Ricardo Newbery wrote:
> 
> On Aug 13, 2007, at 11:42 AM, Wichert Akkerman wrote:
> 
>> I was working on a project today which needed a special caching setup: 
>> content in http://domain/folder/* was also available as 
>> http://folder.domain/*. Adding a script to each rule did not really 
>> appeal to me so I went for a different approach which may also help 
>> clean up the codebase a bit.
>>
>> This is what I did: I added a subscription adapter which can be asked 
>> for URLs that should be purged. It has two methods:
>>
>>  getRelativeUrls
>>     this method returns a list of relative URLs to purge. This could
>>     be used to replace all the hardcoded tests that are currently found
>>     in the content rule
>>
>>  getAbsoluteUrls
>>     I added this because I needed it and I suspected that it can be
>>     useful for others as well: it takes the list of all relative URLs
>>     that are going to be purged and can return a list of absolute URLs
>>     that should also be purged. These URLs will not be post-processed by
>>     the cache tool
>>
>> I'm quite happy with the result: the amount of code needed to support 
>> this is really tiny but it adds a lot of flexibility.
>>
>> Wichert.
> 
> 
> Thanks, this looks like a fantastic addition.  Two questions:
> 
> 
> #1 - Could there be any issues with this code in legacy Plone 2.5 
> installations.  I haven't checked yet but is 
> "zope.component.subscribers" available on all Zope versions allowed by 
> the Plone 2.5 series.  If not, perhaps we need to wrap this in a test.  
> Maybe something like:
> 
>    if getattr(zope.component, 'subscribers', None) is not None:

No, I wrote this code for a Plone 2.5 site so I know it works.

> #2 - Could you write up a short paragraph or two on how to extend a 
> content type to use these adapters?  I would like to include it in the 
> manual's chapter on customizing CacheFu.

To use this you only need to do two things: create a class with those 
two methods and register that with the Zope component architecture. For 
example if you have a class which exposes a 'summary' URL as well you 
can use this code:

clas MyTypePurger(object):
     adapts(IMyType)
     implement(IPurgeUrls)

     def __init__(self, context):
         self.context=context

     def getRelativeUrls(self):
         context=self.context
         portal=getToolByName(context, "portal_url").getPortalObject()
         relpath=context.getPhysicalPath()[len(portal.getPhysicalPath()):]
         relpath="/".join(relpath)+"/"

         return [relpath+"/summary"]

      def getAbsoluteUrls(self, relativeUrls):
         return []


To register this with the component architecture add this to your 
configure.zcml:

      <subscriber
          provides="Products.CacheSetup.interfaces.IPurgeUrls"
          factory=".purgers.MyTypePurger"
          />

Wichert.

-- 
Wichert Akkerman <wichert@...>   It is simple to make things.
http://www.wiggy.net/                  It is hard to make things simple.