Merge pull request #309 from bshifter/filterx-invalid-utf8-fix

Filterx invalid utf8 fix for unset_empties
axoflow · Sep 27, 2024 · d53d84f · d53d84f
2 parents 116def9 + 72321d3
commit d53d84f
Show file tree

Hide file tree

Showing 2 changed files with 35 additions and 0 deletions.
diff --git a/lib/filterx/func-unset-empties.c b/lib/filterx/func-unset-empties.c
@@ -89,6 +89,8 @@ static gboolean _should_unset_string(FilterXFunctionUnsetEmpties *self, FilterXO
   if (check_flag(self->flags, FILTERX_FUNC_UNSET_EMPTIES_FLAG_REPLACE_EMPTY_STRING) && (str_len == 0))
     return TRUE;
 
+  if (!g_utf8_validate(str, -1, NULL))
+    return FALSE;
   if (check_flag(self->flags, FILTERX_FUNC_UNSET_EMPTIES_FLAG_IGNORECASE))
     {
       casefold_str = g_utf8_casefold(str, str_len);
@@ -380,6 +382,12 @@ _handle_target_object(FilterXFunctionUnsetEmpties *self, FilterXObject *target,
           set_flag(&self->flags, FILTERX_FUNC_UNSET_EMPTIES_FLAG_REPLACE_EMPTY_STRING, TRUE);
           return TRUE;
         }
+      if (!g_utf8_validate(str, -1, NULL))
+        {
+          g_set_error(error, FILTERX_FUNCTION_ERROR, FILTERX_FUNCTION_ERROR_CTOR_FAIL,
+                      FILTERX_FUNC_UNSET_EMPTIES_ARG_NAME_TARGETS" strings must be valid utf8 strings! " FILTERX_FUNC_UNSET_EMPTIES_USAGE);
+          return FALSE;
+        }
       if (check_flag(self->flags, FILTERX_FUNC_UNSET_EMPTIES_FLAG_IGNORECASE))
         {
           g_ptr_array_add(self->targets, g_utf8_casefold(str, len));

diff --git a/tests/light/functional_tests/filterx/test_filterx.py b/tests/light/functional_tests/filterx/test_filterx.py
@@ -1516,6 +1516,33 @@ def test_vars(config, syslog_ng):
     assert file_true.read_log() == '{"logmsg_variable":"foo","pipeline_level_variable":"baz","log":{"body":"foobar","attributes":{"attribute":42}},"js_array":[1,2,3,[4,5,6]]}\n'
 
 
+def test_unset_empties_invalid_utf8(config, syslog_ng):
+    (file_true, file_false) = create_config(
+        config, r"""
+            invalid_utf8 = "ba\xC080az";
+            dict = {"foo":invalid_utf8};
+            unset_empties(dict, targets=["baz"]);
+            $MSG = dict;
+    """,
+    )
+    syslog_ng.start(config)
+
+    assert file_true.get_stats()["processed"] == 1
+    assert "processed" not in file_false.get_stats()
+
+
+def test_unset_empties_invalid_utf8_target(config, syslog_ng):
+    (file_true, file_false) = create_config(
+        config, r"""
+            dict = {};
+            unset_empties(dict, targets=["b\xC080az"], recursive=true);
+            $MSG = dict;
+    """,
+    )
+    with pytest.raises(Exception):
+        syslog_ng.start(config)
+
+
 def test_unset_empties(config, syslog_ng):
     (file_true, file_false) = create_config(
         config, r"""