credsweeper 1.11.0__py3-none-any.whl → 1.11.1__py3-none-any.whl

credsweeper/__init__.py

@@ -18,4 +18,4 @@ __all__ = [
     '__version__'
 ]
 
-__version__ = "1.11.0"
+__version__ = "1.11.1"

credsweeper/deep_scanner/deep_scanner.py

@@ -28,6 +28,7 @@ from .pdf_scanner import PdfScanner
 from .pkcs12_scanner import Pkcs12Scanner
 from .pptx_scanner import PptxScanner
 from .tar_scanner import TarScanner
+from .tmx_scanner import TmxScanner
 from .xlsx_scanner import XlsxScanner
 from .xml_scanner import XmlScanner
 from .zip_scanner import ZipScanner
@@ -126,6 +127,10 @@ class DeepScanner(
                 deep_scanners.append(MxfileScanner)
                 deep_scanners.append(XmlScanner)
                 fallback_scanners.append(ByteScanner)
+            elif Util.is_tmx(data):
+                deep_scanners.append(TmxScanner)
+                fallback_scanners.append(XmlScanner)
+                fallback_scanners.append(ByteScanner)
             else:
                 deep_scanners.append(XmlScanner)
                 fallback_scanners.append(ByteScanner)

credsweeper/deep_scanner/tmx_scanner.py

@@ -0,0 +1,45 @@
+import logging
+from abc import ABC
+from typing import List, Optional
+
+from lxml import etree
+
+from credsweeper.common.constants import MIN_DATA_LEN
+from credsweeper.credentials import Candidate
+from credsweeper.deep_scanner.abstract_scanner import AbstractScanner
+from credsweeper.file_handler.data_content_provider import DataContentProvider
+from credsweeper.file_handler.string_content_provider import StringContentProvider
+from credsweeper.utils import Util
+
+logger = logging.getLogger(__name__)
+
+
+class TmxScanner(AbstractScanner, ABC):
+    """Realises tmX files scanning for values only. Image tags are skipped."""
+
+    def data_scan(
+            self,  #
+            data_provider: DataContentProvider,  #
+            depth: int,  #
+            recursive_limit_size: int) -> Optional[List[Candidate]]:
+        """Tries to represent data as xml text and scan as text lines"""
+        try:
+            lines = []
+            # the format is always in single line xlm, so line numbers are not actual
+            tree = etree.fromstring(data_provider.data)
+            for element in tree.iter():
+                tag = Util.extract_element_data(element, "tag")
+                if "Image" in tag:
+                    continue
+                text = Util.extract_element_data(element, "text")
+                if MIN_DATA_LEN > len(text):
+                    continue
+                lines.append(text)
+            tmx_data_provider = StringContentProvider(lines=lines,
+                                                      file_path=data_provider.file_path,
+                                                      file_type=data_provider.file_type,
+                                                      info=f"{data_provider.info}|TMX")
+            return self.scanner.scan(tmx_data_provider)
+        except Exception as exc:
+            logger.warning("Cannot processed tmX file %s %s", str(data_provider.file_path), str(exc))
+        return None

credsweeper/utils/util.py

@@ -517,6 +517,18 @@ class Util:
                 return True
         return False
 
+    @staticmethod
+    def is_tmx(data: Union[bytes, bytearray]) -> bool:
+        """Used to detect tm7,tm6,etc. (ThreadModeling) format."""
+        if isinstance(data, (bytes, bytearray)):
+            for opening_tag, closing_tag in [(b"<ThreatModel", b"</ThreatModel>"),
+                                             (b"<KnowledgeBase", b"</KnowledgeBase>")]:
+                opening_pos = data.find(opening_tag, 0, MAX_LINE_LENGTH)
+                if 0 <= opening_pos < data.find(closing_tag, opening_pos):
+                    # opening and closing tags were found - suppose it is an HTML
+                    return True
+        return False
+
     # A well-formed XML must start from < or a whitespace character
     XML_FIRST_BRACKET_PATTERN = re.compile(rb"^\s*<")
     XML_OPENING_TAG_PATTERN = re.compile(rb"<([0-9A-Za-z_]{1,256})")
@@ -583,14 +595,14 @@ class Util:
         line_nums = []
         tree = etree.fromstringlist(xml_lines)
         for element in tree.iter():
-            tag = Util._extract_element_data(element, "tag")
-            text = Util._extract_element_data(element, "text")
+            tag = Util.extract_element_data(element, "tag")
+            text = Util.extract_element_data(element, "text")
             lines.append(f"{tag} : {text}")
             line_nums.append(element.sourceline)
         return lines, line_nums
 
     @staticmethod
-    def _extract_element_data(element, attr) -> str:
+    def extract_element_data(element: Any, attr: str) -> str:
         """Extract xml element data to string.
 
         Try to extract the xml data and strip() the string.
@@ -605,7 +617,7 @@ class Util:
         """
         element_attr: Any = getattr(element, attr)
         if element_attr is None or not isinstance(element_attr, str):
-            return ""
+            return ''
         return str(element_attr).strip()
 
     @staticmethod

{credsweeper-1.11.0.dist-info → credsweeper-1.11.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: credsweeper
-Version: 1.11.0
+Version: 1.11.1
 Summary: Credential Sweeper
 Project-URL: Homepage, https://github.com/Samsung/CredSweeper
 Project-URL: Bug Tracker, https://github.com/Samsung/CredSweeper/issues

{credsweeper-1.11.0.dist-info → credsweeper-1.11.1.dist-info}/RECORD

@@ -1,4 +1,4 @@
-credsweeper/__init__.py,sha256=QgMuMG73C_jpq7fXbkUOXA9CHYf87zF9458ytKtAUl0,632
+credsweeper/__init__.py,sha256=DNgFBLOXoBUXL0IvKDJswX-CEJfApajSHSuJq_FhRtg,632
 credsweeper/__main__.py,sha256=jlI83ctJJfF0koMqP6u24JASC7MIPA2g1POx7aeuaQ8,17187
 credsweeper/app.py,sha256=sexUp4Qced22AhvbcVlb5C-QtJRoDmUp5qhc_nwj248,21369
 credsweeper/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -21,7 +21,7 @@ credsweeper/deep_scanner/__init__.py,sha256=Lp94BjQPZTgEa77E0v6xZaXZvQf2A-QTHsjq
 credsweeper/deep_scanner/abstract_scanner.py,sha256=RLwANH7C42GMdgq_uyMH895HCUwmZFfCOTmFTzO8ni0,1404
 credsweeper/deep_scanner/byte_scanner.py,sha256=oHeA8mGe995SHqWvONhTDBIE5j50TQASHA9Mv6LHYuQ,1125
 credsweeper/deep_scanner/bzip2_scanner.py,sha256=74RsjmeuffEuxmKl04lXIZt3q_Zvxj-gLHXACqVSU_o,1619
-credsweeper/deep_scanner/deep_scanner.py,sha256=lBfXRsALmI62WRNeo6QULLVSFfv9rP4aa_fOXNYZe_g,17035
+credsweeper/deep_scanner/deep_scanner.py,sha256=l4fj7yuwCl6FWNB3UO4JmMAGEVrHSi-8OH8LIwsLTB4,17263
 credsweeper/deep_scanner/docx_scanner.py,sha256=t0vocPDY54KBjpmQBo53n5KvOISXkB-LlxFbfuRNLMA,4128
 credsweeper/deep_scanner/eml_scanner.py,sha256=iRLr2yvBWGktT2oXxl-haqnhJN3tglO1Mej10hFk0as,3512
 credsweeper/deep_scanner/encoder_scanner.py,sha256=qszql2a-lVuzVN_bNS2EsJ-Zxpqql52o1sJsLnpjX7M,1279
@@ -34,6 +34,7 @@ credsweeper/deep_scanner/pdf_scanner.py,sha256=LMyIoVJPNFOFnAfcZ5Akr7PTWSUBNPT6G
 credsweeper/deep_scanner/pkcs12_scanner.py,sha256=s6WyeLUHxqbMnM3t_eY6GUwn1Yyh_nyehprWM_HgomQ,2142
 credsweeper/deep_scanner/pptx_scanner.py,sha256=aMX6GgnUEShonHjlqhaI5w970b-2yxmKsld5kY1XdeQ,1828
 credsweeper/deep_scanner/tar_scanner.py,sha256=L3a9OUhQQweDNLVbe_LNLhldtVeU8DlS0Ux3ip_KN2w,2425
+credsweeper/deep_scanner/tmx_scanner.py,sha256=6BsMysSSSJrxtssh4bf1e4vwpps7yXDDvByFkyLhC_o,1946
 credsweeper/deep_scanner/xlsx_scanner.py,sha256=Ck8j14OWy9LTXK0GBASCdPq9VhZe5ceUv0uZShFFpo8,2706
 credsweeper/deep_scanner/xml_scanner.py,sha256=Dc5vw8MhOQFppPSMMzJBANKTC1OIu_7UsuLAwPGYQ4c,1302
 credsweeper/deep_scanner/zip_scanner.py,sha256=rWNV43OV8FTpXGMkAlRCwnnaJ-WdiIpreI9FUpx7wb0,2431
@@ -144,9 +145,9 @@ credsweeper/utils/__init__.py,sha256=wPdTkrSBAkR3rppFZ68k6MiT_P7tIHuAb3AcwndJCWg
 credsweeper/utils/entropy_validator.py,sha256=711xCIBGAy-Pb6wqbMpEToa4dOYj5_CmkbKHygLeQrI,2796
 credsweeper/utils/hop_stat.py,sha256=0D7xB1CVAUhseOZWvLZXxn3MYHKZnfnFJ8hj7tONiyU,2978
 credsweeper/utils/pem_key_detector.py,sha256=Z1LJYm4WAqppF2ooj81-lbhrg2woiNKiMk8Nt4DV-G8,7721
-credsweeper/utils/util.py,sha256=koK8Sat8wjWHncOy0MQbnKRUrdxrJ77gt8U6spRG-oA,30451
-credsweeper-1.11.0.dist-info/METADATA,sha256=ntR_qdmXm2ZVenmDOzzlxcaf6lFGL8xOXHf1kNo3fsg,10504
-credsweeper-1.11.0.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-credsweeper-1.11.0.dist-info/entry_points.txt,sha256=SLGNZshvi3zpWPhVmRP-oDXRMRPBS4tzRDy6xYOXwqA,58
-credsweeper-1.11.0.dist-info/licenses/LICENSE,sha256=aU7mGjBKbmRHNLVXXzcPdKmTtBxRwDPtjflQRfN7fFg,1065
-credsweeper-1.11.0.dist-info/RECORD,,
+credsweeper/utils/util.py,sha256=p8Chj7VWJrAP8q_jQhssfm1xoiIN_iCN1uMViq-1JrA,31102
+credsweeper-1.11.1.dist-info/METADATA,sha256=W39T66V_XQYHdIx91-XgGi48QJiiZN7CwGndeUN98FA,10504
+credsweeper-1.11.1.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+credsweeper-1.11.1.dist-info/entry_points.txt,sha256=SLGNZshvi3zpWPhVmRP-oDXRMRPBS4tzRDy6xYOXwqA,58
+credsweeper-1.11.1.dist-info/licenses/LICENSE,sha256=aU7mGjBKbmRHNLVXXzcPdKmTtBxRwDPtjflQRfN7fFg,1065
+credsweeper-1.11.1.dist-info/RECORD,,