* Add ability to get difference between two parts from lua code

2 files changed, 50 insertions, 0 deletions

diff --git a/src/fuzzy.c b/src/fuzzy.c
index cdc9922cc..5a2decb34 100644
--- a/src/fuzzy.c
+++ b/src/fuzzy.c
@@ -321,6 +321,7 @@ fuzzy_init_part (struct mime_text_part *part, memory_pool_t *pool)
 	GList                          *cur_offset;
 	struct process_exception       *cur_ex = NULL;
 	gunichar                        uc;
+	GString *debug;
 
 	cur_offset = part->urls_offset;
 	if (cur_offset != NULL) {
@@ -370,6 +371,7 @@ fuzzy_init_part (struct mime_text_part *part, memory_pool_t *pool)
 		}
 	}
 
+	debug = g_string_sized_new (real_len);
 	new->block_size = fuzzy_blocksize (real_len);
 	new2->block_size = new->block_size * 2;
 
@@ -395,6 +397,7 @@ fuzzy_init_part (struct mime_text_part *part, memory_pool_t *pool)
 				uc = g_utf8_get_char (c);
 				if (g_unichar_isalnum (uc)) {
 					fuzzy_update2 (new, new2, uc);
+					g_string_append_unichar (debug, uc);
 				}
 				c = g_utf8_next_char (c);
 			}
@@ -412,11 +415,13 @@ fuzzy_init_part (struct mime_text_part *part, memory_pool_t *pool)
 			else {
 				if (!g_ascii_isspace (*c) && !g_ascii_ispunct (*c)) {
 					fuzzy_update2 (new, new2, *c);
+					g_string_append_c (debug, *c);
 				}
 				c++;
 			}
 		}
 	}
+	msg_info ("make hash of string: %v", debug);
 	/* Check whether we have more bytes in a rolling window */
 	if (new->rh != 0) {
 		new->hash_pipe[new->hi] = b64[new->h % 64];
diff --git a/src/lua/lua_task.c b/src/lua/lua_task.c
index 8c90feb4b..4916a50ad 100644
--- a/src/lua/lua_task.c
+++ b/src/lua/lua_task.c
@@ -110,6 +110,7 @@ LUA_FUNCTION_DEF (textpart, is_empty);
 LUA_FUNCTION_DEF (textpart, is_html);
 LUA_FUNCTION_DEF (textpart, get_fuzzy);
 LUA_FUNCTION_DEF (textpart, get_language);
+LUA_FUNCTION_DEF (textpart, compare_distance);
 
 static const struct luaL_reg    textpartlib_m[] = {
 	LUA_INTERFACE_DEF (textpart, get_content),
@@ -117,6 +118,7 @@ static const struct luaL_reg    textpartlib_m[] = {
 	LUA_INTERFACE_DEF (textpart, is_html),
 	LUA_INTERFACE_DEF (textpart, get_fuzzy),
 	LUA_INTERFACE_DEF (textpart, get_language),
+	LUA_INTERFACE_DEF (textpart, compare_distance),
 	{"__tostring", lua_class_tostring},
 	{NULL, NULL}
 };
@@ -1341,6 +1343,49 @@ lua_textpart_get_language (lua_State * L)
 	return 1;
 }
 
+static gint
+lua_textpart_compare_distance (lua_State * L)
+{
+	struct mime_text_part          *part = lua_check_textpart (L), *other;
+	void                           *ud = luaL_checkudata (L, 2, "rspamd{textpart}");
+	gint                            diff = -1;
+	GMimeObject                    *parent;
+	const GMimeContentType         *ct;
+
+	luaL_argcheck (L, ud != NULL, 2, "'textpart' expected");
+	other = *((struct mime_text_part **)ud);
+
+	if (part->parent && part->parent == other->parent) {
+		parent = part->parent;
+		ct = g_mime_object_get_content_type (parent);
+#ifndef GMIME24
+		if (ct == NULL || ! g_mime_content_type_is_type (ct, "multipart", "alternative")) {
+#else
+		if (ct == NULL || ! g_mime_content_type_is_type ((GMimeContentType *)ct, "multipart", "alternative")) {
+#endif
+			diff = -1;
+
+		}
+		else {
+			if (!part->is_empty && !other->is_empty) {
+				diff = fuzzy_compare_parts (part, other);
+			}
+			else if ((part->is_empty && !other->is_empty) || (!part->is_empty && other->is_empty)) {
+				/* Empty and non empty parts are different */
+				diff = 0;
+			}
+		}
+	}
+	else {
+			diff = -1;
+	}
+
+
+	lua_pushinteger (L, diff);
+
+	return 1;
+}
+
 /* Image functions */
 static gint
 lua_image_get_width (lua_State *L)