projects
/
oweals
/
busybox.git
/ blobdiff
commit
grep
author
committer
pickaxe
?
search:
re
summary
|
shortlog
|
log
|
commit
|
commitdiff
|
tree
raw
|
inline
| side by side
libbb: fix vasprintf implementation
[oweals/busybox.git]
/
libbb
/
unicode.c
diff --git
a/libbb/unicode.c
b/libbb/unicode.c
index 773a0744e6c183b8230515f8c4129c60fb8ef8b6..544528acd088379460492717cc3af23796412bf2 100644
(file)
--- a/
libbb/unicode.c
+++ b/
libbb/unicode.c
@@
-39,7
+39,7
@@
void FAST_FUNC check_unicode_in_env(void)
unicode_is_enabled = 1;
lang = getenv("LANG");
unicode_is_enabled = 1;
lang = getenv("LANG");
- if (!lang || !
strstr(lang, ".utf8"
))
+ if (!lang || !
(strstr(lang, ".utf") || strstr(lang, ".UTF")
))
return;
unicode_is_enabled = 2;
return;
unicode_is_enabled = 2;
@@
-48,7
+48,7
@@
void FAST_FUNC check_unicode_in_env(void)
static size_t wcrtomb_internal(char *s, wchar_t wc)
{
static size_t wcrtomb_internal(char *s, wchar_t wc)
{
- int n;
+ int n
, i
;
uint32_t v = wc;
if (v <= 0x7f) {
uint32_t v = wc;
if (v <= 0x7f) {
@@
-59,34
+59,26
@@
static size_t wcrtomb_internal(char *s, wchar_t wc)
/* RFC 3629 says that Unicode ends at 10FFFF,
* but we cover entire 32 bits */
/* RFC 3629 says that Unicode ends at 10FFFF,
* but we cover entire 32 bits */
- n = 2;
/* 4000000-FFFFFFFF -> 111111tt 10tttttt 10zzzzzz 10zzyyyy 10yyyyxx 10xxxxxx */
/* 4000000-FFFFFFFF -> 111111tt 10tttttt 10zzzzzz 10zzyyyy 10yyyyxx 10xxxxxx */
- if (v >= 0x4000000) {
- s[5] = (wc & 0x3f) | 0x80;
- wc = (uint32_t)wc >> 6; /* ensuring that high bits are 0 */
- n++;
- }
/* 200000-3FFFFFF -> 111110tt 10zzzzzz 10zzyyyy 10yyyyxx 10xxxxxx */
/* 200000-3FFFFFF -> 111110tt 10zzzzzz 10zzyyyy 10yyyyxx 10xxxxxx */
- if (v >= 0x200000) {
- s[4] = (wc & 0x3f) | 0x80;
- wc >>= 6;
- n++;
- }
/* 10000-1FFFFF -> 11110zzz 10zzyyyy 10yyyyxx 10xxxxxx */
/* 10000-1FFFFF -> 11110zzz 10zzyyyy 10yyyyxx 10xxxxxx */
- if (v >= 0x10000) {
- s[3] = (wc & 0x3f) | 0x80;
- wc >>= 6;
+ /* 800-FFFF -> 1110yyyy 10yyyyxx 10xxxxxx */
+ /* 80-7FF -> 110yyyxx 10xxxxxx */
+
+ /* How many bytes do we need? */
+ n = 2;
+ /* (0x80000000+ would result in n = 7, limiting n to 6) */
+ while (v >= 0x800 && n < 6) {
+ v >>= 5;
n++;
}
n++;
}
- /* 800-FFFF -> 1110yyyy 10yyyyxx 10xxxxxx */
- if (v >= 0x800) {
- s[2] = (wc & 0x3f) | 0x80;
+ /* Fill bytes n-1..1 */
+ i = n;
+ while (--i) {
+ s[i] = (wc & 0x3f) | 0x80;
wc >>= 6;
wc >>= 6;
- n++;
}
}
- /* 80-7FF -> 110yyyxx 10xxxxxx */
- s[1] = (wc & 0x3f) | 0x80;
- wc >>= 6;
+ /* Fill byte 0 */
s[0] = wc | (uint8_t)(0x3f00 >> n);
return n;
}
s[0] = wc | (uint8_t)(0x3f00 >> n);
return n;
}