Misc Fixes for RC5 (#534)

* misc fixes (rc 5):
- banner: only auto init banner if not in top-frame (check for no-frame mode and replay url is set)
- index: 'cdx+' fix for use as internal index: if cdx has a warc filename and offset, don't attempt default live web load
- improved self-redirect: avoid www2 -> www redirect altogether, not just for second redirect
- tests: update tests for improved self-redirect checking
- bump version to pywb-2.4.0-rc5

pywb/static/default_banner.js

@@ -306,8 +306,8 @@ This file is part of pywb, https://github.com/webrecorder/pywb
   // all banners will expose themselves by adding themselves as WBBanner on window
   window.WBBanner = new DefaultBanner();
 
-  // if in replay frame, init immediately
-  if (window.wbinfo) {
+  // if wbinfo.url is set and not-framed, init banner in content frame
+  if (window.wbinfo && window.wbinfo.url && !window.wbinfo.is_framed) {
     if (document.readyState === "loading") {
       document.addEventListener("DOMContentLoaded", function() {
         window.WBBanner.init();

pywb/version.py

@@ -1,4 +1,4 @@
-__version__ = '2.4.0rc4'
+__version__ = '2.4.0-rc5'
 
 if __name__ == '__main__':
     print(__version__)

pywb/warcserver/resource/responseloader.py

@@ -139,18 +139,19 @@ def raise_on_self_redirect(self, params, cdx, status_code, location_url):
         request_url = request_url.split('://', 1)[-1].rstrip('/')
 
         self_redir = False
+        orig_key = params.get('sr-urlkey') or cdx['urlkey']
 
         if request_url == location_url:
             self_redir = True
-        elif params.get('sr-urlkey'):
-            # if new location canonicalized matches old key, also self-redirect
-            if canonicalize(location_url) == params.get('sr-urlkey'):
-                self_redir = True
+
+        # if new location canonicalized matches old key, also self-redirect
+        elif canonicalize(location_url) == orig_key:
+            self_redir = True
 
         if self_redir:
             msg = 'Self Redirect {0} -> {1}'
             msg = msg.format(request_url, location_url)
-            params['sr-urlkey'] = cdx['urlkey']
+            params['sr-urlkey'] = orig_key
             raise LiveResourceException(msg)
 
     @staticmethod
@@ -267,6 +268,9 @@ def __init__(self, forward_proxy_prefix=None, adapter=None):
             self.socks_proxy = None
 
     def load_resource(self, cdx, params):
+        if cdx.get('filename') and cdx.get('offset') is not None:
+            return None
+
         load_url = cdx.get('load_url')
         if not load_url:
             return None

pywb/warcserver/test/test_handlers.py

@@ -220,8 +220,8 @@ def test_agg_select_mem_unrewrite_headers(self):
         buff = BytesIO(resp.body)
         record = ArcWarcRecordLoader().parse_record_stream(buff, no_record_parse=False)
         print(record.http_headers)
-        assert record.http_headers.get_statuscode() == '302'
-        assert record.http_headers.get_header('Location') == 'https://www.iana.org/'
+        assert record.http_headers.get_statuscode() == '200'
+        #assert record.http_headers.get_header('Location') == 'https://www.iana.org/'
 
     @patch('pywb.warcserver.index.indexsource.MementoIndexSource.get_timegate_links', MementoOverrideTests.mock_link_header('select_live'))
     def test_agg_select_live(self):

tests/test_redirects.py

@@ -15,7 +15,7 @@ class TestRedirects(CollsDirMixin, BaseConfigTest):
     def setup_class(cls):
         super(TestRedirects, cls).setup_class('config_test.yaml')
 
-    def create_redirect_record(self, url, redirect_url, timestamp):
+    def create_redirect_record(self, url, redirect_url, timestamp, status='301'):
         warc_headers = {}
         warc_headers['WARC-Date'] = timestamp_to_iso_date(timestamp)
 
@@ -26,7 +26,7 @@ def create_redirect_record(self, url, redirect_url, timestamp):
                         ('Location', redirect_url)
                        ]
 
-        http_headers = StatusAndHeaders('301 Permanent Redirect', headers_list, protocol='HTTP/1.0')
+        http_headers = StatusAndHeaders(status + ' Redirect', headers_list, protocol='HTTP/1.0')
 
         rec = self.writer.create_warc_record(url, 'response',
                                              payload=BytesIO(payload),
@@ -140,4 +140,26 @@ def test_revisit_redirect_skip_self_redir(self, fmod):
         res = self.get('/redir/20190626101112{0}/http://www.example.com/', fmod, status=200)
         assert res.text == 'Some Text'
 
+    def test_init_2(self):
+        filename = os.path.join(self.root_dir, 'redir2.warc.gz')
+        with open(filename, 'wb') as fh:
+            self.writer = WARCWriter(fh, gzip=True)
+
+            redirect = self.create_redirect_record('http://www.example.com/path', 'https://www.example.com/path/', '20191003115920')
+            redirect = self.create_redirect_record('https://www.example.com/path/', 'https://www2.example.com/path', '20191003115927', status='302')
+            response = self.create_response_record('https://www2.example.com/path', '20191024125646', 'Some Text')
+            revisit = self.create_revisit_record('https://www2.example.com/path', '20191024125648', 'https://www2.example.com/path', response.rec_headers['WARC-Date'])
+
+        wb_manager(['init', 'redir2'])
+
+        wb_manager(['add', 'redir2', filename])
+
+        assert os.path.isfile(os.path.join(self.root_dir, self.COLLS_DIR, 'redir2', 'indexes', 'index.cdxj'))
+
+    def test_revisit_redirect_skip_self_redir_2(self, fmod):
+        res = self.get('/redir2/20191024125648{0}/http://www2.example.com/path', fmod, status=200)
+        assert res.text == 'Some Text'
+
+        res = self.get('/redir2/20191024125648{0}/https://www.example.com/path', fmod, status=200)
+        assert res.text == 'Some Text'