unichar

   1 #!/bin/sh
   2
   3 #==============================================================================
   4 # unichar
   5 # File ID: 23f515ba-5d47-11df-bcb0-90e6ba3022ac
   6 #
   7 # Create a searchable SQLite database of all Unicode characters.
   8 #
   9 # Author: Øyvind A. Holm <sunny@sunbase.org>
  10 # License: GNU General Public License version 2 or later.
  11 #==============================================================================
  12
  13 progname=unichar
  14 VERSION=0.5.0
  15
  16 db="$HOME/.unichar.sqlite"
  17 guipdf="$HOME/bin/Lib/unicode/current/charts/CodeCharts.pdf"
  18 uc_url="ftp://unicode.org/Public/UNIDATA/UnicodeData.txt"
  19 pdfprog=evince
  20
  21 opt_convert=0
  22 opt_get=0
  23 opt_gui=0
  24 opt_help=0
  25 opt_init=''
  26 opt_quiet=0
  27 opt_verbose=0
  28 while test -n "$1"; do
  29         case "$1" in
  30         -c|--convert) opt_convert=1; shift ;;
  31         --get) opt_get=1; shift ;;
  32         -g|--gui) opt_gui=1; shift ;;
  33         -h|--help) opt_help=1; shift ;;
  34         --init) opt_init=$2; shift 2 ;;
  35         -q|--quiet) opt_quiet=$(($opt_quiet + 1)); shift ;;
  36         -v|--verbose) opt_verbose=$(($opt_verbose + 1)); shift ;;
  37         --version) echo $progname $VERSION; exit 0 ;;
  38         --) shift; break ;;
  39         *)
  40                 if printf '%s\n' "$1" | grep -q ^-; then
  41                         echo "$progname: $1: Unknown option" >&2
  42                         exit 1
  43                 else
  44                         break
  45                 fi
  46         break ;;
  47         esac
  48 done
  49 opt_verbose=$(($opt_verbose - $opt_quiet))
  50
  51 if test "$opt_help" = "1"; then
  52         test $opt_verbose -gt 0 && { echo; echo $progname $VERSION; }
  53         cat <<END
  54
  55 Search for Unicode characters, different arguments are ANDed against
  56 each other.
  57
  58 Usage:
  59
  60   $progname --init directory_with_Unicode_files [--get]
  61     Create the database and optionally download the necessary files from
  62     unicode.org .
  63
  64   $progname search_string [search_string [...]]
  65     Search through the database after Unicode characters.
  66
  67   $progname -c
  68     See description for -c/--convert.
  69
  70 Options:
  71
  72   -c, --convert
  73     Read $progname output (search result) from stdin and convert it into
  74     proper UTF-8.
  75   --get
  76     Used together with --init. Automatically download the necessary
  77     files from ftp://unicode.org/Public/UNIDATA/ into directory
  78     specified with --init. At the moment only UnicodeData.txt is used.
  79   -g, --gui
  80     Display the Unicode character table with $pdfprog(1).
  81     Location: $guipdf
  82   -h, --help
  83     Show this help.
  84   --init DIRECTORY
  85     Initialise a new database from UnicodeData.txt in DIRECTORY.
  86   -q, --quiet
  87     Be more quiet. Can be repeated to increase silence.
  88   -v, --verbose
  89     Increase level of verbosity. Can be repeated.
  90   --version
  91     Print version information.
  92
  93 For example (with Unicode 8.0.0):
  94
  95   $ $progname latin | wc -l
  96   1492
  97
  98   $ $progname latin ring | wc -l
  99   17
 100
 101   $ $progname die face
 102   2680;DIE FACE-1;So;0;ON;;;;;N;;;;;
 103   2681;DIE FACE-2;So;0;ON;;;;;N;;;;;
 104   2682;DIE FACE-3;So;0;ON;;;;;N;;;;;
 105   2683;DIE FACE-4;So;0;ON;;;;;N;;;;;
 106   2684;DIE FACE-5;So;0;ON;;;;;N;;;;;
 107   2685;DIE FACE-6;So;0;ON;;;;;N;;;;;
 108
 109   $ $progname fac angr
 110   1F620;ANGRY FACE;So;0;ON;;;;;N;;;;;
 111
 112   $ $progname latin above ring | $progname -c
 113   ÅåŮůǺǻẘẙ
 114
 115 END
 116         exit 0
 117 fi
 118
 119 if test -n "$opt_init"; then
 120     if test -e "$db"; then
 121         echo $progname: $db: SQLite database already exists >&2
 122         echo $progname: Please remove it and try again >&2
 123         exit 1
 124     fi
 125
 126     if test "$opt_get" = "1"; then
 127         mkdir -p "$opt_init"
 128         cd "$opt_init" || {
 129             echo $progname: Cannot change directory to \'$opt_init\' >&2
 130             exit 1
 131         }
 132         if wget --version | grep -q "GNU Wget"; then
 133             wget "$uc_url" || {
 134                 echo $progname: Error during download from $uc_url >&2
 135                 exit 1
 136             }
 137         elif curl --version | grep -q '^curl [0-9]'; then
 138             curl "$uc_url" -o UnicodeData.txt || {
 139                 echo $progname: Error during download from $uc_url >&2
 140                 exit 1
 141             }
 142         else
 143             echo $progname: Unable to find a suitable download program >&2
 144             echo $progname: \(searched for wget and curl\) >&2
 145             exit 1
 146         fi
 147         cd - >/dev/null
 148     fi
 149
 150     if test ! -d "$opt_init/."; then
 151         echo $progname: $opt_init: Non-existent or unavailable directory >&2
 152         exit 1
 153     fi
 154
 155     ucdata="$opt_init/UnicodeData.txt"
 156
 157     if test ! -r "$ucdata"; then
 158         echo $progname: $ucdata not found or is not readable >&2
 159         exit 1
 160     fi
 161
 162     sqlite3 "$db" "CREATE TABLE chars (s varchar);"
 163     sqlite3 "$db" "CREATE INDEX idx_chars_s ON chars (s);"
 164     sqlite3 "$db" ".import '$ucdata' chars" && {
 165         echo $progname: $ucdata successfully imported into $db >&2
 166         exit 0
 167     } || {
 168         echo $progname: Something went wrong during import into $db >&2
 169         exit 1
 170     }
 171 fi
 172
 173 if test "$opt_convert" = "1"; then
 174     # Convert from unichar output to actual UTF-8
 175     cut -f 1 -d ';' | fromhex -u
 176     exit 0
 177 fi
 178
 179 if test "$opt_gui" = "1"; then
 180     $pdfprog "$guipdf" &
 181     exit 0
 182 fi
 183
 184 if test ! -r "$db"; then
 185     echo $progname: Database $db not found, >&2
 186     echo you need to run \'$progname --init\' >&2
 187     exit 1
 188 fi
 189
 190 unset and_str search_str
 191 for f in "$@"; do
 192     search_str="$search_str$and_str s LIKE '%$f%'"
 193     and_str=" AND"
 194 done
 195 test -z "$search_str" && {
 196     echo $progname: No search string specified >&2
 197     exit 1
 198 }
 199 sqlite3 "$db" "SELECT * FROM chars WHERE $search_str"
 200
 201 # vim: set ts=8 sw=8 sts=8 noet fo+=w tw=79 fenc=UTF-8 :