[sapphire] Improved stability handling invalid URLs

sapphire/test_worker.py

@@ -4,6 +4,7 @@
 # pylint: disable=protected-access
 
 import socket
+from random import randint
 from threading import Thread, ThreadError
 
 from pytest import mark
@@ -196,11 +197,22 @@ def test_response_01(req, method, scheme, path):
  "req",
  [
  b"a",
+ # Invalid IPv6 URL
  b"GET http://[test/ HTTP/1.1",
  b"GET HTTP/1.1",
  b"GET a a a a a HTTP/1.1",
+ # Invalid characters under NFKC normalization
+ b"GET http://%E2%84%80/ HTTP/1.1",
  ],
 )
 def test_response_02(req):
  """test Request.parse() failures"""
  assert Request.parse(req) is None
+
+
+def test_response_03():
+ """test Request.parse() by passing random urls"""
+ for _ in range(1000):
+ # create random 'netloc', for example '%1A%EF%09'
+ chars = "".join([f"%{randint(0, 255):02X}" for _ in range(randint(1, 8))])
+ Request.parse(f"GET http://{chars}/ HTTP/1.1".encode())

sapphire/worker.py

@@ -47,12 +47,16 @@ def parse(cls, raw_data: bytes) -> Optional["Request"]:
  # TODO: parse headers if needed
 
  try:
+ url_str = req_match.group("url").decode("ascii", errors="replace")
  # unquote() accepts str | bytes as of Python 3.9
- url = urlparse(
- unquote(req_match.group("url").decode("ascii", errors="replace"))
- )
+ url = urlparse(unquote(url_str))
  except ValueError as exc:
- if "Invalid IPv6 URL" not in str(exc): # pragma: no cover
+ msg = str(exc)
+ if (
+ "contains invalid characters under NFKC normalization" not in msg
+ and "Invalid IPv6 URL" not in msg
+ ):
+ LOG.error("Failed to parse URL: %r", url_str)
  raise
  LOG.debug("failed to parse url from request")
  return None