Bring forward a collection of missing tests.

2 files changed, 148 insertions, 4 deletions

diff --git a/generic/tclTest.c b/generic/tclTest.c
index 1676bae..322d6b4 100644
--- a/generic/tclTest.c
+++ b/generic/tclTest.c
@@ -6713,7 +6713,8 @@ TestUtfNextCmd(
     int objc,
     Tcl_Obj *const objv[])
 {
-    int numBytes;
+    int numBytes;	/* Number of bytes supplied in the test string */
+    int offset;		/* Number of bytes we are permitted to read */
     char *bytes;
     const char *result, *first;
     char buffer[32];
@@ -6721,20 +6722,46 @@ TestUtfNextCmd(
     const char *p = tobetested;
     (void)dummy;
 
-    if (objc != 2) {
-	Tcl_WrongNumArgs(interp, 1, objv, "bytes");
+    if (objc < 2 || objc > 3) {
+	Tcl_WrongNumArgs(interp, 1, objv, "bytes ?numBytes?");
 	return TCL_ERROR;
     }
+
     bytes = (char *) Tcl_GetByteArrayFromObj(objv[1], &numBytes);
 
+    offset = numBytes +TCL_UTF_MAX -1;	/* If no constraint is given, allow
+					 * the terminating NUL to limit
+					 * operations. */
+
+    if (objc == 3) {
+	if (TCL_OK != TclGetIntForIndex(interp, objv[2], numBytes, &offset)) {
+	    return TCL_ERROR;
+	}
+	if (offset < 0) {
+	    offset = 0;
+	}
+	if (offset > numBytes +TCL_UTF_MAX -1) {
+	    offset = numBytes +TCL_UTF_MAX -1;
+	}
+    }
+
     if (numBytes > (int)sizeof(buffer)-2) {
-	Tcl_AppendResult(interp, "\"testutfnext\" can only handle 30 bytes", NULL);
+	Tcl_SetObjResult(interp, Tcl_ObjPrintf(
+		"\"testutfnext\" can only handle %d bytes",
+		(int)(sizeof(buffer) - 2)));
 	return TCL_ERROR;
     }
 
     memcpy(buffer + 1, bytes, numBytes);
     buffer[0] = buffer[numBytes + 1] = '\x00';
 
+    if (!Tcl_UtfCharComplete(buffer + 1, offset)) {
+	/* Cannot scan a complete sequence from the data */
+
+	Tcl_SetObjResult(interp, Tcl_NewIntObj(0));
+	return TCL_OK;
+    }
+
     first = TclUtfNext(buffer + 1);
     while ((buffer[0] = *p++) != '\0') {
 	/* Run Tcl_UtfNext with many more possible bytes at src[-1], all should give the same result */
diff --git a/tests/utf.test b/tests/utf.test
index 63ae9ee..c739bb4 100644
--- a/tests/utf.test
+++ b/tests/utf.test
@@ -470,6 +470,123 @@ test utf-6.92 {Tcl_UtfNext, pointing to 2th byte of 4-byte valid sequence} testu
 test utf-6.93 {Tcl_UtfNext, pointing to 2th byte of 4-byte invalid sequence} testutfnext {
     testutfnext \x80\x80\x80
 } 1
+test utf-6.94 {Tcl_UtfNext, read limits} testutfnext {
+    testutfnext G 0
+} 0
+test utf-6.95 {Tcl_UtfNext, read limits} testutfnext {
+    testutfnext \xA0 0
+} 0
+test utf-6.96 {Tcl_UtfNext, read limits} testutfnext {
+    testutfnext AG 1
+} 1
+test utf-6.97 {Tcl_UtfNext, read limits} testutfnext {
+    testutfnext A\xA0 1
+} 1
+test utf-6.98 {Tcl_UtfNext, read limits} testutfnext {
+    testutfnext \xD0\xA0G 1
+} 0
+test utf-6.99 {Tcl_UtfNext, read limits} testutfnext {
+    testutfnext \xD0\xA0G 2
+} 2
+test utf-6.100 {Tcl_UtfNext, read limits} testutfnext {
+    testutfnext \xD0\xA0\xA0 1
+} 0
+test utf-6.101 {Tcl_UtfNext, read limits} testutfnext {
+    testutfnext \xD0\xA0\xA0 2
+} 2
+test utf-6.102 {Tcl_UtfNext, read limits} testutfnext {
+    testutfnext \xE8\xA0\xA0G 1
+} 0
+test utf-6.103 {Tcl_UtfNext, read limits} testutfnext {
+    testutfnext \xE8\xA0\xA0G 2
+} 0
+test utf-6.104 {Tcl_UtfNext, read limits} testutfnext {
+    testutfnext \xE8\xA0\xA0G 3
+} 3
+test utf-6.105 {Tcl_UtfNext, read limits} testutfnext {
+    testutfnext \xE8\xA0\xA0\xA0 1
+} 0
+test utf-6.106 {Tcl_UtfNext, read limits} testutfnext {
+    testutfnext \xE8\xA0\xA0\xA0 2
+} 0
+test utf-6.107 {Tcl_UtfNext, read limits} testutfnext {
+    testutfnext \xE8\xA0\xA0\xA0 3
+} 3
+test utf-6.108.0 {Tcl_UtfNext, read limits} {testutfnext ucs2} {
+    testutfnext \xF2\xA0\xA0\xA0G 1
+} 1
+test utf-6.108.1 {Tcl_UtfNext, read limits} {testutfnext fullutf} {
+    testutfnext \xF2\xA0\xA0\xA0G 1
+} 0
+test utf-6.109.0 {Tcl_UtfNext, read limits} {testutfnext ucs2} {
+    testutfnext \xF2\xA0\xA0\xA0G 2
+} 1
+test utf-6.109.1 {Tcl_UtfNext, read limits} {testutfnext fullutf} {
+    testutfnext \xF2\xA0\xA0\xA0G 2
+} 0
+test utf-6.110.0 {Tcl_UtfNext, read limits} {testutfnext ucs2} {
+    testutfnext \xF2\xA0\xA0\xA0G 3
+} 1
+test utf-6.110.1 {Tcl_UtfNext, read limits} {testutfnext fullutf} {
+    testutfnext \xF2\xA0\xA0\xA0G 3
+} 0
+test utf-6.111.0 {Tcl_UtfNext, read limits} {testutfnext ucs2} {
+    testutfnext \xF2\xA0\xA0\xA0G 4
+} 1
+test utf-6.111.1 {Tcl_UtfNext, read limits} {testutfnext fullutf} {
+    testutfnext \xF2\xA0\xA0\xA0G 4
+} 4
+test utf-6.112.0 {Tcl_UtfNext, read limits} {testutfnext ucs2} {
+    testutfnext \xF2\xA0\xA0\xA0\xA0 1
+} 1
+test utf-6.112.1 {Tcl_UtfNext, read limits} {testutfnext fullutf} {
+    testutfnext \xF2\xA0\xA0\xA0\xA0 1
+} 0
+test utf-6.113.0 {Tcl_UtfNext, read limits} {testutfnext ucs2} {
+    testutfnext \xF2\xA0\xA0\xA0\xA0 2
+} 1
+test utf-6.113.1 {Tcl_UtfNext, read limits} {testutfnext fullutf} {
+    testutfnext \xF2\xA0\xA0\xA0\xA0 2
+} 0
+test utf-6.114.0 {Tcl_UtfNext, read limits} {testutfnext ucs2} {
+    testutfnext \xF2\xA0\xA0\xA0\xA0 3
+} 1
+test utf-6.114.1 {Tcl_UtfNext, read limits} {testutfnext fullutf} {
+    testutfnext \xF2\xA0\xA0\xA0\xA0 3
+} 0
+test utf-6.115.0 {Tcl_UtfNext, read limits} {testutfnext ucs2} {
+    testutfnext \xF2\xA0\xA0\xA0\xA0 4
+} 1
+test utf-6.115.1 {Tcl_UtfNext, read limits} {testutfnext fullutf} {
+    testutfnext \xF2\xA0\xA0\xA0\xA0 4
+} 4
+test utf-6.116 {Tcl_UtfNext, read limits} testutfnext {
+    testutfnext \xA0G 0
+} 0
+test utf-6.117 {Tcl_UtfNext, read limits} testutfnext {
+    testutfnext \xA0G 1
+} 1
+test utf-6.118 {Tcl_UtfNext, read limits} testutfnext {
+    testutfnext \xA0\xA0 1
+} 1
+test utf-6.119 {Tcl_UtfNext, read limits} testutfnext {
+    testutfnext \xA0\xA0G 2
+} 1
+test utf-6.120 {Tcl_UtfNext, read limits} testutfnext {
+    testutfnext \xA0\xA0\xA0 2
+} 1
+test utf-6.121 {Tcl_UtfNext, read limits} testutfnext {
+    testutfnext \xA0\xA0\xA0G 3
+} 1
+test utf-6.122 {Tcl_UtfNext, read limits} testutfnext {
+    testutfnext \xA0\xA0\xA0\xA0 3
+} 1
+test utf-6.123 {Tcl_UtfNext, read limits} testutfnext {
+    testutfnext \xA0\xA0\xA0\xA0G 4
+} 1
+test utf-6.124 {Tcl_UtfNext, read limits} testutfnext {
+    testutfnext \xA0\xA0\xA0\xA0\xA0 4
+} 1
 test utf-6.125 {Tcl_UtfNext, pointing to 2th byte of 5-byte invalid sequence} testutfnext {
     testutfnext \xA0\xA0\xA0\xA0
 } 1