enc/mktable.c

   1 /**********************************************************************
   2   mktable.c
   3 **********************************************************************/
   4 /*-
   5  * Copyright (c) 2002-2016  K.Kosako  <sndgk393 AT ybb DOT ne DOT jp>
   6  * All rights reserved.
   7  *
   8  * Redistribution and use in source and binary forms, with or without
   9  * modification, are permitted provided that the following conditions
  10  * are met:
  11  * 1. Redistributions of source code must retain the above copyright
  12  *    notice, this list of conditions and the following disclaimer.
  13  * 2. Redistributions in binary form must reproduce the above copyright
  14  *    notice, this list of conditions and the following disclaimer in the
  15  *    documentation and/or other materials provided with the distribution.
  16  *
  17  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
  18  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  19  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  20  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
  21  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  22  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  23  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  24  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  25  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  26  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  27  * SUCH DAMAGE.
  28  */
  29
  30 #include <stdlib.h>
  31 #include <stdio.h>
  32 #include <locale.h>
  33
  34 #ifndef __USE_ISOC99
  35 #define __USE_ISOC99
  36 #endif
  37
  38 #include <ctype.h>
  39
  40 #include "regenc.h"
  41
  42 #define ASCII                0
  43 #define UNICODE_ISO_8859_1   1
  44 #define ISO_8859_1           2
  45 #define ISO_8859_2           3
  46 #define ISO_8859_3           4
  47 #define ISO_8859_4           5
  48 #define ISO_8859_5           6
  49 #define ISO_8859_6           7
  50 #define ISO_8859_7           8
  51 #define ISO_8859_8           9
  52 #define ISO_8859_9          10
  53 #define ISO_8859_10         11
  54 #define ISO_8859_11         12
  55 #define ISO_8859_13         13
  56 #define ISO_8859_14         14
  57 #define ISO_8859_15         15
  58 #define ISO_8859_16         16
  59 #define KOI8                17
  60 #define KOI8_R              18
  61
  62 typedef struct {
  63   int   num;
  64   const char* name;
  65 } ENC_INFO;
  66
  67 static ENC_INFO Info[] = {
  68   { ASCII,               "ASCII" },
  69   { UNICODE_ISO_8859_1,  "UNICODE_ISO_8859_1"  },
  70   { ISO_8859_1,  "ISO_8859_1"  },
  71   { ISO_8859_2,  "ISO_8859_2"  },
  72   { ISO_8859_3,  "ISO_8859_3"  },
  73   { ISO_8859_4,  "ISO_8859_4"  },
  74   { ISO_8859_5,  "ISO_8859_5"  },
  75   { ISO_8859_6,  "ISO_8859_6"  },
  76   { ISO_8859_7,  "ISO_8859_7"  },
  77   { ISO_8859_8,  "ISO_8859_8"  },
  78   { ISO_8859_9,  "ISO_8859_9"  },
  79   { ISO_8859_10, "ISO_8859_10" },
  80   { ISO_8859_11, "ISO_8859_11" },
  81   { ISO_8859_13, "ISO_8859_13" },
  82   { ISO_8859_14, "ISO_8859_14" },
  83   { ISO_8859_15, "ISO_8859_15" },
  84   { ISO_8859_16, "ISO_8859_16" },
  85   { KOI8,        "KOI8" },
  86   { KOI8_R,      "KOI8_R" }
  87 };
  88
  89
  90 static int IsAlpha(int enc, int c)
  91 {
  92   if (enc == ASCII)
  93     return isalpha(c);
  94
  95   if (c >= 0x41 && c <= 0x5a) return 1;
  96   if (c >= 0x61 && c <= 0x7a) return 1;
  97
  98   switch (enc) {
  99   case UNICODE_ISO_8859_1:
 100   case ISO_8859_1:
 101   case ISO_8859_9:
 102     if (c == 0xaa) return 1;
 103     if (c == 0xb5) return 1;
 104     if (c == 0xba) return 1;
 105     if (c >= 0xc0 && c <= 0xd6) return 1;
 106     if (c >= 0xd8 && c <= 0xf6) return 1;
 107     if (c >= 0xf8 && c <= 0xff) return 1;
 108     break;
 109
 110   case ISO_8859_2:
 111     if (c == 0xa1 || c == 0xa3) return 1;
 112     if (c == 0xa5 || c == 0xa6) return 1;
 113     if (c >= 0xa9 && c <= 0xac) return 1;
 114     if (c >= 0xae && c <= 0xaf) return 1;
 115     if (c == 0xb1 || c == 0xb3) return 1;
 116     if (c == 0xb5 || c == 0xb6) return 1;
 117     if (c >= 0xb9 && c <= 0xbc) return 1;
 118     if (c >= 0xbe && c <= 0xbf) return 1;
 119     if (c >= 0xc0 && c <= 0xd6) return 1;
 120     if (c >= 0xd8 && c <= 0xf6) return 1;
 121     if (c >= 0xf8 && c <= 0xfe) return 1;
 122     break;
 123
 124   case ISO_8859_3:
 125     if (c == 0xa1) return 1;
 126     if (c == 0xa6) return 1;
 127     if (c >= 0xa9 && c <= 0xac) return 1;
 128     if (c == 0xaf) return 1;
 129     if (c == 0xb1) return 1;
 130     if (c == 0xb5 || c == 0xb6) return 1;
 131     if (c >= 0xb9 && c <= 0xbc) return 1;
 132     if (c == 0xbf) return 1;
 133     if (c >= 0xc0 && c <= 0xc2) return 1;
 134     if (c >= 0xc4 && c <= 0xcf) return 1;
 135     if (c >= 0xd1 && c <= 0xd6) return 1;
 136     if (c >= 0xd8 && c <= 0xe2) return 1;
 137     if (c >= 0xe4 && c <= 0xef) return 1;
 138     if (c >= 0xf1 && c <= 0xf6) return 1;
 139     if (c >= 0xf8 && c <= 0xfe) return 1;
 140     break;
 141
 142   case ISO_8859_4:
 143     if (c >= 0xa1 && c <= 0xa3) return 1;
 144     if (c == 0xa5 || c == 0xa6) return 1;
 145     if (c >= 0xa9 && c <= 0xac) return 1;
 146     if (c == 0xae) return 1;
 147     if (c == 0xb1 || c == 0xb3) return 1;
 148     if (c == 0xb5 || c == 0xb6) return 1;
 149     if (c >= 0xb9 && c <= 0xbf) return 1;
 150     if (c >= 0xc0 && c <= 0xd6) return 1;
 151     if (c >= 0xd8 && c <= 0xf6) return 1;
 152     if (c >= 0xf8 && c <= 0xfe) return 1;
 153     break;
 154
 155   case ISO_8859_5:
 156     if (c >= 0xa1 && c <= 0xcf && c != 0xad) return 1;
 157     if (c >= 0xd0 && c <= 0xff && c != 0xf0 && c != 0xfd) return 1;
 158     break;
 159
 160   case ISO_8859_6:
 161     if (c >= 0xc1 && c <= 0xda) return 1;
 162     if (c >= 0xe0 && c <= 0xf2) return 1;
 163     break;
 164
 165   case ISO_8859_7:
 166     if (c == 0xb6) return 1;
 167     if (c >= 0xb8 && c <= 0xba) return 1;
 168     if (c == 0xbc) return 1;
 169     if (c >= 0xbe && c <= 0xbf) return 1;
 170     if (c == 0xc0) return 1;
 171     if (c >= 0xc1 && c <= 0xdb && c != 0xd2) return 1;
 172     if (c >= 0xdc && c <= 0xfe) return 1;
 173     break;
 174
 175   case ISO_8859_8:
 176     if (c == 0xb5) return 1;
 177     if (c >= 0xe0 && c <= 0xfa) return 1;
 178     break;
 179
 180   case ISO_8859_10:
 181     if (c >= 0xa1 && c <= 0xa6) return 1;
 182     if (c >= 0xa8 && c <= 0xac) return 1;
 183     if (c == 0xae || c == 0xaf) return 1;
 184     if (c >= 0xb1 && c <= 0xb6) return 1;
 185     if (c >= 0xb8 && c <= 0xbc) return 1;
 186     if (c >= 0xbe && c <= 0xff) return 1;
 187     break;
 188
 189   case ISO_8859_11:
 190     if (c >= 0xa1 && c <= 0xda) return 1;
 191     if (c >= 0xdf && c <= 0xfb) return 1;
 192     break;
 193
 194   case ISO_8859_13:
 195     if (c == 0xa8) return 1;
 196     if (c == 0xaa) return 1;
 197     if (c == 0xaf) return 1;
 198     if (c == 0xb5) return 1;
 199     if (c == 0xb8) return 1;
 200     if (c == 0xba) return 1;
 201     if (c >= 0xbf && c <= 0xd6) return 1;
 202     if (c >= 0xd8 && c <= 0xf6) return 1;
 203     if (c >= 0xf8 && c <= 0xfe) return 1;
 204     break;
 205
 206   case ISO_8859_14:
 207     if (c == 0xa1 || c == 0xa2) return 1;
 208     if (c == 0xa4 || c == 0xa5) return 1;
 209     if (c == 0xa6 || c == 0xa8) return 1;
 210     if (c >= 0xaa && c <= 0xac) return 1;
 211     if (c >= 0xaf && c <= 0xb5) return 1;
 212     if (c >= 0xb7 && c <= 0xff) return 1;
 213     break;
 214
 215   case ISO_8859_15:
 216     if (c == 0xaa) return 1;
 217     if (c == 0xb5) return 1;
 218     if (c == 0xba) return 1;
 219     if (c >= 0xc0 && c <= 0xd6) return 1;
 220     if (c >= 0xd8 && c <= 0xf6) return 1;
 221     if (c >= 0xf8 && c <= 0xff) return 1;
 222     if (c == 0xa6) return 1;
 223     if (c == 0xa8) return 1;
 224     if (c == 0xb4) return 1;
 225     if (c == 0xb8) return 1;
 226     if (c == 0xbc) return 1;
 227     if (c == 0xbd) return 1;
 228     if (c == 0xbe) return 1;
 229     break;
 230
 231   case ISO_8859_16:
 232     if (c == 0xa1) return 1;
 233     if (c == 0xa2) return 1;
 234     if (c == 0xa3) return 1;
 235     if (c == 0xa6) return 1;
 236     if (c == 0xa8) return 1;
 237     if (c == 0xaa) return 1;
 238     if (c == 0xac) return 1;
 239     if (c == 0xae) return 1;
 240     if (c == 0xaf) return 1;
 241     if (c == 0xb2) return 1;
 242     if (c == 0xb3) return 1;
 243     if (c == 0xb4) return 1;
 244     if (c >= 0xb8 && c <= 0xba) return 1;
 245     if (c == 0xbc) return 1;
 246     if (c == 0xbd) return 1;
 247     if (c == 0xbe) return 1;
 248     if (c == 0xbf) return 1;
 249     if (c >= 0xc0 && c <= 0xde) return 1;
 250     if (c >= 0xdf && c <= 0xff) return 1;
 251     break;
 252
 253   case KOI8_R:
 254     if (c == 0xa3 || c == 0xb3) return 1;
 255     /* fall */
 256   case KOI8:
 257     if (c >= 0xc0 && c <= 0xff) return 1;
 258     break;
 259
 260   default:
 261     exit(-1);
 262   }
 263
 264   return 0;
 265 }
 266
 267 static int IsBlank(int enc, int c)
 268 {
 269   if (enc == ASCII)
 270     return isblank(c);
 271
 272   if (c == 0x09 || c == 0x20) return 1;
 273
 274   switch (enc) {
 275   case UNICODE_ISO_8859_1:
 276   case ISO_8859_1:
 277   case ISO_8859_2:
 278   case ISO_8859_3:
 279   case ISO_8859_4:
 280   case ISO_8859_5:
 281   case ISO_8859_6:
 282   case ISO_8859_7:
 283   case ISO_8859_8:
 284   case ISO_8859_9:
 285   case ISO_8859_10:
 286   case ISO_8859_11:
 287   case ISO_8859_13:
 288   case ISO_8859_14:
 289   case ISO_8859_15:
 290   case ISO_8859_16:
 291   case KOI8:
 292     if (c == 0xa0) return 1;
 293     break;
 294
 295   case KOI8_R:
 296     if (c == 0x9a) return 1;
 297     break;
 298
 299   default:
 300     exit(-1);
 301   }
 302
 303   return 0;
 304 }
 305
 306 static int IsCntrl(int enc, int c)
 307 {
 308   if (enc == ASCII)
 309     return iscntrl(c);
 310
 311   if (c >= 0x00 && c <= 0x1F) return 1;
 312
 313   switch (enc) {
 314   case UNICODE_ISO_8859_1:
 315     if (c == 0xad) return 1;
 316     /* fall */
 317   case ISO_8859_1:
 318   case ISO_8859_2:
 319   case ISO_8859_3:
 320   case ISO_8859_4:
 321   case ISO_8859_5:
 322   case ISO_8859_6:
 323   case ISO_8859_7:
 324   case ISO_8859_8:
 325   case ISO_8859_9:
 326   case ISO_8859_10:
 327   case ISO_8859_11:
 328   case ISO_8859_13:
 329   case ISO_8859_14:
 330   case ISO_8859_15:
 331   case ISO_8859_16:
 332   case KOI8:
 333     if (c >= 0x7f && c <= 0x9F) return 1;
 334     break;
 335
 336
 337   case KOI8_R:
 338     if (c == 0x7f) return 1;
 339     break;
 340
 341   default:
 342     exit(-1);
 343   }
 344
 345   return 0;
 346 }
 347
 348 static int IsDigit(int enc ARG_UNUSED, int c)
 349 {
 350   if (c >= 0x30 && c <= 0x39) return 1;
 351   return 0;
 352 }
 353
 354 static int IsGraph(int enc, int c)
 355 {
 356   if (enc == ASCII)
 357     return isgraph(c);
 358
 359   if (c >= 0x21 && c <= 0x7e) return 1;
 360
 361   switch (enc) {
 362   case UNICODE_ISO_8859_1:
 363   case ISO_8859_1:
 364   case ISO_8859_2:
 365   case ISO_8859_4:
 366   case ISO_8859_5:
 367   case ISO_8859_9:
 368   case ISO_8859_10:
 369   case ISO_8859_13:
 370   case ISO_8859_14:
 371   case ISO_8859_15:
 372   case ISO_8859_16:
 373     if (c >= 0xa1 && c <= 0xff) return 1;
 374     break;
 375
 376   case ISO_8859_3:
 377     if (c >= 0xa1) {
 378       if (c == 0xa5 || c == 0xae || c == 0xbe || c == 0xc3 || c == 0xd0 ||
 379           c == 0xe3 || c == 0xf0)
 380         return 0;
 381       else
 382         return 1;
 383     }
 384     break;
 385
 386   case ISO_8859_6:
 387     if (c == 0xa4 || c == 0xac || c == 0xad || c == 0xbb || c == 0xbf)
 388       return 1;
 389     if (c >= 0xc1 && c <= 0xda) return 1;
 390     if (c >= 0xe0 && c <= 0xf2) return 1;
 391     break;
 392
 393   case ISO_8859_7:
 394     if (c >= 0xa1 && c <= 0xfe &&
 395         c != 0xa4 && c != 0xa5 && c != 0xaa &&
 396         c != 0xae && c != 0xd2) return 1;
 397     break;
 398
 399   case ISO_8859_8:
 400     if (c >= 0xa2 && c <= 0xfa) {
 401       if (c >= 0xbf && c <= 0xde) return 0;
 402       return 1;
 403     }
 404     break;
 405
 406   case ISO_8859_11:
 407     if (c >= 0xa1 && c <= 0xda) return 1;
 408     if (c >= 0xdf && c <= 0xfb) return 1;
 409     break;
 410
 411   case KOI8:
 412     if (c >= 0xc0 && c <= 0xff) return 1;
 413     break;
 414
 415   case KOI8_R:
 416     if (c >= 0x80 && c <= 0xff && c != 0x9a) return 1;
 417     break;
 418
 419   default:
 420     exit(-1);
 421   }
 422
 423   return 0;
 424 }
 425
 426 static int IsLower(int enc, int c)
 427 {
 428   if (enc == ASCII)
 429     return islower(c);
 430
 431   if (c >= 0x61 && c <= 0x7a) return 1;
 432
 433   switch (enc) {
 434   case UNICODE_ISO_8859_1:
 435   case ISO_8859_1:
 436   case ISO_8859_9:
 437     if (c == 0xaa) return 1;
 438     if (c == 0xb5) return 1;
 439     if (c == 0xba) return 1;
 440     if (c >= 0xdf && c <= 0xf6) return 1;
 441     if (c >= 0xf8 && c <= 0xff) return 1;
 442     break;
 443
 444   case ISO_8859_2:
 445     if (c == 0xb1 || c == 0xb3) return 1;
 446     if (c == 0xb5 || c == 0xb6) return 1;
 447     if (c >= 0xb9 && c <= 0xbc) return 1;
 448     if (c >= 0xbe && c <= 0xbf) return 1;
 449     if (c >= 0xdf && c <= 0xf6) return 1;
 450     if (c >= 0xf8 && c <= 0xfe) return 1;
 451     break;
 452
 453   case ISO_8859_3:
 454     if (c == 0xb1) return 1;
 455     if (c == 0xb5 || c == 0xb6) return 1;
 456     if (c >= 0xb9 && c <= 0xbc) return 1;
 457     if (c == 0xbf) return 1;
 458     if (c == 0xdf) return 1;
 459     if (c >= 0xe0 && c <= 0xe2) return 1;
 460     if (c >= 0xe4 && c <= 0xef) return 1;
 461     if (c >= 0xf1 && c <= 0xf6) return 1;
 462     if (c >= 0xf8 && c <= 0xfe) return 1;
 463     break;
 464
 465   case ISO_8859_4:
 466     if (c == 0xa2) return 1;
 467     if (c == 0xb1 || c == 0xb3) return 1;
 468     if (c == 0xb5 || c == 0xb6) return 1;
 469     if (c >= 0xb9 && c <= 0xbc) return 1;
 470     if (c >= 0xbe && c <= 0xbf) return 1;
 471     if (c == 0xdf) return 1;
 472     if (c >= 0xe0 && c <= 0xf6) return 1;
 473     if (c >= 0xf8 && c <= 0xfe) return 1;
 474     break;
 475
 476   case ISO_8859_5:
 477     if (c >= 0xd0 && c <= 0xff && c != 0xf0 && c != 0xfd) return 1;
 478     break;
 479
 480   case ISO_8859_6:
 481     break;
 482
 483   case ISO_8859_7:
 484     if (c == 0xc0) return 1;
 485     if (c >= 0xdc && c <= 0xfe) return 1;
 486     break;
 487
 488   case ISO_8859_8:
 489     if (c == 0xb5) return 1;
 490     break;
 491
 492   case ISO_8859_10:
 493     if (c >= 0xb1 && c <= 0xb6) return 1;
 494     if (c >= 0xb8 && c <= 0xbc) return 1;
 495     if (c == 0xbe || c == 0xbf) return 1;
 496     if (c >= 0xdf && c <= 0xff) return 1;
 497     break;
 498
 499   case ISO_8859_11:
 500     break;
 501
 502   case ISO_8859_13:
 503     if (c == 0xb5) return 1;
 504     if (c == 0xb8) return 1;
 505     if (c == 0xba) return 1;
 506     if (c == 0xbf) return 1;
 507     if (c >= 0xdf && c <= 0xf6) return 1;
 508     if (c >= 0xf8 && c <= 0xfe) return 1;
 509     break;
 510
 511   case ISO_8859_14:
 512     if (c == 0xa2) return 1;
 513     if (c == 0xa5) return 1;
 514     if (c == 0xab) return 1;
 515     if (c == 0xb1 || c == 0xb3 || c == 0xb5) return 1;
 516     if (c >= 0xb8 && c <= 0xba) return 1;
 517     if (c == 0xbc) return 1;
 518     if (c == 0xbe || c == 0xbf) return 1;
 519     if (c >= 0xdf && c <= 0xff) return 1;
 520     break;
 521
 522   case ISO_8859_15:
 523     if (c == 0xaa) return 1;
 524     if (c == 0xb5) return 1;
 525     if (c == 0xba) return 1;
 526     if (c >= 0xdf && c <= 0xf6) return 1;
 527     if (c >= 0xf8 && c <= 0xff) return 1;
 528     if (c == 0xa8) return 1;
 529     if (c == 0xb8) return 1;
 530     if (c == 0xbd) return 1;
 531     break;
 532
 533   case ISO_8859_16:
 534     if (c == 0xa2) return 1;
 535     if (c == 0xa8) return 1;
 536     if (c == 0xae) return 1;
 537     if (c == 0xb3) return 1;
 538     if (c >= 0xb8 && c <= 0xba) return 1;
 539     if (c == 0xbd) return 1;
 540     if (c == 0xbf) return 1;
 541     if (c >= 0xdf && c <= 0xff) return 1;
 542     break;
 543
 544   case KOI8_R:
 545     if (c == 0xa3) return 1;
 546     /* fall */
 547   case KOI8:
 548     if (c >= 0xc0 && c <= 0xdf) return 1;
 549     break;
 550
 551   default:
 552     exit(-1);
 553   }
 554
 555   return 0;
 556 }
 557
 558 static int IsPrint(int enc, int c)
 559 {
 560   if (enc == ASCII)
 561     return isprint(c);
 562
 563   if (c >= 0x20 && c <= 0x7e) return 1;
 564
 565   switch (enc) {
 566   case UNICODE_ISO_8859_1:
 567     /* if (c >= 0x09 && c <= 0x0d) return 1; */
 568     if (c == 0x85) return 1;
 569     /* fall */
 570   case ISO_8859_1:
 571   case ISO_8859_2:
 572   case ISO_8859_4:
 573   case ISO_8859_5:
 574   case ISO_8859_9:
 575   case ISO_8859_10:
 576   case ISO_8859_13:
 577   case ISO_8859_14:
 578   case ISO_8859_15:
 579   case ISO_8859_16:
 580     if (c >= 0xa0 && c <= 0xff) return 1;
 581     break;
 582
 583   case ISO_8859_3:
 584     if (c >= 0xa0) {
 585       if (c == 0xa5 || c == 0xae || c == 0xbe || c == 0xc3 || c == 0xd0 ||
 586           c == 0xe3 || c == 0xf0)
 587         return 0;
 588       else
 589         return 1;
 590     }
 591     break;
 592
 593   case ISO_8859_6:
 594     if (c == 0xa0) return 1;
 595     if (c == 0xa4 || c == 0xac || c == 0xad || c == 0xbb || c == 0xbf)
 596       return 1;
 597     if (c >= 0xc1 && c <= 0xda) return 1;
 598     if (c >= 0xe0 && c <= 0xf2) return 1;
 599     break;
 600
 601   case ISO_8859_7:
 602     if (c >= 0xa0 && c <= 0xfe &&
 603         c != 0xa4 && c != 0xa5 && c != 0xaa &&
 604         c != 0xae && c != 0xd2) return 1;
 605     break;
 606
 607   case ISO_8859_8:
 608     if (c >= 0xa0 && c <= 0xfa) {
 609       if (c >= 0xbf && c <= 0xde) return 0;
 610       if (c == 0xa1) return 0;
 611       return 1;
 612     }
 613     break;
 614
 615   case ISO_8859_11:
 616     if (c >= 0xa0 && c <= 0xda) return 1;
 617     if (c >= 0xdf && c <= 0xfb) return 1;
 618     break;
 619
 620   case KOI8:
 621     if (c == 0xa0) return 1;
 622     if (c >= 0xc0 && c <= 0xff) return 1;
 623     break;
 624
 625   case KOI8_R:
 626     if (c >= 0x80 && c <= 0xff) return 1;
 627     break;
 628
 629   default:
 630     exit(-1);
 631   }
 632
 633   return 0;
 634 }
 635
 636 static int IsPunct(int enc, int c)
 637 {
 638   if (enc == ASCII)
 639     return ispunct(c);
 640
 641   if (enc == UNICODE_ISO_8859_1) {
 642     if (c == 0x24 || c == 0x2b || c == 0x5e || c == 0x60 ||
 643         c == 0x7c || c == 0x7e) return 1;
 644     if (c >= 0x3c && c <= 0x3e) return 1;
 645   }
 646
 647   if (c >= 0x21 && c <= 0x2f) return 1;
 648   if (c >= 0x3a && c <= 0x40) return 1;
 649   if (c >= 0x5b && c <= 0x60) return 1;
 650   if (c >= 0x7b && c <= 0x7e) return 1;
 651
 652   switch (enc) {
 653   case ISO_8859_1:
 654   case ISO_8859_9:
 655   case ISO_8859_15:
 656     if (c == 0xad) return 1;
 657     /* fall */
 658   case UNICODE_ISO_8859_1:
 659     if (c == 0xa1) return 1;
 660     if (c == 0xab) return 1;
 661     if (c == 0xb7) return 1;
 662     if (c == 0xbb) return 1;
 663     if (c == 0xbf) return 1;
 664     break;
 665
 666   case ISO_8859_2:
 667   case ISO_8859_4:
 668   case ISO_8859_5:
 669   case ISO_8859_14:
 670     if (c == 0xad) return 1;
 671     break;
 672
 673   case ISO_8859_3:
 674   case ISO_8859_10:
 675     if (c == 0xad) return 1;
 676     if (c == 0xb7) return 1;
 677     if (c == 0xbd) return 1;
 678     break;
 679
 680   case ISO_8859_6:
 681     if (c == 0xac) return 1;
 682     if (c == 0xad) return 1;
 683     if (c == 0xbb) return 1;
 684     if (c == 0xbf) return 1;
 685     break;
 686
 687   case ISO_8859_7:
 688     if (c == 0xa1 || c == 0xa2) return 1;
 689     if (c == 0xab) return 1;
 690     if (c == 0xaf) return 1;
 691     if (c == 0xad) return 1;
 692     if (c == 0xb7 || c == 0xbb) return 1;
 693     break;
 694
 695   case ISO_8859_8:
 696     if (c == 0xab) return 1;
 697     if (c == 0xad) return 1;
 698     if (c == 0xb7) return 1;
 699     if (c == 0xbb) return 1;
 700     if (c == 0xdf) return 1;
 701     break;
 702
 703   case ISO_8859_13:
 704     if (c == 0xa1 || c == 0xa5) return 1;
 705     if (c == 0xab || c == 0xad) return 1;
 706     if (c == 0xb4 || c == 0xb7) return 1;
 707     if (c == 0xbb) return 1;
 708     if (c == 0xff) return 1;
 709     break;
 710
 711   case ISO_8859_16:
 712     if (c == 0xa5) return 1;
 713     if (c == 0xab) return 1;
 714     if (c == 0xad) return 1;
 715     if (c == 0xb5) return 1;
 716     if (c == 0xb7) return 1;
 717     if (c == 0xbb) return 1;
 718     break;
 719
 720   case KOI8_R:
 721     if (c == 0x9e) return 1;
 722     break;
 723
 724   case ISO_8859_11:
 725   case KOI8:
 726     break;
 727
 728   default:
 729     exit(-1);
 730   }
 731
 732   return 0;
 733 }
 734
 735 static int IsSpace(int enc, int c)
 736 {
 737   if (enc == ASCII)
 738     return isspace(c);
 739
 740   if (c >= 0x09 && c <= 0x0d) return 1;
 741   if (c == 0x20) return 1;
 742
 743   switch (enc) {
 744   case UNICODE_ISO_8859_1:
 745     if (c == 0x85) return 1;
 746     /* fall */
 747   case ISO_8859_1:
 748   case ISO_8859_2:
 749   case ISO_8859_3:
 750   case ISO_8859_4:
 751   case ISO_8859_5:
 752   case ISO_8859_6:
 753   case ISO_8859_7:
 754   case ISO_8859_8:
 755   case ISO_8859_9:
 756   case ISO_8859_10:
 757   case ISO_8859_11:
 758   case ISO_8859_13:
 759   case ISO_8859_14:
 760   case ISO_8859_15:
 761   case ISO_8859_16:
 762   case KOI8:
 763     if (c == 0xa0) return 1;
 764     break;
 765
 766   case KOI8_R:
 767     if (c == 0x9a) return 1;
 768     break;
 769
 770   default:
 771     exit(-1);
 772   }
 773
 774   return 0;
 775 }
 776
 777 static int IsUpper(int enc, int c)
 778 {
 779   if (enc == ASCII)
 780     return isupper(c);
 781
 782   if (c >= 0x41 && c <= 0x5a) return 1;
 783
 784   switch (enc) {
 785   case UNICODE_ISO_8859_1:
 786   case ISO_8859_1:
 787   case ISO_8859_9:
 788     if (c >= 0xc0 && c <= 0xd6) return 1;
 789     if (c >= 0xd8 && c <= 0xde) return 1;
 790     break;
 791
 792   case ISO_8859_2:
 793     if (c == 0xa1 || c == 0xa3) return 1;
 794     if (c == 0xa5 || c == 0xa6) return 1;
 795     if (c >= 0xa9 && c <= 0xac) return 1;
 796     if (c >= 0xae && c <= 0xaf) return 1;
 797     if (c >= 0xc0 && c <= 0xd6) return 1;
 798     if (c >= 0xd8 && c <= 0xde) return 1;
 799     break;
 800
 801   case ISO_8859_3:
 802     if (c == 0xa1) return 1;
 803     if (c == 0xa6) return 1;
 804     if (c >= 0xa9 && c <= 0xac) return 1;
 805     if (c == 0xaf) return 1;
 806     if (c >= 0xc0 && c <= 0xc2) return 1;
 807     if (c >= 0xc4 && c <= 0xcf) return 1;
 808     if (c >= 0xd1 && c <= 0xd6) return 1;
 809     if (c >= 0xd8 && c <= 0xde) return 1;
 810     break;
 811
 812   case ISO_8859_4:
 813     if (c == 0xa1 || c == 0xa3) return 1;
 814     if (c == 0xa5 || c == 0xa6) return 1;
 815     if (c >= 0xa9 && c <= 0xac) return 1;
 816     if (c == 0xae) return 1;
 817     if (c == 0xbd) return 1;
 818     if (c >= 0xc0 && c <= 0xd6) return 1;
 819     if (c >= 0xd8 && c <= 0xde) return 1;
 820     break;
 821
 822   case ISO_8859_5:
 823     if (c >= 0xa1 && c <= 0xcf && c != 0xad) return 1;
 824     break;
 825
 826   case ISO_8859_6:
 827     break;
 828
 829   case ISO_8859_7:
 830     if (c == 0xb6) return 1;
 831     if (c >= 0xb8 && c <= 0xba) return 1;
 832     if (c == 0xbc) return 1;
 833     if (c >= 0xbe && c <= 0xbf) return 1;
 834     if (c >= 0xc1 && c <= 0xdb && c != 0xd2) return 1;
 835     break;
 836
 837   case ISO_8859_8:
 838   case ISO_8859_11:
 839     break;
 840
 841   case ISO_8859_10:
 842     if (c >= 0xa1 && c <= 0xa6) return 1;
 843     if (c >= 0xa8 && c <= 0xac) return 1;
 844     if (c == 0xae || c == 0xaf) return 1;
 845     if (c >= 0xc0 && c <= 0xde) return 1;
 846     break;
 847
 848   case ISO_8859_13:
 849     if (c == 0xa8) return 1;
 850     if (c == 0xaa) return 1;
 851     if (c == 0xaf) return 1;
 852     if (c >= 0xc0 && c <= 0xd6) return 1;
 853     if (c >= 0xd8 && c <= 0xde) return 1;
 854     break;
 855
 856   case ISO_8859_14:
 857     if (c == 0xa1) return 1;
 858     if (c == 0xa4 || c == 0xa6) return 1;
 859     if (c == 0xa8) return 1;
 860     if (c == 0xaa || c == 0xac) return 1;
 861     if (c == 0xaf || c == 0xb0) return 1;
 862     if (c == 0xb2 || c == 0xb4 || c == 0xb7) return 1;
 863     if (c == 0xbb || c == 0xbd) return 1;
 864     if (c >= 0xc0 && c <= 0xde) return 1;
 865     break;
 866
 867   case ISO_8859_15:
 868     if (c >= 0xc0 && c <= 0xd6) return 1;
 869     if (c >= 0xd8 && c <= 0xde) return 1;
 870     if (c == 0xa6) return 1;
 871     if (c == 0xb4) return 1;
 872     if (c == 0xbc) return 1;
 873     if (c == 0xbe) return 1;
 874     break;
 875
 876   case ISO_8859_16:
 877     if (c == 0xa1) return 1;
 878     if (c == 0xa3) return 1;
 879     if (c == 0xa6) return 1;
 880     if (c == 0xaa) return 1;
 881     if (c == 0xac) return 1;
 882     if (c == 0xaf) return 1;
 883     if (c == 0xb2) return 1;
 884     if (c == 0xb4) return 1;
 885     if (c == 0xbc) return 1;
 886     if (c == 0xbe) return 1;
 887     if (c >= 0xc0 && c <= 0xde) return 1;
 888     break;
 889
 890   case KOI8_R:
 891     if (c == 0xb3) return 1;
 892     /* fall */
 893   case KOI8:
 894     if (c >= 0xe0 && c <= 0xff) return 1;
 895     break;
 896
 897   default:
 898     exit(-1);
 899   }
 900
 901   return 0;
 902 }
 903
 904 static int IsXDigit(int enc, int c)
 905 {
 906   if (enc == ASCII)
 907     return isxdigit(c);
 908
 909   if (c >= 0x30 && c <= 0x39) return 1;
 910   if (c >= 0x41 && c <= 0x46) return 1;
 911   if (c >= 0x61 && c <= 0x66) return 1;
 912   return 0;
 913 }
 914
 915 static int IsWord(int enc, int c)
 916 {
 917   if (enc == ASCII) {
 918     return (isalpha(c) || isdigit(c) || c == 0x5f);
 919   }
 920
 921   if (c >= 0x30 && c <= 0x39) return 1;
 922   if (c >= 0x41 && c <= 0x5a) return 1;
 923   if (c == 0x5f) return 1;
 924   if (c >= 0x61 && c <= 0x7a) return 1;
 925
 926   switch (enc) {
 927   case UNICODE_ISO_8859_1:
 928   case ISO_8859_1:
 929   case ISO_8859_9:
 930     if (c == 0xaa) return 1;
 931     if (c >= 0xb2 && c <= 0xb3) return 1;
 932     if (c == 0xb5) return 1;
 933     if (c >= 0xb9 && c <= 0xba) return 1;
 934     if (c >= 0xbc && c <= 0xbe) return 1;
 935     if (c >= 0xc0 && c <= 0xd6) return 1;
 936     if (c >= 0xd8 && c <= 0xf6) return 1;
 937     if (c >= 0xf8 && c <= 0xff) return 1;
 938     break;
 939
 940   case ISO_8859_2:
 941     if (c == 0xa1 || c == 0xa3) return 1;
 942     if (c == 0xa5 || c == 0xa6) return 1;
 943     if (c >= 0xa9 && c <= 0xac) return 1;
 944     if (c >= 0xae && c <= 0xaf) return 1;
 945     if (c == 0xb1 || c == 0xb3) return 1;
 946     if (c == 0xb5 || c == 0xb6) return 1;
 947     if (c >= 0xb9 && c <= 0xbc) return 1;
 948     if (c >= 0xbe && c <= 0xbf) return 1;
 949     if (c >= 0xc0 && c <= 0xd6) return 1;
 950     if (c >= 0xd8 && c <= 0xf6) return 1;
 951     if (c >= 0xf8 && c <= 0xfe) return 1;
 952     break;
 953
 954   case ISO_8859_3:
 955     if (c == 0xa1) return 1;
 956     if (c == 0xa6) return 1;
 957     if (c >= 0xa9 && c <= 0xac) return 1;
 958     if (c == 0xaf) return 1;
 959     if (c >= 0xb1 && c <= 0xb3) return 1;
 960     if (c == 0xb5 || c == 0xb6) return 1;
 961     if (c >= 0xb9 && c <= 0xbd) return 1;
 962     if (c == 0xbf) return 1;
 963     if (c >= 0xc0 && c <= 0xc2) return 1;
 964     if (c >= 0xc4 && c <= 0xcf) return 1;
 965     if (c >= 0xd1 && c <= 0xd6) return 1;
 966     if (c >= 0xd8 && c <= 0xe2) return 1;
 967     if (c >= 0xe4 && c <= 0xef) return 1;
 968     if (c >= 0xf1 && c <= 0xf6) return 1;
 969     if (c >= 0xf8 && c <= 0xfe) return 1;
 970     break;
 971
 972   case ISO_8859_4:
 973     if (c >= 0xa1 && c <= 0xa3) return 1;
 974     if (c == 0xa5 || c == 0xa6) return 1;
 975     if (c >= 0xa9 && c <= 0xac) return 1;
 976     if (c == 0xae) return 1;
 977     if (c == 0xb1 || c == 0xb3) return 1;
 978     if (c == 0xb5 || c == 0xb6) return 1;
 979     if (c >= 0xb9 && c <= 0xbf) return 1;
 980     if (c >= 0xc0 && c <= 0xd6) return 1;
 981     if (c >= 0xd8 && c <= 0xf6) return 1;
 982     if (c >= 0xf8 && c <= 0xfe) return 1;
 983     break;
 984
 985   case ISO_8859_5:
 986     if (c >= 0xa1 && c <= 0xcf && c != 0xad) return 1;
 987     if (c >= 0xd0 && c <= 0xff && c != 0xf0 && c != 0xfd) return 1;
 988     break;
 989
 990   case ISO_8859_6:
 991     if (c >= 0xc1 && c <= 0xda) return 1;
 992     if (c >= 0xe0 && c <= 0xea) return 1;
 993     if (c >= 0xeb && c <= 0xf2) return 1;
 994     break;
 995
 996   case ISO_8859_7:
 997     if (c == 0xb2 || c == 0xb3) return 1;
 998     if (c == 0xb6) return 1;
 999     if (c >= 0xb8 && c <= 0xba) return 1;
1000     if (c >= 0xbc && c <= 0xbf) return 1;
1001     if (c == 0xc0) return 1;
1002     if (c >= 0xc1 && c <= 0xdb && c != 0xd2) return 1;
1003     if (c >= 0xdc && c <= 0xfe) return 1;
1004     break;
1005
1006   case ISO_8859_8:
1007     if (c == 0xb2 || c == 0xb3 || c == 0xb5 || c == 0xb9) return 1;
1008     if (c >= 0xbc && c <= 0xbe) return 1;
1009     if (c >= 0xe0 && c <= 0xfa) return 1;
1010     break;
1011
1012   case ISO_8859_10:
1013     if (c >= 0xa1 && c <= 0xff) {
1014       if (c != 0xa7 && c != 0xad && c != 0xb0 && c != 0xb7 && c != 0xbd)
1015         return 1;
1016     }
1017     break;
1018
1019   case ISO_8859_11:
1020     if (c >= 0xa1 && c <= 0xda) return 1;
1021     if (c >= 0xdf && c <= 0xfb) return 1;
1022     break;
1023
1024   case ISO_8859_13:
1025     if (c == 0xa8) return 1;
1026     if (c == 0xaa) return 1;
1027     if (c == 0xaf) return 1;
1028     if (c == 0xb2 || c == 0xb3 || c == 0xb5 || c == 0xb9) return 1;
1029     if (c >= 0xbc && c <= 0xbe) return 1;
1030     if (c == 0xb8) return 1;
1031     if (c == 0xba) return 1;
1032     if (c >= 0xbf && c <= 0xd6) return 1;
1033     if (c >= 0xd8 && c <= 0xf6) return 1;
1034     if (c >= 0xf8 && c <= 0xfe) return 1;
1035     break;
1036
1037   case ISO_8859_14:
1038     if (c >= 0xa1 && c <= 0xff) {
1039       if (c == 0xa3 || c == 0xa7 || c == 0xa9 || c == 0xad || c == 0xae ||
1040           c == 0xb6) return 0;
1041       return 1;
1042     }
1043     break;
1044
1045   case ISO_8859_15:
1046     if (c == 0xaa) return 1;
1047     if (c >= 0xb2 && c <= 0xb3) return 1;
1048     if (c == 0xb5) return 1;
1049     if (c >= 0xb9 && c <= 0xba) return 1;
1050     if (c >= 0xbc && c <= 0xbe) return 1;
1051     if (c >= 0xc0 && c <= 0xd6) return 1;
1052     if (c >= 0xd8 && c <= 0xf6) return 1;
1053     if (c >= 0xf8 && c <= 0xff) return 1;
1054     if (c == 0xa6) return 1;
1055     if (c == 0xa8) return 1;
1056     if (c == 0xb4) return 1;
1057     if (c == 0xb8) return 1;
1058     break;
1059
1060   case ISO_8859_16:
1061     if (c == 0xa1) return 1;
1062     if (c == 0xa2) return 1;
1063     if (c == 0xa3) return 1;
1064     if (c == 0xa6) return 1;
1065     if (c == 0xa8) return 1;
1066     if (c == 0xaa) return 1;
1067     if (c == 0xac) return 1;
1068     if (c == 0xae) return 1;
1069     if (c == 0xaf) return 1;
1070     if (c == 0xb2) return 1;
1071     if (c == 0xb3) return 1;
1072     if (c == 0xb4) return 1;
1073     if (c >= 0xb8 && c <= 0xba) return 1;
1074     if (c == 0xbc) return 1;
1075     if (c == 0xbd) return 1;
1076     if (c == 0xbe) return 1;
1077     if (c == 0xbf) return 1;
1078     if (c >= 0xc0 && c <= 0xde) return 1;
1079     if (c >= 0xdf && c <= 0xff) return 1;
1080     break;
1081
1082   case KOI8_R:
1083     if (c == 0x9d) return 1;
1084     if (c == 0xa3 || c == 0xb3) return 1;
1085     /* fall */
1086   case KOI8:
1087     if (c >= 0xc0 && c <= 0xff) return 1;
1088     break;
1089
1090   default:
1091     exit(-1);
1092   }
1093
1094   return 0;
1095 }
1096
1097 static int IsAscii(int enc ARG_UNUSED, int c)
1098 {
1099   if (c >= 0x00 && c <= 0x7f) return 1;
1100   return 0;
1101 }
1102
1103 static int IsNewline(int enc ARG_UNUSED, int c)
1104 {
1105   if (c == 0x0a) return 1;
1106   return 0;
1107 }
1108
1109 static int exec(FILE* fp, ENC_INFO* einfo)
1110 {
1111 #define NCOL  8
1112
1113   int c, val, enc;
1114   int r;
1115
1116   enc = einfo->num;
1117
1118   r = fprintf(fp, "static const unsigned short Enc%s_CtypeTable[256] = {\n",
1119               einfo->name);
1120   if (r < 0) return -1;
1121
1122   for (c = 0; c < 256; c++) {
1123     val = 0;
1124     if (IsNewline(enc, c))  val |= BIT_CTYPE_NEWLINE;
1125     if (IsAlpha (enc, c))   val |= (BIT_CTYPE_ALPHA | BIT_CTYPE_ALNUM);
1126     if (IsBlank (enc, c))   val |= BIT_CTYPE_BLANK;
1127     if (IsCntrl (enc, c))   val |= BIT_CTYPE_CNTRL;
1128     if (IsDigit (enc, c))   val |= (BIT_CTYPE_DIGIT | BIT_CTYPE_ALNUM);
1129     if (IsGraph (enc, c))   val |= BIT_CTYPE_GRAPH;
1130     if (IsLower (enc, c))   val |= BIT_CTYPE_LOWER;
1131     if (IsPrint (enc, c))   val |= BIT_CTYPE_PRINT;
1132     if (IsPunct (enc, c))   val |= BIT_CTYPE_PUNCT;
1133     if (IsSpace (enc, c))   val |= BIT_CTYPE_SPACE;
1134     if (IsUpper (enc, c))   val |= BIT_CTYPE_UPPER;
1135     if (IsXDigit(enc, c))   val |= BIT_CTYPE_XDIGIT;
1136     if (IsWord  (enc, c))   val |= BIT_CTYPE_WORD;
1137     if (IsAscii (enc, c))   val |= BIT_CTYPE_ASCII;
1138
1139     if (c % NCOL == 0) {
1140       r = fputs("  ", fp);
1141       if (r < 0) return -1;
1142     }
1143     r = fprintf(fp, "0x%04x", val);
1144     if (r < 0) return -1;
1145
1146     if (c != 255) {
1147       r = fputs(",", fp);
1148       if (r < 0) return -1;
1149     }
1150     if (c != 0 && c % NCOL == (NCOL-1))
1151       r = fputs("\n", fp);
1152     else
1153       r = fputs(" ", fp);
1154
1155     if (r < 0) return -1;
1156   }
1157   r = fprintf(fp, "};\n");
1158   if (r < 0) return -1;
1159
1160   return 0;
1161 }
1162
1163 extern int main(int argc ARG_UNUSED, char* argv[] ARG_UNUSED)
1164 {
1165   int r;
1166   int i;
1167   FILE* fp = stdout;
1168
1169   setlocale(LC_ALL, "C");
1170   /* setlocale(LC_ALL, "POSIX"); */
1171   /* setlocale(LC_ALL, "en_GB.iso88591"); */
1172   /* setlocale(LC_ALL, "de_BE.iso88591"); */
1173   /* setlocale(LC_ALL, "fr_FR.iso88591"); */
1174
1175   for (i = 0; i < (int )(sizeof(Info)/sizeof(ENC_INFO)); i++) {
1176     r = exec(fp, &Info[i]);
1177     if (r < 0) {
1178       fprintf(stderr, "FAIL exec(): %d\n", r);
1179       return -1;
1180     }
1181   }
1182
1183   return 0;
1184 }