more REST API error handling for url input: bad urls, connection fail…

…ures

https://console.cloud.google.com/errors/13725011683277350339

app.py

@@ -3,6 +3,7 @@
 
 __author__ = 'Ryan Barrett <granary@ryanb.org>'
 
+import httplib
 import json
 import logging
 import urllib
@@ -99,7 +100,15 @@ def get(self):
 
     # fetch url
     url = util.get_required_param(self, 'url')
-    resp = util.urlopen(url)
+    try:
+      resp = util.urlopen(url)
+    except httplib.InvalidURL as e:
+      self.abort(400, str(e))
+    except Exception as e:
+      if util.is_connection_failure(e):
+        self.abort(502, str(e))
+      raise
+
     if url != resp.geturl():
       url = resp.geturl()
       logging.info('Redirected to %s', url)

test_app.py

@@ -1,7 +1,8 @@
 """Unit tests for app.py.
 """
-
+import httplib
 import json
+import socket
 import xml.sax.saxutils
 
 import oauth_dropins.webutil.test
@@ -182,6 +183,19 @@ def test_url_bad_input(self):
     resp = app.application.get_response('/url?url=http://my/posts.json&input=foo')
     self.assert_equals(400, resp.status_int)
 
+  def test_url_bad_url(self):
+    self.expect_urlopen('http://astralandopal.com\\').AndRaise(httplib.InvalidURL(''))
+    self.mox.ReplayAll()
+    resp = app.application.get_response(
+      '/url?url=http://astralandopal.com\\&input=html')
+    self.assert_equals(400, resp.status_int)
+
+  def test_url_fetch_fails(self):
+    self.expect_urlopen('http://my/posts.html').AndRaise(socket.error(''))
+    self.mox.ReplayAll()
+    resp = app.application.get_response('/url?url=http://my/posts.html&input=html')
+    self.assert_equals(502, resp.status_int)
+
   def test_hub(self):
     self.expect_urlopen('http://my/posts.html', HTML % {
       'body_class': '',