* Fix errors in learning implementation

diff --git a/rspamd.conf.sample b/rspamd.conf.sample
index dec751baae4d9ac021ff536cf5a4efbc4b26b794..24bcda5989879947b2b268a507db60ce5e06e9dc 100644 (file)
--- a/rspamd.conf.sample
+++ b/rspamd.conf.sample
@@ -46,7 +46,7 @@ logging {
 
 # Limit for statfile pool size
 # Default: 100M
-statfile_pool_size = 10M;
+statfile_pool_size = 20M;
 
 
 # Sample statfile definition

diff --git a/src/controller.c b/src/controller.c
index 42e790e8aad395733573286783d253f8e329dbcc..2115210e0d20350b71d465740497221415a91978 100644 (file)
--- a/src/controller.c
+++ b/src/controller.c
@@ -35,6 +35,7 @@ enum command_type {
        COMMAND_SHUTDOWN,
        COMMAND_UPTIME,
        COMMAND_LEARN,
+       COMMAND_HELP,
 };
 
 struct controller_command {
@@ -51,11 +52,14 @@ static struct controller_command commands[] = {
        {"shutdown", 1, COMMAND_SHUTDOWN},
        {"uptime", 0, COMMAND_UPTIME},
        {"learn", 1, COMMAND_LEARN},
+       {"help", 0, COMMAND_HELP},
 };
 
 static GCompletion *comp;
 static time_t start_time;
 
+static char greetingbuf[1024];
+
 static 
 void sig_handler (int signo)
 {
@@ -93,6 +97,7 @@ completion_func (gpointer elem)
 static void
 free_session (struct controller_session *session)
 {
+       msg_debug ("free_session: freeing session %p", session);
        bufferevent_disable (session->bev, EV_READ | EV_WRITE);
        bufferevent_free (session->bev);
        memory_pool_delete (session->session_pool);
@@ -144,7 +149,9 @@ process_command (struct controller_command *cmd, char **cmd_args, struct control
                        }
                        break;
                case COMMAND_QUIT:
-                       free_session (session);
+                               session->state = STATE_QUIT;
+                               r = snprintf (out_buf, sizeof (out_buf), "bye" CRLF);
+                               bufferevent_write (session->bev, out_buf, r);
                        break;
                case COMMAND_RELOAD:
                        if (check_auth (cmd, session)) {
@@ -159,6 +166,7 @@ process_command (struct controller_command *cmd, char **cmd_args, struct control
                                r = snprintf (out_buf, sizeof (out_buf), "-- end of stats report" CRLF);
                                bufferevent_write (session->bev, out_buf, r);
                        }
+                       break;
                case COMMAND_SHUTDOWN:
                        if (check_auth (cmd, session)) {
                                r = snprintf (out_buf, sizeof (out_buf), "shutdown request sent" CRLF);
@@ -174,20 +182,20 @@ process_command (struct controller_command *cmd, char **cmd_args, struct control
                                        days = uptime / 86400;
                                        hours = (uptime % 3600) / 60;
                                        minutes = (uptime % 60) / 60;
-                                       r = snprintf (out_buf, sizeof (out_buf), "%dday%s %dhour%s %dminute%s" CRLF, 
+                                       r = snprintf (out_buf, sizeof (out_buf), "%d day%s %d hour%s %d minute%s" CRLF, 
                                                                days, days > 1 ? "s" : " ",
                                                                hours, hours > 1 ? "s" : " ",
                                                                minutes, minutes > 1 ? "s" : " ");
                                }
                                /* If uptime is less than 1 minute print only seconds */
                                else if (uptime / 60 == 0) {
-                                       r = snprintf (out_buf, sizeof (out_buf), "%dsecond%s", (int)uptime, (int)uptime > 1 ? "s" : " ");
+                                       r = snprintf (out_buf, sizeof (out_buf), "%d second%s" CRLF, (int)uptime, (int)uptime > 1 ? "s" : " ");
                                }
                                /* Else print the minutes and seconds. */
                                else {
                                        hours = uptime / 3600;
                                        minutes = (uptime % 60) / 60;
-                                       r = snprintf (out_buf, sizeof (out_buf), "%dhour%s %dminite%s %dsecond%s", 
+                                       r = snprintf (out_buf, sizeof (out_buf), "%d hour%s %d minite%s %d second%s" CRLF, 
                                                                hours, hours > 1 ? "s" : " ",
                                                                minutes, minutes > 1 ? "s" : " ",
                                                                (int)uptime, uptime > 1 ? "s" : " ");
@@ -197,14 +205,14 @@ process_command (struct controller_command *cmd, char **cmd_args, struct control
                        break;
                case COMMAND_LEARN:
                        if (check_auth (cmd, session)) {
-                               arg = *cmd_args++;
+                               arg = *cmd_args;
                                if (!arg || *arg == '\0') {
                                        msg_debug ("process_command: no statfile specified in learn command");
                                        r = snprintf (out_buf, sizeof (out_buf), "learn command requires at least two arguments: stat filename and its size" CRLF);
                                        bufferevent_write (session->bev, out_buf, r);
                                        return;
                                }
-                               arg = *cmd_args;
+                               arg = *(cmd_args + 1);
                                if (arg == NULL || *arg == '\0') {
                                        msg_debug ("process_command: no statfile size specified in learn command");
                                        r = snprintf (out_buf, sizeof (out_buf), "learn command requires at least two arguments: stat filename and its size" CRLF);
@@ -225,10 +233,12 @@ process_command (struct controller_command *cmd, char **cmd_args, struct control
                                        bufferevent_write (session->bev, out_buf, r);
                                        return;
                                }
+                               session->learn_buf->pos = session->learn_buf->buf->begin;
+                               update_buf_size (session->learn_buf);
 
-                               statfile = g_hash_table_lookup (session->cfg->statfiles, arg);
+                               statfile = g_hash_table_lookup (session->cfg->statfiles, *cmd_args);
                                if (statfile == NULL) {
-                                       r = snprintf (out_buf, sizeof (out_buf), "statfile %s is not defined" CRLF, arg);
+                                       r = snprintf (out_buf, sizeof (out_buf), "statfile %s is not defined" CRLF, *cmd_args);
                                        bufferevent_write (session->bev, out_buf, r);
                                        return;
 
@@ -243,7 +253,7 @@ process_command (struct controller_command *cmd, char **cmd_args, struct control
                                /* Get all arguments */
                                while (*cmd_args++) {
                                        arg = *cmd_args;
-                                       if (*arg == '-') {
+                                       if (arg && *arg == '-') {
                                                switch (*(arg + 1)) {
                                                        case 'r':
                                                                arg = *(cmd_args + 1);
@@ -273,10 +283,12 @@ process_command (struct controller_command *cmd, char **cmd_args, struct control
                                                }
                                        }
                                }
-                               session->learn_filename = resolve_stat_filename (session->session_pool, statfile->pattern, session->learn_rcpt, session->learn_from);
+                               session->learn_filename = resolve_stat_filename (session->session_pool, statfile->pattern, 
+                                                                                                                                       session->learn_rcpt, session->learn_from);
                                if (statfile_pool_open (session->worker->srv->statfile_pool, session->learn_filename) == -1) {
                                        /* Try to create statfile */
-                                       if (statfile_pool_create (session->worker->srv->statfile_pool, session->learn_filename, statfile->size) == -1) {
+                                       if (statfile_pool_create (session->worker->srv->statfile_pool, 
+                                                                       session->learn_filename, statfile->size / sizeof (struct stat_file_block)) == -1) {
                                                r = snprintf (out_buf, sizeof (out_buf), "cannot create statfile %s" CRLF, session->learn_filename);
                                                bufferevent_write (session->bev, out_buf, r);
                                                return;
@@ -288,10 +300,20 @@ process_command (struct controller_command *cmd, char **cmd_args, struct control
                                        }
                                }
                                session->state = STATE_LEARN;
-                               r = snprintf (out_buf, sizeof (out_buf), "ok" CRLF);
-                               bufferevent_write (session->bev, out_buf, r);
-                               break;
                        }
+                       break;
+               case COMMAND_HELP:
+                               r = snprintf (out_buf, sizeof (out_buf), 
+                                                       "Rspamd CLI commands (* - privilleged command):" CRLF
+                                                       "    help - this help message" CRLF
+                                                       "(*) learn <statfile> <size> [-r recipient], [-f from] [-n] - learn message to specified statfile" CRLF
+                                                       "    quit - quit CLI session" CRLF
+                                                       "(*) reload - reload rspamd" CRLF
+                                                       "(*) shutdown - shutdown rspamd" CRLF
+                                                       "    stat - show different rspamd stat" CRLF
+                                                       "    uptime - rspamd uptime" CRLF);
+                               bufferevent_write (session->bev, out_buf, r);
+                       break;
        }
 }
 
@@ -338,6 +360,12 @@ read_socket (struct bufferevent *bev, void *arg)
                                                        break;
                                        }
                                }
+                               if (session->state == STATE_COMMAND) {
+                                       session->state = STATE_REPLY;
+                               }
+                               if (session->state != STATE_LEARN) {
+                                       bufferevent_enable (bev, EV_WRITE);
+                               }
                                g_strfreev (params);
                        }
                        if (s != NULL) {
@@ -354,22 +382,23 @@ read_socket (struct bufferevent *bev, void *arg)
                                        if (tokens == NULL) {
                                                i = snprintf (out_buf, sizeof (out_buf), "learn fail, tokenizer error" CRLF);
                                                bufferevent_write (bev, out_buf, i);
-                                               session->state = STATE_COMMAND;
+                                               session->state = STATE_REPLY;
                                                return;
                                        }
                                        session->learn_classifier->learn_func (session->worker->srv->statfile_pool, session->learn_filename, tokens, session->in_class);
                                        session->worker->srv->stat->messages_learned ++;
                                        i = snprintf (out_buf, sizeof (out_buf), "learn ok" CRLF);
                                        bufferevent_write (bev, out_buf, i);
-                                       session->state = STATE_COMMAND;
+                                       bufferevent_enable (bev, EV_WRITE);
+                                       session->state = STATE_REPLY;
                                        break;
                                }
                        }
                        else {
                                i = snprintf (out_buf, sizeof (out_buf), "read error: %d" CRLF, i);
                                bufferevent_write (bev, out_buf, i);
-                               bufferevent_disable (bev, EV_READ);
-                               free_session (session);
+                               bufferevent_enable (bev, EV_WRITE);
+                               session->state = STATE_REPLY;
                        }
                        break;
        }
@@ -378,11 +407,18 @@ read_socket (struct bufferevent *bev, void *arg)
 static void
 write_socket (struct bufferevent *bev, void *arg)
 {
-       char buf[1024], hostbuf[256];
-
-       gethostname (hostbuf, sizeof (hostbuf - 1));
-       hostbuf[sizeof (hostbuf) - 1] = '\0';
-       snprintf (buf, sizeof (buf), "Rspamd version %s is running on %s" CRLF, RVERSION, hostbuf);
+       struct controller_session *session = (struct controller_session *)arg;
+       
+       if (session->state == STATE_QUIT) {
+               msg_info ("closing control connection");
+               /* Free buffers */
+               close (session->sock);
+               free_session (session);
+               return;
+       }
+       else if (session->state == STATE_REPLY) {
+               session->state = STATE_COMMAND;
+       }
        bufferevent_disable (bev, EV_WRITE);
        bufferevent_enable (bev, EV_READ);
 }
@@ -427,6 +463,7 @@ accept_socket (int fd, short what, void *arg)
 
        /* Read event */
        new_session->bev = bufferevent_new (nfd, read_socket, write_socket, err_socket, (void *)new_session);
+       bufferevent_write (new_session->bev, greetingbuf, strlen (greetingbuf));
        bufferevent_enable (new_session->bev, EV_WRITE);
 }
 
@@ -437,6 +474,8 @@ start_controller (struct rspamd_worker *worker)
        int listen_sock, i;
        struct sockaddr_un *un_addr;
        GList *comp_list = NULL;
+       char *hostbuf;
+       long int hostmax;
 
        worker->srv->pid = getpid ();
        worker->srv->type = TYPE_CONTROLLER;
@@ -472,11 +511,17 @@ start_controller (struct rspamd_worker *worker)
        }
        
        /* Init command completion */
-       for (i = 0; i < sizeof (commands) / sizeof (commands[0]) - 1; i ++) {
+       for (i = 0; i < G_N_ELEMENTS (commands); i ++) {
                comp_list = g_list_prepend (comp_list, &commands[i]);
        }
        comp = g_completion_new (completion_func);
        g_completion_add_items (comp, comp_list);
+       /* Fill hostname buf */
+       hostmax = sysconf (_SC_HOST_NAME_MAX) + 1;
+       hostbuf = alloca (hostmax);
+       gethostname (hostbuf, hostmax);
+       hostbuf[hostmax - 1] = '\0';
+       snprintf (greetingbuf, sizeof (greetingbuf), "Rspamd version %s is running on %s" CRLF, RVERSION, hostbuf);
        /* Accept event */
        event_set(&worker->bind_ev, listen_sock, EV_READ | EV_PERSIST, accept_socket, (void *)worker);
        event_add(&worker->bind_ev, NULL);

diff --git a/src/main.c b/src/main.c
index 454f164ead102a7df11c695083f2a245f4be5715..c903d3ede0808576d80d3e6bd86acc41fb7fd6d4 100644 (file)
--- a/src/main.c
+++ b/src/main.c
@@ -418,7 +418,7 @@ main (int argc, char **argv, char **env)
                        if (active_worker != NULL) {
                                msg_info ("main: worker process %d has been successfully started", active_worker->pid);
                                TAILQ_FOREACH_SAFE (cur, &rspamd->workers, next, cur_tmp) {
-                                       if (cur != active_worker && !cur->is_dying) {
+                                       if (cur != active_worker && !cur->is_dying && cur->type == TYPE_WORKER) {
                                                /* Send to old workers SIGUSR2 */
                                                kill (cur->pid, SIGUSR2);
                                                cur->is_dying = 1;

diff --git a/src/main.h b/src/main.h
index abe163b3c767150422187ccfc630c6756e37be92..b2e362c9d2be162f6f78bcae13e7b3bab1209ca5 100644 (file)
--- a/src/main.h
+++ b/src/main.h
@@ -129,6 +129,8 @@ struct controller_session {
        enum {
                STATE_COMMAND,
                STATE_LEARN,
+               STATE_REPLY,
+               STATE_QUIT,
        } state;
        int sock;
        /* Access to authorized commands */

diff --git a/src/tokenizers/osb.c b/src/tokenizers/osb.c
index be930af282b1c413e5e77752b20a053133580fce..afd2febd86ac8d72058f3382d7717d5276483878 100644 (file)
--- a/src/tokenizers/osb.c
+++ b/src/tokenizers/osb.c
@@ -44,7 +44,7 @@ osb_tokenize_text (struct tokenizer *tokenizer, memory_pool_t *pool, f_str_t *in
                }
                hashpipe[0] = fstrhash (&token);
                
-               for (i = 0; i < FEATURE_WINDOW_SIZE - 2; i ++) {
+               for (i = 1; i < FEATURE_WINDOW_SIZE; i ++) {
                        h1 = hashpipe[0]* primes[0] + hashpipe[i] * primes[i<<1];
                    h2 = hashpipe[0] * primes[1] + hashpipe[i] * primes[(i<<1)-1];
                        new = memory_pool_alloc (pool, sizeof (token_node_t));

diff --git a/src/tokenizers/tokenizers.c b/src/tokenizers/tokenizers.c
index 853207af4f3de8f150a839694e546f9cbd71a33e..280ebd47719c2ec3b066ddadf3abc22cae5cdfa4 100644 (file)
--- a/src/tokenizers/tokenizers.c
+++ b/src/tokenizers/tokenizers.c
@@ -50,14 +50,14 @@ get_next_word (f_str_t *buf, f_str_t *token)
                token->begin = buf->begin;
        }
 
+       token->begin = token->begin + token->len;
+       token->len = 0;
+       
        remain = buf->len - (token->begin - buf->begin);
        if (remain <= 0) {
                return NULL;
        }
 
-       token->begin = token->begin + token->len;
-       token->len = 0;
-       
        pos = token->begin;
        /* Skip non graph symbols */
        while (remain-- && !g_ascii_isgraph (*pos ++)) {
@@ -66,6 +66,10 @@ get_next_word (f_str_t *buf, f_str_t *token)
        while (remain-- && g_ascii_isgraph (*pos ++)) {
                token->len ++;
        }
+
+       if (token->len == 0) {
+               return NULL;
+       }
        
        return token;
 }

diff --git a/src/util.c b/src/util.c
index b217800ad786c7b5acb972ad0551edccdadf1946..9444592cf2542de71045aab5b3bb624a019171ab 100644 (file)
--- a/src/util.c
+++ b/src/util.c
@@ -827,10 +827,23 @@ char *
 resolve_stat_filename (memory_pool_t *pool, char *pattern, char *rcpt, char *from)
 {
        int need_to_format = 0, len = 0;
-       int rcptlen = strlen (rcpt);
-       int fromlen = strlen (from);
+       int rcptlen, fromlen;
        char *c = pattern, *new, *s;
        
+       if (rcpt) {
+               rcptlen = strlen (rcpt);
+       }
+       else {
+               rcptlen = 0;
+       }
+
+       if (from) {
+               fromlen = strlen (from);
+       }
+       else {
+               fromlen = 0;
+       }
+
        /* Calculate length */
        while (*c++) {
                if (*c == '%' && *(c + 1) == 'r') {