def convert_tree(t):
label = None
if t.label() == "text":
return slice_text(t[0])
children = []
for elem in t:
if not isinstance(elem, Tree) or (elem.label() != "span" and elem.label() != "rel2par" and elem.label() != "leaf"):
children.append(elem)
if isinstance(elem, Tree) and (label is None or label[0] == "span"):
for sub in (s for s in elem if isinstance(s, Tree)):
if sub.label() == "rel2par":
label = sub
break;
if len(children) == 1:
return convert_tree(children[0])
label_rel = rel2class[label[0].lower()] + "[" + children[0].label()[0:1] + "][" + children[1].label()[0:1] + "]"
if len(children) > 2:
for item in children[1:]:
item._parent = None
return ParseTree(label_rel, [convert_tree(children[0]),
ParseTree(label_rel, [convert_tree(children[1]),
convert_tree(ParseTree("temp", children[2:]))])])
else:
return ParseTree(label_rel, [convert_tree(children[0]), convert_tree(children[1])])
def load_tree_from_file(filename, tokenize = False):
def preprocess_leaf(leaf):
leaf = re.sub('_!(.+?)!_', '\\1', leaf)
if tokenize:
return leaf.split(' ')
else:
return leaf
if filename.endswith('.dis'):
pt = rst_lib.load_tree(filename)
elif filename.endswith('.tree'):
pt = ParseTree.parse(open(filename).read(), leaf_pattern = '_!.+?!_', parse_leaf = preprocess_leaf)
return pt
def load_tree_from_file(filename, tokenize = False):
def preprocess_leaf(leaf):
leaf = re.sub('_!(.+?)!_', '\\1', leaf)
if tokenize:
return leaf.split(' ')
else:
return leaf
if filename.endswith('.dis'):
pt = rst_lib.load_tree(filename)
elif filename.endswith('.tree'):
pt = ParseTree.parse(open(filename).read(), leaf_pattern = '_!.+?!_', parse_leaf = preprocess_leaf)
return pt
def make_new_subtree(label, subtree1, subtree2, deepcopy = False):
if deepcopy:
stump1_clone = copy_subtree(subtree1, True)
stump2_clone = copy_subtree(subtree2, True)
else:
stump1_clone = subtree1
stump2_clone = subtree2
if isinstance(stump1_clone, ParseTree):
stump1_clone._parent = None
if isinstance(stump2_clone, ParseTree):
stump2_clone._parent = None
return ParseTree(label, [stump1_clone, stump2_clone])
def build_tree(self, doc):
if len(doc.edus) == 1:
return [ParseTree("n/a", [doc.edus[0]])]
for i in range(len(doc.sentences)):
sentence = doc.sentences[i]
(start_edu, end_edu) = doc.cuts[i]
if self.verbose:
print 'sentence %d' % i
print 'start_edu', start_edu, 'end_edu', end_edu
self.intra_parser.parse_each_sentence(sentence)
self.multi_parser.parse_document(doc)
return doc.discourse_tree
def build_tree(self, doc):
# print self.use_contextual_features
# Check if only one EDU
if len(doc.edus) == 1:
return [ParseTree("n/a", [doc.edus[0]])]
for i in range(len(doc.sentences)):
sentence = doc.sentences[i]
(start_edu, end_edu) = doc.cuts[i]
if self.verbose:
print('sentence %d' % i)
print('start_edu', start_edu, 'end_edu', end_edu)
self.intra_parser.parse_each_sentence(sentence)
self.multi_parser.parse_document(doc)
return doc.discourse_tree
def simplify_tree(tree, start):
if not tree:
return None
# print 'before', tree
if not isinstance(tree, ParseTree):
t = ParseTree('leaf', [str(start + 1)])
else:
t = tree.__deepcopy__(None)
# print t
L = simplify_tree(tree[0], start)
R = simplify_tree(tree[1], start + len(L.leaves()))
t[0] = L
t[1] = R
# print 'end', t
# print
return t