added identify data api

Author: rootware

pyredactkit/core_redactor.py

@@ -35,6 +35,22 @@ def __init__(self) -> None:
         """
         return None
 
+    def identify_data(self, line: str) -> list:
+        """Function to identify specific option
+        Args:
+            line (str) : line to be supplied to identify pattern
+        Returns:
+            list (list): list of sensitive data found in lines
+        """
+        sensitive_data = []
+        for id in id_object.regexes:
+            redact_pattern = id['pattern']
+            if re.search(redact_pattern, line):
+                pattern_string = re.search(redact_pattern, line)
+                pattern_string = pattern_string.group(0)
+                sensitive_data.append(pattern_string)
+        return sensitive_data
+
     def redact_all(self, line: str) -> tuple:
         """Function to redact specific option
         Args:

pyredactkit/pyredactkit.py

@@ -143,5 +143,9 @@ def main():
         redact_obj.process_text(args.text)
 
 
+def api_return_core_identifier(text: str) -> list:
+    return redact_obj.identify_data(text)
+
+
 if __name__ == "__main__":
     main()

tests/test_common_jobs.py

@@ -1,5 +1,4 @@
 import pytest
-import json
 import os
 from pyredactkit.common_jobs import CommonJobs
 

tests/test_core_redact.py

@@ -60,6 +60,10 @@ def mocker_text_file(mocker):
     mocker.patch(builtin_open, mocked_open)
 
 
+def test_identify_data_should_return_list(redactor_obj):
+    assert type(redactor_obj.identify_data(data)) == list, "identify_data function should return a list"
+
+
 def test_redact_all_function_should_return_string_and_dictionary(redactor_obj):
     set1 = redactor_obj.redact_all(data)
     set2 = ("This is a string", hash_table)

tests/test_custom_redact_engine.py

@@ -58,6 +58,10 @@ def mocker_text_file(mocker):
     mocker.patch(builtin_open, mocked_open)
 
 
+# def test_write_hashmap_should_create_json_file(custom_redactor, tmp_path):
+#     hash_map = {"key": "value"}
+#     custom_redactor.write_hashmap(hash_map, filename='fakefile', savedir=tmp_path)
+
 # def test_redact_custom_function_should_return_string_and_dictionary(custom_redactor):
 #     set1 = custom_redactor.redact_custom(data)
 #     set2 = ("This is a string", hash_table)