🗓 Jul 2, 2024 11:59:14 PM

💚 build steps added/updated
✨ aws_account_id_from_access_key
✨ expand_alpha_range

.github/workflows/tests_multi_os.yml

@@ -56,11 +56,11 @@ jobs:
         
     - name: Install test requirements
       run: |
-        pip install sphinx recommonmark pytest==8.1.1 pytest-cov pyperclip 
+        pip install sphinx recommonmark pytest==8.1.1 pytest-cov==5.0.0 pyperclip 
         
     - name: Test with pytest
       run: |
-        pytest -v --disable-pytest-warnings --cov-report=xml --cov=chepy --cov-config=.coveragerc tests/
+        COVERAGE_CORE=sysmon pytest -v --disable-pytest-warnings --cov-report=xml --cov=chepy --cov-config=.coveragerc tests/
         coverage report -m
         
     - name: Test plugins osx

Makefile

@@ -2,9 +2,9 @@
 
 
 test:
-	python -m pytest --noconftest -v --disable-pytest-warnings --cov-report=xml --cov=chepy --cov-config=.coveragerc tests/
+	COVERAGE_CORE=sysmon python -m pytest --noconftest -v --disable-pytest-warnings --cov-report=xml --cov=chepy --cov-config=.coveragerc tests/
 
 test-all: test
-	python -m pytest --noconftest -v --disable-pytest-warnings tests_plugins/
+	COVERAGE_CORE=sysmon python -m pytest --noconftest -v --disable-pytest-warnings tests_plugins/
 
 # git log --format=%B 4.0.0..5.0.0 | sed '/^\s*$/d' | sort | uniq

chepy/modules/extractors.py

@@ -1,5 +1,6 @@
 import math
 from binascii import unhexlify
+import base64
 from typing import TypeVar, Union, List
 from urllib.parse import urlparse as _pyurlparse
 import lazy_import
@@ -684,3 +685,19 @@ def extract_html_comments(self):
             filter(lambda x: x != "", self._parsel_obj().xpath("//comment()").getall())
         )
         return self
+
+    @ChepyDecorators.call_stack
+    def aws_account_id_from_access_key(self):
+        """Extract AWS account id from access key
+
+        Returns:
+            Chepy: The Chepy object. 
+        """
+        trimmed_AWSKeyID = self._convert_to_str()[4:]
+        x = base64.b32decode(trimmed_AWSKeyID)
+        y = x[0:6]
+        z = int.from_bytes(y, byteorder='big', signed=False)
+        mask = int.from_bytes(unhexlify(b'7fffffffff80'), byteorder='big', signed=False)
+
+        self.state = (z & mask)>>7
+        return self

chepy/modules/extractors.pyi

@@ -44,3 +44,4 @@ class Extractors(ChepyCore):
     def css_selector(self: ExtractorsT, query: str) -> ExtractorsT: ...
     def extract_html_comments(self: ExtractorsT) -> ExtractorsT: ...
     def extract_html_tags(self: ExtractorsT, tag: List[str]) -> ExtractorsT: ...
+    def aws_account_id_from_access_key(self: ExtractorsT) -> ExtractorsT: ...

chepy/modules/utils.py

@@ -819,3 +819,53 @@ def pick(self, *values: Any):
         else:  # pragma: no cover
             raise TypeError("Input should be a list, dictionary, string, or bytes")
         return self
+
+    @ChepyDecorators.call_stack
+    def expand_alpha_range(self, join_by: Union[str, None] = None):
+        """Get all alphanumberic or hex chars for the specified range
+
+        Args:
+            join_by (str, optional): Join by. Defaults to Union[str, None].
+
+        Returns:
+            Chepy: The Chepy object.
+        """
+        alph_str = self._convert_to_str()
+        hold = []
+
+        def expand_range(start, end):
+            return [str(x) for x in range(int(start), int(end) + 1)]
+
+        def expand_char_range(start, end):
+            return [chr(x) for x in range(ord(start), ord(end) + 1)]
+
+        hold = []
+        i = 0
+        length = len(alph_str)
+
+        while i < length:
+            # Match numerical ranges like 10-20
+            num_match = re.match(r"(\d+)-(\d+)", alph_str[i:])
+            if num_match:
+                start, end = num_match.groups()
+                hold.extend(expand_range(start, end))
+                i += len(start) + len(end) + 1  # move past the number range
+            elif i < length - 2 and alph_str[i + 1] == "-" and alph_str[i] != "\\":
+                # Handle character ranges like a-z
+                start = alph_str[i]
+                end = alph_str[i + 2]
+                hold.extend(expand_char_range(start, end))
+                i += 2
+            elif (
+                i < length - 2 and alph_str[i] == "\\" and alph_str[i + 1] == "-"
+            ):  # pragma: no cover
+                hold.append("-")
+                i += 1
+            else:
+                hold.append(alph_str[i])
+            i += 1
+
+        if join_by is not None:
+            hold = join_by.join(hold)
+        self.state = hold
+        return self

chepy/modules/utils.pyi

@@ -41,3 +41,4 @@ class Utils(ChepyCore):
     def drop_bytes(self: UtilsT, start: int, length: int) -> UtilsT: ...
     def without(self: UtilsT, *values: Any) -> UtilsT: ...
     def pick(self: UtilsT, *values: Any) -> UtilsT: ...
+    def expand_alpha_range(self, join_by: Union[str, None]=None) -> UtilsT: ...

tests/test_extractors.py

@@ -239,9 +239,7 @@ def test_find_continuous_patterns():
 
 def test_zero_with_chars_tags():
     assert (
-        Chepy("this 󠁮󠁩󠁣is 󠁣󠁻󠀰just 󠁲󠁟󠀱a 󠀵󠁟󠀱simple 󠀷󠁽text file")
-        .extract_zero_width_chars_tags()
-        .o
+        Chepy("this 󠁮󠁩󠁣is 󠁣󠁻󠀰just 󠁲󠁟󠀱a 󠀵󠁟󠀱simple 󠀷󠁽text file").extract_zero_width_chars_tags().o
         == b"nicc{0r_15_17}"
     )
 
@@ -252,7 +250,7 @@ def test_decode_zero_width():
             "e2808be2808be2808be2808befbbbfe280ace2808b68656c6c6fe2808be2808be2808be2808befbbbfe2808be2808ce2808be2808be2808be2808be280acefbbbfefbbbfe2808be2808be2808be2808befbbbfe2808defbbbfe2808be2808be2808be2808befbbbfe2808be2808ce2808be2808be2808be2808befbbbfe280ace2808c"
         )
         .from_hex()
-        .decode_zero_width("\u200B\u200c\u200d\u202c\ufeff")
+        .decode_zero_width("\u200b\u200c\u200d\u202c\ufeff")
         .o["hidden"]
         == "secret"
     )
@@ -271,7 +269,7 @@ def test_decode_zero_width():
         )
         .from_hex()
         .decode_zero_width(
-            "\u200B\u200C\u200D\u200E\u202A\u202C\u202D\u2062\u2063\ufeff"
+            "\u200b\u200c\u200d\u200e\u202a\u202c\u202d\u2062\u2063\ufeff"
         )
         .o["hidden"]
         == "secret"
@@ -296,3 +294,9 @@ def test_html_tags():
 def test_html_comments():
     c = Chepy("tests/files/test.html").load_file().extract_html_comments()
     assert len(c.o) == 3
+
+
+def test_aws_access_key_from_account_id():
+    assert (
+        Chepy("ASIAQNZGKIQY56JQ7WML").aws_account_id_from_access_key().o == 29608264753
+    )

tests/test_utils.py

@@ -246,17 +246,38 @@ def test_shuffle():
 def test_drop_bytes():
     assert Chepy("hello").drop_bytes(2, 2).o == b"heo"
 
+
 def test_without_pick():
-    data1 = 'hello'
-    data2 = [1,2,'a', 'b']
-    data3 = {'a':1, 2: 3}
+    data1 = "hello"
+    data2 = [1, 2, "a", "b"]
+    data3 = {"a": 1, 2: 3}
     # test without
-    assert Chepy(data1).without('ll').o == b'heo'
-    assert Chepy(data1).without('l', b'l').o == b'heo'
-    assert Chepy(data2).without(1, 'a').o == [2,'b']
-    assert Chepy(data3).without('a').o == {2: 3}
+    assert Chepy(data1).without("ll").o == b"heo"
+    assert Chepy(data1).without("l", b"l").o == b"heo"
+    assert Chepy(data2).without(1, "a").o == [2, "b"]
+    assert Chepy(data3).without("a").o == {2: 3}
     # test pick
-    assert Chepy(data1).pick('ll').o == b'll'
-    assert Chepy(data1).pick('l', b'l').o == b'll'
-    assert Chepy(data2).pick(1, 'a').o == [1,'a']
-    assert Chepy(data3).pick('a').o == {'a': 1}
+    assert Chepy(data1).pick("ll").o == b"ll"
+    assert Chepy(data1).pick("l", b"l").o == b"ll"
+    assert Chepy(data2).pick(1, "a").o == [1, "a"]
+    assert Chepy(data3).pick("a").o == {"a": 1}
+
+
+def test_alpha_range():
+    assert Chepy("a-e").expand_alpha_range().o == ["a", "b", "c", "d", "e"]
+    assert Chepy("a-cA-C0-2").expand_alpha_range().o == [
+        "a",
+        "b",
+        "c",
+        "A",
+        "B",
+        "C",
+        "0",
+        "1",
+        "2",
+    ]
+    assert (
+        Chepy("a-cA-C0-2").expand_alpha_range("").o
+        == "".join(["a", "b", "c", "A", "B", "C", "0", "1", "2"]).encode()
+    )
+    assert Chepy(" a-c:").expand_alpha_range().o == [" ", "a", "b", "c", ":"]