Handle invalid IDNA encoding in hostnames

Fixes #622

libmproxy/protocol/http.py

@@ -584,11 +584,10 @@ class HTTPRequest(HTTPMessage):
             of the request, e.g. if an upstream proxy is in place
 
             If hostheader is set to True, the Host: header will be used as
-            additional (and preferred) data source. This is handy in transparent
-            mode, where only the ip of the destination is known, but not the
-            resolved name. This is disabled by default, as an attacker may spoof
-            the host header to confuse an analyst.
-
+            additional (and preferred) data source. This is handy in
+            transparent mode, where only the IO of the destination is known,
+            but not the resolved name. This is disabled by default, as an
+            attacker may spoof the host header to confuse an analyst.
         """
         host = None
         if hostheader:
@@ -596,7 +595,10 @@ class HTTPRequest(HTTPMessage):
         if not host:
             host = self.host
         if host:
-            return host.encode("idna")
+            try:
+                return host.encode("idna")
+            except ValueError:
+                return host
         else:
             return None
 

test/test_protocol_http.py

@@ -142,6 +142,10 @@ class TestHTTPRequest:
         assert r.pretty_host(True) is None
         assert r.pretty_host(False) is None
 
+        # Invalid IDNA
+        r.headers["host"] = [".disqus.com"]
+        assert r.pretty_host(True) == ".disqus.com"
+
     def test_get_form_for_urlencoded(self):
         r = tutils.treq()
         r.headers.add("content-type", "application/x-www-form-urlencoded")