DSL works just fine, equivalent JSON errors out

[dkoslicki]

This DSL works:

add_qnode(name=methyltransferase, key=n0)
add_qnode(key=n1)
add_qnode(name=plant, key=n3)
add_qedge(key=e0,subject=n0,object=n1,predicate=biolink:subclass_of)
add_qnode(key=n2)
add_qedge(key=e1,subject=n1,object=n2)
add_qedge(key=e2,subject=n2,object=n3)
expand(kp=ARAX/KG2)
resultify()
filter_results(action=limit_number_of_results,max_results=200,prune_kg=true)

copying the accompanying DSL:

{
  "nodes": {
    "n0": {
      "id": "UMLS:C0025831",
      "is_set": false
    },
    "n1": {
      "is_set": false
    },
    "n3": {
      "id": "UMLS_STY:T002",
      "is_set": false
    },
    "n2": {
      "is_set": false
    }
  },
  "edges": {
    "e0": {
      "predicate": "biolink:subclass_of",
      "subject": "n0",
      "object": "n1",
      "exclude": false
    },
    "e1": {
      "subject": "n1",
      "object": "n2",
      "exclude": false
    },
    "e2": {
      "subject": "n2",
      "object": "n3",
      "exclude": false
    }
  }
}

gives the error:
ERROR: Help, two edges at A583. Don't know what to do: 2

[edeutsch]

Yeah, this is something that I was struggling with yesterday and thought I fixed it but then it threw one test into an infinite loop so I backed out the fix, so yes, this is a bug that I need to resolve. Thanks, will work on it.

[dkoslicki]

And just a note: Trung said it is a font issue, but I'm not 100% sure that's the issue (changing the font didn't fix the problem)

[edeutsch]

My guess is the "font issue" is that sometimes JSON query_graphs pasted into Word or similar tools get their simple ASCII quotation marks converted into fancy forward-facing quotes and backwards facing fancy quotes, and then when that gets pasted into our tool, it causes JSON parsing problems.

But not the A583 problem. I know that problem, but my efforts to rectify it so far have been inadequate. The repro above will help me solve it. Will work on it. next week probably. Thanks!

[dkoslicki]

@edeutsch Is this issue still relevant? I.e. should it be left open?

[edeutsch]

yes very much relevant. I need to tackle this.