utf8.c: move display_mode_esc_sequence_len() for use by other functions
[git/git.git] / utf8.c
CommitLineData
9e832665 1#include "git-compat-util.h"
a94410c8 2#include "strbuf.h"
9e832665
JS
3#include "utf8.h"
4
5/* This code is originally from http://www.cl.cam.ac.uk/~mgk25/ucs/ */
6
7struct interval {
8 int first;
9 int last;
10};
11
4247fe79
NTND
12static size_t display_mode_esc_sequence_len(const char *s)
13{
14 const char *p = s;
15 if (*p++ != '\033')
16 return 0;
17 if (*p++ != '[')
18 return 0;
19 while (isdigit(*p) || *p == ';')
20 p++;
21 if (*p++ != 'm')
22 return 0;
23 return p - s;
24}
25
9e832665 26/* auxiliary function for binary search in interval table */
f3fa1838
JH
27static int bisearch(ucs_char_t ucs, const struct interval *table, int max)
28{
9e832665
JS
29 int min = 0;
30 int mid;
31
32 if (ucs < table[0].first || ucs > table[max].last)
33 return 0;
34 while (max >= min) {
35 mid = (min + max) / 2;
36 if (ucs > table[mid].last)
37 min = mid + 1;
38 else if (ucs < table[mid].first)
39 max = mid - 1;
40 else
41 return 1;
42 }
43
44 return 0;
45}
46
47/* The following two functions define the column width of an ISO 10646
48 * character as follows:
49 *
50 * - The null character (U+0000) has a column width of 0.
51 *
52 * - Other C0/C1 control characters and DEL will lead to a return
53 * value of -1.
54 *
55 * - Non-spacing and enclosing combining characters (general
56 * category code Mn or Me in the Unicode database) have a
57 * column width of 0.
58 *
59 * - SOFT HYPHEN (U+00AD) has a column width of 1.
60 *
61 * - Other format characters (general category code Cf in the Unicode
62 * database) and ZERO WIDTH SPACE (U+200B) have a column width of 0.
63 *
64 * - Hangul Jamo medial vowels and final consonants (U+1160-U+11FF)
65 * have a column width of 0.
66 *
67 * - Spacing characters in the East Asian Wide (W) or East Asian
68 * Full-width (F) category as defined in Unicode Technical
69 * Report #11 have a column width of 2.
70 *
71 * - All remaining characters (including all printable
72 * ISO 8859-1 and WGL4 characters, Unicode control characters,
73 * etc.) have a column width of 1.
74 *
28321145 75 * This implementation assumes that ucs_char_t characters are encoded
9e832665
JS
76 * in ISO 10646.
77 */
78
b51be13c 79static int git_wcwidth(ucs_char_t ch)
9e832665
JS
80{
81 /*
82 * Sorted list of non-overlapping intervals of non-spacing characters,
83 * generated by
84 * "uniset +cat=Me +cat=Mn +cat=Cf -00AD +1160-11FF +200B c".
85 */
86 static const struct interval combining[] = {
87 { 0x0300, 0x0357 }, { 0x035D, 0x036F }, { 0x0483, 0x0486 },
88 { 0x0488, 0x0489 }, { 0x0591, 0x05A1 }, { 0x05A3, 0x05B9 },
89 { 0x05BB, 0x05BD }, { 0x05BF, 0x05BF }, { 0x05C1, 0x05C2 },
90 { 0x05C4, 0x05C4 }, { 0x0600, 0x0603 }, { 0x0610, 0x0615 },
91 { 0x064B, 0x0658 }, { 0x0670, 0x0670 }, { 0x06D6, 0x06E4 },
92 { 0x06E7, 0x06E8 }, { 0x06EA, 0x06ED }, { 0x070F, 0x070F },
93 { 0x0711, 0x0711 }, { 0x0730, 0x074A }, { 0x07A6, 0x07B0 },
94 { 0x0901, 0x0902 }, { 0x093C, 0x093C }, { 0x0941, 0x0948 },
95 { 0x094D, 0x094D }, { 0x0951, 0x0954 }, { 0x0962, 0x0963 },
96 { 0x0981, 0x0981 }, { 0x09BC, 0x09BC }, { 0x09C1, 0x09C4 },
97 { 0x09CD, 0x09CD }, { 0x09E2, 0x09E3 }, { 0x0A01, 0x0A02 },
98 { 0x0A3C, 0x0A3C }, { 0x0A41, 0x0A42 }, { 0x0A47, 0x0A48 },
99 { 0x0A4B, 0x0A4D }, { 0x0A70, 0x0A71 }, { 0x0A81, 0x0A82 },
100 { 0x0ABC, 0x0ABC }, { 0x0AC1, 0x0AC5 }, { 0x0AC7, 0x0AC8 },
101 { 0x0ACD, 0x0ACD }, { 0x0AE2, 0x0AE3 }, { 0x0B01, 0x0B01 },
102 { 0x0B3C, 0x0B3C }, { 0x0B3F, 0x0B3F }, { 0x0B41, 0x0B43 },
103 { 0x0B4D, 0x0B4D }, { 0x0B56, 0x0B56 }, { 0x0B82, 0x0B82 },
104 { 0x0BC0, 0x0BC0 }, { 0x0BCD, 0x0BCD }, { 0x0C3E, 0x0C40 },
105 { 0x0C46, 0x0C48 }, { 0x0C4A, 0x0C4D }, { 0x0C55, 0x0C56 },
106 { 0x0CBC, 0x0CBC }, { 0x0CBF, 0x0CBF }, { 0x0CC6, 0x0CC6 },
107 { 0x0CCC, 0x0CCD }, { 0x0D41, 0x0D43 }, { 0x0D4D, 0x0D4D },
108 { 0x0DCA, 0x0DCA }, { 0x0DD2, 0x0DD4 }, { 0x0DD6, 0x0DD6 },
109 { 0x0E31, 0x0E31 }, { 0x0E34, 0x0E3A }, { 0x0E47, 0x0E4E },
110 { 0x0EB1, 0x0EB1 }, { 0x0EB4, 0x0EB9 }, { 0x0EBB, 0x0EBC },
111 { 0x0EC8, 0x0ECD }, { 0x0F18, 0x0F19 }, { 0x0F35, 0x0F35 },
112 { 0x0F37, 0x0F37 }, { 0x0F39, 0x0F39 }, { 0x0F71, 0x0F7E },
113 { 0x0F80, 0x0F84 }, { 0x0F86, 0x0F87 }, { 0x0F90, 0x0F97 },
114 { 0x0F99, 0x0FBC }, { 0x0FC6, 0x0FC6 }, { 0x102D, 0x1030 },
115 { 0x1032, 0x1032 }, { 0x1036, 0x1037 }, { 0x1039, 0x1039 },
116 { 0x1058, 0x1059 }, { 0x1160, 0x11FF }, { 0x1712, 0x1714 },
117 { 0x1732, 0x1734 }, { 0x1752, 0x1753 }, { 0x1772, 0x1773 },
118 { 0x17B4, 0x17B5 }, { 0x17B7, 0x17BD }, { 0x17C6, 0x17C6 },
119 { 0x17C9, 0x17D3 }, { 0x17DD, 0x17DD }, { 0x180B, 0x180D },
120 { 0x18A9, 0x18A9 }, { 0x1920, 0x1922 }, { 0x1927, 0x1928 },
121 { 0x1932, 0x1932 }, { 0x1939, 0x193B }, { 0x200B, 0x200F },
122 { 0x202A, 0x202E }, { 0x2060, 0x2063 }, { 0x206A, 0x206F },
123 { 0x20D0, 0x20EA }, { 0x302A, 0x302F }, { 0x3099, 0x309A },
124 { 0xFB1E, 0xFB1E }, { 0xFE00, 0xFE0F }, { 0xFE20, 0xFE23 },
125 { 0xFEFF, 0xFEFF }, { 0xFFF9, 0xFFFB }, { 0x1D167, 0x1D169 },
126 { 0x1D173, 0x1D182 }, { 0x1D185, 0x1D18B },
127 { 0x1D1AA, 0x1D1AD }, { 0xE0001, 0xE0001 },
128 { 0xE0020, 0xE007F }, { 0xE0100, 0xE01EF }
129 };
130
131 /* test for 8-bit control characters */
132 if (ch == 0)
133 return 0;
134 if (ch < 32 || (ch >= 0x7f && ch < 0xa0))
135 return -1;
136
137 /* binary search in table of non-spacing characters */
138 if (bisearch(ch, combining, sizeof(combining)
139 / sizeof(struct interval) - 1))
140 return 0;
141
142 /*
143 * If we arrive here, ch is neither a combining nor a C0/C1
144 * control character.
145 */
146
147 return 1 +
148 (ch >= 0x1100 &&
149 /* Hangul Jamo init. consonants */
150 (ch <= 0x115f ||
151 ch == 0x2329 || ch == 0x232a ||
152 /* CJK ... Yi */
153 (ch >= 0x2e80 && ch <= 0xa4cf &&
154 ch != 0x303f) ||
155 /* Hangul Syllables */
156 (ch >= 0xac00 && ch <= 0xd7a3) ||
157 /* CJK Compatibility Ideographs */
158 (ch >= 0xf900 && ch <= 0xfaff) ||
159 /* CJK Compatibility Forms */
160 (ch >= 0xfe30 && ch <= 0xfe6f) ||
161 /* Fullwidth Forms */
162 (ch >= 0xff00 && ch <= 0xff60) ||
163 (ch >= 0xffe0 && ch <= 0xffe6) ||
164 (ch >= 0x20000 && ch <= 0x2fffd) ||
165 (ch >= 0x30000 && ch <= 0x3fffd)));
166}
167
168/*
396ccf1f
JH
169 * Pick one ucs character starting from the location *start points at,
170 * and return it, while updating the *start pointer to point at the
44b25b87
JH
171 * end of that character. When remainder_p is not NULL, the location
172 * holds the number of bytes remaining in the string that we are allowed
173 * to pick from. Otherwise we are allowed to pick up to the NUL that
174 * would eventually appear in the string. *remainder_p is also reduced
175 * by the number of bytes we have consumed.
396ccf1f
JH
176 *
177 * If the string was not a valid UTF-8, *start pointer is set to NULL
178 * and the return value is undefined.
9e832665 179 */
5e133b8c 180static ucs_char_t pick_one_utf8_char(const char **start, size_t *remainder_p)
9e832665
JS
181{
182 unsigned char *s = (unsigned char *)*start;
28321145 183 ucs_char_t ch;
44b25b87 184 size_t remainder, incr;
9e832665 185
44b25b87
JH
186 /*
187 * A caller that assumes NUL terminated text can choose
188 * not to bother with the remainder length. We will
189 * stop at the first NUL.
190 */
191 remainder = (remainder_p ? *remainder_p : 999);
192
193 if (remainder < 1) {
194 goto invalid;
195 } else if (*s < 0x80) {
9e832665
JS
196 /* 0xxxxxxx */
197 ch = *s;
44b25b87 198 incr = 1;
9e832665
JS
199 } else if ((s[0] & 0xe0) == 0xc0) {
200 /* 110XXXXx 10xxxxxx */
44b25b87
JH
201 if (remainder < 2 ||
202 (s[1] & 0xc0) != 0x80 ||
203 (s[0] & 0xfe) == 0xc0)
9e832665
JS
204 goto invalid;
205 ch = ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
44b25b87 206 incr = 2;
9e832665
JS
207 } else if ((s[0] & 0xf0) == 0xe0) {
208 /* 1110XXXX 10Xxxxxx 10xxxxxx */
44b25b87
JH
209 if (remainder < 3 ||
210 (s[1] & 0xc0) != 0x80 ||
211 (s[2] & 0xc0) != 0x80 ||
212 /* overlong? */
213 (s[0] == 0xe0 && (s[1] & 0xe0) == 0x80) ||
214 /* surrogate? */
215 (s[0] == 0xed && (s[1] & 0xe0) == 0xa0) ||
216 /* U+FFFE or U+FFFF? */
217 (s[0] == 0xef && s[1] == 0xbf &&
218 (s[2] & 0xfe) == 0xbe))
9e832665
JS
219 goto invalid;
220 ch = ((s[0] & 0x0f) << 12) |
221 ((s[1] & 0x3f) << 6) | (s[2] & 0x3f);
44b25b87 222 incr = 3;
9e832665
JS
223 } else if ((s[0] & 0xf8) == 0xf0) {
224 /* 11110XXX 10XXxxxx 10xxxxxx 10xxxxxx */
44b25b87
JH
225 if (remainder < 4 ||
226 (s[1] & 0xc0) != 0x80 ||
227 (s[2] & 0xc0) != 0x80 ||
228 (s[3] & 0xc0) != 0x80 ||
229 /* overlong? */
230 (s[0] == 0xf0 && (s[1] & 0xf0) == 0x80) ||
231 /* > U+10FFFF? */
232 (s[0] == 0xf4 && s[1] > 0x8f) || s[0] > 0xf4)
9e832665
JS
233 goto invalid;
234 ch = ((s[0] & 0x07) << 18) | ((s[1] & 0x3f) << 12) |
235 ((s[2] & 0x3f) << 6) | (s[3] & 0x3f);
44b25b87 236 incr = 4;
9e832665
JS
237 } else {
238invalid:
239 *start = NULL;
240 return 0;
241 }
242
44b25b87
JH
243 *start += incr;
244 if (remainder_p)
245 *remainder_p = remainder - incr;
396ccf1f
JH
246 return ch;
247}
248
249/*
250 * This function returns the number of columns occupied by the character
251 * pointed to by the variable start. The pointer is updated to point at
44b25b87
JH
252 * the next character. When remainder_p is not NULL, it points at the
253 * location that stores the number of remaining bytes we can use to pick
254 * a character (see pick_one_utf8_char() above).
396ccf1f 255 */
44b25b87 256int utf8_width(const char **start, size_t *remainder_p)
396ccf1f 257{
44b25b87 258 ucs_char_t ch = pick_one_utf8_char(start, remainder_p);
396ccf1f
JH
259 if (!*start)
260 return 0;
b51be13c 261 return git_wcwidth(ch);
9e832665
JS
262}
263
8a9391e9
GT
264/*
265 * Returns the total number of columns required by a null-terminated
266 * string, assuming that the string is utf8. Returns strlen() instead
267 * if the string does not look like a valid utf8 string.
268 */
269int utf8_strwidth(const char *string)
270{
271 int width = 0;
272 const char *orig = string;
273
274 while (1) {
275 if (!string)
276 return strlen(orig);
277 if (!*string)
278 return width;
279 width += utf8_width(&string, NULL);
280 }
281}
282
9e832665
JS
283int is_utf8(const char *text)
284{
285 while (*text) {
286 if (*text == '\n' || *text == '\t' || *text == '\r') {
287 text++;
288 continue;
289 }
44b25b87 290 utf8_width(&text, NULL);
9e832665
JS
291 if (!text)
292 return 0;
293 }
294 return 1;
295}
296
3c0ff44a 297static void strbuf_addchars(struct strbuf *sb, int c, size_t n)
a94410c8 298{
3c0ff44a
RS
299 strbuf_grow(sb, n);
300 memset(sb->buf + sb->len, c, n);
301 strbuf_setlen(sb, sb->len + n);
9e832665
JS
302}
303
37bb5d74
RS
304static void strbuf_add_indented_text(struct strbuf *buf, const char *text,
305 int indent, int indent2)
306{
307 if (indent < 0)
308 indent = 0;
309 while (*text) {
310 const char *eol = strchrnul(text, '\n');
311 if (*eol == '\n')
312 eol++;
3c0ff44a 313 strbuf_addchars(buf, ' ', indent);
68ad5e1e 314 strbuf_add(buf, text, eol - text);
37bb5d74
RS
315 text = eol;
316 indent = indent2;
317 }
318}
319
9e832665
JS
320/*
321 * Wrap the text, if necessary. The variable indent is the indent for the
322 * first line, indent2 is the indent for all other lines.
094e03b0
JS
323 * If indent is negative, assume that already -indent columns have been
324 * consumed (and no extra indent is necessary for the first line).
9e832665 325 */
e0db1765 326void strbuf_add_wrapped_text(struct strbuf *buf,
462749b7 327 const char *text, int indent1, int indent2, int width)
9e832665 328{
462749b7
RS
329 int indent, w, assume_utf8 = 1;
330 const char *bol, *space, *start = text;
331 size_t orig_len = buf->len;
9e832665 332
00d39473 333 if (width <= 0) {
462749b7 334 strbuf_add_indented_text(buf, text, indent1, indent2);
e0db1765 335 return;
00d39473
JH
336 }
337
462749b7
RS
338retry:
339 bol = text;
340 w = indent = indent1;
341 space = NULL;
094e03b0
JS
342 if (indent < 0) {
343 w = -indent;
344 space = text;
345 }
346
9e832665 347 for (;;) {
8a3c63e0
RS
348 char c;
349 size_t skip;
350
351 while ((skip = display_mode_esc_sequence_len(text)))
352 text += skip;
353
354 c = *text;
9e832665 355 if (!c || isspace(c)) {
14e1a4e1 356 if (w <= width || !space) {
9e832665 357 const char *start = bol;
ae0b2702 358 if (!c && text == start)
e0db1765 359 return;
9e832665
JS
360 if (space)
361 start = space;
362 else
3c0ff44a 363 strbuf_addchars(buf, ' ', indent);
68ad5e1e 364 strbuf_add(buf, start, text - start);
094e03b0 365 if (!c)
e0db1765 366 return;
9e832665 367 space = text;
ae0b2702
JS
368 if (c == '\t')
369 w |= 0x07;
370 else if (c == '\n') {
371 space++;
372 if (*space == '\n') {
68ad5e1e 373 strbuf_addch(buf, '\n');
ae0b2702
JS
374 goto new_line;
375 }
376 else if (!isalnum(*space))
377 goto new_line;
378 else
68ad5e1e 379 strbuf_addch(buf, ' ');
ae0b2702 380 }
9e832665
JS
381 w++;
382 text++;
383 }
384 else {
ae0b2702 385new_line:
68ad5e1e 386 strbuf_addch(buf, '\n');
62273826 387 text = bol = space + isspace(*space);
9e832665
JS
388 space = NULL;
389 w = indent = indent2;
390 }
391 continue;
392 }
462749b7 393 if (assume_utf8) {
44b25b87 394 w += utf8_width(&text, NULL);
462749b7
RS
395 if (!text) {
396 assume_utf8 = 0;
397 text = start;
398 strbuf_setlen(buf, orig_len);
399 goto retry;
400 }
401 } else {
9e832665
JS
402 w++;
403 text++;
404 }
405 }
406}
b45974a6 407
e0db1765 408void strbuf_add_wrapped_bytes(struct strbuf *buf, const char *data, int len,
98acc837
JK
409 int indent, int indent2, int width)
410{
411 char *tmp = xstrndup(data, len);
e0db1765 412 strbuf_add_wrapped_text(buf, tmp, indent, indent2, width);
98acc837 413 free(tmp);
98acc837
JK
414}
415
677cfed5
JH
416int is_encoding_utf8(const char *name)
417{
418 if (!name)
419 return 1;
420 if (!strcasecmp(name, "utf-8") || !strcasecmp(name, "utf8"))
421 return 1;
422 return 0;
423}
424
0e18bcd5
JH
425int same_encoding(const char *src, const char *dst)
426{
427 if (is_encoding_utf8(src) && is_encoding_utf8(dst))
428 return 1;
429 return !strcasecmp(src, dst);
430}
431
c0821965
JX
432/*
433 * Wrapper for fprintf and returns the total number of columns required
434 * for the printed string, assuming that the string is utf8.
435 */
436int utf8_fprintf(FILE *stream, const char *format, ...)
437{
438 struct strbuf buf = STRBUF_INIT;
439 va_list arg;
440 int columns;
441
442 va_start(arg, format);
443 strbuf_vaddf(&buf, format, arg);
444 va_end(arg);
445
446 columns = fputs(buf.buf, stream);
447 if (0 <= columns) /* keep the error from the I/O */
448 columns = utf8_strwidth(buf.buf);
449 strbuf_release(&buf);
450 return columns;
451}
452
b45974a6
JH
453/*
454 * Given a buffer and its encoding, return it re-encoded
455 * with iconv. If the conversion fails, returns NULL.
456 */
457#ifndef NO_ICONV
309dbc82 458#if defined(OLD_ICONV) || (defined(__sun__) && !defined(_XPG6))
fd547a97
RJ
459 typedef const char * iconv_ibp;
460#else
461 typedef char * iconv_ibp;
462#endif
76759c7d 463char *reencode_string_iconv(const char *in, size_t insz, iconv_t conv)
b45974a6 464{
76759c7d 465 size_t outsz, outalloc;
fd547a97
RJ
466 char *out, *outpos;
467 iconv_ibp cp;
b45974a6 468
b45974a6
JH
469 outsz = insz;
470 outalloc = outsz + 1; /* for terminating NUL */
471 out = xmalloc(outalloc);
472 outpos = out;
fd547a97 473 cp = (iconv_ibp)in;
b45974a6
JH
474
475 while (1) {
476 size_t cnt = iconv(conv, &cp, &insz, &outpos, &outsz);
477
478 if (cnt == -1) {
479 size_t sofar;
480 if (errno != E2BIG) {
481 free(out);
b45974a6
JH
482 return NULL;
483 }
484 /* insz has remaining number of bytes.
485 * since we started outsz the same as insz,
486 * it is likely that insz is not enough for
487 * converting the rest.
488 */
489 sofar = outpos - out;
490 outalloc = sofar + insz * 2 + 32;
491 out = xrealloc(out, outalloc);
492 outpos = out + sofar;
493 outsz = outalloc - sofar - 1;
494 }
495 else {
496 *outpos = '\0';
497 break;
498 }
499 }
76759c7d
TB
500 return out;
501}
502
503char *reencode_string(const char *in, const char *out_encoding, const char *in_encoding)
504{
505 iconv_t conv;
506 char *out;
507
508 if (!in_encoding)
509 return NULL;
5c680be1 510
76759c7d 511 conv = iconv_open(out_encoding, in_encoding);
5c680be1
JK
512 if (conv == (iconv_t) -1) {
513 /*
514 * Some platforms do not have the variously spelled variants of
515 * UTF-8, so let's fall back to trying the most official
516 * spelling. We do so only as a fallback in case the platform
517 * does understand the user's spelling, but not our official
518 * one.
519 */
520 if (is_encoding_utf8(in_encoding))
521 in_encoding = "UTF-8";
522 if (is_encoding_utf8(out_encoding))
523 out_encoding = "UTF-8";
524 conv = iconv_open(out_encoding, in_encoding);
525 if (conv == (iconv_t) -1)
526 return NULL;
527 }
528
76759c7d 529 out = reencode_string_iconv(in, strlen(in), conv);
b45974a6
JH
530 iconv_close(conv);
531 return out;
532}
533#endif
6cd3c053
KS
534
535/*
536 * Returns first character length in bytes for multi-byte `text` according to
537 * `encoding`.
538 *
539 * - The `text` pointer is updated to point at the next character.
540 * - When `remainder_p` is not NULL, on entry `*remainder_p` is how much bytes
541 * we can consume from text, and on exit `*remainder_p` is reduced by returned
542 * character length. Otherwise `text` is treated as limited by NUL.
543 */
544int mbs_chrlen(const char **text, size_t *remainder_p, const char *encoding)
545{
546 int chrlen;
547 const char *p = *text;
548 size_t r = (remainder_p ? *remainder_p : SIZE_MAX);
549
550 if (r < 1)
551 return 0;
552
553 if (is_encoding_utf8(encoding)) {
554 pick_one_utf8_char(&p, &r);
555
556 chrlen = p ? (p - *text)
557 : 1 /* not valid UTF-8 -> raw byte sequence */;
558 }
559 else {
560 /*
561 * TODO use iconv to decode one char and obtain its chrlen
562 * for now, let's treat encodings != UTF-8 as one-byte
563 */
564 chrlen = 1;
565 }
566
567 *text += chrlen;
568 if (remainder_p)
569 *remainder_p -= chrlen;
570
571 return chrlen;
572}