memento timegate: make timegate headers for /<coll>/<url> behave correctly per-memento spec, (#564)

return 404 if not found, return latest memento header. do this by performing actual response lookup,
but then returning the top frame response if succeeded. addresses ukwa/ukwa-pywb#58

pywb/apps/rewriterapp.py

@@ -345,6 +345,7 @@ class RewriterApp(object):
                                                       content_rw, is_proxy)
 
         response = None
+        keep_frame_response = False
 
         # prefer overrides custom response?
         if pref_mod is not None:
@@ -359,14 +360,23 @@ class RewriterApp(object):
                                                  headers=headers)
             else:
                 wb_url.mod = pref_mod
+        else:
+            if kwargs.get('output'):
+                response = self.handle_timemap(wb_url, kwargs, full_prefix)
+
+            elif wb_url.is_query():
+                response = self.handle_query(environ, wb_url, kwargs, full_prefix)
+
             else:
                 # don't return top-frame response for timegate with exact redirects
-            kwargs['is_timegate_redir'] = is_timegate and redirect_to_exact
+                if not (is_timegate and redirect_to_exact):
+                    keep_frame_response = is_timegate and not redirect_to_exact and not is_proxy
                     response = self.handle_custom_response(environ, wb_url,
                                                            full_prefix, host_prefix,
                                                            kwargs)
 
-        if response:
+
+        if response and not keep_frame_response:
             return self.format_response(response, wb_url, full_prefix, is_timegate, is_proxy)
 
         if is_proxy:
@@ -443,6 +453,11 @@ class RewriterApp(object):
 
             return self.send_redirect(new_path, url_parts, urlrewriter)
 
+        # return top-frame timegate response, with timestamp from cdx
+        if response and keep_frame_response:
+            no_except_close(r.raw)
+            return self.format_response(response, wb_url, full_prefix, is_timegate, is_proxy, cdx['timestamp'])
+
         stream = BufferedReader(r.raw, block_size=BUFF_SIZE)
         record = self.loader.parse_record_stream(stream,
                                                  ensure_http_headers=True)
@@ -560,7 +575,7 @@ class RewriterApp(object):
 
         return response
 
-    def format_response(self, response, wb_url, full_prefix, is_timegate, is_proxy):
+    def format_response(self, response, wb_url, full_prefix, is_timegate, is_proxy, timegate_closest_ts=None):
         memento_ts = None
         if not isinstance(response, WbResponse):
             content_type = 'text/html'
@@ -569,13 +584,13 @@ class RewriterApp(object):
             if not self.is_framed_replay(wb_url):
                 content_type += '; charset=utf-8'
             else:
-                memento_ts = wb_url.timestamp
+                memento_ts = timegate_closest_ts or wb_url.timestamp
 
             response = WbResponse.text_response(response, content_type=content_type)
 
         if self.enable_memento and response.status_headers.statusline.startswith('200'):
             self._add_memento_links(wb_url.url, full_prefix, None, memento_ts,
-                                    response.status_headers, is_timegate, is_proxy)
+                                    response.status_headers, is_timegate, is_proxy, is_memento=not is_timegate)
         return response
 
     def _add_memento_links(self, url, full_prefix, memento_dt, memento_ts,
@@ -873,13 +888,7 @@ class RewriterApp(object):
         return {'metadata': kwargs.get('metadata', {})}
 
     def handle_custom_response(self, environ, wb_url, full_prefix, host_prefix, kwargs):
-        if kwargs.get('output'):
-            return self.handle_timemap(wb_url, kwargs, full_prefix)
-
-        if wb_url.is_query():
-            return self.handle_query(environ, wb_url, kwargs, full_prefix)
-
-        if self.is_framed_replay(wb_url) and not kwargs.get('is_timegate_redir'):
+        if self.is_framed_replay(wb_url):
             extra_params = self.get_top_frame_params(wb_url, kwargs)
             return self.frame_insert_view.get_top_frame(wb_url,
                                                         full_prefix,

tests/test_memento.py

@@ -63,6 +63,28 @@ class TestMemento(MementoMixin, BaseConfigTest):
         assert '"20140127171238"' in resp.text
         assert '"http://www.iana.org/"' in resp.text, resp.text
 
+    def test_memento_top_frame_timegate(self):
+        resp = self.testapp.get('/pywb/http://www.iana.org/_css/2013.1/screen.css')
+
+        # vary header
+        assert VARY in resp.headers
+
+        # no memento header, as not really a memento (top-frame)
+        assert MEMENTO_DATETIME not in resp.headers
+
+        # Memento Headers
+        # memento link
+        dt = 'Mon, 27 Jan 2014 17:12:39 GMT'
+        url = 'http://www.iana.org/_css/2013.1/screen.css'
+
+        links = self.get_links(resp)
+
+        assert self.make_memento_link(url, '20140127171239', dt, 'mp_', include_coll=False) in links
+
+        #timegate link
+        assert self.make_timegate_link(url, '') in links
+
+
     def test_memento_content_replay_exact(self, fmod):
         resp = self.get('/pywb/20140127171238{0}/http://www.iana.org/', fmod)
 
@@ -175,6 +197,15 @@ com,example)/?example=1 20140103030341 {"url": "http://example.com?example=1", "
         resp = self._timemap_get('/pywb/timemap/foo/http://example.com', status=400)
         assert resp.json == {'message': 'output=foo not supported'}
 
+    def test_timegate_error_not_found(self):
+        resp = self.testapp.get('/pywb/http://example.com/x-not-found', status=404)
+        assert resp.status_code == 404
+
+        # No Memento Headers
+        assert VARY not in resp.headers
+        assert MEMENTO_DATETIME not in resp.headers
+        assert 'Link' not in resp.headers
+
     def test_error_bad_accept_datetime(self):
         """
         400 response for bad accept_datetime