ply-utils: Make ply_utf8_character_get_byte_type detect invalid bytes

Somehow I neglected to make ply_utf8_character_get_byte_type
differentiate between valid and invalid utf-8.

This commit fixes that.

diff --git a/src/libply/ply-utils.c b/src/libply/ply-utils.c
index 95b505b170f6ee40e60c2bbf6cafccd238a8c99f..463f61c927205b5c779c73220323bd71a4258189 100644 (file)
--- a/src/libply/ply-utils.c
+++ b/src/libply/ply-utils.c
@@ -757,8 +757,10 @@ ply_utf8_character_get_byte_type (const char byte)
                 byte_type = PLY_UTF8_CHARACTER_BYTE_TYPE_3_BYTES;
         else if ((byte & 0xF8) == 0xF0)
                 byte_type = PLY_UTF8_CHARACTER_BYTE_TYPE_4_BYTES;
-        else
+        else if ((byte & 0xC0) == 0x80)
                 byte_type = PLY_UTF8_CHARACTER_BYTE_TYPE_CONTINUATION;
+        else
+                byte_type = PLY_UTF8_CHARACTER_BYTE_TYPE_INVALID;
 
         return byte_type;
 }