replay/memento: always include 'Content-Location' for in no-redir mode replay (not just for memento timegate), #122

pywb/framework/memento.py

@@ -106,10 +106,10 @@ class MementoRespMixin(object):
                                                                timestamp=ts,
                                                                url=url)
 
-                # Must set content location
-                if is_memento and is_timegate:
-                    self.status_headers.headers.append(('Content-Location',
-                                                        canon_link))
+                # set in replay_views -- Must set content location
+                #if is_memento and is_timegate:
+                #    self.status_headers.headers.append(('Content-Location',
+                #                                        canon_link))
 
                 # don't set memento link for very long urls...
                 if len(canon_link) < 512:

pywb/webapp/replay_views.py

@@ -191,6 +191,16 @@ class ReplayView(object):
                                                        response_iter,
                                                        self.buffer_max_size)
 
+        # Set Content-Location if not exact capture
+        if not self.redir_to_exact:
+            mod = wbrequest.options.get('replay_mod', wbrequest.wb_url.mod)
+            canon_url = (wbrequest.urlrewriter.
+                         get_new_url(timestamp=cdx['timestamp'],
+                                     url=cdx['url'],
+                                     mod=mod))
+
+            status_headers.headers.append(('Content-Location', canon_url))
+
         response = self.response_class(status_headers,
                                        response_iter,
                                        wbrequest=wbrequest,

tests/test_integration.py

@@ -303,23 +303,23 @@ class TestWb:
 
     def test_redirect_non_exact_latest_replay_ts(self):
         resp = self.testapp.get('/pywb-non-exact/http://example.com/')
-        assert resp.status_int == 302
+        assert resp.status_int == 200
 
-        assert resp.headers['Location'].endswith('/http://example.com')
+        assert resp.headers['Content-Location'].endswith('/http://example.com')
 
         # extract ts, which should be current time
-        ts = resp.headers['Location'].rsplit('/http://')[0].rsplit('/', 1)[-1]
-        assert len(ts) == 14, ts
-        resp = resp.follow()
+        ts = resp.headers['Content-Location'].rsplit('/http://')[0].rsplit('/', 1)[-1]
+        assert ts == '20140127171251'
+        #resp = resp.follow()
 
-        self._assert_basic_html(resp)
+        #self._assert_basic_html(resp)
 
         # ensure the current ts is present in the links
         assert '"{0}"'.format(ts) in resp.body
-        assert '/pywb-non-exact/{0}/http://www.iana.org/domains/example'.format(ts) in resp.body
+        assert '/pywb-non-exact/http://www.iana.org/domains/example' in resp.body
 
         # ensure ts is current ts
-        assert timestamp_now() >= ts, ts
+        #assert timestamp_now() >= ts, ts
 
     def test_redirect_relative_3(self):
         # webtest uses Host: localhost:80 by default