[Minor] Rewrite is_utf_mixed_script to use libicu only

Rewrite to use U8_NEXT

1 files changed, 22 insertions, 0 deletions

diff --git a/test/lua/unit/rspamd_util.lua b/test/lua/unit/rspamd_util.lua
index 802b400d2..859316be7 100644
--- a/test/lua/unit/rspamd_util.lua
+++ b/test/lua/unit/rspamd_util.lua
@@ -5,36 +5,42 @@ context("Rspamd util for lua - check generic functions", function()
         {
             input = "test1",
             result = false,
+            mixed_script = false,
             range_start = 0x0000,
             range_end = 0x017f
         },
         {
             input = "test test xxx",
             result = false,
+            mixed_script = false,
             range_start = 0x0000,
             range_end = 0x017f
         },
         {
             input = "АбЫрвАлг",
             result = true,
+            mixed_script = false,
             range_start = 0x0000,
             range_end = 0x017f
         },
         {
             input = "АбЫрвАлг example",
             result = true,
+            mixed_script = true,
             range_start = 0x0000,
             range_end = 0x017f
         },
         {
             input = "example ąłśćżłóę",
             result = false,
+            mixed_script = false,
             range_start = 0x0000,
             range_end = 0x017f
         },
         {
             input = "ąłśćżłóę АбЫрвАлг",
             result = true,
+            mixed_script = true,
             range_start = 0x0000,
             range_end = 0x017f
         },
@@ -64,4 +70,20 @@ context("Rspamd util for lua - check generic functions", function()
         assert_equal(res["letters"], 10)
         assert_equal(res["digits"], 2)
     end)
+
+    for i,c in ipairs(cases) do
+        test("is_utf_mixed_script, test case #" .. i, function()
+          local actual = util.is_utf_mixed_script(c.input)
+
+          assert_equal(c.mixed_script, actual)
+        end)
+    end
+
+    test("is_utf_mixed_script, invalid utf str should return errror", function()
+        assert_error(util.is_utf_mixed_script,'\200\213\202')
+    end)
+
+    test("is_utf_mixed_script, empty str should return errror", function()
+        assert_error(util.is_utf_mixed_script,'\200\213\202')
+    end)
 end)