def transform_to_ZhangShasha_tree(parse_tree, offset = 0, treepos = 0, nuclearity = True):
edges = []
if not isinstance(parse_tree, Tree):
return edges
''' use breadth-first search '''
if nuclearity:
rel_classes = rel_status_classes
else:
rel_classes = class2rel.keys()
rel_classes.append('NO-REL')
if nuclearity:
parent_label = parse_tree.node
else:
parent_label = parse_tree.node[ : -6] if parse_tree.node != 'NO-REL' else parse_tree.node
parent_label = str(rel_classes.index(parent_label))
parent_label += ':' + str(treepos)
parent_label = parent_label.replace('-', '_')
left_treepos = treepos + 1
left = parse_tree[0]
if not isinstance(left, Tree):
left_label = 'e%d' % (offset + 1)
left_span = 1
right_treepos = treepos + 1
else:
if nuclearity:
left_label = left.node
else:
left_label = left.node[ : -6] if left.node != 'NO-REL' else left.node
left_label = str(rel_classes.index(left_label)) + ':' + str(left_treepos)
left_span = len(left.leaves())
right_treepos = treepos + len(left.treepositions()) + len(left.leaves())
right = parse_tree[1]
if not isinstance(right, Tree):
right_label = 'e%d' % (offset + 1 + left_span)
else:
if nuclearity:
right_label = right.node
else:
right_label = right.node[ : -6] if right.node != 'NO-REL' else right.node
right_label = str(rel_classes.index(right_label)) + ':' + str(right_treepos)
edges.append(parent_label + '-' + left_label)
edges.extend(transform_to_ZhangShasha_tree(left, offset, left_treepos))
edges.append(parent_label + '-' + right_label)
edges.extend(transform_to_ZhangShasha_tree(right, offset + left_span, right_treepos))
return edges
def transform_to_shift_reduce_actions(parse_tree, relation = True, nuclearity = True, compressed = True, offset = 1):
if not isinstance(parse_tree, Tree):
if not compressed:
return ['S']
else:
return []
if nuclearity:
rel_classes = rel_status_classes
else:
rel_classes = class2rel.keys()
actions = []
left = parse_tree[0]
right = parse_tree[1]
if isinstance(left, Tree):
left_span = len(left.leaves())
else:
left_span = 1
actions.extend(transform_to_shift_reduce_actions(left, relation, nuclearity, compressed, offset))
actions.extend(transform_to_shift_reduce_actions(right, relation, nuclearity, compressed, offset + left_span))
head_action = 'R'
if relation:
if nuclearity:
action = parse_tree.node
else:
action = parse_tree.node[ : -6]
head_action += str(rel_classes.index(action))
else:
if nuclearity:
action = parse_tree.node[ -6 : ]
head_action += action
if compressed:
left_start = offset
left_end = offset + left_span - 1
right_start = left_end + 1
if isinstance(right, Tree):
right_span = len(right.leaves())
else:
right_span = 1
right_end = right_start + right_span - 1
head_action += '[(%d,%d),(%d,%d)]' % (left_start, left_end, right_start, right_end)
actions.append(head_action)
return actions