class AlphaStage(object):
def __init__(self):
self.tree = Tree()
self.nest_deps = ["aux", "auxpass", "cc", "agent", "det", "advmod", "amod", "poss", "nummod"]
self.add_to_first_deps = ["pcomp", "compound"]
def process_child_token(self, parent_elem_id, token, root_id, pos):
# ignore
if (len(token.dep) == 0) or (token.dep == "punct"):
return
parent_elem = self.tree.get(parent_elem_id)
# nest
if (token.dep in self.nest_deps) or ((token.dep == "prep") and (not parent_elem.has_pos("VERB"))):
child_elem_id = self.process_token(token, root_id)
self.tree.get(root_id).nest(child_elem_id)
self.tree.get(root_id).new_layer()
return
child_elem_id = self.process_token(token)
# add to parent's first child
if token.dep in self.add_to_first_deps:
parent_elem.add_to_first_child(child_elem_id, pos)
return
# add child
parent_elem.add_child(child_elem_id)
def process_token(self, token, root_id=None):
elem_id = self.tree.create_leaf(token)
if root_id is None:
root_id = elem_id
for child_token in token.left_children:
self.process_child_token(elem_id, child_token, root_id, Position.LEFT)
for child_token in token.right_children:
self.process_child_token(elem_id, child_token, root_id, Position.RIGHT)
self.tree.get(elem_id).apply_layers()
return elem_id
def process_sentence(self, sentence):
self.tree.root_id = self.process_token(sentence.root())
self.tree.remove_redundant_nesting()
return self.tree
def __init__(self):
self.tree = Tree()
self.nest_deps = ["aux", "auxpass", "cc", "agent", "det", "advmod", "amod", "poss", "nummod"]
self.add_to_first_deps = ["pcomp", "compound"]
