v1.15.0

.github/workflows/check.yml

@@ -93,7 +93,7 @@ jobs:
         run: |
           banner="$(python -m credsweeper --banner | grep CredSweeper | head -1)"
           echo "banner = '${banner}'"
-          if [ "CredSweeper 1.14.8 crc32:8a4b3391" != "${banner}" ]; then
+          if [ "CredSweeper 1.15.0 crc32:439081df" != "${banner}" ]; then
             echo "Update the check for '${banner}'"
             exit 1
           fi

SECURITY.md

@@ -4,8 +4,8 @@
 
 | Version | Supported          |
 |---------|--------------------|
-| 1.14.x  | :white_check_mark: |
-| <1.14.x | :x:                |
+| 1.15.x  | :white_check_mark: |
+| <1.15.x | :x:                |
 
 ## Reporting a Vulnerability
 

credsweeper/__init__.py

@@ -24,4 +24,4 @@
     "__version__"
 ]
 
-__version__ = "1.14.8"
+__version__ = "1.15.0"

credsweeper/filters/value_blocklist_check.py

@@ -18,6 +18,7 @@ class ValueBlocklistCheck(Filter):
         "string",
         "value",
         "undefined",
+        "uuid",
     ]
 
     def __init__(self, config: Optional[Config] = None) -> None:

credsweeper/rules/config.yaml

@@ -1,16 +1,16 @@
 - name: DOC_GET
   severity: medium
-  confidence: weak
+  confidence: moderate
   type: pattern
   values:
-    - (?P<variable>(\w*(?i:비밀번호|비번|패스워드|키|암호화?|토큰|(?<!by)pass(?!e[dns]|ing|ion|age)|\bpwd?\b|token|secret|key|cred)\w*)\s*(설정은|[=:!]{1,3}))?\s*([._0-9A-Za-z\[\]]*get(env)?\s*\(\s*(?(variable)[^,]+|[\"'\\]*(\\*([\"']|&(quot|apos|#3[49]);)){0,4}(\w*(?i:(?<!by)pass(?!e[dns]|ing|ion|age|\s+[a-z]{3,80})|\bpwd?\b|token|secret|key|cred)\w*))(\\*([\"']|&(quot|apos|#3[49]);)){0,4})\s*(,(\s*default\s*=)?|\)\s*or)\s*([brufl@]{1,2}(?=\\*[\"'&]))?(?P<lq>(\\*([\"']|&(quot|apos|#3[49]);)){1,4})(?P<value>(.(?!(?P=lq))){4,80}.?)
+    - (?P<variable>(\w*(?i:비밀번호|비번|패스워드|키|암호화?|토큰|(?<!by)pass(?!e[dns]|ing|ion|age)|\bpwd?\b|token(?!ize)|secret|key(?!word|board|pad)|cred)\w*)\s*(설정은|[=:!]{1,3}))?\s*([._0-9A-Za-z\[\]]*get(env)?\s*\(\s*(?(variable)[^,]+|[\"'\\]*(\\*([\"']|&(quot|apos|#3[49]);)){0,4}(\w*(?i:(?<!by)pass(?!e[dns]|ing|ion|age|\s+[a-z]{3,64})|\bpwd?\b|token|secret|key|cred)\w*))(\\*([\"']|&(quot|apos|#3[49]);)){0,4})\s*(,(\s*default\s*=)?|\)\s*or)\s*([brufl@]{1,2}(?=\\*[\"'&]))?(?P<lq>(\\*([\"']|&(quot|apos|#3[49]);)){1,4})(?P<value>(.(?!(?P=lq))){4,8000}.?)
   filter_type:
     - ValueAllowlistCheck
+    - ValueBlocklistCheck
     - LineGitBinaryCheck
     - LineUUEPartCheck
     - ValueFilePathCheck
     - ValuePatternCheck(5)
-    - ValueLengthCheck(4,80)
   min_line_len: 8
   required_substrings:
     - pass
@@ -31,17 +31,17 @@
 
 - name: DOC_CREDENTIALS
   severity: medium
-  confidence: weak
+  confidence: moderate
   type: pattern
   values:
-    - (?P<wrap>[\"'`(])?\s*(?P<variable>(\w*(?i:(?<!by)passw?o?r?d?s?(?!e[dns]|ing|ion|age)|pwd?\b|\bp/w\b|token|secret|key|credential)\w*|비밀번호|비번|패스워드|키|암호화?|토큰))[\"'`]*(\s+(?i:is|are|was|were)(\s*[:-])?\s+|\s*(?P<separator>설정은|:=|:(?!:)|=(>|&gt;|(\\\\*u00|%)26gt;)|!==|!=|===|==|=~|=|%3[Dd])\s*)(?P<quote>[\"'`]{1,6})?(?P<value>(?(quote)(?(wrap)[^\"'`)]{4,80}|[^\"'`]{4,80})|(?(wrap)[^\"'`)]{4,80}|\S{4,80})))
+    - (?P<wrap>[\"'`(])?\s*(?P<variable>(\w*(?i:(?<!by)passw?o?r?d?s?(?!e[dns]|ing|ion|age)|pwd?\b|\bp/w\b|token(?!ize)|secret|key(?!word|board|pad)|credential)\w*|비밀번호|비번|패스워드|키|암호화?|토큰))[\"'`]*(\s+(?i:is|are|was|were)(\s*[:-])?\s+|\s*(?P<separator>설정은|:=|:(?!:)|=(>|&gt;|(\\\\*u00|%)26gt;)|!==|!=|===|==|=~|=|%3[Dd])\s*)(?P<quote>[\"'`]{1,6})?(?P<value>(?(quote)(?(wrap)[^\"'`)]{4,8000}|[^\"'`]{4,8000})|(?(wrap)[^\"'`)]{4,8000}|\S{4,8000})))
   filter_type:
     - ValueAllowlistCheck
+    - ValueBlocklistCheck
     - LineGitBinaryCheck
     - LineUUEPartCheck
     - ValueFilePathCheck
     - ValuePatternCheck(5)
-    - ValueLengthCheck(4,80)
   min_line_len: 8
   required_substrings:
     - pass

experiment/hyperparameters.py

@@ -1,10 +1,7 @@
 HP_DICT = {
     "line_lstm_dropout_rate": ((0.4, 0.5, 0.01), 0.47),
-    "line_lstm_recurrent_dropout_rate": ((0.0, 0.3, 0.01), 0.21),
     "variable_lstm_dropout_rate": ((0.4, 0.5, 0.01), 0.42),
-    "variable_lstm_recurrent_dropout_rate": ((0.0, 0.3, 0.01), 0.18),
     "value_lstm_dropout_rate": ((0.4, 0.5, 0.01), 0.47),
-    "value_lstm_recurrent_dropout_rate": ((0.0, 0.3, 0.01), 0.05),
     "dense_a_drop": ((0.0, 0.3, 0.01), 0.21),
     "dense_b_drop": ((0.0, 0.3, 0.01), 0.23),
 }

experiment/ml_model.py

@@ -38,35 +38,32 @@ def get_hyperparam(self, param_name: str, hp=None) -> Any:
     def build(self, hp: Optional[Any]) -> Model:
         """Get keras model with string and feature input and single binary out"""
         line_lstm_dropout_rate = self.get_hyperparam("line_lstm_dropout_rate", hp)
-        line_lstm_recurrent_dropout_rate = self.get_hyperparam("line_lstm_recurrent_dropout_rate", hp)
         variable_lstm_dropout_rate = self.get_hyperparam("variable_lstm_dropout_rate", hp)
-        variable_lstm_recurrent_dropout_rate = self.get_hyperparam("variable_lstm_recurrent_dropout_rate", hp)
         value_lstm_dropout_rate = self.get_hyperparam("value_lstm_dropout_rate", hp)
-        value_lstm_recurrent_dropout_rate = self.get_hyperparam("value_lstm_recurrent_dropout_rate", hp)
         dense_a_drop = self.get_hyperparam("dense_a_drop", hp)
         dense_b_drop = self.get_hyperparam("dense_b_drop", hp)
 
         line_input = Input(shape=(None, self.line_shape[2]), name="line_input", dtype=self.d_type)
         line_lstm = LSTM(units=self.line_shape[1],
                          dtype=self.d_type,
                          dropout=line_lstm_dropout_rate,
-                         recurrent_dropout=line_lstm_recurrent_dropout_rate)
+                         recurrent_dropout=0)
         line_bidirectional = Bidirectional(layer=line_lstm, name="line_bidirectional")
         line_lstm_branch = line_bidirectional(line_input)
 
         variable_input = Input(shape=(None, self.variable_shape[2]), name="variable_input", dtype=self.d_type)
         variable_lstm = LSTM(units=self.variable_shape[1],
                              dtype=self.d_type,
                              dropout=variable_lstm_dropout_rate,
-                             recurrent_dropout=variable_lstm_recurrent_dropout_rate)
+                             recurrent_dropout=0)
         variable_bidirectional = Bidirectional(layer=variable_lstm, name="variable_bidirectional")
         variable_lstm_branch = variable_bidirectional(variable_input)
 
         value_input = Input(shape=(None, self.value_shape[2]), name="value_input", dtype=self.d_type)
         value_lstm = LSTM(units=self.value_shape[1],
                           dtype=self.d_type,
                           dropout=value_lstm_dropout_rate,
-                          recurrent_dropout=value_lstm_recurrent_dropout_rate)
+                          recurrent_dropout=0)
         value_bidirectional = Bidirectional(layer=value_lstm, name="value_bidirectional")
         value_lstm_branch = value_bidirectional(value_input)
 

experiment/model_config_preprocess.py

@@ -13,7 +13,7 @@ def model_config_preprocess(df_all: pd.DataFrame, doc_target: bool) -> Dict[str,
     model_config = Util.json_load(ML_CONFIG_PATH)
     ascii_char_set = ''.join(chr(x) for x in range(0x20, 0x7F))
     extra_char_set = "\x1B\t\n\r"  # ESC code, tab and line end variations
-    doc_char_set = " ●개공기께내는님당드등로메밀번보복본비사생서석성슈스시암에용워으의이작정주지체큰키토패할호화" if doc_target else ''
+    doc_char_set = " ●가개공기께내는님당드등따로메면문밀방번보복본비사생서석성슈스시암에요용워으의이작정주지채체큰키토팅패필하할호화" if doc_target else ''
     model_config["char_set"] = extra_char_set + ascii_char_set + doc_char_set
 
     # check whether all extensions from meta are in ml_config.json

tests/__init__.py

@@ -9,17 +9,17 @@
 # float value of ML threshold is used to display possible lowest values
 ZERO_ML_THRESHOLD = 0.0
 
-# with option --doc & NEGLIGIBLE_ML_THRESHOLD
-SAMPLES_IN_DOC = 934
+# with option --doc & ZERO_ML_THRESHOLD
+SAMPLES_IN_DOC = 930
 
 # credentials count after scan without filters and ML validations
-SAMPLES_REGEX_COUNT = 670
+SAMPLES_REGEX_COUNT = 675
 
 # credentials count after scan with filters and without ML validation
-SAMPLES_FILTERED_COUNT = 554
+SAMPLES_FILTERED_COUNT = 556
 
 # credentials count after default post-processing
-SAMPLES_POST_CRED_COUNT = 508
+SAMPLES_POST_CRED_COUNT = 510
 
 # archived credentials that are not found without --depth
 SAMPLES_IN_DEEP_1 = SAMPLES_POST_CRED_COUNT + 138