fix match with root regex (bb #1421).

git-svn: trunk@4848

ChangeLog

@@ -1,3 +1,9 @@
+Mon Feb 23 14:53:39 EET 2009 (edwin)
+------------------------------------
+ * libclamav/regex_list.c, libclamav/regex_list.h,
+ unit_tests/check_regex.c, unit_tests/input/daily.wdb: fix match with
+ root regex (bb #1421).
+
 Mon Feb 23 14:10:01 EET 2009 (edwin)
 ------------------------------------
  * clamd/session.c: stop scanning when client disconnects in

libclamav/regex_list.c

@@ -272,7 +272,7 @@ int regex_list_match(struct regex_matcher* matcher,char* real_url,const char* di
 	{
 		char *buffer = cli_malloc(buffer_len+1);
 		char *bufrev;
-		int rc = 0;
+		int rc = 0, root;
 		struct cli_ac_data mdata;
 		struct cli_ac_result *res = NULL;
 
@@ -309,9 +309,15 @@ int regex_list_match(struct regex_matcher* matcher,char* real_url,const char* di
 		cli_ac_freedata(&mdata);
 
 		rc = 0;
-		while(res) {
+		root = matcher->root_regex_idx;
+		while(res || root) {
 			struct cli_ac_result *q;
-			regex = res->customdata;
+			if (!res) {
+			    regex = matcher->suffix_regexes[root].head;
+			    root = 0;
+			} else {
+			    regex = res->customdata;
+			}
 			while(!rc && regex) {
 				/* loop over multiple regexes corresponding to
 				 * this suffix */
@@ -324,9 +330,11 @@ int regex_list_match(struct regex_matcher* matcher,char* real_url,const char* di
 				if(rc) *info = regex->pattern;
 				regex = regex->nxt;
 			}
-			q = res;
-			res = res->next;
-			free(q);
+			if (res) {
+			    q = res;
+			    res = res->next;
+			    free(q);
+			}
 		}
 		free(buffer);
 		if(!rc)
@@ -693,6 +701,8 @@ static int add_pattern_suffix(void *cbdata, const char *suffix, size_t suffix_le
 			return CL_EMEM;
 		matcher->suffix_regexes[n].tail = regex;
 		matcher->suffix_regexes[n].head = regex;
+		if (suffix[0] == '/' && suffix[1] == '\0')
+		    matcher->root_regex_idx = n;
 		add_newsuffix(matcher, regex, suffix, suffix_len);
 		cli_dbgmsg(MODULE "added new suffix %s, for regex: %s\n", suffix, regex->pattern);
 	}

libclamav/regex_list.h

@@ -47,6 +47,7 @@ struct regex_matcher {
 	struct hashtable suffix_hash;
 	size_t suffix_cnt;
 	struct regex_list_ht *suffix_regexes;
+	size_t root_regex_idx;
 	size_t regex_cnt;
 	regex_t **all_pregs;
 	struct cli_matcher suffixes;

unit_tests/check_regex.c

@@ -198,6 +198,8 @@ static const struct rtest {
 	{NULL,
 		"http://key.com","go to key.com",2
 	},
+	{":.+\\.paypal\\.(com|de|fr|it)([/?].*)?:.+\\.ebay\\.(at|be|ca|ch|co\\.uk|de|es|fr|ie|in|it|nl|ph|pl|com(\\.(au|cn|hk|my|sg))?)([/?].*)?/",
+	    "http://www.paypal.com", "pics.ebay.com", 1},
 	{NULL, "http://somefakeurl.example.com","someotherdomain-key.com",2},
 	{NULL, "http://somefakeurl.example.com","someotherdomain.key.com",0},
 	{NULL, "http://1.test.example.com/something","test",3},

unit_tests/input/daily.wdb

@@ -28,4 +28,4 @@ X:.+:.+rcm\.amazon\.com([/?].*)?:17-
 X:.+\.ebay\.com([/?].*)?:.+\.ebay\.(at|be|ca|ch|co\.uk|de|es|fr|ie|in|it|nl|ph|pl|com\.(au|cn|hk|my|sg))([/?].*)?:17-
 M:www.paypal.com%a0:www.paypal.com
 M:www.monster.fr:media.monster.com
-X:.+\.paypal\.(com|de|fr|it)([/?].*)?:.+\.ebay\.(at|be|ca|ch|co\.uk|de|es|fr|ie|in|it|nl|ph|pl|com(\.(au|cn|hk|my|sg))?)([/?].*)?:17-
+X:.+\.paypal\.com:.+\.ebay\.com