Another fix to the whole libstat architecture

src/libstat/learn_cache/learn_cache.h

@@ -36,7 +36,8 @@ struct rspamd_config;
 
 struct rspamd_stat_cache {
 	const char *name;
-	gpointer (*init)(struct rspamd_stat_ctx *ctx, struct rspamd_config *cfg);
+	gpointer (*init)(struct rspamd_stat_ctx *ctx,
+			struct rspamd_config *cfg, const ucl_object_t *cf);
 	gint (*process)(struct rspamd_task *task,
 			gboolean is_spam,
 			gpointer ctx);
@@ -45,7 +46,8 @@ struct rspamd_stat_cache {
 };
 
 gpointer rspamd_stat_cache_sqlite3_init(struct rspamd_stat_ctx *ctx,
-		struct rspamd_config *cfg);
+		struct rspamd_config *cfg,
+		const ucl_object_t *cf);
 gint rspamd_stat_cache_sqlite3_process (
 		struct rspamd_task *task,
 		gboolean is_spam, gpointer c);

src/libstat/learn_cache/sqlite3_cache.c

@@ -121,71 +121,51 @@ struct rspamd_stat_sqlite3_ctx {
 
 gpointer
 rspamd_stat_cache_sqlite3_init(struct rspamd_stat_ctx *ctx,
-		struct rspamd_config *cfg)
+		struct rspamd_config *cfg,
+		const ucl_object_t *cf)
 {
 	struct rspamd_stat_sqlite3_ctx *new = NULL;
-	struct rspamd_classifier_config *clf;
-	const ucl_object_t *obj, *elt;
-	GList *cur;
+	const ucl_object_t *elt;
 	gchar dbpath[PATH_MAX];
+	const gchar *path;
 	sqlite3 *sqlite;
-	gboolean has_sqlite_cache = FALSE;
 	GError *err = NULL;
 
-	rspamd_snprintf (dbpath, sizeof (dbpath), SQLITE_CACHE_PATH);
-	cur = cfg->classifiers;
 
-	while (cur) {
-		clf = cur->data;
+	if (cf) {
+		elt = ucl_object_find_key (cf, "path");
 
-		obj = ucl_object_find_key (clf->opts, "cache");
-
-		/* Sqlite3 cache is the default learn cache method */
-		if (obj == NULL) {
-			has_sqlite_cache = TRUE;
-			break;
+		if (elt != NULL) {
+			path = ucl_object_tostring (elt);
+		}
+		else {
+			path = SQLITE_CACHE_PATH;
 		}
-		else if (ucl_object_type (obj) == UCL_OBJECT) {
-			elt = ucl_object_find_key (obj, "name");
+	}
 
-			if (ucl_object_type (elt) == UCL_STRING &&
-				g_ascii_strcasecmp (ucl_object_tostring (elt), "sqlite3") == 0) {
+	rspamd_snprintf (dbpath, sizeof (dbpath), "%s", path);
 
-				has_sqlite_cache = TRUE;
-				elt = ucl_object_find_key (obj, "path");
-				if (elt != NULL && ucl_object_type (elt) == UCL_STRING) {
-					rspamd_snprintf (dbpath, sizeof (dbpath), "%s",
-							ucl_object_tostring (elt));
-				}
-			}
-		}
+	sqlite = rspamd_sqlite3_open_or_create (cfg->cfg_pool,
+			dbpath, create_tables_sql, &err);
 
-		cur = g_list_next (cur);
+	if (sqlite == NULL) {
+		msg_err ("cannot open sqlite3 cache: %e", err);
+		g_error_free (err);
+		err = NULL;
 	}
+	else {
+		new = g_slice_alloc (sizeof (*new));
+		new->db = sqlite;
+		new->prstmt = rspamd_sqlite3_init_prstmt (sqlite, prepared_stmts,
+				RSPAMD_STAT_CACHE_MAX, &err);
 
-	if (has_sqlite_cache) {
-		sqlite = rspamd_sqlite3_open_or_create (cfg->cfg_pool,
-				dbpath, create_tables_sql, &err);
-
-		if (sqlite == NULL) {
-			msg_err_config ("cannot open sqlite3 cache: %e", err);
+		if (new->prstmt == NULL) {
+			msg_err ("cannot open sqlite3 cache: %e", err);
 			g_error_free (err);
 			err = NULL;
-		}
-		else {
-			new = g_slice_alloc (sizeof (*new));
-			new->db = sqlite;
-			new->prstmt = rspamd_sqlite3_init_prstmt (sqlite, prepared_stmts,
-					RSPAMD_STAT_CACHE_MAX, &err);
-
-			if (new->prstmt == NULL) {
-				msg_err_config ("cannot open sqlite3 cache: %e", err);
-				g_error_free (err);
-				err = NULL;
-				sqlite3_close (sqlite);
-				g_slice_free1 (sizeof (*new), new);
-				new = NULL;
-			}
+			sqlite3_close (sqlite);
+			g_slice_free1 (sizeof (*new), new);
+			new = NULL;
 		}
 	}
 

src/libstat/stat_config.c

@@ -98,20 +98,25 @@ rspamd_stat_init (struct rspamd_config *cfg)
 	struct rspamd_statfile_config *stf;
 	struct rspamd_stat_backend *bk;
 	struct rspamd_statfile *st;
+	struct rspamd_classifier *cl;
+	const ucl_object_t *cache_obj = NULL, *cache_name_obj;
+	const gchar *cache_name = NULL;
 
 	if (stat_ctx == NULL) {
 		stat_ctx = g_slice_alloc0 (sizeof (*stat_ctx));
 	}
 
-	stat_ctx->backends = stat_backends;
+	stat_ctx->backends_subrs = stat_backends;
 	stat_ctx->backends_count = G_N_ELEMENTS (stat_backends);
-	stat_ctx->classifiers = stat_classifiers;
+	stat_ctx->classifiers_subrs = stat_classifiers;
 	stat_ctx->classifiers_count = G_N_ELEMENTS (stat_classifiers);
-	stat_ctx->tokenizers = stat_tokenizers;
+	stat_ctx->tokenizers_subrs = stat_tokenizers;
 	stat_ctx->tokenizers_count = G_N_ELEMENTS (stat_tokenizers);
-	stat_ctx->caches = stat_caches;
+	stat_ctx->caches_subrs = stat_caches;
 	stat_ctx->caches_count = G_N_ELEMENTS (stat_caches);
 	stat_ctx->cfg = cfg;
+	stat_ctx->statfiles = g_ptr_array_new ();
+	stat_ctx->classifiers = g_ptr_array_new ();
 	REF_RETAIN (stat_ctx->cfg);
 
 	/* Create statfiles from the classifiers */
@@ -133,16 +138,37 @@ rspamd_stat_init (struct rspamd_config *cfg)
 					clf->tokenizer, NULL);
 		}
 
+		cl = g_slice_alloc0 (sizeof (*cl));
+		cl->cfg = clf;
+		cl->statfiles_ids = g_array_new (FALSE, FALSE, sizeof (gint));
+
+		/* Init classifier cache */
+		if (clf->opts) {
+			cache_obj = ucl_object_find_key (clf->opts, "cache");
+
+			if (cache_obj) {
+				cache_name_obj = ucl_object_find_key (cache_obj, "name");
+			}
+
+			if (cache_name_obj) {
+				cache_name = ucl_object_tostring (cache_name_obj);
+			}
+		}
+
+		cl->cache = rspamd_stat_get_cache (cache_name);
+		g_assert (cl->cache != NULL);
+		cl->cachecf = cl->cache->init (stat_ctx, cfg, cache_obj);
+
 		curst = clf->statfiles;
 
 		while (curst) {
 			stf = curst->data;
 			st = g_slice_alloc0 (sizeof (*st));
-			st->clcf = clf;
+			st->classifier = cl;
 			st->stcf = stf;
-			st->tkcf = stat_ctx->tkcf;
-			st->bkcf = stat_ctx->backends[i].init (stat_ctx, cfg, st);
-			msg_debug_config ("added backend %s", stat_ctx->backends[i].name);
+			st->bkcf = stat_ctx->backends_subrs[i].init (stat_ctx, cfg, st);
+			msg_debug_config ("added backend %s",
+					stat_ctx->backends_subrs[i].name);
 
 			if (st->bkcf == NULL) {
 				msg_err_config ("cannot init backend %s for statfile %s",
@@ -151,19 +177,17 @@ rspamd_stat_init (struct rspamd_config *cfg)
 				g_slice_free1 (sizeof (*st), st);
 			}
 			else {
+				st->id = stat_ctx->statfiles->len;
 				g_ptr_array_add (stat_ctx->statfiles, st);
+				g_array_append_val (cl->statfiles_ids, st->id);
 			}
 
 			curst = curst->next;
 		}
 
-		cur = cur->next;
-	}
+		g_ptr_array_add (stat_ctx->classifiers, cl);
 
-	/* Init caches */
-	for (i = 0; i < stat_ctx->caches_count; i ++) {
-		stat_ctx->caches[i].ctx = stat_ctx->caches[i].init (stat_ctx, cfg);
-		msg_debug_config ("added cache %s", stat_ctx->caches[i].name);
+		cur = cur->next;
 	}
 }
 
@@ -176,9 +200,9 @@ rspamd_stat_close (void)
 	g_assert (stat_ctx != NULL);
 
 	for (i = 0; i < stat_ctx->backends_count; i ++) {
-		if (stat_ctx->backends[i].close != NULL) {
-			stat_ctx->backends[i].close (stat_ctx->backends[i].ctx);
-			msg_debug_config ("closed backend %s", stat_ctx->backends[i].name);
+		if (stat_ctx->backends_subrs[i].close != NULL) {
+			stat_ctx->backends_subrs[i].close (stat_ctx->backends_subrs[i].ctx);
+			msg_debug_config ("closed backend %s", stat_ctx->backends_subrs[i].name);
 		}
 	}
 
@@ -197,8 +221,8 @@ rspamd_stat_get_classifier (const gchar *name)
 	guint i;
 
 	for (i = 0; i < stat_ctx->classifiers_count; i ++) {
-		if (strcmp (name, stat_ctx->classifiers[i].name) == 0) {
-			return &stat_ctx->classifiers[i];
+		if (strcmp (name, stat_ctx->classifiers_subrs[i].name) == 0) {
+			return &stat_ctx->classifiers_subrs[i];
 		}
 	}
 
@@ -215,8 +239,8 @@ rspamd_stat_get_backend (const gchar *name)
 	}
 
 	for (i = 0; i < stat_ctx->backends_count; i ++) {
-		if (strcmp (name, stat_ctx->backends[i].name) == 0) {
-			return &stat_ctx->backends[i];
+		if (strcmp (name, stat_ctx->backends_subrs[i].name) == 0) {
+			return &stat_ctx->backends_subrs[i];
 		}
 	}
 
@@ -233,8 +257,26 @@ rspamd_stat_get_tokenizer (const gchar *name)
 	}
 
 	for (i = 0; i < stat_ctx->tokenizers_count; i ++) {
-		if (strcmp (name, stat_ctx->tokenizers[i].name) == 0) {
-			return &stat_ctx->tokenizers[i];
+		if (strcmp (name, stat_ctx->tokenizers_subrs[i].name) == 0) {
+			return &stat_ctx->tokenizers_subrs[i];
+		}
+	}
+
+	return NULL;
+}
+
+struct rspamd_stat_cache *
+rspamd_stat_get_cache (const gchar *name)
+{
+	guint i;
+
+	if (name == NULL || name[0] == '\0') {
+		name = RSPAMD_DEFAULT_CACHE;
+	}
+
+	for (i = 0; i < stat_ctx->caches_count; i++) {
+		if (strcmp (name, stat_ctx->caches_subrs[i].name) == 0) {
+			return &stat_ctx->caches_subrs[i];
 		}
 	}
 

src/libstat/stat_internal.h

@@ -95,17 +95,18 @@ typedef struct token_node_s {
 
 struct rspamd_stat_ctx {
 	/* Subroutines for all objects */
-	struct rspamd_stat_classifier *classifiers;
+	struct rspamd_stat_classifier *classifiers_subrs;
 	guint classifiers_count;
-	struct rspamd_stat_tokenizer *tokenizers;
+	struct rspamd_stat_tokenizer *tokenizers_subrs;
 	guint tokenizers_count;
-	struct rspamd_stat_backend *backends;
+	struct rspamd_stat_backend *backends_subrs;
 	guint backends_count;
-	struct rspamd_stat_cache *caches;
+	struct rspamd_stat_cache *caches_subrs;
 	guint caches_count;
 
 	/* Runtime configuration */
-	GPtrArray *statfiles; /* struct statfile */
+	GPtrArray *statfiles; /* struct rspamd_statfile */
+	GPtrArray *classifiers; /* struct rspamd_classifier */
 	struct rspamd_config *cfg;
 	/* Global tokenizer */
 	struct rspamd_stat_tokenizer *tokenizer;
@@ -122,6 +123,7 @@ struct rspamd_stat_ctx * rspamd_stat_get_ctx (void);
 struct rspamd_stat_classifier * rspamd_stat_get_classifier (const gchar *name);
 struct rspamd_stat_backend * rspamd_stat_get_backend (const gchar *name);
 struct rspamd_stat_tokenizer * rspamd_stat_get_tokenizer (const gchar *name);
+struct rspamd_stat_cache * rspamd_stat_get_cache (const gchar *name);
 
 static GQuark rspamd_stat_quark (void)
 {