utf8: pick_one_utf8_char()
[git/git.git] / utf8.c
CommitLineData
9e832665
JS
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
6struct interval {
7 int first;
8 int last;
9};
10
11/* auxiliary function for binary search in interval table */
f3fa1838
JH
12static int bisearch(ucs_char_t ucs, const struct interval *table, int max)
13{
9e832665
JS
14 int min = 0;
15 int mid;
16
17 if (ucs < table[0].first || ucs > table[max].last)
18 return 0;
19 while (max >= min) {
20 mid = (min + max) / 2;
21 if (ucs > table[mid].last)
22 min = mid + 1;
23 else if (ucs < table[mid].first)
24 max = mid - 1;
25 else
26 return 1;
27 }
28
29 return 0;
30}
31
32/* The following two functions define the column width of an ISO 10646
33 * character as follows:
34 *
35 * - The null character (U+0000) has a column width of 0.
36 *
37 * - Other C0/C1 control characters and DEL will lead to a return
38 * value of -1.
39 *
40 * - Non-spacing and enclosing combining characters (general
41 * category code Mn or Me in the Unicode database) have a
42 * column width of 0.
43 *
44 * - SOFT HYPHEN (U+00AD) has a column width of 1.
45 *
46 * - Other format characters (general category code Cf in the Unicode
47 * database) and ZERO WIDTH SPACE (U+200B) have a column width of 0.
48 *
49 * - Hangul Jamo medial vowels and final consonants (U+1160-U+11FF)
50 * have a column width of 0.
51 *
52 * - Spacing characters in the East Asian Wide (W) or East Asian
53 * Full-width (F) category as defined in Unicode Technical
54 * Report #11 have a column width of 2.
55 *
56 * - All remaining characters (including all printable
57 * ISO 8859-1 and WGL4 characters, Unicode control characters,
58 * etc.) have a column width of 1.
59 *
28321145 60 * This implementation assumes that ucs_char_t characters are encoded
9e832665
JS
61 * in ISO 10646.
62 */
63
b51be13c 64static int git_wcwidth(ucs_char_t ch)
9e832665
JS
65{
66 /*
67 * Sorted list of non-overlapping intervals of non-spacing characters,
68 * generated by
69 * "uniset +cat=Me +cat=Mn +cat=Cf -00AD +1160-11FF +200B c".
70 */
71 static const struct interval combining[] = {
72 { 0x0300, 0x0357 }, { 0x035D, 0x036F }, { 0x0483, 0x0486 },
73 { 0x0488, 0x0489 }, { 0x0591, 0x05A1 }, { 0x05A3, 0x05B9 },
74 { 0x05BB, 0x05BD }, { 0x05BF, 0x05BF }, { 0x05C1, 0x05C2 },
75 { 0x05C4, 0x05C4 }, { 0x0600, 0x0603 }, { 0x0610, 0x0615 },
76 { 0x064B, 0x0658 }, { 0x0670, 0x0670 }, { 0x06D6, 0x06E4 },
77 { 0x06E7, 0x06E8 }, { 0x06EA, 0x06ED }, { 0x070F, 0x070F },
78 { 0x0711, 0x0711 }, { 0x0730, 0x074A }, { 0x07A6, 0x07B0 },
79 { 0x0901, 0x0902 }, { 0x093C, 0x093C }, { 0x0941, 0x0948 },
80 { 0x094D, 0x094D }, { 0x0951, 0x0954 }, { 0x0962, 0x0963 },
81 { 0x0981, 0x0981 }, { 0x09BC, 0x09BC }, { 0x09C1, 0x09C4 },
82 { 0x09CD, 0x09CD }, { 0x09E2, 0x09E3 }, { 0x0A01, 0x0A02 },
83 { 0x0A3C, 0x0A3C }, { 0x0A41, 0x0A42 }, { 0x0A47, 0x0A48 },
84 { 0x0A4B, 0x0A4D }, { 0x0A70, 0x0A71 }, { 0x0A81, 0x0A82 },
85 { 0x0ABC, 0x0ABC }, { 0x0AC1, 0x0AC5 }, { 0x0AC7, 0x0AC8 },
86 { 0x0ACD, 0x0ACD }, { 0x0AE2, 0x0AE3 }, { 0x0B01, 0x0B01 },
87 { 0x0B3C, 0x0B3C }, { 0x0B3F, 0x0B3F }, { 0x0B41, 0x0B43 },
88 { 0x0B4D, 0x0B4D }, { 0x0B56, 0x0B56 }, { 0x0B82, 0x0B82 },
89 { 0x0BC0, 0x0BC0 }, { 0x0BCD, 0x0BCD }, { 0x0C3E, 0x0C40 },
90 { 0x0C46, 0x0C48 }, { 0x0C4A, 0x0C4D }, { 0x0C55, 0x0C56 },
91 { 0x0CBC, 0x0CBC }, { 0x0CBF, 0x0CBF }, { 0x0CC6, 0x0CC6 },
92 { 0x0CCC, 0x0CCD }, { 0x0D41, 0x0D43 }, { 0x0D4D, 0x0D4D },
93 { 0x0DCA, 0x0DCA }, { 0x0DD2, 0x0DD4 }, { 0x0DD6, 0x0DD6 },
94 { 0x0E31, 0x0E31 }, { 0x0E34, 0x0E3A }, { 0x0E47, 0x0E4E },
95 { 0x0EB1, 0x0EB1 }, { 0x0EB4, 0x0EB9 }, { 0x0EBB, 0x0EBC },
96 { 0x0EC8, 0x0ECD }, { 0x0F18, 0x0F19 }, { 0x0F35, 0x0F35 },
97 { 0x0F37, 0x0F37 }, { 0x0F39, 0x0F39 }, { 0x0F71, 0x0F7E },
98 { 0x0F80, 0x0F84 }, { 0x0F86, 0x0F87 }, { 0x0F90, 0x0F97 },
99 { 0x0F99, 0x0FBC }, { 0x0FC6, 0x0FC6 }, { 0x102D, 0x1030 },
100 { 0x1032, 0x1032 }, { 0x1036, 0x1037 }, { 0x1039, 0x1039 },
101 { 0x1058, 0x1059 }, { 0x1160, 0x11FF }, { 0x1712, 0x1714 },
102 { 0x1732, 0x1734 }, { 0x1752, 0x1753 }, { 0x1772, 0x1773 },
103 { 0x17B4, 0x17B5 }, { 0x17B7, 0x17BD }, { 0x17C6, 0x17C6 },
104 { 0x17C9, 0x17D3 }, { 0x17DD, 0x17DD }, { 0x180B, 0x180D },
105 { 0x18A9, 0x18A9 }, { 0x1920, 0x1922 }, { 0x1927, 0x1928 },
106 { 0x1932, 0x1932 }, { 0x1939, 0x193B }, { 0x200B, 0x200F },
107 { 0x202A, 0x202E }, { 0x2060, 0x2063 }, { 0x206A, 0x206F },
108 { 0x20D0, 0x20EA }, { 0x302A, 0x302F }, { 0x3099, 0x309A },
109 { 0xFB1E, 0xFB1E }, { 0xFE00, 0xFE0F }, { 0xFE20, 0xFE23 },
110 { 0xFEFF, 0xFEFF }, { 0xFFF9, 0xFFFB }, { 0x1D167, 0x1D169 },
111 { 0x1D173, 0x1D182 }, { 0x1D185, 0x1D18B },
112 { 0x1D1AA, 0x1D1AD }, { 0xE0001, 0xE0001 },
113 { 0xE0020, 0xE007F }, { 0xE0100, 0xE01EF }
114 };
115
116 /* test for 8-bit control characters */
117 if (ch == 0)
118 return 0;
119 if (ch < 32 || (ch >= 0x7f && ch < 0xa0))
120 return -1;
121
122 /* binary search in table of non-spacing characters */
123 if (bisearch(ch, combining, sizeof(combining)
124 / sizeof(struct interval) - 1))
125 return 0;
126
127 /*
128 * If we arrive here, ch is neither a combining nor a C0/C1
129 * control character.
130 */
131
132 return 1 +
133 (ch >= 0x1100 &&
134 /* Hangul Jamo init. consonants */
135 (ch <= 0x115f ||
136 ch == 0x2329 || ch == 0x232a ||
137 /* CJK ... Yi */
138 (ch >= 0x2e80 && ch <= 0xa4cf &&
139 ch != 0x303f) ||
140 /* Hangul Syllables */
141 (ch >= 0xac00 && ch <= 0xd7a3) ||
142 /* CJK Compatibility Ideographs */
143 (ch >= 0xf900 && ch <= 0xfaff) ||
144 /* CJK Compatibility Forms */
145 (ch >= 0xfe30 && ch <= 0xfe6f) ||
146 /* Fullwidth Forms */
147 (ch >= 0xff00 && ch <= 0xff60) ||
148 (ch >= 0xffe0 && ch <= 0xffe6) ||
149 (ch >= 0x20000 && ch <= 0x2fffd) ||
150 (ch >= 0x30000 && ch <= 0x3fffd)));
151}
152
153/*
396ccf1f
JH
154 * Pick one ucs character starting from the location *start points at,
155 * and return it, while updating the *start pointer to point at the
156 * end of that character.
157 *
158 * If the string was not a valid UTF-8, *start pointer is set to NULL
159 * and the return value is undefined.
9e832665 160 */
396ccf1f 161ucs_char_t pick_one_utf8_char(const char **start)
9e832665
JS
162{
163 unsigned char *s = (unsigned char *)*start;
28321145 164 ucs_char_t ch;
9e832665
JS
165
166 if (*s < 0x80) {
167 /* 0xxxxxxx */
168 ch = *s;
169 *start += 1;
170 } else if ((s[0] & 0xe0) == 0xc0) {
171 /* 110XXXXx 10xxxxxx */
172 if ((s[1] & 0xc0) != 0x80 ||
173 /* overlong? */
174 (s[0] & 0xfe) == 0xc0)
175 goto invalid;
176 ch = ((s[0] & 0x1f) << 6) | (s[1] & 0x3f);
177 *start += 2;
178 } else if ((s[0] & 0xf0) == 0xe0) {
179 /* 1110XXXX 10Xxxxxx 10xxxxxx */
180 if ((s[1] & 0xc0) != 0x80 ||
181 (s[2] & 0xc0) != 0x80 ||
182 /* overlong? */
183 (s[0] == 0xe0 && (s[1] & 0xe0) == 0x80) ||
184 /* surrogate? */
185 (s[0] == 0xed && (s[1] & 0xe0) == 0xa0) ||
186 /* U+FFFE or U+FFFF? */
187 (s[0] == 0xef && s[1] == 0xbf &&
188 (s[2] & 0xfe) == 0xbe))
189 goto invalid;
190 ch = ((s[0] & 0x0f) << 12) |
191 ((s[1] & 0x3f) << 6) | (s[2] & 0x3f);
192 *start += 3;
193 } else if ((s[0] & 0xf8) == 0xf0) {
194 /* 11110XXX 10XXxxxx 10xxxxxx 10xxxxxx */
195 if ((s[1] & 0xc0) != 0x80 ||
196 (s[2] & 0xc0) != 0x80 ||
197 (s[3] & 0xc0) != 0x80 ||
198 /* overlong? */
199 (s[0] == 0xf0 && (s[1] & 0xf0) == 0x80) ||
200 /* > U+10FFFF? */
201 (s[0] == 0xf4 && s[1] > 0x8f) || s[0] > 0xf4)
202 goto invalid;
203 ch = ((s[0] & 0x07) << 18) | ((s[1] & 0x3f) << 12) |
204 ((s[2] & 0x3f) << 6) | (s[3] & 0x3f);
205 *start += 4;
206 } else {
207invalid:
208 *start = NULL;
209 return 0;
210 }
211
396ccf1f
JH
212 return ch;
213}
214
215/*
216 * This function returns the number of columns occupied by the character
217 * pointed to by the variable start. The pointer is updated to point at
218 * the next character. If it was not valid UTF-8, the pointer is set to
219 * NULL.
220 */
221int utf8_width(const char **start)
222{
223 ucs_char_t ch = pick_one_utf8_char(start);
224 if (!*start)
225 return 0;
b51be13c 226 return git_wcwidth(ch);
9e832665
JS
227}
228
229int is_utf8(const char *text)
230{
231 while (*text) {
232 if (*text == '\n' || *text == '\t' || *text == '\r') {
233 text++;
234 continue;
235 }
236 utf8_width(&text);
237 if (!text)
238 return 0;
239 }
240 return 1;
241}
242
243static void print_spaces(int count)
244{
245 static const char s[] = " ";
246 while (count >= sizeof(s)) {
247 fwrite(s, sizeof(s) - 1, 1, stdout);
248 count -= sizeof(s) - 1;
249 }
250 fwrite(s, count, 1, stdout);
251}
252
253/*
254 * Wrap the text, if necessary. The variable indent is the indent for the
255 * first line, indent2 is the indent for all other lines.
094e03b0
JS
256 * If indent is negative, assume that already -indent columns have been
257 * consumed (and no extra indent is necessary for the first line).
9e832665 258 */
094e03b0 259int print_wrapped_text(const char *text, int indent, int indent2, int width)
9e832665
JS
260{
261 int w = indent, assume_utf8 = is_utf8(text);
262 const char *bol = text, *space = NULL;
263
094e03b0
JS
264 if (indent < 0) {
265 w = -indent;
266 space = text;
267 }
268
9e832665
JS
269 for (;;) {
270 char c = *text;
271 if (!c || isspace(c)) {
272 if (w < width || !space) {
273 const char *start = bol;
274 if (space)
275 start = space;
276 else
277 print_spaces(indent);
278 fwrite(start, text - start, 1, stdout);
094e03b0
JS
279 if (!c)
280 return w;
281 else if (c == '\t')
9e832665
JS
282 w |= 0x07;
283 space = text;
284 w++;
285 text++;
286 }
287 else {
288 putchar('\n');
62273826 289 text = bol = space + isspace(*space);
9e832665
JS
290 space = NULL;
291 w = indent = indent2;
292 }
293 continue;
294 }
295 if (assume_utf8)
296 w += utf8_width(&text);
297 else {
298 w++;
299 text++;
300 }
301 }
302}
b45974a6 303
677cfed5
JH
304int is_encoding_utf8(const char *name)
305{
306 if (!name)
307 return 1;
308 if (!strcasecmp(name, "utf-8") || !strcasecmp(name, "utf8"))
309 return 1;
310 return 0;
311}
312
b45974a6
JH
313/*
314 * Given a buffer and its encoding, return it re-encoded
315 * with iconv. If the conversion fails, returns NULL.
316 */
317#ifndef NO_ICONV
fd547a97
RJ
318#ifdef OLD_ICONV
319 typedef const char * iconv_ibp;
320#else
321 typedef char * iconv_ibp;
322#endif
b45974a6
JH
323char *reencode_string(const char *in, const char *out_encoding, const char *in_encoding)
324{
325 iconv_t conv;
326 size_t insz, outsz, outalloc;
fd547a97
RJ
327 char *out, *outpos;
328 iconv_ibp cp;
b45974a6
JH
329
330 if (!in_encoding)
331 return NULL;
332 conv = iconv_open(out_encoding, in_encoding);
333 if (conv == (iconv_t) -1)
334 return NULL;
335 insz = strlen(in);
336 outsz = insz;
337 outalloc = outsz + 1; /* for terminating NUL */
338 out = xmalloc(outalloc);
339 outpos = out;
fd547a97 340 cp = (iconv_ibp)in;
b45974a6
JH
341
342 while (1) {
343 size_t cnt = iconv(conv, &cp, &insz, &outpos, &outsz);
344
345 if (cnt == -1) {
346 size_t sofar;
347 if (errno != E2BIG) {
348 free(out);
349 iconv_close(conv);
350 return NULL;
351 }
352 /* insz has remaining number of bytes.
353 * since we started outsz the same as insz,
354 * it is likely that insz is not enough for
355 * converting the rest.
356 */
357 sofar = outpos - out;
358 outalloc = sofar + insz * 2 + 32;
359 out = xrealloc(out, outalloc);
360 outpos = out + sofar;
361 outsz = outalloc - sofar - 1;
362 }
363 else {
364 *outpos = '\0';
365 break;
366 }
367 }
368 iconv_close(conv);
369 return out;
370}
371#endif