buildfsa.py
8.28 KB
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
# -*- coding:utf-8 -*-
'''
Created on 21 paź 2013
@author: mlenart
'''
import sys
import logging
import codecs
from morfeuszbuilder.fsa import encode
from morfeuszbuilder.fsa import convertinput
from morfeuszbuilder.fsa.fsa import FSA
from morfeuszbuilder.fsa.serializer import VLengthSerializer1, VLengthSerializer2, SimpleSerializer
from morfeuszbuilder.tagset.tagset import Tagset
from morfeuszbuilder.segrules import rulesParser
from optparse import OptionParser
# class InputFormat():
# ENCODED = 'ENCODED'
# POLIMORF = 'POLIMORF'
# PLAIN = 'PLAIN'
class SerializationMethod():
SIMPLE = 'SIMPLE'
V1 = 'V1'
V2 = 'V2'
def _checkOption(opt, parser, msg):
if opt is None:
print >> sys.stderr, msg
parser.print_help()
exit(1)
def _checkExactlyOneOptionSet(optsList, parser, msg):
if optsList.count(True) != 1:
print >> sys.stderr, msg
parser.print_help()
exit(1)
def _parseOptions():
"""
Parses commandline args
"""
parser = OptionParser()
parser.add_option('-i', '--input-file',
dest='inputFile',
metavar='FILE',
help='path to input file')
parser.add_option('--tagset-file',
dest='tagsetFile',
metavar='FILE',
help='path to the file with tagset')
parser.add_option('--segments-file',
dest='segmentsFile',
metavar='FILE',
help='path to the file with segment rules')
parser.add_option('-o', '--output-file',
dest='outputFile',
metavar='FILE',
help='path to output file')
parser.add_option('-a', '--analyzer',
dest='analyzer',
action='store_true',
default=False,
help='Generate FSA for morphological analysis')
parser.add_option('-g', '--generator',
dest='generator',
action='store_true',
default=False,
help='Generate FSA for morphological synthesis')
parser.add_option('--cpp',
dest='cpp',
action='store_true',
default=False,
help='Encode binary data in c++ file')
parser.add_option('--use-arrays',
dest='useArrays',
action='store_true',
default=False,
help='store states reachable by 2 transitions in arrays (should speed up recognition, available only when --serialization-method=V1)')
parser.add_option('--serialization-method',
dest='serializationMethod',
help="FSA serialization method: \
SIMPLE - fixed-length transitions, fastest and weakest compression \
V1 - variable-length transitions, compressed labels - strongest compression \
V2 - format similar to the default in Jan Daciuk's fsa package - variable-length transitions, non-compressed labels - good compression, good speed")
#~ parser.add_option('--visualize',
#~ dest='visualize',
#~ action='store_true',
#~ default=False,
#~ help='visualize result')
parser.add_option('--train-file',
dest='trainFile',
help='A text file used for training. Should contain words from some large corpus - one word in each line')
parser.add_option('--debug',
dest='debug',
action='store_true',
default=False,
help='output some debugging info')
#~ parser.add_option('--profile',
#~ dest='profile',
#~ action='store_true',
#~ default=False,
#~ help='show profiling graph (required pycallgraph and graphviz')
opts, args = parser.parse_args()
_checkOption(opts.inputFile, parser, "Input file is missing")
_checkOption(opts.outputFile, parser, "Output file is missing")
_checkOption(opts.tagsetFile, parser, "Tagset file is missing")
_checkOption(opts.serializationMethod, parser, "Serialization method file is missing")
_checkExactlyOneOptionSet([opts.analyzer, opts.generator],
parser, 'Must set exactly one FSA type: --analyzer or --generator')
if opts.analyzer:
_checkOption(opts.segmentsFile, parser, "Segment rules file is missing")
if not opts.serializationMethod.upper() in [SerializationMethod.SIMPLE, SerializationMethod.V1, SerializationMethod.V2]:
print >> sys.stderr, '--serialization-method must be one of ('+str([SerializationMethod.SIMPLE, SerializationMethod.V1, SerializationMethod.V2])+')'
parser.print_help()
exit(1)
return opts
def _readPolimorfInput4Analyzer(inputFile, tagset, encoder):
with open(inputFile, 'r') as f:
for entry in convertinput.PolimorfConverter4Analyzer(tagset, encoder, 'utf8').convert(f):
yield entry
def _readPolimorfInput4Generator(inputFile, tagset, encoder):
with open(inputFile, 'r') as f:
for entry in convertinput.PolimorfConverter4Generator(tagset, encoder, 'utf8').convert(f):
yield entry
def _readTrainData(trainFile):
with codecs.open(trainFile, 'r', 'utf8') as f:
for line in f:
yield line.strip()
def _printStats(fsa):
acceptingNum = 0
sinkNum = 0
arrayNum = 0
for s in fsa.dfs():
if s.isAccepting():
acceptingNum += 1
if s.transitionsNum == 0:
sinkNum += 1
if s.serializeAsArray:
arrayNum += 1
logging.info('states num: '+str(fsa.getStatesNum()))
logging.info('transitions num: '+str(fsa.getTransitionsNum()))
logging.info('accepting states num: '+str(acceptingNum))
logging.info('sink states num: '+str(sinkNum))
logging.info('array states num: '+str(arrayNum))
def buildAnalyzerFromPoliMorf(inputFile, tagset):
encoder = encode.MorphEncoder()
fsa = FSA(encoder, tagset)
inputData = _readPolimorfInput4Analyzer(inputFile, tagset, encoder)
for word, data in inputData:
fsa.addEntry(word, data)
fsa.close()
_printStats(fsa)
return fsa
def buildGeneratorFromPoliMorf(inputFile, tagsetFile):
encoder = encode.Encoder4Generator()
tagset = Tagset(tagsetFile)
fsa = FSA(encoder, tagset)
inputData = _readPolimorfInput4Generator(inputFile, tagset, encoder)
for word, data in inputData:
fsa.addEntry(word, data)
fsa.close()
_printStats(fsa)
return fsa
def main(opts):
if opts.debug:
logging.basicConfig(level=logging.DEBUG)
else:
logging.basicConfig(level=logging.INFO)
tagset = Tagset(opts.tagsetFile)
if opts.analyzer:
fsa = buildAnalyzerFromPoliMorf(opts.inputFile, tagset)
segmentRulesManager = rulesParser.RulesParser(tagset).parse(opts.segmentsFile)
additionalData = segmentRulesManager.serialize()
else:
fsa = buildGeneratorFromPoliMorf(opts.inputFile, tagset)
additionalData = bytearray()
if opts.trainFile:
logging.info('training with '+opts.trainFile+' ...')
fsa.train(_readTrainData(opts.trainFile))
logging.info('done training')
serializer = {
SerializationMethod.SIMPLE: SimpleSerializer,
SerializationMethod.V1: VLengthSerializer1,
SerializationMethod.V2: VLengthSerializer2,
}[opts.serializationMethod](fsa)
if opts.cpp:
serializer.serialize2CppFile(opts.outputFile, generator=opts.generator)
else:
serializer.serialize2BinaryFile(opts.outputFile)
# {
# OutputFormat.CPP: serializer.serialize2CppFile,
# OutputFormat.BINARY: serializer.serialize2BinaryFile
# }[opts.outputFormat](opts.outputFile)
logging.info('size: '+str(fsa.initialState.reverseOffset))
if __name__ == '__main__':
import os
print os.getcwd()
opts = _parseOptions()
main(opts)