source: src/compdic/compdic-fst-update @ f4bf33e

if [ $# -lt 2 ]
then
    echo "Usage:"
    echo "        compdic-fst-update <dictionary> <difference> <difference> ..."
    echo "where"
    echo "    <dictionary> - file containig a list of words, one per line, iso-8859-2 encoded"
    echo "    <difference> - a file to which the compiled automaton (cor/kor format) shoul be written"
    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
tmpfst=$tempdir/tmp.fst

touch $dicplus
touch $dicminus

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

echo updating $dict ...
compdic-create-fst $dicplus $fstplus
compdic-create-fst $dicminus $fstminus
fstdifference $dict $fstminus | fstdeterminize > $tmpfst
fstunion $tmpfst $fstplus | fstdeterminize | fstminimize > $tmpfst
mv ${dict} ${dict}~
mv $tmpfst ${dict}


rm -r $tempdir