From: Vsevolod Stakhov <vsevolod@highsecure.ru>
Date: Thu, 19 Feb 2015 16:34:03 +0000 (+0000)
Subject: Improve some cases of IDN urls.
X-Git-Tag: 0.8.2~33
X-Git-Url: http://git.ipfire.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=47c75ed18d3e28f39619ba7c919890d48940a9e4;p=thirdparty%2Frspamd.git

Improve some cases of IDN urls.
---

diff --git a/src/libserver/url.c b/src/libserver/url.c
index d52138f8fe..af52d5e527 100644
--- 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--;
 	}