Extract nodes from json based on user input preserveing a portion of the higher level object as well about glom

[kurtbrose]

First off, it's a recursive problem so we'd need to use Ref in order to get that problem.

Then, it's a filtration problem, so we will need to use the SKIP marker object to drop things which don't match.

Additionally, there are two inputs here: one is the nested nodes, the other is the list of attribute nodes.

This might be better solved with a direct recursion or the remap() recursion helper from boltons. I don't want to over-promise that glom is the right solution. But, I'll take a shot at it.

Reformulating the problem:

• we want to recurse on "child" attribute of dict
• for every dict, if the "node" attribute is in the input set, completely hold onto it and all child nodes
• on the way back out, we want to drop any nodes that weren't included in the result set, or whose children weren't included in the result set

node_spec = Ref("node-spec", 
   Or(
      # case 1: this node is in the input; return it and all children
      And(lambda t: t["node"] in t[S]["input-nodes"], T),
      # case 2: one of the children is in the input
      And((
         A.node,
         "child",
         [Ref("node-spec")],
         Merge(S.node, 

Oof, even as I'm writing this I can tell it's a bad fit for glom; there are multiple inputs and a lot of internal state. This is really just a recursion problem. I'm not going to bother trying to finish it, it would be really tortured.

def filter_node(cur, to_include):
   if cur["node"] in to_include:
      return cur
   children = []
   for child in cur["child"]:
      filtered = filter_node(child)
      if filtered:
         children.append(filtered)
    if children:
      return {**cur, "child": children}
   return None