txt2uc

   1 #!/bin/sh
   2
   3 #==============================================================================
   4 # txt2uc
   5 # File ID: 821a49a8-3776-11e5-8c4f-fefdb24f8e10
   6 #
   7 # Convert UTF-8 text into the format created by unichar(1).
   8 #
   9 # Author: Øyvind A. Holm <sunny@sunbase.org>
  10 # License: GNU General Public License version 2 or later.
  11 #==============================================================================
  12
  13 progname=txt2uc
  14 VERSION=0.3.1
  15
  16 opt_help=0
  17 opt_quiet=0
  18 opt_verbose=0
  19 while test -n "$1"; do
  20         case "$1" in
  21         -h|--help) opt_help=1; shift ;;
  22         -q|--quiet) opt_quiet=$(($opt_quiet + 1)); shift ;;
  23         -v|--verbose) opt_verbose=$(($opt_verbose + 1)); shift ;;
  24         --version) echo $progname $VERSION; exit 0 ;;
  25         --) shift; break ;;
  26         *)
  27                 if printf '%s\n' "$1" | grep -q ^-; then
  28                         echo "$progname: $1: Unknown option" >&2
  29                         exit 1
  30                 else
  31                         break
  32                 fi
  33         break ;;
  34         esac
  35 done
  36 opt_verbose=$(($opt_verbose - $opt_quiet))
  37
  38 if test "$opt_help" = "1"; then
  39         test $opt_verbose -gt 0 && { echo; echo $progname $VERSION; }
  40         cat <<END
  41
  42 Convert UTF-8 text into the format created by unichar(1). Displays all
  43 characters used, also things like combining chars and that bloody BOM.
  44
  45 Usage: $progname [options]
  46
  47 Options:
  48
  49   -h, --help
  50     Show this help.
  51   -q, --quiet
  52     Be more quiet. Can be repeated to increase silence.
  53   -v, --verbose
  54     Increase level of verbosity. Can be repeated.
  55   --version
  56     Print version information.
  57
  58 END
  59         exit 0
  60 fi
  61
  62 tohex -u | zeropad -s 4 -x | fmt -1 | while read f; do
  63         sqlite3 ~/.unichar.sqlite "SELECT * FROM chars WHERE s LIKE '$f;%';"
  64 done
  65
  66 # vim: set ts=8 sw=8 sts=8 noet fo+=w tw=79 fenc=UTF-8 :