epan/dvb_chartbl.c

   1 /* dvb_chartbl.c
   2  * Routines for handling DVB-SI character tables (as defined in EN 300 468)
   3  *
   4  * Wireshark - Network traffic analyzer
   5  * By Gerald Combs <gerald@wireshark.org>
   6  * Copyright 1998 Gerald Combs
   7  *
   8  * SPDX-License-Identifier: GPL-2.0-or-later
   9  */
  10
  11 #include "config.h"
  12
  13 #include <glib.h>
  14
  15 #include <epan/proto.h>
  16 #include <epan/tvbuff.h>
  17 #include <epan/value_string.h>
  18
  19 #include "dvb_chartbl.h"
  20
  21
  22 static const value_string dvb_string_encoding_vals[] = {
  23    { DVB_ENCODING_INVALID,  "Incorrect length for encoding" },
  24    { DVB_ENCODING_RESERVED, "Reserved for future use" },
  25    { DVB_ENCODING_UNKNOWN,  "Unknown/undefined encoding" },
  26
  27    { DVB_ENCODING_LATIN,    "Latin (default table)" },
  28
  29    { DVB_ENCODING_ISO_8859_1,         "ISO/IEC 8859-1 (West European)" },
  30    { DVB_ENCODING_ISO_8859_2,         "ISO/IEC 8859-2 (East European)" },
  31    { DVB_ENCODING_ISO_8859_3,         "ISO/IEC 8859-3 (South European)" },
  32    { DVB_ENCODING_ISO_8859_4,         "ISO/IEC 8859-4 (North and North-East European)" },
  33    { DVB_ENCODING_ISO_8859_5,         "ISO/IEC 8859-5 (Latin/Cyrillic)" },
  34    { DVB_ENCODING_ISO_8859_6,         "ISO/IEC 8859-6 (Latin/Arabic)" },
  35    { DVB_ENCODING_ISO_8859_7,         "ISO/IEC 8859-7 (Latin/Greek)" },
  36    { DVB_ENCODING_ISO_8859_8,         "ISO/IEC 8859-8 (Latin/Hebrew)" },
  37    { DVB_ENCODING_ISO_8859_9,         "ISO/IEC 8859-9 (West European & Turkish)" },
  38    { DVB_ENCODING_ISO_8859_10,        "ISO/IEC 8859-10 (North European)" },
  39    { DVB_ENCODING_ISO_8859_11,        "ISO/IEC 8859-11 (Thai)" },
  40    { DVB_ENCODING_ISO_8859_13,        "ISO/IEC 8859-13 (Baltic)" },
  41    { DVB_ENCODING_ISO_8859_14,        "ISO/IEC 8859-14 (Celtic)" },
  42    { DVB_ENCODING_ISO_8859_15,        "ISO/IEC 8859-15 (West European)" },
  43    { DVB_ENCODING_ISO_10646_BMP,      "ISO/IEC 10646 Basic Multilingual Plane" },
  44    { DVB_ENCODING_KSX_1001,           "KSX 1001-2004 (Korean character set)" },
  45    { DVB_ENCODING_GB_2312,            "GB-2312-1980 (Simplified Chinese)" },
  46    { DVB_ENCODING_ISO_10646_BIG5,     "ISO/IEC 10646 BIG5 subset" },
  47    { DVB_ENCODING_ISO_10646_UTF8_BMP,
  48        "ISO/IEC 10646 Basic Multilingual Plane, UTF-8 encoded" },
  49
  50    { 0, NULL }
  51 };
  52
  53
  54 static dvb_encoding_e
  55 dvb_analyze_string_charset0(uint8_t byte0)
  56 {
  57     switch (byte0) {
  58         case 0x01:
  59             return DVB_ENCODING_ISO_8859_5;
  60         case 0x02:
  61             return DVB_ENCODING_ISO_8859_6;
  62         case 0x03:
  63             return DVB_ENCODING_ISO_8859_7;
  64         case 0x04:
  65             return DVB_ENCODING_ISO_8859_8;
  66         case 0x05:
  67             return DVB_ENCODING_ISO_8859_9;
  68         case 0x06:
  69             return DVB_ENCODING_ISO_8859_10;
  70         case 0x07:
  71             return DVB_ENCODING_ISO_8859_11;
  72         case 0x08:
  73             return DVB_ENCODING_RESERVED; /* was reserved for ISO-8859-12 */
  74         case 0x09:
  75             return DVB_ENCODING_ISO_8859_13;
  76         case 0x0A:
  77             return DVB_ENCODING_ISO_8859_14;
  78         case 0x0B:
  79             return DVB_ENCODING_ISO_8859_15;
  80         case 0x11:
  81             return DVB_ENCODING_ISO_10646_BMP;
  82         case 0x12:
  83             return DVB_ENCODING_KSX_1001;
  84         case 0x13:
  85             return DVB_ENCODING_GB_2312;
  86         case 0x14:
  87             return DVB_ENCODING_ISO_10646_BIG5;
  88         case 0x15:
  89             return DVB_ENCODING_ISO_10646_UTF8_BMP;
  90
  91         default:
  92             return DVB_ENCODING_UNKNOWN;
  93     }
  94 }
  95
  96
  97 static dvb_encoding_e
  98 dvb_analyze_string_charset0_10(uint16_t byte12)
  99 {
 100     switch (byte12) {
 101         case 0x0000:
 102              return DVB_ENCODING_RESERVED;
 103         case 0x0001:
 104          return DVB_ENCODING_ISO_8859_1;
 105         case 0x0002:
 106          return DVB_ENCODING_ISO_8859_2;
 107         case 0x0003:
 108          return DVB_ENCODING_ISO_8859_3;
 109         case 0x0004:
 110          return DVB_ENCODING_ISO_8859_4;
 111         case 0x0005:
 112          return DVB_ENCODING_ISO_8859_5;
 113         case 0x0006:
 114          return DVB_ENCODING_ISO_8859_6;
 115         case 0x0007:
 116          return DVB_ENCODING_ISO_8859_7;
 117         case 0x0008:
 118          return DVB_ENCODING_ISO_8859_8;
 119         case 0x0009:
 120          return DVB_ENCODING_ISO_8859_9;
 121         case 0x000A:
 122          return DVB_ENCODING_ISO_8859_10;
 123         case 0x000B:
 124          return DVB_ENCODING_ISO_8859_11;
 125         case 0x000C:
 126          return DVB_ENCODING_RESERVED;
 127         case 0x000D:
 128          return DVB_ENCODING_ISO_8859_13;
 129         case 0x000E:
 130          return DVB_ENCODING_ISO_8859_14;
 131         case 0x000F:
 132          return DVB_ENCODING_ISO_8859_15;
 133
 134         default: /* 0x10 XX XX */
 135             return DVB_ENCODING_UNKNOWN;
 136     }
 137 }
 138
 139
 140 static dvb_encoding_e
 141 dvb_analyze_string_charset0_1F(uint8_t byte1)
 142 {
 143    /* http://www.dvbservices.com/identifiers/encoding_type_id */
 144
 145     switch (byte1) {
 146        case 0x00: /* 0x1F 0x00 */
 147           return DVB_ENCODING_RESERVED;
 148        case 0x01:
 149        case 0x02:
 150        case 0x03:
 151        case 0x04:
 152           /* XXX: BBC */
 153           return DVB_ENCODING_RESERVED;
 154        case 0x05:
 155        case 0x06:
 156           /* XXX: Malaysian Technical Standards Forum Bhd */
 157           return DVB_ENCODING_RESERVED;
 158
 159        default: /* 0x1F XX */
 160           return DVB_ENCODING_RESERVED;
 161     }
 162 }
 163
 164
 165 unsigned
 166 dvb_analyze_string_charset(tvbuff_t *tvb, int offset, int length, dvb_encoding_e *encoding)
 167 {
 168    if (length >= 1) {
 169       uint8_t byte0 = tvb_get_uint8(tvb, offset + 0);
 170
 171       if (byte0 >= 0x20) {
 172          /* the first byte is a normal character, not the number of a character table */
 173          *encoding = DVB_ENCODING_LATIN;
 174          return 0;
 175
 176       } else if (byte0 == 0x1F) {
 177          if (length >= 2) {
 178             *encoding = dvb_analyze_string_charset0_1F(tvb_get_uint8(tvb, offset + 1));
 179             return 2;
 180          }
 181          *encoding = DVB_ENCODING_INVALID;
 182          return 1;
 183
 184       } else if (byte0 >= 0x16) { /* 16 ... 1E */
 185          *encoding = DVB_ENCODING_RESERVED;
 186          return 1;
 187
 188       } else if (byte0 == 0x10) {
 189          if (length >= 3) {
 190             *encoding = dvb_analyze_string_charset0_10(tvb_get_ntohs(tvb, offset + 1));
 191             return 3;
 192          }
 193          *encoding = DVB_ENCODING_INVALID;
 194          return 1;
 195
 196       } else if ((byte0 >= 0x0C && byte0 <= 0x0F)) {
 197          *encoding = DVB_ENCODING_RESERVED;
 198          return 1;
 199       } else {
 200          *encoding = dvb_analyze_string_charset0(byte0);
 201          return 1;
 202       }
 203    } else
 204       *encoding = DVB_ENCODING_LATIN;
 205
 206    return 0;
 207 }
 208
 209
 210 unsigned
 211 dvb_enc_to_item_enc(dvb_encoding_e encoding)
 212 {
 213    /* XXX: take ISO control codes into account,
 214       e.g. 0x86 - turn emphasis on ; 0x87 - turn emphasis off */
 215
 216    switch (encoding) {
 217       case DVB_ENCODING_ISO_8859_1:
 218          return ENC_ISO_8859_1 | ENC_NA;
 219
 220       case DVB_ENCODING_ISO_8859_2:
 221          return ENC_ISO_8859_2 | ENC_NA;
 222
 223       case DVB_ENCODING_ISO_8859_3:
 224          return ENC_ISO_8859_3 | ENC_NA;
 225
 226       case DVB_ENCODING_ISO_8859_4:
 227          return ENC_ISO_8859_4 | ENC_NA;
 228
 229       case DVB_ENCODING_ISO_8859_5:
 230          return ENC_ISO_8859_5 | ENC_NA;
 231
 232       case DVB_ENCODING_ISO_8859_6:
 233          return ENC_ISO_8859_6 | ENC_NA;
 234
 235       case DVB_ENCODING_ISO_8859_7:
 236          return ENC_ISO_8859_7 | ENC_NA;
 237
 238       case DVB_ENCODING_ISO_8859_8:
 239          return ENC_ISO_8859_8 | ENC_NA;
 240
 241       case DVB_ENCODING_ISO_8859_9:
 242          return ENC_ISO_8859_9 | ENC_NA;
 243
 244       case DVB_ENCODING_ISO_8859_10:
 245          return ENC_ISO_8859_10 | ENC_NA;
 246
 247       case DVB_ENCODING_ISO_8859_11:
 248          return ENC_ISO_8859_11 | ENC_NA;
 249
 250       case DVB_ENCODING_ISO_8859_13:
 251          return ENC_ISO_8859_13 | ENC_NA;
 252
 253       case DVB_ENCODING_ISO_8859_14:
 254          return ENC_ISO_8859_14 | ENC_NA;
 255
 256       case DVB_ENCODING_ISO_8859_15:
 257          return ENC_ISO_8859_15 | ENC_NA;
 258
 259       case DVB_ENCODING_ISO_10646_UTF8_BMP:
 260          return ENC_UTF_8 | ENC_NA;
 261
 262       default: /* not supported */
 263          return ENC_ASCII | ENC_NA;
 264    }
 265 }
 266
 267
 268 void
 269 dvb_add_chartbl(proto_tree *tree, int hf,
 270         tvbuff_t *tvb, int offset, int length, dvb_encoding_e encoding)
 271 {
 272     if (tree == NULL)
 273         return;
 274
 275     if (length==0) {
 276         proto_item *pi;
 277
 278         pi = proto_tree_add_bytes_format(tree, hf, tvb, 0, 0, NULL,
 279                 "Default character table (Latin)");
 280         proto_item_set_generated(pi);
 281     }
 282     else {
 283         proto_tree_add_bytes_format_value(tree, hf,
 284             tvb, offset, length, NULL, "%s (%s)",
 285             val_to_str_const(encoding, dvb_string_encoding_vals, "Unknown"),
 286             tvb_bytes_to_str_punct(PNODE_POOL(tree), tvb, offset, length, ' '));
 287     }
 288 }
 289
 290 /*
 291  * Editor modelines  -  https://www.wireshark.org/tools/modelines.html
 292  *
 293  * Local variables:
 294  * c-basic-offset: 4
 295  * tab-width: 8
 296  * indent-tabs-mode: nil
 297  * End:
 298  *
 299  * vi: set shiftwidth=4 tabstop=8 expandtab:
 300  * :indentSize=4:tabSize=8:noTabs=true:
 301  */