[Fix] When checking for phishing, we need to convert punicode -> UTF8, not vice versa

Previosly, Rspamd has checked for `xn--` signature just to convert it back to `xn--` form.
According to the expected results, it should do the opposite conversion.

Issue: #4065

diff --git a/src/libserver/html/html_url.cxx b/src/libserver/html/html_url.cxx
index 34775ba135fc49da840cbd15e010530212e9e1f9..0b48c7a75ea645c229bd62186b3d68268e3b7786 100644 (file)
--- a/src/libserver/html/html_url.cxx
+++ b/src/libserver/html/html_url.cxx
@@ -92,17 +92,20 @@ convert_idna_hostname_maybe(rspamd_mempool_t *pool, struct rspamd_url *url, bool
        /* Handle IDN url's */
        if (ret.size() > 4 &&
                rspamd_substring_search_caseless(ret.data(), ret.size(), "xn--", 4) != -1) {
+
                const auto buf_capacity = ret.size() * 2 + 1;
                auto *idn_hbuf = (char *)rspamd_mempool_alloc (pool, buf_capacity);
                icu::CheckedArrayByteSink byte_sink{idn_hbuf, (int)buf_capacity};
+
                /* We need to convert it to the normal value first */
                icu::IDNAInfo info;
                auto uc_err = U_ZERO_ERROR;
                auto *udn = get_icu_idna_instance();
-               udn->nameToASCII_UTF8(icu::StringPiece(ret.data(), ret.size()),
+               udn->nameToUnicodeUTF8(icu::StringPiece(ret.data(), ret.size()),
                                byte_sink, info, uc_err);
 
                if (uc_err == U_ZERO_ERROR && !info.hasErrors()) {
+                       /* idn_hbuf is allocated in mempool, so it is safe to use */
                        ret = std::string_view{idn_hbuf, (std::size_t)byte_sink.NumberOfBytesWritten()};
                }
                else {