Changeset 3621 in MondoRescue for branches/3.3/mindi-busybox/libbb/unicode.c

Timestamp:

Dec 20, 2016, 4:07:32 PM (7 years ago)

Author:

Bruno Cornec

Message:

New 3?3 banch for incorporation of latest busybox 1.25. Changing minor version to handle potential incompatibilities.

Location:

branches/3.3

Files:

• . (copied) (copied from branches/3.2 )
• mindi-busybox/libbb/unicode.c (modified) (5 diffs)

branches/3.3/mindi-busybox/libbb/unicode.c

@@ -29 +29 @@
     size_t width;
 
+    /* We pass "" instead of "C" because some libc's have
+     * non-ASCII default locale for setlocale("") call
+     * (this allows users of such libc to have Unicoded
+     * system without having to mess with env).
+     *
+     * We set LC_CTYPE because (a) we may be called with $LC_CTYPE
+     * value in LANG, not with $LC_ALL, (b) internationalized
+     * LC_NUMERIC and LC_TIME are more PITA than benefit
+     * (for one, some utilities have hard time with comma
+     * used as a fractional separator).
+     */
 //TODO: avoid repeated calls by caching last string?
-    setlocale(LC_ALL, (LANG && LANG[0]) ? LANG : "C");
+    setlocale(LC_CTYPE, LANG ? LANG : "");
 
     /* In unicode, this is a one character string */
-// can use unicode_strlen(string) too, but otherwise unicode_strlen() is unused
-    width = mbstowcs(NULL, unicode_0x394, INT_MAX);
+    width = unicode_strlen(unicode_0x394);
     unicode_status = (width == 1 ? UNICODE_ON : UNICODE_OFF);
 }
@@ -40 +50 @@
 void FAST_FUNC init_unicode(void)
 {
-    if (unicode_status == UNICODE_UNKNOWN)
-        reinit_unicode(getenv("LANG"));
+    /* Some people set only $LC_CTYPE, not $LC_ALL, because they want
+     * only Unicode to be activated on their system, not the whole
+     * shebang of wrong decimal points, strange date formats and so on.
+     */
+    if (unicode_status == UNICODE_UNKNOWN) {
+        char *s = getenv("LC_ALL");
+        if (!s) s = getenv("LC_CTYPE");
+        if (!s) s = getenv("LANG");
+        reinit_unicode(s);
+    }
 }
 
@@ -59 +77 @@
 void FAST_FUNC init_unicode(void)
 {
-    if (unicode_status == UNICODE_UNKNOWN)
-        reinit_unicode(getenv("LANG"));
+    if (unicode_status == UNICODE_UNKNOWN) {
+        char *s = getenv("LC_ALL");
+        if (!s) s = getenv("LC_CTYPE");
+        if (!s) s = getenv("LANG");
+        reinit_unicode(s);
+    }
 }
 # endif
@@ -964 +986 @@
 /* The rest is mostly same for libc and for "homegrown" support */
 
-#if 0 // UNUSED
 size_t FAST_FUNC unicode_strlen(const char *string)
 {
@@ -972 +993 @@
     return width;
 }
-#endif
 
 size_t FAST_FUNC unicode_strwidth(const char *string)