SK-2838: fix redaction type in detokenize interface (#249)

Author: saileshwar-skyflow

README.md

@@ -239,8 +239,8 @@ from skyflow.utils.enums import RedactionType
 
 detokenize_request = DetokenizeRequest(
     data=[
-        {'token': 'token1', 'redaction': RedactionType.PLAIN_TEXT},
-        {'token': 'token2', 'redaction': RedactionType.PLAIN_TEXT}
+        {'token': 'token1', 'redaction_type': RedactionType.PLAIN_TEXT},
+        {'token': 'token2', 'redaction_type': RedactionType.PLAIN_TEXT}
     ],
     continue_on_error=True
 )

samples/vault_api/detokenize_records.py

@@ -55,11 +55,11 @@ def perform_detokenization():
         detokenize_data = [
             {
                 'token': '<TOKEN1>',                # Token to be detokenized
-                'redaction': RedactionType.REDACTED
+                'redaction_type': RedactionType.REDACTED
             },
             {
                 'token': '<TOKEN2>',                # Token to be detokenized
-                'redaction': RedactionType.MASKED
+                'redaction_type': RedactionType.MASKED
             }
         ]
 

skyflow/utils/_skyflow_messages.py

@@ -122,7 +122,7 @@ class Error(Enum):
         INVOKE_CONNECTION_FAILED = f"{error_prefix} Invoke Connection operation failed."
 
         INVALID_IDS_TYPE = f"{error_prefix} Validation error. 'ids' has a value of type {{}}. Specify 'ids' as list."
-        INVALID_REDACTION_TYPE = f"{error_prefix} Validation error. 'redaction' has a value of type {{}}. Specify 'redaction' as type Skyflow.RedactionType."
+        INVALID_REDACTION_TYPE = f"{error_prefix} Validation error. 'redaction_type' has a value of type {{}}. Specify 'redaction_type' as type Skyflow.RedactionType."
         INVALID_COLUMN_NAME = f"{error_prefix} Validation error. column_name has a value of type {{}}. Specify 'column' as a string."
         INVALID_COLUMN_VALUE = f"{error_prefix} Validation error. column_values key has a value of type {{}}. Specify column_values key as list."
         INVALID_COLUMN_VALUES = f"{error_prefix} Validation error. column_values key is an empty list. Specify at least one column value when column_name is passed." 
@@ -131,15 +131,15 @@ class Error(Enum):
         INVALID_OFF_SET_VALUE = f"{error_prefix} Validation error. offset key has a value of type {{}}. Specify offset key as integer."
         INVALID_LIMIT_VALUE = f"{error_prefix} Validation error. limit key has a value of type {{}}. Specify limit key as integer."
         INVALID_DOWNLOAD_URL_VALUE = f"{error_prefix} Validation error. download_url key has a value of type {{}}. Specify download_url key as boolean."
-        REDACTION_WITH_TOKENS_NOT_SUPPORTED = f"{error_prefix} Validation error. 'redaction' can't be used when tokens are specified. Remove 'redaction' from payload if tokens are specified."
+        REDACTION_WITH_TOKENS_NOT_SUPPORTED = f"{error_prefix} Validation error. 'redaction_type' can't be used when tokens are specified. Remove 'redaction_type' from payload if tokens are specified."
         TOKENS_GET_COLUMN_NOT_SUPPORTED = f"{error_prefix} Validation error. Column name and/or column values can't be used when tokens are specified. Remove unique column values or tokens from the payload."
         BOTH_IDS_AND_COLUMN_DETAILS_SPECIFIED = f"{error_prefix} Validation error. Both Skyflow IDs and column details can't be specified. Either specify Skyflow IDs or unique column details."
         INVALID_ORDER_BY_VALUE = f"{error_prefix} Validation error. order_by key has a value of type {{}}. Specify order_by key as Skyflow.OrderBy"
 
         UPDATE_FIELD_KEY_ERROR = f"{error_prefix} Validation error. Fields are empty in an update payload. Specify at least one field."
         INVALID_FIELDS_TYPE = f"{error_prefix} Validation error. The 'data' key has a value of type {{}}. Specify 'data' as a dictionary."
         IDS_KEY_ERROR = f"{error_prefix} Validation error. 'ids' key is missing from the payload. Specify an 'ids' key."
-        INVALID_TOKENS_LIST_VALUE = f"{error_prefix} Validation error. The 'data' field is invalid. Specify 'data' as a list of dictionaries containing 'token' and 'redaction'."
+        INVALID_TOKENS_LIST_VALUE = f"{error_prefix} Validation error. The 'data' field is invalid. Specify 'data' as a list of dictionaries containing 'token' and 'redaction_type'."
         INVALID_DATA_FOR_DETOKENIZE = f"{error_prefix}"
         EMPTY_TOKENS_LIST_VALUE = f"{error_prefix} Validation error. Tokens are empty in detokenize payload. Specify at lease one token"
         INVALID_TOKEN_TYPE = f"{ERROR}: [{error_prefix}] Invalid {{}} request. Tokens should be of type string."
@@ -417,6 +417,9 @@ class HttpStatus(Enum):
         BAD_REQUEST = "Bad Request"
 
     class Warning(Enum):
+        DETOKENIZE_REDACTION_KEY_DEPRECATED = (
+            f"{WARN}: [{error_prefix}] 'redaction' key in detokenize data is deprecated and will be removed in a future version. Use 'redaction_type' instead."
+        )
         UPDATE_LOG_LEVEL_DEPRECATED = (
             f"{WARN}: [{error_prefix}] Skyflow.update_log_level() is deprecated. "
             "Use Skyflow.set_log_level() instead."

skyflow/utils/constants.py

@@ -174,6 +174,7 @@ class RequestParameter:
     VALUE = 'value'
     COLUMN_GROUP = 'column_group'
     REDACTION = 'redaction'
+    REDACTION_TYPE = 'redaction_type'
 
 
 class FileUploadField:

skyflow/utils/validations/_validations.py

@@ -11,7 +11,7 @@
     FileUploadField,
     DeidentifyFileRequestField, RequestOperation, ConfigType, SqlCommand, ConfigField, OptionField, CredentialField, Detect
 )
-from skyflow.utils.logger import log_info, log_error_log
+from skyflow.utils.logger import log_info, log_warn, log_error_log
 from skyflow.vault.detect import DeidentifyTextRequest, ReidentifyTextRequest, TokenFormat, Transformations, \
     GetDetectRunRequest, Bleep, DeidentifyFileRequest
 from skyflow.vault.detect._file_input import FileInput
@@ -713,7 +713,17 @@ def validate_detokenize_request(logger, request):
                                invalid_input_error_code)
 
         token = item.get(ResponseField.TOKEN)
-        redaction = item.get(RequestParameter.REDACTION, None)
+
+        has_redaction = RequestParameter.REDACTION in item
+        has_redaction_type = RequestParameter.REDACTION_TYPE in item
+
+        if has_redaction:
+            log_warn(SkyflowMessages.Warning.DETOKENIZE_REDACTION_KEY_DEPRECATED.value, logger)
+
+        if has_redaction_type:
+            redaction = item.get(RequestParameter.REDACTION_TYPE)
+        else:
+            redaction = item.get(RequestParameter.REDACTION, None)
 
         if not isinstance(token, str) or not token:
             raise SkyflowError(SkyflowMessages.Error.INVALID_TOKEN_TYPE.value.format(RequestOperation.DETOKENIZE),

skyflow/vault/controller/_vault.py

@@ -223,7 +223,7 @@ def detokenize(self, request: DetokenizeRequest):
         tokens_list = [
             V1DetokenizeRecordRequest(
                 token=item.get(ResponseField.TOKEN),
-                redaction=item.get(RequestParameter.REDACTION, RedactionType.DEFAULT)
+                redaction=item.get(RequestParameter.REDACTION_TYPE) or item.get(RequestParameter.REDACTION, RedactionType.DEFAULT)
             )
             for item in request.data
         ]

tests/utils/validations/test__validations.py

@@ -1057,11 +1057,36 @@ def test_validate_detokenize_request_invalid_continue_on_error_type(self):
         self.assertEqual(context.exception.message, SkyflowMessages.Error.INVALID_CONTINUE_ON_ERROR_TYPE.value)
 
     def test_validate_detokenize_request_invalid_redaction_type(self):
-        request = DetokenizeRequest(data=[{"token": "token123", "redaction": "invalid"}], continue_on_error=False)
+        request = DetokenizeRequest(data=[{"token": "token123", "redaction_type": "invalid"}], continue_on_error=False)
         with self.assertRaises(SkyflowError) as context:
             validate_detokenize_request(self.logger, request)
         self.assertEqual(context.exception.message, SkyflowMessages.Error.INVALID_REDACTION_TYPE.value.format(str(type("invalid"))))
 
+    def test_validate_detokenize_request_deprecated_redaction_key_emits_warn(self):
+        from unittest.mock import patch
+        request = DetokenizeRequest(data=[{"token": "token123", "redaction": RedactionType.PLAIN_TEXT}], continue_on_error=False)
+        with patch('skyflow.utils.validations._validations.log_warn') as mock_warn:
+            validate_detokenize_request(self.logger, request)
+        mock_warn.assert_called_once()
+        self.assertIn("redaction_type", mock_warn.call_args[0][0])
+
+    def test_validate_detokenize_request_both_keys_prioritizes_redaction_type_and_warns(self):
+        from unittest.mock import patch
+        request = DetokenizeRequest(
+            data=[{"token": "token123", "redaction": RedactionType.PLAIN_TEXT, "redaction_type": RedactionType.MASKED}],
+            continue_on_error=False
+        )
+        with patch('skyflow.utils.validations._validations.log_warn') as mock_warn:
+            validate_detokenize_request(self.logger, request)
+        mock_warn.assert_called_once()
+
+    def test_validate_detokenize_request_redaction_type_only_no_warn(self):
+        from unittest.mock import patch
+        request = DetokenizeRequest(data=[{"token": "token123", "redaction_type": RedactionType.PLAIN_TEXT}], continue_on_error=False)
+        with patch('skyflow.utils.validations._validations.log_warn') as mock_warn:
+            validate_detokenize_request(self.logger, request)
+        mock_warn.assert_not_called()
+
 
     def test_validate_deidentify_file_request_wait_time_negative(self):
         file_input = FileInput(file_path=self.temp_file_path)