source: src/compdic/compdic-update-fst @ acbabee

#! /bin/bash

if [ $# -lt 2 ]
then
    echo "Usage:"
    echo "        compdic-update-fst <dictionary> <difference> <difference> ..."
    echo "where"
    echo "    <dictionary> - fst format automaton"
    echo "    <difference> - diff format file containing paths to be added/removed from dictionary"
    exit 0
fi      

tempdir=`mktemp -d /tmp/compdic.XXXXXX`

dict=$1
shift

dicplus=$tempdir/plus.dic
fstplus=$tempdir/plus.fst
dicminus=$tempdir/minus.dic
fstminus=$tempdir/minus.fst

touch $dicplus
touch $dicminus

while (($#))
do
    echo processing $1 ...
    cat $1 | egrep '^>' | sed -r 's/^> *//;s/[[:space:]].*$//' | canonize >> $dicplus
    cat $1 | egrep '^<' | sed -r 's/^< *//;s/[[:space:]].*$//' | canonize >> $dicminus
    shift
done

echo updating $dict ...

if (( `cat $dicminus | wc -l` ))
then
    tmpfst1=$tempdir/tmp1.fst
    echo "running compdic-dic-to-fst $dicminus $fstminus"
    compdic-dic-to-fst $dicminus $fstminus
    echo "running fstdifference $dict $fstminus | fstdeterminize > $tmpfst1"
    fstdifference $dict $fstminus > $tmpfst1
else
    tmpfst1=$dict
fi

if (( `cat $dicplus | wc -l` ))
then
    tmpfst2=$tempdir/tmp2.fst
    echo "running compdic-dic-to-fst $dicplus $fstplus"
    compdic-dic-to-fst $dicplus $fstplus
    echo "running fstunion $tmpfst1 $fstplus | fstdeterminize | fstminimize > $tmpfst2"
    fstunion $tmpfst1 $fstplus > $tmpfst2
else
    tmpfst2=$tmpfst1
fi



echo "running mv ${dict} ${dict}~"
mv ${dict} ${dict}~

echo "cat ${tmpfst2} | fstrmepsilon | fstdeterminize | fstminimize  > ${dict}"
cat ${tmpfst2} | fstrmepsilon | fstdeterminize | fstminimize  > ${dict}
chmod a+r ${dict}

rm -r $tempdir