Protokoll #24543

{
    "event": "PreToolUse",
    "tool_name": "Edit",
    "tool_input": {
        "file_path": "\/var\/www\/dev.campus.systemische-tools.de\/src\/Infrastructure\/Docs\/ChunkSyncService.php",
        "old_string": "        if (mb_strlen($content) > 1000) {\n            $content = mb_substr($content, 0, 1000) . '...';",
        "new_string": "        if (mb_strlen($content) > Constants::EMBEDDING_TEXT_LIMIT) {\n            $content = mb_substr($content, 0, Constants::EMBEDDING_TEXT_LIMIT) . '...';"
    }
}

{
    "tool_response": {
        "filePath": "\/var\/www\/dev.campus.systemische-tools.de\/src\/Infrastructure\/Docs\/ChunkSyncService.php",
        "oldString": "        if (mb_strlen($content) > 1000) {\n            $content = mb_substr($content, 0, 1000) . '...';",
        "newString": "        if (mb_strlen($content) > Constants::EMBEDDING_TEXT_LIMIT) {\n            $content = mb_substr($content, 0, Constants::EMBEDDING_TEXT_LIMIT) . '...';",
        "originalFile": "<?php\n\ndeclare(strict_types=1);\n\nnamespace Infrastructure\\Docs;\n\n\/\/ @responsibility: Synchronisiert Chunks mit Embeddings nach Qdrant\n\nuse Domain\\Constants;\nuse Infrastructure\\AI\\OllamaService;\nuse Infrastructure\\AI\\QdrantClient;\nuse Infrastructure\\Traits\\JsonDecodeTrait;\nuse PDO;\n\nfinal class ChunkSyncService\n{\n    use JsonDecodeTrait;\n\n    private const string COLLECTION = 'dokumentation_chunks';\n    private const int VECTOR_SIZE = 1024;\n    private const int BATCH_SIZE = 10;\n\n    public function __construct(\n        private PDO $pdo,\n        private OllamaService $ollama,\n        private QdrantClient $qdrant\n    ) {\n    }\n\n    \/**\n     * Ensures the Qdrant collection exists with proper configuration.\n     *\/\n    public function ensureCollection(): bool\n    {\n        return $this->qdrant->ensureCollection(self::COLLECTION, self::VECTOR_SIZE);\n    }\n\n    \/**\n     * Syncs a single chunk to Qdrant.\n     *\/\n    public function syncChunk(int $chunkId): bool\n    {\n        $chunk = $this->getChunk($chunkId);\n\n        if ($chunk === null) {\n            return false;\n        }\n\n        if ($chunk['analysis_status'] !== 'completed') {\n            return false;\n        }\n\n        $doc = $this->getDocument((int) $chunk['dokumentation_id']);\n        $text = $this->prepareTextForEmbedding($chunk, $doc);\n        $embedding = $this->ollama->getEmbedding($text);\n        $payload = $this->buildPayload($chunk, $doc);\n        $qdrantId = $chunk['qdrant_id'] ?? $this->qdrant->generateUuid();\n\n        $success = $this->qdrant->upsertPoint(self::COLLECTION, $qdrantId, $embedding, $payload);\n\n        if ($success && $chunk['qdrant_id'] === null) {\n            $this->updateQdrantId($chunkId, $qdrantId);\n        }\n\n        return $success;\n    }\n\n    \/**\n     * Syncs all analyzed chunks that haven't been synced yet.\n     *\n     * @return array{synced: int, failed: int, errors: array<string>}\n     *\/\n    public function syncAllPending(int $limit = Constants::DEFAULT_LIMIT): array\n    {\n        $this->ensureCollection();\n\n        $results = ['synced' => 0, 'failed' => 0, 'errors' => []];\n        $chunks = $this->getUnsyncedChunks($limit);\n\n        foreach ($chunks as $chunk) {\n            try {\n                if ($this->syncChunk((int) $chunk['id'])) {\n                    $results['synced']++;\n\n                    if ($results['synced'] % self::BATCH_SIZE === 0) {\n                        echo \"Synced {$results['synced']} chunks...\\n\";\n                    }\n                } else {\n                    $results['failed']++;\n                    $results['errors'][] = \"Chunk #{$chunk['id']}: Sync failed\";\n                }\n            } catch (\\RuntimeException $e) {\n                $results['failed']++;\n                $results['errors'][] = \"Chunk #{$chunk['id']}: \" . $e->getMessage();\n            }\n        }\n\n        return $results;\n    }\n\n    \/**\n     * Syncs all chunks (re-sync).\n     *\n     * @return array{synced: int, failed: int, errors: array<string>}\n     *\/\n    public function syncAll(): array\n    {\n        $this->ensureCollection();\n\n        $results = ['synced' => 0, 'failed' => 0, 'errors' => []];\n        $chunks = $this->getAllAnalyzedChunks();\n\n        foreach ($chunks as $chunk) {\n            try {\n                if ($this->syncChunk((int) $chunk['id'])) {\n                    $results['synced']++;\n\n                    if ($results['synced'] % self::BATCH_SIZE === 0) {\n                        echo \"Synced {$results['synced']} chunks...\\n\";\n                    }\n                } else {\n                    $results['failed']++;\n                }\n            } catch (\\RuntimeException $e) {\n                $results['failed']++;\n                $results['errors'][] = \"Chunk #{$chunk['id']}: \" . $e->getMessage();\n            }\n        }\n\n        return $results;\n    }\n\n    \/**\n     * Gets collection statistics.\n     *\n     * @return array{points_count: int, status: string}|null\n     *\/\n    public function getStats(): ?array\n    {\n        return $this->qdrant->getCollectionStats(self::COLLECTION);\n    }\n\n    \/**\n     * Prepares text for embedding.\n     *\n     * @param array<string, mixed> $chunk\n     * @param array<string, mixed> $doc\n     *\/\n    private function prepareTextForEmbedding(array $chunk, array $doc): string\n    {\n        $parts = [];\n\n        $parts[] = 'Dokument: ' . ($doc['title'] ?? '');\n\n        $headingPath = $this->decodeJsonArray($chunk['heading_path'] ?? null);\n        if (!empty($headingPath)) {\n            $parts[] = 'Abschnitt: ' . implode(' > ', $headingPath);\n        }\n\n        $taxonomy = $this->decodeJsonArray($chunk['taxonomy_path'] ?? null);\n        if (!empty($taxonomy)) {\n            $parts[] = 'Kategorie: ' . implode(' > ', $taxonomy);\n        }\n\n        $keywords = $this->decodeJsonArray($chunk['keywords'] ?? null);\n        if (!empty($keywords)) {\n            $parts[] = 'Keywords: ' . implode(', ', $keywords);\n        }\n\n        $content = $chunk['content_clean'] ?? $chunk['content'] ?? '';\n        $content = $this->sanitizeForEmbedding($content);\n        if (mb_strlen($content) > 1000) {\n            $content = mb_substr($content, 0, 1000) . '...';\n        }\n        $parts[] = 'Inhalt: ' . $content;\n\n        $text = implode(\"\\n\\n\", $parts);\n\n        if (mb_strlen($text) > 1800) {\n            $text = mb_substr($text, 0, 1800) . '...';\n        }\n\n        return $text;\n    }\n\n    \/**\n     * Sanitizes text for embedding by removing problematic characters.\n     *\/\n    private function sanitizeForEmbedding(string $text): string\n    {\n        $text = preg_replace('\/[\\x{2500}-\\x{257F}]\/u', ' ', $text) ?? $text;\n        $text = preg_replace('\/[\\x{2580}-\\x{259F}]\/u', ' ', $text) ?? $text;\n        $text = preg_replace('\/[\\x{25A0}-\\x{25FF}]\/u', ' ', $text) ?? $text;\n        $text = preg_replace('\/[\\x00-\\x08\\x0B\\x0C\\x0E-\\x1F\\x7F]\/u', '', $text) ?? $text;\n        $text = preg_replace('\/[ \\t]+\/', ' ', $text) ?? $text;\n        $text = preg_replace('\/\\n{3,}\/', \"\\n\\n\", $text) ?? $text;\n        $text = mb_convert_encoding($text, 'UTF-8', 'UTF-8');\n\n        return trim($text);\n    }\n\n    \/**\n     * Builds the Qdrant payload.\n     *\n     * @param array<string, mixed> $chunk\n     * @param array<string, mixed> $doc\n     * @return array<string, mixed>\n     *\/\n    private function buildPayload(array $chunk, array $doc): array\n    {\n        $content = $chunk['content_clean'] ?? $chunk['content'] ?? '';\n        $content = $this->sanitizeForEmbedding($content);\n        $preview = mb_strlen($content) > 300 ? mb_substr($content, 0, 300) . '...' : $content;\n\n        return [\n            'chunk_id' => (int) $chunk['id'],\n            'doc_id' => (int) $chunk['dokumentation_id'],\n            'chunk_index' => (int) $chunk['chunk_index'],\n            'path' => $doc['path'] ?? '',\n            'title' => $doc['title'] ?? '',\n            'content_preview' => $preview,\n            'heading_path' => $this->decodeJsonArray($chunk['heading_path'] ?? null),\n            'taxonomy_category' => $chunk['taxonomy_category'] ?? null,\n            'taxonomy' => $this->decodeJsonArray($chunk['taxonomy_path'] ?? null),\n            'entities' => $this->decodeJsonArray($chunk['entities'] ?? null),\n            'keywords' => $this->decodeJsonArray($chunk['keywords'] ?? null),\n            'token_count' => (int) ($chunk['token_count'] ?? 0),\n        ];\n    }\n\n    \/**\n     * Updates the qdrant_id in the database.\n     *\/\n    private function updateQdrantId(int $chunkId, string $qdrantId): void\n    {\n        $stmt = $this->pdo->prepare('UPDATE dokumentation_chunks SET qdrant_id = :qid WHERE id = :id');\n        $stmt->execute(['id' => $chunkId, 'qid' => $qdrantId]);\n    }\n\n    \/**\n     * Gets a chunk by ID.\n     *\n     * @return array<string, mixed>|null\n     *\/\n    private function getChunk(int $id): ?array\n    {\n        $stmt = $this->pdo->prepare('SELECT * FROM dokumentation_chunks WHERE id = :id');\n        $stmt->execute(['id' => $id]);\n        $result = $stmt->fetch(PDO::FETCH_ASSOC);\n\n        return $result !== false ? $result : null;\n    }\n\n    \/**\n     * Gets a document by ID.\n     *\n     * @return array<string, mixed>\n     *\/\n    private function getDocument(int $id): array\n    {\n        $stmt = $this->pdo->prepare('SELECT * FROM dokumentation WHERE id = :id');\n        $stmt->execute(['id' => $id]);\n        $result = $stmt->fetch(PDO::FETCH_ASSOC);\n\n        return $result !== false ? $result : [];\n    }\n\n    \/**\n     * Gets unsynced chunks (analyzed but not in Qdrant).\n     *\n     * @return array<array<string, mixed>>\n     *\/\n    private function getUnsyncedChunks(int $limit): array\n    {\n        $stmt = $this->pdo->prepare(\"\n            SELECT * FROM dokumentation_chunks\n            WHERE analysis_status = 'completed' AND qdrant_id IS NULL\n            ORDER BY dokumentation_id, chunk_index\n            LIMIT :limit\n        \");\n        $stmt->bindValue('limit', $limit, PDO::PARAM_INT);\n        $stmt->execute();\n\n        return $stmt->fetchAll(PDO::FETCH_ASSOC);\n    }\n\n    \/**\n     * Gets all analyzed chunks.\n     *\n     * @return array<array<string, mixed>>\n     *\/\n    private function getAllAnalyzedChunks(): array\n    {\n        $stmt = $this->pdo->query(\"\n            SELECT * FROM dokumentation_chunks\n            WHERE analysis_status = 'completed'\n            ORDER BY dokumentation_id, chunk_index\n        \");\n\n        return $stmt->fetchAll(PDO::FETCH_ASSOC);\n    }\n}\n",
        "structuredPatch": [
            {
                "oldStart": 169,
                "oldLines": 8,
                "newStart": 169,
                "newLines": 8,
                "lines": [
                    " ",
                    "         $content = $chunk['content_clean'] ?? $chunk['content'] ?? '';",
                    "         $content = $this->sanitizeForEmbedding($content);",
                    "-        if (mb_strlen($content) > 1000) {",
                    "-            $content = mb_substr($content, 0, 1000) . '...';",
                    "+        if (mb_strlen($content) > Constants::EMBEDDING_TEXT_LIMIT) {",
                    "+            $content = mb_substr($content, 0, Constants::EMBEDDING_TEXT_LIMIT) . '...';",
                    "         }",
                    "         $parts[] = 'Inhalt: ' . $content;",
                    " "
                ]
            }
        ],
        "userModified": false,
        "replaceAll": false
    }
}