More changes to classification stages

diff --git a/src/libstat/stat_internal.h b/src/libstat/stat_internal.h
index c049486fe3f4725cb585ad9cdce84180e898040c..64019678836260c6f37aed6d18352b2f6eb66635 100644 (file)
--- a/src/libstat/stat_internal.h
+++ b/src/libstat/stat_internal.h
@@ -53,6 +53,7 @@ struct rspamd_statfile_runtime {
 
 struct rspamd_classifier_runtime {
        struct rspamd_classifier_config *clcf;
+       struct classifier_ctx *clctx;
        struct rspamd_stat_classifier *cl;
        struct rspamd_stat_backend *backend;
        struct rspamd_tokenizer_runtime *tok;

diff --git a/src/libstat/stat_process.c b/src/libstat/stat_process.c
index 89df10a07d2e06ef0be94b2261dba4942de4b319..db1ed9276167df6e446486b46f9a2febefc13552 100644 (file)
--- a/src/libstat/stat_process.c
+++ b/src/libstat/stat_process.c
@@ -546,7 +546,6 @@ rspamd_stat_classify (struct rspamd_task *task, lua_State *L, guint stage,
        struct rspamd_stat_ctx *st_ctx;
        struct rspamd_statfile_runtime *st_run;
        struct rspamd_classifier_runtime *cl_run;
-       struct classifier_ctx *cl_ctx;
        GList *cl_runtimes;
        GList *cur, *curst;
        gboolean ret = RSPAMD_STAT_PROCESS_OK;
@@ -564,19 +563,37 @@ rspamd_stat_classify (struct rspamd_task *task, lua_State *L, guint stage,
                        }
 
                        task->cl_runtimes = cl_runtimes;
+                       cur = cl_runtimes;
+
+                       /* Finalize backend so it can load tokens delayed if needed */
+                       while (cur) {
+                               cl_run = (struct rspamd_classifier_runtime *) cur->data;
+                               curst = cl_run->st_runtime;
+
+                               while (curst) {
+                                       st_run = curst->data;
+                                       cl_run->backend->finalize_process (task,
+                                                       st_run->backend_runtime,
+                                                       cl_run->backend->ctx);
+                                       curst = g_list_next (curst);
+                               }
+                       }
                }
-
+       }
+       else if (stage == RSPAMD_TASK_STAGE_CLASSIFIERS) {
                cur = cl_runtimes;
 
+               /* The first stage of classification */
                while (cur) {
                        cl_run = (struct rspamd_classifier_runtime *) cur->data;
                        cl_run->stage = RSPAMD_STAT_STAGE_PRE;
 
                        if (cl_run->cl) {
-                               cl_ctx = cl_run->cl->init_func (task->task_pool, cl_run->clcf);
+                               cl_run->clctx = cl_run->cl->init_func (task->task_pool,
+                                               cl_run->clcf);
 
-                               if (cl_ctx != NULL) {
-                                       cl_run->cl->classify_func (cl_ctx, cl_run->tok->tokens,
+                               if (cl_run->clctx != NULL) {
+                                       cl_run->cl->classify_func (cl_run->clctx, cl_run->tok->tokens,
                                                        cl_run, task);
                                }
                        }
@@ -584,8 +601,9 @@ rspamd_stat_classify (struct rspamd_task *task, lua_State *L, guint stage,
                        cur = g_list_next (cur);
                }
        }
-       else if (stage == RSPAMD_TASK_STAGE_CLASSIFIERS) {
+       else if (stage == RSPAMD_TASK_STAGE_CLASSIFIERS_POST) {
                cur = cl_runtimes;
+               /* The second stage of classification */
                while (cur) {
                        cl_run = (struct rspamd_classifier_runtime *) cur->data;
                        cl_run->stage = RSPAMD_STAT_STAGE_POST;
@@ -595,21 +613,6 @@ rspamd_stat_classify (struct rspamd_task *task, lua_State *L, guint stage,
                                continue;
                        }
 
-                       if (cl_run->cl) {
-                               if (cl_ctx != NULL) {
-                                       if (cl_run->cl->classify_func (cl_ctx, cl_run->tok->tokens,
-                                                       cl_run, task)) {
-                                               ret = RSPAMD_STAT_PROCESS_OK;
-                                       }
-                               }
-                       }
-
-                       cur = g_list_next (cur);
-               }
-       }
-       else if (stage == RSPAMD_TASK_STAGE_CLASSIFIERS_POST) {
-               cur = cl_runtimes;
-               while (cur) {
                        cl_run = (struct rspamd_classifier_runtime *) cur->data;
                        cl_run->stage = RSPAMD_STAT_STAGE_POST;
 
@@ -618,14 +621,14 @@ rspamd_stat_classify (struct rspamd_task *task, lua_State *L, guint stage,
                                continue;
                        }
 
-                       curst = cl_run->st_runtime;
-
-                       while (curst) {
-                               st_run = curst->data;
-                               cl_run->backend->finalize_process (task,
-                                               st_run->backend_runtime,
-                                               cl_run->backend->ctx);
-                               curst = g_list_next (curst);
+                       if (cl_run->cl) {
+                               if (cl_run->clctx != NULL) {
+                                       if (cl_run->cl->classify_func (cl_run->clctx,
+                                                       cl_run->tok->tokens,
+                                                       cl_run, task)) {
+                                               ret = RSPAMD_STAT_PROCESS_OK;
+                                       }
+                               }
                        }
 
                        cur = g_list_next (cur);