Improve some cases of IDN urls.

diff --git a/src/libserver/url.c b/src/libserver/url.c
index d52138f8fe1f2f3bdef7a2d99b4330099c5b5e30..af52d5e527090b9eedfcf1aba709d57bbfc2482a 100644 (file)
--- a/src/libserver/url.c
+++ b/src/libserver/url.c
@@ -1037,6 +1037,7 @@ rspamd_web_parse (struct http_parser_url *u, const gchar *str)
                                }
                                else {
                                        st = parse_port;
+                                       c = p + 1;
                                }
                                p ++;
                        }
@@ -1497,21 +1498,20 @@ url_web_end (const gchar *begin,
 
        /* find the end of the domain */
        if (is_atom (*p)) {
-               /* might be a domain or user@domain */
                c = p;
                while (p < end) {
-                       if (!is_atom (*p)) {
+                       if (!is_atom (*p) && !(*p & 0x80)) {
                                break;
                        }
 
                        p++;
 
-                       while (p < end && is_atom (*p)) {
+                       while (p < end && (is_atom (*p) || (*p & 0x80))) {
                                p++;
                        }
 
                        if ((p + 1) < end && *p == '.' &&
-                               (is_atom (*(p + 1)) || *(p + 1) == '/')) {
+                               (is_atom (*(p + 1)) || *(p + 1) == '/' || (*(p + 1) & 0x80))) {
                                p++;
                        }
                }
@@ -1621,10 +1621,6 @@ passwd:
                }
        }
 
-       /* urls are extremely unlikely to end with any
-        * punctuation, so strip any trailing
-        * punctuation off. Also strip off any closing
-        * double-quotes. */
        while (p > pos && strchr (",.:;?!-|}])\"", p[-1])) {
                p--;
        }