Add Direct Neo4j Export Support

[Sagar-Mondal]

Create a new exporter that connects to a live Neo4j database and populates it with the extracted knowledge graph using the official Neo4j Python driver.

[guilhaumelm]

Feature Request: Neo4j Database Exporter

Description

Add a new exporter that connects to a live Neo4j database and populates it with the extracted knowledge graph using the official Neo4j Python driver.

Motivation

Currently, docling-graph supports exporting to various formats (JSON, CSV, Cypher scripts), but lacks direct integration with Neo4j databases. A dedicated Neo4j exporter would enable users to:

• Directly populate Neo4j instances without intermediate files
• Leverage Neo4j's transaction capabilities for reliable data ingestion
• Simplify workflows for production deployments
• Enable real-time graph updates and incremental loading

Proposed Solution

Create a Neo4jExporter class that:

1. Uses the official neo4j Python driver for database connectivity
2. Accepts connection parameters (URI, username, password, database name)
3. Transforms the extracted knowledge graph into Neo4j nodes and relationships
4. Handles batch operations for efficient data loading
5. Supports transaction management and error handling
6. Provides options for merge vs. create operations (to handle updates)

Implementation Details

• Package: docling_graph/db_clients/neo4j_client.py
• Dependencies: neo4j (official Python driver)
• Configuration: Support for connection strings, authentication, and database selection
• API: Follow the existing exporter pattern from docling_graph/core/exporters/

Example Usage

from docling_graph import run_pipeline, PipelineConfig
from docling_graph.db_clients import Neo4jExporter

config = PipelineConfig(
    source="document.pdf",
    template=MyTemplate,
    exporter=Neo4jExporter(
        uri="bolt://localhost:7687",
        username="neo4j",
        password="password",
        database="knowledge_graph"
    )
)

run_pipeline(config)