Usage_list refactored, password for doc (#379)

* [auxiliary] 2023-07-13T10:46:39+03:00 * SourceType applied to unify usage 'doc' and 'src' * tmp * fix * import fix * codestyle * [skip actions] [auxiliary] 2023-07-13T15:20:57+03:00 * min_line_len=10 * fix flake8
Samsung · Jul 17, 2023 · 767546a · 767546a
1 parent 2b3e47d
commit 767546a
Show file tree

Hide file tree

Showing 26 changed files with 369 additions and 255 deletions.
diff --git a/credsweeper/app.py b/credsweeper/app.py
@@ -39,7 +39,7 @@ class CredSweeper:
     """
 
     def __init__(self,
-                 rule_path: Optional[str] = None,
+                 rule_path: Union[None, str, Path] = None,
                  config_path: Optional[str] = None,
                  api_validation: bool = False,
                  json_filename: Union[None, str, Path] = None,
@@ -97,9 +97,7 @@ def __init__(self,
                                             exclude_values=exclude_values)
         self.config = Config(config_dict)
         self.scanner = Scanner(self.config, rule_path)
-        self.doc_scanner = Scanner(self.config, rule_path, ["doc"])
         self.deep_scanner = DeepScanner(self.config, self.scanner)
-        self.deep_doc_scanner = DeepScanner(self.config, self.doc_scanner)
         self.credential_manager = CredentialManager()
         self.json_filename: Union[None, str, Path] = json_filename
         self.xlsx_filename: Union[None, str, Path] = xlsx_filename
@@ -305,12 +303,9 @@ def file_scan(self, content_provider: Union[DiffContentProvider, TextContentProv
             candidates.append(dummy_candidate)
 
         else:
-            if self.config.depth:
+            if self.config.depth or self.config.doc:
                 # deep scan with possible data representation
                 candidates = self.deep_scanner.scan(content_provider, self.config.depth, self.config.size_limit)
-            elif self.config.doc:
-                # document-specific scanning
-                candidates = self.deep_doc_scanner.scan(content_provider, 0, self.config.size_limit)
             else:
                 if content_provider.file_type not in self.config.exclude_containers:
                     # Regular file scanning

diff --git a/credsweeper/filters/group/group.py b/credsweeper/filters/group/group.py
@@ -35,7 +35,7 @@ def filters(self, filters: List[Filter]) -> None:
     @staticmethod
     def get_keyword_base_filters(config: Config) -> List[Filter]:
         """returns base filters"""
-        return [  #
+        filters = [  #
             SeparatorUnusualCheck(),
             ValueAllowlistCheck(),
             ValueArrayDictionaryCheck(),
@@ -46,13 +46,14 @@ def get_keyword_base_filters(config: Config) -> List[Filter]:
             ValueLastWordCheck(),
             ValueLengthCheck(config),
             ValueMethodCheck(),
-            ValueNotAllowedPatternCheck(),
             ValueSimilarityCheck(),
             ValueStringTypeCheck(config),
             ValueTokenCheck(),
             VariableNotAllowedPatternCheck(),
-            ValuePatternCheck(config)
         ]
+        if not config.doc:
+            filters.extend([ValuePatternCheck(config), ValueNotAllowedPatternCheck()])
+        return filters
 
     @staticmethod
     def get_pattern_base_filters(config: Config) -> List[Filter]:

diff --git a/credsweeper/filters/value_not_part_encoded_check.py b/credsweeper/filters/value_not_part_encoded_check.py
@@ -1,6 +1,5 @@
-from typing import Optional
-
 import re
+from typing import Optional
 
 from credsweeper.config import Config
 from credsweeper.credentials import LineData