Use more lenient urllib3.util's parse_url

aikido_zen/helpers/try_parse_url.py

@@ -1,7 +1,7 @@
 """Helper function file"""
 
 from urllib.parse import urlparse
-
+from urllib3.util import parse_url
 
 def try_parse_url(url):
     """Tries to parse the url using urlparse"""
@@ -12,3 +12,13 @@ def try_parse_url(url):
         return None
     except Exception:
         return None
+
+def try_lenient_parse_url(url):
+    """Tries to parse the url using parse_url, which is more lenient than urlparse"""
+    try:
+        parsed_url = parse_url(url)
+        if parsed_url.scheme and parsed_url.host:
+            return parsed_url
+        return None
+    except Exception:
+        return None

aikido_zen/vulnerabilities/ssrf/find_hostname_in_userinput.py

@@ -6,7 +6,7 @@
 from urllib.parse import unquote
 
 from aikido_zen.helpers.get_port_from_url import get_port_from_url
-from aikido_zen.helpers.try_parse_url import try_parse_url
+from aikido_zen.helpers.try_parse_url import try_lenient_parse_url
 
 
 def find_hostname_in_userinput(user_input, hostname_options: List[str], port=None):
@@ -28,9 +28,9 @@ def find_hostname_in_userinput(user_input, hostname_options: List[str], port=Non
         ]
 
     for variant in variants:
-        user_input_url = try_parse_url(variant)
+        user_input_url = try_lenient_parse_url(variant)
         if user_input_url and user_input_url.hostname in hostname_options:
-            user_port = get_port_from_url(user_input_url.geturl())
+            user_port = get_port_from_url(user_input_url.url)
 
             # We were unable to retrieve the port from the URL, likely because it contains an invalid port.
             # Let's assume we have found the hostname in the user input, even though it doesn't match on port.

aikido_zen/vulnerabilities/ssrf/handle_http_response.py

@@ -2,7 +2,7 @@
 
 from aikido_zen.context import get_current_context
 from aikido_zen.helpers.is_redirect_status_code import is_redirect_status_code
-from aikido_zen.helpers.try_parse_url import try_parse_url
+from aikido_zen.helpers.try_parse_url import try_lenient_parse_url
 from .find_hostname_in_context import find_hostname_in_context
 from .get_redirect_origin import get_redirect_origin
 
@@ -29,7 +29,7 @@ def handle_http_response(http_response, source):
     if not isinstance(location, str):
         return
 
-    parsed_location = try_parse_url(location)
+    parsed_location = try_lenient_parse_url(location)
     if not parsed_location:
         return