credsweeper 1.12.1__py3-none-any.whl → 1.13.3__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Potentially problematic release.

This version of credsweeper might be problematic. Click here for more details.

Files changed (51) hide show
  1. credsweeper/__init__.py +1 -1
  2. credsweeper/__main__.py +23 -13
  3. credsweeper/app.py +7 -2
  4. credsweeper/common/keyword_pattern.py +6 -3
  5. credsweeper/common/morpheme_checklist.txt +26 -6
  6. credsweeper/config/config.py +1 -0
  7. credsweeper/credentials/line_data.py +21 -6
  8. credsweeper/deep_scanner/abstract_scanner.py +1 -0
  9. credsweeper/deep_scanner/csv_scanner.py +71 -0
  10. credsweeper/deep_scanner/deep_scanner.py +19 -9
  11. credsweeper/deep_scanner/jks_scanner.py +11 -2
  12. credsweeper/deep_scanner/pkcs_scanner.py +4 -0
  13. credsweeper/deep_scanner/rtf_scanner.py +41 -0
  14. credsweeper/deep_scanner/strings_scanner.py +52 -0
  15. credsweeper/file_handler/byte_content_provider.py +10 -1
  16. credsweeper/file_handler/file_path_extractor.py +2 -0
  17. credsweeper/file_handler/text_content_provider.py +7 -1
  18. credsweeper/filters/__init__.py +1 -1
  19. credsweeper/filters/group/token_pattern.py +2 -2
  20. credsweeper/filters/group/weird_base36_token.py +3 -3
  21. credsweeper/filters/group/weird_base64_token.py +2 -2
  22. credsweeper/filters/value_camel_case_check.py +2 -2
  23. credsweeper/filters/value_file_path_check.py +5 -3
  24. credsweeper/filters/value_github_check.py +3 -2
  25. credsweeper/filters/value_morphemes_check.py +43 -0
  26. credsweeper/filters/value_string_type_check.py +1 -0
  27. credsweeper/ml_model/features/feature.py +1 -18
  28. credsweeper/ml_model/features/file_extension.py +1 -1
  29. credsweeper/ml_model/features/has_html_tag.py +10 -8
  30. credsweeper/ml_model/features/is_secret_numeric.py +4 -3
  31. credsweeper/ml_model/features/rule_name.py +1 -1
  32. credsweeper/ml_model/features/word_in.py +9 -32
  33. credsweeper/ml_model/features/word_in_path.py +2 -3
  34. credsweeper/ml_model/features/word_in_postamble.py +1 -4
  35. credsweeper/ml_model/features/word_in_preamble.py +1 -4
  36. credsweeper/ml_model/features/word_in_transition.py +1 -4
  37. credsweeper/ml_model/features/word_in_value.py +2 -3
  38. credsweeper/ml_model/features/word_in_variable.py +2 -3
  39. credsweeper/ml_model/ml_config.json +15 -8
  40. credsweeper/ml_model/ml_model.onnx +0 -0
  41. credsweeper/ml_model/ml_validator.py +1 -1
  42. credsweeper/rules/config.yaml +174 -207
  43. credsweeper/scanner/scanner.py +12 -7
  44. credsweeper/secret/config.json +18 -5
  45. credsweeper/utils/util.py +21 -18
  46. {credsweeper-1.12.1.dist-info → credsweeper-1.13.3.dist-info}/METADATA +7 -7
  47. {credsweeper-1.12.1.dist-info → credsweeper-1.13.3.dist-info}/RECORD +50 -47
  48. credsweeper/filters/value_couple_keyword_check.py +0 -28
  49. {credsweeper-1.12.1.dist-info → credsweeper-1.13.3.dist-info}/WHEEL +0 -0
  50. {credsweeper-1.12.1.dist-info → credsweeper-1.13.3.dist-info}/entry_points.txt +0 -0
  51. {credsweeper-1.12.1.dist-info → credsweeper-1.13.3.dist-info}/licenses/LICENSE +0 -0
credsweeper/ml_model/ml_config.json CHANGED
@@ -10,6 +10,7 @@
10
10
  "features": [
11
11
  {
12
12
  "type": "RuleSeverity",
13
+ "comment": "INFO=0.0, LOW=0.25, MEDIUM=0.5, HIGH=0.75, CRITICAL=1.0",
13
14
  "kwargs": {}
14
15
  },
15
16
  {
@@ -62,7 +63,7 @@
62
63
  "type": "SearchInAttribute",
63
64
  "comment": "Repeated symbol",
64
65
  "kwargs": {
65
- "pattern": ".*(?:(\\S)(\\S))((\\1.)|(.\\2)){7,}",
66
+ "pattern": "(?:(\\S)(\\S))((\\1.)|(.\\2)){7,}",
66
67
  "attribute": "value"
67
68
  }
68
69
  },
@@ -70,7 +71,7 @@
70
71
  "type": "SearchInAttribute",
71
72
  "comment": "SHA marker",
72
73
  "kwargs": {
73
- "pattern": ".*(?i:sha)[_-]?(224|256|384|512)",
74
+ "pattern": "(?i:sha)[_-]?(224|256|384|512)",
74
75
  "attribute": "value"
75
76
  }
76
77
  },
@@ -126,7 +127,7 @@
126
127
  "type": "SearchInAttribute",
127
128
  "comment": "VariableNotAllowedNameCheck",
128
129
  "kwargs": {
129
- "pattern": "(?i:(filters?|pub(lic)?)_?key)",
130
+ "pattern": "(?i:(sha[_-]?(224|256|384|512)|projects?|filters?|pub(lic)?)_?key)",
130
131
  "attribute": "variable"
131
132
  }
132
133
  },
@@ -134,7 +135,7 @@
134
135
  "type": "SearchInAttribute",
135
136
  "comment": "VariableNotAllowedNameCheck",
136
137
  "kwargs": {
137
- "pattern": "(?i:(id|size|name|type|manager|algorithm|view|error)$)",
138
+ "pattern": "(?i:(id|sum|size|name|type|manager|algorithm|pattern|view|error|date(time)?|time(stamp)?|tag|version|hash|rate)$)",
138
139
  "attribute": "variable"
139
140
  }
140
141
  },
@@ -245,8 +246,10 @@
245
246
  "crypt",
246
247
  "crypted",
247
248
  "decrypt",
249
+ "edited",
248
250
  "encrypt",
249
251
  "example",
252
+ "expire",
250
253
  "fake",
251
254
  "file",
252
255
  "foo",
@@ -260,7 +263,8 @@
260
263
  "pass",
261
264
  "public",
262
265
  "pwd",
263
- "rsa-",
266
+ "redacted",
267
+ "rsa",
264
268
  "salt",
265
269
  "secret",
266
270
  "sha",
@@ -339,6 +343,7 @@
339
343
  "get",
340
344
  "e.g.",
341
345
  "equal",
346
+ "env",
342
347
  "example",
343
348
  "expect",
344
349
  "line",
@@ -484,6 +489,7 @@
484
489
  ".bat",
485
490
  ".bats",
486
491
  ".bazel",
492
+ ".bin",
487
493
  ".build",
488
494
  ".bundle",
489
495
  ".bzl",
@@ -504,7 +510,6 @@
504
510
  ".csp",
505
511
  ".csv",
506
512
  ".dist",
507
- ".doc",
508
513
  ".dockerfile",
509
514
  ".edited",
510
515
  ".eex",
@@ -527,6 +532,8 @@
527
532
  ".gtpl",
528
533
  ".h",
529
534
  ".haml",
535
+ ".har",
536
+ ".hpp",
530
537
  ".hs",
531
538
  ".html",
532
539
  ".idl",
@@ -657,8 +664,8 @@
657
664
  "CMD Password",
658
665
  "CMD Secret",
659
666
  "CMD Token",
667
+ "CURL User Password",
660
668
  "Credential",
661
- "Github Old Token",
662
669
  "Key",
663
670
  "Nonce",
664
671
  "Password",
@@ -671,4 +678,4 @@
671
678
  }
672
679
  }
673
680
  ]
674
- }
681
+ }
credsweeper/ml_model/ml_model.onnx CHANGED
Binary file
credsweeper/ml_model/ml_validator.py CHANGED
@@ -272,7 +272,7 @@ class MlValidator:
272
272
  if head != tail:
273
273
  probability[head:tail] = self._batch_call_model(line_input_list, variable_input_list, value_input_list,
274
274
  features_list)
275
- is_cred = probability > self.threshold
275
+ is_cred = self.threshold <= probability
276
276
  if logger.isEnabledFor(logging.DEBUG):
277
277
  for i, decision in enumerate(is_cred):
278
278
  logger.debug("ML decision: %s with prediction: %s for value: %s", decision, probability[i],