Attached patch fixes a problem with wide character handling. Currently
it is incorrectly emitting byte order marks (not entirely correct ones
either) for random choice of characterset names. The current code uses
the same function to find codesets as it code for charactersets - the
subsequent logic didn't handle characterset name as input (rather than
code name)
Commit:
Fix incorrect output of BOM when converting charactersets by name. Only
affective when iconv/locale enabled.
Signed-off-by: Ed Wildgoose ([email protected])
--- libc/misc/wchar/wchar.c.orig 2011-10-30 20:00:01.000000000 +0000
+++ libc/misc/wchar/wchar.c 2011-10-30 20:02:55.000000000 +0000
@@ -1308,9 +1308,9 @@
&& ((fromcodeset = find_codeset(fromcode)) != 0)) {
if ((px = malloc(sizeof(_UC_iconv_t))) != NULL) {
px->tocodeset = tocodeset;
- px->tobom0 = px->tobom = (tocodeset & 0x10) >> 4;
+ px->tobom0 = px->tobom = (tocodeset >= 0xe0) ?
(tocodeset & 0x10) >> 4 : 0;
px->fromcodeset0 = px->fromcodeset = fromcodeset;
- px->frombom0 = px->frombom = (fromcodeset & 0x10) >> 4;
+ px->frombom0 = px->frombom = (fromcodeset >= 0xe0) ?
(fromcodeset & 0x10) >> 4 : 0;
px->skip_invalid_input = px->tostate.__mask
= px->fromstate.__mask = 0;
return (iconv_t) px;
_______________________________________________
uClibc mailing list
[email protected]
http://lists.busybox.net/mailman/listinfo/uclibc
