xml spec case

Samsung · Jan 16, 2025 · d932db0 · d932db0
1 parent 306190a
commit d932db0
Show file tree

Hide file tree

Showing 2 changed files with 12 additions and 8 deletions.
diff --git a/credsweeper/utils/util.py b/credsweeper/utils/util.py
@@ -481,14 +481,14 @@ def is_mxfile(data: Union[bytes, bytearray]) -> bool:
         return False
 
     # A well-formed XML must start from < or a whitespace character
-    XML_FIRST_TAG_PATTERN = re.compile(rb"^\s*<([0-9A-Za-z_]{1,256})")
-    XML_CLOSE_TAG_PATTERN = re.compile(rb"</[0-9A-Za-z_]{1,256}>")
+    XML_FIRST_BRACKET_PATTERN = re.compile(rb"^\s*<")
+    XML_OPENING_TAG_PATTERN = re.compile(rb"<([0-9A-Za-z_]{1,256})")
 
     @staticmethod
     def is_xml(data: Union[bytes, bytearray]) -> bool:
         """Used to detect xml format from raw bytes"""
-        if isinstance(data, (bytes, bytearray)):
-            if first_bracket_match := Util.XML_FIRST_TAG_PATTERN.search(data, 0, MAX_LINE_LENGTH):
+        if isinstance(data, (bytes, bytearray)) and Util.XML_FIRST_BRACKET_PATTERN.search(data, 0, MAX_LINE_LENGTH):
+            if first_bracket_match := Util.XML_OPENING_TAG_PATTERN.search(data, 0, MAX_LINE_LENGTH):
                 start_pos = first_bracket_match.start()
                 closing_tag = b"</" + first_bracket_match.group(1) + b">"
                 if start_pos < data.find(closing_tag, start_pos):

diff --git a/tests/utils/test_util.py b/tests/utils/test_util.py
@@ -619,10 +619,13 @@ def test_subtext_p(self):
 
     def test_is_xml_n(self):
         self.assertFalse(Util.is_xml(b''))
-        self.assertFalse(Util.is_xml(b'!<>'))
-        self.assertFalse(Util.is_xml(b'<br />'))
-        self.assertFalse(Util.is_xml(bytearray(b'\n' * MAX_LINE_LENGTH) + bytearray(b'    <xml>far far away</xml>')))
-        self.assertFalse(Util.is_xml(b'<html> unmatched tags </xml>'))
+        self.assertFalse(Util.is_xml(b"!<>"))
+        self.assertFalse(Util.is_xml(b"</onlyClosingTagIsFail>"))
+        self.assertFalse(Util.is_xml(b"</p><p>"))
+        self.assertFalse(Util.is_xml(b"<br />"))
+        self.assertFalse(Util.is_xml(bytearray(b'\n' * MAX_LINE_LENGTH) + bytearray(b"    <xml>far far away</xml>")))
+        self.assertFalse(Util.is_xml(b"<html> unmatched tags </xml>"))
+        self.assertFalse(Util.is_xml(b"<?xml version='1.0' encoding='utf-8'?>"))
 
     def test_is_html_n(self):
         self.assertFalse(Util.is_html(b"</html><html>"))
@@ -638,6 +641,7 @@ def test_xml_n(self):
         self.assertFalse(Util.is_html(None))
 
     def test_xml_p(self):
+        self.assertTrue(Util.is_xml(b"<?xml version='1.0' encoding='utf-8'?><xml> matched tags </xml>"))
         data = b"<mxfile atr=0><table></table></mxfile>"
         self.assertTrue(Util.is_xml(data))
         self.assertTrue(Util.is_html(data))