[Feature] Allow `g+:` and `g-:` composite atoms

diff --git a/src/libserver/composites.c b/src/libserver/composites.c
index 83f3a35d49e4637cc40103e30efdcbe9ff37c4b8..d553ce7efeb7b5efcbd330ee63ba99d555c3e969 100644 (file)
--- a/src/libserver/composites.c
+++ b/src/libserver/composites.c
@@ -94,7 +94,7 @@ rspamd_composite_expr_parse (const gchar *line, gsize len,
        /*
         * Composites are just sequences of symbols
         */
-       clen = strcspn (line, ", \t()><+!|&\n");
+       clen = strcspn (line, ", \t()><!|&\n");
        if (clen == 0) {
                /* Invalid composite atom */
                g_set_error (err, rspamd_composites_quark (), 100, "Invalid composite: %s",
@@ -172,13 +172,86 @@ rspamd_composite_process_single_symbol (struct composites_data *cd,
        return rc;
 }
 
+static void
+rspamd_composite_process_symbol_removal (rspamd_expression_atom_t *atom,
+                                                                                struct composites_data *cd,
+                                                                                struct rspamd_symbol_result *ms,
+                                                                                const gchar *beg)
+{
+       gchar t;
+       struct symbol_remove_data *rd, *nrd;
+
+       if (ms == NULL) {
+               return;
+       }
+
+       /*
+        * At this point we know that we need to do something about this symbol,
+        * however, we don't know whether we need to delete it unfortunately,
+        * that depends on the later decisions when the complete expression is
+        * evaluated.
+        */
+       rd = g_hash_table_lookup (cd->symbols_to_remove, ms->name);
+
+       nrd = rspamd_mempool_alloc (cd->task->task_pool, sizeof (*nrd));
+       nrd->sym = ms->name;
+
+       /* By default remove symbols */
+       switch (cd->composite->policy) {
+       case RSPAMD_COMPOSITE_POLICY_REMOVE_ALL:
+       default:
+               nrd->action = (RSPAMD_COMPOSITE_REMOVE_SYMBOL|RSPAMD_COMPOSITE_REMOVE_WEIGHT);
+               break;
+       case RSPAMD_COMPOSITE_POLICY_REMOVE_SYMBOL:
+               nrd->action = RSPAMD_COMPOSITE_REMOVE_SYMBOL;
+               break;
+       case RSPAMD_COMPOSITE_POLICY_REMOVE_WEIGHT:
+               nrd->action = RSPAMD_COMPOSITE_REMOVE_WEIGHT;
+               break;
+       case RSPAMD_COMPOSITE_POLICY_LEAVE:
+               nrd->action = 0;
+               break;
+       }
+
+       for (;;) {
+               t = *beg;
+
+               if (t == '~') {
+                       nrd->action &= ~RSPAMD_COMPOSITE_REMOVE_SYMBOL;
+               }
+               else if (t == '-') {
+                       nrd->action &= ~(RSPAMD_COMPOSITE_REMOVE_WEIGHT|
+                                                        RSPAMD_COMPOSITE_REMOVE_SYMBOL);
+               }
+               else if (t == '^') {
+                       nrd->action |= RSPAMD_COMPOSITE_REMOVE_FORCED;
+               }
+               else {
+                       break;
+               }
+
+               beg ++;
+       }
+
+       nrd->comp = cd->composite;
+       nrd->parent = atom->parent;
+
+       if (rd == NULL) {
+               DL_APPEND (rd, nrd);
+               g_hash_table_insert (cd->symbols_to_remove, (gpointer)ms->name, rd);
+       }
+       else {
+               DL_APPEND (rd, nrd);
+       }
+}
+
 static gdouble
-rspamd_composite_expr_process (struct rspamd_expr_process_data *process_data, rspamd_expression_atom_t *atom)
+rspamd_composite_expr_process (struct rspamd_expr_process_data *process_data,
+               rspamd_expression_atom_t *atom)
 {
        struct composites_data *cd = process_data->cd;
        const gchar *beg = atom->data, *sym = NULL;
-       gchar t;
-       struct symbol_remove_data *rd, *nrd;
+
        struct rspamd_symbol_result *ms = NULL;
        struct rspamd_symbols_group *gr;
        struct rspamd_symbol *sdef;
@@ -225,73 +298,70 @@ rspamd_composite_expr_process (struct rspamd_expr_process_data *process_data, rs
                                rc = rspamd_composite_process_single_symbol (cd, sdef->name, &ms);
 
                                if (rc) {
-                                       break;
+                                       rspamd_composite_process_symbol_removal (atom,
+                                                       cd,
+                                                       ms,
+                                                       beg);
                                }
                        }
                }
        }
-       else {
-               rc = rspamd_composite_process_single_symbol (cd, sym, &ms);
-       }
+       else if (strncmp (sym, "g+:", 3) == 0) {
+               /* Group, positive symbols only */
+               gr = g_hash_table_lookup (cd->task->cfg->groups, sym + 3);
 
-       if (rc != 0 && ms) {
-               /*
-                * At this point we know that we need to do something about this symbol,
-                * however, we don't know whether we need to delete it unfortunately,
-                * that depends on the later decisions when the complete expression is
-                * evaluated.
-                */
-               rd = g_hash_table_lookup (cd->symbols_to_remove, ms->name);
+               if (gr != NULL) {
+                       g_hash_table_iter_init (&it, gr->symbols);
 
-               nrd = rspamd_mempool_alloc (cd->task->task_pool, sizeof (*nrd));
-               nrd->sym = ms->name;
+                       while (g_hash_table_iter_next (&it, &k, &v)) {
+                               sdef = v;
 
-               /* By default remove symbols */
-               switch (cd->composite->policy) {
-               case RSPAMD_COMPOSITE_POLICY_REMOVE_ALL:
-               default:
-                       nrd->action = (RSPAMD_COMPOSITE_REMOVE_SYMBOL|RSPAMD_COMPOSITE_REMOVE_WEIGHT);
-                       break;
-               case RSPAMD_COMPOSITE_POLICY_REMOVE_SYMBOL:
-                       nrd->action = RSPAMD_COMPOSITE_REMOVE_SYMBOL;
-                       break;
-               case RSPAMD_COMPOSITE_POLICY_REMOVE_WEIGHT:
-                       nrd->action = RSPAMD_COMPOSITE_REMOVE_WEIGHT;
-                       break;
-               case RSPAMD_COMPOSITE_POLICY_LEAVE:
-                       nrd->action = 0;
-                       break;
+                               if (sdef->score > 0) {
+                                       rc = rspamd_composite_process_single_symbol (cd,
+                                                       sdef->name,
+                                                       &ms);
+
+                                       if (rc) {
+                                               rspamd_composite_process_symbol_removal (atom,
+                                                               cd,
+                                                               ms,
+                                                               beg);
+                                       }
+                               }
+                       }
                }
+       }
+       else if (strncmp (sym, "g-:", 3) == 0) {
+               /* Group, positive symbols only */
+               gr = g_hash_table_lookup (cd->task->cfg->groups, sym + 3);
 
-               for (;;) {
-                       t = *beg;
-
-                       if (t == '~') {
-                               nrd->action &= ~RSPAMD_COMPOSITE_REMOVE_SYMBOL;
-                       }
-                       else if (t == '-') {
-                               nrd->action &= ~(RSPAMD_COMPOSITE_REMOVE_WEIGHT|
-                                               RSPAMD_COMPOSITE_REMOVE_SYMBOL);
-                       }
-                       else if (t == '^') {
-                               nrd->action |= RSPAMD_COMPOSITE_REMOVE_FORCED;
-                       }
-                       else {
-                               break;
-                       }
+               if (gr != NULL) {
+                       g_hash_table_iter_init (&it, gr->symbols);
 
-                       beg ++;
-               }
+                       while (g_hash_table_iter_next (&it, &k, &v)) {
+                               sdef = v;
 
-               nrd->comp = cd->composite;
-               nrd->parent = atom->parent;
+                               if (sdef->score < 0) {
+                                       rc = rspamd_composite_process_single_symbol (cd, sdef->name, &ms);
 
-               if (rd == NULL) {
-                       DL_APPEND (rd, nrd);
-                       g_hash_table_insert (cd->symbols_to_remove, (gpointer)ms->name, rd);
+                                       if (rc) {
+                                               rspamd_composite_process_symbol_removal (atom,
+                                                               cd,
+                                                               ms,
+                                                               beg);
+                                       }
+                               }
+                       }
                }
-               else {
-                       DL_APPEND (rd, nrd);
+       }
+       else {
+               rc = rspamd_composite_process_single_symbol (cd, sym, &ms);
+
+               if (rc) {
+                       rspamd_composite_process_symbol_removal (atom,
+                                       cd,
+                                       ms,
+                                       beg);
                }
        }