[Feature] Allow to get all types of words from Lua

3 files changed, 163 insertions, 11 deletions

diff --git a/src/lua/lua_common.c b/src/lua/lua_common.c
index dcab21530..fe463b763 100644
--- a/src/lua/lua_common.c
+++ b/src/lua/lua_common.c
@@ -21,9 +21,12 @@
 #include "ottery.h"
 #include "rspamd_control.h"
 #include "lua_thread_pool.h"
+#include "libstat/stat_api.h"
+#include "libserver/rspamd_control.h"
+
 #include <math.h>
 #include <sys/wait.h>
-#include <src/libserver/rspamd_control.h>
+
 
 /* Lua module init function */
 #define MODULE_INIT_FUNC "module_init"
@@ -2403,4 +2406,117 @@ rspamd_lua_try_load_redis (lua_State *L, const ucl_object_t *obj,
 	}
 
 	return FALSE;
+}
+
+gint
+rspamd_lua_push_words (lua_State *L, GArray *words,
+							enum rspamd_lua_words_type how)
+{
+	rspamd_stat_token_t *w;
+	guint i, cnt, fl_cnt;
+
+	lua_createtable (L, words->len, 0);
+
+	for (i = 0, cnt = 1; i < words->len; i ++) {
+		w = &g_array_index (words, rspamd_stat_token_t, i);
+
+		switch (how) {
+		case RSPAMD_LUA_WORDS_STEM:
+			if (w->stemmed.len > 0) {
+				lua_pushlstring (L, w->stemmed.begin, w->stemmed.len);
+				lua_rawseti (L, -2, cnt ++);
+			}
+			break;
+		case RSPAMD_LUA_WORDS_NORM:
+			if (w->normalized.len > 0) {
+				lua_pushlstring (L, w->normalized.begin, w->normalized.len);
+				lua_rawseti (L, -2, cnt ++);
+			}
+			break;
+		case RSPAMD_LUA_WORDS_RAW:
+			if (w->original.len > 0) {
+				lua_pushlstring (L, w->original.begin, w->original.len);
+				lua_rawseti (L, -2, cnt ++);
+			}
+			break;
+		case RSPAMD_LUA_WORDS_FULL:
+			lua_createtable (L, 4, 0);
+
+			if (w->stemmed.len > 0) {
+				lua_pushlstring (L, w->stemmed.begin, w->stemmed.len);
+				lua_rawseti (L, -2, 1);
+			}
+			else {
+				lua_pushstring (L, "");
+				lua_rawseti (L, -2, 1);
+			}
+
+			if (w->normalized.len > 0) {
+				lua_pushlstring (L, w->normalized.begin, w->normalized.len);
+				lua_rawseti (L, -2, 2);
+			}
+			else {
+				lua_pushstring (L, "");
+				lua_rawseti (L, -2, 2);
+			}
+
+			if (w->original.len > 0) {
+				lua_pushlstring (L, w->original.begin, w->original.len);
+				lua_rawseti (L, -2, 3);
+			}
+			else {
+				lua_pushstring (L, "");
+				lua_rawseti (L, -2, 3);
+			}
+
+			/* Flags part */
+			fl_cnt = 1;
+			lua_createtable (L, 4, 0);
+
+			if (w->flags & RSPAMD_STAT_TOKEN_FLAG_NORMALISED) {
+				lua_pushstring (L, "normalised");
+				lua_rawseti (L, -2, fl_cnt ++);
+			}
+			if (w->flags & RSPAMD_STAT_TOKEN_FLAG_BROKEN_UNICODE) {
+				lua_pushstring (L, "broken_unicode");
+				lua_rawseti (L, -2, fl_cnt ++);
+			}
+			if (w->flags & RSPAMD_STAT_TOKEN_FLAG_UTF) {
+				lua_pushstring (L, "utf");
+				lua_rawseti (L, -2, fl_cnt ++);
+			}
+			if (w->flags & RSPAMD_STAT_TOKEN_FLAG_TEXT) {
+				lua_pushstring (L, "text");
+				lua_rawseti (L, -2, fl_cnt ++);
+			}
+			if (w->flags & RSPAMD_STAT_TOKEN_FLAG_HEADER) {
+				lua_pushstring (L, "header");
+				lua_rawseti (L, -2, fl_cnt ++);
+			}
+			if (w->flags & (RSPAMD_STAT_TOKEN_FLAG_META|RSPAMD_STAT_TOKEN_FLAG_LUA_META)) {
+				lua_pushstring (L, "meta");
+				lua_rawseti (L, -2, fl_cnt ++);
+			}
+			if (w->flags & RSPAMD_STAT_TOKEN_FLAG_STOP_WORD) {
+				lua_pushstring (L, "stop_word");
+				lua_rawseti (L, -2, fl_cnt ++);
+			}
+			if (w->flags & RSPAMD_STAT_TOKEN_FLAG_INVISIBLE_SPACES) {
+				lua_pushstring (L, "invisible_spaces");
+				lua_rawseti (L, -2, fl_cnt ++);
+			}
+			if (w->flags & RSPAMD_STAT_TOKEN_FLAG_STEMMED) {
+				lua_pushstring (L, "stemmed");
+				lua_rawseti (L, -2, fl_cnt ++);
+			}
+
+			lua_rawseti (L, -2, 4);
+
+			/* Push to the resulting vector */
+			lua_rawseti (L, -2, cnt ++);
+			break;
+		}
+
+		return 1;
+	}
 }
\ No newline at end of file
diff --git a/src/lua/lua_common.h b/src/lua/lua_common.h
index 2dee888ee..25f5b7ff4 100644
--- a/src/lua/lua_common.h
+++ b/src/lua/lua_common.h
@@ -433,6 +433,22 @@ gboolean rspamd_lua_require_function (lua_State *L, const gchar *modname,
 gboolean rspamd_lua_try_load_redis (lua_State *L, const ucl_object_t *obj,
 		struct rspamd_config *cfg, gint *ref_id);
 
+
+enum rspamd_lua_words_type {
+	RSPAMD_LUA_WORDS_STEM = 0,
+	RSPAMD_LUA_WORDS_NORM,
+	RSPAMD_LUA_WORDS_RAW,
+	RSPAMD_LUA_WORDS_FULL
+};
+/**
+ * Pushes words (rspamd_stat_token_t) to Lua
+ * @param L
+ * @param words
+ * @param how
+ */
+gint rspamd_lua_push_words (lua_State *L, GArray *words,
+		enum rspamd_lua_words_type how);
+
 /* Paths defs */
 #define RSPAMD_CONFDIR_INDEX "CONFDIR"
 #define RSPAMD_LOCAL_CONFDIR_INDEX "LOCAL_CONFDIR"
diff --git a/src/lua/lua_mimepart.c b/src/lua/lua_mimepart.c
index d2ff7e8e4..14111f760 100644
--- a/src/lua/lua_mimepart.c
+++ b/src/lua/lua_mimepart.c
@@ -132,8 +132,16 @@ LUA_FUNCTION_DEF (textpart, get_stats);
 LUA_FUNCTION_DEF (textpart, get_words_count);
 
 /***
- * @method mime_part:get_words()
- * Get words in the part
+ * @method mime_part:get_words([how])
+ * Get words in the part. Optional `how` argument defines type of words returned:
+ * - `stem`: stemmed words (default)
+ * - `norm`: normalised words (utf normalised + lowercased)
+ * - `raw`: raw words in utf (if possible)
+ * - `full`: list of tables, each table has the following fields:
+ *   - [1] - stemmed word
+ *   - [2] - normalised word
+ *   - [3] - raw word
+ *   - [4] - flags (table of strings)
  * @return {table/strings} words in the part
  */
 LUA_FUNCTION_DEF (textpart, get_words);
@@ -759,8 +767,7 @@ lua_textpart_get_words (lua_State *L)
 {
 	LUA_TRACE_POINT;
 	struct rspamd_mime_text_part *part = lua_check_textpart (L);
-	rspamd_stat_token_t *w;
-	guint i;
+	enum rspamd_lua_words_type how = RSPAMD_LUA_WORDS_STEM;
 
 	if (part == NULL) {
 		return luaL_error (L, "invalid arguments");
@@ -770,14 +777,27 @@ lua_textpart_get_words (lua_State *L)
 		lua_createtable (L, 0, 0);
 	}
 	else {
-		lua_createtable (L, part->utf_words->len, 0);
-
-		for (i = 0; i < part->utf_words->len; i ++) {
-			w = &g_array_index (part->utf_words, rspamd_stat_token_t, i);
+		if (lua_type (L, 2) == LUA_TSTRING) {
+			const gchar *how_str = lua_tostring (L, 2);
 
-			lua_pushlstring (L, w->stemmed.begin, w->stemmed.len);
-			lua_rawseti (L, -2, i + 1);
+			if (strcmp (how_str, "stem") == 0) {
+				how = RSPAMD_LUA_WORDS_STEM;
+			}
+			else if (strcmp (how_str, "norm") == 0) {
+				how = RSPAMD_LUA_WORDS_NORM;
+			}
+			else if (strcmp (how_str, "raw") == 0) {
+				how = RSPAMD_LUA_WORDS_RAW;
+			}
+			else if (strcmp (how_str, "full") == 0) {
+				how = RSPAMD_LUA_WORDS_FULL;
+			}
+			else {
+				return luaL_error (L, "unknown words type: %s", how_str);
+			}
 		}
+
+		return rspamd_lua_push_words (L, part->utf_words, how);
 	}
 
 	return 1;