[3613609]: Replace strcasecmp() with UTF-8-aware version.

6 files changed, 62 insertions, 6 deletions

diff --git a/ChangeLog b/ChangeLog
index 3f61f95..bbccf19 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,9 @@
+2013-05-22  Donal K. Fellows  <dkf@users.sf.net>
+
+	* generic/tclUtf.c (TclUtfCasecmp): [Bug 3613609]: Replace problematic
+	uses of strcasecmp with a proper UTF-8-aware version. Affects both
+	[lsearch -nocase] and [lsort -nocase].
+
 2013-05-19  Jan Nijtmans  <nijtmans@users.sf.net>
 
 	* unix/tcl.m4:     Fix for FreeBSD, and remove support for older
diff --git a/generic/tclCmdIL.c b/generic/tclCmdIL.c
index 152e61d..98ec8b4 100644
--- a/generic/tclCmdIL.c
+++ b/generic/tclCmdIL.c
@@ -2807,7 +2807,7 @@ Tcl_LsearchObjCmd(
 	    dataType = INTEGER;
 	    break;
 	case LSEARCH_NOCASE:		/* -nocase */
-	    strCmpFn = strcasecmp;
+	    strCmpFn = TclUtfCasecmp;
 	    noCase = 1;
 	    break;
 	case LSEARCH_NOT:		/* -not */
@@ -3209,7 +3209,7 @@ Tcl_LsearchObjCmd(
 			 */
 
 			if (noCase) {
-			    match = (strcasecmp(bytes, patternBytes) == 0);
+			    match = (TclUtfCasecmp(bytes, patternBytes) == 0);
 			} else {
 			    match = (memcmp(bytes, patternBytes,
 				    (size_t) length) == 0);
@@ -3712,7 +3712,7 @@ Tcl_LsortObjCmd(
 		goto done1;
 	    }
 	    elementArray[i].index.intValue = a;
-	} else if (sortInfo.sortMode == SORTMODE_REAL) {
+	} else if (sortMode == SORTMODE_REAL) {
 	    double a;
 	    if (Tcl_GetDoubleFromObj(sortInfo.interp, indexPtr, &a) != TCL_OK) {
 		sortInfo.resultCode = TCL_ERROR;
@@ -3790,7 +3790,7 @@ Tcl_LsortObjCmd(
     ckfree((char *)elementArray);
 
   done:
-    if (sortInfo.sortMode == SORTMODE_COMMAND) {
+    if (sortMode == SORTMODE_COMMAND) {
 	TclDecrRefCount(sortInfo.compareCmdPtr);
 	TclDecrRefCount(listObj);
 	sortInfo.compareCmdPtr = NULL;
@@ -3932,7 +3932,7 @@ SortCompare(
 	order = strcmp(elemPtr1->index.strValuePtr,
 		elemPtr2->index.strValuePtr);
     } else if (infoPtr->sortMode == SORTMODE_ASCII_NC) {
-	order = strcasecmp(elemPtr1->index.strValuePtr,
+	order = TclUtfCasecmp(elemPtr1->index.strValuePtr,
 		elemPtr2->index.strValuePtr);
     } else if (infoPtr->sortMode == SORTMODE_DICTIONARY) {
 	order = DictionaryCompare(elemPtr1->index.strValuePtr,
diff --git a/generic/tclCmdMZ.c b/generic/tclCmdMZ.c
index 0ad77aa..6fd468c 100644
--- a/generic/tclCmdMZ.c
+++ b/generic/tclCmdMZ.c
@@ -3436,7 +3436,7 @@ Tcl_SwitchObjCmd(
 	    i++;
 	    goto finishedOptions;
 	case OPT_NOCASE:
-	    strCmpFn = strcasecmp;
+	    strCmpFn = TclUtfCasecmp;
 	    noCase = 1;
 	    break;
 
diff --git a/generic/tclInt.h b/generic/tclInt.h
index 92251fe..dc28b97 100644
--- a/generic/tclInt.h
+++ b/generic/tclInt.h
@@ -2762,6 +2762,7 @@ MODULE_SCOPE Tcl_WideInt TclpGetWideClicks(void);
 MODULE_SCOPE double	TclpWideClicksToNanoseconds(Tcl_WideInt clicks);
 #endif
 MODULE_SCOPE Tcl_Obj *	TclDisassembleByteCodeObj(Tcl_Obj *objPtr);
+MODULE_SCOPE int TclUtfCasecmp(CONST char *cs, CONST char *ct);
 
 /*
  *----------------------------------------------------------------
diff --git a/generic/tclUtf.c b/generic/tclUtf.c
index 83900e9..f3d1758 100644
--- a/generic/tclUtf.c
+++ b/generic/tclUtf.c
@@ -1105,6 +1105,46 @@ Tcl_UtfNcasecmp(
 /*
  *----------------------------------------------------------------------
  *
+ * Tcl_UtfNcasecmp --
+ *
+ *	Compare UTF chars of string cs to string ct case insensitively.
+ *	Replacement for strcasecmp in Tcl core, in places where UTF-8 should
+ *	be handled.
+ *
+ * Results:
+ *	Return <0 if cs < ct, 0 if cs == ct, or >0 if cs > ct.
+ *
+ * Side effects:
+ *	None.
+ *
+ *----------------------------------------------------------------------
+ */
+
+int
+TclUtfCasecmp(
+    CONST char *cs,		/* UTF string to compare to ct. */
+    CONST char *ct)		/* UTF string cs is compared to. */
+{
+    while (*cs && *ct) {
+	Tcl_UniChar ch1, ch2;
+
+	cs += TclUtfToUniChar(cs, &ch1);
+	ct += TclUtfToUniChar(ct, &ch2);
+	if (ch1 != ch2) {
+	    ch1 = Tcl_UniCharToLower(ch1);
+	    ch2 = Tcl_UniCharToLower(ch2);
+	    if (ch1 != ch2) {
+		return ch1 - ch2;
+	    }
+	}
+    }
+    return UCHAR(*cs) - UCHAR(*ct);
+}
+
+
+/*
+ *----------------------------------------------------------------------
+ *
  * Tcl_UniCharToUpper --
  *
  *	Compute the uppercase equivalent of the given Unicode character.
diff --git a/tests/cmdIL.test b/tests/cmdIL.test
index b387e71..6fab269 100644
--- a/tests/cmdIL.test
+++ b/tests/cmdIL.test
@@ -394,6 +394,15 @@ test cmdIL-4.34 {SortCompare procedure, -ascii option with -nocase option} {
 test cmdIL-4.35 {SortCompare procedure, -ascii option with -nocase option} {
     lsort -ascii -nocase {d E c B a D35 d300 100 20}
 } {100 20 a B c d d300 D35 E}
+test cmdIL-4.36 {SortCompare procedure, UTF-8 with -nocase option} {
+    scan [lsort -ascii -nocase [list \u101 \u100]] %c%c%c
+} {257 32 256}
+test cmdIL-4.37 {SortCompare procedure, UTF-8 with -nocase option} {
+    scan [lsort -ascii -nocase [list a\u0000a a]] %c%c%c%c%c
+} {97 32 97 0 97}
+test cmdIL-4.38 {SortCompare procedure, UTF-8 with -nocase option} {
+    scan [lsort -ascii -nocase [list a a\u0000a]] %c%c%c%c%c
+} {97 32 97 0 97}
 
 test cmdIL-5.1 {lsort with list style index} {
     lsort -ascii -decreasing -index {0 1} {