def test_print_subtree(self):
"""Test print_subtree() method, which uses udapi.block.write.textmodetrees."""
doc = Document()
data_filename = os.path.join(os.path.dirname(__file__), 'data', 'enh_deps.conllu')
doc.load_conllu(data_filename)
root = doc.bundles[0].get_tree()
expected1 = ("# sent_id = a-mf920901-001-p1s1A\n"
"# text = Slovenská ústava: pro i proti\n"
"─┮\n"
" │ ╭─╼ Slovenská ADJ amod\n"
" ╰─┾ ústava NOUN root\n"
" ┡─╼ : PUNCT punct\n"
" ╰─┮ pro ADP appos\n"
" ┡─╼ i CONJ cc\n"
" ╰─╼ proti ADP conj\n"
"\n")
expected2 = ("─┮\n"
" │ ╭─╼ Slovenská Case=Nom|Degree=Pos|Gender=Fem|Negative=Pos|Number=Sing _\n"
" ╰─┾ ústava Case=Nom|Gender=Fem|Negative=Pos|Number=Sing SpaceAfter=No\n"
" ┡─╼ : _ _\n"
" ╰─┮ pro AdpType=Prep|Case=Acc LId=pro-1\n"
" ┡─╼ i _ LId=i-1\n"
" ╰─╼ proti AdpType=Prep|Case=Dat LId=proti-1\n"
"\n")
# test non-projective tree
root3 = Root()
for i in range(1, 5):
root3.create_child(form=str(i))
nodes = root3.descendants(add_self=1)
nodes[1].parent = nodes[3]
nodes[4].parent = nodes[2]
expected3 = ("─┮\n"
" │ ╭─╼ 1\n"
" ┡─╪───┮ 2\n"
" ╰─┶ 3 │\n"
" ╰─╼ 4\n"
"\n")
try:
sys.stdout = capture = io.StringIO()
root.print_subtree(color=False)
self.assertEqual(capture.getvalue(), expected1)
capture.seek(0)
capture.truncate()
root.print_subtree(color=False, attributes='form,feats,misc',
print_sent_id=False, print_text=False)
self.assertEqual(capture.getvalue(), expected2)
capture.seek(0)
capture.truncate()
root3.print_subtree(color=False, attributes='form', print_sent_id=0, print_text=0)
self.assertEqual(capture.getvalue(), expected3)
finally:
sys.stdout = sys.__stdout__ # pylint: disable=redefined-variable-type
def test_deps_setter(self):
"""Test the deserialization of enhanced dependencies."""
# Create a sample dependency tree.
root = Root()
for _ in range(3):
root.create_child()
nodes = root.descendants()
nodes[0].deps.append({'parent': nodes[1], 'deprel': 'test'})
self.assertEqual(nodes[0].raw_deps, '2:test')
def test_deps_setter(self):
"""Test the deserialization of enhanced dependencies."""
# Create a sample dependency tree.
root = Root()
for _ in range(3):
root.create_child()
nodes = root.descendants()
nodes[0].deps.append({'parent': nodes[1], 'deprel': 'test'})
self.assertEqual(nodes[0].raw_deps, '2:test')
def read_tree(self, document=None):
if self.filehandle is None:
return None
root = Root()
parents = [0]
words = []
form = None
for line in self.filehandle:
line = line.rstrip()
if line == '':
break
if line[0] == '#':
# Are comments allowed in VISL-cg?
continue
if line[0].isspace():
line.lstrip(line)
node, parent_ord = self._node(line, root)
words.append(node)
parents.append(parent_ord)
else:
if words:
words[0].form = form
if len(words) > 1:
split_forms = form.split()
if len(words) == len(split_forms):
for word, split_form in zip(words, split_forms):
word.form = split_form
else:
for word in words[1:]:
word.form = '_'
root.create_multiword_token(words, form=form)
words = []
form = line[2:-2]
if words:
words[0].form = form
for word in words[1:]:
word.form = '_'
nodes = root.descendants(add_self=True)
if len(nodes) == 1:
return None
for node_ord, node in enumerate(nodes[1:], 1):
try:
node.parent = nodes[parents[node_ord]]
except IndexError:
raise ValueError("Node %s HEAD is out of range (%d)" %
(node, parents[node_ord]))
return root
def read_tree(self):
if self.filehandle is None:
return None
root = Root()
parents = [0]
words = []
form = None
for line in self.filehandle:
line = line.rstrip()
if line == '':
break
if line[0] == '#':
root.comment += line[1:] + "\n"
continue
if line[0].isspace():
node, parent_ord = self._node(line.lstrip(), root)
words.append(node)
parents.append(parent_ord)
continue
if words:
words[0].form = form
if len(words) > 1:
split_forms = form.split()
if len(words) == len(split_forms):
for word, split_form in zip(words, split_forms):
word.form = split_form
else:
for word in words[1:]:
word.form = '_'
root.create_multiword_token(words, form=form)
words = []
form = line[2:-2]
if words:
words[0].form = form
for word in words[1:]:
word.form = '_'
nodes = root.descendants(add_self=True)
if len(nodes) == 1:
return None
for node_ord, node in enumerate(nodes[1:], 1):
try:
node.parent = nodes[parents[node_ord]]
except IndexError:
raise ValueError("Node %s HEAD is out of range (%d)" % (node, parents[node_ord]))
return root
def test_print_subtree(self):
"""Test print_subtree() method, which uses udapi.block.write.textmodetrees."""
doc = Document()
data_filename = os.path.join(os.path.dirname(__file__), 'data',
'enh_deps.conllu')
doc.load_conllu(data_filename)
root = doc.bundles[0].get_tree()
expected1 = ("# sent_id = a-mf920901-001-p1s1A\n"
"# text = Slovenská ústava: pro i proti\n"
"─┮\n"
" │ ╭─╼ Slovenská ADJ amod\n"
" ╰─┾ ústava NOUN root\n"
" ┡─╼ : PUNCT punct\n"
" ╰─┮ pro ADP appos\n"
" ┡─╼ i CONJ cc\n"
" ╰─╼ proti ADP conj\n"
"\n")
expected2 = (
"─┮\n"
" │ ╭─╼ Slovenská Case=Nom|Degree=Pos|Gender=Fem|Negative=Pos|Number=Sing _\n"
" ╰─┾ ústava Case=Nom|Gender=Fem|Negative=Pos|Number=Sing SpaceAfter=No\n"
" ┡─╼ : _ _\n"
" ╰─┮ pro AdpType=Prep|Case=Acc LId=pro-1\n"
" ┡─╼ i _ LId=i-1\n"
" ╰─╼ proti AdpType=Prep|Case=Dat LId=proti-1\n"
"\n")
# test non-projective tree
root3 = Root()
for i in range(1, 5):
root3.create_child(form=str(i))
nodes = root3.descendants(add_self=1)
nodes[1].parent = nodes[3]
nodes[4].parent = nodes[2]
expected3 = ("─┮\n"
" │ ╭─╼ 1\n"
" ┡─╪───┮ 2\n"
" ╰─┶ 3 │\n"
" ╰─╼ 4\n"
"\n")
try:
sys.stdout = capture = io.StringIO()
root.print_subtree(color=False)
self.assertEqual(capture.getvalue(), expected1)
capture.seek(0)
capture.truncate()
root.print_subtree(color=False,
attributes='form,feats,misc',
print_sent_id=False,
print_text=False)
self.assertEqual(capture.getvalue(), expected2)
capture.seek(0)
capture.truncate()
root3.print_subtree(color=False,
attributes='form',
print_sent_id=0,
print_text=0)
self.assertEqual(capture.getvalue(), expected3)
finally:
sys.stdout = sys.__stdout__ # pylint: disable=redefined-variable-type