proxy: ip resolver: show 500 error if incorrect coll preconfigured for ip-based settings (todo: make it configurable?)

pywb/framework/proxy_resolvers.py

@@ -1,5 +1,6 @@
 from wbrequestresponse import WbResponse
 from pywb.utils.loaders import extract_client_cookie
+from pywb.utils.wbexception import WbException
 from pywb.utils.statusandheaders import StatusAndHeaders
 from pywb.rewrite.wburl import WbUrl
 
@@ -32,16 +33,16 @@ class BaseCollResolver(object):
 
         # invalid parsing
         if proxy_coll == '':
-            return None, None, None, None, self.select_coll_response(env)
+            return None, None, None, None, self.select_coll_response(env, proxy_coll)
 
         if proxy_coll is None and isinstance(self.use_default_coll, str):
             proxy_coll = self.use_default_coll
 
         if proxy_coll:
-            proxy_coll = '/' + proxy_coll + '/'
+            path = '/' + proxy_coll + '/'
 
             for r in self.routes:
-                matcher, c = r.is_handling(proxy_coll)
+                matcher, c = r.is_handling(path)
                 if matcher:
                     route = r
                     coll = c
@@ -49,7 +50,7 @@ class BaseCollResolver(object):
 
             # if no match, return coll selection response
             if not route:
-                return None, None, None, None, self.select_coll_response(env)
+                return None, None, None, None, self.select_coll_response(env, proxy_coll)
 
         # if 'use_default_coll', find first WbUrl-handling collection
         elif self.use_default_coll:
@@ -59,7 +60,7 @@ class BaseCollResolver(object):
 
         # otherwise, return the appropriate coll selection response
         else:
-            return None, None, None, None, self.select_coll_response(env)
+            return None, None, None, None, self.select_coll_response(env, proxy_coll)
 
         return route, coll, matcher, ts, None
 
@@ -89,7 +90,7 @@ class ProxyAuthResolver(BaseCollResolver):
         proxy_coll = self.read_basic_auth_coll(proxy_auth)
         return proxy_coll, None
 
-    def select_coll_response(self, env):
+    def select_coll_response(self, env, default_coll=None):
         proxy_msg = 'Basic realm="{0}"'.format(self.auth_msg)
 
         headers = [('Content-Type', 'text/plain'),
@@ -136,6 +137,9 @@ class IPCacheResolver(BaseCollResolver):
 
         return ip
 
+    def select_coll_response(self, env, default_coll=None):
+        raise WbException('Invalid Proxy Collection Specified: ' + str(default_coll))
+
     def get_proxy_coll_ts(self, env):
         ip = env['REMOTE_ADDR']
         qs = env.get('pywb.proxy_query')
@@ -202,7 +206,7 @@ class CookieResolver(BaseCollResolver):
         coll, ts, sesh_id = self.get_coll(env)
         return coll, ts
 
-    def select_coll_response(self, env):
+    def select_coll_response(self, env, default_coll=None):
         return self.make_magic_response('auto',
                                         env['REL_REQUEST_URI'],
                                         env)

tests/test_proxy_http_ip.py

@@ -25,11 +25,11 @@ class TestProxyIPResolver(BaseIntegration):
         assert resp.content_type == 'text/plain'
         assert resp.content_length > 0
 
-    def get_url(self, uri, addr='127.0.0.1'):
+    def get_url(self, uri, addr='127.0.0.1', status='*'):
         parts = urlsplit(uri)
         env = dict(REQUEST_URI=uri, QUERY_STRING=parts.query, SCRIPT_NAME='', REMOTE_ADDR=addr)
         # 'Simulating' proxy by settings REQUEST_URI explicitly to full url with empty SCRIPT_NAME
-        return self.testapp.get('/x-ignore-this-x', extra_environ=env)
+        return self.testapp.get('/x-ignore-this-x', extra_environ=env, status=status)
 
     def test_proxy_ip_default_ts(self):
         resp = self.get_url('http://www.iana.org/')
@@ -93,3 +93,13 @@ class TestProxyIPResolver(BaseIntegration):
         resp = self.get_url('http://info.pywb.proxy/')
         assert resp.json == {'ip': '127.0.0.1', 'coll': None, 'ts': None}
 
+    def test_proxy_set_coll_invalid(self):
+        resp = self.get_url('http://info.pywb.proxy/set?coll=invalid')
+        assert resp.content_type == 'application/json'
+        assert resp.json == {'ip': '127.0.0.1', 'coll': 'invalid', 'ts': None}
+
+    def test_proxy_ip_invalid_coll(self):
+        resp = self.get_url('http://www.iana.org/', status=500)
+        assert 'Invalid Proxy Collection Specified: invalid' in resp.body
+
+