def test_dl(self):
self.assertEqual(dam_lev('', ''), 0)
self.assertEqual(dam_lev('', 'a'), 1)
self.assertEqual(dam_lev('a', ''), 1)
self.assertEqual(dam_lev('a', 'b'), 1)
self.assertEqual(dam_lev('a', 'ab'), 1)
self.assertEqual(dam_lev('ab', 'ba'), 1)
self.assertEqual(dam_lev('ab', 'bca'), 2)
self.assertEqual(dam_lev('bca', 'ab'), 2)
self.assertEqual(dam_lev('ab', 'bdca'), 3)
self.assertEqual(dam_lev('bdca', 'ab'), 3)
def test_dl(self):
self.assertEqual(dam_lev('', ''), 0)
self.assertEqual(dam_lev('', 'a'), 1)
self.assertEqual(dam_lev('a', ''), 1)
self.assertEqual(dam_lev('a', 'b'), 1)
self.assertEqual(dam_lev('a', 'ab'), 1)
self.assertEqual(dam_lev('ab', 'ba'), 1)
self.assertEqual(dam_lev('ab', 'bca'), 2)
self.assertEqual(dam_lev('bca', 'ab'), 2)
self.assertEqual(dam_lev('ab', 'bdca'), 3)
self.assertEqual(dam_lev('bdca', 'ab'), 3)
def data_print(answer, recWord):
substitute_costs = np.ones((128, 128), dtype=np.float64)
for i in answer:
print("The word may be : {} the change in char is of : {} digits".
format(
i, int(dam_lev(recWord, i,
substitute_costs=substitute_costs))))
def weighted_ed_rel(self, cand, token, del_costs, ins_costs, sub_costs,
trans_costs):
w_editdist = dam_lev(token,
cand,
delete_costs=del_costs,
insert_costs=ins_costs,
substitute_costs=sub_costs,
transpose_costs=trans_costs)
rel_w_editdist = w_editdist / len(token)
return rel_w_editdist
def __score(self, wrong_word, candidate):
dl_dist = dam_lev(wrong_word,
candidate,
insert_costs=self.insert_costs,
substitute_costs=self.substitute_costs,
delete_costs=self.delete_costs,
transpose_costs=self.transpose_costs) / \
max(len(wrong_word), len(candidate))
log_prior = self.priors[candidate] if candidate in self.priors \
else math.log(float(self.k) / self.N)
return -dl_dist + self.lamda * log_prior
def edit_distance(edp: EditDistanceParams, a: str, b: str,
error: bool) -> np.float64:
"""Damerau-Levenshtein edit distance between two pixel strings."""
res = weighted_levenshtein.dam_lev(
a,
b,
insert_costs=edp.insert_costs,
delete_costs=edp.delete_costs,
substitute_costs=(edp.error_substitute_costs
if error else edp.substitute_costs),
)
# Make sure result can fit in a uint16
assert (0 <= res < 2**16), res
return res
count = 0
# read wiki misspell file line by line
# for line in f_mis:
while count < 4453:
string = f_mis_array[count].strip()
bestv = 10000000
bests = ""
smallest_list = []
# compare with dict lines
for num in range(len(string) - 2, len(string) + 3):
for word in f_len_dict_array[str(num)]:
# for entry in my_dict:
# string is lines in target doc
# entry is lines in dict
thisv = dam_lev(string, word.strip(), delete_costs=higher_costs_1)
# thisv = editdistance.eval(string, word.strip())
if (thisv > 0 and thisv < bestv):
bestv = thisv
bests = word.strip()
smallest_list = [bests]
elif (thisv == bestv):
smallest_list.append(word.strip())
print({
"result_s": smallest_list,
"distance": bestv,
"match": f_cor_array[count].strip() in smallest_list,
"original": f_cor_array[count].strip()
})
# RG3 to RG4
# RG3 to RG5
# RG4 to RG5
Root = ET.Element("xml")
f = open('spineData-ascii.txt')
f.readline() # read and ignore the first line
for line in f: # iterate over the remaining lines
v = line.split('\t')
app = v[0]
rg1_2 = v[1] + '::' + v[3]
if '#fMS' in rg1_2:
print('#fMS is here')
transpose_costs = np.full((128, 128), 0.25, dtype=np.float64)
dist1_2 = dam_lev(v[2], v[4], transpose_costs=transpose_costs)
else:
transpose_costs = np.ones((128, 128), dtype=np.float64)
dist1_2 = dam_lev(v[2], v[4])
rg1_3 = v[1] + '::' + v[5]
if '#fMS' in rg1_3:
print('#fMS is here')
transpose_costs = np.full((128, 128), 0.25, dtype=np.float64)
dist1_3 = dam_lev(v[2], v[6], transpose_costs=transpose_costs)
else:
transpose_costs = np.ones((128, 128), dtype=np.float64)
dist1_3 = dam_lev(v[2], v[6])
rg1_4 = v[1] + '::' + v[7]
if '#fMS' in rg1_4:
count = 0
# read wiki misspell file line by line
# for line in f_mis:
while count < 4453:
string = f_mis_array[count].strip()
bestv = 10000000
bests = ""
smallest_list = []
# compare with dict lines
for num in range(len(string) - 2, len(string) + 3):
for word in f_len_dict_array[str(num)]:
# for entry in my_dict:
# string is lines in target doc
# entry is lines in dict
thisv = dam_lev(string, word.strip(), insert_costs=higher_costs_1)
# thisv = editdistance.eval(string, word.strip())
if (thisv > 0 and thisv < bestv):
bestv = thisv
bests = word.strip()
smallest_list = [bests]
elif (thisv == bestv):
smallest_list.append(word.strip())
print({
"result_s": smallest_list,
"distance": bestv,
"match": f_cor_array[count].strip() in smallest_list,
"original": f_cor_array[count].strip()
})
def _dl(self, x, y):
return dam_lev(x, y, self.iw, self.dw, self.sw, self.tw)
str_build_orders,
open("data/build_orders/TvZ_build_orders_{}_{}.p".format(n, l), "wb"))
pickle.dump(mmrs, open("data/mmr/TvZ_mmr_{}.p".format(n), "wb"))
pickle.dump(player_ids,
open("data/player_ids/TvZ_player_ids_{}.p".format(n), "wb"))
i = 0
for y in range(n):
print(i, "/", n)
for x in range(n):
str_build_orders[x] = str_build_orders[x]
str_build_orders[y] = str_build_orders[y]
if y == x:
distance = 0
else:
distance = dam_lev(str1=str_build_orders[x],
str2=str_build_orders[y],
insert_costs=insert_costs,
delete_costs=delete_costs,
substitute_costs=substitute_costs,
transpose_costs=transpose_costs)
distance = int(distance)
D[y][x] = distance
D[x][y] = distance
#print(str_build_orders[x])
#print(str_build_orders[y])
#print(distance)
i += 1
print("Storing distance matrix")
pickle.dump(D, open("data/distance_matrix/TvZ_{}_{}.p".format(n, l), "wb"))
count = 2779
# read wiki misspell file line by line
# for line in f_mis:
while count < 4453:
string = f_mis_array[count].strip()
bestv = 10000000
bests = ""
smallest_list = []
# compare with dict lines
for num in range(len(string)-2, len(string)+3):
for word in f_len_dict_array[str(num)]:
# for entry in my_dict:
# string is lines in target doc
# entry is lines in dict
thisv = dam_lev(string, word.strip(), transpose_costs=higher_costs_2, insert_costs=higher_costs_1)
# thisv = editdistance.eval(string, word.strip())
if (thisv > 0 and thisv < bestv):
bestv = thisv
bests = word.strip()
smallest_list = [bests]
elif (thisv == bestv):
smallest_list.append(word.strip())
print({"result_s": smallest_list, "distance": bestv, "match": f_cor_array[count].strip() in smallest_list, "original": f_cor_array[count].strip()})
# save to res dict
res_dict[string] = {"result_s": smallest_list, "distance": bestv, "match": f_cor_array[count].strip() in smallest_list, "original": f_cor_array[count].strip()}
count += 1
print("count",count)
def _dl(self, x, y):
return dam_lev(x, y, self.iw, self.dw, self.sw, self.tw)