INF=dict-ru.xml
OUTF=dict-ru-yo.xml
YOF=yo-words

TMPF=$(mktemp)
trap "rm -f $TMPF" EXIT

cat yo.txt | dos2unix | iconv -f cp1251 -t utf8 | sed '1,2d' | sed 's/(.*$//' | sed 's/,/\n/g' | sort | uniq > $YOF

for w in $(cat yo-words); do
	ww=$(echo $w | sed 's/ё/е/')
	out=$(fgrep ">${ww}<" $INF) || continue
	out2=$(echo $out | sed "s/${ww}/${w}/")
	#echo w=$w ww=$ww out=$out out2=$out2
	echo "$out2" >> $TMPF
done

cat $TMPF | sort | uniq > yo.xml