def test_init(self):
with pytest.raises(TypeError):
D()
with pytest.raises(TypeError):
D(1)
D(1, 'some-thing')
D(1, 'some-thing', 0.5, 0, 3, [T('some-token')])
# roots are special: id is None, entity is root, daughters must
# exactly 1 node; rest are None
with pytest.raises(TypeError):
D(None)
with pytest.raises(TypeError):
D(None, 'some-root', 0.5)
with pytest.raises(TypeError):
D(None, 'some-root', start=1)
with pytest.raises(TypeError):
D(None, 'some-root', end=1)
with pytest.raises(ValueError):
D(None,
'some-root',
daughters=[N(1, 'some-thing'),
N(2, 'some-thing')])
with pytest.raises(ValueError):
D(None, 'some-root', daughters=[T('some-token')])
D(None, 'some-root', daughters=[N(1, 'some-thing')])
D(None, 'some-root', None, None, None, daughters=[N(1, 'some-thing')])
# root not as top
with pytest.raises(ValueError):
D(1,
'some-thing',
daughters=[
N(None,
'some-root',
daughters=[N(2, 'a-lex', daughters=[T('some-token')])])
])
def test_attributes(self):
t = D(1, 'some-thing')
assert t.id == 1
assert t.entity == 'some-thing'
assert t.score == -1
assert t.start == -1
assert t.end == -1
assert t.daughters == []
t = D(1, 'some-thing', 0.5, 1, 6, [T('some token')])
assert t.id == 1
assert t.entity == 'some-thing'
assert t.score == 0.5
assert t.start == 1
assert t.end == 6
assert t.daughters == [T('some token')]
t = D(None, 'some-root', daughters=[D(1, 'some-thing')])
assert t.id is None
assert t.entity == 'some-root'
assert t.score is None
assert t.start is None
assert t.end is None
assert len(t.daughters) == 1
def test_fromstring(self):
with pytest.raises(DerivationSyntaxError):
from_string('')
# root with no children
# TODO: this should be a DerivationSyntaxError but the current
# UDF parser doesn't make that straightforward. Revist this
# when/if the UDF parsing changes
with pytest.raises(ValueError):
from_string('(some-root)')
# does not start with `(` or end with `)`
with pytest.raises(DerivationSyntaxError):
from_string(' (1 some-thing -1 -1 -1 ("token"))')
with pytest.raises(DerivationSyntaxError):
from_string(' (1 some-thing -1 -1 -1 ("token")) ')
# uneven parens
with pytest.raises(DerivationSyntaxError):
from_string('(1 some-thing -1 -1 -1 ("token")')
# ok
t = from_string('(1 some-thing -1 -1 -1 ("token"))')
assert t.id == 1
assert t.entity == 'some-thing'
assert t.score == -1.0
assert t.start == -1
assert t.end == -1
assert t.daughters == [T('token')]
# newlines in tree
t = from_string('''(1 some-thing -1 -1 -1
("token"))''')
assert t.id == 1
assert t.entity == 'some-thing'
assert t.score == -1.0
assert t.start == -1
assert t.end == -1
assert t.daughters == [T('token')]
# LKB-style terminals
t = from_string('''(1 some-thing -1 -1 -1
("to ken" 1 2))''')
assert t.id == 1
assert t.entity == 'some-thing'
assert t.score == -1.0
assert t.start == -1
assert t.end == -1
assert t.daughters == [T('to ken')] # start/end ignored
# TFS-style terminals
t = from_string(r'''(1 some-thing -1 -1 -1
("to ken" 2 "token [ +FORM \"to\" ]"
3 "token [ +FORM \"ken\" ]"))''')
assert t.id == 1
assert t.entity == 'some-thing'
assert t.score == -1.0
assert t.start == -1
assert t.end == -1
assert t.daughters == [
T('to ken', [
Tk(2, r'token [ +FORM \"to\" ]'),
Tk(3, r'token [ +FORM \"ken\" ]')
])
]
# longer example
t = from_string(r'''(root
(1 some-thing 0.4 0 5
(2 a-lex 0.8 0 1
("a" 1 "token [ +FORM \"a\" ]"))
(3 bcd-lex 0.5 2 5
("bcd" 2 "token [ +FORM \"bcd\" ]")))
)''')
assert t.entity == 'root'
assert len(t.daughters) == 1
top = t.daughters[0]
assert top.id == 1
assert top.entity == 'some-thing'
assert top.score == 0.4
assert top.start == 0
assert top.end == 5
assert len(top.daughters) == 2
lex = top.daughters[0]
assert lex.id == 2
assert lex.entity == 'a-lex'
assert lex.score == 0.8
assert lex.start == 0
assert lex.end == 1
assert lex.daughters == [T('a', [Tk(1, r'token [ +FORM \"a\" ]')])]
lex = top.daughters[1]
assert lex.id == 3
assert lex.entity == 'bcd-lex'
assert lex.score == 0.5
assert lex.start == 2
assert lex.end == 5
assert lex.daughters == [T('bcd', [Tk(2, r'token [ +FORM \"bcd\" ]')])]
def test_fromstring(self):
with pytest.raises(ValueError): D.from_string('')
# root with no children
with pytest.raises(ValueError): D.from_string('(some-root)')
# does not start with `(` or end with `)`
with pytest.raises(ValueError):
D.from_string(' (1 some-thing -1 -1 -1 ("token"))')
with pytest.raises(ValueError):
D.from_string(' (1 some-thing -1 -1 -1 ("token")) ')
# uneven parens
with pytest.raises(ValueError):
D.from_string('(1 some-thing -1 -1 -1 ("token")')
# ok
t = D.from_string('(1 some-thing -1 -1 -1 ("token"))')
assert t.id == 1
assert t.entity == 'some-thing'
assert t.score == -1.0
assert t.start == -1
assert t.end == -1
assert t.daughters == [T('token')]
# newlines in tree
t = D.from_string('''(1 some-thing -1 -1 -1
("token"))''')
assert t.id == 1
assert t.entity == 'some-thing'
assert t.score == -1.0
assert t.start == -1
assert t.end == -1
assert t.daughters == [T('token')]
# LKB-style terminals
t = D.from_string('''(1 some-thing -1 -1 -1
("to ken" 1 2))''')
assert t.id == 1
assert t.entity == 'some-thing'
assert t.score == -1.0
assert t.start == -1
assert t.end == -1
assert t.daughters == [T('to ken')] # start/end ignored
# TFS-style terminals
t = D.from_string(r'''(1 some-thing -1 -1 -1
("to ken" 2 "token [ +FORM \"to\" ]"
3 "token [ +FORM \"ken\" ]"))''')
assert t.id == 1
assert t.entity == 'some-thing'
assert t.score == -1.0
assert t.start == -1
assert t.end == -1
assert t.daughters == [
T('to ken', [Tk(2, r'token [ +FORM \"to\" ]'),
Tk(3, r'token [ +FORM \"ken\" ]')])
]
# longer example
t = D.from_string(r'''(root
(1 some-thing 0.4 0 5
(2 a-lex 0.8 0 1
("a" 1 "token [ +FORM \"a\" ]"))
(3 bcd-lex 0.5 2 5
("bcd" 2 "token [ +FORM \"bcd\" ]")))
)''')
assert t.entity == 'root'
assert len(t.daughters) == 1
top = t.daughters[0]
assert top.id == 1
assert top.entity == 'some-thing'
assert top.score == 0.4
assert top.start == 0
assert top.end == 5
assert len(top.daughters) == 2
lex = top.daughters[0]
assert lex.id == 2
assert lex.entity == 'a-lex'
assert lex.score == 0.8
assert lex.start == 0
assert lex.end == 1
assert lex.daughters == [T('a', [Tk(1, r'token [ +FORM \"a\" ]')])]
lex = top.daughters[1]
assert lex.id == 3
assert lex.entity == 'bcd-lex'
assert lex.score == 0.5
assert lex.start == 2
assert lex.end == 5
assert lex.daughters == [T('bcd',
[Tk(2, r'token [ +FORM \"bcd\" ]')])]