#!/bin/sh

#
#   SCRIPT WHICH GENERATES THE CHARMAP
#

function generateCharacterMap() {

    # header
    echo '<?php // desc="Character map for Unicode to ASCII conversion"'
    echo 
    echo "    return array( 0x0000 => ''"

    # lookup SMALL characters
    gzip -d < /usr/share/i18n/charmaps/UTF-8.gz \
        | grep -e ' [A-Z] ' \
        | grep -Ee 'LATIN|CYRILIC' \
        | grep -e 'SMALL' \
        | tr ABCDEFGHIJKLMNOPQRSTUVWXYZ abcdefghijklmnopqrstuvwxyz \
        | cut -c3-6,25- \
        | sed "s/^\([^ ]\+\).* \([a-zA-Z]\) .*$/                , 0x\1 => '\2'/"

    # lookup NON SMALL characters
    gzip -d < /usr/share/i18n/charmaps/UTF-8.gz \
        | grep -e ' [A-Z] ' \
        | grep -Ee 'LATIN|CYRILIC' \
        | grep -ve 'SMALL' \
        | cut -c3-6,25- \
        | sed "s/^\([^ ]\+\).* \([a-zA-Z]\) .*$/                , 0x\1 => '\2'/"

    # footer
    echo '                );'
    echo
    echo -n '?>'
}

generateCharacterMap >'./charmap.inc.php'