I'm having trouble to import relationships in a graph.
Let's say I have a few hundreds unique,indexed users that are already created. Then I'd like to create about 120k nodes, each of them being linked to some user through a relationship.
Unfortunately, I'm unable to find a way to batch the import. I'm trying to perform this with the neography ruby gem, but as I'm very new with this environment I wouldn't mind using another way if needed.
What I tried :
@neo.batch(
[:get_node_index, 'user', 'user_id', '1'], #attempt to get the node from index
[:create_node, {"foo => 'bar'}],
[:create_relationship, "has" , "{0}", "{1}"]
) # => fails
,
@neo.batch(
[:create_unique_node, "user", "user_id", "1"], #attempt to create or get the node
[:create_node, {"foo" => "bar"}],
[:create_relationship, "has", "{0}", "{1}"]
) # => fails.
Please note that it is nonetheless possible to batch some create_unique_node commands alone.
The only way I could get the script run is to use
@neo.batch(
[:create_node, {"user_id" => 1}], #works, but duplicates the node
[:create_node, {"foo" => "bar"}],
[:create_relationship, "has", "{0}", "{1}"]
) # => success
However, this will duplicate all my user nodes, which definitely not what I want to achieve. It seems my question is similar to this one, however I don't get at all how am I supposed to use the index when creating the relationships.
Any help would be much appreciated, thanks in advance
Since this question has been upvoted I'm posting the workaround I found since then:
As mentionned in the question, it is possible to batch the create_unique_node to create the nodes. The batch command then returns a list of pointers in which you can get the neo4j ids of each node. I'm not sure to remember wether I had to extract the ids from some hash structure or not, but I'm sure you'll get the point.
So basically, I first created a batch and stored the result in an array :
ids = @neo.batch(
# list of `create_nodes` commands
) #=> returns a list of neo4j ids that you can use further.
To link the nodes, I used a second batch command. Instead of using the failing {id} reference, you can simply use the (absolute) neo4j id of the nodes, so this would look like
[:create_relationship, "something", id1, id2]
where id1 and id2 are given by ids.
This is basically a solution where I'm using absolute ids rather than relative ones....