fixed last failing tests, while beginning test overhaul

This commit is contained in:
Tal Einat 2014-05-06 20:51:04 +03:00
parent 88bead7c7a
commit ed765cff2b
2 changed files with 60 additions and 137 deletions

View File

@ -6,7 +6,8 @@ from tests.test_substitutions_only import TestSubstitionsOnlyBase, \
from fuzzysearch.generic_search import \ from fuzzysearch.generic_search import \
_find_near_matches_generic_linear_programming as fnm_generic_lp, \ _find_near_matches_generic_linear_programming as fnm_generic_lp, \
find_near_matches_generic_ngrams as fnm_generic_ngrams, \ find_near_matches_generic_ngrams as fnm_generic_ngrams, \
has_near_match_generic_ngrams as hnm_generic_ngrams has_near_match_generic_ngrams as hnm_generic_ngrams, \
find_near_matches_generic
class TestGenericSearchLpAsLevenshtein(TestFindNearMatchesLevenshteinBase, class TestGenericSearchLpAsLevenshtein(TestFindNearMatchesLevenshteinBase,
@ -53,16 +54,19 @@ class TestGenericSearchBase(object):
max_l_dist=None): max_l_dist=None):
raise NotImplementedError raise NotImplementedError
def expectedOutcomes(self, search_result, expected_outcomes):
raise NotImplementedError
def test_empty_sequence(self): def test_empty_sequence(self):
self.assertEqual(self.search('PATTERN', '', 0, 0, 0), []) self.assertEqual(self.search('PATTERN', '', 0, 0, 0, 0), [])
def test_empty_subsequence_exeption(self): def test_empty_subsequence_exeption(self):
with self.assertRaises(ValueError): with self.assertRaises(ValueError):
self.search('', 'TEXT', 0, 0, 0) self.search('', 'TEXT', 0, 0, 0, 0)
def test_match_identical_sequence(self): def test_match_identical_sequence(self):
self.assertEqual( self.assertEqual(
self.search('PATTERN', 'PATTERN', 0, 0, 0), self.search('PATTERN', 'PATTERN', 0, 0, 0, 0),
[Match(start=0, end=len('PATTERN'), dist=0)], [Match(start=0, end=len('PATTERN'), dist=0)],
) )
@ -72,7 +76,7 @@ class TestGenericSearchBase(object):
expected_match = Match(start=10, end=17, dist=0) expected_match = Match(start=10, end=17, dist=0)
self.assertEqual( self.assertEqual(
self.search(substring, text, 0, 0, 0), self.search(substring, text, 0, 0, 0, 0),
[expected_match], [expected_match],
) )
@ -80,23 +84,24 @@ class TestGenericSearchBase(object):
# sequence = 'abcdefg' # sequence = 'abcdefg'
# pattern = 'bde' # pattern = 'bde'
self.assertEqual( self.expectedOutcomes(
self.search('def', 'abcddefg', 0, 0, 0), self.search('def', 'abcddefg', 0, 0, 0, 0),
[Match(start=4, end=7, dist=0)], [Match(start=4, end=7, dist=0)],
) )
self.assertEqual( self.expectedOutcomes(
self.search('def', 'abcddefg', 1, 0, 0), self.search('def', 'abcddefg', 1, 0, 0, 1),
[Match(start=4, end=7, dist=0)], [Match(start=4, end=7, dist=0)],
) )
self.assertIn( self.expectedOutcomes(
Match(start=4, end=7, dist=0), self.search('def', 'abcddefg', 0, 0, 1, 1),
self.search('def', 'abcddefg', 0, 0, 1), [Match(start=4, end=7, dist=0),
Match(start=5, end=7, dist=1)]
) )
self.assertEqual( self.expectedOutcomes(
self.search('def', 'abcddefg', 0, 1, 0, 0), self.search('def', 'abcddefg', 0, 1, 0, 1),
[Match(start=4, end=7, dist=0)], [Match(start=4, end=7, dist=0)],
) )
@ -105,25 +110,36 @@ class TestGenericSearchBase(object):
# pattern = 'bde' # pattern = 'bde'
self.assertEqual( self.assertEqual(
self.search('bde', 'abcdefg', 0, 1, 0), self.search('bde', 'abcdefg', 0, 1, 0, 1),
[Match(start=1, end=5, dist=1)], [Match(start=1, end=5, dist=1)],
) )
self.assertEqual( self.assertEqual(
self.search('bde', 'abcdefg', 0, 0, 0), self.search('bde', 'abcdefg', 0, 0, 0, 0),
[], [],
) )
self.assertEqual( self.assertEqual(
self.search('bde', 'abcdefg', 1, 0, 0), self.search('bde', 'abcdefg', 1, 0, 0, 1),
[Match(start=2, end=5, dist=1)], [Match(start=2, end=5, dist=1)],
) )
self.assertEqual( self.assertEqual(
self.search('bde', 'abcdefg', 0, 0, 1), self.search('bde', 'abcdefg', 0, 0, 1, 1),
[Match(start=3, end=5, dist=1)], [Match(start=3, end=5, dist=1)],
) )
class TestGenericSearch(TestGenericSearchBase, unittest.TestCase):
def search(self, pattern, sequence, max_subs, max_ins, max_dels,
max_l_dist=None):
return list(find_near_matches_generic(pattern, sequence,
max_subs, max_ins,
max_dels, max_l_dist))
def expectedOutcomes(self, search_result, expected_outcomes):
return search_result == expected_outcomes
def test_valid_none_arguments(self): def test_valid_none_arguments(self):
# check that no exception is raised when some values are None # check that no exception is raised when some values are None
self.assertEqual( self.assertEqual(
@ -179,11 +195,14 @@ class TestGenericSearchLp(TestGenericSearchBase, unittest.TestCase):
return list(fnm_generic_lp(pattern, sequence, return list(fnm_generic_lp(pattern, sequence,
max_subs, max_ins, max_dels, max_l_dist)) max_subs, max_ins, max_dels, max_l_dist))
def expectedOutcomes(self, search_result, expected_outcomes):
return search_result == expected_outcomes
def test_double_first_item_two_results(self): def test_double_first_item_two_results(self):
# sequence = 'abcdefg' # sequence = 'abcdefg'
# pattern = 'bde' # pattern = 'bde'
self.assertListEqual( self.assertListEqual(
self.search('def', 'abcddefg', 0, 1, 0), self.search('def', 'abcddefg', 0, 1, 0, 1),
[Match(start=3, end=7, dist=1), Match(start=4, end=7, dist=0)], [Match(start=3, end=7, dist=1), Match(start=4, end=7, dist=0)],
) )
@ -212,6 +231,13 @@ class TestGenericSearchNgrams(TestGenericSearchBase, unittest.TestCase):
return fnm_generic_ngrams(pattern, sequence, return fnm_generic_ngrams(pattern, sequence,
max_subs, max_ins, max_dels, max_l_dist) max_subs, max_ins, max_dels, max_l_dist)
def expectedOutcomes(self, search_result, expected_outcomes):
best_from_groups = [
get_best_match_in_group(group)
for group in group_matches(search_result)
]
return search_result == best_from_groups
def test_missing_second_item_complex(self): def test_missing_second_item_complex(self):
self.assertTrue( self.assertTrue(
set(self.search('bde', 'abcdefg', 1, 1, 1, 1)).issubset([ set(self.search('bde', 'abcdefg', 1, 1, 1, 1)).issubset([
@ -236,6 +262,9 @@ class TestHasNearMatchGenericNgrams(TestGenericSearchBase, unittest.TestCase):
return hnm_generic_ngrams(pattern, sequence, return hnm_generic_ngrams(pattern, sequence,
max_subs, max_ins, max_dels, max_l_dist) max_subs, max_ins, max_dels, max_l_dist)
def expectedOutcomes(self, search_result, expected_outcomes):
return search_result == bool(expected_outcomes)
def assertEqual(self, actual_value, expected_value, *args, **kwargs): def assertEqual(self, actual_value, expected_value, *args, **kwargs):
return super(TestHasNearMatchGenericNgrams, self).assertEqual( return super(TestHasNearMatchGenericNgrams, self).assertEqual(
actual_value, bool(expected_value), *args, **kwargs) actual_value, bool(expected_value), *args, **kwargs)
@ -244,9 +273,6 @@ class TestHasNearMatchGenericNgrams(TestGenericSearchBase, unittest.TestCase):
return super(TestHasNearMatchGenericNgrams, self).assertEqual( return super(TestHasNearMatchGenericNgrams, self).assertEqual(
actual_value, bool(expected_value), *args, **kwargs) actual_value, bool(expected_value), *args, **kwargs)
def assertIn(self, member, container, *args, **kwargs):
return super(TestHasNearMatchGenericNgrams, self).assertTrue(container)
def test_missing_second_item_complex(self): def test_missing_second_item_complex(self):
# skip this because ngrams search requires that the subsequence's # skip this because ngrams search requires that the subsequence's
# length is greater than the maximum Levenshtein distance # length is greater than the maximum Levenshtein distance

View File

@ -2,6 +2,7 @@ from tests.compat import unittest
from tests.test_levenshtein import TestFindNearMatchesLevenshteinBase from tests.test_levenshtein import TestFindNearMatchesLevenshteinBase
from fuzzysearch.common import Match, get_best_match_in_group, group_matches from fuzzysearch.common import Match, get_best_match_in_group, group_matches
from tests.test_substitutions_only import TestSubstitionsOnlyBase from tests.test_substitutions_only import TestSubstitionsOnlyBase
from tests.test_generic_search import TestGenericSearchBase
try: try:
from fuzzysearch._generic_search import \ from fuzzysearch._generic_search import \
@ -65,120 +66,6 @@ else:
self).test_double_first_item() self).test_double_first_item()
class TestGenericSearchBase(object):
def test_empty_sequence(self):
self.assertEqual([], self.search('PATTERN', '', 0, 0, 0))
def test_empty_subsequence_exeption(self):
with self.assertRaises(ValueError):
self.search('', 'TEXT', 0, 0, 0)
def test_match_identical_sequence(self):
self.assertEqual(
[Match(start=0, end=len('PATTERN'), dist=0)],
self.search('PATTERN', 'PATTERN', 0, 0, 0),
)
def test_substring(self):
substring = 'PATTERN'
text = 'aaaaaaaaaaPATTERNaaaaaaaaa'
expected_match = Match(start=10, end=17, dist=0)
self.assertEqual(
[expected_match],
self.search(substring, text, 0, 0, 0)
)
def test_double_first_item(self):
# sequence = 'abcdefg'
# pattern = 'bde'
self.assertEqual(
[Match(start=4, end=7, dist=0)],
self.search('def', 'abcddefg', 0, 0, 0),
)
self.assertEqual(
[Match(start=4, end=7, dist=0)],
self.search('def', 'abcddefg', 1, 0, 0),
)
self.assertIn(
Match(start=4, end=7, dist=0),
self.search('def', 'abcddefg', 0, 0, 1),
)
self.assertEqual(
[Match(start=4, end=7, dist=0)],
self.search('def', 'abcddefg', 0, 1, 0, 0),
)
def test_missing_second_item(self):
# sequence = 'abcdefg'
# pattern = 'bde'
self.assertEqual(
self.search('bde', 'abcdefg', 0, 1, 0),
[Match(start=1, end=5, dist=1)],
)
self.assertEqual(
self.search('bde', 'abcdefg', 0, 0, 0),
[],
)
self.assertEqual(
self.search('bde', 'abcdefg', 1, 0, 0),
[Match(start=2, end=5, dist=1)],
)
self.assertEqual(
self.search('bde', 'abcdefg', 0, 0, 1),
[Match(start=3, end=5, dist=1)],
)
def test_valid_none_arguments(self):
# check that no exception is raised when some values are None
self.assertEqual(
self.search('a', 'b', 0, None, None, 0),
[],
)
self.assertEqual(
self.search('a', 'b', None, 0, None, 0),
[],
)
self.assertEqual(
self.search('a', 'b', None, None, 0, 0),
[],
)
self.assertEqual(
self.search('a', 'b', 0, 0, None, 0),
[],
)
self.assertEqual(
self.search('a', 'b', 0, None, 0, 0),
[],
)
self.assertEqual(
self.search('a', 'b', None, 0, 0, 0),
[],
)
self.assertEqual(
self.search('a', 'b', None, None, None, 0),
[],
)
self.assertEqual(
self.search('a', 'b', 0, 0, 0, None),
[],
)
class TestGenericSearchLp(TestGenericSearchBase, unittest.TestCase): class TestGenericSearchLp(TestGenericSearchBase, unittest.TestCase):
def search(self, pattern, sequence, max_subs, max_ins, max_dels, def search(self, pattern, sequence, max_subs, max_ins, max_dels,
max_l_dist=None): max_l_dist=None):
@ -187,6 +74,9 @@ else:
max_subs, max_ins, max_subs, max_ins,
max_dels, max_l_dist)) max_dels, max_l_dist))
def expectedOutcomes(self, search_result, expected_outcomes):
return search_result == expected_outcomes
def test_double_first_item_two_results(self): def test_double_first_item_two_results(self):
# sequence = 'abcdefg' # sequence = 'abcdefg'
# pattern = 'bde' # pattern = 'bde'
@ -227,6 +117,13 @@ else:
) )
] ]
def expectedOutcomes(self, search_result, expected_outcomes):
best_from_groups = [
get_best_match_in_group(group)
for group in group_matches(search_result)
]
return search_result == best_from_groups
def test_missing_second_item_complex(self): def test_missing_second_item_complex(self):
self.assertTrue( self.assertTrue(
set(self.search('bde', 'abcdefg', 1, 1, 1, 1)).issubset([ set(self.search('bde', 'abcdefg', 1, 1, 1, 1)).issubset([