Imagine a GraphQL query like this:
Query {
projects(platforms: [String], languages: [String])
}
And a Neo4j graph containing nodes labeled Project, Platform and Language.
When none of the optional arguments are passed, the resolver should return the list of all projects in the graph with no filtering (possibly with pagination in future). When either or all arguments are passed, those should be applied in the underlying query; i.e. we should get projects that are implemented in given languages or for given platforms.
I am using Bolt.Sips driver for Neo4j in my Elixir backend, and this driver offers minimal functionality as for wrapping the Cypher query language, so for the most part I need to write queries on my own. So far I have this implementation of the resolver, since I have yet to implement the second argument:
def get_projects!(_parent, %{languages: _} = args, _resolution) do
"""
match (node:Project)<-[:LANGUAGE_FOR]-(l:Language)
where l.name in {props}.languages
return node
"""
|> generic_query!(%{props: args})
end
def get_projects!(_parent, _args, _resolution) do
get_all_entities_of_type!("Project")
end
where generic_query!/2 and get_all_entities_of_type!/1 are just wrappers for Bolt.Sips query execution.
My question is: what's the best way to construct a Cypher query in this case? Should I just write a bunch of functions that will construct a query string before execution? Maybe it's just my general lack of GraphQL experience and query schemas like this are to be constructed differently? Any tips would be appreciated.
