NLTK tree data structure, finding a node, it's parent or children

Question

I am using nltk's Tree data structure to work with parsetree strings.

from nltk.tree import Tree
parsed = Tree('(ROOT (S (NP (PRP It)) (VP (VBZ is) (ADJP (RB so) (JJ nice))) (. .)))')

The data structure, however, seems to be limited. Is it possible to get a node by it's string value and then navigate to top or bottom?

For example suppose you want to get the node with string value 'nice' and then see what's its parent, children, etc. Is it achievable via nltk's Tree?

Answer 1

For NLTK 3.0, you want to use the ParentedTree subclass.

http://www.nltk.org/api/nltk.html#nltk.tree.ParentedTree

Using the sample tree you've given, create a ParentedTree and search for the node you want:

from nltk.tree import ParentedTree
ptree = ParentedTree.fromstring('(ROOT (S (NP (PRP It)) \
        (VP (VBZ is) (ADJP (RB so) (JJ nice))) (. .)))')

leaf_values = ptree.leaves()

if 'nice' in leaf_values:
    leaf_index = leaf_values.index('nice')
    tree_location = ptree.leaf_treeposition(leaf_index)
    print tree_location
    print ptree[tree_location]

You can iterate through the tree directly to get the child subtrees. The parent() method is used to find the parent tree for the given subtree.

Here's an example using a deeper tree for child and parent:

from nltk.tree import ParentedTree
ptree = ParentedTree.fromstring('(ROOT (S (NP (JJ Congressional) \
    (NNS representatives)) (VP (VBP are) (VP (VBN motivated) \
    (PP (IN by) (NP (NP (ADJ shiny) (NNS money))))))) (. .))')

def traverse(t):
    try:
        t.label()
    except AttributeError:
        return
    else:

        if t.height() == 2:   #child nodes
            print t.parent()
            return

        for child in t:
            traverse(child)

traverse(ptree)