[910] in BarnOwl Developers
[D-O-H] r896 - branches/barnowl_unicode/owl
daemon@ATHENA.MIT.EDU (asedeno@MIT.EDU)
Thu Oct 29 18:11:02 2009
Resent-From: nelhage@mit.edu
Resent-To: barnowl-dev-mtg@charon.mit.edu
To: dirty-owl-hackers@mit.edu
From: asedeno@MIT.EDU
Reply-to: dirty-owl-hackers@MIT.EDU
Date: Thu, 17 Jan 2008 14:55:23 -0500 (EST)
Author: asedeno
Date: 2008-01-17 14:55:22 -0500 (Thu, 17 Jan 2008)
New Revision: 896
Modified:
branches/barnowl_unicode/owl/owl.c
Log:
unicode/glib branch
better compliance with UTF-8 processing. Stop trying to pull in a UTF-8
character as soon as we know something has gone wrong.
Modified: branches/barnowl_unicode/owl/owl.c
===================================================================
--- branches/barnowl_unicode/owl/owl.c 2008-01-17 14:51:48 UTC (rev 895)
+++ branches/barnowl_unicode/owl/owl.c 2008-01-17 19:55:22 UTC (rev 896)
@@ -561,14 +561,26 @@
else bytes = 1;
for (i = 1; i < bytes; i++) {
- utf8buf[i] = wgetch(typwin);
- }
- if (g_utf8_validate(utf8buf, -1, NULL)) {
- j.uch = g_utf8_get_char(utf8buf);
- }
- else {
- j.ch = ERR;
- }
+ int tmp = wgetch(typwin);
+ /* If what we got was not a byte, or not a continuation byte */
+ if (tmp > 0xff || !(tmp & 0x80 && ~tmp & 0x40)) {
+ /* ill-formed UTF-8 code unit subsequence, put back the
+ char we just got. */
+ ungetch(tmp);
+ j.ch = ERR;
+ break;
+ }
+ utf8buf[i] = tmp;
+ }
+
+ if (j.ch != ERR) {
+ if (g_utf8_validate(utf8buf, -1, NULL)) {
+ j.uch = g_utf8_get_char(utf8buf);
+ }
+ else {
+ j.ch = ERR;
+ }
+ }
}
else if (j.ch <= 0x7f) {
j.uch = j.ch;
