cyr2lat

#!/bin/sh
# translit - Serbian alphabet transilteration

# transliterate input from UTF-8 Serbian Cyrillic to Latin:
# cyr2lat [ file ... ]
#
# transliterate input from UTF-8 Serbian Latin to 7-bit ASCII:
# lat2ascii [ file ... ]
#
# Requires: AWK that supports UTF-8 (looks for GNU AWK)
#
# Author: Vladimir Vrzić
#
# translit is free software; you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation; either version 2, or (at your option)
# any later version.

searchpath() {
# Usage: searchpath progname path
  IFS=${PATH_SEPARATOR-":"} prog=':'
  for dir in $2; do
    # try `progname' with all well known extensions
    for ext in '' '.exe'; do
      try="$dir/$1$ext"
      test -f "$try" && test -x "$try" && prog="$try" && break
    done
    test "$prog" = ":" || break
  done
  echo "$prog"
}

AWK=$(searchpath gawk "$PATH")
if [ "$AWK" = ":" ]; then
  echo >&2 "$0: Cannot find gawk in PATH"
  exit 1
fi

awk_cyr2lat_map='
BEGIN {
  m["А"] = "A" ; m["а"] = "a"
  m["Б"] = "B" ; m["б"] = "b"
  m["В"] = "V" ; m["в"] = "v"
  m["Г"] = "G" ; m["г"] = "g"
  m["Д"] = "D" ; m["д"] = "d"
  m["Ђ"] = "Đ" ; m["ђ"] = "đ"
  m["Е"] = "E" ; m["е"] = "e"
  m["Ж"] = "Ž" ; m["ж"] = "ž"
  m["З"] = "Z" ; m["з"] = "z"
  m["И"] = "I" ; m["и"] = "i"
  m["Ј"] = "J" ; m["ј"] = "j"
  m["К"] = "K" ; m["к"] = "k"
  m["Л"] = "L" ; m["л"] = "l"
  m["Љ"] = "Lj"; m["љ"] = "lj"
  m["М"] = "M" ; m["м"] = "m"
  m["Н"] = "N" ; m["н"] = "n"
  m["Њ"] = "Nj"; m["њ"] = "nj"
  m["О"] = "O" ; m["о"] = "o"
  m["П"] = "P" ; m["п"] = "p"
  m["Р"] = "R" ; m["р"] = "r"
  m["С"] = "S" ; m["с"] = "s"
  m["Т"] = "T" ; m["т"] = "t"
  m["Ћ"] = "Ć" ; m["ћ"] = "ć"
  m["У"] = "U" ; m["у"] = "u"
  m["Ф"] = "F" ; m["ф"] = "f"
  m["Х"] = "H" ; m["х"] = "h"
  m["Ц"] = "C" ; m["ц"] = "c"
  m["Ч"] = "Č" ; m["ч"] = "č"
  m["Џ"] = "Dž"; m["џ"] = "dž"
  m["Ш"] = "Š" ; m["ш"] = "š"
  FS=""
}
'
awk_lat2ascii_map='
BEGIN {
  m["Č"] = "C" ; m["č"] = "c"
  m["Ć"] = "C" ; m["ć"] = "c"
  m["Đ"] = "Dj"; m["đ"] = "dj"
  m["Š"] = "S" ; m["š"] = "s"
  m["Ž"] = "Z" ; m["ž"] = "z"
  FS=""
}
'
awk_each_line='
{
  for (i = 1; i <= NF; i++)
    if (m[$i]) printf("%s", m[$i])
    else printf("%s", $i)
  printf("\n")
}
'

awk_begin="$awk_cyr2lat_map"
if [ "$(basename "$0")" = 'lat2ascii' ]; then
  awk_begin="$awk_lat2ascii_map"
fi

exec $AWK "$awk_begin $awk_each_line" "$@"