[Fix] Try to fix false positive URL detections in text parts

diff --git a/src/libmime/message.c b/src/libmime/message.c
index dbc9921d92152f2c25d084f110280fc755132eba..791bd6837b739e0461e7c5dd9a3d141f8eb687ef 100644 (file)
--- a/src/libmime/message.c
+++ b/src/libmime/message.c
@@ -1268,7 +1268,6 @@ process_text_part (struct rspamd_task *task,
                                type,
                                text_part);
                text_part->orig = part_content;
-               rspamd_url_text_extract (task->task_pool, task, text_part, FALSE);
                g_ptr_array_add (task->text_parts, text_part);
        }
        else {
@@ -1310,6 +1309,10 @@ process_text_part (struct rspamd_task *task,
                        c = p + 1;
                }
        }
+
+       if (!IS_PART_HTML (text_part)) {
+               rspamd_url_text_extract (task->task_pool, task, text_part, FALSE);
+       }
 }
 
 struct mime_foreach_data {

diff --git a/src/libserver/url.c b/src/libserver/url.c
index fe70585beae6d0cd915b2442408e199d96a70774..70a5f3c9bf4abd22af5f5ed1c7f7026cdf48ce22 100644 (file)
--- a/src/libserver/url.c
+++ b/src/libserver/url.c
@@ -1795,10 +1795,9 @@ url_tld_end (struct url_callback_data *cb,
 {
        const gchar *p;
 
-       /* A url must be finished by tld, so it must be followed by space character */
        p = pos + match->m_len;
 
-       if (p == cb->end || g_ascii_isspace (*p) || *p == ',') {
+       if (p == cb->end) {
                match->m_len = p - match->m_begin;
                return TRUE;
        }
@@ -2302,7 +2301,7 @@ rspamd_url_text_extract (rspamd_mempool_t *pool,
 {
        struct rspamd_url_mimepart_cbdata mcbd;
 
-       if (part->content == NULL || part->content->len == 0) {
+       if (part->stripped_content == NULL || part->stripped_content->len == 0) {
                msg_warn_task ("got empty text part");
                return;
        }
@@ -2310,8 +2309,8 @@ rspamd_url_text_extract (rspamd_mempool_t *pool,
        mcbd.task = task;
        mcbd.part = part;
 
-       rspamd_url_find_multiple (task->task_pool, part->content->data,
-                       part->content->len, is_html,
+       rspamd_url_find_multiple (task->task_pool, part->stripped_content->data,
+                       part->stripped_content->len, is_html,
                        rspamd_url_text_part_callback, &mcbd);
 
        /* Handle offsets of this part */

diff --git a/test/lua/unit/url.lua b/test/lua/unit/url.lua
index 06082afe0036832045819788de0c4d20740339cf..de274425d6e08fa96600b5d04a1c7d54e9b4f454 100644 (file)
--- a/test/lua/unit/url.lua
+++ b/test/lua/unit/url.lua
@@ -17,8 +17,8 @@ context("URL check functions", function()
   test("Extract urls from text", function()
     local pool = mpool.create()
     local cases = {
-      {"test.com text", {"test.com", nil}},
-      {" test.com text", {"test.com", nil}},
+      {"test.com", {"test.com", nil}},
+      {" test.com", {"test.com", nil}},
       {"<test.com> text", {"test.com", nil}},
       {"test.com. text", {"test.com", nil}},
       {"mailto:A.User@example.com text", {"example.com", "A.User"}},