Examining the current path from within a filter closure?

[Paul Grebenc]

I've been experimenting with the possibility of examining the current path from within a loop's while closure, so that I can base a loop's continuation on portions of the path followed up to that point.  There is a 'path' property available, which makes it possible to examine the current path from within the closure.

I am wondering, is there a way to do something similar with a filter closure?

For example, can I selectively filter something, for example, based on a comparison of the value of a property of the current vertex/edge, with the value of that property contained a specific number of steps back in the current path?

Thanks,

Paul

[Marko Rodriguez]

Hi,

> I've been experimenting with the possibility of examining the current path from within a loop's while closure, so that I can base a loop's continuation on portions of the path followed up to that point. There is a 'path' property available, which makes it possible to examine the current path from within the closure.

Yep. With Gremlin 2.x, be sure to set Pipeline.enablePath() as when path calls are in a closure, its impossible for the compiler to know that you want paths enabled. Path calculations are expensive and thus, assumed to be off. Note that when you do like paths(), simplePath(), etc., it knows you want paths enabled.

> I am wondering, is there a way to do something similar with a filter closure?

Huh. No, as you only have reference to 'it' which is the current object. ............................................................ could be possible, but would require some redos. You could do it with step() as step's 'it' is the previous pipe. it.getCurrentPath().

HTH,
Marko.

http://thinkaurelius.com

[MC]

Marco,

Is this still the best way to achieve the desired behavior, access the current path from within a filter closure? I'm assuming things have changed since 2012. Is there any example code for how to do this?

I would like to use the current path to filter out certain edges.

MC

[Marko Rodriguez]

Hi,

Are you using TinkerPop3 or TinkerPop2?

Marko.

http://markorodriguez.com

--
You received this message because you are subscribed to the Google Groups "Gremlin-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to gremlin-user...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/gremlin-users/659c25c2-3f21-4991-9b75-fd1a82a551f2%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[MC]

Marko,

TP2.

MC

[Marko Rodriguez]

Hi MC,

I'll have to defer to Kuppitz as I honestly don't know -- I haven't touched TinkerPop2 in 1.5 years :D.

Marko.

http://markorodriguez.com

To view this discussion on the web visit https://groups.google.com/d/msgid/gremlin-users/8ffd0a2a-54a7-4c90-912e-20fdde361cf0%40googlegroups.com.

[Daniel Kuppitz]

Not sure about the best approach, but playing with it for a while I liked this variant most:

// starting at v[1], find all vertices at a distance that is a multiple of 1
// distance is defined by the cumulative edge weights between the two vertices
// stop at a distance greater than 5
g.v(1).as("source").ifThenElse {it instanceof Vertex} {it} {it.last()} \
 .bothE().bothV().dedup().as("target").path().loop("source") {
    it.getObject().grep(Edge)*.getProperty("weight").sum() <= 5
  } {
    def distance = it.getObject().grep(Edge)*.getProperty("weight").sum()
    0 == ((int) distance * 10) % 10
  }.select(["target"])

Output:

gremlin> g.v(1).as("source").ifThenElse {it instanceof Vertex} {it} {it.last()} \
gremlin>  .bothE().bothV().dedup().as("target").path().loop("source") {
gremlin>     it.getObject().grep(Edge)*.getProperty("weight").sum() <= 5
gremlin>   } {
gremlin>     def distance = it.getObject().grep(Edge)*.getProperty("weight").sum()
gremlin>     0 == ((int) distance * 10) % 10
gremlin>   }.select(["target"])
==>[target:v[4]]
==>[target:v[5]]

Basically simply create the path outside the closure, so that it's passed as the current object to the closures.

Cheers,
Daniel