merge core-8-branch

1 files changed, 50 insertions, 13 deletions

diff --git a/tests/utf.test b/tests/utf.test
index 95775a8..e820359 100644
--- a/tests/utf.test
+++ b/tests/utf.test
@@ -21,7 +21,7 @@ testConstraint testbytestring [llength [info commands testbytestring]]
 catch {unset x}
 
 # Some tests require support for 4-byte UTF-8 sequences
-testConstraint fullutf [expr {[format %c 0x010000] != "\ufffd"}]
+testConstraint tip389 [expr {[string length \U010000] == 2}]
 
 test utf-1.1 {Tcl_UniCharToUtf: 1 byte sequences} testbytestring {
     expr {"\x01" eq [testbytestring "\x01"]}
@@ -41,9 +41,21 @@ test utf-1.5 {Tcl_UniCharToUtf: overflowed Tcl_UniChar} testbytestring {
 test utf-1.6 {Tcl_UniCharToUtf: negative Tcl_UniChar} testbytestring {
     expr {[format %c -1] eq [testbytestring "\xef\xbf\xbd"]}
 } 1
-test utf-1.7 {Tcl_UniCharToUtf: 4 byte sequences} -constraints {fullutf testbytestring} -body {
+test utf-1.7 {Tcl_UniCharToUtf: 4 byte sequences} -constraints testbytestring -body {
     expr {"\U014e4e" eq [testbytestring "\xf0\x94\xb9\x8e"]}
 } -result 1
+test utf-1.8 {Tcl_UniCharToUtf: 3 byte sequence, upper surrogate} testbytestring {
+    expr {"\ud842" eq [testbytestring "\xed\xa1\x82"]}
+} 1
+test utf-1.9 {Tcl_UniCharToUtf: 3 byte sequence, lower surrogate} testbytestring {
+    expr {"\udc42" eq [testbytestring "\xed\xb1\x82"]}
+} 1
+test utf-1.10 {Tcl_UniCharToUtf: 3 byte sequence, upper surrogate} testbytestring {
+    expr {[format %c 0xd842] eq [testbytestring "\xed\xa1\x82"]}
+} 1
+test utf-1.11 {Tcl_UniCharToUtf: 3 byte sequence, lower surrogate} testbytestring {
+    expr {[format %c 0xdc42] eq [testbytestring "\xed\xb1\x82"]}
+} 1
 
 test utf-2.1 {Tcl_UtfToUniChar: low ascii} {
     string length "abc"
@@ -66,10 +78,10 @@ test utf-2.6 {Tcl_UtfToUniChar: lead (3-byte) followed by 1 trail} testbytestrin
 test utf-2.7 {Tcl_UtfToUniChar: lead (3-byte) followed by 2 trail} testbytestring {
     string length [testbytestring "\xE4\xb9\x8e"]
 } {1}
-test utf-2.8 {Tcl_UtfToUniChar: lead (4-byte) followed by 3 trail} -constraints {fullutf testbytestring} -body {
+test utf-2.8 {Tcl_UtfToUniChar: lead (4-byte) followed by 3 trail} -constraints {tip389 testbytestring} -body {
     string length [testbytestring "\xF0\x90\x80\x80"]
 } -result {2}
-test utf-2.9 {Tcl_UtfToUniChar: lead (4-byte) followed by 3 trail} -constraints {fullutf testbytestring} -body {
+test utf-2.9 {Tcl_UtfToUniChar: lead (4-byte) followed by 3 trail} -constraints {tip389 testbytestring} -body {
     string length [testbytestring "\xF4\x8F\xBF\xBF"]
 } -result {2}
 test utf-2.10 {Tcl_UtfToUniChar: lead (4-byte) followed by 3 trail, underflow} testbytestring {
@@ -146,6 +158,12 @@ test utf-8.3 {Tcl_UniCharAtIndex: index > 0} {
 test utf-8.4 {Tcl_UniCharAtIndex: index > 0} {
     string index \u4e4e\u25a\xff\u543 2
 } "\uff"
+test utf-8.5 {Tcl_UniCharAtIndex: upper surrogate} {
+    string index \ud842 0
+} "\ud842"
+test utf-8.5 {Tcl_UniCharAtIndex: lower surrogate} {
+    string index \udc42 0
+} "\udc42"
 
 test utf-9.1 {Tcl_UtfAtIndex: index = 0} {
     string range abcd 0 2
@@ -228,15 +246,13 @@ bsCheck \U4e21	20001
 bsCheck \U004e21	20001
 bsCheck \U00004e21	20001
 bsCheck \U0000004e21	78
-if {[testConstraint fullutf]} {
-    bsCheck \U00110000	69632
-    bsCheck \U01100000	69632
-    bsCheck \U11000000	69632
-    bsCheck \U0010FFFF	1114111
-    bsCheck \U010FFFF0	1114111
-    bsCheck \U10FFFF00	1114111
-    bsCheck \UFFFFFFFF	1048575
-}
+bsCheck \U00110000	69632
+bsCheck \U01100000	69632
+bsCheck \U11000000	69632
+bsCheck \U0010FFFF	1114111
+bsCheck \U010FFFF0	1114111
+bsCheck \U10FFFF00	1114111
+bsCheck \UFFFFFFFF	1048575
 
 test utf-11.1 {Tcl_UtfToUpper} {
     string toupper {}
@@ -250,6 +266,12 @@ 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-11.6 {Tcl_UtfToUpper low/high surrogate)} {
+    string toupper \udc24\ud824
+} \udc24\ud824
 
 test utf-12.1 {Tcl_UtfToLower} {
     string tolower {}
@@ -263,6 +285,12 @@ 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-12.6 {Tcl_UtfToUpper low/high surrogate)} {
+    string tolower \udc24\ud824
+} \udc24\ud824
 
 test utf-13.1 {Tcl_UtfToTitle} {
     string totitle {}
@@ -276,6 +304,15 @@ 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-13.7 {Tcl_UtfToTitle low/high surrogate)} {
+    string totitle \udc24\ud824
+} \udc24\ud824
 
 test utf-14.1 {Tcl_UtfNcasecmp} {
     string compare -nocase a b