ui,pop-ups: fix duplicated a-z class in auto-generated rule names

ui/opensnitch/dialogs/prompt/utils.py

@@ -1,5 +1,6 @@
 from slugify import slugify
 import os
+import re
 import ipaddress
 
 from PyQt6.QtCore import QCoreApplication as QC
@@ -23,7 +24,20 @@ def get_rule_name(rule, is_list):
         rule_temp_name = "%s-list" % rule_temp_name
     else:
         rule_temp_name = "%s-simple" % rule_temp_name
-    rule_temp_name = slugify("%s %s" % (rule_temp_name, rule.operator.data))
+
+    rule_data = rule.operator.data
+    if rule.operator.type == Config.RULE_TYPE_REGEXP:
+        # Alnum character classes (e.g. [0-9A-Za-z]) don't carry any
+        # identifying info for a rule name, and slugify() lowercases
+        # everything, so a mixed-case range like [0-9A-Za-z] turns into
+        # "0-9a-za-z", which looks like a duplicated a-z class. Strip them
+        # out instead of slugifying them (#1587).
+        # Only match classes made up of digits/letters/hyphens, so we don't
+        # touch escaped brackets or other regexp syntax a user may have
+        # typed by hand (e.g. r'\[test\]' or r'[a-z\]]').
+        rule_data = re.sub(r'\[[0-9A-Za-z\-]+\]', '', rule_data)
+
+    rule_temp_name = slugify("%s %s" % (rule_temp_name, rule_data))
 
     return rule_temp_name[:128]
 

ui/tests/dialogs/test_prompt_utils.py

@@ -0,0 +1,52 @@
+#
+# pytest -v tests/dialogs/test_prompt_utils.py
+#
+
+import opensnitch.proto as proto
+ui_pb2, ui_pb2_grpc = proto.import_()
+
+from opensnitch.config import Config
+from opensnitch.dialogs.prompt import utils
+
+
+def _make_rule(data, rule_type=Config.RULE_TYPE_REGEXP):
+    rule = ui_pb2.Rule(name="user.choice")
+    rule.action = Config.ACTION_ALLOW
+    rule.duration = Config.DURATION_ONCE
+    rule.operator.type = rule_type
+    rule.operator.operand = Config.OPERAND_PROCESS_PATH
+    rule.operator.data = data
+    return rule
+
+
+class TestPromptUtils():
+
+    def test_get_rule_name_appimage_regexp_no_duplicated_class(self):
+        """ Regexps built for appimage/snap paths use a mixed-case character
+        class ([0-9A-Za-z]). slugify() lowercases everything, which used to
+        turn it into "0-9a-za-z" in the rule name, looking like a duplicated
+        a-z class (#1587).
+        """
+        rule = _make_rule(r'^/tmp/\.mount_handy_[0-9A-Za-z]+\/.*handy$')
+        name = utils.get_rule_name(rule, False)
+
+        assert "0-9a-za-z" not in name
+        assert "handy" in name
+
+    def test_get_rule_name_simple_rule_unaffected(self):
+        """ Non-regexp rules aren't touched by the character-class stripping.
+        """
+        rule = _make_rule("www.google.com", rule_type=Config.RULE_TYPE_SIMPLE)
+        name = utils.get_rule_name(rule, False)
+
+        assert name == "allow-once-simple-www-google-com"
+
+    def test_get_rule_name_escaped_brackets_not_mangled(self):
+        """ A hand-typed regexp matching a literal bracket must not be
+        treated as a character class and stripped.
+        """
+        rule = _make_rule(r'^/opt/\[legacy\]/app$')
+        name = utils.get_rule_name(rule, False)
+
+        assert "legacy" in name
+        assert "app" in name