ALT Linux Sisyphus

#!/bin/sh
#
# 	i2myspell - Ispell->MySpell affix table and dictionary converter
#
#	Version 1.4 (2002.3.8.)
#
#	(c) Copyright 2002 LАszlС NИmeth, All Rights Reserved
#
#	This program is free software; you can redistribute it and/or
#	modify it under the terms of the BSD License (see http://www.opensource.org)
#
case $# in
 0|1) echo "i2myspell ispell->myspell affix table and dictionary converter
generate Myspell aff file:
1. i2myspell ispell_hash_dictionary characterset upper lower [trial_range] 
    [compound_control_flag] [compoundmin]
generate Myspell dic file:
2. i2myspell -d ispell_dict_file

Example: 
1. i2myspell magyar ISO8859-2 A-Zаимсжузэш a-zАИМСЖУЗЭШ МСЗ Y 2 > hu_HU.aff
2. i2myspell -d magyar.dict > hu_HU.dic

ispell_hash_dictionary argument will become alternate dictionary argument 
of ispell -d flag, because i2myspell use Ispell generated affix table.
(See Ispell -D affix dump flag.)"; exit;;
esac

export LC_ALL=C # set correct sort, because sorting with hungarian locale is wrong

case $1 in
-d)

# dict file uniq
sed 's#/# #' $2 | sort -r -k 1 | uniq | grep -v ^$ |
awk '{ 
    if (p!=$1) { 
	printf "\n%s", $0; 
	p=$1 } 
    else { 
	if ($2!="") printf "/%s", $2; 
    }
}' | sed 's#/##g
s# #/#' >/tmp/i2my$$.1
cat /tmp/i2my$$.1 | wc -l | tr -cd '0-9\n'
tail +2 /tmp/i2my$$.1
echo
rm -f /tmp/i2my$$.1

exit;;
esac

ispell -d $1 -D | # dump affix table from the ispell hash
sed 's/ //g
/prefixes/,/suffixes/s/flag[*]\(.\):/PFX \1 Y /
/prefixes/,/suffixes/s/flag\(.\):/PFX \1 N /
/suffixes/,//s/flag[*]\(.\):/SFX \1 Y /
/suffixes/,//s/flag\(.\):/SFX \1 N /
s/\([^ ]*\).>.\([^,]*\)$/0 \2 \1/
s/\([^ ]*\).>.-\([^,]*\),\(.*\)$/\2 \3 \1/' |
tee /tmp/i2my$$.1 | cut -c -7 > /tmp/i2my$$.2

# myspell affix table header
echo "SET $2"
if [ -n "$5" ]; then echo "TRY $5"; fi;
if [ -n "$7" ]; then echo "COMPOUNDMIN $7"; fi;
if [ -n "$6" ]; then echo "COMPOUNDFLAG $6"; fi;

cut -c 8- /tmp/i2my$$.1 | 
tr $3 $4 |
paste -d "" /tmp/i2my$$.2 - | 
egrep -v '^(suffixes|prefixes|flagmarker)' |
sort -k 2 | awk '
    NR==1 { o1=$1; o2=$2; o3=$3; n[o2]=-1; }
    { 
    printf "%s %s   %-4s %-12s %-25s %s\n", $1, $2, $4, $5, $6, $7;
    if ($2==o2) { 
	n[$2]++; 
    } else { 
	printf "%s %s %s %s\n", o1, o2, o3, n[o2]+1; 
	o1=$1; o2=$2; o3=$3;
    }
} END { printf "%s %s %s %s\n", o1, o2, o3, n[o2]+1; } ' | tee x |
sed 's/  *$//' | sort -r | sed 's/\(^.*[0-9]\)$/\
\1/'
rm /tmp/i2my$$.*