[Minor] Store the fact that we have utf8 only regexps in corpus

src/libserver/re_cache.c

@@ -78,11 +78,14 @@ static const guchar rspamd_hs_magic[] = {'r', 's', 'h', 's', 'r', 'e', '1', '1'}
 struct rspamd_re_class {
 	guint64 id;
 	enum rspamd_re_type type;
+	gboolean has_utf8; /* if there are any utf8 regexps */
 	gpointer type_data;
 	gsize type_len;
 	GHashTable *re;
-	gchar hash[rspamd_cryptobox_HASHBYTES + 1];
 	rspamd_cryptobox_hash_state_t *st;
+
+	gchar hash[rspamd_cryptobox_HASHBYTES + 1];
+
 #ifdef WITH_HYPERSCAN
 	hs_database_t *hs_db;
 	hs_scratch_t *hs_scratch;
@@ -298,6 +301,10 @@ rspamd_re_cache_add (struct rspamd_re_cache *cache, rspamd_regexp_t *re,
 		g_hash_table_insert (re_class->re, rspamd_regexp_get_id (nre), nre);
 	}
 
+	if (rspamd_regexp_get_flags (re) & RSPAMD_REGEXP_FLAG_UTF) {
+		re_class->has_utf8 = TRUE;
+	}
+
 	return nre;
 }
 

src/libutil/regexp.c

@@ -379,6 +379,7 @@ rspamd_regexp_new (const gchar *pattern, const gchar *flags,
 				break;
 			case 'u':
 				rspamd_flags &= ~RSPAMD_REGEXP_FLAG_RAW;
+				rspamd_flags |= RSPAMD_REGEXP_FLAG_UTF;
 #ifndef WITH_PCRE2
 				regexp_flags |= PCRE_FLAG(UTF8);
 #else
@@ -392,6 +393,7 @@ rspamd_regexp_new (const gchar *pattern, const gchar *flags,
 				break;
 			case 'r':
 				rspamd_flags |= RSPAMD_REGEXP_FLAG_RAW;
+				rspamd_flags &= ~RSPAMD_REGEXP_FLAG_UTF;
 #ifndef WITH_PCRE2
 				regexp_flags &= ~PCRE_FLAG(UTF8);
 #else

src/libutil/regexp.h

@@ -33,6 +33,7 @@
 #define RSPAMD_REGEXP_FLAG_FULL_MATCH (1 << 3)
 #define RSPAMD_REGEXP_FLAG_PCRE_ONLY (1 << 4)
 #define RSPAMD_REGEXP_FLAG_DISABLE_JIT (1 << 5)
+#define RSPAMD_REGEXP_FLAG_UTF (1 << 6)
 
 struct rspamd_config;