Create a group vertex for each group, and create outgoing edges to the group vertices

I have a gremlin query which groups vertices based on two properties

g.V().hasLabel("PERSON").
  group().
    by(values('favorite_brand', 'favorite_color').fold()).
  next()

It returns a list of each group mapped to the list of the vertices in group

1   {('adidas', 'blue'): [v[123], v[456]]}
2   {('nike', 'red'): [v[789]]}

How can I: for each group, create a vertex with an outgoing edge to all the vertices in that group and also set the new group vertex properties to be the same

For example for the above, I would create two new vertices. Group Vertex 1 would have 'favorite_brand' as adidas and 'favorite_color' as blue and would have two outgoing edges to the two vertices 123 and 456.

Same for Group Vertex 2

Is there a way in gremlin to carry this query or do I have to store the returned hashmap in a variable and for loop in my lambda to create new vertices? I'm familiar with addV step but how would I iterate through each element in the hashmap and then access the list value? Thanks!

I have look at the tinkerpop official documentation to understand group step but then didn't find enough information on how to iterate through results and perform actions

Solution

Not having your data, the answer below is built using the air-routes data set. The initial group can be built using:

g.V().hasId('3','8','12','13','14').
  group().
    by(values('region', 'country').fold()).
  unfold()

which yields

1   {('US-CA', 'US'): [v[13]]}
2   {('US-NY', 'US'): [v[12], v[14]]}
3   {('US-TX', 'US'): [v[3], v[8]]}

From there we can build a query to unroll the group while creating the new nodes and edges.

g.V().hasId('3','8','12','13','14').
  group().
    by(values('region', 'country').fold()).
  unfold().as('grp').
  addV('group').as('new-node').
    property('region',select(keys).limit(local,1)).
    property('country',select(keys).tail(local)).
  sideEffect(select('grp').select(values).unfold().addE('new-edge').from('new-node'))

which shows us the nodes created but will also have created the edges inside the sideEffect.

1   v[26c2d653-5e9c-3ec9-0854-6ed2a212c63b]
2   v[80c2d653-5e9c-b838-063c-82f6d21cd6e5]
3   v[42c2d653-5e9d-31c7-2c02-c7bf72fe8e38]

We can use the query below to verify everything has worked.

g.V().hasLabel('group').out().path().by(valueMap()).by(id())

Which returns

1   path[{'region': ['US-CA'], 'country': ['US']}, 13]
2   path[{'region': ['US-NY'], 'country': ['US']}, 12]
3   path[{'region': ['US-NY'], 'country': ['US']}, 14]
4   path[{'region': ['US-TX'], 'country': ['US']}, 3]
5   path[{'region': ['US-TX'], 'country': ['US']}, 8]

I used Amazon Neptune to build this answer but it should work on other TinkerPop compliant stores.