1 /* vi: set sw=4 ts=4: */
3 * Unicode support routines.
5 * Copyright (C) 2009 Denys Vlasenko
7 * Licensed under GPL version 2, see file LICENSE in this tarball for details.
12 uint8_t unicode_status;
14 size_t FAST_FUNC bb_mbstrlen(const char *string)
16 size_t width = mbstowcs(NULL, string, INT_MAX);
17 if (width == (size_t)-1L)
18 return strlen(string);
22 #if ENABLE_LOCALE_SUPPORT
24 /* Unicode support using libc */
26 void FAST_FUNC init_unicode(void)
28 /* In unicode, this is a one character string */
29 static const char unicode_0x394[] = { 0xce, 0x94, 0 };
31 if (unicode_status != UNICODE_UNKNOWN)
34 unicode_status = bb_mbstrlen(unicode_0x394) == 1 ? UNICODE_ON : UNICODE_OFF;
39 /* Crude "locale support" which knows only C and Unicode locales */
41 # if ENABLE_FEATURE_CHECK_UNICODE_IN_ENV
42 void FAST_FUNC init_unicode(void)
46 if (unicode_status != UNICODE_UNKNOWN)
49 unicode_status = UNICODE_OFF;
50 lang = getenv("LANG");
51 if (!lang || !(strstr(lang, ".utf") || strstr(lang, ".UTF")))
53 unicode_status = UNICODE_ON;
57 static size_t wcrtomb_internal(char *s, wchar_t wc)
67 /* RFC 3629 says that Unicode ends at 10FFFF,
68 * but we cover entire 32 bits */
70 /* 4000000-FFFFFFFF -> 111111tt 10tttttt 10zzzzzz 10zzyyyy 10yyyyxx 10xxxxxx */
71 /* 200000-3FFFFFF -> 111110tt 10zzzzzz 10zzyyyy 10yyyyxx 10xxxxxx */
72 /* 10000-1FFFFF -> 11110zzz 10zzyyyy 10yyyyxx 10xxxxxx */
73 /* 800-FFFF -> 1110yyyy 10yyyyxx 10xxxxxx */
74 /* 80-7FF -> 110yyyxx 10xxxxxx */
76 /* How many bytes do we need? */
78 /* (0x80000000+ would result in n = 7, limiting n to 6) */
79 while (v >= 0x800 && n < 6) {
83 /* Fill bytes n-1..1 */
86 s[i] = (wc & 0x3f) | 0x80;
90 s[0] = wc | (uint8_t)(0x3f00 >> n);
94 size_t FAST_FUNC wcrtomb(char *s, wchar_t wc, mbstate_t *ps UNUSED_PARAM)
96 if (unicode_status != UNICODE_ON) {
101 return wcrtomb_internal(s, wc);
104 size_t FAST_FUNC wcstombs(char *dest, const wchar_t *src, size_t n)
108 if (unicode_status != UNICODE_ON) {
119 while (n >= MB_CUR_MAX) {
121 size_t len = wcrtomb_internal(dest, wc);
129 char tbuf[MB_CUR_MAX];
131 size_t len = wcrtomb_internal(tbuf, wc);
135 memcpy(dest, tbuf, len);
144 size_t FAST_FUNC mbstowcs(wchar_t *dest, const char *src, size_t n)
148 if (unicode_status != UNICODE_ON) {
150 unsigned char c = *src++;
163 unsigned c = (unsigned char) *src++;
174 /* 80-7FF -> 110yyyxx 10xxxxxx */
175 /* 800-FFFF -> 1110yyyy 10yyyyxx 10xxxxxx */
176 /* 10000-1FFFFF -> 11110zzz 10zzyyyy 10yyyyxx 10xxxxxx */
177 /* 200000-3FFFFFF -> 111110tt 10zzzzzz 10zzyyyy 10yyyyxx 10xxxxxx */
178 /* 4000000-FFFFFFFF -> 111111tt 10tttttt 10zzzzzz 10zzyyyy 10yyyyxx 10xxxxxx */
183 } while ((c & 0x80) && bytes < 6);
186 c = (uint8_t)(c) >> bytes;
189 unsigned ch = (unsigned char) *src++;
190 if ((ch & 0xc0) != 0x80) {
193 c = (c << 6) + (ch & 0x3f);
197 /* Need to check that c isn't produced by overlong encoding */
198 /* Example: 11000000 10000000 converts to NUL */
199 /* 11110000 10000000 10000100 10000000 converts to 0x100 */
200 /* correct encoding: 11000100 10000000 */
201 if (c <= 0x7f) { /* crude check */
203 //or maybe: c = 0xfffd; /* replacement character */
214 int FAST_FUNC iswspace(wint_t wc)
216 return (unsigned)wc <= 0x7f && isspace(wc);
219 int FAST_FUNC iswalnum(wint_t wc)
221 return (unsigned)wc <= 0x7f && isalnum(wc);
224 int FAST_FUNC iswpunct(wint_t wc)
226 return (unsigned)wc <= 0x7f && ispunct(wc);