fold: unicode support. Based on a patch by Tomas Heinrich <heinrich.tomas@gmail.com>
General Unicode support is tweaked to expose unicode_status.
function old new delta
init_unicode - 77 +77
write2stdout - 19 +19
adjust_column 68 71 +3
unicode_status - 1 +1
unicode_is_enabled 1 - -1
grep_main 780 773 -7
fold_main 619 552 -67
check_unicode_in_env 77 - -77
------------------------------------------------------------------------------
(add/remove: 3/2 grow/shrink: 1/2 up/down: 100/-152) Total: -52 bytes
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
diff --git a/libbb/unicode.c b/libbb/unicode.c
index 544528a..9d316df 100644
--- a/libbb/unicode.c
+++ b/libbb/unicode.c
@@ -7,7 +7,9 @@
* Licensed under GPL version 2, see file LICENSE in this tarball for details.
*/
#include "libbb.h"
-# include "unicode.h"
+#include "unicode.h"
+
+uint8_t unicode_status;
size_t FAST_FUNC bb_mbstrlen(const char *string)
{
@@ -17,32 +19,38 @@
return width;
}
-#if !ENABLE_LOCALE_SUPPORT
+#if ENABLE_LOCALE_SUPPORT
+
+/* Unicode support using libc */
+
+void FAST_FUNC init_unicode(void)
+{
+ /* In unicode, this is a one character string */
+ static const char unicode_0x394[] = { 0xce, 0x94, 0 };
+
+ if (unicode_status != UNICODE_UNKNOWN)
+ return;
+
+ unicode_status = bb_mbstrlen(unicode_0x394) == 1 ? UNICODE_ON : UNICODE_OFF;
+}
+
+#else
/* Crude "locale support" which knows only C and Unicode locales */
-/* unicode_is_enabled:
- * 0: not known yet,
- * 1: not unicode (IOW: assuming one char == one byte)
- * 2: unicode
- */
-# if !ENABLE_FEATURE_CHECK_UNICODE_IN_ENV
-# define unicode_is_enabled 2
-# else
-static smallint unicode_is_enabled;
-void FAST_FUNC check_unicode_in_env(void)
+# if ENABLE_FEATURE_CHECK_UNICODE_IN_ENV
+void FAST_FUNC init_unicode(void)
{
char *lang;
- if (unicode_is_enabled)
+ if (unicode_status != UNICODE_UNKNOWN)
return;
- unicode_is_enabled = 1;
+ unicode_status = UNICODE_OFF;
lang = getenv("LANG");
if (!lang || !(strstr(lang, ".utf") || strstr(lang, ".UTF")))
return;
-
- unicode_is_enabled = 2;
+ unicode_status = UNICODE_ON;
}
# endif
@@ -85,7 +93,7 @@
size_t FAST_FUNC wcrtomb(char *s, wchar_t wc, mbstate_t *ps UNUSED_PARAM)
{
- if (unicode_is_enabled != 2) {
+ if (unicode_status != UNICODE_ON) {
*s = wc;
return 1;
}
@@ -97,7 +105,7 @@
{
size_t org_n = n;
- if (unicode_is_enabled != 2) {
+ if (unicode_status != UNICODE_ON) {
while (n) {
wchar_t c = *src++;
*dest++ = c;
@@ -137,7 +145,7 @@
{
size_t org_n = n;
- if (unicode_is_enabled != 2) {
+ if (unicode_status != UNICODE_ON) {
while (n) {
unsigned char c = *src++;