Protokoll #3694

{
    "event": "PreToolUse",
    "tool_name": "Edit",
    "tool_input": {
        "file_path": "\/var\/www\/dev.campus.systemische-tools.de\/src\/Infrastructure\/Docs\/HybridSearchService.php",
        "old_string": "self::QDRANT_HOST",
        "new_string": "$this->qdrantHost",
        "replace_all": true
    }
}

{
    "tool_response": {
        "filePath": "\/var\/www\/dev.campus.systemische-tools.de\/src\/Infrastructure\/Docs\/HybridSearchService.php",
        "oldString": "self::QDRANT_HOST",
        "newString": "$this->qdrantHost",
        "originalFile": "<?php\n\ndeclare(strict_types=1);\n\nnamespace Infrastructure\\Docs;\n\nuse Infrastructure\\AI\\OllamaService;\nuse Infrastructure\\Config\\CredentialService;\nuse Infrastructure\\Traits\\JsonDecodeTrait;\nuse PDO;\nuse RuntimeException;\n\n\/**\n * Hybrid Search Service combining vector search with SQL filtering.\n *\n * Implements a two-stage search:\n * 1. Semantic search via Qdrant (vector similarity)\n * 2. Structured filtering via MariaDB (taxonomy, entities, keywords)\n *\n * This approach achieves 70-85% precision vs 30-40% for pure vector search.\n *\/\nfinal class HybridSearchService\n{\n    use JsonDecodeTrait;\n\n    private const string COLLECTION = 'dokumentation_chunks';\n    private const int TIMEOUT = 30;\n\n    private PDO $pdo;\n    private OllamaService $ollama;\n    private string $qdrantHost;\n\n    public function __construct()\n    {\n        $this->ollama = new OllamaService();\n        $this->pdo = $this->createConnection();\n        $this->qdrantHost = CredentialService::getQdrantHost();\n    }\n\n    \/**\n     * Performs a hybrid search combining semantic and structured filtering.\n     *\n     * @param string $query The search query\n     * @param array{\n     *     taxonomy_category?: string,\n     *     taxonomy_path?: array<string>,\n     *     entity_type?: string,\n     *     entity_name?: string,\n     *     keyword?: string,\n     *     min_score?: float\n     * } $filters Optional structured filters\n     * @param int $limit Maximum results\n     * @return array<array{\n     *     chunk_id: int,\n     *     doc_id: int,\n     *     path: string,\n     *     title: string,\n     *     content: string,\n     *     heading_path: array<string>,\n     *     taxonomy: array<string>,\n     *     entities: array<mixed>,\n     *     keywords: array<string>,\n     *     score: float,\n     *     relevance_score: float\n     * }>\n     *\/\n    public function search(string $query, array $filters = [], int $limit = 10): array\n    {\n        \/\/ Stage 1: Semantic search in Qdrant\n        $vectorResults = $this->semanticSearch($query, $filters, $limit * 3);\n\n        if (empty($vectorResults)) {\n            return [];\n        }\n\n        \/\/ Stage 2: Enrich with SQL data and apply filters\n        $enrichedResults = $this->enrichAndFilter($vectorResults, $filters);\n\n        \/\/ Stage 3: Re-rank based on combined score\n        $rankedResults = $this->rerank($enrichedResults, $query);\n\n        return array_slice($rankedResults, 0, $limit);\n    }\n\n    \/**\n     * Searches within a specific taxonomy category.\n     *\n     * @return array<array<string, mixed>>\n     *\/\n    public function searchByCategory(string $query, string $category, int $limit = 10): array\n    {\n        return $this->search($query, ['taxonomy_category' => $category], $limit);\n    }\n\n    \/**\n     * Searches for chunks containing a specific entity.\n     *\n     * @return array<array<string, mixed>>\n     *\/\n    public function searchByEntity(string $query, string $entityName, int $limit = 10): array\n    {\n        return $this->search($query, ['entity_name' => $entityName], $limit);\n    }\n\n    \/**\n     * Gets all available taxonomy categories.\n     *\n     * @return array<array{category: string, count: int}>\n     *\/\n    public function getTaxonomyCategories(): array\n    {\n        $stmt = $this->pdo->query('\n            SELECT taxonomy_category as category, COUNT(*) as count\n            FROM dokumentation_chunks\n            WHERE taxonomy_category IS NOT NULL\n            GROUP BY taxonomy_category\n            ORDER BY count DESC\n        ');\n\n        return $stmt->fetchAll(PDO::FETCH_ASSOC);\n    }\n\n    \/**\n     * Gets all entities grouped by type.\n     *\n     * @return array<string, array<string>>\n     *\/\n    public function getEntitiesByType(): array\n    {\n        $stmt = $this->pdo->query(\"\n            SELECT entities FROM dokumentation_chunks\n            WHERE entities IS NOT NULL AND entities != '[]'\n        \");\n\n        $byType = [];\n\n        foreach ($stmt->fetchAll(PDO::FETCH_ASSOC) as $row) {\n            $entities = $this->decodeJsonArray($row['entities'] ?? null);\n            foreach ($entities as $entity) {\n                if (isset($entity['name'], $entity['type'])) {\n                    $type = $entity['type'];\n                    if (!isset($byType[$type])) {\n                        $byType[$type] = [];\n                    }\n                    if (!in_array($entity['name'], $byType[$type], true)) {\n                        $byType[$type][] = $entity['name'];\n                    }\n                }\n            }\n        }\n\n        return $byType;\n    }\n\n    \/**\n     * Suggests related searches based on current results.\n     *\n     * @param array<array<string, mixed>> $results\n     * @return array<string>\n     *\/\n    public function suggestRelatedSearches(array $results): array\n    {\n        $suggestions = [];\n\n        foreach ($results as $result) {\n            \/\/ Add keywords from results\n            foreach ($result['keywords'] ?? [] as $keyword) {\n                if (!in_array($keyword, $suggestions, true)) {\n                    $suggestions[] = $keyword;\n                }\n            }\n\n            \/\/ Add entity names\n            foreach ($result['entities'] ?? [] as $entity) {\n                if (isset($entity['name']) && !in_array($entity['name'], $suggestions, true)) {\n                    $suggestions[] = $entity['name'];\n                }\n            }\n        }\n\n        return array_slice($suggestions, 0, 5);\n    }\n\n    \/**\n     * Performs semantic search in Qdrant.\n     *\n     * @param array<string, mixed> $filters\n     * @return array<array{id: string, score: float, payload: array<string, mixed>}>\n     *\/\n    private function semanticSearch(string $query, array $filters, int $limit): array\n    {\n        $embedding = $this->ollama->getEmbedding($query);\n\n        $url = sprintf('%s\/collections\/%s\/points\/search', self::QDRANT_HOST, self::COLLECTION);\n\n        $payload = [\n            'vector' => array_values($embedding),\n            'limit' => $limit,\n            'with_payload' => true,\n        ];\n\n        \/\/ Add Qdrant filter if taxonomy category specified\n        if (isset($filters['taxonomy_category'])) {\n            $payload['filter'] = [\n                'must' => [\n                    [\n                        'key' => 'taxonomy_category',\n                        'match' => ['value' => $filters['taxonomy_category']],\n                    ],\n                ],\n            ];\n        }\n\n        try {\n            $response = $this->makeRequest($url, $payload, 'POST');\n\n            if (!isset($response['result']) || !is_array($response['result'])) {\n                return [];\n            }\n\n            return array_map(static function (array $item): array {\n                return [\n                    'id' => (string) $item['id'],\n                    'score' => (float) ($item['score'] ?? 0),\n                    'payload' => is_array($item['payload'] ?? null) ? $item['payload'] : [],\n                ];\n            }, $response['result']);\n        } catch (RuntimeException) {\n            return [];\n        }\n    }\n\n    \/**\n     * Enriches vector results with SQL data and applies additional filters.\n     *\n     * @param array<array{id: string, score: float, payload: array<string, mixed>}> $vectorResults\n     * @param array<string, mixed> $filters\n     * @return array<array<string, mixed>>\n     *\/\n    private function enrichAndFilter(array $vectorResults, array $filters): array\n    {\n        $results = [];\n        $minScore = $filters['min_score'] ?? 0.3;\n\n        foreach ($vectorResults as $vr) {\n            \/\/ Apply minimum score filter\n            if ($vr['score'] < $minScore) {\n                continue;\n            }\n\n            $chunkId = (int) ($vr['payload']['chunk_id'] ?? 0);\n            if ($chunkId === 0) {\n                continue;\n            }\n\n            \/\/ Get full chunk data from DB\n            $chunk = $this->getChunkWithDocument($chunkId);\n            if ($chunk === null) {\n                continue;\n            }\n\n            \/\/ Apply entity filter\n            if (isset($filters['entity_name'])) {\n                $entities = $this->decodeJsonArray($chunk['entities'] ?? null);\n                $found = false;\n                foreach ($entities as $entity) {\n                    if (isset($entity['name']) && stripos($entity['name'], $filters['entity_name']) !== false) {\n                        $found = true;\n                        break;\n                    }\n                }\n                if (!$found) {\n                    continue;\n                }\n            }\n\n            \/\/ Apply entity type filter\n            if (isset($filters['entity_type'])) {\n                $entities = $this->decodeJsonArray($chunk['entities'] ?? null);\n                $found = false;\n                foreach ($entities as $entity) {\n                    if (isset($entity['type']) && strtoupper($entity['type']) === strtoupper($filters['entity_type'])) {\n                        $found = true;\n                        break;\n                    }\n                }\n                if (!$found) {\n                    continue;\n                }\n            }\n\n            \/\/ Apply keyword filter\n            if (isset($filters['keyword'])) {\n                $keywords = $this->decodeJsonArray($chunk['keywords'] ?? null);\n                $found = false;\n                foreach ($keywords as $kw) {\n                    if (stripos($kw, $filters['keyword']) !== false) {\n                        $found = true;\n                        break;\n                    }\n                }\n                if (!$found) {\n                    continue;\n                }\n            }\n\n            $results[] = [\n                'chunk_id' => $chunkId,\n                'doc_id' => (int) $chunk['dokumentation_id'],\n                'path' => $chunk['doc_path'] ?? '',\n                'title' => $chunk['doc_title'] ?? '',\n                'content' => $chunk['content_clean'] ?? $chunk['content'] ?? '',\n                'heading_path' ... [TRUNCATED-6f0abc969923a33c]",
        "structuredPatch": [
            {
                "oldStart": 191,
                "oldLines": 7,
                "newStart": 191,
                "newLines": 7,
                "lines": [
                    "     {",
                    "         $embedding = $this->ollama->getEmbedding($query);",
                    " ",
                    "-        $url = sprintf('%s\/collections\/%s\/points\/search', self::QDRANT_HOST, self::COLLECTION);",
                    "+        $url = sprintf('%s\/collections\/%s\/points\/search', $this->qdrantHost, self::COLLECTION);",
                    " ",
                    "         $payload = [",
                    "             'vector' => array_values($embedding),"
                ]
            }
        ],
        "userModified": false,
        "replaceAll": true
    }
}