[Fix] Fix emoji joiner FP

Issue: #4290

diff --git a/src/libutil/cxx/utf8_util.cxx b/src/libutil/cxx/utf8_util.cxx
index 8c727e9ad6a3f7f1f50d6b3500843b2017c62c25..0e7cd39d7d53a5dfd730b5e43c223617bff279d0 100644 (file)
--- a/src/libutil/cxx/utf8_util.cxx
+++ b/src/libutil/cxx/utf8_util.cxx
@@ -85,8 +85,10 @@ rspamd_normalise_unicode_inplace(char *start, size_t *len)
        if (!zw_spaces.isFrozen()) {
                /* Add zw spaces to the set */
                zw_spaces.add(0x200B);
+               /* TODO: ZW non joiner, it might be used for ligatures, so it should possibly be excluded as well */
                zw_spaces.add(0x200C);
-               zw_spaces.add(0x200D);
+               /* See github issue #4290 for explanation. It seems that the ZWJ has many legit use cases */
+               //zw_spaces.add(0x200D);
                zw_spaces.add(0xFEF);
                zw_spaces.add(0x00AD);
                zw_spaces.freeze();