optimize cleanup search query shop api extends

src/Commerce/CommerceQueryParser.php

@@ -239,10 +239,148 @@ final readonly class CommerceQueryParser
         );
 
         $tokens = $this->normalizeSearchTokens($tokens);
+        $tokens = $this->compactShopSearchTokens($tokens);
 
         return trim(implode(' ', $tokens));
     }
 
+    /**
+     * Keep the Store API query narrow without relying on endless spelling-specific stop words.
+     *
+     * Direct product queries often contain user instructions such as "show all as a list".
+     * Shopware search performs best when the query only contains product-defining tokens:
+     * model numbers, the immediately related model name, brands, and semantic commerce terms.
+     *
+     * @param string[] $tokens
+     * @return string[]
+     */
+    private function compactShopSearchTokens(array $tokens): array
+    {
+        $tokens = array_values(array_filter(
+            $tokens,
+            fn(string $token): bool => !$this->isQueryNoiseToken($token)
+        ));
+
+        if ($tokens === []) {
+            return [];
+        }
+
+        $keep = [];
+
+        foreach ($tokens as $index => $token) {
+            if ($this->isModelNumberToken($token)) {
+                $keep[$index] = true;
+
+                for ($offset = 1; $offset <= $this->config->getModelContextTokenWindow(); $offset++) {
+                    $previousIndex = $index - $offset;
+
+                    if (!isset($tokens[$previousIndex]) || !$this->isLikelyModelContextToken($tokens[$previousIndex])) {
+                        break;
+                    }
+
+                    $keep[$previousIndex] = true;
+                }
+
+                $nextIndex = $index + 1;
+                if (isset($tokens[$nextIndex]) && $this->isModelSuffixToken($tokens[$nextIndex])) {
+                    $keep[$nextIndex] = true;
+                }
+            }
+
+            if ($this->isSemanticShopToken($token) || $this->isKnownBrandToken($token)) {
+                $keep[$index] = true;
+            }
+        }
+
+        if ($keep === []) {
+            return $this->limitShopSearchTokens($tokens);
+        }
+
+        ksort($keep);
+
+        $compacted = [];
+        foreach (array_keys($keep) as $index) {
+            $compacted[] = $tokens[$index];
+        }
+
+        return $this->limitShopSearchTokens(array_values(array_unique($compacted)));
+    }
+
+    private function isQueryNoiseToken(string $token): bool
+    {
+        $token = trim(mb_strtolower($token, 'UTF-8'));
+
+        if ($token === '') {
+            return true;
+        }
+
+        if (preg_match($this->config->getContainsDigitPattern(), $token) === 1) {
+            return false;
+        }
+
+        if (mb_strlen($token) <= $this->config->getMinMeaningfulAlphaTokenLength()) {
+            return true;
+        }
+
+        if ($this->isSearchControlToken($token)) {
+            return true;
+        }
+
+        return preg_match($this->config->getInstructionOrPresentationTokenPattern(), $token) === 1;
+    }
+
+    private function isModelNumberToken(string $token): bool
+    {
+        return preg_match($this->config->getModelNumberTokenPattern(), $token) === 1;
+    }
+
+    private function isLikelyModelContextToken(string $token): bool
+    {
+        if ($this->isQueryNoiseToken($token)) {
+            return false;
+        }
+
+        if ($this->isSemanticShopToken($token)) {
+            return false;
+        }
+
+        return preg_match($this->config->getModelContextTokenPattern(), $token) === 1;
+    }
+
+    private function isModelSuffixToken(string $token): bool
+    {
+        if ($this->isQueryNoiseToken($token)) {
+            return false;
+        }
+
+        return preg_match($this->config->getModelSuffixTokenPattern(), $token) === 1;
+    }
+
+    private function isSemanticShopToken(string $token): bool
+    {
+        return in_array($token, $this->config->getSemanticShopSearchTokens(), true);
+    }
+
+    private function isKnownBrandToken(string $token): bool
+    {
+        return in_array($token, $this->config->getKnownBrands(), true);
+    }
+
+    /**
+     * @param string[] $tokens
+     * @return string[]
+     */
+    private function limitShopSearchTokens(array $tokens): array
+    {
+        $limit = $this->config->getMaxShopSearchTokens();
+
+        if ($limit <= 0 || count($tokens) <= $limit) {
+            return $tokens;
+        }
+
+        return array_slice($tokens, 0, $limit);
+    }
+
     private function shouldUseHistoryContext(string $prompt): bool
     {
         return preg_match($this->config->getHistoryContextValuePattern(), $prompt) === 1;