[67aa9a2070] Tcl_UtfToUniChar returns single byte for invalid UTF-8 input as documented.

2 files changed, 34 insertions, 3 deletions

diff --git a/generic/tclUtf.c b/generic/tclUtf.c
index b33bf6a..eec4068 100644
--- a/generic/tclUtf.c
+++ b/generic/tclUtf.c
@@ -292,7 +292,9 @@ Tcl_UtfToUniChar(
 	     */
 
 	    *chPtr = (Tcl_UniChar) (((byte & 0x1F) << 6) | (src[1] & 0x3F));
-	    return 2;
+	    if ((*chPtr == 0) || (*chPtr > 0x7f)) {
+		return 2;
+	    }
 	}
 
 	/*
@@ -307,7 +309,9 @@ Tcl_UtfToUniChar(
 
 	    *chPtr = (Tcl_UniChar) (((byte & 0x0F) << 12)
 		    | ((src[1] & 0x3F) << 6) | (src[2] & 0x3F));
-	    return 3;
+	    if (*chPtr > 0x7ff) {
+		return 3;
+	    }
 	}
 
 	/*
@@ -324,7 +328,9 @@ Tcl_UtfToUniChar(
 
 	    *chPtr = (Tcl_UniChar) (((byte & 0x0E) << 18) | ((src[1] & 0x3F) << 12)
 		    | ((src[2] & 0x3F) << 6) | (src[3] & 0x3F));
-	    return 4;
+	    if ((*chPtr <= 0x10ffff) && (*chPtr > 0xffff)) {
+		return 4;
+	    }
 	}
 
 	/*
diff --git a/tests/encoding.test b/tests/encoding.test
index d9ba072..eb2a08f 100644
--- a/tests/encoding.test
+++ b/tests/encoding.test
@@ -448,6 +448,31 @@ test encoding-24.3 {EscapeFreeProc on open channels} {stdio} {
     list $count [viewable $line]
 } [list 3 "\u4e4e\u4e5e\u4e5f (\\u4e4e\\u4e5e\\u4e5f)"]
 
+test encoding-24.4 {Parse valid or invalid utf-8} {
+    string length [encoding convertfrom utf-8 "\xc0\x80"]
+} 1
+test encoding-24.5 {Parse valid or invalid utf-8} {
+    string length [encoding convertfrom utf-8 "\xc0\x81"]
+} 2
+test encoding-24.6 {Parse valid or invalid utf-8} {
+    string length [encoding convertfrom utf-8 "\xc1\xbf"]
+} 2
+test encoding-24.7 {Parse valid or invalid utf-8} {
+    string length [encoding convertfrom utf-8 "\xc2\x80"]
+} 1
+test encoding-24.8 {Parse valid or invalid utf-8} {
+    string length [encoding convertfrom utf-8 "\xe0\x80\x80"]
+} 3
+test encoding-24.9 {Parse valid or invalid utf-8} {
+    string length [encoding convertfrom utf-8 "\xe0\x9f\xbf"]
+} 3
+test encoding-24.10 {Parse valid or invalid utf-8} {
+    string length [encoding convertfrom utf-8 "\xe0\xa0\x80"]
+} 1
+test encoding-24.10 {Parse valid or invalid utf-8} {
+    string length [encoding convertfrom utf-8 "\xef\xbf\xbf"]
+} 1
+
 file delete [file join [temporaryDirectory] iso2022.txt]
 
 #