PNLP-8004: mask numbers in bubbles and pronounceText (#177)

PNLP-8004: mask numbers in bubbles and pronounceText
(cherry picked from commit a996013)

core/message/from_message.py

@@ -11,10 +11,11 @@
 import core.logging.logger_constants as log_const
 from core.logging.logger_utils import log
 from core.utils.masking_message import masking
-from core.utils.utils import current_time_ms
+from core.utils.utils import current_time_ms, mask_numbers
 from core.message.msg_validator import MessageValidator
 
 from smart_kit.configs import get_app_config
+from smart_kit.configs import settings
 
 
 class Headers:
@@ -252,7 +253,9 @@ def generate_new_callback_id(self) -> str:
 
     @property
     def masked_value(self) -> str:
-        masked_data = masking(self.as_dict, self.masking_fields)
+        mask_numbers_flag = settings.Settings()["template_settings"].get("mask_numbers", False)
+        masked_data = mask_numbers(masking(self.as_dict, self.masking_fields)) if mask_numbers_flag else \
+            masking(self.as_dict, self.masking_fields)
         return json.dumps(masked_data, ensure_ascii=False)
 
     @property

core/utils/utils.py

@@ -1,4 +1,5 @@
 # coding=utf-8
+import copy
 import datetime
 import gc
 import json
@@ -8,7 +9,7 @@
 
 from collections import OrderedDict
 from math import isnan, isinf
-from typing import Optional
+from typing import Optional, Any, Dict
 from time import time
 
 from scenarios.user.user_model import User
@@ -143,3 +144,15 @@ def deep_update_dict(original, update):
         elif isinstance(value, dict):
             deep_update_dict(value, update[key])
     return update
+
+
+def mask_numbers(message: Dict[str, Any]) -> Dict[str, Any]:
+    masked_message = copy.deepcopy(message)
+    items = masked_message.get("payload", {}).get("items", [])
+    for item in items:
+        if "bubble" in item and "text" in item["bubble"]:
+            item["bubble"]["text"] = re.sub(r"\d+(?:[.,]\d+)?", "*number*", item["bubble"]["text"])
+    pronounce_text = masked_message.get("payload", {}).get("pronounceText")
+    if pronounce_text is not None:
+        masked_message["payload"]["pronounceText"] = re.sub(r"\d+(?:[.,]\d+)?", "*number*", pronounce_text)
+    return masked_message

smart_kit/message/smartapp_to_message.py

@@ -6,6 +6,8 @@
 from copy import copy
 
 from core.utils.masking_message import masking
+from core.utils.utils import mask_numbers
+from smart_kit.configs import settings
 from smart_kit.utils import SmartAppToMessage_pb2
 
 if TYPE_CHECKING:
@@ -69,7 +71,9 @@ def as_protobuf_message(data_as_dict):
 
     @cached_property
     def masked_value(self):
-        masked_data = masking(self.as_dict, self.masking_fields)
+        mask_numbers_flag = settings.Settings()["template_settings"].get("mask_numbers", False)
+        masked_data = mask_numbers(masking(self.as_dict, self.masking_fields)) if mask_numbers_flag else \
+            masking(self.as_dict, self.masking_fields)
         if self.command.loader == "json.dumps":
             return json.dumps(masked_data, ensure_ascii=False)
         elif self.command.loader == "protobuf":

smart_kit/template/static/configs/template_config.yml

@@ -21,3 +21,4 @@ self_service_with_state_save_messages: true
 project_id: template-app-id
 consumer_topic: "app"
 kafka_message_key_recovery_log_level: "DEBUG"
+mask_numbers: false

tests/core_tests/test_utils/test_mask_numbers.py

@@ -0,0 +1,37 @@
+from unittest import TestCase
+
+from core.utils.utils import mask_numbers
+
+
+class TestMaskNumbers(TestCase):
+    def test_1(self):
+        message = {
+            "payload": {
+                "pronounceText": "номер телефона: +79990000000",
+                "items": [
+                    {
+                        "bubble": {
+                            "text": "номер телефона: +7 (999) 000-00-00"
+                        }
+                    }
+                ]
+            }
+        }
+
+        masked_message = mask_numbers(message)
+        self.assertEqual(masked_message["payload"]["pronounceText"], "номер телефона: +*number*")
+        self.assertEqual(masked_message["payload"]["items"][0]["bubble"]["text"],
+                         "номер телефона: +*number* (*number*) *number*-*number*-*number*")
+
+        self.assertEqual(message, {
+            "payload": {
+                "pronounceText": "номер телефона: +79990000000",
+                "items": [
+                    {
+                        "bubble": {
+                            "text": "номер телефона: +7 (999) 000-00-00"
+                        }
+                    }
+                ]
+            }
+        })