def test_reconcile_TIMEX(self):
s = Timex3XmlDocument("<root>This is some annotated text.</root>")
t = Timex(type="date")
t1 = Timex(id=1)
t2 = Timex(id=2)
t3 = Timex(id=3)
t.value = "20100710"
t.id = 6
t.mod = "BEFORE"
t.freq = "1M"
t.comment = "Test"
t.quant = "EVERY"
t.temporal_function = True
t.document_role = "MODIFICATION_TIME"
t.begin_timex = t1
t.end_timex = t2
t.context = t3
s.reconcile(
[
[
("This", "POS", set()),
("is", "POS", set()),
("some", "POS", {t}),
("annotated", "POS", {t}),
("text", "POS", {t}),
(".", "POS", set()),
]
]
)
self.assertEquals(
str(s),
xml.dom.minidom.parseString(
'<root>This is <TIMEX3 tid="t6" beginPoint="t1" endPoint="t2" anchorTimeID="t3" functionInDocument="MODIFICATION_TIME" temporalFunction="true" type="DATE" value="20100710" mod="BEFORE" freq="1M" comment="Test" quant="EVERY">some annotated text</TIMEX3>.</root>'
).toxml(),
)
def test_reconcile_TIMEX(self):
s = Timex3XmlDocument('<root>This is some annotated text.</root>')
t = Timex(type='date')
t1 = Timex(id=1)
t2 = Timex(id=2)
t3 = Timex(id=3)
t.value = "20100710"
t.id = 6
t.mod = "BEFORE"
t.freq = "1M"
t.comment = "Test"
t.quant = 'EVERY'
t.temporal_function = True
t.document_role = 'MODIFICATION_TIME'
t.begin_timex = t1
t.end_timex = t2
t.context = t3
s.reconcile([[('This', 'POS', set()), ('is', 'POS', set()),
('some', 'POS', {t}), ('annotated', 'POS', {t}),
('text', 'POS', {t}), ('.', 'POS', set())]])
self.assertEqual(
str(s),
xml.dom.minidom.parseString(
'<root>This is <TIMEX3 tid="t6" beginPoint="t1" endPoint="t2" anchorTimeID="t3" functionInDocument="MODIFICATION_TIME" temporalFunction="true" type="DATE" value="20100710" mod="BEFORE" freq="1M" comment="Test" quant="EVERY">some annotated text</TIMEX3>.</root>'
).toxml())
def _timex_from_node(self, node):
"""
Given a node representing a TIMEX3 element, return a timex object
representing it
"""
t = Timex()
if node.hasAttribute('tid'):
t.id = int(node.getAttribute('tid')[1:])
if node.hasAttribute('value'):
t.value = node.getAttribute('value')
if node.hasAttribute('mod'):
t.mod = node.getAttribute('mod')
if node.hasAttribute('type'):
t.type = node.getAttribute('type')
if node.hasAttribute('freq'):
t.freq = node.getAttribute('freq')
if node.hasAttribute('quant'):
t.quant = node.getAttribute('quant')
if node.hasAttribute('comment'):
t.comment = node.getAttribute('comment')
if node.getAttribute('temporalFunction'):
t.temporal_function = True
if node.hasAttribute('functionInDocument'):
t.document_role = node.getAttribute('functionInDocument')
if node.hasAttribute('beginPoint'):
t.begin_timex = int(node.getAttribute('beginPoint')[1:])
if node.hasAttribute('endPoint'):
t.end_timex = int(node.getAttribute('endPoint')[1:])
if node.hasAttribute('anchorTimeID'):
t.context = int(node.getAttribute('anchorTimeID')[1:])
return t
def _timex_from_node(self, node):
"""
Given a node representing a TIMEX3 element, return a timex object
representing it
"""
t = Timex()
if node.hasAttribute('tid'):
t.id = int(node.getAttribute('tid')[1:])
if node.hasAttribute('value'):
t.value = node.getAttribute('value')
if node.hasAttribute('mod'):
t.mod = node.getAttribute('mod')
if node.hasAttribute('type'):
t.type = node.getAttribute('type')
if node.hasAttribute('freq'):
t.freq = node.getAttribute('freq')
if node.hasAttribute('quant'):
t.quant = node.getAttribute('quant')
if node.hasAttribute('comment'):
t.comment = node.getAttribute('comment')
if node.getAttribute('temporalFunction'):
t.temporal_function = True
if node.hasAttribute('functionInDocument'):
t.document_role = node.getAttribute('functionInDocument')
if node.hasAttribute('beginPoint'):
t.begin_timex = int(node.getAttribute('beginPoint')[1:])
if node.hasAttribute('endPoint'):
t.end_timex = int(node.getAttribute('endPoint')[1:])
if node.hasAttribute('anchorTimeID'):
t.context = int(node.getAttribute('anchorTimeID')[1:])
return t
def test_assign_IDs_consecutive(self):
# Get some sample IDs
ts = set([Timex(), Timex(), Timex()])
at = Timex()
at.id = 2
ts.add(at)
add_timex_ids(ts)
# Get the assigned IDs
tids = set()
for t in ts:
tids.add(t.id)
# Should be exactly 4 unique IDs and pre-assigned one hasn't changed
self.assertEquals(len(tids), 4)
self.assertEquals(2, at.id)
# Should be consecutive for new ones
self.assertTrue(1 in tids)
self.assertTrue(2 in tids)
self.assertTrue(3 in tids)
self.assertTrue(4 in tids)
