summaryrefslogtreecommitdiffstats
path: root/generic/tclEncoding.c
diff options
context:
space:
mode:
authorjan.nijtmans <nijtmans@users.sourceforge.net>2019-12-09 11:31:25 (GMT)
committerjan.nijtmans <nijtmans@users.sourceforge.net>2019-12-09 11:31:25 (GMT)
commit54b0e77083c64793570653ba6e506f173eef82f0 (patch)
treefb1dc054e0c14e1911abe23c1729f46f1a7f00b6 /generic/tclEncoding.c
parent39bf69006a6847b4068bd3b9b8debda81197502a (diff)
parentc66173309fa8adc4c4159bdeae016c7f9e2cbae0 (diff)
downloadtcl-54b0e77083c64793570653ba6e506f173eef82f0.zip
tcl-54b0e77083c64793570653ba6e506f173eef82f0.tar.gz
tcl-54b0e77083c64793570653ba6e506f173eef82f0.tar.bz2
Merge 8.6
Diffstat (limited to 'generic/tclEncoding.c')
-rw-r--r--generic/tclEncoding.c19
1 files changed, 13 insertions, 6 deletions
diff --git a/generic/tclEncoding.c b/generic/tclEncoding.c
index 350a13c..a326856 100644
--- a/generic/tclEncoding.c
+++ b/generic/tclEncoding.c
@@ -2409,17 +2409,24 @@ UtfToUtfProc(
dst += Tcl_UniCharToUtf(*chPtr, dst);
} else {
src += TclUtfToUniChar(src, chPtr);
- if ((*chPtr | 0x3FF) == 0xDBFF) {
- /* A high surrogate character is detected, handle especially */
+ if ((*chPtr | 0x7FF) == 0xDFFF) {
+ /* A surrogate character is detected, handle especially */
Tcl_UniChar low = *chPtr;
- if (src <= srcEnd-3) {
- Tcl_UtfToUniChar(src, &low);
- }
- if ((low | 0x3FF) != 0xDFFF) {
+ size_t len = (src <= srcEnd-3) ? Tcl_UtfToUniChar(src, &low) : 0;
+ if (((low | 0x3FF) != 0xDFFF) || !(*chPtr & 0x800)) {
*dst++ = (char) (((*chPtr >> 12) | 0xE0) & 0xEF);
*dst++ = (char) (((*chPtr >> 6) | 0x80) & 0xBF);
*dst++ = (char) ((*chPtr | 0x80) & 0xBF);
continue;
+ } else if (TCL_UTF_MAX <= 4) {
+ int full = (((*chPtr & 0x3FF) << 10) | (low & 0x3FF)) + 0x10000;
+ *dst++ = (char) (((full >> 18) | 0xF0) & 0xF7);
+ *dst++ = (char) (((full >> 12) | 0x80) & 0xBF);
+ *dst++ = (char) (((full >> 6) | 0x80) & 0xBF);
+ *dst++ = (char) ((full | 0x80) & 0xBF);
+ *chPtr = 0;
+ src += len;
+ continue;
}
}
dst += Tcl_UniCharToUtf(*chPtr, dst);