Implement special "string totitle" for Extended Georgian characters (new behavior in Unicode 11)

3 files changed, 22 insertions, 4 deletions

diff --git a/generic/tclUtf.c b/generic/tclUtf.c
index 319bfa0..1d73a7a 100644
--- a/generic/tclUtf.c
+++ b/generic/tclUtf.c
@@ -1034,7 +1034,10 @@ Tcl_UtfToTitle(
 	    lowChar = (((lowChar & 0x3ff) << 10) | (ch & 0x3ff)) + 0x10000;
 	}
 #endif
-	lowChar = Tcl_UniCharToLower(lowChar);
+	/* Special exception for Gregorian characters, which don't have titlecase */
+	if ((lowChar < 0x1C90) || (lowChar >= 0x1CC0)) {
+	    lowChar = Tcl_UniCharToLower(lowChar);
+	}
 
 	if (bytes < TclUtfCount(lowChar)) {
 	    memcpy(dst, src, (size_t) bytes);
@@ -1355,8 +1358,9 @@ Tcl_UniCharToLower(
 {
     if (!UNICODE_OUT_OF_RANGE(ch)) {
 	int info = GetUniCharInfo(ch);
+	int mode = GetCaseType(info);
 
-	if (GetCaseType(info) & 0x02) {
+	if ((mode & 0x02) && (mode != 0x7)) {
 	    ch += GetDelta(info);
 	}
     }
@@ -1392,7 +1396,9 @@ Tcl_UniCharToTitle(
 	     * Subtract or add one depending on the original case.
 	     */
 
-	    ch += ((mode & 0x4) ? -1 : 1);
+	    if (mode != 0x7) {
+		ch += ((mode & 0x4) ? -1 : 1);
+	    }
 	} else if (mode == 0x4) {
 	    ch -= GetDelta(info);
 	}
diff --git a/tests/utf.test b/tests/utf.test
index af471e1..39818cc 100644
--- a/tests/utf.test
+++ b/tests/utf.test
@@ -250,6 +250,9 @@ test utf-11.3 {Tcl_UtfToUpper} {
 test utf-11.4 {Tcl_UtfToUpper} {
     string toupper \u01e3ab
 } \u01e2AB
+test utf-11.5 {Tcl_UtfToUpper Georgian (new in Unicode 11)} {
+    string toupper \u10d0\u1c90
+} \u1c90\u1c90
 
 test utf-12.1 {Tcl_UtfToLower} {
     string tolower {}
@@ -263,6 +266,9 @@ test utf-12.3 {Tcl_UtfToLower} {
 test utf-12.4 {Tcl_UtfToLower} {
     string tolower \u01e2AB
 } \u01e3ab
+test utf-12.5 {Tcl_UtfToLower Georgian (new in Unicode 11)} {
+    string tolower \u10d0\u1c90
+} \u10d0\u10d0
 
 test utf-13.1 {Tcl_UtfToTitle} {
     string totitle {}
@@ -276,6 +282,12 @@ test utf-13.3 {Tcl_UtfToTitle} {
 test utf-13.4 {Tcl_UtfToTitle} {
     string totitle \u01f3ab
 } \u01f2ab
+test utf-13.5 {Tcl_UtfToTitle Georgian (new in Unicode 11)} {
+    string totitle \u10d0\u1c90
+} \u10d0\u1c90
+test utf-13.6 {Tcl_UtfToTitle Georgian (new in Unicode 11)} {
+    string totitle \u1c90\u10d0
+} \u1c90\u10d0
 
 test utf-14.1 {Tcl_UtfNcasecmp} {
     string compare -nocase a b
diff --git a/tools/uniClass.tcl b/tools/uniClass.tcl
index 8047894..86ec931 100644
--- a/tools/uniClass.tcl
+++ b/tools/uniClass.tcl
@@ -66,7 +66,7 @@ proc genTable {type} {
     for {set i 0} {$i <= 0x10ffff} {incr i} {
     if {$i == 0xd800} {
 	# Skip surrogates
-	set i 0xdc00
+	set i 0xe000
     }
 	if {[string is $type [format %c $i]]} {
 	    if {$i == ($last + 1)} {