bmeg / ophion

Language for making graph queries from data

Version Matrix


Language for making graph queries from data



Given a graph adhering to the tinkerpop interface, we want to make remote queries against it without exposing a gremlin console. To this end, Ophion provides a json-encoded schema for making graph queries as data:

 [{"label": "person"},
  {"as": "people"},
  {"outEdge": "created"},
  {"has": "weight", "within": [1.0]},
  {"inVertex": "software"},
  {"as": "software"},
  {"select": ["people", "software"]}]}

Given some json like this, you can parse it into a program that executes the given query when run:

import ophion.Ophion._
import org.apache.tinkerpop.gremlin.tinkergraph.structure.TinkerFactory

val graph = TinkerFactory.createModern
val traversal = graph.traversal.V()

val example = """{"query":
 [{"label": "person"},
  {"as": "people"},
  {"outEdge": "created"},
  {"has": "weight", "within": [1.0]},
  {"inVertex": "software"},
  {"as": "software"},
  {"select": ["people", "software"]}]}"""

val query = Query.fromString(example).compose
val result = query.foldMap(operationInterpreter(traversal)).head.toList

Gremlin functionality currently supported:

selecting vertexes or edges by label

{"label": "alphabet"}

values - getting just certain values rather than the whole vertex or edge property map

{"values": ["color", "pattern", "hair", "volume"]}

limit and range

{"limit": 13}
{"begin": 84, "end": 87}

has...within predicates

{"has": "arrow", "within": ["feathered", "of time", "prehistoric", "guide"]}

traversing from and to edges and vertexes

{"in": "pointing at me"}
{"out": "who I'm pointing at"}
{"inEdge": "hop up to the edge"}
{"outEdge": "same but with outgoing"}
{"inVertex": "from an edge, go to the vertex at one end"}
{"outVertex": "go to the vertex on the other end"}

count and counting by a property

{"count": ""}
..., {"groupCount": ""}, {"by": "color"}, ...

You can label different group counts and retrieve them with cap:

{"groupCount": "color"},
{"by": "color"},
{"groupCount": "faces"},
{"by": "faces"},
{"cap", ["color", "faces"]}, ...

marking a point in a traversal with as and selecting those points at the end of the traversal.

{"as": "birth place"}
... (do some stuff)
{"as": "resting place"}
{"select": ["birth place", "resting place"]} --> just get the endpoints

Ophion hopes to support the entire gremlin spec in the future, but for now these features already enable all of our current graph query use cases.