Add prefix and suffix matching for strlist filter.

Signed-off-by: Florent Bruneau <florent.bruneau@polytechnique.org>

diff --git a/common/trie.c b/common/trie.c
index cb39cb2..f58bd30 100644 (file)
--- a/common/trie.c
+++ b/common/trie.c
@@ -111,7 +111,11 @@ static inline bool trie_entry_match(const trie_t *trie,
 static inline bool trie_entry_prefix(const trie_t *trie,
                                      const trie_entry_t *entry, const char *key)
 {
-    return !!(strncmp(array_ptr(trie->c, entry->c_offset), key, entry->c_len) == 0);
+    int len = entry->c_len;
+    if (len > 0 && array_elt(trie->c, entry->c_offset + len - 1) == '\0') {
+        --len;
+    }
+    return !!(strncmp(array_ptr(trie->c, entry->c_offset), key, len) == 0);
 }
 
 static inline bool trie_entry_is_leaf(const trie_entry_t *entry)
@@ -366,6 +370,11 @@ static inline void trie_entry_inspect(const trie_t *trie, bool show_content,
     static int leaves    = 0;
     static int depth_sum = 0;
 
+    if (entry == array_ptr(trie->entries, 0)) {
+      max_depth = 0;
+      leaves    = 0;
+      depth_sum = 0;
+    }
     if (trie_entry_is_leaf(entry)) {
         if (level > max_depth) {
             max_depth = level;

diff --git a/common/tst-trie.c b/common/tst-trie.c
index 990b151..78977a8 100644 (file)
--- a/common/tst-trie.c
+++ b/common/tst-trie.c
@@ -129,7 +129,7 @@ int main(int argc, char *argv[])
      */
     if (argc > 1) {
         trie = create_trie_from_file(argv[1]);
-        trie_inspect(trie, false);
+        trie_inspect(trie, true);
         trie_delete(&trie);
     }
     return 0;

diff --git a/example/postlicyd.conf b/example/postlicyd.conf
index fded2bd..ca17b60 100644 (file)
--- a/example/postlicyd.conf
+++ b/example/postlicyd.conf
@@ -101,15 +101,22 @@ spamhaus_and_abuseat {
 
 #     - strlist: match strings from the query against a list of list.
 #        Parameters:
-#           - file: (no)?lock:(pre|suf)fix:weight:filename
+#           - file: (no)?lock:(partial-)?(pre|suf)fix:weight:filename
 #             declare a file to load. If lock is given, the list is locked into the
 #             RAM. Prefix/Suffix is a parameter to tell the matcher which is the most
 #             efficient storage order. The strings are internally stored into a trie that
 #             allow high compression if a lot of prefix are shared by several strings. If
 #             you choose "prefix", string are stored in the natural order in memory and
 #             prefix compression is performed. If you choose "suffix", strings are stored
-#             in reverse order in memory and suffix compression is performed. The weight
-#             is a number giving the weight of this list in the string score.
+#             in reverse order in memory and suffix compression is performed. If you add "partial-"
+#             to the match order, the entry will match if the file contains a prefix (resp. suffix)
+#             of the string. The weight is a number giving the weight of this list in the string score.
+#               e.g.:
+#                * a file that contains ".polytechnique.org" in "partial-suffix" mode will match
+#                 all subdomains of "polytechnique.org".
+#                * a file that contains "postmaster@" in "partial-prefix" mode will match all
+#                 postmaster emails.
+#                * a file open without "partial-" modifier match exact strings.
 #           - soft_threshold: score (default: 1)
 #             minimum score to match the soft_match return value
 #           - hard_threshold: score (default: 1)
@@ -140,7 +147,7 @@ client_whitelist {
   type  = strlist;
 
   # configuration
-  file    = lock:1:/var/spool/postlicyd/client_whitelist;
+  file    = lock:1:suffix:/var/spool/postlicyd/client_whitelist;
   fields  = client_name;
 
   # hooks

diff --git a/postlicyd/data/test.conf b/postlicyd/data/test.conf
index 6e0a7a0..8f4c9a0 100644 (file)
--- a/postlicyd/data/test.conf
+++ b/postlicyd/data/test.conf
@@ -55,6 +55,17 @@ hostnames2 {
   on_fail = postfix:OK;
 }
 
+hostnames3 {
+  type = strlist;
+
+  fields = client_name;
+  file   = nolock:partial-suffix:1:data/test_hostnames_4;
+
+  on_hard_match = postfix:OK;
+  on_fail = postfix:OK;
+}
+
+
 emails1 {
   type = strlist;
 
@@ -87,6 +98,16 @@ emails2 {
   on_fail = postfix:ok;
 }
 
+emails3 {
+  type = strlist;
+
+  fields = sender;
+  file   = nolock:partial-prefix:1:data/test_emails_4;
+
+  on_hard_match = postfix:ok;
+  on_fail = postfix:ok;
+}
+
 ips1 {
   type = iplist;
 

diff --git a/postlicyd/data/testcase_1 b/postlicyd/data/testcase_1
index a562889..4356694 100644 (file)
--- a/postlicyd/data/testcase_1
+++ b/postlicyd/data/testcase_1
@@ -27,6 +27,8 @@ match1=match
 match2=fail
 hostnames1=fail
 hostnames2=fail
+hostnames3=fail
 emails1=fail
 emails2=fail
+emails3=fail
 ips1=hard_match

diff --git a/postlicyd/data/testcase_2 b/postlicyd/data/testcase_2
index 55fdab7..674df87 100644 (file)
--- a/postlicyd/data/testcase_2
+++ b/postlicyd/data/testcase_2
@@ -3,7 +3,7 @@ protocol_state=RCPT
 protocol_name=SMTP
 helo_name=example.org
 queue_id=8045F2AB23
-sender=contact@example.com
+sender=postmaster@example.com
 recipient=contact@example.org
 recipient_count=0
 client_address=1.2.3.4
@@ -27,6 +27,8 @@ match1=fail
 match2=match
 hostnames1=soft_match
 hostnames2=hard_match
+hostnames3=hard_match
 emails1=soft_match
 emails2=hard_match
+emails3=hard_match
 ips1=hard_match

diff --git a/postlicyd/data/testcase_3 b/postlicyd/data/testcase_3
index efe3754..c6c2670 100644 (file)
--- a/postlicyd/data/testcase_3
+++ b/postlicyd/data/testcase_3
@@ -27,6 +27,8 @@ match1=fail
 match2=match
 hostnames1=fail
 hostnames2=hard_match
+hostnames3=hard_match
 emails1=fail
 emails2=fail
+emails3=fail
 ips1=error

diff --git a/postlicyd/data/testcase_4 b/postlicyd/data/testcase_4
index beed42d..2ab60ee 100644 (file)
--- a/postlicyd/data/testcase_4
+++ b/postlicyd/data/testcase_4
@@ -27,6 +27,8 @@ match1=fail
 match2=fail
 hostnames1=fail
 hostnames2=hard_match
+hostnames3=hard_match
 emails1=fail
 emails2=fail
+emails3=fail
 ips1=error

diff --git a/postlicyd/data/testcase_5 b/postlicyd/data/testcase_5
index 9b96b98..d28b7f4 100644 (file)
--- a/postlicyd/data/testcase_5
+++ b/postlicyd/data/testcase_5
@@ -27,6 +27,8 @@ match1=fail
 match2=match
 hostnames1=fail
 hostnames2=hard_match
+hostnames3=hard_match
 emails1=fail
 emails2=fail
+emails3=fail
 ips1=fail

diff --git a/postlicyd/data/testcase_6 b/postlicyd/data/testcase_6
index 45213d7..616b315 100644 (file)
--- a/postlicyd/data/testcase_6
+++ b/postlicyd/data/testcase_6
@@ -3,7 +3,7 @@ protocol_state=RCPT
 protocol_name=SMTP
 helo_name=tata.example.org
 queue_id=8045F2AB23
-sender=contact@exemple.com
+sender=postmaster@exemple.com
 recipient=contact@exemple.org
 recipient_count=0
 client_address=2.3.4.5
@@ -27,6 +27,8 @@ match1=fail
 match2=fail
 hostnames1=fail
 hostnames2=hard_match
+hostnames3=hard_match
 emails1=fail
 emails2=fail
+emails3=hard_match
 ips1=soft_match

diff --git a/postlicyd/strlist.c b/postlicyd/strlist.c
index 875ee23..4ca2057 100644 (file)
--- a/postlicyd/strlist.c
+++ b/postlicyd/strlist.c
@@ -43,6 +43,7 @@ typedef struct strlist_config_t {
     PA(trie_t) tries;
     A(int)     weights;
     A(bool)    reverses;
+    A(bool)    partiales;
 
     int soft_threshold;
     int hard_threshold;
@@ -69,6 +70,7 @@ static void strlist_config_delete(strlist_config_t **config)
         array_deep_wipe((*config)->tries, trie_delete);
         array_wipe((*config)->weights);
         array_wipe((*config)->reverses);
+        array_wipe((*config)->partiales);
         p_delete(config);
     }
 }
@@ -177,6 +179,7 @@ static bool strlist_filter_constructor(filter_t *filter)
             bool lock = false;
             int  weight = 0;
             bool reverse = false;
+            bool partial = false;
             trie_t *trie = NULL;
             const char *current = param->value;
             const char *p = m_strchrnul(param->value, ':');
@@ -198,6 +201,11 @@ static bool strlist_filter_constructor(filter_t *filter)
                     break;
 
                   case 1:
+                    if (p - current > (ssize_t)strlen("partial-") 
+                        && strncmp(current, "partial-", strlen("partial-")) == 0) {
+                        partial = true;
+                        current += strlen("partial-");
+                    }
                     if ((p - current) == 6 && strncmp(current, "suffix", 6) == 0) {
                         reverse = true;
                     } else if ((p - current) == 6 && strncmp(current, "prefix", 6) == 0) {
@@ -222,6 +230,7 @@ static bool strlist_filter_constructor(filter_t *filter)
                     array_add(config->tries, trie);
                     array_add(config->weights, weight);
                     array_add(config->reverses, reverse);
+                    array_add(config->partiales, partial);
                     break;
                 }
                 if (i != 3) {
@@ -326,8 +335,10 @@ static filter_result_t strlist_filter(const filter_t *filter, const query_t *que
         for (uint32_t i = 0 ; i < config->tries.len ; ++i) {                   \
             const int weight   = array_elt(config->weights, i);                \
             const trie_t *trie = array_elt(config->tries, i);                  \
-            const bool rev = array_elt(config->reverses, i);                   \
-            if (trie_lookup(trie, rev ? reverse : normal)) {                   \
+            const bool rev     = array_elt(config->reverses, i);               \
+            const bool part    = array_elt(config->partiales, i);              \
+            if ((!part && trie_lookup(trie, rev ? reverse : normal))           \
+                || (part && trie_prefix(trie, rev ? reverse : normal))) {      \
                 sum += weight;                                                 \
             }                                                                  \
         }                                                                      \