replay api: make ReplayView overridable in WBHandler subclass,

allow custom content loader callable

pywb/warc/resolvingloader.py

@@ -9,7 +9,7 @@ class ResolvingLoader:
         self.path_resolvers = make_best_resolvers(paths)
         self.record_loader = record_loader
 
-    def resolve_headers_and_payload(self, cdx, failed_files, cdx_loader):
+    def __call__(self, cdx, failed_files, cdx_loader, *args):
         """
         Resolve headers and payload for a given capture
         In the simple case, headers and payload are in the same record.

pywb/warc/test/test_loading.py

@@ -357,9 +357,7 @@ def load_from_cdx_test(cdx, revisit_func=load_orig_cdx, reraise=False,
     cdx = CDXObject(cdx)
 
     try:
-        (headers, stream) = resolve_loader.resolve_headers_and_payload(cdx,
-                                                                       failed_files,
-                                                                       revisit_func)
+        (headers, stream) = resolve_loader(cdx, failed_files, revisit_func)
         print headers
         sys.stdout.write(stream.readline())
         sys.stdout.write(stream.readline())

pywb/webapp/handlers.py

@@ -112,6 +112,12 @@ class WBHandler(SearchPageWbUrlHandler):
                                create_template(config.get('not_found_html'),
                                'Not Found Error'))
 
+        self.replay = self._init_replay_view(config)
+
+        self.fallback_handler = None
+        self.fallback_name = config.get('fallback')
+
+    def _init_replay_view(self, config):
         cookie_maker = config.get('cookie_maker')
         record_loader = ArcWarcRecordLoader(cookie_maker=cookie_maker)
 
@@ -120,10 +126,7 @@ class WBHandler(SearchPageWbUrlHandler):
         resolving_loader = ResolvingLoader(paths=paths,
                                            record_loader=record_loader)
 
-        self.replay = ReplayView(resolving_loader, config)
-
-        self.fallback_handler = None
-        self.fallback_name = config.get('fallback')
+        return ReplayView(resolving_loader, config)
 
     def resolve_refs(self, handler_dict):
         if self.fallback_name:

pywb/webapp/replay_views.py

@@ -1,4 +1,6 @@
 import re
+import logging
+
 from io import BytesIO
 from urlparse import urlsplit
 
@@ -87,8 +89,9 @@ class ReplayView(object):
                                                       failed_files)
 
             except (CaptureException, ArchiveLoadFailed) as ce:
-                import traceback
-                traceback.print_exc()
+                #import traceback
+                #traceback.print_exc()
+                logging.debug(ce)
                 last_e = ce
                 pass
 
@@ -131,10 +134,10 @@ class ReplayView(object):
         return response
 
     def replay_capture(self, wbrequest, cdx, cdx_loader, failed_files):
-        (status_headers, stream) = (self.content_loader.
-                                    resolve_headers_and_payload(cdx,
-                                                                failed_files,
-                                                                cdx_loader))
+        (status_headers, stream) = (self.content_loader(cdx,
+                                                        failed_files,
+                                                        cdx_loader,
+                                                        wbrequest))
 
         # check and reject self-redirect
         self._reject_self_redirect(wbrequest, cdx, status_headers)