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
|
#!/usr/bin/python3
import os
import os.path
import sqlite3
from argparse import ArgumentParser
from operator import itemgetter
from math import log
from sys import float_info
import utils
from myconfig import MyConfig
from dirwalk import walkIndex
config = MyConfig()
#change cwd to the word recognizer directory
words_dir = config.getWordRecognizerDir()
os.chdir(words_dir)
#chdir done
############################################################
# Create Bigram Database #
############################################################
CREATE_BIGRAM_DDL = '''
CREATE TABLE bigram (
prefix TEXT NOT NULL,
postfix TEXT NOT NULL,
freq INTEGER NOT NULL
);
'''
CREATE_BIGRAM_PREFIX_INDEX_DDL = '''
CREATE INDEX bigram_prefix_index on bigram(prefix);
'''
CREATE_BIGRAM_POSTFIX_INDEX_DDL = '''
CREATE INDEX bigram_postfix_index on bigram(postfix);
'''
SELECT_ALL_NGRAM_DML = '''
SELECT words, freq FROM ngram;
'''
INSERT_BIGRAM_DML = '''
INSERT INTO bigram(prefix, postfix, freq) VALUES (?, ?, ?);
'''
def createBigramSqlite(workdir):
print(workdir, 'create bigram')
filename = config.getBigramFileName()
filepath = workdir + os.sep + filename
print(filepath)
if os.access(filepath, os.F_OK):
os.unlink(filepath)
conn = sqlite3.connect(filepath)
cur = conn.cursor()
cur.execute(CREATE_BIGRAM_DDL)
cur.execute(CREATE_BIGRAM_PREFIX_INDEX_DDL)
cur.execute(CREATE_BIGRAM_POSTFIX_INDEX_DDL)
conn.commit()
if conn:
conn.close()
def populateBigramSqlite(workdir):
print(workdir, 'populate bigram')
sep = config.getWordSep()
filename = config.getBigramFileName()
filepath = workdir + os.sep + filename
bigram_conn = sqlite3.connect(filepath)
bigram_cur = bigram_conn.cursor()
length = 2
filename = config.getNgramFileName(length)
filepath = workdir + os.sep + filename
ngram_conn = sqlite3.connect(filepath)
ngram_cur = ngram_conn.cursor()
#begin processing
rows = ngram_cur.execute(SELECT_ALL_NGRAM_DML).fetchall()
for row in rows:
(words_str, freq) = row
words = words_str.strip(sep).split(sep, 1)
assert len(words) == length
(prefix, postfix) = words
bigram_cur.execute(INSERT_BIGRAM_DML, (prefix, postfix, freq))
#print(prefix, postfix, freq)
bigram_conn.commit()
ngram_conn.commit()
if bigram_conn:
bigram_conn.close()
if ngram_conn:
ngram_conn.close()
############################################################
# Information Entropy Model #
############################################################
def computeEntropy(freqs):
#print(freqs)
totalfreq = sum(freqs)
freqs = [ freq / float(totalfreq) for freq in freqs ]
assert abs(1 - sum(freqs)) < len(freqs) * float_info.epsilon
entropy = - sum([ freq * log(freq) for freq in freqs ])
return entropy
############################################################
# Get Threshold Pass #
############################################################
SELECT_PREFIX_DML = '''
SELECT prefix, freq FROM bigram WHERE postfix = ? ;
'''
SELECT_POSTFIX_DML = '''
SELECT postfix, freq FROM bigram WHERE prefix = ? ;
'''
def computePrefixEntropy(cur, word):
rows = cur.execute(SELECT_PREFIX_DML, (word, )).fetchall()
if 0 == len(rows):
return 0.
freqs = []
for row in rows:
(prefix, freq) = row
assert freq >= 1
freqs.append(freq)
return computeEntropy(freqs)
def computePostfixEntropy(cur, word):
rows = cur.execute(SELECT_POSTFIX_DML, (word, )).fetchall()
if 0 == len(rows):
return 0.
freqs = []
for row in rows:
(postfix, freq) = row
assert freq >= 1
freqs.append(freq)
return computeEntropy(freqs)
def computeThreshold(cur, tag):
wordswithentropy = []
wordlistfile = open(config.getWordsListFileName(), "r")
for oneline in wordlistfile.readlines():
oneline = oneline.rstrip(os.linesep)
if len(oneline) == 0:
continue
word = oneline
entropy = 0.
if "prefix" == tag:
entropy = computePrefixEntropy(cur, word)
elif "postfix" == tag:
entropy = computePostfixEntropy(cur, word)
else:
raise "invalid tag value."
#print(word, entropy)
if entropy < config.getMinimumEntropy():
continue
wordswithentropy.append((word, entropy))
wordlistfile.close()
#ascending sort
wordswithentropy.sort(key=itemgetter(1))
pos = int(len(wordswithentropy) * config.getNewWordThreshold())
(word, threshold) = wordswithentropy[-pos]
print(word, threshold)
return threshold
############################################################
# Get Word Pass #
############################################################
|