the download title against the product titles.
*
* Deliberately conservative: a download is linked ONLY when the match is
* unique. Matching per download title, in this order:
*
* 1. title minus a trailing document-type phrase ("... Datasheet",
* "... Brochure", "... User Guide", ...) == product title
* 2. additionally with a leading brand word (VITEC / Datapath) removed
* 3. additionally with a trailing role word (Transmitter / Receiver /
* Encoder / Decoder / Card) removed - "Aligo TX100 Transmitter
* Datasheet" -> "TX100"
*
* All comparisons on a normalized form (lowercase, alphanumerics only).
* Existing links are never touched, nothing is removed; a download that
* matches nothing (or more than one product) is reported for manual review.
* Idempotent: already-linked pairs are skipped.
*
* vendor/bin/typo3 vitec:link-product-downloads --dry-run
* vendor/bin/typo3 vitec:link-product-downloads
*/
#[AsCommand(
name: 'vitec:link-product-downloads',
description: 'Link download records to products by title match (unique matches only, add-only)'
)]
final class LinkProductDownloadsCommand extends Command
{
private const PRODUCT_TABLE = 'tx_vitec_domain_model_product';
private const DOWNLOAD_TABLE = 'tx_vitec_domain_model_download';
private const MM_TABLE = 'tx_vitec_product_download_mm';
/** Trailing document-type phrases, longest first. */
private const DOC_SUFFIXES = [
'quick start guide', 'quick reference guide', 'application note',
'success story', 'user guide', 'white paper', 'case study',
'whats new', "what's new", 'datasheet', 'data sheet', 'brochure',
'firmware', 'software', 'manual', 'flyer', 'guide',
];
/** Trailing role words a download title may carry beyond the product name. */
private const ROLE_SUFFIXES = [
'capture card', 'graphics card', 'encoder card', 'decoder card',
'transmitter', 'receiver', 'transcoder', 'end-point', 'end point',
'encoder', 'decoder', 'blade', 'card',
];
/**
* Known live-site names that differ from OUR (authoritative) dev names -
* normalized document name => dev product title. Dev names never change
* (decision 2026-09-04), so the bridge lives here.
*/
private const ALIASES = [
'visionscsdi4' => 'VisionSC-SD14',
'visioniosdi4' => 'VisionIO-SD14',
'visionsdi2' => 'VisionSD12',
'prismsff' => 'SFF PRISM',
'sffprism' => 'SFF PRISM',
'avediaep6' => 'Avedia End-Points',
'eztvep6' => 'EZ TV End-Points',
'xp2526' => 'X-Player End-Points',
'xp2650' => 'X-Player End-Points',
];
/**
* Series documents: regex on the lowercased document name (doc-type
* suffix already stripped) => dev product title. May match SEVERAL
* records of that title (the Avedia Modular Chassis exists twice by
* design - encoder branch and RF branch - and gets its datasheets on
* both).
*/
private const SERIES_RULES = [
'/^rf gateway g45\d+/' => '45-series RF gateways',
'/^encoder e38\d+/' => '38-Series Encoders',
'/^chassis c1(101|103|210)$/' => 'Avedia Modular Chassis',
];
protected function configure(): void
{
$this->addOption('dry-run', null, InputOption::VALUE_NONE, 'Report the plan - write nothing');
}
protected function execute(InputInterface $input, OutputInterface $output): int
{
Bootstrap::initializeBackendAuthentication();
$dryRun = (bool)$input->getOption('dry-run');
$connection = GeneralUtility::makeInstance(ConnectionPool::class)->getConnectionForTable(self::PRODUCT_TABLE);
$products = $connection->fetchAllAssociative(
'SELECT uid, title FROM ' . self::PRODUCT_TABLE . ' WHERE deleted = 0'
);
$productByNorm = [];
foreach ($products as $product) {
$productByNorm[$this->normalize((string)$product['title'])][] = (int)$product['uid'];
}
$titleByUid = array_column($products, 'title', 'uid');
$downloads = $connection->fetchAllAssociative(
'SELECT uid, title FROM ' . self::DOWNLOAD_TABLE . ' WHERE deleted = 0'
);
$existing = [];
foreach ($connection->fetchAllAssociative('SELECT uid_local, uid_foreign FROM ' . self::MM_TABLE) as $row) {
$existing[(int)$row['uid_local']][] = (int)$row['uid_foreign'];
}
$additions = []; // product uid => download uids to add
$unmatched = [];
$ambiguous = [];
foreach ($downloads as $download) {
$downloadUid = (int)$download['uid'];
$name = $this->stripDocSuffix((string)$download['title']);
if ($name === null) {
// No document-type suffix - not a per-product document
// (company brochures etc.); leave those to manual linking.
$unmatched[] = $download['title'] . ' (kein Dokumenttyp-Suffix)';
continue;
}
// Aliases and series rules may map to several records on purpose;
// generic title matches must stay unique.
$multiAllowed = false;
$candidates = [];
$aliasTitle = self::ALIASES[$this->normalize($name)] ?? null;
if ($aliasTitle === null) {
foreach (self::SERIES_RULES as $pattern => $title) {
if (preg_match($pattern, mb_strtolower($name))) {
$aliasTitle = $title;
break;
}
}
}
if ($aliasTitle !== null) {
$candidates = $productByNorm[$this->normalize($aliasTitle)] ?? [];
$multiAllowed = true;
}
if ($candidates === []) {
$candidates = $this->matchProduct($name, $productByNorm);
}
if ($candidates === []) {
$unmatched[] = $download['title'];
continue;
}
if (count($candidates) > 1 && !$multiAllowed) {
$ambiguous[] = $download['title'];
continue;
}
foreach ($candidates as $productUid) {
if (in_array($downloadUid, $existing[$productUid] ?? [], true)) {
continue; // already linked
}
$additions[$productUid][] = $downloadUid;
$output->writeln(sprintf('%s "%s" (dl %d) -> %d %s',
$dryRun ? 'plan ' : 'link ', $download['title'], $downloadUid, $productUid, $titleByUid[$productUid]));
}
}
foreach ($unmatched as $title) {
$output->writeln('<comment>kein Treffer: ' . $title . '</comment>');
}
foreach ($ambiguous as $title) {
$output->writeln('<comment>MEHRDEUTIG: ' . $title . '</comment>');
}
$pairCount = array_sum(array_map('count', $additions));
$output->writeln(sprintf("\n%d neue Verknuepfung(en) auf %d Produkt(e); %d ohne Treffer, %d mehrdeutig.",
$pairCount, count($additions), count($unmatched), count($ambiguous)));
if ($additions === []) {
return Command::SUCCESS;
}
if ($dryRun) {
$output->writeln('DRY RUN - nichts geschrieben.');
return Command::SUCCESS;
}
$datamap = [];
foreach ($additions as $productUid => $downloadUids) {
$merged = array_merge($existing[$productUid] ?? [], $downloadUids);
$datamap[self::PRODUCT_TABLE][$productUid]['downloads'] = implode(',', array_unique($merged));
}
$dataHandler = GeneralUtility::makeInstance(DataHandler::class);
$dataHandler->start($datamap, []);
$dataHandler->process_datamap();
if ($dataHandler->errorLog !== []) {
foreach ($dataHandler->errorLog as $error) {
$output->writeln('<error>' . $error . '</error>');
}
return Command::FAILURE;
}
$output->writeln('Fertig. Cache leeren: vendor/bin/typo3 cache:flush');
return Command::SUCCESS;
}
/** Title without its trailing document-type phrase; null when none found. */
private function stripDocSuffix(string $title): ?string
{
$norm = mb_strtolower(trim($title));
foreach (self::DOC_SUFFIXES as $suffix) {
if (str_ends_with($norm, $suffix)) {
return trim(mb_substr(trim($title), 0, mb_strlen(trim($title)) - mb_strlen($suffix)));
}
}
return null;
}
/**
* Unique product uids matching the document name, applying the passes
* described in the class comment.
*
* @param array<string,int[]> $productByNorm
* @return int[]
*/
private function matchProduct(string $name, array $productByNorm): array
{
$variants = [$name];
// leading brand word
$stripped = preg_replace('/^(vitec|datapath)\s+/i', '', $name);
if ($stripped !== $name) {
$variants[] = $stripped;
}
// trailing role word (on both variants)
foreach ($variants as $variant) {
foreach (self::ROLE_SUFFIXES as $role) {
if (preg_match('/\s+' . preg_quote($role, '/') . '$/i', $variant)) {
$variants[] = trim(preg_replace('/\s+' . preg_quote($role, '/') . '$/i', '', $variant));
}
}
}
foreach ($variants as $variant) {
$norm = $this->normalize($variant);
// aliases also apply to the stripped variants ("VisionSC SDI4
// Capture Card" -> "VisionSC SDI4" -> alias -> VisionSC-SD14)
$aliasTitle = self::ALIASES[$norm] ?? null;
if ($aliasTitle !== null) {
$norm = $this->normalize($aliasTitle);
}
$hit = $productByNorm[$norm] ?? [];
if ($hit !== []) {
return $hit;
}
}
return [];
}
/**
* lowercase, alphanumerics only - "TX1/F" and "tx1 f" compare equal.
* "+" survives as "plus", otherwise "MGW Diamond+ OG" and
* "MGW Diamond OG" would collide.
*/
private function normalize(string $value): string
{
return (string)preg_replace('/[^a-z0-9]+/', '', str_replace('+', 'plus', mb_strtolower($value)));
}
}