def dummy_constituent_tree(token_yield,
full_token_yield,
dummy_label,
dummy_root,
label=None):
"""
:param token_yield: connected yield of a parse tree
:type token_yield: list[ConstituentTerminal]
:param full_token_yield: full yield of the parse tree
:type full_token_yield: list[ConstituentTerminal]
:return: dummy constituent tree
:rtype: ConstituentTree
generates a dummy tree for a given yield using dummy_label as inner node symbol
"""
tree = ConstituentTree(label)
# create all leaves and punctuation
for token in full_token_yield:
if token not in token_yield:
tree.add_punct(full_token_yield.index(token), token.pos(),
token.form())
else:
tree.add_leaf(full_token_yield.index(token), token.pos(),
token.form())
# generate root node
root_id = 'n0'
tree.add_node(root_id, ConstituentCategory(dummy_root))
tree.add_to_root(root_id)
parent = root_id
if len(token_yield) > 1:
i = 1
# generate inner nodes of branching tree
for token in token_yield[:-2]:
node = ConstituentCategory(str(dummy_label))
tree.add_node('n' + str(i), node)
tree.add_child(parent, 'n' + str(i))
tree.add_child(parent, full_token_yield.index(token))
parent = 'n' + str(i)
i += 1
token = token_yield[len(token_yield) - 2]
tree.add_child(parent, full_token_yield.index(token))
token = token_yield[len(token_yield) - 1]
tree.add_child(parent, full_token_yield.index(token))
elif len(token_yield) == 1:
tree.add_child(parent, full_token_yield.index(token_yield[0]))
return tree
def main():
# train_path = '../res/SPMRL_SHARED_2014_NO_ARABIC/GERMAN_SPMRL/gold/xml/train/train.German.gold.xml'
# corpus = sentence_names_to_hybridtrees(["s" + str(i) for i in range(1, 10)], file_name=train_path, hold=False)
train_path = '../res/SPMRL_SHARED_2014_NO_ARABIC/GERMAN_SPMRL/gold/xml/dev/dev.German.gold.xml'
names = ["s" + str(i) for i in range(40675, 40700)]
names = ['s40564']
corpus = sentence_names_to_hybridtrees(names, path=train_path, hold=False)
cp = TreeComparator()
tree_sys = ConstituentTree()
tree_sys.add_node('0', ConstituentCategory('PN'))
tree_sys.add_node('1', corpus[0].token_yield()[0], order=True)
tree_sys.add_punct("3", '$.', '.')
tree_sys.add_to_root('0')
tree_sys.add_child('0', '1')
param = build_param()
for i, hybridtree in enumerate(corpus):
print(i)
# discotree = convert_tree(hybridtree)
tree, sent = convert_tree(hybridtree)
tree2, sent2 = convert_tree(tree_sys)
if i == 11:
pass
# print(discotree)
# print(discotree.draw())
# print(DrawTree(discotree, discotree.sent))
print(DrawTree(tree, sent))
print(' '.join(map(lambda x: x.form(), hybridtree.full_token_yield())))
print(DrawTree(tree2, sent2))
print(tree[::-1])
print('POS', tree.pos())
result = TreePairResult(i, tree, sent, tree2, sent2, param)
print(result.scores())
print("Comparator: ", cp.compare_hybridtrees(hybridtree, hybridtree))
def constituent_tree_1_pos_stripped():
tree = ConstituentTree("s1")
tree.add_leaf("f1", "--", "hat")
tree.add_leaf("f2", "--", "schnell")
tree.add_leaf("f3", "--", "gearbeitet")
tree.add_punct("f4", "--", ".")
tree.set_label("V", "V")
tree.add_child("V", "f1")
tree.add_child("V", "f3")
tree.set_label("ADV", "ADV")
tree.add_child("ADV", "f2")
tree.set_label("VP", "VP")
tree.add_child("VP", "V")
tree.add_child("VP", "ADV")
tree.add_to_root("VP")
return tree
def flat_dummy_constituent_tree(token_yield,
full_token_yield,
dummy_label,
dummy_root,
label=None,
gold_pos=True):
"""
:param token_yield: connected yield of a parse tree
:type token_yield: list[ConstituentTerminal]
:param full_token_yield: full yield of the parse tree
:type full_token_yield: list[ConstituentTerminal]
:return: dummy constituent tree
:rtype: ConstituentTree
generates a flat dummy tree for a given yield where all nodes are attached under the root
"""
tree = ConstituentTree(label)
# generate root node
root_id = 'n_root'
tree.add_node(root_id, ConstituentCategory(dummy_root))
tree.add_to_root(root_id)
parent = root_id
# create all leaves and punctuation
for token in full_token_yield:
pos = token.pos() if gold_pos else '--'
if token not in token_yield:
tree.add_punct(full_token_yield.index(token), pos, token.form())
else:
idx = full_token_yield.index(token)
tree.add_leaf(idx,
pos,
token.form(),
morph=token.morph_feats(),
lemma=token.lemma())
tree.add_child(parent, idx)
return tree
def constituent_tree_2():
tree = ConstituentTree("s2")
tree.add_leaf("l1", "N", "John")
tree.add_leaf("l2", "V", "hit")
tree.add_leaf("l3", "D", "the")
tree.add_leaf("l4", "N", "Ball")
tree.add_punct("l5", "PUNC", ".")
tree.set_label("NP", "NP")
tree.add_child("NP", "l3")
tree.add_child("NP", "l4")
tree.set_label("VP", "VP")
tree.add_child("VP", "l2")
tree.add_child("VP", "NP")
tree.set_label("S", "S")
tree.add_child("S", "l1")
tree.add_child("S", "VP")
tree.add_to_root("S")
return tree
def setUp(self):
tree = ConstituentTree("s1")
tree.add_leaf("f1",
"VAFIN",
"hat",
morph=[("number", "Sg"), ("person", "3"),
("tense", "Past"), ("mood", "Ind")])
tree.add_leaf("f2", "ADV", "schnell", morph=[("degree", "Pos")])
tree.add_leaf("f3", "VVPP", "gearbeitet")
tree.add_punct("f4", "PUNC", ".")
tree.add_child("VP2", "f1")
tree.add_child("VP2", "f3")
tree.add_child("ADVP", "f2")
tree.add_child("VP1", "VP2")
tree.add_child("VP1", "ADVP")
tree.set_label("VP2", "VP")
tree.set_label("VP1", "VP")
tree.set_label("ADVP", "ADVP")
self.tree = tree
tree2 = ConstituentTree("s2")
tree2.add_leaf("f1",
"VAFIN",
"haben",
morph=[("number", "Pl"), ("person", "3"),
("tense", "Past"), ("mood", "Ind")])
tree2.add_leaf("f2", "ADV", "gut", morph=[("degree", "Pos")])
tree2.add_leaf("f3", "VVPP", "gekocht")
tree2.add_punct("f4", "PUNC", ".")
tree2.add_child("VP2", "f1")
tree2.add_child("VP2", "f3")
tree2.add_child("ADVP", "f2")
tree2.add_child("VP1", "VP2")
tree2.add_child("VP1", "ADVP")
tree2.set_label("VP2", "VP")
tree2.set_label("VP1", "VP")
tree2.set_label("ADVP", "ADVP")
tree2.add_to_root("VP1")
self.tree2 = tree2
self.tree3 = ConstituentTree("s3")
self.tree3.add_leaf("f1",
"ADJA",
"Allgemeiner",
edge="NK",
morph=[("number", "Sg")])
self.tree3.add_leaf("f2",
"ADJA",
"Deutscher",
edge="NK",
morph=[("degree", "Pos"), ("number", "Sg")])
self.tree3.add_leaf("f3",
"NN",
"Fahrrad",
edge="NK",
morph=[("number", "Sg"), ("gender", "Neut")])
self.tree3.add_leaf("f4",
"NN",
"Club",
edge="NK",
morph=[("number", "Sg"), ("gender", "Neut")])
for i in range(1, 5):
self.tree3.add_child("NP", "f" + str(i))
self.tree3.set_label("NP", "NP")
self.tree3.add_to_root("NP")