如何在单次Gremlin遍历中同时获取顶点属性集并导出GraphML子图

Great question! In Gremlin, you can handle both tasks in one traversal by using side-effect operations or aggregation steps—these let you run secondary actions (like exporting your subgraph) without breaking the main traversal flow that collects your vertex property data. Here are two practical approaches:

Approach 1: Use aggregate + sideEffect #

This method first aggregates your vertex properties, then uses a side effect to export the subgraph, and finally returns the aggregated property collection.

g.V().hasLabel('person') // Start with your target vertices (adjust filter as needed)
  .aggregate('vertexProps').by(elementMap()) // Aggregate vertex details (use valueMap() if you don't need id/label)
  .sideEffect(
    // Side effect to export the subgraph to GraphML
    subgraph('subGraph') // Define the subgraph scope (customize edges/vertices here)
      .cap('subGraph') // Extract the subgraph object
      .writeGraph('output.graphml') // Export to your target GraphML file
  )
  .cap('vertexProps') // Return the aggregated vertex property collection

Breakdown: #

• aggregate('vertexProps').by(elementMap()): Stores each vertex's full details (id, label, properties) in an in-memory collection named vertexProps. Swap elementMap() with valueMap() if you only want property key-value pairs.
• sideEffect(...): Runs the subgraph export logic without altering the main traversal flow. For example, add .outE().inV() inside subgraph() if you need to include outgoing edges and their target vertices.
• cap('vertexProps'): Ends the main traversal by returning the aggregated vertex properties as a list.

Approach 2: Use fold + as + sideEffect #

If you prefer to first collect all vertices, then handle both tasks, this approach folds the vertex stream into a list, runs the export as a side effect, and then extracts the properties:

g.V().hasLabel('person')
  .fold() // Collapse vertices into a single list
  .as('vertexList') // Label this list for later reference
  .sideEffect(
    unfold() // Expand the list back into individual vertices
      .subgraph('subGraph').bothE().bothV() // Include all connected edges/vertices in the subgraph
      .cap('subGraph')
      .writeGraph('output.graphml')
  )
  .select('vertexList') // Retrieve the stored vertex list
  .unfold().elementMap().fold() // Extract properties and return as a structured collection

Key Notes: #

• Ensure your Gremlin version (TinkerPop 3.x+) and graph database support writeGraph (TinkerGraph, Neptune, and most major graph databases do).
• Customize the subgraph() logic to match your needs—narrow it down to specific edge labels with .outE('follows').inV() if you don't want all connected edges.
• If working with a remote graph (like Neptune), adjust the export path to a location accessible by the graph server, or use client-side logic to handle export after retrieving the subgraph.

内容的提问来源于stack exchange，提问作者Mihai Bucica