🗓 Jun 4, 2023 8:28:27 PM

✨ find_continuous_patterns
✨ find_longest_continious_pattern
🧪 tests added/updated
🚧 5.2.0
💚 build steps added/updated

.github/workflows/tests_multi_os.yml

@@ -18,9 +18,10 @@ jobs:
     steps:
     - uses: actions/checkout@v2.3.4
     - name: Set up Python ${{ matrix.python-version }}
-      uses: actions/setup-python@v2.2.1
+      uses: actions/setup-python@v4
       with:
         python-version: ${{ matrix.python-version }}
+        cache: 'pip'
     # - uses: actions/cache@v1
     #   id: devcache
     #   with:

chepy/__version__.py

@@ -1,2 +1,2 @@
-__version__ = "5.1.0"  # pragma: no cover
+__version__ = "5.2.0"  # pragma: no cover
 __author__ = "Hapsida @securisec"  # pragma: no cover

chepy/modules/extractors.py

@@ -1,4 +1,4 @@
-from typing import TypeVar
+from typing import TypeVar, Union
 from urllib.parse import urlparse as _pyurlparse
 
 import regex as re
@@ -44,7 +44,7 @@ def extract_hashes(self) -> ExtractorsT:
         return self
 
     @ChepyDecorators.call_stack
-    def extract_strings(self, length: int = 4, join_by: str='\n') -> ExtractorsT:
+    def extract_strings(self, length: int = 4, join_by: str = "\n") -> ExtractorsT:
         """Extract strings from state
 
         Args:
@@ -70,9 +70,7 @@ def extract_strings(self, length: int = 4, join_by: str='\n') -> ExtractorsT:
         return self
 
     @ChepyDecorators.call_stack
-    def extract_ips(
-        self, is_binary: bool = False
-    ) -> ExtractorsT:
+    def extract_ips(self, is_binary: bool = False) -> ExtractorsT:
         """Extract ipv4 and ipv6 addresses
 
         Args:
@@ -116,7 +114,10 @@ def extract_email(self, is_binary: bool = False) -> ExtractorsT:
         pattern = b"(^[a-zA-Z0-9_.+-]+@[a-zA-Z0-9-]+\.[a-zA-Z0-9-.]+$)"
         if is_binary:
             matched = list(
-                filter(lambda x: re.search(pattern, x), self.extract_strings().o.encode().splitlines())
+                filter(
+                    lambda x: re.search(pattern, x),
+                    self.extract_strings().o.encode().splitlines(),
+                )
             )
         else:  # pragma: no cover
             matched = list(
@@ -438,3 +439,62 @@ def extract_base64(self, min: int = 20) -> ExtractorsT:
         else:
             self.state = found[0]
         return self
+
+    @ChepyDecorators.call_stack
+    def find_continuous_patterns(self, str2: Union[str, bytes], min_value: int = 10):
+        """Find continius patterns between the state as a string and the provided str2
+
+        Args:
+            str2 (Union[str, bytes]): String to find matches against
+            min_value (int, optional): Minimum value of continious matches. Defaults to 10.
+
+        Returns:
+            Chepy: The Chepy object.
+        """
+        str1 = self._convert_to_bytes()
+        if isinstance(str2, str):
+            str2 = str2.encode()
+        combined_data = str1 + str2
+        data_length = len(combined_data)
+        patterns = []
+
+        for length in range(1, data_length + 1):
+            for start in range(data_length - length + 1):
+                pattern = combined_data[start : start + length]
+
+                if pattern in str1 and pattern in str2 and len(pattern) > min_value:
+                    patterns.append(pattern)
+
+        self.state = patterns
+        return self
+
+    @ChepyDecorators.call_stack
+    def find_longest_continious_pattern(self, str2: str):
+        """Find longest continious pattern
+
+        Args:
+            str2 (Union[str, bytes]): String to find match against
+
+        Returns:
+            Chepy: The Chepy object.
+        """
+        str1 = self._convert_to_bytes()
+        if isinstance(str2, str):
+            str2 = str2.encode()
+        combined_data = str1 + str2
+        data_length = len(combined_data)
+        matches = []
+
+        for length in range(1, data_length + 1):
+            for start in range(data_length - length + 1):
+                pattern = combined_data[start : start + length]
+
+                if (
+                    pattern in str1
+                    and pattern in str2
+                    and len(pattern) > len(matches[-1:])
+                ):
+                    matches.append(pattern)
+
+        self.state = max(matches, key=len) if matches else ""
+        return self

chepy/modules/extractors.pyi

@@ -1,5 +1,5 @@
 from ..core import ChepyCore
-from typing import Any, TypeVar
+from typing import Any, TypeVar, Union
 
 ExtractorsT = TypeVar('ExtractorsT', bound='Extractors')
 
@@ -34,3 +34,5 @@ class Extractors(ChepyCore):
     def extract_dsa_private(self: ExtractorsT) -> ExtractorsT: ...
     def extract_jwt_token(self: ExtractorsT) -> ExtractorsT: ...
     def extract_base64(self: ExtractorsT, min: int=...) -> ExtractorsT: ...
+    def find_continuous_patterns(self: ExtractorsT, str2: Union[str, bytes], min_value: int=...) -> ExtractorsT: ...
+    def find_longest_continious_pattern(self: ExtractorsT, str2: Union[str, bytes]) -> ExtractorsT: ...

tests/test_extractors.py

@@ -2,7 +2,10 @@
 
 
 def test_extract_strings():
-    assert Chepy("tests/files/hello").load_file().extract_strings().o.splitlines()[0] == '__PAGEZERO'
+    assert (
+        Chepy("tests/files/hello").load_file().extract_strings().o.splitlines()[0]
+        == "__PAGEZERO"
+    )
 
 
 def test_extract_hashes():
@@ -214,3 +217,21 @@ def test_extract_b64():
     when an unknown printer took a galley of type c2VjdXJpc2VjLnRlc3QuZGF0YQo= and scrambled it to make a type specimen book. 
     """
     assert Chepy(data).extract_base64().o == "c2VjdXJpc2VjLnRlc3QuZGF0YQo="
+
+
+def test_find_longest_continious_pattern():
+    str1 = "Helhello worldlo World"
+    str2 = "hello world"
+    assert Chepy(str1).find_longest_continious_pattern(str2).o == b"hello world"
+    str1 = b"Helhello worldlo World"
+    str2 = b"hello world"
+    assert Chepy(str1).find_longest_continious_pattern(str2).o == b"hello world"
+
+
+def test_find_continuous_patterns():
+    str1 = "Helhello worldlo World"
+    str2 = "hello world"
+    assert len(Chepy(str1).find_continuous_patterns(str2, 3).o) == 73
+    str1 = b"Helhello worldlo World"
+    str2 = b"hello world"
+    assert len(Chepy(str1).find_continuous_patterns(str2, 3).o) == 73