irstlm 6.00.05-2 / usr / lib / irstlm / bin / build-lm.sh

This file is indexed.

/usr/lib/irstlm/bin/build-lm.sh is in irstlm 6.00.05-2.

This file is owned by root:root, with mode 0o755.

The actual contents of the file can be viewed below. 

  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
#! /bin/bash

set -m # Enable Job Control

function usage()
{
    cmnd=$(basename $0);
    cat<<EOF

$cmnd - estimates a language model file and saves it in intermediate ARPA format

USAGE:
       $cmnd [options]

OPTIONS:
       -i|--InputFile          Input training file e.g. 'gunzip -c train.gz'
       -o|--OutputFile         Output gzipped LM, e.g. lm.gz
       -k|--Parts              Number of splits (default 5)
       -n|--NgramSize          Order of language model (default 3)
       -d|--Dictionary         Define subdictionary for n-grams (optional, default is without any subdictionary)
       -s|--LanguageModelType  Smoothing methods: witten-bell (default), shift-beta, improved-shift-beta, stupid-backoff; kneser-ney and improved-kneser-ney still accepted for back-compatibility, but mapped into shift-beta and improved-shift-beta, respectively
       -p|--PruneSingletons    Prune singleton n-grams (default false)
       -f|--PruneFrequencyThreshold      Pruning frequency threshold for each level; comma-separated list of values; (default is '0,0,...,0', for all levels)
       -t|--TmpDir             Directory for temporary files (default ./stat_PID)
       -l|--LogFile            File to store logging info (default /dev/null)
       -u|--uniform            Use uniform word frequency for dictionary splitting (default false)
       -b|--boundaries         Include sentence boundary n-grams (optional, default false)
       -v|--verbose            Verbose
       -h|-?|--help            Show this message

EOF
}

if [ ! $IRSTLM ]; then
   echo "Set IRSTLM environment variable with path to irstlm"
   exit 2
fi

#paths to scripts and commands in irstlm
scr=$IRSTLM/bin
bin=$IRSTLM/bin
gzip=`which gzip 2> /dev/null`;
gunzip=`which gunzip 2> /dev/null`;

#check irstlm installation
if [ ! -e $bin/dict -o  ! -e $scr/split-dict.pl ]; then
   echo "$IRSTLM does not contain a proper installation of IRSTLM"
   exit 3
fi

#default parameters
logfile=/dev/null
tmpdir=stat_$$
order=3
parts=3
inpfile="";
outfile=""
verbose="";
smoothing="witten-bell";
prune="";
prune_thr_str="";
boundaries="";
dictionary="";
uniform="-f=y";
backoff=""

while [ "$1" != "" ]; do
    case $1 in
        -i | --InputFile )          shift;
																inpfile=$1;
																;;
        -o | --OutputFile )         shift;
																outfile=$1;
                                ;;
        -n | --NgramSize )           shift;
																order=$1;
                                ;;
        -k | --Parts )          shift;
																parts=$1;
                                ;;
        -d | --Dictionary )     shift;
                                dictionary="-sd=$1";
                                ;;
        -s | --LanguageModelType )        shift;
																				  smoothing=$1;
                                          ;;
        -f | --PruneFrequencyThreshold )  shift;
																          prune_thr_str="--PruneFrequencyThreshold=$1";
                                          ;;
        -p | --PruneSingletons )     prune='--prune-singletons';
																			;;
        -l | --LogFile )        shift;
																logfile=$1;
                                ;;
        -t | --TmpDir )         shift;
																tmpdir=$1;
                                ;;
        -u | --uniform )        uniform=' ';
                                ;;
        -b | --boundaries )     boundaries='--cross-sentence';
																;;
        -v | --verbose )        verbose='--verbose';
                                ;;
        -h | -? | --help )      usage;
                                exit 0;
                                ;;
        * )                     usage;
                                exit 1;
    esac
    shift
done

case $smoothing in
witten-bell) 
smoothing="--witten-bell";
;; 
kneser-ney)
## kneser-ney still accepted for back-compatibility, but mapped into shift-beta
smoothing="--shift-beta";
;;
improved-kneser-ney)
## improved-kneser-ney still accepted for back-compatibility, but mapped into improved-shift-beta
smoothing="--improved-shift-beta"; 
;;
shift-beta)
smoothing="--shift-beta";
;;
improved-shift-beta)
smoothing="--improved-shift-beta";
;;
stupid-backoff)
smoothing="--stupid-backoff";
backoff="--backoff"
;;
*) 
echo "wrong smoothing setting; '$smoothing' does not exist";
exit 4
esac
			

echo "LOGFILE:$logfile"
			 

if [ $verbose ] ; then
echo inpfile='"'$inpfile'"' outfile=$outfile order=$order parts=$parts tmpdir=$tmpdir prune=$prune smoothing=$smoothing dictionary=$dictionary verbose=$verbose prune_thr_str=$prune_thr_str  >> $logfile 2>&1
fi

if [ ! "$inpfile" -o ! "$outfile" ] ; then
    usage
    exit 5
fi
 
if [ -e $outfile ]; then
   echo "Output file $outfile already exists! either remove or rename it."
   exit 6
fi

if [ -e $logfile -a $logfile != "/dev/null" -a $logfile != "/dev/stdout" ]; then
   echo "Logfile $logfile already exists! either remove or rename it."
   exit 7
fi

echo "BIS LOGFILE:$logfile" >> $logfile 2>&1

#check tmpdir
tmpdir_created=0;
if [ ! -d $tmpdir ]; then
   echo "Temporary directory $tmpdir does not exist"  >> $logfile 2>&1
   echo "creating $tmpdir"  >> $logfile 2>&1
   mkdir -p $tmpdir
   tmpdir_created=1
else
   echo "Cleaning temporary directory $tmpdir" >> $logfile 2>&1
    rm $tmpdir/* 2> /dev/null
    if [ $? != 0 ]; then
        echo "Warning: some temporary files could not be removed" >> $logfile 2>&1
    fi
fi


echo "Extracting dictionary from training corpus" >> $logfile 2>&1
$bin/dict -i="$inpfile" -o=$tmpdir/dictionary $uniform -sort=no 2> $logfile

echo "Splitting dictionary into $parts lists" >> $logfile 2>&1
$scr/split-dict.pl --input $tmpdir/dictionary --output $tmpdir/dict. --parts $parts >> $logfile 2>&1

echo "Extracting n-gram statistics for each word list" >> $logfile 2>&1
echo "Important: dictionary must be ordered according to order of appearance of words in data" >> $logfile 2>&1
echo "used to generate n-gram blocks,  so that sub language model blocks results ordered too" >> $logfile 2>&1

for sdict in $tmpdir/dict.*;do
sdict=`basename $sdict`
echo "Extracting n-gram statistics for $sdict" >> $logfile 2>&1
if [ $smoothing = "--shift-beta" -o $smoothing = "--improved-shift-beta" ]; then
additional_parameters="-iknstat=$tmpdir/ikn.stat.$sdict"
else
additional_parameters=""
fi

$bin/ngt -i="$inpfile" -n=$order -gooout=y -o="$gzip -c > $tmpdir/ngram.${sdict}.gz" -fd="$tmpdir/$sdict" $dictionary $additional_parameters >> $logfile 2>&1 &

#$bin/ngt -i="$inpfile" -n=$order -gooout=y -o="$gzip -c > $tmpdir/ngram.${sdict}.gz" -fd="$tmpdir/$sdict" $dictionary -iknstat="$tmpdir/ikn.stat.$sdict" >> $logfile 2>&1 &
#else
#$bin/ngt -i="$inpfile" -n=$order -gooout=y -o="$gzip -c > $tmpdir/ngram.${sdict}.gz" -fd="$tmpdir/$sdict" $dictionary >> $logfile 2>&1 &
#fi
done

# Wait for all parallel jobs to finish
while [ 1 ]; do fg 2> /dev/null; [ $? == 1 ] && break; done

echo "Estimating language models for each word list" >> $logfile 2>&1
for sdict in `ls $tmpdir/dict.*` ; do
sdict=`basename $sdict`
echo "Estimating language models for $sdict" >> $logfile 2>&1

if [ $smoothing = "--shift-beta" -o $smoothing = "--improved-shift-beta" ]; then
additional_smoothing_parameters="cat $tmpdir/ikn.stat.dict.*"
additional_parameters="$backoff"
else
additional_smoothing_parameters=""
additional_parameters=""
fi
$scr/build-sublm.pl $verbose $prune $prune_thr_str $smoothing "$additional_smoothing_parameters" --size $order --ngrams "$gunzip -c $tmpdir/ngram.${sdict}.gz" -sublm $tmpdir/lm.$sdict $additional_parameters >> $logfile 2>&1 &

#if [ $smoothing = "--shift-beta" -o $smoothing = "--improved-shift-beta" ]; then
#$scr/build-sublm.pl $verbose $prune $prune_thr_str $smoothing "cat $tmpdir/ikn.stat.dict.*" --size $order --ngrams "$gunzip -c $tmpdir/ngram.${sdict}.gz" -sublm $tmpdir/lm.$sdict $backoff >> $logfile 2>&1 &
#else
#$scr/build-sublm.pl $verbose $prune $prune_thr_str $smoothing --size $order --ngrams "$gunzip -c $tmpdir/ngram.${sdict}.gz" -sublm $tmpdir/lm.$sdict >> $logfile 2>&1 &
#fi

done

# Wait for all parallel jobs to finish
while [ 1 ]; do fg 2> /dev/null; [ $? == 1 ] && break; done

echo "Merging language models into $outfile" >> $logfile 2>&1
$scr/merge-sublm.pl --size $order --sublm $tmpdir/lm.dict -lm $outfile $backoff  >> $logfile 2>&1

echo "Cleaning temporary directory $tmpdir" >> $logfile 2>&1
rm $tmpdir/* 2> /dev/null

if [ $tmpdir_created -eq 1 ]; then
    echo "Removing temporary directory $tmpdir" >> $logfile 2>&1
    rmdir $tmpdir 2> /dev/null
    if [ $? != 0 ]; then
        echo "Warning: the temporary directory could not be removed." >> $logfile 2>&1
    fi
fi
 
exit 0

APT Browse - Built by Thomas Orozco.