-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathspellsuggest.py
261 lines (186 loc) · 9.11 KB
/
spellsuggest.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
# -*- coding: utf-8 -*-
import re
import os
from trie import Trie
from levenshtein_damerau_threshold import dp_levenshtein_backwards, dp_restricted_damerau_backwards, dp_intermediate_damerau_backwards
import numpy as np
import json
class SpellSuggester:
"""
Clase que implementa el método suggest para la búsqueda de términos.
"""
def __init__(self, vocab_file_path):
"""Método constructor de la clase SpellSuggester
Construye una lista de términos únicos (vocabulario),
que además se utiliza para crear un trie.
Args:
vocab_file (str): ruta del fichero de texto para cargar el vocabulario.
"""
self.vocabulary = self.build_vocab(vocab_file_path, tokenizer=re.compile("\W+"))
def build_vocab(self, vocab_file_path, tokenizer):
"""Método para crear el vocabulario.
Se tokeniza por palabras el fichero de texto,
se eliminan palabras duplicadas y se ordena
lexicográficamente.
Args:
vocab_file (str): ruta del fichero de texto para cargar el vocabulario.
tokenizer (re.Pattern): expresión regular para la tokenización.
"""
if os.path.isdir(vocab_file_path):
vocab = set()
for dir, subdirs, files in os.walk(vocab_file_path):
for filename in files:
if filename.endswith('.json'):
fullname = os.path.join(dir, filename)
with open(fullname) as fh:
news_list = json.load(fh)
for new in news_list:
words = set(tokenizer.split(new['article'].lower()))
for word in words:
vocab.add(word)
vocab.discard('') # por si acaso
return sorted(vocab)
else:
with open(vocab_file_path, "r", encoding='utf-8') as fr:
vocab = set(tokenizer.split(fr.read().lower()))
vocab.discard('') # por si acaso
return sorted(vocab)
def suggest(self, term, distance="levenshtein", threshold=None):
"""Método para sugerir palabras similares siguiendo la tarea 3.
A completar.
Args:
term (str): término de búsqueda.
distance (str): algoritmo de búsqueda a utilizar
{"levenshtein", "restricted", "intermediate"}.
threshold (int): threshold para limitar la búsqueda
puede utilizarse con los algoritmos de distancia mejorada de la tarea 2
o filtrando la salida de las distancias de la tarea 2
"""
assert distance in ["levenshtein", "restricted", "intermediate"]
results = {} # diccionario termino:distancia
for word in self.vocabulary:
if(distance == "levenshtein"):
if(threshold != None):
dist = dp_levenshtein_backwards(word, term, threshold)
if(dist <= threshold):
results[word] = dist
else:
dist = dp_levenshtein_backwards(word, term)
results[word] = dist
elif(distance == "restricted"):
if(threshold != None):
dist = dp_restricted_damerau_backwards(word, term, threshold)
if(dist <= threshold):
results[word] = dist
else:
dist = dp_restricted_damerau_backwards(word, term)
results[word] = dist
elif(distance == "intermediate"):
if(threshold != None):
dist = dp_intermediate_damerau_backwards(word, term, threshold)
if(dist <= threshold):
results[word] = dist
else:
dist = dp_intermediate_damerau_backwards(word, term)
results[word] = dist
return results
class TrieSpellSuggester(SpellSuggester):
"""
Clase que implementa el método suggest para la búsqueda de términos y añade el trie
"""
def __init__(self, vocab_file_path):
super().__init__(vocab_file_path)
self.trie = Trie(self.vocabulary)
def dp_levenshtein_backwards_trie(self, x, threshold = 4):
D = np.zeros((self.trie.get_num_states() + 1, len(x) + 1))
for i in range(1, self.trie.get_num_states() + 1):
D[i, 0] = D[self.trie.get_parent(i), 0] + 1
for j in range(1, len(x) + 1):
D[0, j] = D[0, j - 1] + 1
for i in range(1, self.trie.get_num_states() + 1):
D[i, j] = min(
D[self.trie.get_parent(i), j] + 1,
D[i, j - 1] + 1,
D[self.trie.get_parent(i), j - 1] + (self.trie.get_label(i) != x[j-1])
)
if(min(D[:, j]) > threshold):
return threshold + 1
return D[:, len(x)]
def dp_restricted_damerau_backwards_trie(self, x, threshold = 4):
D = np.zeros((self.trie.get_num_states() + 1, len(x) + 1))
for i in range(1, self.trie.get_num_states() + 1):
D[i, 0] = D[self.trie.get_parent(i), 0] + 1
for j in range(1, len(x) + 1):
D[0, j] = D[0, j - 1] + 1
for i in range(1, self.trie.get_num_states() + 1):
D[i, j] = min(
D[self.trie.get_parent(i), j] + 1,
D[i, j - 1] + 1,
D[self.trie.get_parent(i), j - 1] + (self.trie.get_label(i) != x[j-1])
)
if i > 1 and j > 1 and x[j - 2] == self.trie.get_label(i) and x[j - 1] == self.trie.get_label(self.trie.get_parent(i)):
D[i, j] = min(
D[i, j],
D[self.trie.get_parent(self.trie.get_parent(i)), j - 2] + 1
)
if(min(D[:, j]) > threshold):
return threshold + 1
return D[:, len(x)]
def dp_intermediate_damerau_backwards_trie(self, x, threshold = 4):
D = np.zeros((self.trie.get_num_states() + 1, len(x) + 1))
for i in range(1, self.trie.get_num_states() + 1):
D[i, 0] = D[self.trie.get_parent(i), 0] + 1
for j in range(1, len(x) + 1):
D[0, j] = D[0, j - 1] + 1
for i in range(1, self.trie.get_num_states() + 1):
D[i, j] = min(
D[self.trie.get_parent(i), j] + 1,
D[i, j - 1] + 1,
D[self.trie.get_parent(i), j - 1] + (self.trie.get_label(i) != x[j-1])
)
if i > 1 and j > 1 and self.trie.get_label(self.trie.get_parent(i)) == x[j - 1] and self.trie.get_label(i) == x[j-2]:
D[i, j] = min(
D[i, j],
D[self.trie.get_parent(self.trie.get_parent(i)), j - 2] + 1
)
if(i > 2 and j > 1 and self.trie.get_label(self.trie.get_parent(self.trie.get_parent(i))) == x[j - 1] and self.trie.get_label(i) == x[j - 2]):
D[i, j] = min(
D[i, j],
D[self.trie.get_parent(self.trie.get_parent(i)), j - 2] + 1
)
if(i > 1 and j > 2 and self.trie.get_label(i) == x[j - 3] and self.trie.get_label(self.trie.get_parent(i)) == x[j - 1]):
D[i, j] = min(
D[i, j],
D[self.trie.get_parent(self.trie.get_parent(i)), j - 2] + 1
)
if(min(D[:, j]) > threshold):
return threshold + 1
return D[:, len(x)]
def suggest(self, term, distance="levenshtein", threshold=None):
assert distance in ["levenshtein", "restricted", "intermediate"]
if distance == "levenshtein":
if (threshold != None):
distances = self.dp_levenshtein_backwards_trie(term, threshold)
else:
distances = self.dp_levenshtein_backwards_trie(term)
elif distance == "restricted":
if (threshold != None):
distances = self.dp_restricted_damerau_backwards_trie(term, threshold)
else:
distances = self.dp_restricted_damerau_backwards_trie(term)
elif distance == "intermediate":
if (threshold != None):
distances = self.dp_intermediate_damerau_backwards_trie(term, threshold)
else:
distances = self.dp_intermediate_damerau_backwards_trie(term)
results = {} # diccionario termino:distancia
if type(distances) is np.ndarray:
for i in range(0, self.trie.get_num_states() + 1):
if(self.trie.is_final(i) and distances[i] <= threshold ):
word = self.trie.get_output(i)
results[word] = distances[i]
return results
if __name__ == "__main__":
spellsuggester = TrieSpellSuggester("./corpora/quijote.txt")
print(spellsuggester.suggest("alábese"))
# cuidado, la salida es enorme print(suggester.trie)