Fix error introduced in [3e5e37f83b058f3d] for Tcl_UniCharToUtf, and add test.

2 files changed, 25 insertions, 1 deletions

diff --git a/generic/tclUtf.c b/generic/tclUtf.c
index ab27f1b..bef32f0 100644
--- a/generic/tclUtf.c
+++ b/generic/tclUtf.c
@@ -256,7 +256,7 @@ Tcl_UniCharToUtf(
 		    ch += 0x40;
 		    /* Fill buffer with specific 3-byte (invalid) byte combination,
 		       so following low surrogate can recognize it and combine */
-		    buf[2] = (char) (        0x03 & ch);
+		    buf[2] = (char) ((ch << 4) & 0x30);
 		    buf[1] = (char) (0x80 | (0x3F & (ch >> 2)));
 		    buf[0] = (char) (0xF0 | (0x07 & (ch >> 8)));
 		    return 1;
diff --git a/tests/encoding.test b/tests/encoding.test
index 1971360..8351c91 100644
--- a/tests/encoding.test
+++ b/tests/encoding.test
@@ -487,6 +487,30 @@ test encoding-16.8 {Utf32ToUtfProc} -body {
     list $val [format %x [scan $val %c]]
 } -result "\uFFFD fffd"
 
+test encoding-16.8 {
+    Utf16ToUtfProc, Tcl_UniCharToUtf, surrogate pairs in utf-16
+} -body {
+    apply [list {} {
+	for {set i 0xD800} {$i < 0xDBFF} {incr i} {
+	    for {set j 0xDC00} {$j < 0xDFFF} {incr j} {
+		set string [binary format S2 [list $i $j]]
+		set status [catch {
+		    set decoded [encoding convertfrom utf-16be $string]
+		    set encoded [encoding convertto utf-16be $decoded]
+		}]
+		if {$status || ( $encoded ne $string )} {
+		    return [list [format %x $i] [format %x $j]]
+		}
+	    }
+	}
+	return done
+    } [namespace current]]
+} -result done
+
+
+
+
+
 test encoding-17.1 {UtfToUtf16Proc} -body {
     encoding convertto utf-16 "\U460DC"
 } -result "\xD8\xD8\xDC\xDC"