#!/usr/local/bin/perl
#
# slavonic-sort
# Author: Justin Zamora
#
# This program sorts a transliterated Church Slavonic text according
# to the ordering of the letters of the Church Slavonic alphabet.
# It uses the transliteration used on the "Help Me Learn Church Slavonic"
# web site, http://justin.zamora.com/slavonic
#
# This program is designed for sorting a vocabulary list.  It assumes
# that a dash surrounded by spaces separates a transliterated word
# from its definition.  If there is no dash on the line, it will treat
# the entire line as transliterated text.

if (@ARGV && $#ARGV != 0)
{
   die "Usage: $0 file\n";
}

while (<>)
{
$line = $_;

# Separate the word from the definition
($word, $def, @rest) = split(/ - /, $line);
if (@rest)
{
   print "Entry for $word has extra dash\n";
   exit 1;
}
if (!$def)
{
   chomp($word);
}

# Remove diacritics
$word =~ s/[\/\^\\\`]//go;

# Generate and prepend sortable collating sequence
$word =~ s/^ja/R/go;
$word =~ s/ksja$/3=T/go;
$word =~ s/psja$/9=T/go;
$word =~ s/shch/H/go;
$word =~ s/^wt/B/go;
$word =~ s/zh/*/go;
$word =~ s/ch/D/go;
$word =~ s/sh/G/go;	

# Handle i before another i
do
{
    $count = $word =~ s/ii/Ii/o;
}
while ($count > 0);

# Handle i before a vowel
$word =~ s/ia/2#/go;
$word =~ s/i[eE]/2)/go;
$word =~ s/iI/22/go;
$word =~ s/i[oO]/27/go;
$word =~ s/iy/2L/go;
$word =~ s/iw/28/go;
$word =~ s/iu/2?/go;
$word =~ s/ije/2N/go;
$word =~ s/iju/2Q/go;
$word =~ s/ija/2T/go;
$word =~ s/ij/21/go;

$word =~ s/je/N/go;
$word =~ s/ju/Q/go;
$word =~ s/ja/T/go;
$word =~ s/ks/U/go;
$word =~ s/ps/W/go;
$word =~ s/i/0/go;

$word =~ tr|abvgdeEZzjIkKlmnoOwpPrsStufxc"y'FV|#$%&())+/123345677899<==>?@ACJLMXY|;

push @words, ($word . " | " . $line);

}

@sorted = sort @words;

for $line (@sorted)
{
   $line =~ s/^.*\| //;
   print $line;
}