optimize data retrieve by customfields und enricher

2026-04-13 16:11:19 +02:00
parent 0a05ccaee3
commit f7685c6fb5
10 changed files with 234 additions and 32 deletions
--- a/src/Agent/AgentRunner.php
+++ b/src/Agent/AgentRunner.php
@@ -58,7 +58,7 @@ final readonly class AgentRunner
            //$includeFullContext = false;
            if ($includeFullContext) {

-                yield $this->systemMsg("Ich analyse deine Anfrage...", "think");
+                yield $this->systemMsg("Ich analysiere deine Anfrage...", "think");

                $promptSwagSearch = '
                Erzeuge aus dem folgenden Nutzereingabetext einen kurzen Suchtext für die Shopware-6-Suche.
@@ -80,6 +80,7 @@ final readonly class AgentRunner
                ';

                $this->thinkSuppressor->reset();
+
                foreach ($this->ollamaClient->stream($promptSwagSearch) as $swagToken) {

                    if (!is_string($swagToken)) {
@@ -138,7 +139,6 @@ final readonly class AgentRunner

            yield $this->systemMsg("Denke nach...", "think");

-
            // ---------------------------------------------------------
            // 5) Build final prompt
            // ---------------------------------------------------------
--- a/src/Agent/PromptBuilder.php
+++ b/src/Agent/PromptBuilder.php
@@ -75,6 +75,7 @@ final readonly class PromptBuilder
        // 3) LIVE SHOP RESULTS (AUTHORITATIVE FOR PRODUCTS)
        // ------------------------------------------------------------
        $shopBlock = '';
+        $isDetailed = !(count($shopResults) > 5);

        if ($shopResults !== []) {
            $lines = [];
@@ -113,6 +114,18 @@ final readonly class PromptBuilder
                    $parts[] = "URL: " . $product->url;
                }

+                if ($product->productImage) {
+                    $parts[] = "productImage: " . $product->productImage;
+                }
+
+                if ($isDetailed && $product->description) {
+                    $parts[] = "description: " . $product->description;
+                }
+
+                if ($product->customFields) {
+                    $parts[] = "Meta-Informationen: " . $product->customFields;
+                }
+
                $lines[] = implode("\n", $parts);
            }

--- a/src/Commerce/Dto/ShopProductResult.php
+++ b/src/Commerce/Dto/ShopProductResult.php
@@ -10,15 +10,18 @@ final readonly class ShopProductResult
     * @param string[] $highlights
     */
    public function __construct(
-        public string $id,
-        public string $name,
+        public string  $id,
+        public string  $name,
        public ?string $productNumber = null,
        public ?string $manufacturer = null,
        public ?string $price = null,
-        public ?bool $available = null,
+        public ?bool   $available = null,
        public ?string $url = null,
-        public array $highlights = [],
-        public ?string $description = null
-    ) {
+        public array   $highlights = [],
+        public ?string $description = null,
+        public ?string $productImage = null,
+        public ?string $customFields = null,
+    )
+    {
    }
 }
--- a/src/Commerce/ShopSearchService.php
+++ b/src/Commerce/ShopSearchService.php
@@ -7,6 +7,11 @@ namespace App\Commerce;
 use App\Commerce\Dto\ShopProductResult;
 use App\Shopware\ShopwareCriteriaBuilder;
 use App\Shopware\StoreApiClient;
+use Symfony\Contracts\HttpClient\Exception\ClientExceptionInterface;
+use Symfony\Contracts\HttpClient\Exception\DecodingExceptionInterface;
+use Symfony\Contracts\HttpClient\Exception\RedirectionExceptionInterface;
+use Symfony\Contracts\HttpClient\Exception\ServerExceptionInterface;
+use Symfony\Contracts\HttpClient\Exception\TransportExceptionInterface;

 final readonly class ShopSearchService
 {
@@ -29,12 +34,18 @@ final readonly class ShopSearchService
        if (!$this->enabled) {
            return [];
        }
-
+        $response = [];
        $query = $this->queryParser->parse($originalPrompt, $commerceIntent);
        $criteria = $this->criteriaBuilder->build($query, $this->maxResults);
-        $response = $this->storeApiClient->searchProducts($criteria);
+        try {
+            $response = $this->storeApiClient->searchProducts($criteria);
+        } catch (ClientExceptionInterface|DecodingExceptionInterface|RedirectionExceptionInterface|ServerExceptionInterface|TransportExceptionInterface $e) {

-        return $this->mapProducts($response);
+        }
+
+        $result = $this->mapProducts($response);;
+
+        return $result;
    }

    /**
@@ -63,6 +74,8 @@ final readonly class ShopSearchService
                url: $this->baseUrl . $this->extractUrl($row),
                highlights: $this->extractHighlights($row),
                description: $this->cleanUpDescription($row),
+                productImage: $row['cover']['media']['thumbnails'][0]['url'] ?? 'no-image',
+                customFields: $this->getRelevantCustomFields($row['customFields'])
            );
        }

@@ -72,6 +85,15 @@ final readonly class ShopSearchService
        ));
    }

+    private function getRelevantCustomFields($customField): string
+    {
+        $result = ($customField['migration_Backup_product_attr1'] ?? '') . ': ' . ($customField['migration_Backup_product_attr2'] ?? '');
+        $result .= ' | Einsatzgebiete: ' . ($customField['migration_Backup_product_attr4'] ?? '');
+        $result .= ' | Sprachen: ' . ($customField['migration_Backup_product_attr5'] ?? '');
+
+        return $result;
+    }
+
    private function cleanUpDescription($description): string
    {
        if (isset($description['translated']['description'])) {
--- a/src/Knowledge/Retrieval/NdjsonHybridRetriever.php
+++ b/src/Knowledge/Retrieval/NdjsonHybridRetriever.php
@@ -41,8 +41,11 @@ final class NdjsonHybridRetriever implements RetrieverInterface
        private readonly SalesIntentLite                 $salesIntentLite,
        private readonly CatalogIntentLite               $catalogIntent,
        private readonly IntentRouteResolver             $routeResolver,
-        private readonly EntityCatalogService            $entityCatalogService
-    ) {}
+        private readonly EntityCatalogService            $entityCatalogService,
+        private readonly QueryEnricher                   $queryEnricher,
+    )
+    {
+    }

    // =========================================================
    // PUBLIC API
@@ -126,10 +129,11 @@ final class NdjsonHybridRetriever implements RetrieverInterface
    // =========================================================

    private function execute(
-        string $prompt,
+        string                $prompt,
        ModelGenerationConfig $config,
-        bool $withScores
-    ): array {
+        bool                  $withScores
+    ): array
+    {

        $entityLabel = $this->catalogIntent->detect($prompt);
        $salesIntent = $this->detectSalesIntent($prompt);
@@ -195,11 +199,12 @@ final class NdjsonHybridRetriever implements RetrieverInterface
    // =========================================================

    private function runCore(
-        string $prompt,
+        string                $prompt,
        ModelGenerationConfig $config,
-        bool $withScores,
-        string $salesIntent
-    ): array {
+        bool                  $withScores,
+        string                $salesIntent
+    ): array
+    {

        $limit = max(1, min($config->getRetrievalMaxChunks(), self::HARD_MAX_CHUNKS));
        $vectorTopKBase = max(1, min($config->getRetrievalVectorTopK(), self::HARD_MAX_VECTORK));
@@ -207,6 +212,7 @@ final class NdjsonHybridRetriever implements RetrieverInterface
        $isListQuery = $this->intentLite->isListQuery($prompt);

        $cleanQuery = $this->queryCleaner->clean($prompt);
+        $cleanQuery = $this->queryEnricher->enrichPrompt($cleanQuery);

        if ($cleanQuery === '') {
            return [
@@ -316,9 +322,10 @@ final class NdjsonHybridRetriever implements RetrieverInterface

    private function computeThresholdAndTopK(
        string $salesIntent,
-        bool $isListQuery,
-        int $vectorTopKBase
-    ): array {
+        bool   $isListQuery,
+        int    $vectorTopKBase
+    ): array
+    {

        $threshold = self::VECTOR_SCORE_THRESHOLD;
        $topK = $vectorTopKBase;
@@ -344,9 +351,10 @@ final class NdjsonHybridRetriever implements RetrieverInterface
        array $globalHits,
        array $scopedHits,
        float $threshold,
-        bool $boostScoped,
-        bool $captureRaw
-    ): array {
+        bool  $boostScoped,
+        bool  $captureRaw
+    ): array
+    {

        $rrfScores = [];
        $rawScores = [];
--- a/src/Knowledge/Retrieval/QueryEnricher.php
+++ b/src/Knowledge/Retrieval/QueryEnricher.php
@@ -0,0 +1,128 @@
+<?php
+
+declare(strict_types=1);
+
+namespace App\Knowledge\Retrieval;
+
+final class QueryEnricher
+{
+    public function enrichPrompt(string $query): string
+    {
+        // Return early if the input is empty or contains only whitespace.
+        if (trim($query) === '') {
+            return '';
+        }
+
+        // Keep the original query untouched for the final output.
+        $originalQuery = $query;
+
+        // Normalize the query for case-insensitive matching.
+        $normalizedQuery = $this->normalize($query);
+
+        // Expect an associative array like:
+        // [
+        //     'hose'  => 'jeans',
+        //     'jacke' => 'mantel',
+        // ]
+        $mapping = $this->enrichQueryList();
+
+        // Build a bidirectional lookup table:
+        // key   -> value
+        // value -> key
+        $lookup = $this->buildBidirectionalLookup($mapping);
+
+        // Split the query into searchable words/tokens.
+        $tokens = $this->tokenize($normalizedQuery);
+
+        $matches = [];
+
+        foreach ($tokens as $token) {
+            // If the token exists in the lookup table, add the mapped counterpart.
+            if (isset($lookup[$token])) {
+                $matches[] = $lookup[$token];
+            }
+        }
+
+        // Remove duplicates while preserving order.
+        $matches = array_values(array_unique($matches));
+
+        // If nothing was found, return the original query unchanged.
+        if ($matches === []) {
+            return $originalQuery;
+        }
+
+        // Append the matched counterpart terms to the original prompt.
+        return $originalQuery . " | Pseudonyme: " . implode(', ', $matches);
+    }
+
+    /**
+     * Normalize a string for case-insensitive comparison.
+     */
+    private function normalize(string $value): string
+    {
+        return mb_strtolower(trim($value), 'UTF-8');
+    }
+
+    /**
+     * Tokenize the query into words.
+     * Splits on everything that is not a letter or number.
+     */
+    private function tokenize(string $value): array
+    {
+        return preg_split('/[^\p{L}\p{N}]+/u', $value, -1, PREG_SPLIT_NO_EMPTY) ?: [];
+    }
+
+    /**
+     * Build a lookup table that works in both directions.
+     *
+     * Example:
+     * [
+     *     'hose'  => 'jeans',
+     *     'jacke' => 'mantel',
+     * ]
+     *
+     * becomes:
+     * [
+     *     'hose'   => 'jeans',
+     *     'jeans'  => 'hose',
+     *     'jacke'  => 'mantel',
+     *     'mantel' => 'jacke',
+     * ]
+     */
+    private function buildBidirectionalLookup(array $mapping): array
+    {
+        $lookup = [];
+
+        foreach ($mapping as $key => $value) {
+            $key = trim((string)$key);
+            $value = trim((string)$value);
+
+            // Skip incomplete pairs.
+            if ($key === '' || $value === '') {
+                continue;
+            }
+
+            $normalizedKey = $this->normalize($key);
+            $normalizedValue = $this->normalize($value);
+
+            // If the key is found in the query, return the value.
+            $lookup[$normalizedKey] = $value;
+
+            // If the value is found in the query, return the key.
+            $lookup[$normalizedValue] = $key;
+        }
+
+        return $lookup;
+    }
+
+    public function enrichQueryList(): array
+    {
+        return [
+            'Wasserhärte' => "Resthärte",
+            'Gerät' => 'Modell',
+            'Indikator' => 'Chemie',
+            'Wasserhärte-Grenzwert'=>'Resthärte',
+            'Resthärte-Grenzwert'=>'Wasserhärte'
+        ];
+    }
+}
--- a/src/Shopware/ShopwareCriteriaBuilder.php
+++ b/src/Shopware/ShopwareCriteriaBuilder.php
@@ -28,7 +28,9 @@ final class ShopwareCriteriaBuilder
                    'calculatedPrice',
                    'seoUrls',
                    'manufacturer',
-                    'translated.name'
+                    'translated.name',
+                    'cover',
+                    'customFields'
                ],
                'product_manufacturer' => [
                    'name',
@@ -43,10 +45,30 @@ final class ShopwareCriteriaBuilder
                'seo_url' => [
                    'seoPathInfo',
                ],
+                'product_media' => [
+                    'id',
+                    'media'
+                ],
+                'media' => [
+                    'id',
+                    'url',
+                    'thumbnails',
+                    'alt',
+                    'title'
+                ]
            ],
            'associations' => [
                'manufacturer' => new \stdClass(),
                'seoUrls' => new \stdClass(),
+                'cover' => [
+                    'associations' => [
+                        'media' => [
+                            'associations' => [
+                                "thumbnails" => new \stdClass()
+                            ]
+                        ]
+                    ]
+                ]
            ],
            'sort' => [
                [