def test_boundary_edit_distance(self):
'''
Test boundary_edit_distance.
'''
edits = ([(1, 'b'), (1, 'b'), (1, 'b')], [], [(9, 10, 1)])
self.assertEquals(edits, boundary_edit_distance(
boundary_string_from_masses(self.masses_an1),
boundary_string_from_masses(self.masses_an2)))
def boundary_string_from_boundary_indices(segmentation, doc_length):
"""converts boundary indices to segeval-compatible boundary strings
Args:
segmentation (list of int): list of segmentation boundary indices
doc_length (int): length of the segmented document
Returns:
tuple: Boundary string
"""
i = 1
tokens_in_segment = 0
masses = []
current_seg_index = 0
while i < doc_length:
tokens_in_segment += 1
if current_seg_index < len(
segmentation) and i > segmentation[current_seg_index] - 1:
masses.append(tokens_in_segment)
tokens_in_segment = 0
current_seg_index += 1
i += 1
masses.append(doc_length - segmentation[-1])
return segeval.boundary_string_from_masses(tuple(masses))
def test_boundary_string_from_masses(self):
'''
Test boundary_string_from_masses.
'''
self.assertEquals((
frozenset([]), frozenset([]), frozenset([]), frozenset([]),
frozenset([]), frozenset([]), frozenset([]), frozenset([]),
frozenset([]), frozenset([]), frozenset([1]), frozenset([])),
boundary_string_from_masses(self.masses_an1))
