[Fix] Deal with 8bit characters in email addresses

6 files changed, 64 insertions, 37 deletions

diff --git a/src/libmime/email_addr.c b/src/libmime/email_addr.c
index c857794c8..59e99f531 100644
--- a/src/libmime/email_addr.c
+++ b/src/libmime/email_addr.c
@@ -169,7 +169,8 @@ static gboolean
 rspamd_email_address_parse_heuristic (const char *data, size_t len,
 		struct rspamd_email_address *addr)
 {
-	const gchar *p = data;
+	const gchar *p = data, *at = NULL, *end = data + len;
+	gboolean ret = FALSE;
 
 	memset (addr, 0, sizeof (*addr));
 
@@ -179,19 +180,36 @@ rspamd_email_address_parse_heuristic (const char *data, size_t len,
 		addr->addr = p + 1;
 		addr->raw = p;
 		addr->raw_len = len;
+		ret = TRUE;
 
-		return TRUE;
+		p = p + 1;
+		len = addr->addr_len;
+		end = p + len;
 	}
 	else if (len > 0) {
 		addr->addr = p;
 		addr->addr_len = len;
 		addr->raw = p;
 		addr->raw_len = len;
+		ret = TRUE;
+	}
+
+	if (ret) {
+		at = memchr (p, '@', len);
+
+		if (at != NULL && at + 1 < end) {
+			addr->domain = at + 1;
+			addr->domain_len = end - (at + 1);
+			addr->user = p;
+			addr->user_len = at - p;
+		}
 
-		return TRUE;
+		if (rspamd_str_has_8bit (p, len)) {
+			addr->flags |= RSPAMD_EMAIL_ADDR_HAS_8BIT;
+		}
 	}
 
-	return FALSE;
+	return ret;
 }
 
 GPtrArray *
@@ -329,6 +347,9 @@ rspamd_email_address_from_mime (rspamd_mempool_t *pool,
 				state = skip_spaces;
 				next_state = parse_name;
 			}
+			else if (*p == '@') {
+				seen_at = TRUE;
+			}
 			p ++;
 			break;
 		case skip_spaces:
diff --git a/src/libmime/email_addr.h b/src/libmime/email_addr.h
index f6a16d682..b4f192ee7 100644
--- a/src/libmime/email_addr.h
+++ b/src/libmime/email_addr.h
@@ -32,6 +32,7 @@ enum rspamd_email_address_flags {
 	RSPAMD_EMAIL_ADDR_HAS_BACKSLASH = (1 << 6),
 	RSPAMD_EMAIL_ADDR_ADDR_ALLOCATED = (1 << 7),
 	RSPAMD_EMAIL_ADDR_USER_ALLOCATED = (1 << 8),
+	RSPAMD_EMAIL_ADDR_HAS_8BIT = (1 << 9),
 };
 
 /*
diff --git a/src/libmime/mime_encoding.c b/src/libmime/mime_encoding.c
index 11f764e11..97eede726 100644
--- a/src/libmime/mime_encoding.c
+++ b/src/libmime/mime_encoding.c
@@ -455,36 +455,6 @@ rspamd_mime_charset_utf_check (rspamd_ftok_t *charset,
 	return FALSE;
 }
 
-/* https://graphics.stanford.edu/~seander/bithacks.html#HasMoreInWord */
-#define hasmore(x,n) (((x)+~0UL/255*(127-(n))|(x))&~0UL/255*128)
-
-static inline gboolean
-rspamd_mime_has_8bit (const guchar *beg, gsize len)
-{
-	unsigned long *w;
-	gsize i, leftover = len % sizeof (*w);
-
-	w = (unsigned long *)beg;
-
-	for (i = 0; i < len / sizeof (*w); i ++) {
-		if (hasmore (*w, 127)) {
-			return TRUE;
-		}
-
-		w ++;
-	}
-
-	beg = (const guchar *)w;
-
-	for (i = 0; i < leftover; i ++) {
-		if (beg[i] > 127) {
-			return TRUE;
-		}
-	}
-
-	return FALSE;
-}
-
 GByteArray *
 rspamd_mime_text_part_maybe_convert (struct rspamd_task *task,
 		struct rspamd_mime_text_part *text_part)
@@ -498,7 +468,7 @@ rspamd_mime_text_part_maybe_convert (struct rspamd_task *task,
 	rspamd_ftok_t charset_tok;
 	struct rspamd_mime_part *part = text_part->mime_part;
 
-	if (rspamd_mime_has_8bit (text_part->raw.begin, text_part->raw.len)) {
+	if (rspamd_str_has_8bit (text_part->raw.begin, text_part->raw.len)) {
 		text_part->flags |= RSPAMD_MIME_TEXT_PART_FLAG_8BIT;
 	}
 
@@ -508,7 +478,7 @@ rspamd_mime_text_part_maybe_convert (struct rspamd_task *task,
 	memcpy (part_content->data, text_part->parsed.begin, text_part->parsed.len);
 	part_content->len = text_part->parsed.len;
 
-	if (rspamd_mime_has_8bit (text_part->parsed.begin, text_part->parsed.len)) {
+	if (rspamd_str_has_8bit (text_part->parsed.begin, text_part->parsed.len)) {
 		text_part->flags |= RSPAMD_MIME_TEXT_PART_FLAG_8BIT_ENCODED;
 	}
 
diff --git a/src/libutil/str_util.h b/src/libutil/str_util.h
index 2fec42987..473b5cbbb 100644
--- a/src/libutil/str_util.h
+++ b/src/libutil/str_util.h
@@ -339,4 +339,35 @@ const void *rspamd_memrchr (const void *m, gint c, gsize len);
  */
 gsize rspamd_memcspn (const gchar *s, const gchar *e, gsize len);
 
+
+/* https://graphics.stanford.edu/~seander/bithacks.html#HasMoreInWord */
+#define rspamd_str_hasmore(x,n) ((((x)+~0UL/255*(127-(n)))|(x))&~0UL/255*128)
+
+static inline gboolean
+rspamd_str_has_8bit (const guchar *beg, gsize len)
+{
+	unsigned long *w;
+	gsize i, leftover = len % sizeof (*w);
+
+	w = (unsigned long *)beg;
+
+	for (i = 0; i < len / sizeof (*w); i ++) {
+		if (rspamd_str_hasmore (*w, 127)) {
+			return TRUE;
+		}
+
+		w ++;
+	}
+
+	beg = (const guchar *)w;
+
+	for (i = 0; i < leftover; i ++) {
+		if (beg[i] > 127) {
+			return TRUE;
+		}
+	}
+
+	return FALSE;
+}
+
 #endif /* SRC_LIBUTIL_STR_UTIL_H_ */
diff --git a/src/libutil/util.h b/src/libutil/util.h
index 48381ed92..605822fee 100644
--- a/src/libutil/util.h
+++ b/src/libutil/util.h
@@ -517,4 +517,6 @@ gdouble rspamd_normalize_probability (gdouble x, gdouble bias);
 guint64 rspamd_tm_to_time (const struct tm *tm, glong tz);
 
 #define PTR_ARRAY_FOREACH(ar, i, cur) for ((i) = 0; (ar) != NULL && (i) < (ar)->len && (((cur) = g_ptr_array_index((ar), (i))) || 1); ++(i))
+
+
 #endif
diff --git a/src/ragel/smtp_addr_parser.rl b/src/ragel/smtp_addr_parser.rl
index 7e8498966..501ee82f4 100644
--- a/src/ragel/smtp_addr_parser.rl
+++ b/src/ragel/smtp_addr_parser.rl
@@ -54,7 +54,9 @@
   }
 
   action Valid_addr {
-    addr->flags |= RSPAMD_EMAIL_ADDR_VALID;
+    if (addr->addr_len > 0) {
+      addr->flags |= RSPAMD_EMAIL_ADDR_VALID;
+    }
   }
 
   action Addr_has_angle {