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 typedef unsigned int ucs_char_t;  /* assuming 32bit int */
   7
   8 struct interval {
   9   int first;
  10   int last;
  11 };
  12
  13 /* auxiliary function for binary search in interval table */
  14 static int bisearch(ucs_char_t ucs, const struct interval *table, int max) {
  15         int min = 0;
  16         int mid;
  17
  18         if (ucs < table[0].first || ucs > table[max].last)
  19                 return 0;
  20         while (max >= min) {
  21                 mid = (min + max) / 2;
  22                 if (ucs > table[mid].last)
  23                         min = mid + 1;
  24                 else if (ucs < table[mid].first)
  25                         max = mid - 1;
  26                 else
  27                         return 1;
  28         }
  29
  30         return 0;
  31 }
  32
  33 /* The following two functions define the column width of an ISO 10646
  34  * character as follows:
  35  *
  36  *    - The null character (U+0000) has a column width of 0.
  37  *
  38  *    - Other C0/C1 control characters and DEL will lead to a return
  39  *      value of -1.
  40  *
  41  *    - Non-spacing and enclosing combining characters (general
  42  *      category code Mn or Me in the Unicode database) have a
  43  *      column width of 0.
  44  *
  45  *    - SOFT HYPHEN (U+00AD) has a column width of 1.
  46  *
  47  *    - Other format characters (general category code Cf in the Unicode
  48  *      database) and ZERO WIDTH SPACE (U+200B) have a column width of 0.
  49  *
  50  *    - Hangul Jamo medial vowels and final consonants (U+1160-U+11FF)
  51  *      have a column width of 0.
  52  *
  53  *    - Spacing characters in the East Asian Wide (W) or East Asian
  54  *      Full-width (F) category as defined in Unicode Technical
  55  *      Report #11 have a column width of 2.
  56  *
  57  *    - All remaining characters (including all printable
  58  *      ISO 8859-1 and WGL4 characters, Unicode control characters,
  59  *      etc.) have a column width of 1.
  60  *
  61  * This implementation assumes that ucs_char_t characters are encoded
  62  * in ISO 10646.
  63  */
  64
  65 static int wcwidth(ucs_char_t ch)
  66 {
  67         /*
  68          * Sorted list of non-overlapping intervals of non-spacing characters,
  69          * generated by
  70          *   "uniset +cat=Me +cat=Mn +cat=Cf -00AD +1160-11FF +200B c".
  71          */
  72         static const struct interval combining[] = {
  73                 { 0x0300, 0x0357 }, { 0x035D, 0x036F }, { 0x0483, 0x0486 },
  74                 { 0x0488, 0x0489 }, { 0x0591, 0x05A1 }, { 0x05A3, 0x05B9 },
  75                 { 0x05BB, 0x05BD }, { 0x05BF, 0x05BF }, { 0x05C1, 0x05C2 },
  76                 { 0x05C4, 0x05C4 }, { 0x0600, 0x0603 }, { 0x0610, 0x0615 },
  77                 { 0x064B, 0x0658 }, { 0x0670, 0x0670 }, { 0x06D6, 0x06E4 },
  78                 { 0x06E7, 0x06E8 }, { 0x06EA, 0x06ED }, { 0x070F, 0x070F },
  79                 { 0x0711, 0x0711 }, { 0x0730, 0x074A }, { 0x07A6, 0x07B0 },
  80                 { 0x0901, 0x0902 }, { 0x093C, 0x093C }, { 0x0941, 0x0948 },
  81                 { 0x094D, 0x094D }, { 0x0951, 0x0954 }, { 0x0962, 0x0963 },
  82                 { 0x0981, 0x0981 }, { 0x09BC, 0x09BC }, { 0x09C1, 0x09C4 },
  83                 { 0x09CD, 0x09CD }, { 0x09E2, 0x09E3 }, { 0x0A01, 0x0A02 },
  84                 { 0x0A3C, 0x0A3C }, { 0x0A41, 0x0A42 }, { 0x0A47, 0x0A48 },
  85                 { 0x0A4B, 0x0A4D }, { 0x0A70, 0x0A71 }, { 0x0A81, 0x0A82 },
  86                 { 0x0ABC, 0x0ABC }, { 0x0AC1, 0x0AC5 }, { 0x0AC7, 0x0AC8 },
  87                 { 0x0ACD, 0x0ACD }, { 0x0AE2, 0x0AE3 }, { 0x0B01, 0x0B01 },
  88                 { 0x0B3C, 0x0B3C }, { 0x0B3F, 0x0B3F }, { 0x0B41, 0x0B43 },
  89                 { 0x0B4D, 0x0B4D }, { 0x0B56, 0x0B56 }, { 0x0B82, 0x0B82 },
  90                 { 0x0BC0, 0x0BC0 }, { 0x0BCD, 0x0BCD }, { 0x0C3E, 0x0C40 },
  91                 { 0x0C46, 0x0C48 }, { 0x0C4A, 0x0C4D }, { 0x0C55, 0x0C56 },
  92                 { 0x0CBC, 0x0CBC }, { 0x0CBF, 0x0CBF }, { 0x0CC6, 0x0CC6 },
  93                 { 0x0CCC, 0x0CCD }, { 0x0D41, 0x0D43 }, { 0x0D4D, 0x0D4D },
  94                 { 0x0DCA, 0x0DCA }, { 0x0DD2, 0x0DD4 }, { 0x0DD6, 0x0DD6 },
  95                 { 0x0E31, 0x0E31 }, { 0x0E34, 0x0E3A }, { 0x0E47, 0x0E4E },
  96                 { 0x0EB1, 0x0EB1 }, { 0x0EB4, 0x0EB9 }, { 0x0EBB, 0x0EBC },
  97                 { 0x0EC8, 0x0ECD }, { 0x0F18, 0x0F19 }, { 0x0F35, 0x0F35 },
  98                 { 0x0F37, 0x0F37 }, { 0x0F39, 0x0F39 }, { 0x0F71, 0x0F7E },
  99                 { 0x0F80, 0x0F84 }, { 0x0F86, 0x0F87 }, { 0x0F90, 0x0F97 },
 100                 { 0x0F99, 0x0FBC }, { 0x0FC6, 0x0FC6 }, { 0x102D, 0x1030 },
 101                 { 0x1032, 0x1032 }, { 0x1036, 0x1037 }, { 0x1039, 0x1039 },
 102                 { 0x1058, 0x1059 }, { 0x1160, 0x11FF }, { 0x1712, 0x1714 },
 103                 { 0x1732, 0x1734 }, { 0x1752, 0x1753 }, { 0x1772, 0x1773 },
 104                 { 0x17B4, 0x17B5 }, { 0x17B7, 0x17BD }, { 0x17C6, 0x17C6 },
 105                 { 0x17C9, 0x17D3 }, { 0x17DD, 0x17DD }, { 0x180B, 0x180D },
 106                 { 0x18A9, 0x18A9 }, { 0x1920, 0x1922 }, { 0x1927, 0x1928 },
 107                 { 0x1932, 0x1932 }, { 0x1939, 0x193B }, { 0x200B, 0x200F },
 108                 { 0x202A, 0x202E }, { 0x2060, 0x2063 }, { 0x206A, 0x206F },
 109                 { 0x20D0, 0x20EA }, { 0x302A, 0x302F }, { 0x3099, 0x309A },
 110                 { 0xFB1E, 0xFB1E }, { 0xFE00, 0xFE0F }, { 0xFE20, 0xFE23 },
 111                 { 0xFEFF, 0xFEFF }, { 0xFFF9, 0xFFFB }, { 0x1D167, 0x1D169 },
 112                 { 0x1D173, 0x1D182 }, { 0x1D185, 0x1D18B },
 113                 { 0x1D1AA, 0x1D1AD }, { 0xE0001, 0xE0001 },
 114                 { 0xE0020, 0xE007F }, { 0xE0100, 0xE01EF }
 115         };
 116
 117         /* test for 8-bit control characters */
 118         if (ch == 0)
 119                 return 0;
 120         if (ch < 32 || (ch >= 0x7f && ch < 0xa0))
 121                 return -1;
 122
 123         /* binary search in table of non-spacing characters */
 124         if (bisearch(ch, combining, sizeof(combining)
 125                                 / sizeof(struct interval) - 1))
 126                 return 0;
 127
 128         /*
 129          * If we arrive here, ch is neither a combining nor a C0/C1
 130          * control character.
 131          */
 132
 133         return 1 +
 134                 (ch >= 0x1100 &&
 135                     /* Hangul Jamo init. consonants */
 136                  (ch <= 0x115f ||
 137                   ch == 0x2329 || ch == 0x232a ||
 138                   /* CJK ... Yi */
 139                   (ch >= 0x2e80 && ch <= 0xa4cf &&
 140                    ch != 0x303f) ||
 141                   /* Hangul Syllables */
 142                   (ch >= 0xac00 && ch <= 0xd7a3) ||
 143                   /* CJK Compatibility Ideographs */
 144                   (ch >= 0xf900 && ch <= 0xfaff) ||
 145                   /* CJK Compatibility Forms */
 146                   (ch >= 0xfe30 && ch <= 0xfe6f) ||
 147                   /* Fullwidth Forms */
 148                   (ch >= 0xff00 && ch <= 0xff60) ||
 149                   (ch >= 0xffe0 && ch <= 0xffe6) ||
 150                   (ch >= 0x20000 && ch <= 0x2fffd) ||
 151                   (ch >= 0x30000 && ch <= 0x3fffd)));
 152 }
 153
 154 /*
 155  * This function returns the number of columns occupied by the character
 156  * pointed to by the variable start. The pointer is updated to point at
 157  * the next character. If it was not valid UTF-8, the pointer is set to NULL.
 158  */
 159 int utf8_width(const char **start)
 160 {
 161         unsigned char *s = (unsigned char *)*start;
 162         ucs_char_t ch;
 163
 164         if (*s < 0x80) {
 165                 /* 0xxxxxxx */
 166                 ch = *s;
 167                 *start += 1;
 168         } else if ((s[0] & 0xe0) == 0xc0) {
 169                 /* 110XXXXx 10xxxxxx */
 170                 if ((s[1] & 0xc0) != 0x80 ||
 171                                 /* overlong? */
 172                                 (s[0] & 0xfe) == 0xc0)
 173                         goto invalid;
 174                 ch = ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
 175                 *start += 2;
 176         } else if ((s[0] & 0xf0) == 0xe0) {
 177                 /* 1110XXXX 10Xxxxxx 10xxxxxx */
 178                 if ((s[1] & 0xc0) != 0x80 ||
 179                                 (s[2] & 0xc0) != 0x80 ||
 180                                 /* overlong? */
 181                                 (s[0] == 0xe0 && (s[1] & 0xe0) == 0x80) ||
 182                                 /* surrogate? */
 183                                 (s[0] == 0xed && (s[1] & 0xe0) == 0xa0) ||
 184                                 /* U+FFFE or U+FFFF? */
 185                                 (s[0] == 0xef && s[1] == 0xbf &&
 186                                  (s[2] & 0xfe) == 0xbe))
 187                         goto invalid;
 188                 ch = ((s[0] & 0x0f) << 12) |
 189                         ((s[1] & 0x3f) << 6) | (s[2] & 0x3f);
 190                 *start += 3;
 191         } else if ((s[0] & 0xf8) == 0xf0) {
 192                 /* 11110XXX 10XXxxxx 10xxxxxx 10xxxxxx */
 193                 if ((s[1] & 0xc0) != 0x80 ||
 194                                 (s[2] & 0xc0) != 0x80 ||
 195                                 (s[3] & 0xc0) != 0x80 ||
 196                                 /* overlong? */
 197                                 (s[0] == 0xf0 && (s[1] & 0xf0) == 0x80) ||
 198                                 /* > U+10FFFF? */
 199                                 (s[0] == 0xf4 && s[1] > 0x8f) || s[0] > 0xf4)
 200                         goto invalid;
 201                 ch = ((s[0] & 0x07) << 18) | ((s[1] & 0x3f) << 12) |
 202                         ((s[2] & 0x3f) << 6) | (s[3] & 0x3f);
 203                 *start += 4;
 204         } else {
 205 invalid:
 206                 *start = NULL;
 207                 return 0;
 208         }
 209
 210         return wcwidth(ch);
 211 }
 212
 213 int is_utf8(const char *text)
 214 {
 215         while (*text) {
 216                 if (*text == '\n' || *text == '\t' || *text == '\r') {
 217                         text++;
 218                         continue;
 219                 }
 220                 utf8_width(&text);
 221                 if (!text)
 222                         return 0;
 223         }
 224         return 1;
 225 }
 226
 227 static void print_spaces(int count)
 228 {
 229         static const char s[] = "                    ";
 230         while (count >= sizeof(s)) {
 231                 fwrite(s, sizeof(s) - 1, 1, stdout);
 232                 count -= sizeof(s) - 1;
 233         }
 234         fwrite(s, count, 1, stdout);
 235 }
 236
 237 /*
 238  * Wrap the text, if necessary. The variable indent is the indent for the
 239  * first line, indent2 is the indent for all other lines.
 240  */
 241 void print_wrapped_text(const char *text, int indent, int indent2, int width)
 242 {
 243         int w = indent, assume_utf8 = is_utf8(text);
 244         const char *bol = text, *space = NULL;
 245
 246         for (;;) {
 247                 char c = *text;
 248                 if (!c || isspace(c)) {
 249                         if (w < width || !space) {
 250                                 const char *start = bol;
 251                                 if (space)
 252                                         start = space;
 253                                 else
 254                                         print_spaces(indent);
 255                                 fwrite(start, text - start, 1, stdout);
 256                                 if (!c) {
 257                                         putchar('\n');
 258                                         return;
 259                                 } else if (c == '\t')
 260                                         w |= 0x07;
 261                                 space = text;
 262                                 w++;
 263                                 text++;
 264                         }
 265                         else {
 266                                 putchar('\n');
 267                                 text = bol = space + 1;
 268                                 space = NULL;
 269                                 w = indent = indent2;
 270                         }
 271                         continue;
 272                 }
 273                 if (assume_utf8)
 274                         w += utf8_width(&text);
 275                 else {
 276                         w++;
 277                         text++;
 278                 }
 279         }
 280 }
 281
 282 int is_encoding_utf8(const char *name)
 283 {
 284         if (!name)
 285                 return 1;
 286         if (!strcasecmp(name, "utf-8") || !strcasecmp(name, "utf8"))
 287                 return 1;
 288         return 0;
 289 }
 290
 291 /*
 292  * Given a buffer and its encoding, return it re-encoded
 293  * with iconv.  If the conversion fails, returns NULL.
 294  */
 295 #ifndef NO_ICONV
 296 char *reencode_string(const char *in, const char *out_encoding, const char *in_encoding)
 297 {
 298         iconv_t conv;
 299         size_t insz, outsz, outalloc;
 300         char *out, *outpos, *cp;
 301
 302         if (!in_encoding)
 303                 return NULL;
 304         conv = iconv_open(out_encoding, in_encoding);
 305         if (conv == (iconv_t) -1)
 306                 return NULL;
 307         insz = strlen(in);
 308         outsz = insz;
 309         outalloc = outsz + 1; /* for terminating NUL */
 310         out = xmalloc(outalloc);
 311         outpos = out;
 312         cp = (char *)in;
 313
 314         while (1) {
 315                 size_t cnt = iconv(conv, &cp, &insz, &outpos, &outsz);
 316
 317                 if (cnt == -1) {
 318                         size_t sofar;
 319                         if (errno != E2BIG) {
 320                                 free(out);
 321                                 iconv_close(conv);
 322                                 return NULL;
 323                         }
 324                         /* insz has remaining number of bytes.
 325                          * since we started outsz the same as insz,
 326                          * it is likely that insz is not enough for
 327                          * converting the rest.
 328                          */
 329                         sofar = outpos - out;
 330                         outalloc = sofar + insz * 2 + 32;
 331                         out = xrealloc(out, outalloc);
 332                         outpos = out + sofar;
 333                         outsz = outalloc - sofar - 1;
 334                 }
 335                 else {
 336                         *outpos = '\0';
 337                         break;
 338                 }
 339         }
 340         iconv_close(conv);
 341         return out;
 342 }
 343 #endif