GraphDrawing | Graph Analysis and Drawing Library - Grad | Data Visualization library

you can flag nodes for removal, and then remove them:

Upgrade to the newest version of networkx, it is working as of networkx version 2.6.2.

write_graphml takes a dynamic_properties. Let's configure that:

Live On Coliru

The Java SimpleXML library looks like it is primarily for serializing Java objects to XML. It does support deserialization (parsing XML to Java) - but the documentation says:

org.simpleframework.xml.stream - Provides an XML parsing layer over a set of third party libraries. The stream package provides an event based XML parsing framework that can be used with several third party libraries such as DOM, StAX and the XML pull API.

In other words, it delegates to other packages.

So, here is a StAX approach. This uses the cursor-based StAX XML Stream Reader (not the alternative StAX event reader), because as it says here in the "Comparing Cursor and Iterator APIs" section:

If you are programming for a particularly memory-constrained environment, like Java ME, you can make smaller, more efficient code with the cursor API.

Here is the approach. It just parses the input file and prints out the relevant parts. It does not attempt to do anything more than that - just to demonstrate how to access the data items you need:

Without some sample data it's hard to tell what might be completely wrong but here's some things to consider based on the Gremlin in your question:

• shortestPath()-step is a an OLAP step and therefore requires that you define g using withComputer() - an example can be found here.
• I notice that you search for your starting vertex in two different ways in each of your queries. In the first case you search by a property named "id" and in the second you search by the vertex identifier (referenced as T.id when used in has()). Perhaps that initial starting vertex isn't found properly?
• The second traversal looks invalid as it tries to do a loop (i.e. it has an until() condition) but does not specify a repeat() operation.

When my traversals don't return what I expect, I try to simplify them to the point where I can see where the traversal is filtering away the traversers that I expected. In other words, have Gremlin Console open and your graph connected. Ensure that g.V("startVertexId") returns the vertex you expect. When satisfied of the result, add a step, run g.V("startVertexId").out() and validate that. Continue with that pattern until you find out where your results are not returning. You can also run your traversal with profile() step and it should show you where the traversers are filtering away, but sometimes it's easier to recognize by picking the traversal apart and simplifying.

I tested your traversal with the data you recently supplied in your question and when I make the adjustments as suggested in my answer and comments I don't seem to have a problem getting results:

A nifty little tick if you want to create a builder instance scoped to a namespace is to use Nokogiri::XML::Builder.with(doc.root):

You explicitly select the first occurrence of the node: FirstOrDefault()
You have to select the nodes in a loop: