unit test sanitize_dict_before_log

cdci_data_analysis/flask_app/app.py

@@ -15,7 +15,7 @@
 import random
 import hashlib
 import validators
-
+import re
 import logging
 
 from raven.contrib.flask import Sentry
@@ -225,10 +225,13 @@ def remove_nested_keys(D, keys):
 
 def sanitize_dict_before_log(dict_to_sanitize):
     sensitive_keys = ['token']  # Add any other sensitive keys here
+    allowed_characters = r'[^A-Za-z0-9 ]'
+    replacement_character = ''
     sanitized_values = {}
     for key, value in dict_to_sanitize.items():
         if key not in sensitive_keys:
-            value = str(value).replace('\n', '').replace('\r', '')
+            # value = str(value).replace('\n', '').replace('\r', '')
+            value = re.sub(allowed_characters, replacement_character, str(value))
             sanitized_values[key] = value
     return sanitized_values
 

tests/test_server_basic.py

@@ -28,6 +28,7 @@
 from cdci_data_analysis.analysis.renku_helper import clone_renku_repo, checkout_branch_renku_repo, check_job_id_branch_is_present, get_repo_path, generate_commit_request_url, create_new_notebook_with_code, generate_nb_hash, create_renku_ini_config_obj, generate_ini_file_hash
 from cdci_data_analysis.analysis.drupal_helper import execute_drupal_request, get_drupal_request_headers, get_revnum, get_observations_for_time_range, generate_gallery_jwt_token, get_user_id, get_source_astrophysical_entity_id_by_source_name
 from cdci_data_analysis.plugins.dummy_plugin.data_server_dispatcher import DataServerQuery, ReturnProgressProductQuery
+from cdci_data_analysis.flask_app.app import sanitize_dict_before_log
 
 # logger
 logger = logging.getLogger(__name__)
@@ -70,6 +71,24 @@ def remove_args_from_dic(arg_dic, remove_keys):
     tem=0,
 )
 
+@pytest.mark.fast
+def test_sanitize_dict_before_log():
+
+    test_dict = {
+        'token': 'mytoken',
+        'field': 'myfield\n\r',
+        'username': 'myusername',
+        'email': '[email protected]'
+    }
+
+    expected_dict = {
+        'field': 'myfield',
+        'username': 'myusername',
+        'email': 'myemailexamplecom'
+    }
+
+    sanitized_dict = sanitize_dict_before_log(test_dict)
+    assert sanitized_dict == expected_dict
 
 @pytest.mark.fast
 def test_js9(dispatcher_live_fixture):