Paths between nodes excluding certain edges

[Luiz Celso Gomes Jr.]

Hi, all!

I need to find the paths between two nodes following both in/out edges but excluding edges whose labels are in a list. Below is the solution I came up with, but it isn't working. Is there an easier way of doing that?

Thanks!

def g = TinkerGraphFactory.createTinkerGraph()

def origs = [g.v(4)]

def dest = '2'

def NOTFOLLOW = ["bla"]

def paths = origs._().as('start')

.copySplit(

_().inE.filter{!(it.label in NOTFOLLOW)}.outV, 

_().outE.filter{!(it.label in NOTFOLLOW)}.inV

).fairMerge()

.simplePath().loop('start'){it.loops<=3}{(it.object.id==dest)}

.path().toList()

paths.each{println it}

[Luiz Celso Gomes Jr.]

It's funny that if I don't use the "loop" step, just copying the code multiple times, everything works perfectly. Would that be a bug? Or am I doing something wrong in the loop parameters?

E.g., this works:

paths = origs._().as('start')

.copySplit(

_().inE.filter{!(it.label in NOTFOLLOW)}.outV, 

_().outE.filter{!(it.label in NOTFOLLOW)}.inV

).fairMerge()

.simplePath()

.copySplit(

_().inE.filter{!(it.label in NOTFOLLOW)}.outV, 

_().outE.filter{!(it.label in NOTFOLLOW)}.inV

).fairMerge()

.simplePath()

        .path().toList()

I do need the loop step though..

:(

Thanks

[Stephen Mallette]

I took the recipe that finds paths between vertices in gremlindocs

http://gremlindocs.com/#recipes/paths-between-two-vertices

and modified it slightly so that it has multiple starts looking for
destinations at v[5] and has edge label filtering:

gremlin> origs = [g.v(1),g.v(3)]
==>v[1]
==>v[3]
gremlin> origs._().both('knows','created').loop(1){it.loops<=3 &&
!(it.object.id in ['1','5'])}.has('id','5').path
==>[v[1], v[4], v[5]]
==>[v[1], v[3], v[4], v[5]]
==>[v[3], v[4], v[5]]

perhaps that will help you solve your problem...

Stephen

On Fri, Nov 9, 2012 at 2:34 PM, Luiz Celso Gomes Jr.

> --
>
>

[Luiz Celso Gomes Jr.]

That's nice Stephen, thanks. The problem is that I don't know which edges to traverse. I just know which edges not to traverse. It's been surprisingly hard to find a way around it.

:(

--

[Marko Rodriguez]

Hi,

What do you mean, "edges not to traverse." By label? If so, you can do:

x = ['a','b','c']

origs._().both.filter{!x.contains(it.label)}.loop(2){ ... }

In a future release, we will be support has and hasNot with ORing values.

https://github.com/tinkerpop/gremlin/issues/308

HTH,

Marko.

http://markorodriguez.com

--
 
 

[Luiz Celso Gomes Jr.]

Thanks Marko, but you are filtering out nodes, not edges, right? I need to avoid certain edges (that's why I had that complicated code in the first email, so that I could use filter in the edges).

Makes sense?

Thanks!

--
 
 

[Marko Rodriguez]

Hi,

Sorry, my code snippet is wrong. Do this:

x = ['a','b','c']

origs._().bothE.filter{!x.contains(it.label)}.bothV.loop(2){ ... }

Therefore, only edges that do not have labels specified in x.

HTH,

Marko.

http://markorodriguez.com

--
 
 

[Luiz Celso Gomes Jr.]

I think this doubles the number of nodes, because bothV returns the head and tail of any edge not filtered. I'm pretty sure about that, but I'll test again when I get to my computer.

Maybe the best solution would be to add a variant of the both (and in/out) step in gremlin? I think I could try that in the near future...

Marko, what about the code in the first email? Do you see any obivious error? That code would do it for me, but I was getting a weird behavior. If you think the code makes some sense, I'll look into whats going on.

Thanks a bunch!

http://markorodriguez.com

--
 
 

--
 
 

[Stephen Mallette]

I've been tinkering with this issue since it posted and can't quite
get it right. Luiz, did you ever get it figured out?

Stephen

On Sun, Nov 11, 2012 at 6:45 PM, Luiz Celso Gomes Jr.

> --
>
>