64bits fixes.
[apps/pfixtools.git] / postlicyd / strlist.c
index 875ee23..0c790cb 100644 (file)
 #include "trie.h"
 #include "file.h"
 #include "str.h"
+#include "rbl.h"
 #include "policy_tokens.h"
 
 typedef struct strlist_config_t {
     PA(trie_t) tries;
     A(int)     weights;
     A(bool)    reverses;
+    A(bool)    partiales;
+
+    A(char)     hosts;
+    A(int)      host_offsets;
+    A(int)      host_weights;
 
     int soft_threshold;
     int hard_threshold;
 
     unsigned is_email         :1;
+    unsigned is_hostname      :1;
+
     unsigned match_sender     :1;
     unsigned match_recipient  :1;
 
-    unsigned is_hostname      :1;
     unsigned match_helo       :1;
     unsigned match_client     :1;
     unsigned match_reverse    :1;
@@ -69,6 +76,10 @@ 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);
+        array_wipe((*config)->hosts);
+        array_wipe((*config)->host_offsets);
+        array_wipe((*config)->host_weights);
         p_delete(config);
     }
 }
@@ -109,8 +120,8 @@ static trie_t *strlist_create(const char *file, bool reverse, bool lock)
         --end;
     }
     if (end != map.end) {
-        syslog(LOG_WARNING, "file %s miss a final \\n, ignoring last line",
-               file);
+        warn("file %s miss a final \\n, ignoring last line",
+             file);
     }
 
     db = trie_new();
@@ -120,7 +131,7 @@ static trie_t *strlist_create(const char *file, bool reverse, bool lock)
             eol = end;
         }
         if (eol - p >= BUFSIZ) {
-            syslog(LOG_ERR, "unreasonnable long line");
+            err("unreasonnable long line");
             file_map_close(&map);
             trie_delete(&db);
             return NULL;
@@ -145,6 +156,86 @@ static trie_t *strlist_create(const char *file, bool reverse, bool lock)
     return db;
 }
 
+static bool strlist_create_from_rhbl(const char *file, bool lock,
+                                     trie_t **phosts, trie_t **pdomains)
+{
+    trie_t *hosts, *domains;
+    uint32_t host_count, domain_count;
+    file_map_t map;
+    const char *p, *end;
+    char line[BUFSIZ];
+
+    if (!file_map_open(&map, file, false)) {
+        return false;
+    }
+    p   = map.map;
+    end = map.end;
+    while (end > p && end[-1] != '\n') {
+        --end;
+    }
+    if (end != map.end) {
+        warn("file %s miss a final \\n, ignoring last line",
+             file);
+    }
+
+    hosts = trie_new();
+    host_count = 0;
+    domains = trie_new();
+    domain_count = 0;
+    while (p < end && p != NULL) {
+        const char *eol = (char *)memchr(p, '\n', end - p);
+        if (eol == NULL) {
+            eol = end;
+        }
+        if (eol - p >= BUFSIZ) {
+            err("unreasonnable long line");
+            file_map_close(&map);
+            trie_delete(&hosts);
+            trie_delete(&domains);
+            return false;
+        }
+        if (*p != '#') {
+            const char *eos = eol;
+            while (p < eos && isspace(*p)) {
+                ++p;
+            }
+            while (p < eos && isspace(eos[-1])) {
+                --eos;
+            }
+            if (p < eos) {
+                if (isalnum(*p)) {
+                    strlist_copy(line, p, eos - p, true);
+                    trie_insert(hosts, line);
+                    ++host_count;
+                } else if (*p == '*') {
+                    ++p;
+                    strlist_copy(line, p, eos - p, true);
+                    trie_insert(domains, line);
+                    ++domain_count;
+                }
+            }
+        }
+        p = eol + 1;
+    }
+    file_map_close(&map);
+    if (host_count > 0) {
+        trie_compile(hosts, lock);
+        *phosts = hosts;
+    } else {
+        trie_delete(&hosts);
+        *phosts = NULL;
+    }
+    if (domain_count > 0) {
+        trie_compile(domains, lock);
+        *pdomains = domains;
+    } else {
+        trie_delete(&domains);
+        *pdomains = NULL;
+    }
+    return hosts != NULL || domains != NULL;
+
+}
+
 
 static bool strlist_filter_constructor(filter_t *filter)
 {
@@ -152,7 +243,7 @@ static bool strlist_filter_constructor(filter_t *filter)
 
 #define PARSE_CHECK(Expr, Str, ...)                                            \
     if (!(Expr)) {                                                             \
-        syslog(LOG_ERR, Str, ##__VA_ARGS__);                                   \
+        err(Str, ##__VA_ARGS__);                                               \
         strlist_config_delete(&config);                                        \
         return false;                                                          \
     }
@@ -162,7 +253,7 @@ static bool strlist_filter_constructor(filter_t *filter)
     foreach (filter_param_t *param, filter->params) {
         switch (param->type) {
           /* file parameter is:
-           *  [no]lock:(prefix|suffix):weight:filename
+           *  [no]lock:(partial-)(prefix|suffix):weight:filename
            *  valid options are:
            *    - lock:   memlock the database in memory.
            *    - nolock: don't memlock the database in memory.
@@ -177,6 +268,7 @@ 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, ':');
@@ -193,18 +285,23 @@ static bool strlist_filter_constructor(filter_t *filter)
                         lock = false;
                     } else {
                         PARSE_CHECK(false, "illegal locking state %.*s",
-                                    p - current, current);
+                                    (int)(p - current), current);
                     }
                     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) {
                         reverse = false;
                     } else {
                         PARSE_CHECK(false, "illegal character order value %.*s",
-                                    p - current, current);
+                                    (int)(p - current), current);
                     }
                     break;
 
@@ -212,7 +309,7 @@ static bool strlist_filter_constructor(filter_t *filter)
                     weight = strtol(current, &next, 10);
                     PARSE_CHECK(next == p && weight >= 0 && weight <= 1024,
                                 "illegal weight value %.*s",
-                                (p - current), current);
+                                (int)(p - current), current);
                     break;
 
                   case 3:
@@ -222,6 +319,7 @@ 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) {
@@ -231,6 +329,105 @@ static bool strlist_filter_constructor(filter_t *filter)
             }
           } break;
 
+          /* rbldns parameter is:
+           *  [no]lock::weight:filename
+           *  valid options are:
+           *    - lock:   memlock the database in memory.
+           *    - nolock: don't memlock the database in memory.
+           *    - \d+:    a number describing the weight to give to the match
+           *              the given list [mandatory]
+           *  directly import a file issued from a rhbl in rbldns format.
+           */
+          case ATK_RBLDNS: {
+            bool lock = false;
+            int  weight = 0;
+            trie_t *trie_hosts   = NULL;
+            trie_t *trie_domains = NULL;
+            const char *current = param->value;
+            const char *p = m_strchrnul(param->value, ':');
+            char *next = NULL;
+            for (int i = 0 ; i < 3 ; ++i) {
+                PARSE_CHECK(i == 2 || *p,
+                            "file parameter must contains a locking state "
+                            "and a weight option");
+                switch (i) {
+                  case 0:
+                    if ((p - current) == 4 && strncmp(current, "lock", 4) == 0) {
+                        lock = true;
+                    } else if ((p - current) == 6 && strncmp(current, "nolock", 6) == 0) {
+                        lock = false;
+                    } else {
+                        PARSE_CHECK(false, "illegal locking state %.*s",
+                                    (int)(p - current), current);
+                    }
+                    break;
+
+                  case 1:
+                    weight = strtol(current, &next, 10);
+                    PARSE_CHECK(next == p && weight >= 0 && weight <= 1024,
+                                "illegal weight value %.*s",
+                                (int)(p - current), current);
+                    break;
+
+                  case 2:
+                    PARSE_CHECK(strlist_create_from_rhbl(current, lock,
+                                                         &trie_hosts, &trie_domains),
+                                "cannot load string list from rhbl %s", current);
+                    if (trie_hosts != NULL) {
+                        array_add(config->tries, trie_hosts);
+                        array_add(config->weights, weight);
+                        array_add(config->reverses, true);
+                        array_add(config->partiales, false);
+                    }
+                    if (trie_domains != NULL) {
+                        array_add(config->tries, trie_domains);
+                        array_add(config->weights, weight);
+                        array_add(config->reverses, true);
+                        array_add(config->partiales, true);
+                    }
+                    config->is_hostname = true;
+                    break;
+                }
+                if (i != 2) {
+                    current = p + 1;
+                    p = m_strchrnul(current, ':');
+                }
+            }
+          } break;
+
+          /* dns parameter.
+           *  weight:hostname.
+           * define a RBL to use through DNS resolution.
+           */
+          case ATK_DNS: {
+            int  weight = 0;
+            const char *current = param->value;
+            const char *p = m_strchrnul(param->value, ':');
+            char *next = NULL;
+            for (int i = 0 ; i < 2 ; ++i) {
+                PARSE_CHECK(i == 1 || *p,
+                            "host parameter must contains a weight option");
+                switch (i) {
+                  case 0:
+                    weight = strtol(current, &next, 10);
+                    PARSE_CHECK(next == p && weight >= 0 && weight <= 1024,
+                                "illegal weight value %.*s",
+                                (int)(p - current), current);
+                    break;
+
+                  case 1:
+                    array_add(config->host_offsets, array_len(config->hosts));
+                    array_append(config->hosts, current, strlen(current) + 1);
+                    array_add(config->host_weights, weight);
+                    break;
+                }
+                if (i != 1) {
+                    current = p + 1;
+                    p = m_strchrnul(current, ':');
+                }
+            }
+          } break;
+
           /* hard_threshold parameter is an integer.
            *  If the matching score is greater or equal than this threshold,
            *  the hook "hard_match" is called.
@@ -267,11 +464,13 @@ static bool strlist_filter_constructor(filter_t *filter)
                   CASE(HELO_NAME, helo, hostname);
                   CASE(CLIENT_NAME, client, hostname);
                   CASE(REVERSE_CLIENT_NAME, reverse, hostname);
+                  CASE(SENDER_DOMAIN, sender, hostname);
+                  CASE(RECIPIENT_DOMAIN, recipient, hostname);
                   CASE(SENDER, sender, email);
                   CASE(RECIPIENT, recipient, email);
 #undef CASE
                   default:
-                    PARSE_CHECK(false, "unknown field %.*s", p - current, current);
+                    PARSE_CHECK(false, "unknown field %.*s", (int)(p - current), current);
                     break;
                 }
                 if (!*p) {
@@ -288,7 +487,7 @@ static bool strlist_filter_constructor(filter_t *filter)
 
     PARSE_CHECK(config->is_email != config->is_hostname,
                 "matched field MUST be emails XOR hostnames");
-    PARSE_CHECK(config->tries.len,
+    PARSE_CHECK(config->tries.len || config->host_offsets.len,
                 "no file parameter in the filter %s", filter->name);
     filter->data = config;
     return true;
@@ -307,15 +506,17 @@ static filter_result_t strlist_filter(const filter_t *filter, const query_t *que
     char normal[BUFSIZ];
     const strlist_config_t *config = filter->data;
     int sum = 0;
+    bool error = true;
+
+
     if (config->is_email && 
         ((config->match_sender && query->state < SMTP_MAIL)
         || (config->match_recipient && query->state != SMTP_RCPT))) {
-        syslog(LOG_WARNING, "trying to match an email against a field that is not "
-               "available in current protocol state");
+        warn("trying to match an email against a field that is not "
+             "available in current protocol state");
         return HTK_ABORT;
     } else if (config->is_hostname && config->match_helo && query->state < SMTP_HELO) {
-        syslog(LOG_WARNING, "trying to match hostname against helo before helo "
-               "is received");
+        warn("trying to match hostname against helo before helo is received");
         return HTK_ABORT;
     }
 #define LOOKUP(Flag, Field)                                                    \
@@ -326,21 +527,67 @@ 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;                                                 \
+                if (sum >= config->hard_threshold) {                           \
+                    return HTK_HARD_MATCH;                                     \
+                }                                                              \
             }                                                                  \
+            error = false;                                                     \
         }                                                                      \
     }
+#define DNS(Flag, Field)                                                       \
+    if (config->match_ ## Flag) {                                              \
+        const int len = m_strlen(query->Field);                                \
+        strlist_copy(normal, query->Field, len, false);                        \
+        for (uint32_t i = 0 ; len > 0 && i < config->host_offsets.len ; ++i) { \
+            const char *rbl    = array_ptr(config->hosts,                      \
+                                           array_elt(config->host_offsets, i));\
+            const int weight   = array_elt(config->host_weights, i);           \
+            switch (rhbl_check(normal, rbl)) {                                 \
+              case RBL_FOUND:                                                  \
+                error = false;                                                 \
+                sum += weight;                                                 \
+                if (sum >= config->hard_threshold) {                           \
+                    return HTK_HARD_MATCH;                                     \
+                }                                                              \
+                break;                                                         \
+              case RBL_NOTFOUND:                                               \
+                error = false;                                                 \
+                break;                                                         \
+              case RBL_ERROR:                                                  \
+                warn("rbl %s unavailable", rbl);                               \
+                break;                                                         \
+            }                                                                  \
+        }                                                                      \
+    }
+
     if (config->is_email) {
         LOOKUP(sender, sender);
         LOOKUP(recipient, recipient);
+        DNS(sender, sender);
+        DNS(recipient, recipient);
     } else if (config->is_hostname) {
         LOOKUP(helo, helo_name);
         LOOKUP(client, client_name);
         LOOKUP(reverse, reverse_client_name);
+        LOOKUP(recipient, recipient_domain);
+        LOOKUP(sender, sender_domain);
+        DNS(helo, helo_name);
+        DNS(client, client_name);
+        DNS(reverse, reverse_client_name);
+        DNS(recipient, recipient_domain);
+        DNS(sender, sender_domain);
     }
+#undef  DNS
 #undef  LOOKUP
+    if (error) {
+        err("filter %s: all the rbls returned an error", filter->name);
+        return HTK_ERROR;
+    }
     if (sum >= config->hard_threshold) {
         return HTK_HARD_MATCH;
     } else if (sum >= config->soft_threshold) {
@@ -365,6 +612,8 @@ static int strlist_init(void)
     /* Parameters.
      */
     (void)filter_param_register(type, "file");
+    (void)filter_param_register(type, "rbldns");
+    (void)filter_param_register(type, "dns");
     (void)filter_param_register(type, "hard_threshold");
     (void)filter_param_register(type, "soft_threshold");
     (void)filter_param_register(type, "fields");