def create_ne_segments(values, starts, ends, document, segstorage):
segments = []
segment = None
for value, start, end in izip(values, starts, ends):
if value.startswith('B-'):
ne_value = value[2:]
if segment is not None:
segments.append(segment)
segment = Segment(u'ne_type', ne_value, document, start, end)
segment = Segment(u'ne_type', ne_value, document, start, end)
elif value.startswith('I-'):
segment.end = end
elif value.startswith('O') and segment is not None:
segments.append(segment)
segment = None
if segment is not None:
segment.end = len(document.text)
segments.append(segment)
segstorage.save(segments)
def test_to_dict(self):
A = Segment(self.name(), self.value(), self.document(), self.start(), self.end())
self.assertEqual(Segment.to_dict(A), self.dictionary())
def test_from_dict(self):
A = Segment(self.name(), self.value(), self.document(), self.start(), self.end())
B = Segment.from_dict(self.dictionary())
self.assertEqual(A, B)