assemblyline 4.7.5.dev122__tar.gz → 4.7.5.dev123__tar.gz

{assemblyline-4.7.5.dev122/assemblyline.egg-info → assemblyline-4.7.5.dev123}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: assemblyline
-Version: 4.7.5.dev122
+Version: 4.7.5.dev123
 Summary: Assemblyline 4 - Automated malware analysis framework
 Home-page: https://github.com/CybercentreCanada/assemblyline-base
 Author: CCCS Assemblyline development team

assemblyline-4.7.5.dev123/assemblyline/VERSION

@@ -0,0 +1 @@
+4.7.5.dev123

{assemblyline-4.7.5.dev122 → assemblyline-4.7.5.dev123}/assemblyline/common/net.py

@@ -43,29 +43,31 @@ def is_valid_domain(domain: str) -> bool:
     if "@" in domain or not domain:
         return False
 
+    # Normalize domain to punycode uppercase
+    if not domain.isascii():
+        try:
+            domain = domain.encode('idna').decode('ascii')
+        except ValueError:
+            return False
+    domain = domain.upper()
+
     labels = domain.split(".")
+    if len(labels) < 2:
+        return False
+
     for label in labels:
         if not label or label.startswith("-") or label.endswith("-"):
             return False
         if not all(c.isalnum() or c == "-" for c in label):
             return False
 
-    if len(labels) < 2:
-        return False
-
-    tld = labels[-1].upper()
-    if not tld.isascii():
-        try:
-            tld = tld.encode('idna').decode('ascii').upper()
-        except ValueError:
-            return False
+    tld = labels[-1]
 
     combined_tlds = find_top_level_domains()
     if tld in combined_tlds:
         return True
 
-    domain_upper = domain.upper()
-    return any(domain_upper.endswith(d) for d in TLDS_SPECIAL_BY_DOMAIN)
+    return any(domain.endswith(d) for d in TLDS_SPECIAL_BY_DOMAIN)
 
 
 def is_valid_ip(ip: str) -> bool:

{assemblyline-4.7.5.dev122 → assemblyline-4.7.5.dev123}/assemblyline/odm/base.py

@@ -78,8 +78,14 @@ SHA256_REGEX = r"^[a-f0-9]{64}$"
 MAC_REGEX = r"^(?:(?:[0-9a-f]{2}-){5}[0-9a-f]{2}|(?:[0-9a-f]{2}:){5}[0-9a-f]{2})$"
 URI_PATH = r"([/?#]\S*)"
 # Used for finding URIs in a blob
-URI_REGEX = f"((?:(?:[A-Za-z][A-Za-z0-9+.-]*:)//)(?:[^/?#\\s]*@)?({IP_REGEX}|{DOMAIN_REGEX})(?::\\d{{1,5}})?" \
-            f"{URI_PATH}?)"
+URI_REGEX = (
+    f"((?:(?:[A-Za-z][A-Za-z0-9+.-]*:)//)(?:[^/?#\\s]*@)?({IPV4_REGEX}|"
+    # Not using DOMAIN_REGEX so wildcard domains are handled without bad backtracking.
+    r"(?:[A-Za-z0-9\u00a1-\U0010ffff_-]{1,63}[.])*"
+    r"(?:[A-Za-z0-9\u00a1-\U0010ffff][A-Za-z0-9\u00a1-\U0010ffff_-]{0,62})?[A-Za-z0-9\u00a1-\U0010ffff]"
+    r"\.(?:[Xx][Nn]--)?(?:[A-Za-z0-9\u00a1-\U0010ffff]{2,}\.?)"
+    f"|\\[{IPV6_REGEX}\\])(?::\\d{{1,5}})?{URI_PATH}?)"
+)
 # Used for direct matching
 FULL_URI = f"^{URI_REGEX}$"
 UNC_PATH_REGEX = r"^\\\\[a-zA-Z0-9-_\s]{1,63}(?:\.[a-zA-Z0-9-_\s]{1,63}){0,3}" \
@@ -546,11 +552,12 @@ class URI(Keyword):
             raise ValueError(f"[{self.name or self.parent_name}] '{value}' not match the "
                              f"validator: {self.validation_regex.pattern}")
 
-        if not is_valid_domain(match.group(2)) and not is_valid_ip(match.group(2)):
-            raise ValueError(f"[{self.name or self.parent_name}] '{match.group(2)}' in URI '{value}'"
+        hostname = match.group(2)
+        if not is_valid_domain('.'.join(hostname.rsplit('.', 2)[-2:])) and not is_valid_ip(hostname) and not hostname.startswith('['):
+            raise ValueError(f"[{self.name or self.parent_name}] '{hostname}' in URI '{value}'"
                              " is not a valid Domain or IP.")
 
-        return match.group(0).replace(match.group(2), match.group(2).lower())
+        return match.group(0).replace(hostname, hostname.lower())
 
 
 class UNCPath(ValidatedKeyword):

{assemblyline-4.7.5.dev122 → assemblyline-4.7.5.dev123/assemblyline.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: assemblyline
-Version: 4.7.5.dev122
+Version: 4.7.5.dev123
 Summary: Assemblyline 4 - Automated malware analysis framework
 Home-page: https://github.com/CybercentreCanada/assemblyline-base
 Author: CCCS Assemblyline development team

assemblyline-4.7.5.dev122/assemblyline/VERSION

@@ -1 +0,0 @@
-4.7.5.dev122