utf8.c

   1 #include "git-compat-util.h"
   2 #include "utf8.h"
   3
   4 /* This code is originally from http://www.cl.cam.ac.uk/~mgk25/ucs/ */
   5
   6 struct interval {
   7   int first;
   8   int last;
   9 };
  10
  11 /* auxiliary function for binary search in interval table */
  12 static int bisearch(wchar_t ucs, const struct interval *table, int max) {
  13         int min = 0;
  14         int mid;
  15
  16         if (ucs < table[0].first || ucs > table[max].last)
  17                 return 0;
  18         while (max >= min) {
  19                 mid = (min + max) / 2;
  20                 if (ucs > table[mid].last)
  21                         min = mid + 1;
  22                 else if (ucs < table[mid].first)
  23                         max = mid - 1;
  24                 else
  25                         return 1;
  26         }
  27
  28         return 0;
  29 }
  30
  31 /* The following two functions define the column width of an ISO 10646
  32  * character as follows:
  33  *
  34  *    - The null character (U+0000) has a column width of 0.
  35  *
  36  *    - Other C0/C1 control characters and DEL will lead to a return
  37  *      value of -1.
  38  *
  39  *    - Non-spacing and enclosing combining characters (general
  40  *      category code Mn or Me in the Unicode database) have a
  41  *      column width of 0.
  42  *
  43  *    - SOFT HYPHEN (U+00AD) has a column width of 1.
  44  *
  45  *    - Other format characters (general category code Cf in the Unicode
  46  *      database) and ZERO WIDTH SPACE (U+200B) have a column width of 0.
  47  *
  48  *    - Hangul Jamo medial vowels and final consonants (U+1160-U+11FF)
  49  *      have a column width of 0.
  50  *
  51  *    - Spacing characters in the East Asian Wide (W) or East Asian
  52  *      Full-width (F) category as defined in Unicode Technical
  53  *      Report #11 have a column width of 2.
  54  *
  55  *    - All remaining characters (including all printable
  56  *      ISO 8859-1 and WGL4 characters, Unicode control characters,
  57  *      etc.) have a column width of 1.
  58  *
  59  * This implementation assumes that wchar_t characters are encoded
  60  * in ISO 10646.
  61  */
  62
  63 static int wcwidth(wchar_t ch)
  64 {
  65         /*
  66          * Sorted list of non-overlapping intervals of non-spacing characters,
  67          * generated by
  68          *   "uniset +cat=Me +cat=Mn +cat=Cf -00AD +1160-11FF +200B c".
  69          */
  70         static const struct interval combining[] = {
  71                 { 0x0300, 0x0357 }, { 0x035D, 0x036F }, { 0x0483, 0x0486 },
  72                 { 0x0488, 0x0489 }, { 0x0591, 0x05A1 }, { 0x05A3, 0x05B9 },
  73                 { 0x05BB, 0x05BD }, { 0x05BF, 0x05BF }, { 0x05C1, 0x05C2 },
  74                 { 0x05C4, 0x05C4 }, { 0x0600, 0x0603 }, { 0x0610, 0x0615 },
  75                 { 0x064B, 0x0658 }, { 0x0670, 0x0670 }, { 0x06D6, 0x06E4 },
  76                 { 0x06E7, 0x06E8 }, { 0x06EA, 0x06ED }, { 0x070F, 0x070F },
  77                 { 0x0711, 0x0711 }, { 0x0730, 0x074A }, { 0x07A6, 0x07B0 },
  78                 { 0x0901, 0x0902 }, { 0x093C, 0x093C }, { 0x0941, 0x0948 },
  79                 { 0x094D, 0x094D }, { 0x0951, 0x0954 }, { 0x0962, 0x0963 },
  80                 { 0x0981, 0x0981 }, { 0x09BC, 0x09BC }, { 0x09C1, 0x09C4 },
  81                 { 0x09CD, 0x09CD }, { 0x09E2, 0x09E3 }, { 0x0A01, 0x0A02 },
  82                 { 0x0A3C, 0x0A3C }, { 0x0A41, 0x0A42 }, { 0x0A47, 0x0A48 },
  83                 { 0x0A4B, 0x0A4D }, { 0x0A70, 0x0A71 }, { 0x0A81, 0x0A82 },
  84                 { 0x0ABC, 0x0ABC }, { 0x0AC1, 0x0AC5 }, { 0x0AC7, 0x0AC8 },
  85                 { 0x0ACD, 0x0ACD }, { 0x0AE2, 0x0AE3 }, { 0x0B01, 0x0B01 },
  86                 { 0x0B3C, 0x0B3C }, { 0x0B3F, 0x0B3F }, { 0x0B41, 0x0B43 },
  87                 { 0x0B4D, 0x0B4D }, { 0x0B56, 0x0B56 }, { 0x0B82, 0x0B82 },
  88                 { 0x0BC0, 0x0BC0 }, { 0x0BCD, 0x0BCD }, { 0x0C3E, 0x0C40 },
  89                 { 0x0C46, 0x0C48 }, { 0x0C4A, 0x0C4D }, { 0x0C55, 0x0C56 },
  90                 { 0x0CBC, 0x0CBC }, { 0x0CBF, 0x0CBF }, { 0x0CC6, 0x0CC6 },
  91                 { 0x0CCC, 0x0CCD }, { 0x0D41, 0x0D43 }, { 0x0D4D, 0x0D4D },
  92                 { 0x0DCA, 0x0DCA }, { 0x0DD2, 0x0DD4 }, { 0x0DD6, 0x0DD6 },
  93                 { 0x0E31, 0x0E31 }, { 0x0E34, 0x0E3A }, { 0x0E47, 0x0E4E },
  94                 { 0x0EB1, 0x0EB1 }, { 0x0EB4, 0x0EB9 }, { 0x0EBB, 0x0EBC },
  95                 { 0x0EC8, 0x0ECD }, { 0x0F18, 0x0F19 }, { 0x0F35, 0x0F35 },
  96                 { 0x0F37, 0x0F37 }, { 0x0F39, 0x0F39 }, { 0x0F71, 0x0F7E },
  97                 { 0x0F80, 0x0F84 }, { 0x0F86, 0x0F87 }, { 0x0F90, 0x0F97 },
  98                 { 0x0F99, 0x0FBC }, { 0x0FC6, 0x0FC6 }, { 0x102D, 0x1030 },
  99                 { 0x1032, 0x1032 }, { 0x1036, 0x1037 }, { 0x1039, 0x1039 },
 100                 { 0x1058, 0x1059 }, { 0x1160, 0x11FF }, { 0x1712, 0x1714 },
 101                 { 0x1732, 0x1734 }, { 0x1752, 0x1753 }, { 0x1772, 0x1773 },
 102                 { 0x17B4, 0x17B5 }, { 0x17B7, 0x17BD }, { 0x17C6, 0x17C6 },
 103                 { 0x17C9, 0x17D3 }, { 0x17DD, 0x17DD }, { 0x180B, 0x180D },
 104                 { 0x18A9, 0x18A9 }, { 0x1920, 0x1922 }, { 0x1927, 0x1928 },
 105                 { 0x1932, 0x1932 }, { 0x1939, 0x193B }, { 0x200B, 0x200F },
 106                 { 0x202A, 0x202E }, { 0x2060, 0x2063 }, { 0x206A, 0x206F },
 107                 { 0x20D0, 0x20EA }, { 0x302A, 0x302F }, { 0x3099, 0x309A },
 108                 { 0xFB1E, 0xFB1E }, { 0xFE00, 0xFE0F }, { 0xFE20, 0xFE23 },
 109                 { 0xFEFF, 0xFEFF }, { 0xFFF9, 0xFFFB }, { 0x1D167, 0x1D169 },
 110                 { 0x1D173, 0x1D182 }, { 0x1D185, 0x1D18B },
 111                 { 0x1D1AA, 0x1D1AD }, { 0xE0001, 0xE0001 },
 112                 { 0xE0020, 0xE007F }, { 0xE0100, 0xE01EF }
 113         };
 114
 115         /* test for 8-bit control characters */
 116         if (ch == 0)
 117                 return 0;
 118         if (ch < 32 || (ch >= 0x7f && ch < 0xa0))
 119                 return -1;
 120
 121         /* binary search in table of non-spacing characters */
 122         if (bisearch(ch, combining, sizeof(combining)
 123                                 / sizeof(struct interval) - 1))
 124                 return 0;
 125
 126         /*
 127          * If we arrive here, ch is neither a combining nor a C0/C1
 128          * control character.
 129          */
 130
 131         return 1 +
 132                 (ch >= 0x1100 &&
 133                     /* Hangul Jamo init. consonants */
 134                  (ch <= 0x115f ||
 135                   ch == 0x2329 || ch == 0x232a ||
 136                   /* CJK ... Yi */
 137                   (ch >= 0x2e80 && ch <= 0xa4cf &&
 138                    ch != 0x303f) ||
 139                   /* Hangul Syllables */
 140                   (ch >= 0xac00 && ch <= 0xd7a3) ||
 141                   /* CJK Compatibility Ideographs */
 142                   (ch >= 0xf900 && ch <= 0xfaff) ||
 143                   /* CJK Compatibility Forms */
 144                   (ch >= 0xfe30 && ch <= 0xfe6f) ||
 145                   /* Fullwidth Forms */
 146                   (ch >= 0xff00 && ch <= 0xff60) ||
 147                   (ch >= 0xffe0 && ch <= 0xffe6) ||
 148                   (ch >= 0x20000 && ch <= 0x2fffd) ||
 149                   (ch >= 0x30000 && ch <= 0x3fffd)));
 150 }
 151
 152 /*
 153  * This function returns the number of columns occupied by the character
 154  * pointed to by the variable start. The pointer is updated to point at
 155  * the next character. If it was not valid UTF-8, the pointer is set to NULL.
 156  */
 157 int utf8_width(const char **start)
 158 {
 159         unsigned char *s = (unsigned char *)*start;
 160         wchar_t ch;
 161
 162         if (*s < 0x80) {
 163                 /* 0xxxxxxx */
 164                 ch = *s;
 165                 *start += 1;
 166         } else if ((s[0] & 0xe0) == 0xc0) {
 167                 /* 110XXXXx 10xxxxxx */
 168                 if ((s[1] & 0xc0) != 0x80 ||
 169                                 /* overlong? */
 170                                 (s[0] & 0xfe) == 0xc0)
 171                         goto invalid;
 172                 ch = ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
 173                 *start += 2;
 174         } else if ((s[0] & 0xf0) == 0xe0) {
 175                 /* 1110XXXX 10Xxxxxx 10xxxxxx */
 176                 if ((s[1] & 0xc0) != 0x80 ||
 177                                 (s[2] & 0xc0) != 0x80 ||
 178                                 /* overlong? */
 179                                 (s[0] == 0xe0 && (s[1] & 0xe0) == 0x80) ||
 180                                 /* surrogate? */
 181                                 (s[0] == 0xed && (s[1] & 0xe0) == 0xa0) ||
 182                                 /* U+FFFE or U+FFFF? */
 183                                 (s[0] == 0xef && s[1] == 0xbf &&
 184                                  (s[2] & 0xfe) == 0xbe))
 185                         goto invalid;
 186                 ch = ((s[0] & 0x0f) << 12) |
 187                         ((s[1] & 0x3f) << 6) | (s[2] & 0x3f);
 188                 *start += 3;
 189         } else if ((s[0] & 0xf8) == 0xf0) {
 190                 /* 11110XXX 10XXxxxx 10xxxxxx 10xxxxxx */
 191                 if ((s[1] & 0xc0) != 0x80 ||
 192                                 (s[2] & 0xc0) != 0x80 ||
 193                                 (s[3] & 0xc0) != 0x80 ||
 194                                 /* overlong? */
 195                                 (s[0] == 0xf0 && (s[1] & 0xf0) == 0x80) ||
 196                                 /* > U+10FFFF? */
 197                                 (s[0] == 0xf4 && s[1] > 0x8f) || s[0] > 0xf4)
 198                         goto invalid;
 199                 ch = ((s[0] & 0x07) << 18) | ((s[1] & 0x3f) << 12) |
 200                         ((s[2] & 0x3f) << 6) | (s[3] & 0x3f);
 201                 *start += 4;
 202         } else {
 203 invalid:
 204                 *start = NULL;
 205                 return 0;
 206         }
 207
 208         return wcwidth(ch);
 209 }
 210
 211 int is_utf8(const char *text)
 212 {
 213         while (*text) {
 214                 if (*text == '\n' || *text == '\t' || *text == '\r') {
 215                         text++;
 216                         continue;
 217                 }
 218                 utf8_width(&text);
 219                 if (!text)
 220                         return 0;
 221         }
 222         return 1;
 223 }
 224
 225 static void print_spaces(int count)
 226 {
 227         static const char s[] = "                    ";
 228         while (count >= sizeof(s)) {
 229                 fwrite(s, sizeof(s) - 1, 1, stdout);
 230                 count -= sizeof(s) - 1;
 231         }
 232         fwrite(s, count, 1, stdout);
 233 }
 234
 235 /*
 236  * Wrap the text, if necessary. The variable indent is the indent for the
 237  * first line, indent2 is the indent for all other lines.
 238  */
 239 void print_wrapped_text(const char *text, int indent, int indent2, int width)
 240 {
 241         int w = indent, assume_utf8 = is_utf8(text);
 242         const char *bol = text, *space = NULL;
 243
 244         for (;;) {
 245                 char c = *text;
 246                 if (!c || isspace(c)) {
 247                         if (w < width || !space) {
 248                                 const char *start = bol;
 249                                 if (space)
 250                                         start = space;
 251                                 else
 252                                         print_spaces(indent);
 253                                 fwrite(start, text - start, 1, stdout);
 254                                 if (!c) {
 255                                         putchar('\n');
 256                                         return;
 257                                 } else if (c == '\t')
 258                                         w |= 0x07;
 259                                 space = text;
 260                                 w++;
 261                                 text++;
 262                         }
 263                         else {
 264                                 putchar('\n');
 265                                 text = bol = space + 1;
 266                                 space = NULL;
 267                                 w = indent = indent2;
 268                         }
 269                         continue;
 270                 }
 271                 if (assume_utf8)
 272                         w += utf8_width(&text);
 273                 else {
 274                         w++;
 275                         text++;
 276                 }
 277         }
 278 }
 279
 280 int is_encoding_utf8(const char *name)
 281 {
 282         if (!name)
 283                 return 1;
 284         if (!strcasecmp(name, "utf-8") || !strcasecmp(name, "utf8"))
 285                 return 1;
 286         return 0;
 287 }
 288
 289 /*
 290  * Given a buffer and its encoding, return it re-encoded
 291  * with iconv.  If the conversion fails, returns NULL.
 292  */
 293 #ifndef NO_ICONV
 294 char *reencode_string(const char *in, const char *out_encoding, const char *in_encoding)
 295 {
 296         iconv_t conv;
 297         size_t insz, outsz, outalloc;
 298         char *out, *outpos, *cp;
 299
 300         if (!in_encoding)
 301                 return NULL;
 302         conv = iconv_open(out_encoding, in_encoding);
 303         if (conv == (iconv_t) -1)
 304                 return NULL;
 305         insz = strlen(in);
 306         outsz = insz;
 307         outalloc = outsz + 1; /* for terminating NUL */
 308         out = xmalloc(outalloc);
 309         outpos = out;
 310         cp = (char *)in;
 311
 312         while (1) {
 313                 size_t cnt = iconv(conv, &cp, &insz, &outpos, &outsz);
 314
 315                 if (cnt == -1) {
 316                         size_t sofar;
 317                         if (errno != E2BIG) {
 318                                 free(out);
 319                                 iconv_close(conv);
 320                                 return NULL;
 321                         }
 322                         /* insz has remaining number of bytes.
 323                          * since we started outsz the same as insz,
 324                          * it is likely that insz is not enough for
 325                          * converting the rest.
 326                          */
 327                         sofar = outpos - out;
 328                         outalloc = sofar + insz * 2 + 32;
 329                         out = xrealloc(out, outalloc);
 330                         outpos = out + sofar;
 331                         outsz = outalloc - sofar - 1;
 332                 }
 333                 else {
 334                         *outpos = '\0';
 335                         break;
 336                 }
 337         }
 338         iconv_close(conv);
 339         return out;
 340 }
 341 #endif