initial changes to backend url quality check to account for some VHM …

castle/cms/archival.py

@@ -199,7 +199,7 @@ def _get_vhm_base_url(public_url, site_path):
     parsed = urlparse(public_url)
     port = parsed.port
     if port is None:
-        port = 80 if parsed.scheme == 'http' else 403
+        port = 80 if parsed.scheme == 'http' else 443
     return '/VirtualHostBase/{scheme}/{hostname}:{port}{site_path}/VirtualHostRoot'.format(
         scheme=parsed.scheme,
         hostname=parsed.hostname,
@@ -276,7 +276,7 @@ def get_vhm_url(self, url):
         final_path = '/' + '/'.join(parsed_path)
         return self.vhm_base + final_path
 
-    def __call__(self, url, use_vhm=True, require_public_url=True):
+    def __call__(self, url, use_vhm=True, require_public_url=True, root=None):
         url = normalize_url(url)
         if not url:
             return
@@ -298,7 +298,7 @@ def __call__(self, url, use_vhm=True, require_public_url=True):
 
         if use_vhm:
             url = self.get_vhm_url(url)
-        resp = subrequest(url)
+        resp = subrequest(url, root=root)
         if resp.getStatus() == 404:
             return
 

castle/cms/browser/content/__init__.py

@@ -60,6 +60,8 @@
 from zope.component.hooks import getSite
 from zope.container.interfaces import INameChooser
 from zope.interface.declarations import noLongerProvides
+from zope.globalrequest import getRequest
+from zope.globalrequest import setRequest
 
 
 try:
@@ -805,6 +807,7 @@ def get_invalid_backend_urls_found(self, content):
 
 class QualityCheckContent(BrowserView):
     QUALITY_CHECK_URL = '/@@quality-check'
+    ERROR = 'ERROR'
 
     @property
     def formatted_url(self):
@@ -816,26 +819,39 @@ def formatted_url(self):
         except Exception:
             return None
 
-    @property
-    def subrequest_results(self):
-        opener = SubrequestUrlOpener(
-            site=api.portal.get(),
-            check_blacklist=False,
-        )
-        return opener(self.formatted_url, require_public_url=False)
-
     @property
     def contains_backend_urls(self):
-        subrequest_results = self.subrequest_results
-        subrequest_html = subrequest_results['data']
-        backend_utils = BackendUrlUtils()
-        backend_urls_found = backend_utils.get_invalid_backend_urls_found(subrequest_html)
-        if len(backend_urls_found) > 0:
-            logger.warn('There were backend urls found in the html')
-            logger.info('Backend urls found: {}'.format(repr(backend_urls_found)))
-            logger.info('Data searched for: ' + self.formatted_url)
-            return True
-        return False
+        app = api.portal.get()
+        orig_req = getRequest()
+        # we subvert the original request to remove any configured VHM meta that the
+        # plone.subrequest.subrequest() method will use to construct the request it'll
+        # end up using
+        req_for_sub = orig_req.copy() # didn't work locally - maybe clone()?
+        req_for_sub['VIRTUAL_URL_PARTS'] = None
+        setRequest(req_for_sub)
+        try:
+            opener = SubrequestUrlOpener(site=app, check_blacklist=False)
+            # we pass in the app as the "root" object to get a base path from, since it'll just
+            # be an empty path (or a single slash, which will be neglegable)
+            #
+            # the subverted VIRTUAL_URL_PARTS and the portal root object as the root/context to
+            # use combined will mean that the calculated VHM path for the configured frontend
+            # can be used from within the context of a backend url
+            subrequest_results = opener(self.formatted_url, require_public_url=False, root=app)
+            subrequest_html = subrequest_results['data']
+            backend_utils = BackendUrlUtils()
+            backend_urls_found = backend_utils.get_invalid_backend_urls_found(subrequest_html)
+            if len(backend_urls_found) > 0:
+                logger.warn('There were backend urls found in the html')
+                logger.info('Backend urls found: {}'.format(repr(backend_urls_found)))
+                logger.info('Data searched for: ' + self.formatted_url)
+                return True
+            return False
+        except Exception:
+            return self.ERROR
+        finally:
+            # and we revert back to the original request
+            setRequest(orig_req)
 
     @property
     def are_links_valid(self):
@@ -878,7 +894,7 @@ def __call__(self):
             'headersOrdered': self.are_headers_ordered(html),
             'html': html_parser.unescape(html),
             'isTemplate': self.context in get_template_repository_info()['templates'],
-            # 'containsBackendUrls': self.contains_backend_urls,
+            'containsBackendUrls': self.contains_backend_urls,  # boolean or 'ERROR'
         })