I've made an "dirty" hack how a traversal (query) DSL could look like in
Scala. I pushed the code to
here<https://github.com/FaKod/neo4j-scala/blob/ScalaTraverserHack/examples/eu/fakod/examples/TheMatrixTraversalHack.scala>
.
The "query" uses a collections of nodes (nodeMap is a HashMap mapping
Strings to Node instances):
val startNodes = nodeMap("Neo") :: nodeMap("Morpheus") ::
nodeMap("Trinity") :: Nil
The Traverser can be started by calling method start on this list of nodes:
startNodes.start[Matrix](_.name.length > 3)(--("CODED_BY") -- ("KNOWS") -->)
- start takes type parameters. In This case its the case class
Matrix(defined by case
class Matrix(name: String, profession: String)) used while creating the
nodes (a neo4j-scala feature).
- The Term (_.name.length > 3) is basically the returnable evaluator,
which dereferences the name parameter of the Matrix instance and checks its
length.
- The Term (--("CODED_BY") -- ("KNOWS") -->) evaluates to the
relationship type and direction array (as parameterlist to call method
traverse)
The return of the call is a List of Matrix instances. Its possible to use
Scala collection API to apply operations on this.
Like sorting:
startNodes.start[Matrix](_.name.length > 3)(--("CODED_BY") -- ("KNOWS")
-->) sortWith (_.name < _.name)
The length of all names:
startNodes.start[Matrix](_.name.length > 3)(--("KNOWS")
-->).map(_.name).foldLeft(0)(_ + _.length)
Append all names:
startNodes.start[Matrix](_.name.length > 3)(--("CODED_BY") -- ("KNOWS")
-->).map(_.name).foldLeft("")(_ + _)
The Output of the Github examples
here<https://github.com/FaKod/neo4j-scala/blob/ScalaTraverserHack/examples/eu/fakod/examples/TheMatrixTraversalHack.scala>is:
Relations CODED_BY and KNOWS, sorted by name: List(Matrix(Agent
Smith,Program), Matrix(Cypher,Hacker), Matrix(Morpheus,Hacker), Matrix(The
Architect,Whatever), Matrix(Trinity,Hacker))
Relations KNOWS, length of all names: 32
Relations CODED_BY and KNOWS, all names appended:
TrinityMorpheusCypherAgent SmithThe Architect
This is really a hack. It should be possible to get rid most of the
relation parentheses and to run the traverser multithreaded automatically
for every node.
I hope this is helpful :-)
Christopher
On Fri, Nov 18, 2011 at 3:16 PM, Christopher Schmidt <
fakod...@googlemail.com> wrote:
> Hi Andres,
>
> worries is the wrong term. I think a query language has its use case and
> its absolutely necessary for Neo4j. It maps common use cases to a powerful
> language.
>
> But using it programmatically is something different. Thats what I mean
> with abstraction (in the sense of software development). It is a different
> language, with a different semantic used in a compiled language (Java or
> Scala); and this is not a natural fit.
>
> Do I have something in mind: No - no concrete examples.
> But obviously the traverser api with return and stop evaluators is too
> generic (together with the lack of common result list operations, like
> sort, filter etc).
>
> In case of Scala it should be possible (using the powerful collection api)
> to create a DSL for complex queries and aggregation of results (that is
> able to support multi core processing as well, which is a important feature
> imho).
>
> I will try to invest some time. Maybe I can get to an example...
>
>
>
> On Thu, Nov 17, 2011 at 2:44 PM, Andres Taylor <
> andres.tay...@neotechnology.com> wrote:
>
>> On Thu, Nov 17, 2011 at 1:21 PM, Christopher Schmidt <
>> fakod...@googlemail.com> wrote:
>>
>> > Cypher has some
>> > similarities to SQL as a special query language. And to be honest, I
>> think
>> > using it programmatically will cause the same symtoms as SQL with
>> respect
>> > to runtime errors, type safety, abstraction etc.
>> >
>>
>> I understand your worries about runtime errors and type safety. I'd love
>> to
>> know what you mean by "abstraction etc". Do you have any concrete examples
>> in mind?
>>
>> Andrés
>> _______________________________________________
>> Neo4j mailing list
>> User@lists.neo4j.org
>> https://lists.neo4j.org/mailman/listinfo/user
>>
>
>
>
> --
> Christopher
> twitter: @fakod
> blog: http://blog.fakod.eu
>
>
--
Christopher
twitter: @fakod
blog: http://blog.fakod.eu
_______________________________________________
Neo4j mailing list
User@lists.neo4j.org
https://lists.neo4j.org/mailman/listinfo/user
