From: Vsevolod Stakhov <vsevolod@highsecure.ru>
Date: Mon, 12 Nov 2018 11:16:45 +0000 (+0000)
Subject: [Feature] Add support of words regexps
X-Git-Tag: 1.8.2~54
X-Git-Url: http://git.ipfire.org/gitweb.cgi?a=commitdiff_plain;h=40958cbc99827ff63ba60f5b97c00104ecb47661;p=thirdparty%2Frspamd.git

[Feature] Add support of words regexps
---

diff --git a/src/libmime/mime_expressions.c b/src/libmime/mime_expressions.c
index 596d959c6e..c6d258c494 100644
--- a/src/libmime/mime_expressions.c
+++ b/src/libmime/mime_expressions.c
@@ -239,6 +239,10 @@ rspamd_parse_long_option (const gchar *start, gsize len,
 		ret = TRUE;
 		a->type = RSPAMD_RE_SARAWBODY;
 	}
+	else if (TYPE_CHECK (start, "words", len)) {
+		ret = TRUE;
+		a->type = RSPAMD_RE_WORDS;
+	}
 	else if (TYPE_CHECK (start, "selector", len)) {
 		ret = TRUE;
 		a->type = RSPAMD_RE_SELECTOR;
diff --git a/src/libserver/re_cache.c b/src/libserver/re_cache.c
index 64f53773d2..7b7cabb696 100644
--- a/src/libserver/re_cache.c
+++ b/src/libserver/re_cache.c
@@ -23,6 +23,7 @@
 #include "libutil/util.h"
 #include "libutil/regexp.h"
 #include "lua/lua_common.h"
+#include "libstat/stat_api.h"
 
 #include "khash.h"
 
@@ -1199,6 +1200,46 @@ rspamd_re_cache_exec_re (struct rspamd_task *task,
 			g_free (lenvec);
 		}
 		break;
+	case RSPAMD_RE_WORDS:
+		if (task->text_parts->len > 0) {
+			cnt = 0;
+
+			PTR_ARRAY_FOREACH (task->text_parts, i, part) {
+				if (part->utf_words) {
+					cnt += part->utf_words->len;
+				}
+			}
+
+			if (cnt > 0) {
+				scvec = g_malloc (sizeof (*scvec) * cnt);
+				lenvec = g_malloc (sizeof (*lenvec) * cnt);
+
+				cnt = 0;
+
+				PTR_ARRAY_FOREACH (task->text_parts, i, part) {
+					guint j;
+					rspamd_stat_token_t *tok;
+
+
+					if (part->utf_words) {
+						for (j = 0; j < part->utf_words->len; j ++) {
+							tok = &g_array_index (part->utf_words, rspamd_stat_token_t, j);
+							scvec[cnt] = tok->begin;
+							lenvec[cnt++] = tok->len;
+						}
+					}
+				}
+
+				ret = rspamd_re_cache_process_regexp_data (rt, re,
+						task, scvec, lenvec, cnt, TRUE);
+
+				msg_debug_re_task ("checking sa words regexp: %s -> %d",
+						rspamd_regexp_get_pattern (re), ret);
+				g_free (scvec);
+				g_free (lenvec);
+			}
+		}
+		break;
 	case RSPAMD_RE_SELECTOR:
 		if (rspamd_re_cache_process_selector (task, rt,
 				re_class->type_data,
diff --git a/src/libserver/re_cache.h b/src/libserver/re_cache.h
index c14b29ef0d..596ea08c27 100644
--- a/src/libserver/re_cache.h
+++ b/src/libserver/re_cache.h
@@ -35,6 +35,7 @@ enum rspamd_re_type {
 	RSPAMD_RE_BODY, /* full in SA */
 	RSPAMD_RE_SABODY, /* body in SA */
 	RSPAMD_RE_SARAWBODY, /* rawbody in SA */
+	RSPAMD_RE_WORDS, /* normalized words */
 	RSPAMD_RE_SELECTOR, /* use lua selector to process regexp */
 	RSPAMD_RE_MAX
 };