def testEditDist(
self): # - - - - - - - - - - - - - - - - - - - - - - - - -
"""Test 'EditDist' approximate string comparator"""
for pair in self.string_pairs:
approx_str_value = stringcmp.editdist(pair[0], pair[1])
assert (isinstance(approx_str_value,float)), \
'"EditDist" does not return a floating point number for: '+ \
str(pair)
assert (approx_str_value >= 0.0), \
'"EditDist" returns a negative number for: '+str(pair)
assert (approx_str_value <= 1.0), \
'"EditDist" returns a number larger than 1.0 for: '+str(pair)
approx_str_value_1 = stringcmp.editdist(pair[0], pair[1])
approx_str_value_2 = stringcmp.editdist(pair[1], pair[0])
assert (approx_str_value_1 == approx_str_value_2), \
'"EditDist" returns different values for pair and swapped ' + \
'pair: '+str(pair)+': '+str(approx_str_value_1)+', '+ \
str(approx_str_value_2)
# Check for value 1.0 if the strings are the same
#
if (pair[0] == pair[1]):
assert (approx_str_value == 1.0), \
'"EditDist" does not return 1.0 if strings are equal: '+ \
str(pair)
def testEditDist(self): # - - - - - - - - - - - - - - - - - - - - - - - - -
"""Test 'EditDist' approximate string comparator"""
for pair in self.string_pairs:
approx_str_value = stringcmp.editdist(pair[0],pair[1])
assert (isinstance(approx_str_value,float)), \
'"EditDist" does not return a floating point number for: '+ \
str(pair)
assert (approx_str_value >= 0.0), \
'"EditDist" returns a negative number for: '+str(pair)
assert (approx_str_value <= 1.0), \
'"EditDist" returns a number larger than 1.0 for: '+str(pair)
approx_str_value_1 = stringcmp.editdist(pair[0],pair[1])
approx_str_value_2 = stringcmp.editdist(pair[1],pair[0])
assert (approx_str_value_1 == approx_str_value_2), \
'"EditDist" returns different values for pair and swapped ' + \
'pair: '+str(pair)+': '+str(approx_str_value_1)+', '+ \
str(approx_str_value_2)
# Check for value 1.0 if the strings are the same
#
if (pair[0] == pair[1]):
assert (approx_str_value == 1.0), \
'"EditDist" does not return 1.0 if strings are equal: '+ \
str(pair)
def testBagDist(self): # - - - - - - - - - - - - - - - - - - - - - - - - -
"""Test 'BagDist' approximate string comparator"""
for pair in self.string_pairs:
approx_str_value = stringcmp.bagdist(pair[0], pair[1])
assert isinstance(
approx_str_value, float
), '"BagDist" does not return a floating point number for: ' + str(
pair)
assert (approx_str_value >= 0.0
), '"BagDist" returns a negative number for: ' + str(pair)
assert (
approx_str_value <= 1.0
), '"BagDist" returns a number larger than 1.0 for: ' + str(pair)
approx_str_value_1 = stringcmp.bagdist(pair[0], pair[1])
approx_str_value_2 = stringcmp.bagdist(pair[1], pair[0])
assert approx_str_value_1 == approx_str_value_2, (
'"BagDist" returns different values for pair and swapped ' +
"pair: " + str(pair) + ": " + str(approx_str_value_1) + ", " +
str(approx_str_value_2))
# Check for value 1.0 if the strings are the same
#
if pair[0] == pair[1]:
assert (
approx_str_value == 1.0
), '"BagDist" does not return 1.0 if strings are equal: ' + str(
pair)
# Check if bad distance is always larger than edit distance
editdist_str_value = stringcmp.editdist(pair[0], pair[1])
assert (
approx_str_value >= editdist_str_value
), '"BagDist" value is smaller than "EditDist" value for: ' + str(
pair)
def distance(ovv,cand):
ovv = get_reduced(ovv)
return editdist(ovv,cand)
##------posqgram 3------##
start_time7 = time.time()
s += ' %.3f' % (stringcmp.posqgram(twoNames[0], twoNames[1], 3))
time_used7 = time.time() - start_time7
s += ' %.10f' % (time_used7)
##------sgram------##
start_time8 = time.time()
s += ' %.3f' % (stringcmp.sgram(twoNames[0], twoNames[1],
[[0], [0, 1], [1, 2]]))
time_used8 = time.time() - start_time8
s += ' %.10f' % (time_used8)
##------editdist------##
start_time9 = time.time()
s += ' %.3f' % (stringcmp.editdist(twoNames[0], twoNames[1]))
time_used9 = time.time() - start_time9
s += ' %.10f' % (time_used9)
##------mod_editdist------##
start_time10 = time.time()
s += ' %.3f' % (stringcmp.mod_editdist(twoNames[0], twoNames[1]))
time_used10 = time.time() - start_time10
s += ' %.10f' % (time_used10)
##------bagdist------##
start_time11 = time.time()
s += ' %.3f' % (stringcmp.bagdist(twoNames[0], twoNames[1]))
time_used11 = time.time() - start_time11
s += ' %.10f' % (time_used11)