aboutsummaryrefslogtreecommitdiffstats
path: root/src/mds-kbdc/string.c
diff options
context:
space:
mode:
authorMattias Andrée <maandree@operamail.com>2014-12-06 19:28:59 +0100
committerMattias Andrée <maandree@operamail.com>2014-12-06 19:29:31 +0100
commit977cc038e1d0cfa9455146bcdab568d32401f37f (patch)
tree52f9a8f5f083bf8f8c30d8ad6e7fff040cfbcc06 /src/mds-kbdc/string.c
parentmds-kbdc: parse numerals (diff)
downloadmds-977cc038e1d0cfa9455146bcdab568d32401f37f.tar.gz
mds-977cc038e1d0cfa9455146bcdab568d32401f37f.tar.bz2
mds-977cc038e1d0cfa9455146bcdab568d32401f37f.tar.xz
mds-kbdc: parse_keys
Signed-off-by: Mattias Andrée <maandree@operamail.com>
Diffstat (limited to '')
-rw-r--r--src/mds-kbdc/string.c13
1 files changed, 10 insertions, 3 deletions
diff --git a/src/mds-kbdc/string.c b/src/mds-kbdc/string.c
index 69f8f25..d6befe3 100644
--- a/src/mds-kbdc/string.c
+++ b/src/mds-kbdc/string.c
@@ -98,22 +98,29 @@ char* string_encode(const char32_t* restrict string)
char* restrict rc;
/* Allocated Modified UTF-8 string. */
- if (xmalloc(rc, 6 * n + 1, char))
+ if (xmalloc(rc, 7 * n + 1, char))
return NULL;
/* Convert to Modified UTF-8. */
for (i = j = 0; i < n; i++)
{
#define _c(s) rc[j++] = (char)(((c >> (s)) & 0x3F) | 0x80)
-#define _t(s) c < (char32_t)(1L << s)
+#define _t(s) (0 < c) && ((uint32_t)c < (uint32_t)(1ULL << s))
char32_t c = string[i];
if (c == 0) rc[j++] = (char)0xC0, rc[j++] = (char)0x80;
else if (_t( 7)) rc[j++] = (char)c;
else if (_t(11)) rc[j++] = (char)((c >> 6) | 0xC0), _c( 0);
else if (_t(16)) rc[j++] = (char)((c >> 12) | 0xE0), _c( 6), _c( 0);
else if (_t(21)) rc[j++] = (char)((c >> 18) | 0xF0), _c(12), _c( 6), _c( 0);
+ /* UTF-8 actually ends here, fits 32 planes. */
else if (_t(26)) rc[j++] = (char)((c >> 24) | 0xF8), _c(18), _c(12), _c( 6), _c(0);
- else rc[j++] = (char)((c >> 30) | 0xFC), _c(24), _c(18), _c(12), _c(6), _c(0);
+ else if (_t(31)) rc[j++] = (char)((c >> 30) | 0xFC), _c(24), _c(18), _c(12), _c(6), _c(0);
+ /* The orginal UTF-8 specification ended here, fits 31 bits.
+ * However, we added another byte so we can fit 32 bits
+ * (actually we ca now fit 36 bits.)
+ * However, we only needed this in `string_decode` which would
+ * not require any changed, but we added it here for symmetry. */
+ else rc[j++] = (char)((c >> 30) | 0xFE), _c(30), _c(24), _c(18), _c(12), _c(6), _c(0);
#undef _t
#undef _c
}