Added UTF-8 BOM detection

- This prevents things like the following from happening, by warning the
  user if an incompatible file is loaded:
  http://hercules.ws/board/topic/5126-mapname-selft-announcer/?p=33068
- We're erroring out instead of ignoring it, as explained in the source
  comment.
- Special thanks to jaBote.

Signed-off-by: Haru <haru@dotalux.com>

1 files changed, 12 insertions, 0 deletions

diff --git a/src/map/npc.c b/src/map/npc.c
index f77359b5f..3018cceeb 100644
--- a/src/map/npc.c
+++ b/src/map/npc.c
@@ -3951,6 +3951,18 @@ int npc_parsesrcfile(const char* filepath, bool runOnInit) {
 	}
 	fclose(fp);
 
+	if ((unsigned char)buffer[0] == 0xEF && (unsigned char)buffer[1] == 0xBB && (unsigned char)buffer[2] == 0xBF) {
+		// UTF-8 BOM. This is most likely an error on the user's part, because:
+		// - BOM is discouraged in UTF-8, and the only place where you see it is Notepad and such.
+		// - It's unlikely that the user wants to use UTF-8 data here, since we don't really support it, nor does the client by default.
+		// - If the user really wants to use UTF-8 (instead of latin1, EUC-KR, SJIS, etc), then they can still do it <without BOM>.
+		// More info at http://unicode.org/faq/utf_bom.html#bom5 and http://en.wikipedia.org/wiki/Byte_order_mark#UTF-8
+		ShowError("npc_parsesrcfile: Detected unsupported UTF-8 BOM in file '%s'. Stopping (please consider using another character set.)\n", filepath);
+		aFree(buffer);
+		fclose(fp);
+		return -1;
+	}
+
 	// parse buffer
 	for( p = script->skip_space(buffer); p && *p ; p = script->skip_space(p) )
 	{