libbb/unicode.c

   1 /* vi: set sw=4 ts=4: */
   2 /*
   3  * Unicode support routines.
   4  *
   5  * Copyright (C) 2009 Denys Vlasenko
   6  *
   7  * Licensed under GPL version 2, see file LICENSE in this tarball for details.
   8  */
   9 #include "libbb.h"
  10 #include "unicode.h"
  11
  12 /* If it's not #defined as a constant in unicode.h... */
  13 #ifndef unicode_status
  14 uint8_t unicode_status;
  15 #endif
  16
  17 /* This file is compiled only if FEATURE_ASSUME_UNICODE is on.
  18  * We check other options and decide whether to use libc support
  19  * via locale, or use our own logic:
  20  */
  21
  22 #if ENABLE_LOCALE_SUPPORT
  23
  24 /* Unicode support using libc locale support. */
  25
  26 void FAST_FUNC init_unicode(void)
  27 {
  28         /* In unicode, this is a one character string */
  29         static const char unicode_0x394[] = { 0xce, 0x94, 0 };
  30
  31         if (unicode_status != UNICODE_UNKNOWN)
  32                 return;
  33
  34         unicode_status = unicode_strlen(unicode_0x394) == 1 ? UNICODE_ON : UNICODE_OFF;
  35 }
  36
  37 #else
  38
  39 /* Homegrown Unicode support. It knows only C and Unicode locales. */
  40
  41 # if ENABLE_FEATURE_CHECK_UNICODE_IN_ENV
  42 void FAST_FUNC init_unicode(void)
  43 {
  44         char *lang;
  45
  46         if (unicode_status != UNICODE_UNKNOWN)
  47                 return;
  48
  49         unicode_status = UNICODE_OFF;
  50         lang = getenv("LANG");
  51         if (!lang || !(strstr(lang, ".utf") || strstr(lang, ".UTF")))
  52                 return;
  53         unicode_status = UNICODE_ON;
  54 }
  55 # endif
  56
  57 static size_t wcrtomb_internal(char *s, wchar_t wc)
  58 {
  59         int n, i;
  60         uint32_t v = wc;
  61
  62         if (v <= 0x7f) {
  63                 *s = v;
  64                 return 1;
  65         }
  66
  67         /* RFC 3629 says that Unicode ends at 10FFFF,
  68          * but we cover entire 32 bits */
  69
  70         /* 4000000-FFFFFFFF -> 111111tt 10tttttt 10zzzzzz 10zzyyyy 10yyyyxx 10xxxxxx */
  71         /* 200000-3FFFFFF -> 111110tt 10zzzzzz 10zzyyyy 10yyyyxx 10xxxxxx */
  72         /* 10000-1FFFFF -> 11110zzz 10zzyyyy 10yyyyxx 10xxxxxx */
  73         /* 800-FFFF -> 1110yyyy 10yyyyxx 10xxxxxx */
  74         /* 80-7FF -> 110yyyxx 10xxxxxx */
  75
  76         /* How many bytes do we need? */
  77         n = 2;
  78         /* (0x80000000+ would result in n = 7, limiting n to 6) */
  79         while (v >= 0x800 && n < 6) {
  80                 v >>= 5;
  81                 n++;
  82         }
  83         /* Fill bytes n-1..1 */
  84         i = n;
  85         while (--i) {
  86                 s[i] = (wc & 0x3f) | 0x80;
  87                 wc >>= 6;
  88         }
  89         /* Fill byte 0 */
  90         s[0] = wc | (uint8_t)(0x3f00 >> n);
  91         return n;
  92 }
  93 size_t FAST_FUNC wcrtomb(char *s, wchar_t wc, mbstate_t *ps UNUSED_PARAM)
  94 {
  95         if (unicode_status != UNICODE_ON) {
  96                 *s = wc;
  97                 return 1;
  98         }
  99
 100         return wcrtomb_internal(s, wc);
 101 }
 102 size_t FAST_FUNC wcstombs(char *dest, const wchar_t *src, size_t n)
 103 {
 104         size_t org_n = n;
 105
 106         if (unicode_status != UNICODE_ON) {
 107                 while (n) {
 108                         wchar_t c = *src++;
 109                         *dest++ = c;
 110                         if (c == 0)
 111                                 break;
 112                         n--;
 113                 }
 114                 return org_n - n;
 115         }
 116
 117         while (n >= MB_CUR_MAX) {
 118                 wchar_t wc = *src++;
 119                 size_t len = wcrtomb_internal(dest, wc);
 120
 121                 if (wc == L'\0')
 122                         return org_n - n;
 123                 dest += len;
 124                 n -= len;
 125         }
 126         while (n) {
 127                 char tbuf[MB_CUR_MAX];
 128                 wchar_t wc = *src++;
 129                 size_t len = wcrtomb_internal(tbuf, wc);
 130
 131                 if (len > n)
 132                         len = n;
 133                 memcpy(dest, tbuf, len);
 134                 if (wc == L'\0')
 135                         return org_n - n;
 136                 dest += len;
 137                 n -= len;
 138         }
 139         return org_n - n;
 140 }
 141
 142 #define ERROR_WCHAR (~(wchar_t)0)
 143
 144 static const char *mbstowc_internal(wchar_t *res, const char *src)
 145 {
 146         int bytes;
 147         unsigned c = (unsigned char) *src++;
 148
 149         if (c <= 0x7f) {
 150                 *res = c;
 151                 return src;
 152         }
 153
 154         /* 80-7FF -> 110yyyxx 10xxxxxx */
 155         /* 800-FFFF -> 1110yyyy 10yyyyxx 10xxxxxx */
 156         /* 10000-1FFFFF -> 11110zzz 10zzyyyy 10yyyyxx 10xxxxxx */
 157         /* 200000-3FFFFFF -> 111110tt 10zzzzzz 10zzyyyy 10yyyyxx 10xxxxxx */
 158         /* 4000000-FFFFFFFF -> 111111tt 10tttttt 10zzzzzz 10zzyyyy 10yyyyxx 10xxxxxx */
 159         bytes = 0;
 160         do {
 161                 c <<= 1;
 162                 bytes++;
 163         } while ((c & 0x80) && bytes < 6);
 164         if (bytes == 1) {
 165                 /* A bare "continuation" byte. Say, 80 */
 166                 *res = ERROR_WCHAR;
 167                 return src;
 168         }
 169         c = (uint8_t)(c) >> bytes;
 170
 171         while (--bytes) {
 172                 unsigned ch = (unsigned char) *src;
 173                 if ((ch & 0xc0) != 0x80) {
 174                         /* Missing "continuation" byte. Example: e0 80 */
 175                         *res = ERROR_WCHAR;
 176                         return src;
 177                 }
 178                 c = (c << 6) + (ch & 0x3f);
 179                 src++;
 180         }
 181
 182         /* TODO */
 183         /* Need to check that c isn't produced by overlong encoding */
 184         /* Example: 11000000 10000000 converts to NUL */
 185         /* 11110000 10000000 10000100 10000000 converts to 0x100 */
 186         /* correct encoding: 11000100 10000000 */
 187         if (c <= 0x7f) { /* crude check */
 188                 *res = ERROR_WCHAR;
 189                 return src;
 190         }
 191
 192         *res = c;
 193         return src;
 194 }
 195 size_t FAST_FUNC mbstowcs(wchar_t *dest, const char *src, size_t n)
 196 {
 197         size_t org_n = n;
 198
 199         if (unicode_status != UNICODE_ON) {
 200                 while (n) {
 201                         unsigned char c = *src++;
 202
 203                         if (dest)
 204                                 *dest++ = c;
 205                         if (c == 0)
 206                                 break;
 207                         n--;
 208                 }
 209                 return org_n - n;
 210         }
 211
 212         while (n) {
 213                 wchar_t wc;
 214                 src = mbstowc_internal(&wc, src);
 215                 if (wc == ERROR_WCHAR) /* error */
 216                         return (size_t) -1L;
 217                 if (dest)
 218                         *dest++ = wc;
 219                 if (wc == 0) /* end-of-string */
 220                         break;
 221                 n--;
 222         }
 223
 224         return org_n - n;
 225 }
 226
 227 int FAST_FUNC iswspace(wint_t wc)
 228 {
 229         return (unsigned)wc <= 0x7f && isspace(wc);
 230 }
 231
 232 int FAST_FUNC iswalnum(wint_t wc)
 233 {
 234         return (unsigned)wc <= 0x7f && isalnum(wc);
 235 }
 236
 237 int FAST_FUNC iswpunct(wint_t wc)
 238 {
 239         return (unsigned)wc <= 0x7f && ispunct(wc);
 240 }
 241
 242 #include "unicode_wcwidth.c"
 243
 244 # if ENABLE_UNICODE_BIDI_SUPPORT
 245 int FAST_FUNC unicode_bidi_isrtl(wint_t wc)
 246 {
 247         /* ranges taken from
 248          * http://www.unicode.org/Public/5.2.0/ucd/extracted/DerivedBidiClass.txt
 249          * Bidi_Class=Left_To_Right | Bidi_Class=Arabic_Letter
 250          */
 251         static const struct interval rtl_b[] = {
 252 #  define BIG_(a,b) { a, b },
 253 #  define PAIR(a,b)
 254 #  define ARRAY \
 255                 PAIR(0x0590, 0x0590) \
 256                 PAIR(0x05BE, 0x05BE) \
 257                 PAIR(0x05C0, 0x05C0) \
 258                 PAIR(0x05C3, 0x05C3) \
 259                 PAIR(0x05C6, 0x05C6) \
 260                 BIG_(0x05C8, 0x05FF) \
 261                 PAIR(0x0604, 0x0605) \
 262                 PAIR(0x0608, 0x0608) \
 263                 PAIR(0x060B, 0x060B) \
 264                 PAIR(0x060D, 0x060D) \
 265                 BIG_(0x061B, 0x064A) \
 266                 PAIR(0x065F, 0x065F) \
 267                 PAIR(0x066D, 0x066F) \
 268                 BIG_(0x0671, 0x06D5) \
 269                 PAIR(0x06E5, 0x06E6) \
 270                 PAIR(0x06EE, 0x06EF) \
 271                 BIG_(0x06FA, 0x070E) \
 272                 PAIR(0x0710, 0x0710) \
 273                 BIG_(0x0712, 0x072F) \
 274                 BIG_(0x074B, 0x07A5) \
 275                 BIG_(0x07B1, 0x07EA) \
 276                 PAIR(0x07F4, 0x07F5) \
 277                 BIG_(0x07FA, 0x0815) \
 278                 PAIR(0x081A, 0x081A) \
 279                 PAIR(0x0824, 0x0824) \
 280                 PAIR(0x0828, 0x0828) \
 281                 BIG_(0x082E, 0x08FF) \
 282                 PAIR(0x200F, 0x200F) \
 283                 PAIR(0x202B, 0x202B) \
 284                 PAIR(0x202E, 0x202E) \
 285                 BIG_(0xFB1D, 0xFB1D) \
 286                 BIG_(0xFB1F, 0xFB28) \
 287                 BIG_(0xFB2A, 0xFD3D) \
 288                 BIG_(0xFD40, 0xFDCF) \
 289                 BIG_(0xFDC8, 0xFDCF) \
 290                 BIG_(0xFDF0, 0xFDFC) \
 291                 BIG_(0xFDFE, 0xFDFF) \
 292                 BIG_(0xFE70, 0xFEFE)
 293                 /* Probably not necessary
 294                 {0x10800, 0x1091E},
 295                 {0x10920, 0x10A00},
 296                 {0x10A04, 0x10A04},
 297                 {0x10A07, 0x10A0B},
 298                 {0x10A10, 0x10A37},
 299                 {0x10A3B, 0x10A3E},
 300                 {0x10A40, 0x10A7F},
 301                 {0x10B36, 0x10B38},
 302                 {0x10B40, 0x10E5F},
 303                 {0x10E7F, 0x10FFF},
 304                 {0x1E800, 0x1EFFF}
 305                 */
 306                 ARRAY
 307 #  undef BIG_
 308 #  undef PAIR
 309         };
 310 #  define BIG_(a,b)
 311 #  define PAIR(a,b) (a << 2) | (b-a),
 312         static const uint16_t rtl_p[] = { ARRAY };
 313 #  undef BIG_
 314 #  undef PAIR
 315 #  define BIG_(a,b) char big_##a[b < 0x4000 && b-a <= 3 ? -1 : 1];
 316 #  define PAIR(a,b) char pair##a[b >= 0x4000 || b-a > 3 ? -1 : 1];
 317         struct CHECK { ARRAY };
 318 #  undef BIG_
 319 #  undef PAIR
 320 #  undef ARRAY
 321
 322         if (in_interval_table(wc, rtl_b, ARRAY_SIZE(rtl_b) - 1))
 323                 return 1;
 324         if (in_uint16_table(wc, rtl_p, ARRAY_SIZE(rtl_p) - 1))
 325                 return 1;
 326         return 0;
 327 }
 328
 329 #  if ENABLE_UNICODE_NEUTRAL_TABLE
 330 int FAST_FUNC unicode_bidi_is_neutral_wchar(wint_t wc)
 331 {
 332         /* ranges taken from
 333          * http://www.unicode.org/Public/5.2.0/ucd/extracted/DerivedBidiClass.txt
 334          * Bidi_Classes: Paragraph_Separator, Segment_Separator,
 335          * White_Space, Other_Neutral, European_Number, European_Separator,
 336          * European_Terminator, Arabic_Number, Common_Separator
 337          */
 338         static const struct interval neutral_b[] = {
 339 #  define BIG_(a,b) { a, b },
 340 #  define PAIR(a,b)
 341 #  define ARRAY \
 342                 BIG_(0x0009, 0x000D) \
 343                 BIG_(0x001C, 0x0040) \
 344                 BIG_(0x005B, 0x0060) \
 345                 PAIR(0x007B, 0x007E) \
 346                 PAIR(0x0085, 0x0085) \
 347                 BIG_(0x00A0, 0x00A9) \
 348                 PAIR(0x00AB, 0x00AC) \
 349                 BIG_(0x00AE, 0x00B4) \
 350                 PAIR(0x00B6, 0x00B9) \
 351                 BIG_(0x00BB, 0x00BF) \
 352                 PAIR(0x00D7, 0x00D7) \
 353                 PAIR(0x00F7, 0x00F7) \
 354                 PAIR(0x02B9, 0x02BA) \
 355                 BIG_(0x02C2, 0x02CF) \
 356                 BIG_(0x02D2, 0x02DF) \
 357                 BIG_(0x02E5, 0x02FF) \
 358                 PAIR(0x0374, 0x0375) \
 359                 PAIR(0x037E, 0x037E) \
 360                 PAIR(0x0384, 0x0385) \
 361                 PAIR(0x0387, 0x0387) \
 362                 PAIR(0x03F6, 0x03F6) \
 363                 PAIR(0x058A, 0x058A) \
 364                 PAIR(0x0600, 0x0603) \
 365                 PAIR(0x0606, 0x0607) \
 366                 PAIR(0x0609, 0x060A) \
 367                 PAIR(0x060C, 0x060C) \
 368                 PAIR(0x060E, 0x060F) \
 369                 BIG_(0x0660, 0x066C) \
 370                 PAIR(0x06DD, 0x06DD) \
 371                 PAIR(0x06E9, 0x06E9) \
 372                 BIG_(0x06F0, 0x06F9) \
 373                 PAIR(0x07F6, 0x07F9) \
 374                 PAIR(0x09F2, 0x09F3) \
 375                 PAIR(0x09FB, 0x09FB) \
 376                 PAIR(0x0AF1, 0x0AF1) \
 377                 BIG_(0x0BF3, 0x0BFA) \
 378                 BIG_(0x0C78, 0x0C7E) \
 379                 PAIR(0x0CF1, 0x0CF2) \
 380                 PAIR(0x0E3F, 0x0E3F) \
 381                 PAIR(0x0F3A, 0x0F3D) \
 382                 BIG_(0x1390, 0x1400) \
 383                 PAIR(0x1680, 0x1680) \
 384                 PAIR(0x169B, 0x169C) \
 385                 PAIR(0x17DB, 0x17DB) \
 386                 BIG_(0x17F0, 0x17F9) \
 387                 BIG_(0x1800, 0x180A) \
 388                 PAIR(0x180E, 0x180E) \
 389                 PAIR(0x1940, 0x1940) \
 390                 PAIR(0x1944, 0x1945) \
 391                 BIG_(0x19DE, 0x19FF) \
 392                 PAIR(0x1FBD, 0x1FBD) \
 393                 PAIR(0x1FBF, 0x1FC1) \
 394                 PAIR(0x1FCD, 0x1FCF) \
 395                 PAIR(0x1FDD, 0x1FDF) \
 396                 PAIR(0x1FED, 0x1FEF) \
 397                 PAIR(0x1FFD, 0x1FFE) \
 398                 BIG_(0x2000, 0x200A) \
 399                 BIG_(0x2010, 0x2029) \
 400                 BIG_(0x202F, 0x205F) \
 401                 PAIR(0x2070, 0x2070) \
 402                 BIG_(0x2074, 0x207E) \
 403                 BIG_(0x2080, 0x208E) \
 404                 BIG_(0x20A0, 0x20B8) \
 405                 PAIR(0x2100, 0x2101) \
 406                 PAIR(0x2103, 0x2106) \
 407                 PAIR(0x2108, 0x2109) \
 408                 PAIR(0x2114, 0x2114) \
 409                 PAIR(0x2116, 0x2118) \
 410                 BIG_(0x211E, 0x2123) \
 411                 PAIR(0x2125, 0x2125) \
 412                 PAIR(0x2127, 0x2127) \
 413                 PAIR(0x2129, 0x2129) \
 414                 PAIR(0x212E, 0x212E) \
 415                 PAIR(0x213A, 0x213B) \
 416                 BIG_(0x2140, 0x2144) \
 417                 PAIR(0x214A, 0x214D) \
 418                 BIG_(0x2150, 0x215F) \
 419                 PAIR(0x2189, 0x2189) \
 420                 BIG_(0x2190, 0x2335) \
 421                 BIG_(0x237B, 0x2394) \
 422                 BIG_(0x2396, 0x23E8) \
 423                 BIG_(0x2400, 0x2426) \
 424                 BIG_(0x2440, 0x244A) \
 425                 BIG_(0x2460, 0x249B) \
 426                 BIG_(0x24EA, 0x26AB) \
 427                 BIG_(0x26AD, 0x26CD) \
 428                 BIG_(0x26CF, 0x26E1) \
 429                 PAIR(0x26E3, 0x26E3) \
 430                 BIG_(0x26E8, 0x26FF) \
 431                 PAIR(0x2701, 0x2704) \
 432                 PAIR(0x2706, 0x2709) \
 433                 BIG_(0x270C, 0x2727) \
 434                 BIG_(0x2729, 0x274B) \
 435                 PAIR(0x274D, 0x274D) \
 436                 PAIR(0x274F, 0x2752) \
 437                 BIG_(0x2756, 0x275E) \
 438                 BIG_(0x2761, 0x2794) \
 439                 BIG_(0x2798, 0x27AF) \
 440                 BIG_(0x27B1, 0x27BE) \
 441                 BIG_(0x27C0, 0x27CA) \
 442                 PAIR(0x27CC, 0x27CC) \
 443                 BIG_(0x27D0, 0x27FF) \
 444                 BIG_(0x2900, 0x2B4C) \
 445                 BIG_(0x2B50, 0x2B59) \
 446                 BIG_(0x2CE5, 0x2CEA) \
 447                 BIG_(0x2CF9, 0x2CFF) \
 448                 BIG_(0x2E00, 0x2E99) \
 449                 BIG_(0x2E9B, 0x2EF3) \
 450                 BIG_(0x2F00, 0x2FD5) \
 451                 BIG_(0x2FF0, 0x2FFB) \
 452                 BIG_(0x3000, 0x3004) \
 453                 BIG_(0x3008, 0x3020) \
 454                 PAIR(0x3030, 0x3030) \
 455                 PAIR(0x3036, 0x3037) \
 456                 PAIR(0x303D, 0x303D) \
 457                 PAIR(0x303E, 0x303F) \
 458                 PAIR(0x309B, 0x309C) \
 459                 PAIR(0x30A0, 0x30A0) \
 460                 PAIR(0x30FB, 0x30FB) \
 461                 BIG_(0x31C0, 0x31E3) \
 462                 PAIR(0x321D, 0x321E) \
 463                 BIG_(0x3250, 0x325F) \
 464                 PAIR(0x327C, 0x327E) \
 465                 BIG_(0x32B1, 0x32BF) \
 466                 PAIR(0x32CC, 0x32CF) \
 467                 PAIR(0x3377, 0x337A) \
 468                 PAIR(0x33DE, 0x33DF) \
 469                 PAIR(0x33FF, 0x33FF) \
 470                 BIG_(0x4DC0, 0x4DFF) \
 471                 BIG_(0xA490, 0xA4C6) \
 472                 BIG_(0xA60D, 0xA60F) \
 473                 BIG_(0xA673, 0xA673) \
 474                 BIG_(0xA67E, 0xA67F) \
 475                 BIG_(0xA700, 0xA721) \
 476                 BIG_(0xA788, 0xA788) \
 477                 BIG_(0xA828, 0xA82B) \
 478                 BIG_(0xA838, 0xA839) \
 479                 BIG_(0xA874, 0xA877) \
 480                 BIG_(0xFB29, 0xFB29) \
 481                 BIG_(0xFD3E, 0xFD3F) \
 482                 BIG_(0xFDFD, 0xFDFD) \
 483                 BIG_(0xFE10, 0xFE19) \
 484                 BIG_(0xFE30, 0xFE52) \
 485                 BIG_(0xFE54, 0xFE66) \
 486                 BIG_(0xFE68, 0xFE6B) \
 487                 BIG_(0xFF01, 0xFF20) \
 488                 BIG_(0xFF3B, 0xFF40) \
 489                 BIG_(0xFF5B, 0xFF65) \
 490                 BIG_(0xFFE0, 0xFFE6) \
 491                 BIG_(0xFFE8, 0xFFEE) \
 492                 BIG_(0xFFF9, 0xFFFD)
 493                 /*
 494                 {0x10101, 0x10101},
 495                 {0x10140, 0x1019B},
 496                 {0x1091F, 0x1091F},
 497                 {0x10B39, 0x10B3F},
 498                 {0x10E60, 0x10E7E},
 499                 {0x1D200, 0x1D241},
 500                 {0x1D245, 0x1D245},
 501                 {0x1D300, 0x1D356},
 502                 {0x1D6DB, 0x1D6DB},
 503                 {0x1D715, 0x1D715},
 504                 {0x1D74F, 0x1D74F},
 505                 {0x1D789, 0x1D789},
 506                 {0x1D7C3, 0x1D7C3},
 507                 {0x1D7CE, 0x1D7FF},
 508                 {0x1F000, 0x1F02B},
 509                 {0x1F030, 0x1F093},
 510                 {0x1F100, 0x1F10A}
 511                 */
 512                 ARRAY
 513 #  undef BIG_
 514 #  undef PAIR
 515         };
 516 #  define BIG_(a,b)
 517 #  define PAIR(a,b) (a << 2) | (b-a),
 518         static const uint16_t neutral_p[] = { ARRAY };
 519 #  undef BIG_
 520 #  undef PAIR
 521 #  define BIG_(a,b) char big_##a[b < 0x4000 && b-a <= 3 ? -1 : 1];
 522 #  define PAIR(a,b) char pair##a[b >= 0x4000 || b-a > 3 ? -1 : 1];
 523         struct CHECK { ARRAY };
 524 #  undef BIG_
 525 #  undef PAIR
 526 #  undef ARRAY
 527
 528         if (in_interval_table(wc, neutral_b, ARRAY_SIZE(neutral_b) - 1))
 529                 return 1;
 530         if (in_uint16_table(wc, neutral_p, ARRAY_SIZE(neutral_p) - 1))
 531                 return 1;
 532         return 0;
 533 }
 534 #  endif
 535
 536 # endif /* UNICODE_BIDI_SUPPORT */
 537
 538 #endif /* Homegrown Unicode support */
 539
 540
 541 /* The rest is mostly same for libc and for "homegrown" support */
 542
 543 size_t FAST_FUNC unicode_strlen(const char *string)
 544 {
 545         size_t width = mbstowcs(NULL, string, INT_MAX);
 546         if (width == (size_t)-1L)
 547                 return strlen(string);
 548         return width;
 549 }
 550
 551 static char* FAST_FUNC unicode_conv_to_printable2(uni_stat_t *stats, const char *src, unsigned width, int flags)
 552 {
 553         char *dst;
 554         unsigned dst_len;
 555         unsigned uni_count;
 556         unsigned uni_width;
 557
 558         if (unicode_status != UNICODE_ON) {
 559                 char *d;
 560                 if (flags & UNI_FLAG_PAD) {
 561                         d = dst = xmalloc(width + 1);
 562                         while ((int)--width >= 0) {
 563                                 unsigned char c = *src;
 564                                 if (c == '\0') {
 565                                         do
 566                                                 *d++ = ' ';
 567                                         while ((int)--width >= 0);
 568                                         break;
 569                                 }
 570                                 *d++ = (c >= ' ' && c < 0x7f) ? c : '?';
 571                                 src++;
 572                         }
 573                         *d = '\0';
 574                 } else {
 575                         d = dst = xstrndup(src, width);
 576                         while (*d) {
 577                                 unsigned char c = *d;
 578                                 if (c < ' ' || c >= 0x7f)
 579                                         *d = '?';
 580                                 d++;
 581                         }
 582                 }
 583                 if (stats)
 584                         stats->byte_count = stats->unicode_count = (d - dst);
 585                 return dst;
 586         }
 587
 588         dst = NULL;
 589         uni_count = uni_width = 0;
 590         dst_len = 0;
 591         while (1) {
 592                 int w;
 593                 wchar_t wc;
 594
 595 #if ENABLE_LOCALE_SUPPORT
 596                 {
 597                         mbstate_t mbst = { 0 };
 598                         ssize_t rc = mbsrtowcs(&wc, &src, 1, &mbst);
 599                         /* If invalid sequence is seen: -1 is returned,
 600                          * src points to the invalid sequence, errno = EILSEQ.
 601                          * Else number of wchars (excluding terminating L'\0')
 602                          * written to dest is returned.
 603                          * If len (here: 1) non-L'\0' wchars stored at dest,
 604                          * src points to the next char to be converted.
 605                          * If string is completely converted: src = NULL.
 606                          */
 607                         if (rc == 0) /* end-of-string */
 608                                 break;
 609                         if (rc < 0) { /* error */
 610                                 src++;
 611                                 goto subst;
 612                         }
 613                         if (!iswprint(wc))
 614                                 goto subst;
 615                 }
 616 #else
 617                 src = mbstowc_internal(&wc, src);
 618                 /* src is advanced to next mb char
 619                  * wc == ERROR_WCHAR: invalid sequence is seen
 620                  * else: wc is set
 621                  */
 622                 if (wc == ERROR_WCHAR) /* error */
 623                         goto subst;
 624                 if (wc == 0) /* end-of-string */
 625                         break;
 626 #endif
 627                 if (CONFIG_LAST_SUPPORTED_WCHAR && wc > CONFIG_LAST_SUPPORTED_WCHAR)
 628                         goto subst;
 629                 w = wcwidth(wc);
 630                 if ((ENABLE_UNICODE_COMBINING_WCHARS && w < 0) /* non-printable wchar */
 631                  || (!ENABLE_UNICODE_COMBINING_WCHARS && w <= 0)
 632                  || (!ENABLE_UNICODE_WIDE_WCHARS && w > 1)
 633                 ) {
 634  subst:
 635                         wc = CONFIG_SUBST_WCHAR;
 636                         w = 1;
 637                 }
 638                 width -= w;
 639                 /* Note: if width == 0, we still may add more chars,
 640                  * they may be zero-width or combining ones */
 641                 if ((int)width < 0) {
 642                         /* can't add this wc, string would become longer than width */
 643                         width += w;
 644                         break;
 645                 }
 646
 647                 uni_count++;
 648                 uni_width += w;
 649                 dst = xrealloc(dst, dst_len + MB_CUR_MAX);
 650 #if ENABLE_LOCALE_SUPPORT
 651                 {
 652                         mbstate_t mbst = { 0 };
 653                         dst_len += wcrtomb(&dst[dst_len], wc, &mbst);
 654                 }
 655 #else
 656                 dst_len += wcrtomb_internal(&dst[dst_len], wc);
 657 #endif
 658         }
 659
 660         /* Pad to remaining width */
 661         if (flags & UNI_FLAG_PAD) {
 662                 dst = xrealloc(dst, dst_len + width + 1);
 663                 uni_count += width;
 664                 uni_width += width;
 665                 while ((int)--width >= 0) {
 666                         dst[dst_len++] = ' ';
 667                 }
 668         }
 669         dst[dst_len] = '\0';
 670         if (stats) {
 671                 stats->byte_count = dst_len;
 672                 stats->unicode_count = uni_count;
 673                 stats->unicode_width = uni_width;
 674         }
 675
 676         return dst;
 677 }
 678 char* FAST_FUNC unicode_conv_to_printable(uni_stat_t *stats, const char *src)
 679 {
 680         return unicode_conv_to_printable2(stats, src, INT_MAX, 0);
 681 }
 682 char* FAST_FUNC unicode_conv_to_printable_maxwidth(uni_stat_t *stats, const char *src, unsigned maxwidth)
 683 {
 684         return unicode_conv_to_printable2(stats, src, maxwidth, 0);
 685 }
 686 char* FAST_FUNC unicode_conv_to_printable_fixedwidth(uni_stat_t *stats, const char *src, unsigned width)
 687 {
 688         return unicode_conv_to_printable2(stats, src, width, UNI_FLAG_PAD);
 689 }
 690
 691 #ifdef UNUSED
 692 unsigned FAST_FUNC unicode_padding_to_width(unsigned width, const char *src)
 693 {
 694         if (unicode_status != UNICODE_ON) {
 695                 return width - strnlen(src, width);
 696         }
 697
 698         while (1) {
 699                 int w;
 700                 wchar_t wc;
 701
 702 #if ENABLE_LOCALE_SUPPORT
 703                 {
 704                         mbstate_t mbst = { 0 };
 705                         ssize_t rc = mbsrtowcs(&wc, &src, 1, &mbst);
 706                         if (rc <= 0) /* error, or end-of-string */
 707                                 return width;
 708                 }
 709 #else
 710                 src = mbstowc_internal(&wc, src);
 711                 if (wc == ERROR_WCHAR || wc == 0) /* error, or end-of-string */
 712                         return width;
 713 #endif
 714                 w = wcwidth(wc);
 715                 if (w < 0) /* non-printable wchar */
 716                         return width;
 717                 width -= w;
 718                 if ((int)width <= 0) /* string is longer than width */
 719                         return 0;
 720         }
 721 }
 722 #endif