[Minor] Add methods to extract real charset from text parts

2 files changed, 36 insertions, 0 deletions

diff --git a/src/libmime/mime_encoding.c b/src/libmime/mime_encoding.c
index 8788403f3..c30cbe3e3 100644
--- a/src/libmime/mime_encoding.c
+++ b/src/libmime/mime_encoding.c
@@ -690,6 +690,7 @@ rspamd_mime_text_part_maybe_convert (struct rspamd_task *task,
 			}
 
 			checked = TRUE;
+			text_part->real_charset = charset;
 		}
 		else {
 			SET_PART_UTF (text_part);
@@ -697,6 +698,7 @@ rspamd_mime_text_part_maybe_convert (struct rspamd_task *task,
 			rspamd_mime_text_part_ucs_from_utf (task, text_part);
 			rspamd_mime_text_part_normalise (task, text_part);
 			rspamd_mime_text_part_maybe_renormalise (task, text_part);
+			text_part->real_charset = UTF8_CHARSET;
 
 			return;
 		}
@@ -710,6 +712,7 @@ rspamd_mime_text_part_maybe_convert (struct rspamd_task *task,
 					MIN (RSPAMD_CHARSET_MAX_CONTENT, part_content->len));
 			msg_info_task ("detected charset: %s", charset);
 			checked = TRUE;
+			text_part->real_charset = charset;
 		}
 	}
 
@@ -730,6 +733,7 @@ rspamd_mime_text_part_maybe_convert (struct rspamd_task *task,
 		rspamd_mime_text_part_ucs_from_utf (task, text_part);
 		rspamd_mime_text_part_normalise (task, text_part);
 		rspamd_mime_text_part_maybe_renormalise (task, text_part);
+		text_part->real_charset = UTF8_CHARSET;
 
 		return;
 	}
@@ -748,6 +752,8 @@ rspamd_mime_text_part_maybe_convert (struct rspamd_task *task,
 			text_part->utf_raw_content = part_content;
 			return;
 		}
+
+		text_part->real_charset = charset;
 	}
 
 	SET_PART_UTF (text_part);
diff --git a/src/lua/lua_mimepart.c b/src/lua/lua_mimepart.c
index 33d1a3011..5a1bd0860 100644
--- a/src/lua/lua_mimepart.c
+++ b/src/lua/lua_mimepart.c
@@ -162,6 +162,13 @@ LUA_FUNCTION_DEF (textpart, get_html);
  * @return {string} short abbreviation (such as `ru`) for the script's language
  */
 LUA_FUNCTION_DEF (textpart, get_language);
+
+/***
+ * @method text_part:get_charset()
+ * Returns part real charset
+ * @return {string} charset of the part
+ */
+LUA_FUNCTION_DEF (textpart, get_charset);
 /***
  * @method text_part:get_languages()
  * Returns array of tables of all languages detected for a part:
@@ -205,6 +212,7 @@ static const struct luaL_reg textpartlib_m[] = {
 	LUA_INTERFACE_DEF (textpart, is_html),
 	LUA_INTERFACE_DEF (textpart, get_html),
 	LUA_INTERFACE_DEF (textpart, get_language),
+	LUA_INTERFACE_DEF (textpart, get_charset),
 	LUA_INTERFACE_DEF (textpart, get_languages),
 	LUA_INTERFACE_DEF (textpart, get_mimepart),
 	LUA_INTERFACE_DEF (textpart, get_stats),
@@ -842,6 +850,28 @@ lua_textpart_get_language (lua_State * L)
 }
 
 static gint
+lua_textpart_get_charset (lua_State * L)
+{
+	LUA_TRACE_POINT;
+	struct rspamd_mime_text_part *part = lua_check_textpart (L);
+
+	if (part != NULL) {
+		if (part->real_charset != NULL) {
+			lua_pushstring (L, part->real_charset);
+			return 1;
+		}
+		else {
+			lua_pushnil (L);
+		}
+	}
+	else {
+		return luaL_error (L, "invalid arguments");
+	}
+
+	return 1;
+}
+
+static gint
 lua_textpart_get_languages (lua_State * L)
 {
 	LUA_TRACE_POINT;