Revision: 361
Author: bslatkin
Date: Thu Jun 3 16:20:53 2010
Log: hub ignores improper char encodings in etag header values
http://code.google.com/p/pubsubhubbub/source/detail?r=361
Modified:
/trunk/hub/main.py
/trunk/hub/main_test.py
=======================================
--- /trunk/hub/main.py Thu Jun 3 16:00:36 2010
+++ /trunk/hub/main.py Thu Jun 3 16:20:53 2010
@@ -1275,9 +1275,21 @@
format: The last parsing format that worked correctly for this feed.
Should be 'rss' or 'atom'.
"""
- self.content_type = headers.get('Content-Type', '').lower()
- self.last_modified = headers.get('Last-Modified')
- self.etag = headers.get('ETag')
+ try:
+ self.content_type = headers.get('Content-Type', '').lower()
+ except UnicodeDecodeError:
+ logging.exception('Content-Type header had bad encoding')
+
+ try:
+ self.last_modified = headers.get('Last-Modified')
+ except UnicodeDecodeError:
+ logging.exception('Last-Modified header had bad encoding')
+
+ try:
+ self.etag = headers.get('ETag')
+ except UnicodeDecodeError:
+ logging.exception('ETag header had bad encoding')
+
if header_footer is not None:
self.header_footer = header_footer
if format is not None:
=======================================
--- /trunk/hub/main_test.py Thu Jun 3 16:00:36 2010
+++ /trunk/hub/main_test.py Thu Jun 3 16:20:53 2010
@@ -2031,6 +2031,33 @@
self.assertEquals(data.replace('\n', ''),
event.payload.replace('\n', ''))
self.assertEquals('application/atom+xml', event.content_type)
+ def testPullWithUnicodeEtag(self):
+ """Tests when the ETag header has a unicode value.
+
+ The ETag value should be ignored because non-ascii ETag values are
invalid.
+ """
+ data = ('<?xml version="1.0" encoding="utf-8"?>\n<feed><my
header="data"/>'
+ '<entry><id>1</id><updated>123</updated>wooh</entry></feed>')
+ topic = 'http://example.com/my-topic'
+ callback = 'http://example.com/my-subscriber'
+ self.assertTrue(Subscription.insert(callback,
topic, 'token', 'secret'))
+ FeedToFetch.insert([topic])
+ urlfetch_test_stub.instance.expect('get', topic, 200, data,
+ response_headers={
+ 'ETag': '\xe3\x83\x96\xe3\x83\xad\xe3\x82\xb0\xe8\xa1\x86',
+ 'Content-Type': 'application/atom+xml',
+ })
+ self.handle('post', ('topic', topic))
+ feed = FeedToFetch.get_by_key_name(get_hash_key_name(topic))
+ self.assertTrue(feed is None)
+ event = EventToDeliver.all().get()
+ self.assertEquals(data.replace('\n', ''),
event.payload.replace('\n', ''))
+ self.assertEquals('application/atom+xml', event.content_type)
+ self.assertEquals(
+ {'Connection': 'cache-control',
+ 'Cache-Control': 'no-cache no-store max-age=1'},
+ FeedRecord.all().get().get_request_headers())
+
def testPullGoodRss(self):
"""Tests when the RSS XML can parse just fine."""
data = ('<?xml version="1.0" encoding="utf-8"?>\n'
