[Bug 3466099] BOM in Unicode

2 files changed, 36 insertions, 4 deletions

diff --git a/generic/tclEncoding.c b/generic/tclEncoding.c
index 15411d8..49418c9 100644
--- a/generic/tclEncoding.c
+++ b/generic/tclEncoding.c
@@ -979,13 +979,13 @@ Tcl_GetEncodingNames(
 int
 Tcl_SetSystemEncoding(
     Tcl_Interp *interp,		/* Interp for error reporting, if not NULL. */
-    const char *name)		/* The name of the desired encoding, or NULL
+    const char *name)		/* The name of the desired encoding, or NULL/""
 				 * to reset to default encoding. */
 {
     Tcl_Encoding encoding;
     Encoding *encodingPtr;
 
-    if (name == NULL) {
+    if (!name || !*name) {
 	Tcl_MutexLock(&encodingMutex);
 	encoding = defaultEncoding;
 	encodingPtr = (Encoding *) encoding;
diff --git a/generic/tclIOUtil.c b/generic/tclIOUtil.c
index fa616b3..62553f2 100644
--- a/generic/tclIOUtil.c
+++ b/generic/tclIOUtil.c
@@ -1729,7 +1729,22 @@ Tcl_FSEvalFileEx(
 
     objPtr = Tcl_NewObj();
     Tcl_IncrRefCount(objPtr);
-    if (Tcl_ReadChars(chan, objPtr, -1, 0) < 0) {
+    /* Try to read first character of stream, so we can
+     * check for utf-8 BOM to be handled especially.
+     */
+    if (Tcl_ReadChars(chan, objPtr, 1, 0) < 0) {
+	Tcl_Close(interp, chan);
+	Tcl_AppendResult(interp, "couldn't read file \"",
+		Tcl_GetString(pathPtr), "\": ", Tcl_PosixError(interp), NULL);
+	goto end;
+    }
+    string = Tcl_GetString(objPtr);
+    /*
+     * If first character is not a BOM, append the remaining characters,
+     * otherwise replace them [Bug 3466099].
+     */
+    if (Tcl_ReadChars(chan, objPtr, -1,
+	    memcmp(string, "\xef\xbf\xbe", 3)) < 0) {
 	Tcl_Close(interp, chan);
 	Tcl_AppendResult(interp, "couldn't read file \"",
 		Tcl_GetString(pathPtr), "\": ", Tcl_PosixError(interp), NULL);
@@ -1798,6 +1813,7 @@ TclNREvalFile(
     Tcl_Obj *oldScriptFile, *objPtr;
     Interp *iPtr;
     Tcl_Channel chan;
+    const char *string;
 
     if (Tcl_FSGetNormalizedPath(interp, pathPtr) == NULL) {
 	return TCL_ERROR;
@@ -1839,13 +1855,29 @@ TclNREvalFile(
 
     objPtr = Tcl_NewObj();
     Tcl_IncrRefCount(objPtr);
-    if (Tcl_ReadChars(chan, objPtr, -1, 0) < 0) {
+    /* Try to read first character of stream, so we can
+     * check for utf-8 BOM to be handled especially.
+     */
+    if (Tcl_ReadChars(chan, objPtr, 1, 0) < 0) {
 	Tcl_Close(interp, chan);
 	Tcl_AppendResult(interp, "couldn't read file \"",
 		Tcl_GetString(pathPtr), "\": ", Tcl_PosixError(interp), NULL);
 	Tcl_DecrRefCount(objPtr);
 	return TCL_ERROR;
     }
+    string = Tcl_GetString(objPtr);
+    /*
+     * If first character is not a BOM, append the remaining characters,
+     * otherwise replace them [Bug 3466099].
+     */
+    if (Tcl_ReadChars(chan, objPtr, -1,
+	    memcmp(string, "\xef\xbf\xbe", 3)) < 0) {
+	Tcl_Close(interp, chan);
+	Tcl_AppendResult(interp, "couldn't read file \"",
+			Tcl_GetString(pathPtr), "\": ", Tcl_PosixError(interp), NULL);
+	Tcl_DecrRefCount(objPtr);
+	return TCL_ERROR;
+    }
 
     if (Tcl_Close(interp, chan) != TCL_OK) {
 	Tcl_DecrRefCount(objPtr);