#! /usr/bin/gawk -f
# Last edited on 2004-02-04 03:27:48 by stolfi

BEGIN{ 
  split("", ct);
  name["91"] = "HEBREW_ACCENT_ETNAHTA";
  name["92"] = "HEBREW_ACCENT_SEGOL";
  name["93"] = "HEBREW_ACCENT_SHALSHELET";
  name["94"] = "HEBREW_ACCENT_ZAQEF_QATAN";
  name["95"] = "HEBREW_ACCENT_ZAQEF_GADOL";
  name["96"] = "HEBREW_ACCENT_TIPEHA";
  name["97"] = "HEBREW_ACCENT_REVIA";
  name["98"] = "HEBREW_ACCENT_ZARQA";
  name["99"] = "HEBREW_ACCENT_PASHTA";
  name["9A"] = "HEBREW_ACCENT_YETIV";
  name["9B"] = "HEBREW_ACCENT_TEVIR";
  name["9C"] = "HEBREW_ACCENT_GERESH";
  name["9D"] = "HEBREW_ACCENT_GERESH_MUQDAM";
  name["9E"] = "HEBREW_ACCENT_GERSHAYIM";
  name["9F"] = "HEBREW_ACCENT_QARNEY_PARA";
  name["A0"] = "HEBREW_ACCENT_TELISHA_GEDOLA";
  name["A1"] = "HEBREW_ACCENT_PAZER";
  name["A3"] = "HEBREW_ACCENT_MUNAH";
  name["A4"] = "HEBREW_ACCENT_MAHAPAKH";
  name["A5"] = "HEBREW_ACCENT_MERKHA";
  name["A6"] = "HEBREW_ACCENT_MERKHA_KEFULA";
  name["A7"] = "HEBREW_ACCENT_DARGA";
  name["A8"] = "HEBREW_ACCENT_QADMA";
  name["A9"] = "HEBREW_ACCENT_TELISHA_QETANA";
  name["AA"] = "HEBREW_ACCENT_YERAH_BEN_YOMO";
  name["AB"] = "HEBREW_ACCENT_OLE";
  name["AC"] = "HEBREW_ACCENT_ILUY";
  name["AD"] = "HEBREW_ACCENT_DEHI";
  name["AE"] = "HEBREW_ACCENT_ZINOR";
  name["AF"] = "HEBREW_MARK_MASORA_CIRCLE";
  name["B0"] = "HEBREW_POINT_SHEVA";
  name["B1"] = "HEBREW_POINT_HATAF_SEGOL";
  name["B2"] = "HEBREW_POINT_HATAF_PATAH";
  name["B3"] = "HEBREW_POINT_HATAF_QAMATS";
  name["B4"] = "HEBREW_POINT_HIRIQ";
  name["B5"] = "HEBREW_POINT_TSERE";
  name["B6"] = "HEBREW_POINT_SEGOL";
  name["B7"] = "HEBREW_POINT_PATAH";
  name["B8"] = "HEBREW_POINT_QAMATS";
  name["B9"] = "HEBREW_POINT_HOLAM";
  name["BB"] = "HEBREW_POINT_QUBUTS";
  name["BC"] = "HEBREW_POINT_DAGESH_OR_MAPIQ";
  name["BD"] = "HEBREW_POINT_METEG";
  name["BE"] = "HEBREW_PUNCTUATION_MAQAF";
  name["BF"] = "HEBREW_POINT_RAFE";
  name["C0"] = "HEBREW_PUNCTUATION_PASEQ";
  name["C1"] = "HEBREW_POINT_SHIN_DOT";
  name["C2"] = "HEBREW_POINT_SIN_DOT";
  name["C3"] = "HEBREW_PUNCTUATION_SOF_PASUQ";
  name["C4"] = "HEBREW_MARK_UPPER_DOT";
  name["D0"] = "HEBREW_LETTER_ALEF";
  name["D1"] = "HEBREW_LETTER_BET";
  name["D2"] = "HEBREW_LETTER_GIMEL";
  name["D3"] = "HEBREW_LETTER_DALET";
  name["D4"] = "HEBREW_LETTER_HE";
  name["D5"] = "HEBREW_LETTER_VAV";
  name["D6"] = "HEBREW_LETTER_ZAYIN";
  name["D7"] = "HEBREW_LETTER_HET";
  name["D8"] = "HEBREW_LETTER_TET";
  name["D9"] = "HEBREW_LETTER_YOD";
  name["DA"] = "HEBREW_LETTER_FINAL_KAF";
  name["DB"] = "HEBREW_LETTER_KAF";
  name["DC"] = "HEBREW_LETTER_LAMED";
  name["DD"] = "HEBREW_LETTER_FINAL_MEM";
  name["DE"] = "HEBREW_LETTER_MEM";
  name["DF"] = "HEBREW_LETTER_FINAL_NUN";
  name["E0"] = "HEBREW_LETTER_NUN";
  name["E1"] = "HEBREW_LETTER_SAMEKH";
  name["E2"] = "HEBREW_LETTER_AYIN";
  name["E3"] = "HEBREW_LETTER_FINAL_PE";
  name["E4"] = "HEBREW_LETTER_PE";
  name["E5"] = "HEBREW_LETTER_FINAL_TSADI";
  name["E6"] = "HEBREW_LETTER_TSADI";
  name["E7"] = "HEBREW_LETTER_QOF";
  name["E8"] = "HEBREW_LETTER_RESH";
  name["E9"] = "HEBREW_LETTER_SHIN";
  name["EA"] = "HEBREW_LETTER_TAV";
  name["F0"] = "HEBREW_LIGATURE_YIDDISH_DOUBLE_VAV";
  name["F1"] = "HEBREW_LIGATURE_YIDDISH_VAV_YOD";
  name["F2"] = "HEBREW_LIGATURE_YIDDISH_DOUBLE_YOD";
  name["F3"] = "HEBREW_PUNCTUATION_GERESH";
  name["F4"] = "HEBREW_PUNCTUATION_GERSHAYIM";
  name["__"] = "SPACE";
  name["**"] = "INVALID_CHARACTER";
}

/^ *([\#]|$)/ { next; }

/[«»]/ {
  lin = $0;
  gsub(/^[^«»]*[«]/, "«", lin);
  gsub(/[»][^«»]*$/, "»", lin);
  gsub(/[»][^«»]*[«]/, "» «", lin);
  $0 = lin;
  for (i = 1; i <= NF; i++) 
    { code = $(i);
      if (! (code ~ /^[«][_\/*0-9A-F][_\/*0-9A-F][»]$/))
        { printf "bad code = \"%s\"\n", code > "/dev/stderr"; }
      code = substr(code, 2,2);
      ct[code]++;
    }
  next;
}

END{
  for (code in ct) 
    { printf " %7d  %s  %s\n", ct[code], code, name[code]; }
}