contrib/less/charset.c

   1 /*
   2  * Copyright (C) 1984-2015  Mark Nudelman
   3  *
   4  * You may distribute under the terms of either the GNU General Public
   5  * License or the Less License, as specified in the README file.
   6  *
   7  * For more information, see the README file.
   8  */
   9
  10
  11 /*
  12  * Functions to define the character set
  13  * and do things specific to the character set.
  14  */
  15
  16 #include "less.h"
  17 #if HAVE_LOCALE
  18 #include <locale.h>
  19 #include <ctype.h>
  20 #include <langinfo.h>
  21 #endif
  22
  23 #include "charset.h"
  24
  25 public int utf_mode = 0;
  26
  27 /*
  28  * Predefined character sets,
  29  * selected by the LESSCHARSET environment variable.
  30  */
  31 struct charset {
  32         char *name;
  33         int *p_flag;
  34         char *desc;
  35 } charsets[] = {
  36         { "ascii",              NULL,       "8bcccbcc18b95.b" },
  37         { "utf-8",              &utf_mode,  "8bcccbcc18b95.b126.bb" },
  38         { "iso8859",            NULL,       "8bcccbcc18b95.33b." },
  39         { "latin3",             NULL,       "8bcccbcc18b95.33b5.b8.b15.b4.b12.b18.b12.b." },
  40         { "arabic",             NULL,       "8bcccbcc18b95.33b.3b.7b2.13b.3b.b26.5b19.b" },
  41         { "greek",              NULL,       "8bcccbcc18b95.33b4.2b4.b3.b35.b44.b" },
  42         { "greek2005",          NULL,       "8bcccbcc18b95.33b14.b35.b44.b" },
  43         { "hebrew",             NULL,       "8bcccbcc18b95.33b.b29.32b28.2b2.b" },
  44         { "koi8-r",             NULL,       "8bcccbcc18b95.b." },
  45         { "KOI8-T",             NULL,       "8bcccbcc18b95.b8.b6.b8.b.b.5b7.3b4.b4.b3.b.b.3b." },
  46         { "georgianps",         NULL,       "8bcccbcc18b95.3b11.4b12.2b." },
  47         { "tcvn",               NULL,       "b..b...bcccbccbbb7.8b95.b48.5b." },
  48         { "TIS-620",            NULL,       "8bcccbcc18b95.b.4b.11b7.8b." },
  49         { "next",               NULL,       "8bcccbcc18b95.bb125.bb" },
  50         { "dos",                NULL,       "8bcccbcc12bc5b95.b." },
  51         { "windows-1251",       NULL,       "8bcccbcc12bc5b95.b24.b." },
  52         { "windows-1252",       NULL,       "8bcccbcc12bc5b95.b.b11.b.2b12.b." },
  53         { "windows-1255",       NULL,       "8bcccbcc12bc5b95.b.b8.b.5b9.b.4b." },
  54         { "ebcdic",             NULL,       "5bc6bcc7bcc41b.9b7.9b5.b..8b6.10b6.b9.7b9.8b8.17b3.3b9.7b9.8b8.6b10.b.b.b." },
  55         { "IBM-1047",           NULL,       "4cbcbc3b9cbccbccbb4c6bcc5b3cbbc4bc4bccbc191.b" },
  56         { NULL, NULL, NULL }
  57 };
  58
  59 /*
  60  * Support "locale charmap"/nl_langinfo(CODESET) values, as well as others.
  61  */
  62 struct cs_alias {
  63         char *name;
  64         char *oname;
  65 } cs_aliases[] = {
  66         { "UTF-8",              "utf-8" },
  67         { "ANSI_X3.4-1968",     "ascii" },
  68         { "US-ASCII",           "ascii" },
  69         { "latin1",             "iso8859" },
  70         { "ISO-8859-1",         "iso8859" },
  71         { "latin9",             "iso8859" },
  72         { "ISO-8859-15",        "iso8859" },
  73         { "latin2",             "iso8859" },
  74         { "ISO-8859-2",         "iso8859" },
  75         { "ISO-8859-3",         "latin3" },
  76         { "latin4",             "iso8859" },
  77         { "ISO-8859-4",         "iso8859" },
  78         { "cyrillic",           "iso8859" },
  79         { "ISO-8859-5",         "iso8859" },
  80         { "ISO-8859-6",         "arabic" },
  81         { "ISO-8859-7",         "greek" },
  82         { "IBM9005",            "greek2005" },
  83         { "ISO-8859-8",         "hebrew" },
  84         { "latin5",             "iso8859" },
  85         { "ISO-8859-9",         "iso8859" },
  86         { "latin6",             "iso8859" },
  87         { "ISO-8859-10",        "iso8859" },
  88         { "latin7",             "iso8859" },
  89         { "ISO-8859-13",        "iso8859" },
  90         { "latin8",             "iso8859" },
  91         { "ISO-8859-14",        "iso8859" },
  92         { "latin10",            "iso8859" },
  93         { "ISO-8859-16",        "iso8859" },
  94         { "IBM437",             "dos" },
  95         { "EBCDIC-US",          "ebcdic" },
  96         { "IBM1047",            "IBM-1047" },
  97         { "KOI8-R",             "koi8-r" },
  98         { "KOI8-U",             "koi8-r" },
  99         { "GEORGIAN-PS",        "georgianps" },
 100         { "TCVN5712-1",         "tcvn" },
 101         { "NEXTSTEP",           "next" },
 102         { "windows",            "windows-1252" }, /* backward compatibility */
 103         { "CP1251",             "windows-1251" },
 104         { "CP1252",             "windows-1252" },
 105         { "CP1255",             "windows-1255" },
 106         { NULL, NULL }
 107 };
 108
 109 #define IS_BINARY_CHAR  01
 110 #define IS_CONTROL_CHAR 02
 111
 112 static char chardef[256];
 113 static char *binfmt = NULL;
 114 static char *utfbinfmt = NULL;
 115 public int binattr = AT_STANDOUT;
 116
 117
 118 /*
 119  * Define a charset, given a description string.
 120  * The string consists of 256 letters,
 121  * one for each character in the charset.
 122  * If the string is shorter than 256 letters, missing letters
 123  * are taken to be identical to the last one.
 124  * A decimal number followed by a letter is taken to be a
 125  * repetition of the letter.
 126  *
 127  * Each letter is one of:
 128  *      . normal character
 129  *      b binary character
 130  *      c control character
 131  */
 132         static void
 133 ichardef(s)
 134         char *s;
 135 {
 136         register char *cp;
 137         register int n;
 138         register char v;
 139
 140         n = 0;
 141         v = 0;
 142         cp = chardef;
 143         while (*s != '\0')
 144         {
 145                 switch (*s++)
 146                 {
 147                 case '.':
 148                         v = 0;
 149                         break;
 150                 case 'c':
 151                         v = IS_CONTROL_CHAR;
 152                         break;
 153                 case 'b':
 154                         v = IS_BINARY_CHAR|IS_CONTROL_CHAR;
 155                         break;
 156
 157                 case '0': case '1': case '2': case '3': case '4':
 158                 case '5': case '6': case '7': case '8': case '9':
 159                         n = (10 * n) + (s[-1] - '0');
 160                         continue;
 161
 162                 default:
 163                         error("invalid chardef", NULL_PARG);
 164                         quit(QUIT_ERROR);
 165                         /*NOTREACHED*/
 166                 }
 167
 168                 do
 169                 {
 170                         if (cp >= chardef + sizeof(chardef))
 171                         {
 172                                 error("chardef longer than 256", NULL_PARG);
 173                                 quit(QUIT_ERROR);
 174                                 /*NOTREACHED*/
 175                         }
 176                         *cp++ = v;
 177                 } while (--n > 0);
 178                 n = 0;
 179         }
 180
 181         while (cp < chardef + sizeof(chardef))
 182                 *cp++ = v;
 183 }
 184
 185 /*
 186  * Define a charset, given a charset name.
 187  * The valid charset names are listed in the "charsets" array.
 188  */
 189         static int
 190 icharset(name, no_error)
 191         register char *name;
 192         int no_error;
 193 {
 194         register struct charset *p;
 195         register struct cs_alias *a;
 196
 197         if (name == NULL || *name == '\0')
 198                 return (0);
 199
 200         /* First see if the name is an alias. */
 201         for (a = cs_aliases;  a->name != NULL;  a++)
 202         {
 203                 if (strcmp(name, a->name) == 0)
 204                 {
 205                         name = a->oname;
 206                         break;
 207                 }
 208         }
 209
 210         for (p = charsets;  p->name != NULL;  p++)
 211         {
 212                 if (strcmp(name, p->name) == 0)
 213                 {
 214                         ichardef(p->desc);
 215                         if (p->p_flag != NULL)
 216                                 *(p->p_flag) = 1;
 217                         return (1);
 218                 }
 219         }
 220
 221         if (!no_error) {
 222                 error("invalid charset name", NULL_PARG);
 223                 quit(QUIT_ERROR);
 224         }
 225         return (0);
 226 }
 227
 228 #if HAVE_LOCALE
 229 /*
 230  * Define a charset, given a locale name.
 231  */
 232         static void
 233 ilocale()
 234 {
 235         register int c;
 236
 237         for (c = 0;  c < (int) sizeof(chardef);  c++)
 238         {
 239                 if (isprint(c))
 240                         chardef[c] = 0;
 241                 else if (iscntrl(c))
 242                         chardef[c] = IS_CONTROL_CHAR;
 243                 else
 244                         chardef[c] = IS_BINARY_CHAR|IS_CONTROL_CHAR;
 245         }
 246 }
 247 #endif
 248
 249 /*
 250  * Define the printing format for control (or binary utf) chars.
 251  */
 252         static void
 253 setbinfmt(s, fmtvarptr, default_fmt)
 254         char *s;
 255         char **fmtvarptr;
 256         char *default_fmt;
 257 {
 258         if (s && utf_mode)
 259         {
 260                 /* It would be too hard to account for width otherwise.  */
 261                 char *t = s;
 262                 while (*t)
 263                 {
 264                         if (*t < ' ' || *t > '~')
 265                         {
 266                                 s = default_fmt;
 267                                 goto attr;
 268                         }
 269                         t++;
 270                 }
 271         }
 272
 273         /* %n is evil */
 274         if (s == NULL || *s == '\0' ||
 275             (*s == '*' && (s[1] == '\0' || s[2] == '\0' || strchr(s + 2, 'n'))) ||
 276             (*s != '*' && strchr(s, 'n')))
 277                 s = default_fmt;
 278
 279         /*
 280          * Select the attributes if it starts with "*".
 281          */
 282  attr:
 283         if (*s == '*')
 284         {
 285                 switch (s[1])
 286                 {
 287                 case 'd':  binattr = AT_BOLD;      break;
 288                 case 'k':  binattr = AT_BLINK;     break;
 289                 case 's':  binattr = AT_STANDOUT;  break;
 290                 case 'u':  binattr = AT_UNDERLINE; break;
 291                 default:   binattr = AT_NORMAL;    break;
 292                 }
 293                 s += 2;
 294         }
 295         *fmtvarptr = s;
 296 }
 297
 298 /*
 299  *
 300  */
 301         static void
 302 set_charset()
 303 {
 304         char *s;
 305
 306         /*
 307          * See if environment variable LESSCHARSET is defined.
 308          */
 309         s = lgetenv("LESSCHARSET");
 310         if (icharset(s, 0))
 311                 return;
 312
 313         /*
 314          * LESSCHARSET is not defined: try LESSCHARDEF.
 315          */
 316         s = lgetenv("LESSCHARDEF");
 317         if (s != NULL && *s != '\0')
 318         {
 319                 ichardef(s);
 320                 return;
 321         }
 322
 323 #if HAVE_LOCALE
 324 #ifdef CODESET
 325         /*
 326          * Try using the codeset name as the charset name.
 327          */
 328         s = nl_langinfo(CODESET);
 329         if (icharset(s, 1))
 330                 return;
 331 #endif
 332 #endif
 333
 334 #if HAVE_STRSTR
 335         /*
 336          * Check whether LC_ALL, LC_CTYPE or LANG look like UTF-8 is used.
 337          */
 338         if ((s = lgetenv("LC_ALL")) != NULL ||
 339             (s = lgetenv("LC_CTYPE")) != NULL ||
 340             (s = lgetenv("LANG")) != NULL)
 341         {
 342                 if (   strstr(s, "UTF-8") != NULL || strstr(s, "utf-8") != NULL
 343                     || strstr(s, "UTF8")  != NULL || strstr(s, "utf8")  != NULL)
 344                         if (icharset("utf-8", 1))
 345                                 return;
 346         }
 347 #endif
 348
 349 #if HAVE_LOCALE
 350         /*
 351          * Get character definitions from locale functions,
 352          * rather than from predefined charset entry.
 353          */
 354         ilocale();
 355 #if MSDOS_COMPILER
 356         /*
 357          * Default to "dos".
 358          */
 359         (void) icharset("dos", 1);
 360 #else
 361         /*
 362          * Default to "latin1".
 363          */
 364         (void) icharset("latin1", 1);
 365 #endif
 366 #endif
 367 }
 368
 369 /*
 370  * Initialize charset data structures.
 371  */
 372         public void
 373 init_charset()
 374 {
 375         char *s;
 376
 377 #if HAVE_LOCALE
 378         setlocale(LC_ALL, "");
 379 #endif
 380
 381         set_charset();
 382
 383         s = lgetenv("LESSBINFMT");
 384         setbinfmt(s, &binfmt, "*s<%02X>");
 385
 386         s = lgetenv("LESSUTFBINFMT");
 387         setbinfmt(s, &utfbinfmt, "<U+%04lX>");
 388 }
 389
 390 /*
 391  * Is a given character a "binary" character?
 392  */
 393         public int
 394 binary_char(c)
 395         LWCHAR c;
 396 {
 397         if (utf_mode)
 398                 return (is_ubin_char(c));
 399         c &= 0377;
 400         return (chardef[c] & IS_BINARY_CHAR);
 401 }
 402
 403 /*
 404  * Is a given character a "control" character?
 405  */
 406         public int
 407 control_char(c)
 408         LWCHAR c;
 409 {
 410         c &= 0377;
 411         return (chardef[c] & IS_CONTROL_CHAR);
 412 }
 413
 414 /*
 415  * Return the printable form of a character.
 416  * For example, in the "ascii" charset '\3' is printed as "^C".
 417  */
 418         public char *
 419 prchar(c)
 420         LWCHAR c;
 421 {
 422         /* {{ This buffer can be overrun if LESSBINFMT is a long string. }} */
 423         static char buf[32];
 424
 425         c &= 0377;
 426         if ((c < 128 || !utf_mode) && !control_char(c))
 427                 SNPRINTF1(buf, sizeof(buf), "%c", (int) c);
 428         else if (c == ESC)
 429                 strcpy(buf, "ESC");
 430 #if IS_EBCDIC_HOST
 431         else if (!binary_char(c) && c < 64)
 432                 SNPRINTF1(buf, sizeof(buf), "^%c",
 433                 /*
 434                  * This array roughly inverts CONTROL() #defined in less.h,
 435                  * and should be kept in sync with CONTROL() and IBM-1047.
 436                  */
 437                 "@ABC.I.?...KLMNO"
 438                 "PQRS.JH.XY.."
 439                 "\\]^_"
 440                 "......W[.....EFG"
 441                 "..V....D....TU.Z"[c]);
 442 #else
 443         else if (c < 128 && !control_char(c ^ 0100))
 444                 SNPRINTF1(buf, sizeof(buf), "^%c", (int) (c ^ 0100));
 445 #endif
 446         else
 447                 SNPRINTF1(buf, sizeof(buf), binfmt, c);
 448         return (buf);
 449 }
 450
 451 /*
 452  * Return the printable form of a UTF-8 character.
 453  */
 454         public char *
 455 prutfchar(ch)
 456         LWCHAR ch;
 457 {
 458         static char buf[32];
 459
 460         if (ch == ESC)
 461                 strcpy(buf, "ESC");
 462         else if (ch < 128 && control_char(ch))
 463         {
 464                 if (!control_char(ch ^ 0100))
 465                         SNPRINTF1(buf, sizeof(buf), "^%c", ((char) ch) ^ 0100);
 466                 else
 467                         SNPRINTF1(buf, sizeof(buf), binfmt, (char) ch);
 468         } else if (is_ubin_char(ch))
 469         {
 470                 SNPRINTF1(buf, sizeof(buf), utfbinfmt, ch);
 471         } else
 472         {
 473                 char *p = buf;
 474                 if (ch >= 0x80000000)
 475                         ch = 0xFFFD; /* REPLACEMENT CHARACTER */
 476                 put_wchar(&p, ch);
 477                 *p = '\0';
 478         }
 479         return (buf);
 480 }
 481
 482 /*
 483  * Get the length of a UTF-8 character in bytes.
 484  */
 485         public int
 486 utf_len(ch)
 487         char ch;
 488 {
 489         if ((ch & 0x80) == 0)
 490                 return 1;
 491         if ((ch & 0xE0) == 0xC0)
 492                 return 2;
 493         if ((ch & 0xF0) == 0xE0)
 494                 return 3;
 495         if ((ch & 0xF8) == 0xF0)
 496                 return 4;
 497         if ((ch & 0xFC) == 0xF8)
 498                 return 5;
 499         if ((ch & 0xFE) == 0xFC)
 500                 return 6;
 501         /* Invalid UTF-8 encoding. */
 502         return 1;
 503 }
 504
 505 /*
 506  * Does the parameter point to the lead byte of a well-formed UTF-8 character?
 507  */
 508         public int
 509 is_utf8_well_formed(s, slen)
 510         unsigned char *s;
 511         int slen;
 512 {
 513         int i;
 514         int len;
 515
 516         if (IS_UTF8_INVALID(s[0]))
 517                 return (0);
 518
 519         len = utf_len((char) s[0]);
 520         if (len > slen)
 521                 return (0);
 522         if (len == 1)
 523                 return (1);
 524         if (len == 2)
 525         {
 526                 if (s[0] < 0xC2)
 527                     return (0);
 528         } else
 529         {
 530                 unsigned char mask;
 531                 mask = (~((1 << (8-len)) - 1)) & 0xFF;
 532                 if (s[0] == mask && (s[1] & mask) == 0x80)
 533                         return (0);
 534         }
 535
 536         for (i = 1;  i < len;  i++)
 537                 if (!IS_UTF8_TRAIL(s[i]))
 538                         return (0);
 539         return (1);
 540 }
 541
 542 /*
 543  * Return number of invalid UTF-8 sequences found in a buffer.
 544  */
 545         public int
 546 utf_bin_count(data, len)
 547         unsigned char *data;
 548         int len;
 549 {
 550         int bin_count = 0;
 551         while (len > 0)
 552         {
 553                 if (is_utf8_well_formed(data, len))
 554                 {
 555                         int clen = utf_len(*data);
 556                         data += clen;
 557                         len -= clen;
 558                 } else
 559                 {
 560                         /* Skip to next lead byte. */
 561                         bin_count++;
 562                         do {
 563                                 ++data;
 564                                 --len;
 565                         } while (len > 0 && !IS_UTF8_LEAD(*data));
 566                 }
 567         }
 568         return (bin_count);
 569 }
 570
 571 /*
 572  * Get the value of a UTF-8 character.
 573  */
 574         public LWCHAR
 575 get_wchar(p)
 576         char *p;
 577 {
 578         switch (utf_len(p[0]))
 579         {
 580         case 1:
 581         default:
 582                 /* 0xxxxxxx */
 583                 return (LWCHAR)
 584                         (p[0] & 0xFF);
 585         case 2:
 586                 /* 110xxxxx 10xxxxxx */
 587                 return (LWCHAR) (
 588                         ((p[0] & 0x1F) << 6) |
 589                         (p[1] & 0x3F));
 590         case 3:
 591                 /* 1110xxxx 10xxxxxx 10xxxxxx */
 592                 return (LWCHAR) (
 593                         ((p[0] & 0x0F) << 12) |
 594                         ((p[1] & 0x3F) << 6) |
 595                         (p[2] & 0x3F));
 596         case 4:
 597                 /* 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx */
 598                 return (LWCHAR) (
 599                         ((p[0] & 0x07) << 18) |
 600                         ((p[1] & 0x3F) << 12) |
 601                         ((p[2] & 0x3F) << 6) |
 602                         (p[3] & 0x3F));
 603         case 5:
 604                 /* 111110xx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx */
 605                 return (LWCHAR) (
 606                         ((p[0] & 0x03) << 24) |
 607                         ((p[1] & 0x3F) << 18) |
 608                         ((p[2] & 0x3F) << 12) |
 609                         ((p[3] & 0x3F) << 6) |
 610                         (p[4] & 0x3F));
 611         case 6:
 612                 /* 1111110x 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx */
 613                 return (LWCHAR) (
 614                         ((p[0] & 0x01) << 30) |
 615                         ((p[1] & 0x3F) << 24) |
 616                         ((p[2] & 0x3F) << 18) |
 617                         ((p[3] & 0x3F) << 12) |
 618                         ((p[4] & 0x3F) << 6) |
 619                         (p[5] & 0x3F));
 620         }
 621 }
 622
 623 /*
 624  * Store a character into a UTF-8 string.
 625  */
 626         public void
 627 put_wchar(pp, ch)
 628         char **pp;
 629         LWCHAR ch;
 630 {
 631         if (!utf_mode || ch < 0x80)
 632         {
 633                 /* 0xxxxxxx */
 634                 *(*pp)++ = (char) ch;
 635         } else if (ch < 0x800)
 636         {
 637                 /* 110xxxxx 10xxxxxx */
 638                 *(*pp)++ = (char) (0xC0 | ((ch >> 6) & 0x1F));
 639                 *(*pp)++ = (char) (0x80 | (ch & 0x3F));
 640         } else if (ch < 0x10000)
 641         {
 642                 /* 1110xxxx 10xxxxxx 10xxxxxx */
 643                 *(*pp)++ = (char) (0xE0 | ((ch >> 12) & 0x0F));
 644                 *(*pp)++ = (char) (0x80 | ((ch >> 6) & 0x3F));
 645                 *(*pp)++ = (char) (0x80 | (ch & 0x3F));
 646         } else if (ch < 0x200000)
 647         {
 648                 /* 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx */
 649                 *(*pp)++ = (char) (0xF0 | ((ch >> 18) & 0x07));
 650                 *(*pp)++ = (char) (0x80 | ((ch >> 12) & 0x3F));
 651                 *(*pp)++ = (char) (0x80 | ((ch >> 6) & 0x3F));
 652                 *(*pp)++ = (char) (0x80 | (ch & 0x3F));
 653         } else if (ch < 0x4000000)
 654         {
 655                 /* 111110xx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx */
 656                 *(*pp)++ = (char) (0xF0 | ((ch >> 24) & 0x03));
 657                 *(*pp)++ = (char) (0x80 | ((ch >> 18) & 0x3F));
 658                 *(*pp)++ = (char) (0x80 | ((ch >> 12) & 0x3F));
 659                 *(*pp)++ = (char) (0x80 | ((ch >> 6) & 0x3F));
 660                 *(*pp)++ = (char) (0x80 | (ch & 0x3F));
 661         } else
 662         {
 663                 /* 1111110x 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx */
 664                 *(*pp)++ = (char) (0xF0 | ((ch >> 30) & 0x01));
 665                 *(*pp)++ = (char) (0x80 | ((ch >> 24) & 0x3F));
 666                 *(*pp)++ = (char) (0x80 | ((ch >> 18) & 0x3F));
 667                 *(*pp)++ = (char) (0x80 | ((ch >> 12) & 0x3F));
 668                 *(*pp)++ = (char) (0x80 | ((ch >> 6) & 0x3F));
 669                 *(*pp)++ = (char) (0x80 | (ch & 0x3F));
 670         }
 671 }
 672
 673 /*
 674  * Step forward or backward one character in a string.
 675  */
 676         public LWCHAR
 677 step_char(pp, dir, limit)
 678         char **pp;
 679         signed int dir;
 680         char *limit;
 681 {
 682         LWCHAR ch;
 683         int len;
 684         char *p = *pp;
 685
 686         if (!utf_mode)
 687         {
 688                 /* It's easy if chars are one byte. */
 689                 if (dir > 0)
 690                         ch = (LWCHAR) ((p < limit) ? *p++ : 0);
 691                 else
 692                         ch = (LWCHAR) ((p > limit) ? *--p : 0);
 693         } else if (dir > 0)
 694         {
 695                 len = utf_len(*p);
 696                 if (p + len > limit)
 697                 {
 698                         ch = 0;
 699                         p = limit;
 700                 } else
 701                 {
 702                         ch = get_wchar(p);
 703                         p += len;
 704                 }
 705         } else
 706         {
 707                 while (p > limit && IS_UTF8_TRAIL(p[-1]))
 708                         p--;
 709                 if (p > limit)
 710                         ch = get_wchar(--p);
 711                 else
 712                         ch = 0;
 713         }
 714         *pp = p;
 715         return ch;
 716 }
 717
 718 /*
 719  * Unicode characters data
 720  * Actual data is in the generated *.uni files.
 721  */
 722
 723 #define DECLARE_RANGE_TABLE_START(name) \
 724     static struct wchar_range name##_array[] = {
 725 #define DECLARE_RANGE_TABLE_END(name) \
 726     }; struct wchar_range_table name##_table = { name##_array, sizeof(name##_array)/sizeof(*name##_array) };
 727
 728 DECLARE_RANGE_TABLE_START(compose)
 729 #include "compose.uni"
 730 DECLARE_RANGE_TABLE_END(compose)
 731
 732 DECLARE_RANGE_TABLE_START(ubin)
 733 #include "ubin.uni"
 734 DECLARE_RANGE_TABLE_END(ubin)
 735
 736 DECLARE_RANGE_TABLE_START(wide)
 737 #include "wide.uni"
 738 DECLARE_RANGE_TABLE_END(wide)
 739
 740 /* comb_table is special pairs, not ranges. */
 741 static struct wchar_range comb_table[] = {
 742         {0x0644,0x0622}, {0x0644,0x0623}, {0x0644,0x0625}, {0x0644,0x0627},
 743 };
 744
 745
 746         static int
 747 is_in_table(ch, table)
 748         LWCHAR ch;
 749         struct wchar_range_table *table;
 750 {
 751         int hi;
 752         int lo;
 753
 754         /* Binary search in the table. */
 755         if (ch < table->table[0].first)
 756                 return 0;
 757         lo = 0;
 758         hi = table->count - 1;
 759         while (lo <= hi)
 760         {
 761                 int mid = (lo + hi) / 2;
 762                 if (ch > table->table[mid].last)
 763                         lo = mid + 1;
 764                 else if (ch < table->table[mid].first)
 765                         hi = mid - 1;
 766                 else
 767                         return 1;
 768         }
 769         return 0;
 770 }
 771
 772 /*
 773  * Is a character a UTF-8 composing character?
 774  * If a composing character follows any char, the two combine into one glyph.
 775  */
 776         public int
 777 is_composing_char(ch)
 778         LWCHAR ch;
 779 {
 780         return is_in_table(ch, &compose_table);
 781 }
 782
 783 /*
 784  * Should this UTF-8 character be treated as binary?
 785  */
 786         public int
 787 is_ubin_char(ch)
 788         LWCHAR ch;
 789 {
 790         return is_in_table(ch, &ubin_table);
 791 }
 792
 793 /*
 794  * Is this a double width UTF-8 character?
 795  */
 796         public int
 797 is_wide_char(ch)
 798         LWCHAR ch;
 799 {
 800         return is_in_table(ch, &wide_table);
 801 }
 802
 803 /*
 804  * Is a character a UTF-8 combining character?
 805  * A combining char acts like an ordinary char, but if it follows
 806  * a specific char (not any char), the two combine into one glyph.
 807  */
 808         public int
 809 is_combining_char(ch1, ch2)
 810         LWCHAR ch1;
 811         LWCHAR ch2;
 812 {
 813         /* The table is small; use linear search. */
 814         int i;
 815         for (i = 0;  i < sizeof(comb_table)/sizeof(*comb_table);  i++)
 816         {
 817                 if (ch1 == comb_table[i].first &&
 818                     ch2 == comb_table[i].last)
 819                         return 1;
 820         }
 821         return 0;
 822 }
 823