From: Vsevolod Stakhov <vsevolod@rambler-co.ru>
Date: Tue, 14 Jul 2009 13:09:49 +0000 (+0400)
Subject: * Fix symbol planning, add cache_file directive to config file
X-Git-Tag: 0.2.7~88
X-Git-Url: https://source.dussan.org/?a=commitdiff_plain;h=5f5254ff214fd77a6fdc8cce9269c29837fd162a;p=rspamd.git

* Fix symbol planning, add cache_file directive to config file
* Fix some errors
---

diff --git a/src/cfg_file.l b/src/cfg_file.l
index 06e1a33e7..7f2931ccc 100644
--- a/src/cfg_file.l
+++ b/src/cfg_file.l
@@ -58,6 +58,7 @@ metric							return METRIC;
 name							return NAME;
 required_score					return REQUIRED_SCORE;
 function						return FUNCTION;
+cache_file						return CACHE_FILE;
 control							return CONTROL;
 password						return PASSWORD;
 lmtp							return LMTP;
diff --git a/src/cfg_file.y b/src/cfg_file.y
index af54a9fb4..7a2aa3cb7 100644
--- a/src/cfg_file.y
+++ b/src/cfg_file.y
@@ -49,7 +49,7 @@ struct rspamd_view *cur_view = NULL;
 %token  READ_SERVERS WRITE_SERVER DIRECTORY_SERVERS MAILBOX_QUERY USERS_QUERY LASTLOGIN_QUERY
 %token  MEMCACHED WORKER TYPE REQUIRE MODULE
 %token  MODULE_OPT PARAM VARIABLE
-%token  FILTERS FACTORS METRIC NAME
+%token  FILTERS FACTORS METRIC NAME CACHE_FILE
 %token  REQUIRED_SCORE FUNCTION FRACT COMPOSITES CONTROL PASSWORD
 %token  LOGGING LOG_TYPE LOG_TYPE_CONSOLE LOG_TYPE_SYSLOG LOG_TYPE_FILE
 %token  LOG_LEVEL LOG_LEVEL_DEBUG LOG_LEVEL_INFO LOG_LEVEL_WARNING LOG_LEVEL_ERROR LOG_FACILITY LOG_FILENAME
@@ -362,6 +362,7 @@ metriccmd:
 	| metricfunction
 	| metricscore
 	| metricclassifier
+	| metriccache
 	;
 	
 metricname:
@@ -416,6 +417,15 @@ metricclassifier:
 	}
 	;
 
+metriccache:
+	CACHE_FILE EQSIGN QUOTEDSTRING {
+		if (cur_metric == NULL) {
+			cur_metric = memory_pool_alloc0 (cfg->cfg_pool, sizeof (struct metric));
+		}
+		cur_metric->cache_filename = memory_pool_strdup (cfg->cfg_pool, $3);
+	}
+	;
+
 factors:
 	FACTORS OBRACE factorsbody EBRACE
 	;
diff --git a/src/cfg_utils.c b/src/cfg_utils.c
index 5688d44a0..7307db803 100644
--- a/src/cfg_utils.c
+++ b/src/cfg_utils.c
@@ -171,7 +171,6 @@ parse_bind_line (struct config_file *cfg, struct worker_conf *cf, char *str)
 void
 init_defaults (struct config_file *cfg)
 {
-	struct metric *def_metric;
 
 	cfg->memcached_error_time = DEFAULT_UPSTREAM_ERROR_TIME;
 	cfg->memcached_dead_time = DEFAULT_UPSTREAM_DEAD_TIME;
@@ -189,14 +188,6 @@ init_defaults (struct config_file *cfg)
 	cfg->statfiles = g_hash_table_new (g_str_hash, g_str_equal);
 	cfg->cfg_params = g_hash_table_new (g_str_hash, g_str_equal);
 
-	def_metric = memory_pool_alloc (cfg->cfg_pool, sizeof (struct metric));
-	def_metric->name = "default";
-	def_metric->func_name = "factors";
-	def_metric->func = factor_consolidation_func;
-	def_metric->required_score = DEFAULT_SCORE;
-	def_metric->classifier = get_classifier ("winnow");
-	g_hash_table_insert (cfg->metrics, "default", def_metric);
-
 }
 
 void
@@ -479,6 +470,7 @@ void
 post_load_config (struct config_file *cfg)
 {
 	struct timespec ts;
+	struct metric *def_metric;
 
 	g_hash_table_foreach (cfg->variables, substitute_all_variables, cfg);
 	g_hash_table_foreach (cfg->modules_opts, substitute_module_variables, cfg);
@@ -499,6 +491,18 @@ post_load_config (struct config_file *cfg)
 	if (cfg->clock_res > 3) {
 		cfg->clock_res = 3;
 	}
+
+	if (g_hash_table_lookup (cfg->metrics, DEFAULT_METRIC) == NULL) {
+		def_metric = memory_pool_alloc (cfg->cfg_pool, sizeof (struct metric));
+		def_metric->name = DEFAULT_METRIC;
+		def_metric->func_name = "factors";
+		def_metric->func = factor_consolidation_func;
+		def_metric->required_score = DEFAULT_SCORE;
+		def_metric->classifier = get_classifier ("winnow");
+		cfg->metrics_list = g_list_prepend (cfg->metrics_list, def_metric);
+		g_hash_table_insert (cfg->metrics, DEFAULT_METRIC, def_metric);
+	}
+
 }
 
 
diff --git a/src/filter.c b/src/filter.c
index 2bfd2bc36..a1894603d 100644
--- a/src/filter.c
+++ b/src/filter.c
@@ -50,6 +50,8 @@ insert_result (struct worker_task *task, const char *metric_name, const char *sy
 	struct metric *metric;
 	struct metric_result *metric_res;
 	struct symbol *s;
+	struct cache_item *item;
+	int i;
 
 	metric = g_hash_table_lookup (task->worker->srv->cfg->metrics, metric_name);
 	if (metric == NULL) {
@@ -95,6 +97,17 @@ insert_result (struct worker_task *task, const char *metric_name, const char *sy
 
 		g_hash_table_insert (metric_res->symbols, (gpointer)symbol, s);
 	}
+
+	/* Process cache item */
+	if (metric->cache) {
+		for (i = 0; i < metric->cache->used_items; i ++) {
+			item = &metric->cache->items[i];
+
+			if (flag > 0 && strcmp (item->s->symbol, symbol) == 0) {
+				item->s->frequency ++;
+			}
+		}
+	}
 }
 
 /*
@@ -219,6 +232,7 @@ check_metric_is_spam (struct worker_task *task, struct metric *metric)
 
 	res = g_hash_table_lookup (task->results, metric->name);
 	if (res) {
+		metric_process_callback_forced (metric->name, res, task);
 		return res->score >= metric->required_score;
 	}
 
diff --git a/src/main.c b/src/main.c
index beefabdcb..72d5b1d62 100644
--- a/src/main.c
+++ b/src/main.c
@@ -616,7 +616,9 @@ main (int argc, char **argv, char **env)
 	l = g_list_first (cfg->metrics_list);
 	while (l) {
 		metric = l->data;
-		init_symbols_cache (cfg->cfg_pool, metric->cache, metric->cache_filename);
+		if (!init_symbols_cache (cfg->cfg_pool, metric->cache, metric->cache_filename)) {
+			exit (EXIT_FAILURE);
+		}
 		l = g_list_next (l);
 	}
 	
diff --git a/src/plugins/chartable.c b/src/plugins/chartable.c
index 19c46c7b7..951285070 100644
--- a/src/plugins/chartable.c
+++ b/src/plugins/chartable.c
@@ -108,10 +108,10 @@ chartable_module_config (struct config_file *cfg)
 	/* Search in factors hash table */
 	w = g_hash_table_lookup (cfg->factors, chartable_module_ctx->symbol);
 	if (w == NULL) {
-		register_symbol (metric->cache, chartable_module_ctx->symbol, 1, chartable_symbol_callback, NULL);
+		register_symbol (&metric->cache, chartable_module_ctx->symbol, 1, chartable_symbol_callback, NULL);
 	}
 	else {
-		register_symbol (metric->cache, chartable_module_ctx->symbol, *w, chartable_symbol_callback, NULL);
+		register_symbol (&metric->cache, chartable_module_ctx->symbol, *w, chartable_symbol_callback, NULL);
 	}
 
 	return res;
diff --git a/src/plugins/emails.c b/src/plugins/emails.c
index 6b789916b..ef886ecf5 100644
--- a/src/plugins/emails.c
+++ b/src/plugins/emails.c
@@ -116,10 +116,10 @@ emails_module_config (struct config_file *cfg)
 	/* Search in factors hash table */
 	w = g_hash_table_lookup (cfg->factors, email_module_ctx->symbol);
 	if (w == NULL) {
-		register_symbol (metric->cache, email_module_ctx->symbol, 1, emails_symbol_callback, NULL);
+		register_symbol (&metric->cache, email_module_ctx->symbol, 1, emails_symbol_callback, NULL);
 	}
 	else {
-		register_symbol (metric->cache, email_module_ctx->symbol, *w, emails_symbol_callback, NULL);
+		register_symbol (&metric->cache, email_module_ctx->symbol, *w, emails_symbol_callback, NULL);
 	}
 
 	return res;
diff --git a/src/plugins/regexp.c b/src/plugins/regexp.c
index 7b123dd33..23ec3908c 100644
--- a/src/plugins/regexp.c
+++ b/src/plugins/regexp.c
@@ -194,10 +194,10 @@ regexp_module_config (struct config_file *cfg)
 		/* Search in factors hash table */
 		w = g_hash_table_lookup (cfg->factors, cur->param);
 		if (w == NULL) {
-			register_symbol (metric->cache, cur->param, 1, process_regexp_item, cur_item);
+			register_symbol (&metric->cache, cur->param, 1, process_regexp_item, cur_item);
 		}
 		else {
-			register_symbol (metric->cache, cur->param, *w, process_regexp_item, cur_item);
+			register_symbol (&metric->cache, cur->param, *w, process_regexp_item, cur_item);
 		}
 		
 		cur_opt = g_list_next (cur_opt);
diff --git a/src/plugins/surbl.c b/src/plugins/surbl.c
index 691c57681..15bdc149b 100644
--- a/src/plugins/surbl.c
+++ b/src/plugins/surbl.c
@@ -195,10 +195,10 @@ surbl_module_config (struct config_file *cfg)
 				/* Search in factors hash table */
 				w = g_hash_table_lookup (cfg->factors, new_suffix->symbol);
 				if (w == NULL) {
-					register_symbol (metric->cache, new_suffix->symbol, 1, surbl_test_url, NULL);
+					register_symbol (&metric->cache, new_suffix->symbol, 1, surbl_test_url, new_suffix);
 				}
 				else {
-					register_symbol (metric->cache, new_suffix->symbol, *w, surbl_test_url, NULL);
+					register_symbol (&metric->cache, new_suffix->symbol, *w, surbl_test_url, new_suffix);
 				}
 			}
 		}
@@ -227,10 +227,10 @@ surbl_module_config (struct config_file *cfg)
 		surbl_module_ctx->suffixes = g_list_prepend (surbl_module_ctx->suffixes, new_suffix);
 		w = g_hash_table_lookup (cfg->factors, new_suffix->symbol);
 		if (w == NULL) {
-			register_symbol (metric->cache, new_suffix->symbol, 1, surbl_test_url, new_suffix);
+			register_symbol (&metric->cache, new_suffix->symbol, 1, surbl_test_url, new_suffix);
 		}
 		else {
-			register_symbol (metric->cache, new_suffix->symbol, *w, surbl_test_url, new_suffix);
+			register_symbol (&metric->cache, new_suffix->symbol, *w, surbl_test_url, new_suffix);
 		}
 	}
 
diff --git a/src/symbols_cache.c b/src/symbols_cache.c
index 60cdd2644..5ed0a92d9 100644
--- a/src/symbols_cache.c
+++ b/src/symbols_cache.c
@@ -46,7 +46,7 @@ int
 cache_cmp (const void *p1, const void *p2)
 {
 	const struct cache_item *i1 = p1, *i2 = p2;
-
+	
 	return strcmp (i1->s->symbol, i2->s->symbol);
 }
 
@@ -70,22 +70,20 @@ static void
 grow_cache (struct symbols_cache *cache)
 {
 	guint old = cache->cur_items, i;
+	void *new;
 
 	cache->cur_items = cache->cur_items * 2;
-	cache->items = g_renew (struct cache_item, cache->items, cache->cur_items);
+	new = g_new0 (struct cache_item, cache->cur_items);
+	memcpy (new, cache->items, old * sizeof (struct cache_item));
+	g_free (cache->items);
+	cache->items = new;
+
 	/* Create new saved_cache_items */
 	for (i = old - 1; i < cache->cur_items; i ++) {
-		cache->items[i].s = g_malloc (sizeof (struct saved_cache_item));
+		cache->items[i].s = g_new0 (struct saved_cache_item, 1);
 	}
 }
 
-static void
-truncate_cache (struct symbols_cache *cache)
-{
-	cache->items = g_renew (struct cache_item, cache->items, cache->used_items);
-	cache->cur_items = cache->used_items;
-}
-
 static GChecksum *
 get_mem_cksum (struct symbols_cache *cache)
 {
@@ -186,41 +184,37 @@ create_cache_file (struct symbols_cache *cache, const char *filename, int fd)
 }
 
 void 
-register_symbol (struct symbols_cache *cache, const char *name, double weight, symbol_func_t func, gpointer user_data)
+register_symbol (struct symbols_cache **cache, const char *name, double weight, symbol_func_t func, gpointer user_data)
 {
 	struct cache_item *item = NULL;
 	int i;
 	
-	if (cache == NULL) {
-		cache = g_new0 (struct symbols_cache, 1);
+	if (*cache == NULL) {
+		*cache = g_new0 (struct symbols_cache, 1);
 	}
-	if (cache->items == NULL) {
-		cache->cur_items = MIN_CACHE;
-		cache->used_items = 0;
-		cache->items = g_new0 (struct cache_item, cache->cur_items);
-		for (i = 0; i < cache->cur_items; i ++) {
-			cache->items[i].s = g_malloc (sizeof (struct saved_cache_item));
-		}
-	}
-
-	for (i = 0; i < cache->cur_items; i ++) {
-		if (cache->items[i].s->symbol[0] != '\0') {
-			item = &cache->items[i];
+	if ((*cache)->items == NULL) {
+		(*cache)->cur_items = MIN_CACHE;
+		(*cache)->used_items = 0;
+		(*cache)->items = g_new0 (struct cache_item, (*cache)->cur_items);
+		for (i = 0; i < (*cache)->cur_items; i ++) {
+			(*cache)->items[i].s = g_new0 (struct saved_cache_item, 1);
 		}
 	}
 	
-	if (item == NULL) {
-		grow_cache (cache);
+	if ((*cache)->used_items >= (*cache)->cur_items) {
+		grow_cache (*cache);
 		/* Call once more */
 		register_symbol (cache, name, weight, func, user_data);
 		return;
 	}
 
+	item = &(*cache)->items[(*cache)->used_items];
+	
 	g_strlcpy (item->s->symbol, name, sizeof (item->s->symbol));
 	item->func = func;
 	item->user_data = user_data;
 	item->s->weight = weight;
-	cache->used_items ++;
+	(*cache)->used_items ++;
 	set_counter (item->s->symbol, 0);
 }
 
@@ -237,7 +231,6 @@ init_symbols_cache (memory_pool_t *pool, struct symbols_cache *cache, const char
 		return FALSE;
 	}
 	
-	truncate_cache (cache);
 	/* Sort items in cache */
 	qsort (cache->items, cache->used_items, sizeof (struct cache_item), cache_cmp);
 
@@ -288,7 +281,7 @@ init_symbols_cache (memory_pool_t *pool, struct symbols_cache *cache, const char
 
 	g_checksum_get_digest (cksum, mem_sum, &cklen);
 	/* Now try to read file sum */
-	if (lseek (fd, SEEK_END, -(cklen)) == -1) {
+	if (lseek (fd, -(cklen), SEEK_END) == -1) {
 		close (fd);
 		g_free (mem_sum);
 		g_checksum_free (cksum);
@@ -340,6 +333,7 @@ call_symbol_callback (struct worker_task *task, struct symbols_cache *cache, str
 			return FALSE;
 		}
 		if (cache->uses ++ >= MAX_USES) {
+			msg_info ("call_symbols_callback: resort symbols cache");
 			memory_pool_wlock_rwlock (cache->lock);
 			cache->uses = 0;
 			/* Resort while having write lock */
@@ -350,7 +344,7 @@ call_symbol_callback (struct worker_task *task, struct symbols_cache *cache, str
 	}
 	else {
 		/* Next pointer */
-		if (*saved_item - cache->items == cache->used_items) {
+		if (*saved_item - cache->items >= cache->used_items - 1) {
 			/* No more items in cache */
 			return FALSE;
 		}
@@ -358,7 +352,6 @@ call_symbol_callback (struct worker_task *task, struct symbols_cache *cache, str
 		item = *saved_item + 1;
 		memory_pool_runlock_rwlock (cache->lock);
 	}
-
 	if (check_view (task->cfg->views, item->s->symbol, task)) {
 #ifdef HAVE_CLOCK_PROCESS_CPUTIME_ID
 		clock_gettime (CLOCK_PROCESS_CPUTIME_ID, &ts1);
@@ -379,7 +372,6 @@ call_symbol_callback (struct worker_task *task, struct symbols_cache *cache, str
 
 		diff = (ts2.tv_sec - ts1.tv_sec) * 1000000 + (ts2.tv_nsec - ts1.tv_nsec) / 1000;
 		item->s->avg_time = set_counter (item->s->symbol, diff);
-		item->s->frequency ++;
 	}
 
 	*saved_item = item;
diff --git a/src/symbols_cache.h b/src/symbols_cache.h
index e8e0be24f..65028cfcc 100644
--- a/src/symbols_cache.h
+++ b/src/symbols_cache.h
@@ -41,7 +41,7 @@ gboolean init_symbols_cache (memory_pool_t *pool, struct symbols_cache *cache, c
  * @param func pointer to handler
  * @param user_data pointer to user_data
  */
-void register_symbol (struct symbols_cache *cache, const char *name, double weight, symbol_func_t func, gpointer user_data);
+void register_symbol (struct symbols_cache **cache, const char *name, double weight, symbol_func_t func, gpointer user_data);
 
 /**
  * Call function for cached symbol using saved callback