ml info

credsweeper/app.py

@@ -10,7 +10,7 @@
 # Directory of credsweeper sources MUST be placed before imports to avoid circular import error
 APP_PATH = Path(__file__).resolve().parent
 
-from credsweeper.common.constants import KeyValidationOption, Severity, ThresholdPreset, DiffRowType
+from credsweeper.common.constants import Severity, ThresholdPreset, DiffRowType
 from credsweeper.config import Config
 from credsweeper.credentials import Candidate, CredentialManager, CandidateKey
 from credsweeper.deep_scanner.deep_scanner import DeepScanner
@@ -368,11 +368,9 @@ def post_processing(self) -> None:
                     for candidate in group_candidates:
                         if candidate.use_ml:
                             if is_cred[i]:
-                                candidate.ml_validation = KeyValidationOption.VALIDATED_KEY
                                 candidate.ml_probability = probability[i]
                                 new_cred_list.append(candidate)
                         else:
-                            candidate.ml_validation = KeyValidationOption.NOT_AVAILABLE
                             new_cred_list.append(candidate)
             else:
                 logger.info("Skipping ML validation due not applicable")
@@ -435,7 +433,8 @@ def export_results(self, change_type: Optional[DiffRowType] = None) -> None:
                 for line_data in credential.line_data_list:
                     # bright rule name and path or info
                     print(Style.BRIGHT + credential.rule_name +
-                          f" {line_data.info or line_data.path}:{line_data.line_num}" + Style.RESET_ALL)
+                          f" {line_data.info or line_data.path}:{line_data.line_num} {credential.ml_info}" +
+                          Style.RESET_ALL)
                     print(line_data.get_colored_line(hashed=self.hashed, subtext=self.subtext))
 
         if is_exported is False:

credsweeper/common/constants.py

@@ -97,14 +97,6 @@ class Chars(Enum):
 ENTROPY_LIMIT_BASE3x = 3
 
 
-class KeyValidationOption(Enum):
-    """API validation state"""
-    INVALID_KEY = 0
-    VALIDATED_KEY = 1
-    UNDECIDED = 2
-    NOT_AVAILABLE = 3
-
-
 class GroupType(Enum):
     """Group type - used in Group constructor for load predefined set of filters"""
     KEYWORD = "keyword"

credsweeper/credentials/candidate.py

@@ -3,7 +3,7 @@
 from json.encoder import py_encode_basestring_ascii
 from typing import Any, Dict, List, Optional
 
-from credsweeper.common.constants import KeyValidationOption, Severity, Confidence
+from credsweeper.common.constants import Severity, Confidence
 from credsweeper.config import Config
 from credsweeper.credentials.line_data import LineData
 
@@ -39,7 +39,6 @@ def __init__(self,
         self.config = config
         self.use_ml = use_ml
         self.confidence = confidence
-        self.ml_validation = KeyValidationOption.NOT_AVAILABLE
         self.ml_probability: Optional[float] = None
 
     def compare(self, other: 'Candidate') -> bool:
@@ -48,7 +47,6 @@ def compare(self, other: 'Candidate') -> bool:
                 and self.severity == other.severity \
                 and self.confidence == other.confidence \
                 and self.use_ml == other.use_ml \
-                and self.ml_validation == other.ml_validation \
                 and self.ml_probability == other.ml_probability \
                 and len(self.line_data_list) == len(other.line_data_list):
             for i, j in zip(self.line_data_list, other.line_data_list):
@@ -79,7 +77,7 @@ def to_str(self, subtext: bool = False, hashed: bool = False) -> str:
                f" | severity: {self.severity.value}" \
                f" | confidence: {self.confidence.value}" \
                f" | line_data_list: [{', '.join([x.to_str(subtext, hashed) for x in self.line_data_list])}]" \
-               f" | ml_validation: {self.ml_validation.name}"
+               f" | ml_validation: {self.ml_validation}"
 
     def __str__(self):
         return self.to_str()
@@ -95,7 +93,7 @@ def to_json(self, hashed: bool, subtext: bool) -> Dict:
 
         """
         full_output = {
-            "ml_validation": self.ml_validation.name,
+            "ml_validation": self.ml_validation,
             "patterns": [pattern.pattern for pattern in self.patterns],
             "ml_probability": self.ml_probability,
             "rule": self.rule_name,
@@ -139,3 +137,29 @@ def get_dummy_candidate(cls, config: Config, file_path: str, file_type: str, inf
             severity=Severity.INFO,  #
             config=config,  #
             confidence=Confidence.MODERATE)
+
+    @property
+    def ml_validation(self) -> str:
+        """Temporally replaced self.ml_validation"""
+        if not self.use_ml:
+            return "NOT_AVAILABLE"
+        elif isinstance(self.ml_probability, float):
+            return "VALIDATED_KEY"
+        elif self.ml_probability is None:
+            return "UNDECIDED"
+        else:
+            return "INVALID_KEY"
+
+    @property
+    def ml_info(self) -> str:
+        """Used to generate short info about ML of the candidate
+
+        Returns:
+            NA - Not applicable ML for the credential type
+            None - ML was not calculated
+            float - the probability
+        """
+        if not self.use_ml:
+            return "NA"
+        else:
+            return str(self.ml_probability)

tests/test_app.py

@@ -27,7 +27,7 @@ def setUp(self):
     @staticmethod
     def _m_credsweeper(args) -> Tuple[str, str]:
         with subprocess.Popen(
-            [sys.executable, "-m", "credsweeper", *args],  #
+                [sys.executable, "-m", "credsweeper", *args],  #
                 cwd=APP_PATH.parent,  #
                 stdout=subprocess.PIPE,  #
                 stderr=subprocess.PIPE) as proc:
@@ -179,14 +179,16 @@ def test_it_works_with_multiline_in_patch_p(self) -> None:
     # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # #
 
     def test_it_works_with_patch_color_p(self) -> None:
-        target_path = str(SAMPLES_PATH / "password.patch")
+        target_path = str(SAMPLES_PATH / "uuid-update.patch")
         _stdout, _stderr = self._m_credsweeper(["--diff_path", target_path, "--log", "silence", "--color"])
         output = " ".join(_stdout.split()[:-1])
         expected = """
-                    \x1b[1mPassword .changes/1.16.98.json:added:3\x1b[0m 
-                    "\x1b[94mpassword\x1b[0m"\x1b[92m:\x1b[0m "\x1b[93mdkajco1\x1b[0m"
-                    Added File Credentials: 1 Deleted File Credentials: 0 Time Elapsed:
-                    """
+                   \x1b[1mUUID uuid:added:1 NA\x1b[0m
+                   \x1b[93mbace4d19-fa7e-dead-beef-9129474bcd81\x1b[0m
+                   \x1b[1mUUID uuid:deleted:1 NA\x1b[0m
+                   \x1b[93mbace4d19-fa7e-beef-cafe-9129474bcd81\x1b[0m
+                   Added File Credentials: 1 Deleted File Credentials: 1 Time Elapsed:
+                   """
         expected = " ".join(expected.split())
         self.assertEqual(expected, output)