1/* 2 * compat.c - compatibility routines for the deprived 3 * 4 * This file is part of zsh, the Z shell. 5 * 6 * Copyright (c) 1992-1997 Paul Falstad 7 * All rights reserved. 8 * 9 * Permission is hereby granted, without written agreement and without 10 * license or royalty fees, to use, copy, modify, and distribute this 11 * software and to distribute modified versions of this software for any 12 * purpose, provided that the above copyright notice and the following 13 * two paragraphs appear in all copies of this software. 14 * 15 * In no event shall Paul Falstad or the Zsh Development Group be liable 16 * to any party for direct, indirect, special, incidental, or consequential 17 * damages arising out of the use of this software and its documentation, 18 * even if Paul Falstad and the Zsh Development Group have been advised of 19 * the possibility of such damage. 20 * 21 * Paul Falstad and the Zsh Development Group specifically disclaim any 22 * warranties, including, but not limited to, the implied warranties of 23 * merchantability and fitness for a particular purpose. The software 24 * provided hereunder is on an "as is" basis, and Paul Falstad and the 25 * Zsh Development Group have no obligation to provide maintenance, 26 * support, updates, enhancements, or modifications. 27 * 28 */ 29 30#include "zsh.mdh" 31#include "compat.pro" 32 33/* Return pointer to first occurence of string t * 34 * in string s. Return NULL if not present. */ 35 36#ifndef HAVE_STRSTR 37char * 38strstr(const char *s, const char *t) 39{ 40 char *p1, *p2; 41 42 for (; *s; s++) { 43 for (p1 = s, p2 = t; *p2; p1++, p2++) 44 if (*p1 != *p2) 45 break; 46 if (!*p2) 47 return (char *)s; 48 } 49 return NULL; 50} 51#endif 52 53 54#ifndef HAVE_GETHOSTNAME 55int 56gethostname(char *name, size_t namelen) 57{ 58 struct utsname uts; 59 60 uname(&uts); 61 if(strlen(uts.nodename) >= namelen) { 62 errno = EINVAL; 63 return -1; 64 } 65 strcpy(name, uts.nodename); 66 return 0; 67} 68#endif 69 70 71#ifndef HAVE_GETTIMEOFDAY 72int 73gettimeofday(struct timeval *tv, struct timezone *tz) 74{ 75 tv->tv_usec = 0; 76 tv->tv_sec = (long)time((time_t) 0); 77 return 0; 78} 79#endif 80 81 82/* compute the difference between two calendar times */ 83 84#ifndef HAVE_DIFFTIME 85double 86difftime(time_t t2, time_t t1) 87{ 88 return ((double)t2 - (double)t1); 89} 90#endif 91 92 93#ifndef HAVE_STRERROR 94extern char *sys_errlist[]; 95 96/* Get error message string associated with a particular * 97 * error number, and returns a pointer to that string. * 98 * This is not a particularly robust version of strerror. */ 99 100char * 101strerror(int errnum) 102{ 103 return (sys_errlist[errnum]); 104} 105#endif 106 107 108#if 0 109/* pathconf(_PC_PATH_MAX) is not currently useful to zsh. The value * 110 * returned varies depending on a number of factors, e.g. the amount * 111 * of memory available to the operating system at a given time; thus * 112 * it can't be used for buffer allocation, or even as an indication * 113 * of whether an attempt to use or create a given pathname may fail * 114 * at any future time. * 115 * * 116 * The call is also permitted to fail if the argument path is not an * 117 * existing directory, so even to make sense of that one must search * 118 * for a valid directory somewhere in the path and adjust. Even if * 119 * it succeeds, the return value is relative to the input directory, * 120 * and therefore potentially relative to the length of the shortest * 121 * path either to that directory or to our working directory. * 122 * * 123 * Finally, see the note below for glibc; detection of pathconf() is * 124 * not by itself an indication that it works reliably. */ 125 126/* The documentation for pathconf() says something like: * 127 * The limit is returned, if one exists. If the system does * 128 * not have a limit for the requested resource, -1 is * 129 * returned, and errno is unchanged. If there is an error, * 130 * -1 is returned, and errno is set to reflect the nature of * 131 * the error. * 132 * * 133 * System calls are not permitted to set errno to 0; but we must (or * 134 * some other flag value) in order to determine that the resource is * 135 * unlimited. What use is leaving errno unchanged? Instead, define * 136 * a wrapper that resets errno to 0 and returns 0 for "the system * 137 * does not have a limit," so that -1 always means a real error. */ 138 139/**/ 140mod_export long 141zpathmax(char *dir) 142{ 143#ifdef HAVE_PATHCONF 144 long pathmax; 145 146 errno = 0; 147 if ((pathmax = pathconf(dir, _PC_PATH_MAX)) >= 0) { 148 /* Some versions of glibc pathconf return a hardwired value! */ 149 return pathmax; 150 } else if (errno == EINVAL || errno == ENOENT || errno == ENOTDIR) { 151 /* Work backward to find a directory, until we run out of path. */ 152 char *tail = strrchr(dir, '/'); 153 while (tail > dir && tail[-1] == '/') 154 --tail; 155 if (tail > dir) { 156 *tail = 0; 157 pathmax = zpathmax(dir); 158 *tail = '/'; 159 } else { 160 errno = 0; 161 if (tail) 162 pathmax = pathconf("/", _PC_PATH_MAX); 163 else 164 pathmax = pathconf(".", _PC_PATH_MAX); 165 } 166 if (pathmax > 0) { 167 long taillen = (tail ? strlen(tail) : (strlen(dir) + 1)); 168 if (taillen < pathmax) 169 return pathmax - taillen; 170 else 171 errno = ENAMETOOLONG; 172 } 173 } 174 if (errno) 175 return -1; 176 else 177 return 0; /* pathmax should be considered unlimited */ 178#else 179 long dirlen = strlen(dir); 180 181 /* The following is wrong if dir is not an absolute path. */ 182 return ((long) ((dirlen >= PATH_MAX) ? 183 ((errno = ENAMETOOLONG), -1) : 184 ((errno = 0), PATH_MAX - dirlen))); 185#endif 186} 187#endif /* 0 */ 188 189#ifdef HAVE_SYSCONF 190/* 191 * This is replaced by a macro from system.h if not HAVE_SYSCONF. 192 * 0 is returned by sysconf if _SC_OPEN_MAX is unavailable; 193 * -1 is returned on error 194 * 195 * Neither of these should happen, but resort to OPEN_MAX rather 196 * than return 0 or -1 just in case. 197 * 198 * We'll limit the open maximum to ZSH_INITIAL_OPEN_MAX to 199 * avoid probing ridiculous numbers of file descriptors. 200 */ 201 202/**/ 203mod_export long 204zopenmax(void) 205{ 206 long openmax; 207 208 if ((openmax = sysconf(_SC_OPEN_MAX)) < 1) { 209 openmax = OPEN_MAX; 210 } else if (openmax > OPEN_MAX) { 211 /* On some systems, "limit descriptors unlimited" or the * 212 * equivalent will set openmax to a huge number. Unless * 213 * there actually is a file descriptor > OPEN_MAX already * 214 * open, nothing in zsh requires the true maximum, and in * 215 * fact it causes inefficiency elsewhere if we report it. * 216 * So, report the maximum of OPEN_MAX or the largest open * 217 * descriptor (is there a better way to find that?). */ 218 long i, j = OPEN_MAX; 219 if (openmax > ZSH_INITIAL_OPEN_MAX) 220 openmax = ZSH_INITIAL_OPEN_MAX; 221 for (i = j; i < openmax; i += (errno != EINTR)) { 222 errno = 0; 223 if (fcntl(i, F_GETFL, 0) < 0 && 224 (errno == EBADF || errno == EINTR)) 225 continue; 226 j = i; 227 } 228 openmax = j; 229 } 230 231 return (max_zsh_fd > openmax) ? max_zsh_fd : openmax; 232} 233#endif 234 235/* 236 * Rationalise the current directory, returning the string. 237 * 238 * If "d" is not NULL, it is used to store information about the 239 * directory. The returned name is also present in d->dirname and is in 240 * permanently allocated memory. The handling of this case depends on 241 * whether the fchdir() system call is available; if it is, it is assumed 242 * the caller is able to restore the current directory. On successfully 243 * identifying the directory the function returns immediately rather 244 * than ascending the hierarchy. 245 * 246 * If "d" is NULL, no assumption about the caller's behaviour is 247 * made. The returned string is in heap memory. This case is 248 * always handled by changing directory up the hierarchy. 249 * 250 * On Cygwin or other systems where USE_GETCWD is defined (at the 251 * time of writing only QNX), we skip all the above and use the 252 * getcwd() system call. 253 */ 254 255/**/ 256mod_export char * 257zgetdir(struct dirsav *d) 258{ 259 char nbuf[PATH_MAX+3]; 260 char *buf; 261 int bufsiz, pos; 262 struct stat sbuf; 263 ino_t pino; 264 dev_t pdev; 265#if !defined(__CYGWIN__) && !defined(USE_GETCWD) 266 struct dirent *de; 267 DIR *dir; 268 dev_t dev; 269 ino_t ino; 270 int len; 271#endif 272 273 buf = zhalloc(bufsiz = PATH_MAX); 274 pos = bufsiz - 1; 275 buf[pos] = '\0'; 276 strcpy(nbuf, "../"); 277 if (stat(".", &sbuf) < 0) { 278 return NULL; 279 } 280 281 /* Record the initial inode and device */ 282 pino = sbuf.st_ino; 283 pdev = sbuf.st_dev; 284 if (d) 285 d->ino = pino, d->dev = pdev; 286#if !defined(__CYGWIN__) && !defined(USE_GETCWD) 287#ifdef HAVE_FCHDIR 288 else 289#endif 290 holdintr(); 291 292 for (;;) { 293 /* Examine the parent of the current directory. */ 294 if (stat("..", &sbuf) < 0) 295 break; 296 297 /* Inode and device of curtent directory */ 298 ino = pino; 299 dev = pdev; 300 /* Inode and device of current directory's parent */ 301 pino = sbuf.st_ino; 302 pdev = sbuf.st_dev; 303 304 /* If they're the same, we've reached the root directory. */ 305 if (ino == pino && dev == pdev) { 306 if (!buf[pos]) 307 buf[--pos] = '/'; 308 if (d) { 309#ifndef HAVE_FCHDIR 310 zchdir(buf + pos); 311 noholdintr(); 312#endif 313 return d->dirname = ztrdup(buf + pos); 314 } 315 zchdir(buf + pos); 316 noholdintr(); 317 return buf + pos; 318 } 319 320 /* Search the parent for the current directory. */ 321 if (!(dir = opendir(".."))) 322 break; 323 324 while ((de = readdir(dir))) { 325 char *fn = de->d_name; 326 /* Ignore `.' and `..'. */ 327 if (fn[0] == '.' && 328 (fn[1] == '\0' || 329 (fn[1] == '.' && fn[2] == '\0'))) 330 continue; 331#ifdef HAVE_STRUCT_DIRENT_D_STAT 332 if(de->d_stat.st_dev == dev && de->d_stat.st_ino == ino) { 333 /* Found the directory we're currently in */ 334 strncpy(nbuf + 3, fn, PATH_MAX); 335 break; 336 } 337#else /* !HAVE_STRUCT_DIRENT_D_STAT */ 338# ifdef HAVE_STRUCT_DIRENT_D_INO 339 if (dev != pdev || (ino_t) de->d_ino == ino) 340# endif /* HAVE_STRUCT_DIRENT_D_INO */ 341 { 342 /* Maybe found directory, need to check device & inode */ 343 strncpy(nbuf + 3, fn, PATH_MAX); 344 lstat(nbuf, &sbuf); 345 if (sbuf.st_dev == dev && sbuf.st_ino == ino) 346 break; 347 } 348#endif /* !HAVE_STRUCT_DIRENT_D_STAT */ 349 } 350 closedir(dir); 351 if (!de) 352 break; /* Not found */ 353 /* 354 * We get the "/" free just by copying from nbuf+2 instead 355 * of nbuf+3, which is where we copied the path component. 356 * This means buf[pos] is always a "/". 357 */ 358 len = strlen(nbuf + 2); 359 pos -= len; 360 while (pos <= 1) { 361 char *newbuf = zhalloc(2*bufsiz); 362 memcpy(newbuf + bufsiz, buf, bufsiz); 363 buf = newbuf; 364 pos += bufsiz; 365 bufsiz *= 2; 366 } 367 memcpy(buf + pos, nbuf + 2, len); 368#ifdef HAVE_FCHDIR 369 if (d) 370 return d->dirname = ztrdup(buf + pos + 1); 371#endif 372 if (chdir("..")) 373 break; 374 } 375 376 /* 377 * Fix up the directory, if necessary. 378 * We're changing back down the hierarchy, ignore the 379 * "/" at buf[pos]. 380 */ 381 if (d) { 382#ifndef HAVE_FCHDIR 383 if (buf[pos]) 384 zchdir(buf + pos + 1); 385 noholdintr(); 386#endif 387 return NULL; 388 } 389 390 if (buf[pos]) 391 zchdir(buf + pos + 1); 392 noholdintr(); 393 394#else /* __CYGWIN__, USE_GETCWD cases */ 395 396 if (!getcwd(buf, bufsiz)) { 397 if (d) { 398 return NULL; 399 } 400 } else { 401 if (d) { 402 return d->dirname = ztrdup(buf); 403 } 404 return buf; 405 } 406#endif 407 408 /* 409 * Something bad happened. 410 * This has been seen when inside a special directory, 411 * such as the Netapp .snapshot directory, that doesn't 412 * appear as a directory entry in the parent directory. 413 * We'll just need our best guess. 414 * 415 * We only get here from zgetcwd(); let that fall back to pwd. 416 */ 417 418 return NULL; 419} 420 421/* 422 * Try to find the current directory. 423 * If we couldn't work it out internally, fall back to getcwd(). 424 * If it fails, fall back to pwd; if zgetcwd() is being used 425 * to set pwd, pwd should be NULL and we just return ".". 426 */ 427 428/**/ 429char * 430zgetcwd(void) 431{ 432 char *ret = zgetdir(NULL); 433#ifdef HAVE_GETCWD 434 if (!ret) { 435#ifdef GETCWD_CALLS_MALLOC 436 char *cwd = getcwd(NULL, 0); 437 if (cwd) { 438 ret = dupstring(cwd); 439 free(cwd); 440 } 441#else 442 char *cwdbuf = zalloc(PATH_MAX); 443 ret = getcwd(cwdbuf, PATH_MAX); 444 if (ret) 445 ret = dupstring(ret); 446 free(cwdbuf); 447#endif /* GETCWD_CALLS_MALLOC */ 448 } 449#endif /* HAVE_GETCWD */ 450 if (!ret) 451 ret = pwd; 452 if (!ret) 453 ret = dupstring("."); 454 return ret; 455} 456 457/* chdir with arbitrary long pathname. Returns 0 on success, -1 on normal * 458 * failure and -2 when chdir failed and the current directory is lost. */ 459 460/**/ 461mod_export int 462zchdir(char *dir) 463{ 464 char *s; 465 int currdir = -2; 466 467 for (;;) { 468 if (!*dir || chdir(dir) == 0) { 469#ifdef HAVE_FCHDIR 470 if (currdir >= 0) 471 close(currdir); 472#endif 473 return 0; 474 } 475 if ((errno != ENAMETOOLONG && errno != ENOMEM) || 476 strlen(dir) < PATH_MAX) 477 break; 478 for (s = dir + PATH_MAX - 1; s > dir && *s != '/'; s--) 479 ; 480 if (s == dir) 481 break; 482#ifdef HAVE_FCHDIR 483 if (currdir == -2) 484 currdir = open(".", O_RDONLY|O_NOCTTY); 485#endif 486 *s = '\0'; 487 if (chdir(dir) < 0) { 488 *s = '/'; 489 break; 490 } 491#ifndef HAVE_FCHDIR 492 currdir = -1; 493#endif 494 *s = '/'; 495 while (*++s == '/') 496 ; 497 dir = s; 498 } 499#ifdef HAVE_FCHDIR 500 if (currdir >= 0) { 501 if (fchdir(currdir) < 0) { 502 close(currdir); 503 return -2; 504 } 505 close(currdir); 506 return -1; 507 } 508#endif 509 return currdir == -2 ? -1 : -2; 510} 511 512/* 513 * How to print out a 64 bit integer. This isn't needed (1) if longs 514 * are 64 bit, since ordinary %ld will work (2) if we couldn't find a 515 * 64 bit type anyway. 516 */ 517/**/ 518#ifdef ZSH_64_BIT_TYPE 519/**/ 520mod_export char * 521output64(zlong val) 522{ 523 static char llbuf[DIGBUFSIZE]; 524 convbase(llbuf, val, 0); 525 return llbuf; 526} 527/**/ 528#endif /* ZSH_64_BIT_TYPE */ 529 530#ifndef HAVE_STRTOUL 531 532/* 533 * Copyright (c) 1990, 1993 534 * The Regents of the University of California. All rights reserved. 535 * 536 * Redistribution and use in source and binary forms, with or without 537 * modification, are permitted provided that the following conditions 538 * are met: 539 * 1. Redistributions of source code must retain the above copyright 540 * notice, this list of conditions and the following disclaimer. 541 * 2. Redistributions in binary form must reproduce the above copyright 542 * notice, this list of conditions and the following disclaimer in the 543 * documentation and/or other materials provided with the distribution. 544 * 3. Neither the name of the University nor the names of its contributors 545 * may be used to endorse or promote products derived from this software 546 * without specific prior written permission. 547 * 548 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 549 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 550 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 551 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 552 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 553 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 554 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 555 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 556 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 557 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 558 * SUCH DAMAGE. 559 */ 560 561/* 562 * Convert a string to an unsigned long integer. 563 * 564 * Ignores `locale' stuff. Assumes that the upper and lower case 565 * alphabets and digits are each contiguous. 566 */ 567unsigned long 568strtoul(nptr, endptr, base) 569 const char *nptr; 570 char **endptr; 571 int base; 572{ 573 const char *s; 574 unsigned long acc, cutoff; 575 int c; 576 int neg, any, cutlim; 577 578 /* endptr may be NULL */ 579 580 s = nptr; 581 do { 582 c = (unsigned char) *s++; 583 } while (isspace(c)); 584 if (c == '-') { 585 neg = 1; 586 c = *s++; 587 } else { 588 neg = 0; 589 if (c == '+') 590 c = *s++; 591 } 592 if ((base == 0 || base == 16) && 593 c == '0' && (*s == 'x' || *s == 'X')) { 594 c = s[1]; 595 s += 2; 596 base = 16; 597 } 598 if (base == 0) 599 base = c == '0' ? 8 : 10; 600 601 cutoff = ULONG_MAX / (unsigned long)base; 602 cutlim = (int)(ULONG_MAX % (unsigned long)base); 603 for (acc = 0, any = 0;; c = (unsigned char) *s++) { 604 if (isdigit(c)) 605 c -= '0'; 606 else if (isalpha(c)) { 607 c -= isupper(c) ? 'A' - 10 : 'a' - 10; 608 } else 609 break; 610 if (c >= base) 611 break; 612 if (any < 0) 613 continue; 614 if (acc > cutoff || (acc == cutoff && c > cutlim)) { 615 any = -1; 616 acc = ULONG_MAX; 617 errno = ERANGE; 618 } else { 619 any = 1; 620 acc *= (unsigned long)base; 621 acc += c; 622 } 623 } 624 if (neg && any > 0) 625 acc = -acc; 626 if (endptr != NULL) 627 *endptr = any ? s - 1 : nptr; 628 return (acc); 629} 630#endif /* HAVE_STRTOUL */ 631 632/**/ 633#if defined(BROKEN_WCWIDTH) && (defined(__STDC_ISO_10646__) || defined(__APPLE__)) 634 635/* 636 * This is an implementation of wcwidth() and wcswidth() (defined in 637 * IEEE Std 1002.1-2001) for Unicode. 638 * 639 * http://www.opengroup.org/onlinepubs/007904975/functions/wcwidth.html 640 * http://www.opengroup.org/onlinepubs/007904975/functions/wcswidth.html 641 * 642 * In fixed-width output devices, Latin characters all occupy a single 643 * "cell" position of equal width, whereas ideographic CJK characters 644 * occupy two such cells. Interoperability between terminal-line 645 * applications and (teletype-style) character terminals using the 646 * UTF-8 encoding requires agreement on which character should advance 647 * the cursor by how many cell positions. No established formal 648 * standards exist at present on which Unicode character shall occupy 649 * how many cell positions on character terminals. These routines are 650 * a first attempt of defining such behavior based on simple rules 651 * applied to data provided by the Unicode Consortium. 652 * 653 * For some graphical characters, the Unicode standard explicitly 654 * defines a character-cell width via the definition of the East Asian 655 * FullWidth (F), Wide (W), Half-width (H), and Narrow (Na) classes. 656 * In all these cases, there is no ambiguity about which width a 657 * terminal shall use. For characters in the East Asian Ambiguous (A) 658 * class, the width choice depends purely on a preference of backward 659 * compatibility with either historic CJK or Western practice. 660 * Choosing single-width for these characters is easy to justify as 661 * the appropriate long-term solution, as the CJK practice of 662 * displaying these characters as double-width comes from historic 663 * implementation simplicity (8-bit encoded characters were displayed 664 * single-width and 16-bit ones double-width, even for Greek, 665 * Cyrillic, etc.) and not any typographic considerations. 666 * 667 * Much less clear is the choice of width for the Not East Asian 668 * (Neutral) class. Existing practice does not dictate a width for any 669 * of these characters. It would nevertheless make sense 670 * typographically to allocate two character cells to characters such 671 * as for instance EM SPACE or VOLUME INTEGRAL, which cannot be 672 * represented adequately with a single-width glyph. The following 673 * routines at present merely assign a single-cell width to all 674 * neutral characters, in the interest of simplicity. This is not 675 * entirely satisfactory and should be reconsidered before 676 * establishing a formal standard in this area. At the moment, the 677 * decision which Not East Asian (Neutral) characters should be 678 * represented by double-width glyphs cannot yet be answered by 679 * applying a simple rule from the Unicode database content. Setting 680 * up a proper standard for the behavior of UTF-8 character terminals 681 * will require a careful analysis not only of each Unicode character, 682 * but also of each presentation form, something the author of these 683 * routines has avoided to do so far. 684 * 685 * http://www.unicode.org/unicode/reports/tr11/ 686 * 687 * Markus Kuhn -- 2007-05-26 (Unicode 5.0) 688 * 689 * Permission to use, copy, modify, and distribute this software 690 * for any purpose and without fee is hereby granted. The author 691 * disclaims all warranties with regard to this software. 692 * 693 * Latest version: http://www.cl.cam.ac.uk/~mgk25/ucs/wcwidth.c 694 */ 695 696struct interval { 697 int first; 698 int last; 699}; 700 701/* auxiliary function for binary search in interval table */ 702static int bisearch(wchar_t ucs, const struct interval *table, int max) { 703 int min = 0; 704 int mid; 705 706 if (ucs < table[0].first || ucs > table[max].last) 707 return 0; 708 while (max >= min) { 709 mid = (min + max) / 2; 710 if (ucs > table[mid].last) 711 min = mid + 1; 712 else if (ucs < table[mid].first) 713 max = mid - 1; 714 else 715 return 1; 716 } 717 718 return 0; 719} 720 721 722/* The following two functions define the column width of an ISO 10646 723 * character as follows: 724 * 725 * - The null character (U+0000) has a column width of 0. 726 * 727 * - Other C0/C1 control characters and DEL will lead to a return 728 * value of -1. 729 * 730 * - Non-spacing and enclosing combining characters (general 731 * category code Mn or Me in the Unicode database) have a 732 * column width of 0. 733 * 734 * - SOFT HYPHEN (U+00AD) has a column width of 1. 735 * 736 * - Other format characters (general category code Cf in the Unicode 737 * database) and ZERO WIDTH SPACE (U+200B) have a column width of 0. 738 * 739 * - Hangul Jamo medial vowels and final consonants (U+1160-U+11FF) 740 * have a column width of 0. 741 * 742 * - Spacing characters in the East Asian Wide (W) or East Asian 743 * Full-width (F) category as defined in Unicode Technical 744 * Report #11 have a column width of 2. 745 * 746 * - All remaining characters (including all printable 747 * ISO 8859-1 and WGL4 characters, Unicode control characters, 748 * etc.) have a column width of 1. 749 * 750 * This implementation assumes that wchar_t characters are encoded 751 * in ISO 10646. 752 */ 753 754/**/ 755int 756mk_wcwidth(wchar_t ucs) 757{ 758 /* sorted list of non-overlapping intervals of non-spacing characters */ 759 /* generated by "uniset +cat=Me +cat=Mn +cat=Cf -00AD +1160-11FF +200B c" */ 760 static const struct interval combining[] = { 761 { 0x0300, 0x036F }, { 0x0483, 0x0486 }, { 0x0488, 0x0489 }, 762 { 0x0591, 0x05BD }, { 0x05BF, 0x05BF }, { 0x05C1, 0x05C2 }, 763 { 0x05C4, 0x05C5 }, { 0x05C7, 0x05C7 }, { 0x0600, 0x0603 }, 764 { 0x0610, 0x0615 }, { 0x064B, 0x065E }, { 0x0670, 0x0670 }, 765 { 0x06D6, 0x06E4 }, { 0x06E7, 0x06E8 }, { 0x06EA, 0x06ED }, 766 { 0x070F, 0x070F }, { 0x0711, 0x0711 }, { 0x0730, 0x074A }, 767 { 0x07A6, 0x07B0 }, { 0x07EB, 0x07F3 }, { 0x0901, 0x0902 }, 768 { 0x093C, 0x093C }, { 0x0941, 0x0948 }, { 0x094D, 0x094D }, 769 { 0x0951, 0x0954 }, { 0x0962, 0x0963 }, { 0x0981, 0x0981 }, 770 { 0x09BC, 0x09BC }, { 0x09C1, 0x09C4 }, { 0x09CD, 0x09CD }, 771 { 0x09E2, 0x09E3 }, { 0x0A01, 0x0A02 }, { 0x0A3C, 0x0A3C }, 772 { 0x0A41, 0x0A42 }, { 0x0A47, 0x0A48 }, { 0x0A4B, 0x0A4D }, 773 { 0x0A70, 0x0A71 }, { 0x0A81, 0x0A82 }, { 0x0ABC, 0x0ABC }, 774 { 0x0AC1, 0x0AC5 }, { 0x0AC7, 0x0AC8 }, { 0x0ACD, 0x0ACD }, 775 { 0x0AE2, 0x0AE3 }, { 0x0B01, 0x0B01 }, { 0x0B3C, 0x0B3C }, 776 { 0x0B3F, 0x0B3F }, { 0x0B41, 0x0B43 }, { 0x0B4D, 0x0B4D }, 777 { 0x0B56, 0x0B56 }, { 0x0B82, 0x0B82 }, { 0x0BC0, 0x0BC0 }, 778 { 0x0BCD, 0x0BCD }, { 0x0C3E, 0x0C40 }, { 0x0C46, 0x0C48 }, 779 { 0x0C4A, 0x0C4D }, { 0x0C55, 0x0C56 }, { 0x0CBC, 0x0CBC }, 780 { 0x0CBF, 0x0CBF }, { 0x0CC6, 0x0CC6 }, { 0x0CCC, 0x0CCD }, 781 { 0x0CE2, 0x0CE3 }, { 0x0D41, 0x0D43 }, { 0x0D4D, 0x0D4D }, 782 { 0x0DCA, 0x0DCA }, { 0x0DD2, 0x0DD4 }, { 0x0DD6, 0x0DD6 }, 783 { 0x0E31, 0x0E31 }, { 0x0E34, 0x0E3A }, { 0x0E47, 0x0E4E }, 784 { 0x0EB1, 0x0EB1 }, { 0x0EB4, 0x0EB9 }, { 0x0EBB, 0x0EBC }, 785 { 0x0EC8, 0x0ECD }, { 0x0F18, 0x0F19 }, { 0x0F35, 0x0F35 }, 786 { 0x0F37, 0x0F37 }, { 0x0F39, 0x0F39 }, { 0x0F71, 0x0F7E }, 787 { 0x0F80, 0x0F84 }, { 0x0F86, 0x0F87 }, { 0x0F90, 0x0F97 }, 788 { 0x0F99, 0x0FBC }, { 0x0FC6, 0x0FC6 }, { 0x102D, 0x1030 }, 789 { 0x1032, 0x1032 }, { 0x1036, 0x1037 }, { 0x1039, 0x1039 }, 790 { 0x1058, 0x1059 }, { 0x1160, 0x11FF }, { 0x135F, 0x135F }, 791 { 0x1712, 0x1714 }, { 0x1732, 0x1734 }, { 0x1752, 0x1753 }, 792 { 0x1772, 0x1773 }, { 0x17B4, 0x17B5 }, { 0x17B7, 0x17BD }, 793 { 0x17C6, 0x17C6 }, { 0x17C9, 0x17D3 }, { 0x17DD, 0x17DD }, 794 { 0x180B, 0x180D }, { 0x18A9, 0x18A9 }, { 0x1920, 0x1922 }, 795 { 0x1927, 0x1928 }, { 0x1932, 0x1932 }, { 0x1939, 0x193B }, 796 { 0x1A17, 0x1A18 }, { 0x1B00, 0x1B03 }, { 0x1B34, 0x1B34 }, 797 { 0x1B36, 0x1B3A }, { 0x1B3C, 0x1B3C }, { 0x1B42, 0x1B42 }, 798 { 0x1B6B, 0x1B73 }, { 0x1DC0, 0x1DCA }, { 0x1DFE, 0x1DFF }, 799 { 0x200B, 0x200F }, { 0x202A, 0x202E }, { 0x2060, 0x2063 }, 800 { 0x206A, 0x206F }, { 0x20D0, 0x20EF }, { 0x302A, 0x302F }, 801 { 0x3099, 0x309A }, { 0xA806, 0xA806 }, { 0xA80B, 0xA80B }, 802 { 0xA825, 0xA826 }, { 0xFB1E, 0xFB1E }, { 0xFE00, 0xFE0F }, 803 { 0xFE20, 0xFE23 }, { 0xFEFF, 0xFEFF }, { 0xFFF9, 0xFFFB }, 804 { 0x10A01, 0x10A03 }, { 0x10A05, 0x10A06 }, { 0x10A0C, 0x10A0F }, 805 { 0x10A38, 0x10A3A }, { 0x10A3F, 0x10A3F }, { 0x1D167, 0x1D169 }, 806 { 0x1D173, 0x1D182 }, { 0x1D185, 0x1D18B }, { 0x1D1AA, 0x1D1AD }, 807 { 0x1D242, 0x1D244 }, { 0xE0001, 0xE0001 }, { 0xE0020, 0xE007F }, 808 { 0xE0100, 0xE01EF } 809 }; 810 811 /* test for 8-bit control characters */ 812 if (ucs == 0) 813 return 0; 814 if (ucs < 32 || (ucs >= 0x7f && ucs < 0xa0)) 815 return -1; 816 817 /* binary search in table of non-spacing characters */ 818 if (bisearch(ucs, combining, 819 sizeof(combining) / sizeof(struct interval) - 1)) 820 return 0; 821 822 /* if we arrive here, ucs is not a combining or C0/C1 control character */ 823 824 return 1 + 825 (ucs >= 0x1100 && 826 (ucs <= 0x115f || /* Hangul Jamo init. consonants */ 827 ucs == 0x2329 || ucs == 0x232a || 828 (ucs >= 0x2e80 && ucs <= 0xa4cf && 829 ucs != 0x303f) || /* CJK ... Yi */ 830 (ucs >= 0xac00 && ucs <= 0xd7a3) || /* Hangul Syllables */ 831 (ucs >= 0xf900 && ucs <= 0xfaff) || /* CJK Compatibility Ideographs */ 832 (ucs >= 0xfe10 && ucs <= 0xfe19) || /* Vertical forms */ 833 (ucs >= 0xfe30 && ucs <= 0xfe6f) || /* CJK Compatibility Forms */ 834 (ucs >= 0xff00 && ucs <= 0xff60) || /* Fullwidth Forms */ 835 (ucs >= 0xffe0 && ucs <= 0xffe6) || 836 (ucs >= 0x20000 && ucs <= 0x2fffd) || 837 (ucs >= 0x30000 && ucs <= 0x3fffd))); 838} 839 840 841/* 842 * The following functions are part of the original wcwidth.c: 843 * we don't use them but I've kept them in case - pws. 844 */ 845#if 0 846int mk_wcswidth(const wchar_t *pwcs, size_t n) 847{ 848 int w, width = 0; 849 850 for (;*pwcs && n-- > 0; pwcs++) 851 if ((w = mk_wcwidth(*pwcs)) < 0) 852 return -1; 853 else 854 width += w; 855 856 return width; 857} 858 859 860/* 861 * The following functions are the same as mk_wcwidth() and 862 * mk_wcswidth(), except that spacing characters in the East Asian 863 * Ambiguous (A) category as defined in Unicode Technical Report #11 864 * have a column width of 2. This variant might be useful for users of 865 * CJK legacy encodings who want to migrate to UCS without changing 866 * the traditional terminal character-width behaviour. It is not 867 * otherwise recommended for general use. 868 */ 869int mk_wcwidth_cjk(wchar_t ucs) 870{ 871 /* sorted list of non-overlapping intervals of East Asian Ambiguous 872 * characters, generated by "uniset +WIDTH-A -cat=Me -cat=Mn -cat=Cf c" */ 873 static const struct interval ambiguous[] = { 874 { 0x00A1, 0x00A1 }, { 0x00A4, 0x00A4 }, { 0x00A7, 0x00A8 }, 875 { 0x00AA, 0x00AA }, { 0x00AE, 0x00AE }, { 0x00B0, 0x00B4 }, 876 { 0x00B6, 0x00BA }, { 0x00BC, 0x00BF }, { 0x00C6, 0x00C6 }, 877 { 0x00D0, 0x00D0 }, { 0x00D7, 0x00D8 }, { 0x00DE, 0x00E1 }, 878 { 0x00E6, 0x00E6 }, { 0x00E8, 0x00EA }, { 0x00EC, 0x00ED }, 879 { 0x00F0, 0x00F0 }, { 0x00F2, 0x00F3 }, { 0x00F7, 0x00FA }, 880 { 0x00FC, 0x00FC }, { 0x00FE, 0x00FE }, { 0x0101, 0x0101 }, 881 { 0x0111, 0x0111 }, { 0x0113, 0x0113 }, { 0x011B, 0x011B }, 882 { 0x0126, 0x0127 }, { 0x012B, 0x012B }, { 0x0131, 0x0133 }, 883 { 0x0138, 0x0138 }, { 0x013F, 0x0142 }, { 0x0144, 0x0144 }, 884 { 0x0148, 0x014B }, { 0x014D, 0x014D }, { 0x0152, 0x0153 }, 885 { 0x0166, 0x0167 }, { 0x016B, 0x016B }, { 0x01CE, 0x01CE }, 886 { 0x01D0, 0x01D0 }, { 0x01D2, 0x01D2 }, { 0x01D4, 0x01D4 }, 887 { 0x01D6, 0x01D6 }, { 0x01D8, 0x01D8 }, { 0x01DA, 0x01DA }, 888 { 0x01DC, 0x01DC }, { 0x0251, 0x0251 }, { 0x0261, 0x0261 }, 889 { 0x02C4, 0x02C4 }, { 0x02C7, 0x02C7 }, { 0x02C9, 0x02CB }, 890 { 0x02CD, 0x02CD }, { 0x02D0, 0x02D0 }, { 0x02D8, 0x02DB }, 891 { 0x02DD, 0x02DD }, { 0x02DF, 0x02DF }, { 0x0391, 0x03A1 }, 892 { 0x03A3, 0x03A9 }, { 0x03B1, 0x03C1 }, { 0x03C3, 0x03C9 }, 893 { 0x0401, 0x0401 }, { 0x0410, 0x044F }, { 0x0451, 0x0451 }, 894 { 0x2010, 0x2010 }, { 0x2013, 0x2016 }, { 0x2018, 0x2019 }, 895 { 0x201C, 0x201D }, { 0x2020, 0x2022 }, { 0x2024, 0x2027 }, 896 { 0x2030, 0x2030 }, { 0x2032, 0x2033 }, { 0x2035, 0x2035 }, 897 { 0x203B, 0x203B }, { 0x203E, 0x203E }, { 0x2074, 0x2074 }, 898 { 0x207F, 0x207F }, { 0x2081, 0x2084 }, { 0x20AC, 0x20AC }, 899 { 0x2103, 0x2103 }, { 0x2105, 0x2105 }, { 0x2109, 0x2109 }, 900 { 0x2113, 0x2113 }, { 0x2116, 0x2116 }, { 0x2121, 0x2122 }, 901 { 0x2126, 0x2126 }, { 0x212B, 0x212B }, { 0x2153, 0x2154 }, 902 { 0x215B, 0x215E }, { 0x2160, 0x216B }, { 0x2170, 0x2179 }, 903 { 0x2190, 0x2199 }, { 0x21B8, 0x21B9 }, { 0x21D2, 0x21D2 }, 904 { 0x21D4, 0x21D4 }, { 0x21E7, 0x21E7 }, { 0x2200, 0x2200 }, 905 { 0x2202, 0x2203 }, { 0x2207, 0x2208 }, { 0x220B, 0x220B }, 906 { 0x220F, 0x220F }, { 0x2211, 0x2211 }, { 0x2215, 0x2215 }, 907 { 0x221A, 0x221A }, { 0x221D, 0x2220 }, { 0x2223, 0x2223 }, 908 { 0x2225, 0x2225 }, { 0x2227, 0x222C }, { 0x222E, 0x222E }, 909 { 0x2234, 0x2237 }, { 0x223C, 0x223D }, { 0x2248, 0x2248 }, 910 { 0x224C, 0x224C }, { 0x2252, 0x2252 }, { 0x2260, 0x2261 }, 911 { 0x2264, 0x2267 }, { 0x226A, 0x226B }, { 0x226E, 0x226F }, 912 { 0x2282, 0x2283 }, { 0x2286, 0x2287 }, { 0x2295, 0x2295 }, 913 { 0x2299, 0x2299 }, { 0x22A5, 0x22A5 }, { 0x22BF, 0x22BF }, 914 { 0x2312, 0x2312 }, { 0x2460, 0x24E9 }, { 0x24EB, 0x254B }, 915 { 0x2550, 0x2573 }, { 0x2580, 0x258F }, { 0x2592, 0x2595 }, 916 { 0x25A0, 0x25A1 }, { 0x25A3, 0x25A9 }, { 0x25B2, 0x25B3 }, 917 { 0x25B6, 0x25B7 }, { 0x25BC, 0x25BD }, { 0x25C0, 0x25C1 }, 918 { 0x25C6, 0x25C8 }, { 0x25CB, 0x25CB }, { 0x25CE, 0x25D1 }, 919 { 0x25E2, 0x25E5 }, { 0x25EF, 0x25EF }, { 0x2605, 0x2606 }, 920 { 0x2609, 0x2609 }, { 0x260E, 0x260F }, { 0x2614, 0x2615 }, 921 { 0x261C, 0x261C }, { 0x261E, 0x261E }, { 0x2640, 0x2640 }, 922 { 0x2642, 0x2642 }, { 0x2660, 0x2661 }, { 0x2663, 0x2665 }, 923 { 0x2667, 0x266A }, { 0x266C, 0x266D }, { 0x266F, 0x266F }, 924 { 0x273D, 0x273D }, { 0x2776, 0x277F }, { 0xE000, 0xF8FF }, 925 { 0xFFFD, 0xFFFD }, { 0xF0000, 0xFFFFD }, { 0x100000, 0x10FFFD } 926 }; 927 928 /* binary search in table of non-spacing characters */ 929 if (bisearch(ucs, ambiguous, 930 sizeof(ambiguous) / sizeof(struct interval) - 1)) 931 return 2; 932 933 return mk_wcwidth(ucs); 934} 935 936 937int mk_wcswidth_cjk(const wchar_t *pwcs, size_t n) 938{ 939 int w, width = 0; 940 941 for (;*pwcs && n-- > 0; pwcs++) 942 if ((w = mk_wcwidth_cjk(*pwcs)) < 0) 943 return -1; 944 else 945 width += w; 946 947 return width; 948} 949#endif /* 0 */ 950 951/**/ 952#endif /* BROKEN_WCWIDTH && (__STDC_ISO_10646__ || __APPLE__) */ 953 954