Protokoll #24723

{
    "event": "PreToolUse",
    "tool_name": "Read",
    "tool_input": {
        "file_path": "\/var\/www\/dev.campus.systemische-tools.de\/src\/Infrastructure\/Config\/PipelineStepConfig.php"
    }
}

{
    "tool_response": {
        "type": "text",
        "file": {
            "filePath": "\/var\/www\/dev.campus.systemische-tools.de\/src\/Infrastructure\/Config\/PipelineStepConfig.php",
            "content": "<?php\n\ndeclare(strict_types=1);\n\nnamespace Infrastructure\\Config;\n\n\/\/ @responsibility: Konfiguration für Pipeline-Schritte und Collections\n\nfinal class PipelineStepConfig\n{\n    \/**\n     * Get all available step types with their configuration.\n     *\n     * @return array<string, array<string, mixed>>\n     *\/\n    public static function getStepTypes(): array\n    {\n        return [\n            \/\/ Phase 1: Vorverarbeitung\n            'detect' => [\n                'label' => 'Erkennung',\n                'description' => 'Dateien scannen und Format prüfen',\n                'phase' => 'Vorverarbeitung',\n                'storage' => null,\n            ],\n            'validate' => [\n                'label' => 'Validierung',\n                'description' => 'Datei-Prüfung auf Lesbarkeit und Korruption',\n                'phase' => 'Vorverarbeitung',\n                'storage' => null,\n            ],\n            'page_split' => [\n                'label' => 'Seitenzerlegung',\n                'description' => 'PDF in Einzelseiten zerlegen für Referenz und Vision-Analyse',\n                'phase' => 'Vorverarbeitung',\n                'storage' => 'ki_content.document_pages',\n            ],\n            'vision_analyze' => [\n                'label' => 'Bildanalyse',\n                'description' => 'Seiten via Vision-Modell analysieren, Bilder und Grafiken erkennen',\n                'phase' => 'Vorverarbeitung',\n                'storage' => 'ki_content.document_pages (vision_analysis)',\n                'uses_vision' => true,\n            ],\n            'extract' => [\n                'label' => 'Textextraktion',\n                'description' => 'Text extrahieren, OCR für Bilder mit Text',\n                'phase' => 'Vorverarbeitung',\n                'storage' => null,\n            ],\n            'structure' => [\n                'label' => 'Strukturerkennung',\n                'description' => 'Überschriften, Listen und Hierarchie erkennen',\n                'phase' => 'Vorverarbeitung',\n                'storage' => 'ki_content.document_sections',\n            ],\n            'segment' => [\n                'label' => 'Abschnitte',\n                'description' => 'Logische Dokumentgliederung nach Struktur',\n                'phase' => 'Vorverarbeitung',\n                'storage' => 'ki_content.document_sections',\n            ],\n            'chunk' => [\n                'label' => 'Textbausteine',\n                'description' => 'Chunks erstellen (max 800 Token) mit Seitenreferenz',\n                'phase' => 'Vorverarbeitung',\n                'storage' => 'ki_content.chunks',\n            ],\n            \/\/ Phase 2: Speicherung & Vektorisierung\n            'metadata_store' => [\n                'label' => 'DB-Speicherung',\n                'description' => 'Dokument, Seiten und Chunks in MariaDB speichern',\n                'phase' => 'Speicherung',\n                'storage' => 'ki_content.documents, .document_pages, .chunks',\n            ],\n            'embed' => [\n                'label' => 'Vektorisierung',\n                'description' => 'Embeddings erstellen für Vektor-Suche',\n                'phase' => 'Speicherung',\n                'storage' => 'Qdrant: {collection}',\n                'fixed_model' => 'mxbai-embed-large (1024-dim)',\n                'has_collection' => true,\n            ],\n            'collection_setup' => [\n                'label' => 'Collection',\n                'description' => 'Qdrant-Collection einrichten falls nötig',\n                'phase' => 'Speicherung',\n                'storage' => 'Qdrant: {collection}',\n            ],\n            'vector_store' => [\n                'label' => 'Vektorspeicherung',\n                'description' => 'Vektoren in Qdrant mit MariaDB-ID als Referenz',\n                'phase' => 'Speicherung',\n                'storage' => 'Qdrant: {collection}',\n            ],\n            'index_optimize' => [\n                'label' => 'Index-Optimierung',\n                'description' => 'HNSW-Index für schnelle Suche optimieren',\n                'phase' => 'Speicherung',\n                'storage' => 'Qdrant: {collection}',\n            ],\n            \/\/ Phase 3: Wissensextraktion\n            'knowledge_page' => [\n                'label' => 'Seiten-Wissen',\n                'description' => 'Pro Seite: Entitäten → Semantik → Ontologie → Taxonomie',\n                'phase' => 'Wissen',\n                'storage' => 'ki_content.page_knowledge, .entities, .entity_semantics',\n                'uses_llm' => true,\n            ],\n            'knowledge_section' => [\n                'label' => 'Abschnitt-Wissen',\n                'description' => 'Pro Kapitel: Aggregierte Wissensrepräsentation',\n                'phase' => 'Wissen',\n                'storage' => 'ki_content.section_knowledge',\n                'uses_llm' => true,\n            ],\n            'knowledge_document' => [\n                'label' => 'Dokument-Wissen',\n                'description' => 'Konsolidierte Gesamtsicht des Dokuments',\n                'phase' => 'Wissen',\n                'storage' => 'ki_content.document_knowledge',\n                'uses_llm' => true,\n            ],\n            'knowledge_validate' => [\n                'label' => 'Wissens-Validierung',\n                'description' => 'Abgleich mit DB, Duplikate zusammenführen, neue validieren',\n                'phase' => 'Wissen',\n                'storage' => 'ki_content.entities (merged)',\n            ],\n            \/\/ Legacy Analyse-Schritte\n            'entity_extract' => [\n                'label' => 'Entitäten (Legacy)',\n                'description' => 'Personen, Organisationen, Konzepte, Methoden erkennen',\n                'phase' => 'Analyse',\n                'storage' => 'ki_content.chunk_entities',\n                'uses_llm' => true,\n            ],\n            'relation_extract' => [\n                'label' => 'Beziehungen (Legacy)',\n                'description' => 'Relationen zwischen Entitäten extrahieren',\n                'phase' => 'Analyse',\n                'storage' => 'ki_content.entity_relations',\n                'uses_llm' => true,\n            ],\n            'taxonomy_build' => [\n                'label' => 'Taxonomie (Legacy)',\n                'description' => 'Hierarchische Kategorisierung aufbauen',\n                'phase' => 'Analyse',\n                'storage' => 'ki_content.chunk_taxonomy, .taxonomy_terms',\n                'uses_llm' => true,\n            ],\n            'semantic_analyze' => [\n                'label' => 'Semantik (Legacy)',\n                'description' => 'Bedeutungs-Analyse, Konzepte und Definitionen',\n                'phase' => 'Analyse',\n                'storage' => 'ki_content.chunk_semantics',\n                'uses_llm' => true,\n            ],\n            'summarize' => [\n                'label' => 'Zusammenfassung',\n                'description' => 'Dokument- und Seiten-Zusammenfassungen erstellen',\n                'phase' => 'Analyse',\n                'storage' => 'ki_content.documents (summary), .document_pages',\n                'uses_llm' => true,\n            ],\n            'question_generate' => [\n                'label' => 'Fragengenerierung',\n                'description' => 'Beispielfragen für RAG-Chat erstellen',\n                'phase' => 'Analyse',\n                'storage' => 'ki_content.generated_questions',\n                'uses_llm' => true,\n            ],\n            'finalize' => [\n                'label' => 'Abschluss',\n                'description' => 'Status finalisieren und Job beenden',\n                'phase' => 'Analyse',\n                'storage' => 'ki_content.documents (status)',\n            ],\n            'analyze' => [\n                'label' => 'Analyse (Legacy)',\n                'description' => 'Kombinierte Analyse (veraltet)',\n                'phase' => 'Analyse',\n                'storage' => 'ki_content.chunk_entities, .chunk_semantics',\n                'uses_llm' => true,\n            ],\n        ];\n    }\n\n    \/**\n     * Get available Qdrant collections.\n     *\n     * @return array<string, string>\n     *\/\n    public static function getCollections(): array\n    {\n        return [\n            'documents' => 'Documents (Schulungsunterlagen)',\n            'mail' => 'Mail (E-Mails)',\n            'entities' => 'Entities (Entitäten)',\n            'knowledge' => 'Knowledge (Wissensbasis)',\n        ];\n    }\n\n    \/**\n     * Get default steps for new pipelines.\n     *\n     * @return array<array<string, mixed>>\n     *\/\n    public static function getDefaultSteps(): array\n    {\n        return [\n            ['step_type' => 'detect', 'config' => ['hash_algorithm' => 'sha256'], 'sort_order' => 1, 'enabled' => 1],\n            ['step_type' => 'extract', 'config' => ['ocr_enabled' => true, 'ocr_language' => 'deu'], 'sort_order' => 2, 'enabled' => 1],\n            ['step_type' => 'chunk', 'config' => ['min_size' => 100, 'max_size' => 2000, 'overlap' => 0.1], 'sort_order' => 3, 'enabled' => 1],\n            ['step_type' => 'embed', 'config' => ['model' => 'mxbai-embed-large', 'collection' => 'documents', 'dimensions' => 1024], 'sort_order' => 4, 'enabled' => 1],\n            ['step_type' => 'analyze', 'config' => ['extract_entities' => true, 'extract_relations' => true, 'classify_taxonomy' => true], 'sort_order' => 5, 'enabled' => 0],\n        ];\n    }\n\n    \/**\n     * Get default file extensions.\n     *\n     * @return array<string>\n     *\/\n    public static function getDefaultExtensions(): array\n    {\n        return ['.pdf', '.docx', '.pptx', '.md', '.txt'];\n    }\n\n    \/**\n     * Parse extension string into array.\n     *\n     * @param string $input\n     * @return array<string>\n     *\/\n    public static function parseExtensions(string $input): array\n    {\n        $extensions = [];\n        $parts = preg_split('\/[\\s,;]+\/', $input);\n\n        if ($parts === false) {\n            return self::getDefaultExtensions();\n        }\n\n        foreach ($parts as $ext) {\n            $ext = trim($ext);\n            if ($ext === '') {\n           ... [TRUNCATED-27741db15bfc7a28]",
            "numLines": 259,
            "startLine": 1,
            "totalLines": 259
        }
    }
}