seq handler: option to filter out bad responses (not 2xx or 3xx) and proceed to next handler. if at last handler, return error response

dockerfile: remove volume, port from base image
error page: add cors headers for error page

Dockerfile

@@ -19,11 +19,11 @@ RUN python setup.py install
 RUN mkdir /webarchive
 COPY config.yaml /webarchive/
 
-VOLUME /webarchive
+#VOLUME /webarchive
 
 WORKDIR /webarchive
 
-EXPOSE 8080
+#EXPOSE 8080
 
 CMD ["uwsgi", "/uwsgi/uwsgi.ini"]
 

pywb/apps/rewriterapp.py

@@ -467,7 +467,9 @@ class RewriterApp(object):
     def _not_found_response(self, environ, url):
         resp = self.not_found_view.render_to_string(environ, url=url)
 
-        return WbResponse.text_response(resp, status='404 Not Found', content_type='text/html')
+        return WbResponse.text_response(resp, status='404 Not Found',
+                                        content_type='text/html',
+                                        headers=[('Access-Control-Allow-Origin', '*')])
 
     def _error_response(self, environ, msg='', details='', status='404 Not Found'):
         resp = self.error_view.render_to_string(environ,

pywb/apps/wbrequestresponse.py

@@ -43,11 +43,16 @@ class WbResponse(object):
         return WbResponse(status_headers, value=stream)
 
     @staticmethod
-    def text_response(text, status='200 OK', content_type='text/plain; charset=utf-8'):
+    def text_response(text, status='200 OK', content_type='text/plain; charset=utf-8', headers=None):
         encoded_text = text.encode('utf-8')
-        status_headers = StatusAndHeaders(status,
+
-                                          [('Content-Type', content_type),
+        def_headers = [('Content-Type', content_type),
-                                           ('Content-Length', str(len(encoded_text)))])
+                       ('Content-Length', str(len(encoded_text)))]
+
+        if headers:
+            def_headers += headers
+
+        status_headers = StatusAndHeaders(status, def_headers)
 
         return WbResponse(status_headers, value=[encoded_text])
 

pywb/warcserver/basewarcserver.py

@@ -7,11 +7,13 @@ from werkzeug.exceptions import HTTPException
 import requests
 import traceback
 import json
-
+import logging
 import six
 
 JSON_CT = 'application/json; charset=utf-8'
 
+logger = logging.getLogger('warcserver')
+
 
 #=============================================================================
 class BaseWarcServer(object):
@@ -19,6 +21,9 @@ class BaseWarcServer(object):
         self.route_dict = {}
         self.debug = kwargs.get('debug', False)
 
+        if self.debug:
+            logger.setLevel(logging.DEBUG)
+
         self.url_map = Map()
 
         def list_routes(environ):

pywb/warcserver/handlers.py

@@ -84,7 +84,7 @@ class IndexHandler(object):
             return None, None, errs
 
         cdx_iter, errs = self._load_index_source(params)
-        if not cdx_iter:
+        if not cdx_iter or errs:
             return None, None, errs
 
         content_type, res = handler(cdx_iter, fields)
@@ -152,8 +152,9 @@ class DefaultResourceHandler(ResourceHandler):
 
 #=============================================================================
 class HandlerSeq(object):
-    def __init__(self, handlers):
+    def __init__(self, handlers, filter_errors=True):
         self.handlers = handlers
+        self.filter_errors = filter_errors
 
     def get_supported_modes(self):
         if self.handlers:
@@ -163,12 +164,26 @@ class HandlerSeq(object):
 
     def __call__(self, params):
         all_errs = {}
+        err_res = None
+        err_out_headers = None
         for handler in self.handlers:
             out_headers, res, errs = handler(params)
+            if out_headers and self.filter_errors:
+                status = out_headers.get('Warcserver-Status')
+                if status and not status.startswith(('1', '2', '3')):
+                    errs = {'status_error': status}
+                    err_res = res
+                    err_out_headers = out_headers
+                    res = None
+
             all_errs.update(errs)
+
             if res is not None:
                 return out_headers, res, all_errs
 
+        if err_res and err_out_headers:
+            return err_out_headers, err_res, all_errs
+
         return None, None, all_errs
 
 

pywb/warcserver/resource/responseloader.py

@@ -56,6 +56,9 @@ class BaseLoader(object):
 
         out_headers['Warcserver-Cdx'] = to_native_str(cdx.to_cdxj().rstrip())
         out_headers['Warcserver-Source-Coll'] = to_native_str(source)
+        status = cdx.get('status')
+        if status:
+            out_headers['Warcserver-Status'] = str(status)
 
         if not warc_headers:
             if other_headers:
@@ -318,6 +321,8 @@ class LiveWebLoader(BaseLoader):
                                status=upstream_res.status,
                                reason=upstream_res.reason)
 
+        cdx['status'] = upstream_res.status
+
         http_headers_buff = status
 
         orig_resp = upstream_res._original_response