def wer(hyp: List[str], ref: List[str]) -> Tuple[float]:
"""
Compute edit distance between two str list
Args:
hlist: list[str], hypothesis
rlist: list[str], reference
Return:
float: three error types (sub/ins/del)
"""
error, match, ops = ed.edit_distance_backpointer(hyp, ref)
sub_err, ins_err, del_err, equal = 0, 0, 0, 0
for op in ops:
if op[0] == "delete":
del_err += 1
elif op[0] == "replace":
sub_err += 1
elif op[0] == "insert":
ins_err += 1
else:
equal += 1
if sub_err + del_err + ins_err != error:
raise RuntimeError("Bugs: sub_err + del_err + ins_err != #error")
if equal != match:
raise RuntimeError("Bugs: equal != #match")
return (sub_err, ins_err, del_err)
def test_edit_distance0(self):
"""Test edit distance between 'ab' and 'dab'."""
a = ['a', 'b']
b = ['d', 'a', 'b']
self.assertEqual(edit_distance(a, b), (1, 2))
bp_expected_result = (1, 2, [['insert', 0, 0, 0, 1],
['equal', 0, 1, 1, 2],
['equal', 1, 2, 2, 3]])
self.assertEqual(edit_distance_backpointer(a, b), bp_expected_result)
def test_edit_distance3(self):
"""Test for 'are you at work now'."""
a = ['are', 'you', 'at', 'work', 'now']
b = ['i', 'feel', 'are', 'saying']
bp_expected_result = (5, 0, [['delete', 0, 1, 0, 0],
['replace', 1, 2, 0, 1],
['replace', 2, 3, 1, 2],
['replace', 3, 4, 2, 3],
['replace', 4, 5, 3, 4]])
self.assertTrue(edit_distance_backpointer(a, b) == bp_expected_result)
def test_edit_distance3(self):
"""Test for 'are you at work now'."""
a = ['are', 'you', 'at', 'work', 'now']
b = ['i', 'feel', 'are', 'saying']
bp_expected_result = (5, 0, [['delete', 0, 1, 0, 0],
['replace', 1, 2, 0, 1],
['replace', 2, 3, 1, 2],
['replace', 3, 4, 2, 3],
['replace', 4, 5, 3, 4]])
self.assertTrue(edit_distance_backpointer(a, b) == bp_expected_result)
def test_edit_distance0(self):
"""Test edit distance between 'ab' and 'dab'."""
a = ["a", "b"]
b = ["d", "a", "b"]
self.assertEqual(edit_distance(a, b), (1, 2))
bp_expected_result = (
1,
2,
[["insert", 0, 0, 0, 1], ["equal", 0, 1, 1, 2], ["equal", 1, 2, 2, 3]],
)
self.assertEqual(edit_distance_backpointer(a, b), bp_expected_result)
def test_edit_distance2(self):
"""Test edit distance for 'hi my name is andy'."""
a = ['hi', 'my', 'name', 'is', 'andy']
b = ['hi', "i'm", 'my', "name's", 'sandy']
self.assertTrue(edit_distance(a, b) == (4, 1))
bp_expected_result = (4, 1, [['equal', 0, 1, 0, 1],
['replace', 1, 2, 1, 2],
['replace', 2, 3, 2, 3],
['replace', 3, 4, 3, 4],
['replace', 4, 5, 4, 5]])
self.assertTrue(edit_distance_backpointer(a, b) == bp_expected_result)
def test_edit_distance1(self):
"""Test edit distance between 'ab' and 'acdab'."""
a = ['a', 'b']
b = ['a', 'c', 'd', 'a', 'b']
self.assertTrue(edit_distance(a, b) == (3, 2))
bp_expected_result = (3, 2, [['insert', 0, 0, 0, 1],
['insert', 0, 0, 1, 2],
['insert', 0, 0, 2, 3],
['equal', 0, 1, 3, 4],
['equal', 1, 2, 4, 5]])
self.assertTrue(edit_distance_backpointer(a, b) == bp_expected_result)
def test_edit_distance2(self):
"""Test edit distance for 'hi my name is andy'."""
a = ['hi', 'my', 'name', 'is', 'andy']
b = ['hi', "i'm", 'my', "name's", 'sandy']
self.assertTrue(edit_distance(a, b) == (4, 1))
bp_expected_result = (4, 1, [['equal', 0, 1, 0, 1],
['replace', 1, 2, 1, 2],
['replace', 2, 3, 2, 3],
['replace', 3, 4, 3, 4],
['replace', 4, 5, 4, 5]])
self.assertTrue(edit_distance_backpointer(a, b) == bp_expected_result)
def test_edit_distance1(self):
"""Test edit distance between 'ab' and 'acdab'."""
a = ['a', 'b']
b = ['a', 'c', 'd', 'a', 'b']
self.assertTrue(edit_distance(a, b) == (3, 2))
bp_expected_result = (3, 2, [['insert', 0, 0, 0, 1],
['insert', 0, 0, 1, 2],
['insert', 0, 0, 2, 3],
['equal', 0, 1, 3, 4],
['equal', 1, 2, 4, 5]])
self.assertTrue(edit_distance_backpointer(a, b) == bp_expected_result)
def test_edit_distance_highest_match(self):
"""Test edit distance for 'hi my name is andy', maximizing matches rather than
minimizing edits."""
a = ['hi', 'my', 'name', 'is', 'andy']
b = ['hi', "i'm", 'my', "name's", 'sandy']
self.assertTrue(edit_distance(a, b, action_function=highest_match_action) == (4, 2))
bp_expected_result = (4, 2, [['equal', 0, 1, 0, 1],
['insert', 0, 0, 1, 2],
['equal', 1, 2, 2, 3],
['delete', 2, 3, 2, 2],
['replace', 3, 4, 3, 4],
['replace', 4, 5, 4, 5]])
self.assertTrue(edit_distance_backpointer(a, b, action_function=highest_match_action) == bp_expected_result)
def test_edit_distance3(self):
"""Test for 'are you at work now'."""
a = ["are", "you", "at", "work", "now"]
b = ["i", "feel", "are", "saying"]
bp_expected_result = (
5,
0,
[
["delete", 0, 1, 0, 0],
["replace", 1, 2, 0, 1],
["replace", 2, 3, 1, 2],
["replace", 3, 4, 2, 3],
["replace", 4, 5, 3, 4],
],
)
self.assertEqual(edit_distance_backpointer(a, b), bp_expected_result)
def test_edit_distance2(self):
"""Test edit distance for 'hi my name is andy'."""
a = ["hi", "my", "name", "is", "andy"]
b = ["hi", "i'm", "my", "name's", "sandy"]
self.assertEqual(edit_distance(a, b), (4, 1))
bp_expected_result = (
4,
1,
[
["equal", 0, 1, 0, 1],
["replace", 1, 2, 1, 2],
["replace", 2, 3, 2, 3],
["replace", 3, 4, 3, 4],
["replace", 4, 5, 4, 5],
],
)
self.assertEqual(edit_distance_backpointer(a, b), bp_expected_result)
def test_edit_distance1(self):
"""Test edit distance between 'ab' and 'acdab'."""
a = ["a", "b"]
b = ["a", "c", "d", "a", "b"]
self.assertEqual(edit_distance(a, b), (3, 2))
bp_expected_result = (
3,
2,
[
["insert", 0, 0, 0, 1],
["insert", 0, 0, 1, 2],
["insert", 0, 0, 2, 3],
["equal", 0, 1, 3, 4],
["equal", 1, 2, 4, 5],
],
)
self.assertEqual(edit_distance_backpointer(a, b), bp_expected_result)
def test_edit_distance_highest_match(self):
"""Test edit distance for 'hi my name is andy', maximizing matches rather than
minimizing edits."""
a = ['hi', 'my', 'name', 'is', 'andy']
b = ['hi', "i'm", 'my', "name's", 'sandy']
self.assertTrue(
edit_distance(a, b, action_function=highest_match_action) == (4,
2))
bp_expected_result = (4, 2, [['equal', 0, 1, 0, 1],
['insert', 0, 0, 1, 2],
['equal', 1, 2, 2, 3],
['delete', 2, 3, 2, 2],
['replace', 3, 4, 3, 4],
['replace', 4, 5, 4, 5]])
self.assertTrue(
edit_distance_backpointer(
a, b, action_function=highest_match_action) ==
bp_expected_result)
def test_edit_distance_highest_match(self):
"""Test edit distance for 'hi my name is andy', maximizing matches rather than
minimizing edits."""
a = ["hi", "my", "name", "is", "andy"]
b = ["hi", "i'm", "my", "name's", "sandy"]
self.assertEqual(
edit_distance(a, b, action_function=highest_match_action), (4, 2)
)
bp_expected_result = (
4,
2,
[
["equal", 0, 1, 0, 1],
["insert", 1, 1, 1, 2],
["equal", 1, 2, 2, 3],
["delete", 2, 3, 2, 2],
["replace", 3, 4, 3, 4],
["replace", 4, 5, 4, 5],
],
)
self.assertEqual(
edit_distance_backpointer(a, b, action_function=highest_match_action),
bp_expected_result,
)
