Add UPC, SKU, and manufacturer index builder

2026-07-15 11:15:57 -05:00
parent c0c74dfb6b
commit 085d1c2352
5 changed files with 162 additions and 0 deletions
--- a/tools/build-indexes.php
+++ b/tools/build-indexes.php
@@ -0,0 +1,140 @@
+<?php
+
+declare(strict_types=1);
+
+$root = dirname(__DIR__);
+$catalogDir = $root . '/catalog';
+$indexesDir = $root . '/indexes';
+
+$upcIndex = [];
+$skuIndex = [];
+$manufacturerIndex = [];
+$errors = [];
+$checked = 0;
+
+if (!is_dir($catalogDir)) {
+    fwrite(STDERR, "Catalog directory not found: {$catalogDir}\n");
+    exit(1);
+}
+
+if (!is_dir($indexesDir) && !mkdir($indexesDir, 0775, true) && !is_dir($indexesDir)) {
+    fwrite(STDERR, "Unable to create indexes directory: {$indexesDir}\n");
+    exit(1);
+}
+
+$iterator = new RecursiveIteratorIterator(
+    new RecursiveDirectoryIterator(
+        $catalogDir,
+        FilesystemIterator::SKIP_DOTS
+    )
+);
+
+foreach ($iterator as $file) {
+    if (!$file->isFile() || strtolower($file->getExtension()) !== 'json') {
+        continue;
+    }
+
+    $checked++;
+    $path = $file->getPathname();
+    $relativePath = substr($path, strlen($root) + 1);
+
+    $raw = file_get_contents($path);
+
+    if ($raw === false) {
+        $errors[] = "{$relativePath}: unable to read file";
+        continue;
+    }
+
+    try {
+        $record = json_decode($raw, true, 512, JSON_THROW_ON_ERROR);
+    } catch (JsonException $exception) {
+        $errors[] = "{$relativePath}: invalid JSON: {$exception->getMessage()}";
+        continue;
+    }
+
+    if (!is_array($record)) {
+        $errors[] = "{$relativePath}: root value must be an object";
+        continue;
+    }
+
+    $upc = $record['upc'] ?? null;
+    $sku = $record['sku'] ?? null;
+    $manufacturer = $record['manufacturer'] ?? null;
+
+    if (is_string($upc) && $upc !== '') {
+        if (isset($upcIndex[$upc])) {
+            $errors[] = "{$relativePath}: duplicate UPC {$upc}, already used by {$upcIndex[$upc]}";
+        } else {
+            $upcIndex[$upc] = $relativePath;
+        }
+    }
+
+    if (
+        is_string($manufacturer) &&
+        trim($manufacturer) !== ''
+    ) {
+        $manufacturerKey = strtolower(trim($manufacturer));
+
+        if (!isset($manufacturerIndex[$manufacturerKey])) {
+            $manufacturerIndex[$manufacturerKey] = [];
+        }
+
+        $manufacturerIndex[$manufacturerKey][] = $relativePath;
+
+        if (is_string($sku) && trim($sku) !== '') {
+            $skuKey = $manufacturerKey . ':' . strtolower(trim($sku));
+
+            if (isset($skuIndex[$skuKey])) {
+                $errors[] = "{$relativePath}: duplicate SKU key {$skuKey}, already used by {$skuIndex[$skuKey]}";
+            } else {
+                $skuIndex[$skuKey] = $relativePath;
+            }
+        }
+    }
+}
+
+if ($errors !== []) {
+    fwrite(STDERR, "Index build failed:\n\n");
+
+    foreach ($errors as $error) {
+        fwrite(STDERR, " - {$error}\n");
+    }
+
+    fwrite(STDERR, "\nChecked {$checked} product file(s).\n");
+    exit(1);
+}
+
+ksort($upcIndex, SORT_STRING);
+ksort($skuIndex, SORT_STRING);
+ksort($manufacturerIndex, SORT_STRING);
+
+foreach ($manufacturerIndex as &$paths) {
+    sort($paths, SORT_STRING);
+}
+unset($paths);
+
+$files = [
+    $indexesDir . '/upc-index.json' => $upcIndex,
+    $indexesDir . '/sku-index.json' => $skuIndex,
+    $indexesDir . '/manufacturer-index.json' => $manufacturerIndex,
+];
+
+foreach ($files as $outputPath => $data) {
+    $json = json_encode(
+        $data,
+        JSON_PRETTY_PRINT |
+        JSON_UNESCAPED_SLASHES |
+        JSON_THROW_ON_ERROR
+    );
+
+    if (file_put_contents($outputPath, $json . PHP_EOL) === false) {
+        fwrite(STDERR, "Unable to write {$outputPath}\n");
+        exit(1);
+    }
+}
+
+echo "Index build complete.\n";
+echo "Products checked: {$checked}\n";
+echo "UPC entries: " . count($upcIndex) . "\n";
+echo "SKU entries: " . count($skuIndex) . "\n";
+echo "Manufacturers: " . count($manufacturerIndex) . "\n";