hippotat: Convert an explicitly configured URL to ASCII.

[hippotat.git] / hippotat

diff --git a/hippotat b/hippotat
index 200cf3a2a4b333ba797c7b4d36719289687209ab..6aa3121c7ad68a72367930015817a9d91955e460 100755 (executable)
--- a/hippotat
+++ b/hippotat
@@ -26,6 +26,7 @@ from hippotatlib import *
 
 import twisted.web
 import twisted.web.client
+import urllib.parse
 
 import io
 
@@ -43,7 +44,12 @@ class GeneralResponseConsumer(twisted.internet.protocol.Protocol):
     self._log(DBG.HTTP_CTRL, 'connectionMade')
 
   def connectionLostOK(self, reason):
-    return reason.check(twisted.web.client.ResponseDone)
+    return (reason.check(twisted.web.client.ResponseDone) or
+            reason.check(twisted.web.client.PotentialDataLoss))
+    # twisted.web.client.PotentialDataLoss is an entirely daft
+    # exception.  It will occur every time if the origin server does
+    # not provide a Content-Length.  (hippotatd does, of course, but
+    # the HTTP transaction might be proxied.)
 
 class ResponseConsumer(GeneralResponseConsumer):
   def __init__(self, cl, req, resp):
@@ -218,8 +224,7 @@ class Client():
       cl.log(DBG.HTTP_FULL, 'requesting: ' + str(mime))
 
       hh = { 'User-Agent': ['hippotat'],
-             'Content-Type': ['multipart/form-data; boundary="b"'],
-             'Content-Length': [str(len(mime))] }
+             'Content-Type': ['multipart/form-data; boundary="b"'] }
 
       bytesreader = io.BytesIO(mime)
       producer = twisted.web.client.FileBodyProducer(bytesreader)
@@ -241,6 +246,17 @@ class Client():
 
 clients = [ ]
 
+def encode_url(urlstr):
+  # Oh, this is a disaster.  We're given a URL as a `str', but the underlying
+  # machinery insists on having `bytes'.  Assume we've been given a sensible
+  # URL, with escaping in all of the necessary places, except that it may
+  # contain non-ASCII characters: then encode as UTF-8 and squash the top-
+  # bit-set bytes down to percent escapes.
+  #
+  # This conses like it's going out of fashion, but it gets the job done.
+  return b''.join(bytes([b]) if b < 128 else '%%%02X' % b
+                  for b in urlstr.encode('utf-8'))
+
 def process_cfg(_opts, putative_servers, putative_clients):
   global clients
 
@@ -265,7 +281,7 @@ def process_cfg(_opts, putative_servers, putative_clients):
       try: c.ifname     = srch(cfg_get_raw, 'ifname_client')
       except NoOptionError: pass
 
-      try: c.url = srch(cfg.get,'url')
+      try: c.url = encode_url(srch(cfg.get,'url'))
       except NoOptionError:
         cfg_process_saddrs(c, ss)
         c.url = c.saddrs[0].url()