harden queries

src/Commerce/CommerceQueryParser.php

@@ -5,77 +5,37 @@ declare(strict_types=1);
 namespace App\Commerce;
 
 use App\Commerce\Dto\CommerceSearchQuery;
+use App\Knowledge\Retrieval\QueryCleaner;
+use App\Knowledge\Text\TextNormalizer;
 
 final class CommerceQueryParser
 {
-    /**
-     * @var string[]
-     */
-    private array $knownColors = [
-        'schwarz',
-        'weiß',
-        'weiss',
-        'rot',
-        'blau',
-        'grün',
-        'gruen',
-        'gelb',
-        'grau',
-        'beige',
-        'rosa',
-        'pink',
-        'orange',
-        'braun',
-    ];
+    public function __construct(
+        private readonly TextNormalizer $textNormalizer,
+        private readonly QueryCleaner   $queryCleaner,
+    )
+    {
 
-    /**
-     * @var string[]
-     */
-    private array $knownCategories = [
-        'sneaker',
-        'schuhe',
-        'hoodie',
-        't-shirt',
-        'shirt',
-        'jacke',
-        'regenjacke',
-        'trinkflasche',
-        'flasche',
-        'rucksack',
-        'tasche',
-        'mütze',
-        'muetze',
-        'kappe',
-        'hose',
-        'pullover',
-    ];
+    }
 
     /**
      * @var string[]
      */
     private array $knownBrands = [
-        'nike',
-        'adidas',
-        'puma',
-        'reebok',
-        'under armour',
-        'new balance',
+        'heyl',
+        'horiba'
     ];
 
     public function parse(string $originalPrompt, string $intent): CommerceSearchQuery
     {
         $normalized = $this->normalize($originalPrompt);
-
         [$priceMin, $priceMax] = $this->extractPriceRange($normalized);
         $sizes = $this->extractSizes($normalized);
-        $colors = $this->extractColors($normalized);
         $brand = $this->extractBrand($normalized);
-        $category = $this->extractCategory($normalized);
         $properties = [];
 
         $searchText = $this->buildSearchText(
             $normalized,
-            $colors,
             $sizes,
             $brand,
             $priceMin,
@@ -86,9 +46,7 @@ final class CommerceQueryParser
             originalPrompt: $originalPrompt,
             normalizedPrompt: $normalized,
             searchText: $searchText !== '' ? $searchText : $normalized,
-            category: $category,
             brand: $brand,
-            colors: $colors,
             sizes: $sizes,
             properties: $properties,
             priceMin: $priceMin,
@@ -100,7 +58,9 @@ final class CommerceQueryParser
 
     private function normalize(string $prompt): string
     {
-        $value = mb_strtolower(trim($prompt));
+        $value = $this->textNormalizer->normalize($prompt);
+        $value = $this->queryCleaner->clean($value);
+        $value = mb_strtolower(trim($value));
         $value = str_replace(['€'], ' euro ', $value);
         $value = preg_replace('/[^\p{L}\p{N}\s.,\-]/u', ' ', $value) ?? $value;
         $value = preg_replace('/\s+/u', ' ', $value) ?? $value;
@@ -157,23 +117,7 @@ final class CommerceQueryParser
             }
         }
 
-        return array_values(array_unique(array_filter($sizes, static fn ($v) => $v !== '')));
-    }
-
-    /**
-     * @return string[]
-     */
-    private function extractColors(string $prompt): array
-    {
-        $colors = [];
-
-        foreach ($this->knownColors as $color) {
-            if (preg_match('/\b' . preg_quote($color, '/') . '\b/u', $prompt) === 1) {
-                $colors[] = $color;
-            }
-        }
-
-        return array_values(array_unique($colors));
+        return array_values(array_unique(array_filter($sizes, static fn($v) => $v !== '')));
     }
 
     private function extractBrand(string $prompt): ?string
@@ -184,32 +128,21 @@ final class CommerceQueryParser
             }
         }
 
-        if (preg_match('/\bmarke\s+([a-z0-9][a-z0-9\s\-]+)/u', $prompt, $m) === 1) {
+        if (preg_match('/\bheyl\s+([a-z0-9][a-z0-9\s\-]+)/u', $prompt, $m) === 1) {
             return trim($m[1]);
         }
 
         return null;
     }
 
-    private function extractCategory(string $prompt): ?string
-    {
-        foreach ($this->knownCategories as $category) {
-            if (preg_match('/\b' . preg_quote($category, '/') . '\b/u', $prompt) === 1) {
-                return $category;
-            }
-        }
-
-        return null;
-    }
-
     private function buildSearchText(
-        string $prompt,
-        array $colors,
-        array $sizes,
+        string  $prompt,
+        array   $sizes,
         ?string $brand,
-        ?float $priceMin,
-        ?float $priceMax
-    ): string {
+        ?float  $priceMin,
+        ?float  $priceMax
+    ): string
+    {
         $text = ' ' . $prompt . ' ';
 
         $phrasesToRemove = [
@@ -231,10 +164,6 @@ final class CommerceQueryParser
             $text = str_replace($phrase, ' ', $text);
         }
 
-        foreach ($colors as $color) {
-            $text = preg_replace('/\b' . preg_quote($color, '/') . '\b/u', ' ', $text) ?? $text;
-        }
-
         foreach ($sizes as $size) {
             $text = preg_replace('/\b' . preg_quote($size, '/') . '\b/u', ' ', $text) ?? $text;
         }
@@ -253,7 +182,7 @@ final class CommerceQueryParser
 
         $text = preg_replace('/\s+/u', ' ', $text) ?? $text;
         $text = trim($text, " \t\n\r\0\x0B-.,");
-        $tokens = array_filter(explode(' ', $text), static fn (string $token): bool => mb_strlen($token) > 1);
+        $tokens = array_filter(explode(' ', $text), static fn(string $token): bool => mb_strlen($token) > 1);
 
         return trim(implode(' ', $tokens));
     }
@@ -262,6 +191,6 @@ final class CommerceQueryParser
     {
         $value = str_replace(',', '.', trim($value));
 
-        return is_numeric($value) ? (float) $value : null;
+        return is_numeric($value) ? (float)$value : null;
     }
 }