def all_metrics(string1, string2, longer_prob):
pre_scale = 0.1
typo_scale = 10
ans = string_metrics(string1, string2,
boost_threshold=0.7, pre_len=4,
typo_table=adjwt, typo_scale=typo_scale,
pre_scale=pre_scale, longer_prob=longer_prob)
(len1, len2, num_matches, half_transposes,
typo_score, pre_matches, adjust_long) = ans
assert not (longer_prob or adjust_long) or longer_prob
weight_jaro = fn_jaro(len1, len2, num_matches, half_transposes, 0, 1)
weight_typo = fn_jaro(len1, len2, num_matches, half_transposes,
typo_score, typo_scale)
weight_winkler = fn_winkler(weight_jaro, pre_matches, pre_scale)
weight_winkler_typo = fn_winkler(weight_typo, pre_matches, pre_scale)
weight_longer = weight_winkler_typo
if adjust_long:
weight_longer = fn_longer(weight_longer, len1, len2,
num_matches, pre_matches)
return (len1, len2, num_matches, half_transposes,
typo_score, pre_matches, adjust_long,
weight_jaro, weight_typo, weight_winkler,
weight_winkler_typo, weight_longer)
def all_metrics(string1, string2, longer_prob):
pre_scale = 0.1
typo_scale = 10
ans = string_metrics(string1,
string2,
boost_threshold=0.7,
pre_len=4,
typo_table=adjwt,
typo_scale=typo_scale,
pre_scale=pre_scale,
longer_prob=longer_prob)
(len1, len2, num_matches, half_transposes, typo_score, pre_matches,
adjust_long) = ans
assert not (longer_prob or adjust_long) or longer_prob
weight_jaro = fn_jaro(len1, len2, num_matches, half_transposes, 0, 1)
weight_typo = fn_jaro(len1, len2, num_matches, half_transposes, typo_score,
typo_scale)
weight_winkler = fn_winkler(weight_jaro, pre_matches, pre_scale)
weight_winkler_typo = fn_winkler(weight_typo, pre_matches, pre_scale)
weight_longer = weight_winkler_typo
if adjust_long:
weight_longer = fn_longer(weight_longer, len1, len2, num_matches,
pre_matches)
return (len1, len2, num_matches, half_transposes, typo_score, pre_matches,
adjust_long, weight_jaro, weight_typo, weight_winkler,
weight_winkler_typo, weight_longer)
def test():
import jaro
for test in jaro_tests:
# s1, s2, m, t, jaro, wink = test
s1, s2 = test[:2]
string_metrics = jaro.string_metrics(s1, s2)
(len1, len2, num_matches, half_transposes,
typo_score, pre_matches, adjust_long) = string_metrics
weight_jaro = jaro.metric_jaro(s1, s2)
weight_winkler = jaro.metric_jaro_winkler(s1, s2)
weight_original = jaro.metric_original(s1, s2)
# TODO: Test for the custom function?
weights = [weight_jaro, weight_winkler, weight_original]
check = [num_matches, half_transposes]
check.extend(['%7.5f' % w for w in weights])
if check != list(test[2:]):
print
print s1, s2
print check
print test[2:]
raise AssertionError
strings = []
for s in [s1, s2]:
if s.strip() == '':
s = '-'*(len(s)+1)
strings.append(s.ljust(12))
for n in [num_matches, half_transposes]:
strings.append(str(n).rjust(2))
for w in weights:
strings.append(' %7.5f' % w)
print ' '.join(strings)
def test():
import jaro
for test in jaro_tests:
# s1, s2, m, t, jaro, wink = test
s1, s2 = test[:2]
string_metrics = jaro.string_metrics(s1, s2)
(len1, len2, num_matches, half_transposes, typo_score, pre_matches,
adjust_long) = string_metrics
weight_jaro = jaro.metric_jaro(s1, s2)
weight_winkler = jaro.metric_jaro_winkler(s1, s2)
weight_original = jaro.metric_original(s1, s2)
# TODO: Test for the custom function?
weights = [weight_jaro, weight_winkler, weight_original]
check = [num_matches, half_transposes]
check.extend(['%7.5f' % w for w in weights])
if check != list(test[2:]):
print
print s1, s2
print check
print test[2:]
raise AssertionError
strings = []
for s in [s1, s2]:
if s.strip() == '':
s = '-' * (len(s) + 1)
strings.append(s.ljust(12))
for n in [num_matches, half_transposes]:
strings.append(str(n).rjust(2))
for w in weights:
strings.append(' %7.5f' % w)
print ' '.join(strings)