]>
Commit | Line | Data |
---|---|---|
9e832665 | 1 | #include "git-compat-util.h" |
a94410c8 | 2 | #include "strbuf.h" |
9e832665 JS |
3 | #include "utf8.h" |
4 | ||
d05b08cd | 5 | /* This code is originally from https://www.cl.cam.ac.uk/~mgk25/ucs/ */ |
9e832665 | 6 | |
aab2a1ae TB |
7 | static const char utf16_be_bom[] = {'\xFE', '\xFF'}; |
8 | static const char utf16_le_bom[] = {'\xFF', '\xFE'}; | |
9 | static const char utf32_be_bom[] = {'\0', '\0', '\xFE', '\xFF'}; | |
10 | static const char utf32_le_bom[] = {'\xFF', '\xFE', '\0', '\0'}; | |
11 | ||
9e832665 | 12 | struct interval { |
a68a67de JK |
13 | ucs_char_t first; |
14 | ucs_char_t last; | |
9e832665 JS |
15 | }; |
16 | ||
1640632b | 17 | size_t display_mode_esc_sequence_len(const char *s) |
4247fe79 NTND |
18 | { |
19 | const char *p = s; | |
20 | if (*p++ != '\033') | |
21 | return 0; | |
22 | if (*p++ != '[') | |
23 | return 0; | |
24 | while (isdigit(*p) || *p == ';') | |
25 | p++; | |
26 | if (*p++ != 'm') | |
27 | return 0; | |
28 | return p - s; | |
29 | } | |
30 | ||
9e832665 | 31 | /* auxiliary function for binary search in interval table */ |
f3fa1838 JH |
32 | static int bisearch(ucs_char_t ucs, const struct interval *table, int max) |
33 | { | |
9e832665 JS |
34 | int min = 0; |
35 | int mid; | |
36 | ||
37 | if (ucs < table[0].first || ucs > table[max].last) | |
38 | return 0; | |
39 | while (max >= min) { | |
19716b21 | 40 | mid = min + (max - min) / 2; |
9e832665 JS |
41 | if (ucs > table[mid].last) |
42 | min = mid + 1; | |
43 | else if (ucs < table[mid].first) | |
44 | max = mid - 1; | |
45 | else | |
46 | return 1; | |
47 | } | |
48 | ||
49 | return 0; | |
50 | } | |
51 | ||
52 | /* The following two functions define the column width of an ISO 10646 | |
53 | * character as follows: | |
54 | * | |
55 | * - The null character (U+0000) has a column width of 0. | |
56 | * | |
57 | * - Other C0/C1 control characters and DEL will lead to a return | |
58 | * value of -1. | |
59 | * | |
60 | * - Non-spacing and enclosing combining characters (general | |
61 | * category code Mn or Me in the Unicode database) have a | |
62 | * column width of 0. | |
63 | * | |
64 | * - SOFT HYPHEN (U+00AD) has a column width of 1. | |
65 | * | |
66 | * - Other format characters (general category code Cf in the Unicode | |
67 | * database) and ZERO WIDTH SPACE (U+200B) have a column width of 0. | |
68 | * | |
69 | * - Hangul Jamo medial vowels and final consonants (U+1160-U+11FF) | |
70 | * have a column width of 0. | |
71 | * | |
72 | * - Spacing characters in the East Asian Wide (W) or East Asian | |
73 | * Full-width (F) category as defined in Unicode Technical | |
74 | * Report #11 have a column width of 2. | |
75 | * | |
76 | * - All remaining characters (including all printable | |
77 | * ISO 8859-1 and WGL4 characters, Unicode control characters, | |
78 | * etc.) have a column width of 1. | |
79 | * | |
28321145 | 80 | * This implementation assumes that ucs_char_t characters are encoded |
9e832665 JS |
81 | * in ISO 10646. |
82 | */ | |
83 | ||
b51be13c | 84 | static int git_wcwidth(ucs_char_t ch) |
9e832665 JS |
85 | { |
86 | /* | |
87 | * Sorted list of non-overlapping intervals of non-spacing characters, | |
9e832665 | 88 | */ |
e233bef4 | 89 | #include "unicode-width.h" |
9e832665 JS |
90 | |
91 | /* test for 8-bit control characters */ | |
92 | if (ch == 0) | |
93 | return 0; | |
94 | if (ch < 32 || (ch >= 0x7f && ch < 0xa0)) | |
95 | return -1; | |
96 | ||
97 | /* binary search in table of non-spacing characters */ | |
fa364ad7 | 98 | if (bisearch(ch, zero_width, ARRAY_SIZE(zero_width) - 1)) |
9e832665 JS |
99 | return 0; |
100 | ||
08460345 | 101 | /* binary search in table of double width characters */ |
fa364ad7 | 102 | if (bisearch(ch, double_width, ARRAY_SIZE(double_width) - 1)) |
08460345 | 103 | return 2; |
9e832665 | 104 | |
08460345 | 105 | return 1; |
9e832665 JS |
106 | } |
107 | ||
108 | /* | |
396ccf1f JH |
109 | * Pick one ucs character starting from the location *start points at, |
110 | * and return it, while updating the *start pointer to point at the | |
44b25b87 JH |
111 | * end of that character. When remainder_p is not NULL, the location |
112 | * holds the number of bytes remaining in the string that we are allowed | |
113 | * to pick from. Otherwise we are allowed to pick up to the NUL that | |
114 | * would eventually appear in the string. *remainder_p is also reduced | |
115 | * by the number of bytes we have consumed. | |
396ccf1f JH |
116 | * |
117 | * If the string was not a valid UTF-8, *start pointer is set to NULL | |
118 | * and the return value is undefined. | |
9e832665 | 119 | */ |
5e133b8c | 120 | static ucs_char_t pick_one_utf8_char(const char **start, size_t *remainder_p) |
9e832665 JS |
121 | { |
122 | unsigned char *s = (unsigned char *)*start; | |
28321145 | 123 | ucs_char_t ch; |
44b25b87 | 124 | size_t remainder, incr; |
9e832665 | 125 | |
44b25b87 JH |
126 | /* |
127 | * A caller that assumes NUL terminated text can choose | |
128 | * not to bother with the remainder length. We will | |
129 | * stop at the first NUL. | |
130 | */ | |
131 | remainder = (remainder_p ? *remainder_p : 999); | |
132 | ||
133 | if (remainder < 1) { | |
134 | goto invalid; | |
135 | } else if (*s < 0x80) { | |
9e832665 JS |
136 | /* 0xxxxxxx */ |
137 | ch = *s; | |
44b25b87 | 138 | incr = 1; |
9e832665 JS |
139 | } else if ((s[0] & 0xe0) == 0xc0) { |
140 | /* 110XXXXx 10xxxxxx */ | |
44b25b87 JH |
141 | if (remainder < 2 || |
142 | (s[1] & 0xc0) != 0x80 || | |
143 | (s[0] & 0xfe) == 0xc0) | |
9e832665 JS |
144 | goto invalid; |
145 | ch = ((s[0] & 0x1f) << 6) | (s[1] & 0x3f); | |
44b25b87 | 146 | incr = 2; |
9e832665 JS |
147 | } else if ((s[0] & 0xf0) == 0xe0) { |
148 | /* 1110XXXX 10Xxxxxx 10xxxxxx */ | |
44b25b87 JH |
149 | if (remainder < 3 || |
150 | (s[1] & 0xc0) != 0x80 || | |
151 | (s[2] & 0xc0) != 0x80 || | |
152 | /* overlong? */ | |
153 | (s[0] == 0xe0 && (s[1] & 0xe0) == 0x80) || | |
154 | /* surrogate? */ | |
155 | (s[0] == 0xed && (s[1] & 0xe0) == 0xa0) || | |
156 | /* U+FFFE or U+FFFF? */ | |
157 | (s[0] == 0xef && s[1] == 0xbf && | |
158 | (s[2] & 0xfe) == 0xbe)) | |
9e832665 JS |
159 | goto invalid; |
160 | ch = ((s[0] & 0x0f) << 12) | | |
161 | ((s[1] & 0x3f) << 6) | (s[2] & 0x3f); | |
44b25b87 | 162 | incr = 3; |
9e832665 JS |
163 | } else if ((s[0] & 0xf8) == 0xf0) { |
164 | /* 11110XXX 10XXxxxx 10xxxxxx 10xxxxxx */ | |
44b25b87 JH |
165 | if (remainder < 4 || |
166 | (s[1] & 0xc0) != 0x80 || | |
167 | (s[2] & 0xc0) != 0x80 || | |
168 | (s[3] & 0xc0) != 0x80 || | |
169 | /* overlong? */ | |
170 | (s[0] == 0xf0 && (s[1] & 0xf0) == 0x80) || | |
171 | /* > U+10FFFF? */ | |
172 | (s[0] == 0xf4 && s[1] > 0x8f) || s[0] > 0xf4) | |
9e832665 JS |
173 | goto invalid; |
174 | ch = ((s[0] & 0x07) << 18) | ((s[1] & 0x3f) << 12) | | |
175 | ((s[2] & 0x3f) << 6) | (s[3] & 0x3f); | |
44b25b87 | 176 | incr = 4; |
9e832665 JS |
177 | } else { |
178 | invalid: | |
179 | *start = NULL; | |
180 | return 0; | |
181 | } | |
182 | ||
44b25b87 JH |
183 | *start += incr; |
184 | if (remainder_p) | |
185 | *remainder_p = remainder - incr; | |
396ccf1f JH |
186 | return ch; |
187 | } | |
188 | ||
189 | /* | |
190 | * This function returns the number of columns occupied by the character | |
191 | * pointed to by the variable start. The pointer is updated to point at | |
44b25b87 JH |
192 | * the next character. When remainder_p is not NULL, it points at the |
193 | * location that stores the number of remaining bytes we can use to pick | |
194 | * a character (see pick_one_utf8_char() above). | |
396ccf1f | 195 | */ |
44b25b87 | 196 | int utf8_width(const char **start, size_t *remainder_p) |
396ccf1f | 197 | { |
44b25b87 | 198 | ucs_char_t ch = pick_one_utf8_char(start, remainder_p); |
396ccf1f JH |
199 | if (!*start) |
200 | return 0; | |
b51be13c | 201 | return git_wcwidth(ch); |
9e832665 JS |
202 | } |
203 | ||
8a9391e9 GT |
204 | /* |
205 | * Returns the total number of columns required by a null-terminated | |
206 | * string, assuming that the string is utf8. Returns strlen() instead | |
207 | * if the string does not look like a valid utf8 string. | |
208 | */ | |
522cc87f | 209 | int utf8_strnwidth(const char *string, size_t len, int skip_ansi) |
8a9391e9 | 210 | { |
8a9391e9 | 211 | const char *orig = string; |
937b71cc | 212 | size_t width = 0; |
8a9391e9 | 213 | |
2bc1e7ec | 214 | while (string && string < orig + len) { |
937b71cc PS |
215 | int glyph_width; |
216 | size_t skip; | |
17d23e8a | 217 | |
2bc1e7ec NTND |
218 | while (skip_ansi && |
219 | (skip = display_mode_esc_sequence_len(string)) != 0) | |
220 | string += skip; | |
17d23e8a PS |
221 | |
222 | glyph_width = utf8_width(&string, NULL); | |
223 | if (glyph_width > 0) | |
224 | width += glyph_width; | |
8a9391e9 | 225 | } |
937b71cc PS |
226 | |
227 | /* | |
228 | * TODO: fix the interface of this function and `utf8_strwidth()` to | |
229 | * return `size_t` instead of `int`. | |
230 | */ | |
231 | return cast_size_t_to_int(string ? width : len); | |
2bc1e7ec NTND |
232 | } |
233 | ||
234 | int utf8_strwidth(const char *string) | |
235 | { | |
522cc87f | 236 | return utf8_strnwidth(string, strlen(string), 0); |
8a9391e9 GT |
237 | } |
238 | ||
9e832665 JS |
239 | int is_utf8(const char *text) |
240 | { | |
241 | while (*text) { | |
242 | if (*text == '\n' || *text == '\t' || *text == '\r') { | |
243 | text++; | |
244 | continue; | |
245 | } | |
44b25b87 | 246 | utf8_width(&text, NULL); |
9e832665 JS |
247 | if (!text) |
248 | return 0; | |
249 | } | |
250 | return 1; | |
251 | } | |
252 | ||
37bb5d74 RS |
253 | static void strbuf_add_indented_text(struct strbuf *buf, const char *text, |
254 | int indent, int indent2) | |
255 | { | |
256 | if (indent < 0) | |
257 | indent = 0; | |
258 | while (*text) { | |
259 | const char *eol = strchrnul(text, '\n'); | |
260 | if (*eol == '\n') | |
261 | eol++; | |
3c0ff44a | 262 | strbuf_addchars(buf, ' ', indent); |
68ad5e1e | 263 | strbuf_add(buf, text, eol - text); |
37bb5d74 RS |
264 | text = eol; |
265 | indent = indent2; | |
266 | } | |
267 | } | |
268 | ||
9e832665 JS |
269 | /* |
270 | * Wrap the text, if necessary. The variable indent is the indent for the | |
271 | * first line, indent2 is the indent for all other lines. | |
094e03b0 JS |
272 | * If indent is negative, assume that already -indent columns have been |
273 | * consumed (and no extra indent is necessary for the first line). | |
9e832665 | 274 | */ |
e0db1765 | 275 | void strbuf_add_wrapped_text(struct strbuf *buf, |
462749b7 | 276 | const char *text, int indent1, int indent2, int width) |
9e832665 | 277 | { |
462749b7 RS |
278 | int indent, w, assume_utf8 = 1; |
279 | const char *bol, *space, *start = text; | |
280 | size_t orig_len = buf->len; | |
9e832665 | 281 | |
00d39473 | 282 | if (width <= 0) { |
462749b7 | 283 | strbuf_add_indented_text(buf, text, indent1, indent2); |
e0db1765 | 284 | return; |
00d39473 JH |
285 | } |
286 | ||
462749b7 RS |
287 | retry: |
288 | bol = text; | |
289 | w = indent = indent1; | |
290 | space = NULL; | |
094e03b0 JS |
291 | if (indent < 0) { |
292 | w = -indent; | |
293 | space = text; | |
294 | } | |
295 | ||
9e832665 | 296 | for (;;) { |
8a3c63e0 RS |
297 | char c; |
298 | size_t skip; | |
299 | ||
300 | while ((skip = display_mode_esc_sequence_len(text))) | |
301 | text += skip; | |
302 | ||
303 | c = *text; | |
9e832665 | 304 | if (!c || isspace(c)) { |
14e1a4e1 | 305 | if (w <= width || !space) { |
9e832665 | 306 | const char *start = bol; |
ae0b2702 | 307 | if (!c && text == start) |
e0db1765 | 308 | return; |
9e832665 JS |
309 | if (space) |
310 | start = space; | |
311 | else | |
3c0ff44a | 312 | strbuf_addchars(buf, ' ', indent); |
68ad5e1e | 313 | strbuf_add(buf, start, text - start); |
094e03b0 | 314 | if (!c) |
e0db1765 | 315 | return; |
9e832665 | 316 | space = text; |
ae0b2702 JS |
317 | if (c == '\t') |
318 | w |= 0x07; | |
319 | else if (c == '\n') { | |
320 | space++; | |
321 | if (*space == '\n') { | |
68ad5e1e | 322 | strbuf_addch(buf, '\n'); |
ae0b2702 JS |
323 | goto new_line; |
324 | } | |
325 | else if (!isalnum(*space)) | |
326 | goto new_line; | |
327 | else | |
68ad5e1e | 328 | strbuf_addch(buf, ' '); |
ae0b2702 | 329 | } |
9e832665 JS |
330 | w++; |
331 | text++; | |
332 | } | |
333 | else { | |
ae0b2702 | 334 | new_line: |
68ad5e1e | 335 | strbuf_addch(buf, '\n'); |
62273826 | 336 | text = bol = space + isspace(*space); |
9e832665 JS |
337 | space = NULL; |
338 | w = indent = indent2; | |
339 | } | |
340 | continue; | |
341 | } | |
462749b7 | 342 | if (assume_utf8) { |
44b25b87 | 343 | w += utf8_width(&text, NULL); |
462749b7 RS |
344 | if (!text) { |
345 | assume_utf8 = 0; | |
346 | text = start; | |
347 | strbuf_setlen(buf, orig_len); | |
348 | goto retry; | |
349 | } | |
350 | } else { | |
9e832665 JS |
351 | w++; |
352 | text++; | |
353 | } | |
354 | } | |
355 | } | |
b45974a6 | 356 | |
e0db1765 | 357 | void strbuf_add_wrapped_bytes(struct strbuf *buf, const char *data, int len, |
98acc837 JK |
358 | int indent, int indent2, int width) |
359 | { | |
360 | char *tmp = xstrndup(data, len); | |
e0db1765 | 361 | strbuf_add_wrapped_text(buf, tmp, indent, indent2, width); |
98acc837 | 362 | free(tmp); |
98acc837 JK |
363 | } |
364 | ||
a7f01c6b NTND |
365 | void strbuf_utf8_replace(struct strbuf *sb_src, int pos, int width, |
366 | const char *subst) | |
367 | { | |
f930a239 PS |
368 | const char *src = sb_src->buf, *end = sb_src->buf + sb_src->len; |
369 | struct strbuf dst; | |
370 | int w = 0; | |
a7f01c6b | 371 | |
f930a239 | 372 | strbuf_init(&dst, sb_src->len); |
a7f01c6b NTND |
373 | |
374 | while (src < end) { | |
f930a239 | 375 | const char *old; |
81c2d4c3 | 376 | int glyph_width; |
a7f01c6b NTND |
377 | size_t n; |
378 | ||
379 | while ((n = display_mode_esc_sequence_len(src))) { | |
f930a239 | 380 | strbuf_add(&dst, src, n); |
a7f01c6b | 381 | src += n; |
a7f01c6b NTND |
382 | } |
383 | ||
43087596 NTND |
384 | if (src >= end) |
385 | break; | |
386 | ||
a7f01c6b | 387 | old = src; |
81c2d4c3 PS |
388 | glyph_width = utf8_width((const char**)&src, NULL); |
389 | if (!src) /* broken utf-8, do nothing */ | |
9a012bf3 | 390 | goto out; |
81c2d4c3 PS |
391 | |
392 | /* | |
393 | * In case we see a control character we copy it into the | |
394 | * buffer, but don't add it to the width. | |
395 | */ | |
396 | if (glyph_width < 0) | |
397 | glyph_width = 0; | |
398 | ||
399 | if (glyph_width && w >= pos && w < pos + width) { | |
a7f01c6b | 400 | if (subst) { |
f930a239 | 401 | strbuf_addstr(&dst, subst); |
a7f01c6b NTND |
402 | subst = NULL; |
403 | } | |
f930a239 PS |
404 | } else { |
405 | strbuf_add(&dst, old, src - old); | |
a7f01c6b | 406 | } |
f930a239 | 407 | |
81c2d4c3 | 408 | w += glyph_width; |
a7f01c6b | 409 | } |
f930a239 PS |
410 | |
411 | strbuf_swap(sb_src, &dst); | |
9a012bf3 | 412 | out: |
f930a239 | 413 | strbuf_release(&dst); |
a7f01c6b NTND |
414 | } |
415 | ||
2f0c4a36 LS |
416 | /* |
417 | * Returns true (1) if the src encoding name matches the dst encoding | |
418 | * name directly or one of its alternative names. E.g. UTF-16BE is the | |
419 | * same as UTF16BE. | |
420 | */ | |
421 | static int same_utf_encoding(const char *src, const char *dst) | |
422 | { | |
89f8caba RS |
423 | if (skip_iprefix(src, "utf", &src) && skip_iprefix(dst, "utf", &dst)) { |
424 | skip_prefix(src, "-", &src); | |
425 | skip_prefix(dst, "-", &dst); | |
426 | return !strcasecmp(src, dst); | |
2f0c4a36 LS |
427 | } |
428 | return 0; | |
429 | } | |
430 | ||
677cfed5 JH |
431 | int is_encoding_utf8(const char *name) |
432 | { | |
433 | if (!name) | |
434 | return 1; | |
2f0c4a36 | 435 | if (same_utf_encoding("utf-8", name)) |
677cfed5 JH |
436 | return 1; |
437 | return 0; | |
438 | } | |
439 | ||
0e18bcd5 JH |
440 | int same_encoding(const char *src, const char *dst) |
441 | { | |
2f0c4a36 LS |
442 | static const char utf8[] = "UTF-8"; |
443 | ||
444 | if (!src) | |
445 | src = utf8; | |
446 | if (!dst) | |
447 | dst = utf8; | |
448 | if (same_utf_encoding(src, dst)) | |
0e18bcd5 JH |
449 | return 1; |
450 | return !strcasecmp(src, dst); | |
451 | } | |
452 | ||
c0821965 JX |
453 | /* |
454 | * Wrapper for fprintf and returns the total number of columns required | |
455 | * for the printed string, assuming that the string is utf8. | |
456 | */ | |
457 | int utf8_fprintf(FILE *stream, const char *format, ...) | |
458 | { | |
459 | struct strbuf buf = STRBUF_INIT; | |
460 | va_list arg; | |
461 | int columns; | |
462 | ||
463 | va_start(arg, format); | |
464 | strbuf_vaddf(&buf, format, arg); | |
465 | va_end(arg); | |
466 | ||
467 | columns = fputs(buf.buf, stream); | |
468 | if (0 <= columns) /* keep the error from the I/O */ | |
469 | columns = utf8_strwidth(buf.buf); | |
470 | strbuf_release(&buf); | |
471 | return columns; | |
472 | } | |
473 | ||
b45974a6 JH |
474 | /* |
475 | * Given a buffer and its encoding, return it re-encoded | |
476 | * with iconv. If the conversion fails, returns NULL. | |
477 | */ | |
478 | #ifndef NO_ICONV | |
309dbc82 | 479 | #if defined(OLD_ICONV) || (defined(__sun__) && !defined(_XPG6)) |
fd547a97 RJ |
480 | typedef const char * iconv_ibp; |
481 | #else | |
482 | typedef char * iconv_ibp; | |
483 | #endif | |
aab2a1ae TB |
484 | char *reencode_string_iconv(const char *in, size_t insz, iconv_t conv, |
485 | size_t bom_len, size_t *outsz_p) | |
b45974a6 | 486 | { |
76759c7d | 487 | size_t outsz, outalloc; |
fd547a97 RJ |
488 | char *out, *outpos; |
489 | iconv_ibp cp; | |
b45974a6 | 490 | |
b45974a6 | 491 | outsz = insz; |
aab2a1ae | 492 | outalloc = st_add(outsz, 1 + bom_len); /* for terminating NUL */ |
b45974a6 | 493 | out = xmalloc(outalloc); |
aab2a1ae | 494 | outpos = out + bom_len; |
fd547a97 | 495 | cp = (iconv_ibp)in; |
b45974a6 JH |
496 | |
497 | while (1) { | |
498 | size_t cnt = iconv(conv, &cp, &insz, &outpos, &outsz); | |
499 | ||
df5213b7 | 500 | if (cnt == (size_t) -1) { |
b45974a6 JH |
501 | size_t sofar; |
502 | if (errno != E2BIG) { | |
503 | free(out); | |
b45974a6 JH |
504 | return NULL; |
505 | } | |
506 | /* insz has remaining number of bytes. | |
507 | * since we started outsz the same as insz, | |
508 | * it is likely that insz is not enough for | |
509 | * converting the rest. | |
510 | */ | |
511 | sofar = outpos - out; | |
77aa03d6 | 512 | outalloc = st_add3(sofar, st_mult(insz, 2), 32); |
b45974a6 JH |
513 | out = xrealloc(out, outalloc); |
514 | outpos = out + sofar; | |
515 | outsz = outalloc - sofar - 1; | |
516 | } | |
517 | else { | |
518 | *outpos = '\0'; | |
b782bbab NTND |
519 | if (outsz_p) |
520 | *outsz_p = outpos - out; | |
b45974a6 JH |
521 | break; |
522 | } | |
523 | } | |
76759c7d TB |
524 | return out; |
525 | } | |
526 | ||
3270741e JH |
527 | static const char *fallback_encoding(const char *name) |
528 | { | |
529 | /* | |
530 | * Some platforms do not have the variously spelled variants of | |
531 | * UTF-8, so let's fall back to trying the most official | |
532 | * spelling. We do so only as a fallback in case the platform | |
533 | * does understand the user's spelling, but not our official | |
534 | * one. | |
535 | */ | |
536 | if (is_encoding_utf8(name)) | |
537 | return "UTF-8"; | |
538 | ||
df375588 JH |
539 | /* |
540 | * Even though latin-1 is still seen in e-mail | |
541 | * headers, some platforms only install ISO-8859-1. | |
542 | */ | |
543 | if (!strcasecmp(name, "latin-1")) | |
544 | return "ISO-8859-1"; | |
545 | ||
3270741e JH |
546 | return name; |
547 | } | |
548 | ||
c7d017d7 | 549 | char *reencode_string_len(const char *in, size_t insz, |
b782bbab | 550 | const char *out_encoding, const char *in_encoding, |
c7d017d7 | 551 | size_t *outsz) |
76759c7d TB |
552 | { |
553 | iconv_t conv; | |
554 | char *out; | |
aab2a1ae TB |
555 | const char *bom_str = NULL; |
556 | size_t bom_len = 0; | |
76759c7d TB |
557 | |
558 | if (!in_encoding) | |
559 | return NULL; | |
5c680be1 | 560 | |
aab2a1ae TB |
561 | /* UTF-16LE-BOM is the same as UTF-16 for reading */ |
562 | if (same_utf_encoding("UTF-16LE-BOM", in_encoding)) | |
563 | in_encoding = "UTF-16"; | |
564 | ||
565 | /* | |
566 | * For writing, UTF-16 iconv typically creates "UTF-16BE-BOM" | |
567 | * Some users under Windows want the little endian version | |
79444c92 | 568 | * |
569 | * We handle UTF-16 and UTF-32 ourselves only if the platform does not | |
570 | * provide a BOM (which we require), since we want to match the behavior | |
571 | * of the system tools and libc as much as possible. | |
aab2a1ae TB |
572 | */ |
573 | if (same_utf_encoding("UTF-16LE-BOM", out_encoding)) { | |
574 | bom_str = utf16_le_bom; | |
575 | bom_len = sizeof(utf16_le_bom); | |
576 | out_encoding = "UTF-16LE"; | |
577 | } else if (same_utf_encoding("UTF-16BE-BOM", out_encoding)) { | |
578 | bom_str = utf16_be_bom; | |
579 | bom_len = sizeof(utf16_be_bom); | |
580 | out_encoding = "UTF-16BE"; | |
79444c92 | 581 | #ifdef ICONV_OMITS_BOM |
582 | } else if (same_utf_encoding("UTF-16", out_encoding)) { | |
583 | bom_str = utf16_be_bom; | |
584 | bom_len = sizeof(utf16_be_bom); | |
585 | out_encoding = "UTF-16BE"; | |
586 | } else if (same_utf_encoding("UTF-32", out_encoding)) { | |
587 | bom_str = utf32_be_bom; | |
588 | bom_len = sizeof(utf32_be_bom); | |
589 | out_encoding = "UTF-32BE"; | |
590 | #endif | |
aab2a1ae TB |
591 | } |
592 | ||
76759c7d | 593 | conv = iconv_open(out_encoding, in_encoding); |
5c680be1 | 594 | if (conv == (iconv_t) -1) { |
3270741e JH |
595 | in_encoding = fallback_encoding(in_encoding); |
596 | out_encoding = fallback_encoding(out_encoding); | |
597 | ||
5c680be1 JK |
598 | conv = iconv_open(out_encoding, in_encoding); |
599 | if (conv == (iconv_t) -1) | |
600 | return NULL; | |
601 | } | |
aab2a1ae | 602 | out = reencode_string_iconv(in, insz, conv, bom_len, outsz); |
b45974a6 | 603 | iconv_close(conv); |
aab2a1ae TB |
604 | if (out && bom_str && bom_len) |
605 | memcpy(out, bom_str, bom_len); | |
b45974a6 JH |
606 | return out; |
607 | } | |
608 | #endif | |
6cd3c053 | 609 | |
10ecb82e LS |
610 | static int has_bom_prefix(const char *data, size_t len, |
611 | const char *bom, size_t bom_len) | |
612 | { | |
613 | return data && bom && (len >= bom_len) && !memcmp(data, bom, bom_len); | |
614 | } | |
615 | ||
10ecb82e LS |
616 | int has_prohibited_utf_bom(const char *enc, const char *data, size_t len) |
617 | { | |
618 | return ( | |
619 | (same_utf_encoding("UTF-16BE", enc) || | |
620 | same_utf_encoding("UTF-16LE", enc)) && | |
621 | (has_bom_prefix(data, len, utf16_be_bom, sizeof(utf16_be_bom)) || | |
622 | has_bom_prefix(data, len, utf16_le_bom, sizeof(utf16_le_bom))) | |
623 | ) || ( | |
624 | (same_utf_encoding("UTF-32BE", enc) || | |
625 | same_utf_encoding("UTF-32LE", enc)) && | |
626 | (has_bom_prefix(data, len, utf32_be_bom, sizeof(utf32_be_bom)) || | |
627 | has_bom_prefix(data, len, utf32_le_bom, sizeof(utf32_le_bom))) | |
628 | ); | |
629 | } | |
630 | ||
c6e48652 LS |
631 | int is_missing_required_utf_bom(const char *enc, const char *data, size_t len) |
632 | { | |
633 | return ( | |
634 | (same_utf_encoding(enc, "UTF-16")) && | |
635 | !(has_bom_prefix(data, len, utf16_be_bom, sizeof(utf16_be_bom)) || | |
636 | has_bom_prefix(data, len, utf16_le_bom, sizeof(utf16_le_bom))) | |
637 | ) || ( | |
638 | (same_utf_encoding(enc, "UTF-32")) && | |
639 | !(has_bom_prefix(data, len, utf32_be_bom, sizeof(utf32_be_bom)) || | |
640 | has_bom_prefix(data, len, utf32_le_bom, sizeof(utf32_le_bom))) | |
641 | ); | |
642 | } | |
643 | ||
6cd3c053 KS |
644 | /* |
645 | * Returns first character length in bytes for multi-byte `text` according to | |
646 | * `encoding`. | |
647 | * | |
648 | * - The `text` pointer is updated to point at the next character. | |
649 | * - When `remainder_p` is not NULL, on entry `*remainder_p` is how much bytes | |
650 | * we can consume from text, and on exit `*remainder_p` is reduced by returned | |
651 | * character length. Otherwise `text` is treated as limited by NUL. | |
652 | */ | |
653 | int mbs_chrlen(const char **text, size_t *remainder_p, const char *encoding) | |
654 | { | |
655 | int chrlen; | |
656 | const char *p = *text; | |
657 | size_t r = (remainder_p ? *remainder_p : SIZE_MAX); | |
658 | ||
659 | if (r < 1) | |
660 | return 0; | |
661 | ||
662 | if (is_encoding_utf8(encoding)) { | |
663 | pick_one_utf8_char(&p, &r); | |
664 | ||
665 | chrlen = p ? (p - *text) | |
666 | : 1 /* not valid UTF-8 -> raw byte sequence */; | |
667 | } | |
668 | else { | |
669 | /* | |
670 | * TODO use iconv to decode one char and obtain its chrlen | |
671 | * for now, let's treat encodings != UTF-8 as one-byte | |
672 | */ | |
673 | chrlen = 1; | |
674 | } | |
675 | ||
676 | *text += chrlen; | |
677 | if (remainder_p) | |
678 | *remainder_p -= chrlen; | |
679 | ||
680 | return chrlen; | |
681 | } | |
6162a1d3 JK |
682 | |
683 | /* | |
6aaf956b JK |
684 | * Pick the next char from the stream, ignoring codepoints an HFS+ would. |
685 | * Note that this is _not_ complete by any means. It's just enough | |
6162a1d3 JK |
686 | * to make is_hfs_dotgit() work, and should not be used otherwise. |
687 | */ | |
688 | static ucs_char_t next_hfs_char(const char **in) | |
689 | { | |
690 | while (1) { | |
691 | ucs_char_t out = pick_one_utf8_char(in, NULL); | |
692 | /* | |
693 | * check for malformed utf8. Technically this | |
694 | * gets converted to a percent-sequence, but | |
695 | * returning 0 is good enough for is_hfs_dotgit | |
696 | * to realize it cannot be .git | |
697 | */ | |
698 | if (!*in) | |
699 | return 0; | |
700 | ||
701 | /* these code points are ignored completely */ | |
702 | switch (out) { | |
703 | case 0x200c: /* ZERO WIDTH NON-JOINER */ | |
704 | case 0x200d: /* ZERO WIDTH JOINER */ | |
705 | case 0x200e: /* LEFT-TO-RIGHT MARK */ | |
706 | case 0x200f: /* RIGHT-TO-LEFT MARK */ | |
707 | case 0x202a: /* LEFT-TO-RIGHT EMBEDDING */ | |
708 | case 0x202b: /* RIGHT-TO-LEFT EMBEDDING */ | |
709 | case 0x202c: /* POP DIRECTIONAL FORMATTING */ | |
710 | case 0x202d: /* LEFT-TO-RIGHT OVERRIDE */ | |
711 | case 0x202e: /* RIGHT-TO-LEFT OVERRIDE */ | |
712 | case 0x206a: /* INHIBIT SYMMETRIC SWAPPING */ | |
713 | case 0x206b: /* ACTIVATE SYMMETRIC SWAPPING */ | |
714 | case 0x206c: /* INHIBIT ARABIC FORM SHAPING */ | |
715 | case 0x206d: /* ACTIVATE ARABIC FORM SHAPING */ | |
716 | case 0x206e: /* NATIONAL DIGIT SHAPES */ | |
717 | case 0x206f: /* NOMINAL DIGIT SHAPES */ | |
718 | case 0xfeff: /* ZERO WIDTH NO-BREAK SPACE */ | |
719 | continue; | |
720 | } | |
721 | ||
6aaf956b | 722 | return out; |
6162a1d3 JK |
723 | } |
724 | } | |
725 | ||
0fc333ba JK |
726 | static int is_hfs_dot_generic(const char *path, |
727 | const char *needle, size_t needle_len) | |
6162a1d3 JK |
728 | { |
729 | ucs_char_t c; | |
730 | ||
6aaf956b JK |
731 | c = next_hfs_char(&path); |
732 | if (c != '.') | |
733 | return 0; | |
6aaf956b JK |
734 | |
735 | /* | |
736 | * there's a great deal of other case-folding that occurs | |
0fc333ba JK |
737 | * in HFS+, but this is enough to catch our fairly vanilla |
738 | * hard-coded needles. | |
6aaf956b | 739 | */ |
0fc333ba JK |
740 | for (; needle_len > 0; needle++, needle_len--) { |
741 | c = next_hfs_char(&path); | |
742 | ||
743 | /* | |
744 | * We know our needles contain only ASCII, so we clamp here to | |
745 | * make the results of tolower() sane. | |
746 | */ | |
747 | if (c > 127) | |
748 | return 0; | |
749 | if (tolower(c) != *needle) | |
750 | return 0; | |
751 | } | |
752 | ||
6162a1d3 JK |
753 | c = next_hfs_char(&path); |
754 | if (c && !is_dir_sep(c)) | |
755 | return 0; | |
756 | ||
757 | return 1; | |
758 | } | |
dde843e7 | 759 | |
0fc333ba JK |
760 | /* |
761 | * Inline wrapper to make sure the compiler resolves strlen() on literals at | |
762 | * compile time. | |
763 | */ | |
764 | static inline int is_hfs_dot_str(const char *path, const char *needle) | |
765 | { | |
766 | return is_hfs_dot_generic(path, needle, strlen(needle)); | |
767 | } | |
768 | ||
769 | int is_hfs_dotgit(const char *path) | |
770 | { | |
771 | return is_hfs_dot_str(path, "git"); | |
772 | } | |
773 | ||
774 | int is_hfs_dotgitmodules(const char *path) | |
775 | { | |
776 | return is_hfs_dot_str(path, "gitmodules"); | |
777 | } | |
778 | ||
779 | int is_hfs_dotgitignore(const char *path) | |
780 | { | |
781 | return is_hfs_dot_str(path, "gitignore"); | |
782 | } | |
783 | ||
784 | int is_hfs_dotgitattributes(const char *path) | |
785 | { | |
786 | return is_hfs_dot_str(path, "gitattributes"); | |
787 | } | |
788 | ||
801ed010 JK |
789 | int is_hfs_dotmailmap(const char *path) |
790 | { | |
791 | return is_hfs_dot_str(path, "mailmap"); | |
792 | } | |
793 | ||
dde843e7 JH |
794 | const char utf8_bom[] = "\357\273\277"; |
795 | ||
796 | int skip_utf8_bom(char **text, size_t len) | |
797 | { | |
798 | if (len < strlen(utf8_bom) || | |
799 | memcmp(*text, utf8_bom, strlen(utf8_bom))) | |
800 | return 0; | |
801 | *text += strlen(utf8_bom); | |
802 | return 1; | |
803 | } | |
110dcda5 KN |
804 | |
805 | void strbuf_utf8_align(struct strbuf *buf, align_type position, unsigned int width, | |
806 | const char *s) | |
807 | { | |
522cc87f | 808 | size_t slen = strlen(s); |
110dcda5 KN |
809 | int display_len = utf8_strnwidth(s, slen, 0); |
810 | int utf8_compensation = slen - display_len; | |
811 | ||
812 | if (display_len >= width) { | |
813 | strbuf_addstr(buf, s); | |
814 | return; | |
815 | } | |
816 | ||
817 | if (position == ALIGN_LEFT) | |
818 | strbuf_addf(buf, "%-*s", width + utf8_compensation, s); | |
819 | else if (position == ALIGN_MIDDLE) { | |
820 | int left = (width - display_len) / 2; | |
821 | strbuf_addf(buf, "%*s%-*s", left, "", width - left + utf8_compensation, s); | |
822 | } else if (position == ALIGN_RIGHT) | |
823 | strbuf_addf(buf, "%*s", width + utf8_compensation, s); | |
824 | } |