def test_movement_index(self):
for tracetype in ["T", "ICH", "CL"]:
print(tracetype)
# Regular trace
t = T.parse("(NP *%s*-1)" % (tracetype,))
self.assertEqual(t.metadata.index, 1)
self.assertEqual(t.metadata.idx_type, util.IDX_REGULAR)
self.assertEqual(t.text, "*%s*" % (tracetype,))
self.assertEqual(t.label, "NP")
# NB: traces cannot be gaps!
t = T.parse("(NP *%s*=1)" % (tracetype,))
self.assertIsNone(t.metadata.index)
self.assertIsNone(t.metadata.idx_type)
self.assertEqual(t.text, "*%s*=1" % (tracetype,))
self.assertEqual(t.label, "NP")
# Invalid trace
t = T.parse("(NP *%s*-X)" % (tracetype,))
self.assertIsNone(t.metadata.index)
self.assertIsNone(t.metadata.idx_type)
self.assertEqual(t.text, "*%s*-X" % (tracetype,))
self.assertEqual(t.label, "NP")
# Invalid trace (gap)
t = T.parse("(NP *%s*=X)" % (tracetype,))
self.assertIsNone(t.metadata.index)
self.assertIsNone(t.metadata.idx_type)
self.assertEqual(t.text, "*%s*=X" % (tracetype,))
self.assertEqual(t.label, "NP")
def test_str_indices(self):
t = T.parse("( (IP=1 (FOO bar)))")
self.assertEqual(str(t), "( (IP=1 (FOO bar)))")
t = T.parse("( (IP (NP-SBJ *T*-1)))")
self.assertEqual(str(t), "( (IP (NP-SBJ *T*-1)))")
# TODO: the below is a real issue...what to do about leaves at the top
# level???
raise SkipTest
t = T.parse("( (IP=1 foo))")
self.assertEqual(str(t), "( (IP=1 foo))")
def tree(self, line, cell):
# TODO: are we sure we like this method of assigning the tree to a
# variable?
t = tree.parse(cell)
if line is not None and line.isidentifier() and not keyword.iskeyword(line):
self.shell.user_ns[line] = t
return t
def test_parse(self):
self.assertIsNone(T.parse(""))
self.assertIsNone(T.parse(" \n "))
self.assertRaises(T.ParseError, lambda: T.parse("(FOO"))
self.assertRaises(T.ParseError, lambda: T.parse("(FOO))"))
self.assertRaises(T.ParseError, lambda: T.parse("(FOO bar))"))
self.assertRaises(T.ParseError, lambda: T.parse("(FOO)"))
self.assertRaises(T.ParseError, lambda: T.parse("(FOO bar baz)"))
def test_str(self):
t = T.parse("""( (IP (NP (D I)) (VBP love) (NP (NPR Python) (NPR programming))))""")
s = str(t)
self.assertIsInstance(s, str)
self.assertMultiLineEqual(s, textwrap.dedent(
"""
( (IP (NP (D I))
(VBP love)
(NP (NPR Python)
(NPR programming))))
""").strip())
def test_id(self):
t = T.parse("""( (IP (NP (D I)) (VBP love)
(NP (NPR Python) (NPR programming)))(ID foo))""")
# Test that ID is parsed
self.assertEqual(t.metadata["ID"], "foo")
# Test that str works
s = str(t)
self.assertIsInstance(s, str)
self.assertMultiLineEqual(s, textwrap.dedent(
"""
( (IP (NP (D I))
(VBP love)
(NP (NPR Python)
(NPR programming)))
(ID foo))
""").strip())
t2 = T.parse("""( (ID foo)(IP (NP (D I)) (VBP love)
(NP (NPR Python) (NPR programming))))""")
# Test that the order of the ID node doesn't matter to parsing
self.assertEqual(t2.metadata.id, "foo")
self.assertEqual(t, t2)
self.assertEqual(s, str(t2))
def test_urtext(self):
# TODO: woefully incomplete
tree = "( (IP-MAT (X *T*) (X FOO) (X *con*) (XP (X bar) (X BAZ) (. .)) (CODE dddd)))"
self.assertEqual(T.parse(tree).urtext, "FOO bar BAZ.")
def test_eq(self):
t = T.parse("(NP (N foo))")
t2 = T.parse("(NP (N foo))")
self.assertEqual(t, t2)
def test_parse_1(self):
t = T.parse("( (ID foo) (IP (NP (PRO it)) (VBP works)))")
self.assertIsInstance(t, T.NonTerminal)
self.assertEqual(t.metadata.id, 'foo')
self.assertEqual(t, NT("IP", [NT("NP", [L("PRO", "it")]), L("VBP", "works")], {"ID": "foo"}))
