Make dialog parser and keyword lines work with 4 byte unicode

novelwriter/common.py

@@ -493,6 +493,21 @@ def decodeMimeHandles(mimeData: QMimeData) -> list[str]:
     return mimeData.data(nwConst.MIME_HANDLE).data().decode().split("|")
 
 
+def utf16CharMap(text: str) -> list[int]:
+    """Compute mapping from Python string index to QString index.
+    Python strings are always one character per position in either
+    ASCII, UCS-2 or UCS-4. QStrings are in UTF-16, so wide characters
+    use 2 indices, and thus creates an offset.
+    """
+    posMap = list(range(0, len(text) + 1))
+    offset = 0
+    for i, c in enumerate(text):
+        if ord(c) > 0xffff:
+            offset += 1
+        posMap[i + 1] = i + 1 + offset
+    return posMap
+
+
 ##
 #  Encoder Functions
 ##