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
|
#!/usr/bin/python3
import os
import os.path
import sys
from subprocess import Popen, PIPE
from argparse import ArgumentParser
from operator import itemgetter
import utils
from myconfig import MyConfig
config = MyConfig()
#change cwd to the libpinyin data directory
libpinyin_dir = config.getToolsDir()
libpinyin_sub_dir = os.path.join(libpinyin_dir, 'data')
os.chdir(libpinyin_sub_dir)
#chdir done
def handleError(error):
sys.exit(error)
def handleOneModel(modelfile, reportfile):
modelfilestatuspath = modelfile + config.getStatusPostfix()
modelfilestatus = utils.load_status(modelfilestatuspath)
if not utils.check_epoch(modelfilestatus, 'Generate'):
raise utils.EpochError('Please generate first.\n')
if utils.check_epoch(modelfilestatus, 'Estimate'):
return
reporthandle = open(reportfile, 'wb')
result_line_prefix = "average lambda:"
avg_lambda = 0.
#begin processing
cmdline = ['../utils/training/estimate_k_mixture_model', \
'--deleted-bigram-file', \
config.getEstimatesModel(), \
'--bigram-file', \
modelfile]
subprocess = Popen(cmdline, shell=False, stdout=PIPE, \
close_fds=True)
for line in subprocess.stdout.readlines():
reporthandle.writelines([line])
#remove trailing '\n'
line = line.decode('utf-8')
line = line.rstrip(os.linesep)
if line.startswith(result_line_prefix):
avg_lambda = float(line[len(result_line_prefix):])
reporthandle.close()
(pid, status) = os.waitpid(subprocess.pid, 0)
if status != 0:
sys.exit('estimate k mixture model returns error.')
#end processing
print('average lambda:', avg_lambda)
modelfilestatus['EstimateScore'] = avg_lambda
utils.sign_epoch(modelfilestatus, 'Estimate')
utils.store_status(modelfilestatuspath, modelfilestatus)
def walkThroughModels(path):
for root, dirs, files in os.walk(path, topdown=True, onerror=handleError):
for onefile in files:
filepath = os.path.join(root, onefile)
if onefile.endswith(config.getModelPostfix()):
subpath = os.path.relpath(filepath, path)
reportfile = filepath + config.getReportPostfix()
print("Processing " + subpath)
handleOneModel(filepath, reportfile)
print("Processed " + subpath)
elif onefile.endswith(config.getStatusPostfix()):
pass
elif onefile.endswith(config.getIndexPostfix()):
pass
elif onefile.endswith(config.getReportPostfix()):
pass
else:
print('Unexpected file:' + filepath)
def gatherModels(path, indexname):
indexfilestatuspath = indexname + config.getStatusPostfix()
indexfilestatus = utils.load_status(indexfilestatuspath)
if utils.check_epoch(indexfilestatuspath, 'Estimate'):
return
#begin processing
indexfile = open(indexname, "w")
for root, dirs, files in os.walk(path, topdown=True, onerror=handleError):
for onefile in files:
filepath = os.path.join(root, onefile)
if onefile.endswith(config.getModelPostfix()):
#append one record to index file
subdir = os.path.relpath(root, path)
statusfilepath = filepath + config.getStatusPostfix()
status = utils.load_status(statusfilepath)
if not (utils.check_epoch(status, 'Estimate') and \
'EstimateScore' in status):
raise utils.EpochError('Unknown Error:\n' + \
'Try re-run estimate.\n')
avg_lambda = status['EstimateScore']
line = subdir + '#' + onefile + '#' + str(avg_lambda)
indexfile.writelines([line, os.linesep])
#record written
elif onefile.endswith(config.getStatusPostfix()):
pass
elif onefile.endswith(config.getIndexPostfix()):
pass
elif onefile.endswith(config.getReportPostfix()):
pass
else:
print('Unexpected file:' + filepath)
indexfile.close()
#end processing
utils.sign_epoch(indexfilestatus, 'Estimate')
utils.store_status(indexfilestatuspath, indexfilestatus)
def sortModels(indexname, sortedindexname):
sortedindexfilestatuspath = sortedindexname + config.getStatusPostfix()
sortedindexfilestatus = utils.load_status(sortedindexfilestatuspath)
if utils.check_epoch(sortedindexfilestatus, 'Estimate'):
return
#begin processing
records = []
indexfile = open(indexname, 'r')
for line in indexfile.readlines():
#remove the trailing '\n'
line = line.rstrip(os.linesep)
(subdir, modelname, score) = line.split('#', 2)
score = float(score)
records.append((subdir, modelname, score))
indexfile.close()
records.sort(key=itemgetter(2), reverse=True)
sortedindexfile = open(sortedindexname, 'w')
for record in records:
(subdir, modelname, score) = record
line = subdir + '#' + modelname + '#' + str(score)
sortedindexfile.writelines([line, os.linesep])
sortedindexfile.close()
#end processing
utils.sign_epoch(sortedindexfilestatus, 'Estimate')
utils.store_status(sortedindexfilestatuspath, sortedindexfilestatus)
if __name__ == '__main__':
parser = ArgumentParser(description='Estimate model candidates.')
parser.add_argument('--modeldir', action='store', \
help='model directory', \
default=config.getModelDir())
args = parser.parse_args()
print(args)
print("estimating")
walkThroughModels(args.modeldir)
print("gathering")
indexname = os.path.join(args.modeldir, config.getEstimateIndex())
gatherModels(args.modeldir, indexname)
print("sorting")
sortedindexname = os.path.join(args.modeldir, \
config.getSortedEstimateIndex())
sortModels(indexname, sortedindexname)
print("done")
|