Bug: web public url_filter

[wmoore]

I have these values set up

jrs-macbookpro-25107:OMERO will$ omero config get
omero.web.debug=True
omero.web.public.enabled=True
omero.web.public.password=ome
omero.web.public.url_filter=webtest
omero.web.public.user=emdb_public

However, when I go to localhost:8000/webtest
I don't get logged in (redirected to webclient login).

If I do this, then the public login works for that url:

diff --git a/components/tools/OmeroWeb/omeroweb/decorators.py b/components/tools/OmeroWeb/omeroweb/decorators.py
index 7acbf48..bf41ef1 100644
--- a/components/tools/OmeroWeb/omeroweb/decorators.py
+++ b/components/tools/OmeroWeb/omeroweb/decorators.py
@@ -155,7 +155,9 @@ class login_required(object):
         Verifies that the URL for the resource being requested falls within
         the scope of the OMERO.webpublic URL filter.
         """
-        return settings.PUBLIC_URL_FILTER.match(request.path) is not None
+        p = request.path.strip("/")
+        return settings.PUBLIC_URL_FILTER.match(p) is not None
 
     def get_connection(self, server_id, request):

I'm no regex expert, but I'm sure there's a nicer way to fix this one?

[jburel]

Moved from sprint 2012-06-19 (17)

[cxallan]

Closed by upstream commit on sprint18-bugfixes:

• ​https://github.com/aleksandra-tarkowska/openmicroscopy/commit/85f04ef083af284934100da75fa05580c472338e

[Chris Allan <callan@…>]

(In [41c8f373f925e8d552ed64600d7810b5489ba2ca/ome.git] on branch develop) Add a safety net for partial OMERO.webpublic configuration. (See #9139)

[Chris Allan <callan@…>]

(In [85f04ef083af284934100da75fa05580c472338e/ome.git] on branch develop) Switch OMERO.webpublic URL filter to re.search. (Fixes #9139)

Previously the OMERO.webpublic URL filter was using re.match, which
would not allow more flexible regular expression matching. Switching to
re.search means that "exotic" matching techniques can be used, but does
mean that more explicit regular expressions are needed when attempting
to filter based on a base URL:

• 'webtest' matches '/webtest' but also '/webclient/webtest'
• 'dataset' matches '/webtest/dataset' and also '/webclient/dataset'
• '/webtest' matches '/webtest...' but also '/webclient/webtest'
• '^{/webtest' matches '/webtest...' but not '/webclient/webtest'}

We'll have to be very explicit about this in the documentation if we
choose to keep the re.search.