@@ -1,3 +1,8 @@

+Wed Feb 18 22:58:14 EET 2009 (edwin)

+------------------------------------

+ * libclamav/phishcheck.c, unit_tests/check_regex.c: improve URL

+ handling

+

 Wed Feb 18 22:24:22 EET 2009 (edwin)

 ------------------------------------

  * libclamav/: reorder fields (bb #1144)

@@ -1002,9 +1002,10 @@ static inline int validate_uri_ialpha(const char *start, const char *end)

 /*

  * Only those URLs are identified as URLs for which phishing detection can be performed.

  */

-static int isURL(const char* URL, int accept_anyproto)

+static int isURL(char* URL, int accept_anyproto)

 {

-	const char *start = NULL, *p, *q, *end;

+	char *last_tld_end = NULL, *q;

+	const char *start = NULL, *p, *end;

 	if(!URL)

 		return 0;

@@ -1055,6 +1056,8 @@ static int isURL(const char* URL, int accept_anyproto)

 		if(q) {

 			if(!validate_uri_xpalphas_nodot(p, q))

 				return 0;

+			if (accept_anyproto && in_tld_set(p, q-p))

+			    last_tld_end = q;

 			p = q+1;

 		}

 	} while(q);

@@ -1063,7 +1066,16 @@ static int isURL(const char* URL, int accept_anyproto)

 	if (end < p)

 		end = p;

 	while (*end == ' ' && end > p) --end;

-	return !!in_tld_set(p, end - p);

+

+	if (in_tld_set(p, end - p))

+	    return 1;

+	if (!accept_anyproto)

+	    return 0;

+	if (last_tld_end) {

+	    *last_tld_end = '\0';

+	    return 1;

+	}

+	return 0;

 }

 /*

@@ -1132,6 +1144,7 @@ static int url_get_host(struct url_check* url,struct url_check* host_url,int isR

 	if(!isReal) {

 		url->pre_fixup.host_start = start - URL;

 		url->pre_fixup.host_end = end - URL;

+		url->pre_fixup.pre_displayLink.data[url->pre_fixup.host_end] = '\0';

 	}

 	return CL_PHISH_NODECISION;

 }

@@ -226,6 +226,7 @@ static const struct rtest {

 		"http://pages.ebay.de@fake.example.com","pages.ebay.de",0},

 	{NULL,"http://key.com","https://key.com",0},

 	{NULL,"http://key.com%00fake.example.com","https://key.com",0},

+	{NULL,"http://key.com.example.com","key.com.invalid"}

 };

 #ifdef CHECK_HAVE_LOOPS

@@ -434,7 +435,6 @@ static struct uc {

     {"http://example.com/blah/..","example.com/",""},

     {"http://example.com/blah/../x","example.com/","x"},

     {"http://example.com/./a","example.com/","a"}

-

 };

 START_TEST (test_url_canon)