[Fix] Fix urls and emails hashes

4 files changed, 28 insertions, 9 deletions

diff --git a/src/libserver/html.c b/src/libserver/html.c
index 72e8cbd9d..449766a0c 100644
--- a/src/libserver/html.c
+++ b/src/libserver/html.c
@@ -2150,7 +2150,7 @@ rspamd_html_process_part_full (rspamd_mempool_t *pool, struct html_content *hc,
 							}
 
 							if (displayed_url) {
-								if (url->protocol == PROTOCOL_MAILTO) {
+								if (displayed_url->protocol == PROTOCOL_MAILTO) {
 									target_tbl = emails;
 								}
 								else {
@@ -2158,7 +2158,8 @@ rspamd_html_process_part_full (rspamd_mempool_t *pool, struct html_content *hc,
 								}
 
 								if (target_tbl != NULL) {
-									turl = g_hash_table_lookup (target_tbl, url);
+									turl = g_hash_table_lookup (target_tbl,
+											displayed_url);
 
 									if (turl != NULL) {
 										/* Here, we assume the following:
@@ -2172,6 +2173,10 @@ rspamd_html_process_part_full (rspamd_mempool_t *pool, struct html_content *hc,
 											turl->flags &= ~RSPAMD_URL_FLAG_FROM_TEXT;
 										}
 									}
+									else {
+										g_hash_table_insert (target_tbl,
+												displayed_url, displayed_url);
+									}
 								}
 							}
 						}
diff --git a/src/libserver/task.c b/src/libserver/task.c
index 886e1b004..76d9639b3 100644
--- a/src/libserver/task.c
+++ b/src/libserver/task.c
@@ -105,7 +105,7 @@ rspamd_task_new (struct rspamd_worker *worker, struct rspamd_config *cfg)
 	rspamd_mempool_add_destructor (new_task->task_pool,
 		(rspamd_mempool_destruct_t) g_hash_table_unref,
 		new_task->raw_headers);
-	new_task->emails = g_hash_table_new (rspamd_url_hash, rspamd_emails_cmp);
+	new_task->emails = g_hash_table_new (rspamd_email_hash, rspamd_emails_cmp);
 	rspamd_mempool_add_destructor (new_task->task_pool,
 		(rspamd_mempool_destruct_t) g_hash_table_unref,
 		new_task->emails);
diff --git a/src/libserver/url.c b/src/libserver/url.c
index c4c5fd038..420c0f959 100644
--- a/src/libserver/url.c
+++ b/src/libserver/url.c
@@ -2584,7 +2584,24 @@ rspamd_url_hash (gconstpointer u)
 		rspamd_cryptobox_fast_hash_update (&st, url->string, url->urllen);
 	}
 
-	rspamd_cryptobox_fast_hash_update (&st, &url->flags, sizeof (url->flags));
+	return rspamd_cryptobox_fast_hash_final (&st);
+}
+
+guint
+rspamd_email_hash (gconstpointer u)
+{
+	const struct rspamd_url *url = u;
+	rspamd_cryptobox_fast_hash_state_t st;
+
+	rspamd_cryptobox_fast_hash_init (&st, rspamd_hash_seed ());
+
+	if (url->hostlen > 0) {
+		rspamd_cryptobox_fast_hash_update (&st, url->host, url->hostlen);
+	}
+
+	if (url->userlen > 0) {
+		rspamd_cryptobox_fast_hash_update (&st, url->user, url->userlen);
+	}
 
 	return rspamd_cryptobox_fast_hash_final (&st);
 }
@@ -2621,17 +2638,13 @@ gboolean
 rspamd_urls_cmp (gconstpointer a, gconstpointer b)
 {
 	const struct rspamd_url *u1 = a, *u2 = b;
-	int r;
+	int r = 0;
 
 	if (u1->urllen != u2->urllen) {
 		return FALSE;
 	}
 	else {
 		r = memcmp (u1->string, u2->string, u1->urllen);
-		if (r == 0 && u1->flags != u2->flags) {
-			/* Always insert phished urls to the tree */
-			return FALSE;
-		}
 	}
 
 	return r == 0;
diff --git a/src/libserver/url.h b/src/libserver/url.h
index 3fab46c5e..7af0bdfb8 100644
--- a/src/libserver/url.h
+++ b/src/libserver/url.h
@@ -180,6 +180,7 @@ void rspamd_url_add_tag (struct rspamd_url *url, const gchar *tag,
 		rspamd_mempool_t *pool);
 
 guint rspamd_url_hash (gconstpointer u);
+guint rspamd_email_hash (gconstpointer u);
 
 /* Compare two emails for building emails hash */
 gboolean rspamd_emails_cmp (gconstpointer a, gconstpointer b);