26 May
2012
26 May
'12
2:52 p.m.
I am writing a macro that performs a relatively expensive calculation
to generate content for display (a PNG image) based on user input. The
basic macro works fine, but I'd like to cache the generated content to
save recalculating it when I don't need to.
My idea is that I generate a SHA1 digest of the macro content, and
persistently cache the output generated for that input. I'm thinking
of using an LRU cache to avoid keeping massive amounts of data that'll
never get used again. The question I have is where I can put the cache
so that it is saved with the page, and I can access and update it from
the macro code. I suspect that the answer is to attach an object to
the page somehow, but I've been looking round the documentation and
can't find any examples of how I might do this (TBH, I'm struggling to
find my way round the documentation, so it may be that I've missed
something obvious - if so, then by all means point me to the right
place :-))
As the object is a PNG image, I'm currently base64-encoding it, and
using a data URL to display it. But if it's possible to cache the data
in such a way that the raw data is (or can be made) directly
addressible via a URL, that would be even better.
Can anyone help?
Thanks,
Paul.
26 May
26 May
3:02 p.m.
Paul
Check the cache macro which you can use inside your own macro
http://extensions.xwiki.org/xwiki/bin/view/Extension/Cache+Macro
Envoyé de mon iPhone
Le 26 mai 2012 à 14:52, Paul Moore <p.f.moore(a)gmail.com> a écrit :
I am writing a macro that performs a relatively
expensive calculation
to generate content for display (a PNG image) based on user input. The
basic macro works fine, but I'd like to cache the generated content to
save recalculating it when I don't need to.
My idea is that I generate a SHA1 digest of the macro content, and
persistently cache the output generated for that input. I'm thinking
of using an LRU cache to avoid keeping massive amounts of data that'll
never get used again. The question I have is where I can put the cache
so that it is saved with the page, and I can access and update it from
the macro code. I suspect that the answer is to attach an object to
the page somehow, but I've been looking round the documentation and
can't find any examples of how I might do this (TBH, I'm struggling to
find my way round the documentation, so it may be that I've missed
something obvious - if so, then by all means point me to the right
place :-))
As the object is a PNG image, I'm currently base64-encoding it, and
using a data URL to display it. But if it's possible to cache the data
in such a way that the raw data is (or can be made) directly
addressible via a URL, that would be even better.
Can anyone help?
Thanks,
Paul.
_______________________________________________
users mailing list
users(a)xwiki.org
http://lists.xwiki.org/mailman/listinfo/users
4:34 p.m.
On 26 May 2012 14:02, Ludovic Dubost <ludovic(a)xwiki.com> wrote:
Paul
Check the cache macro which you can use inside your own macro
http://extensions.xwiki.org/xwiki/bin/view/Extension/Cache+Macro
Interesting, thanks. I'm not quite sure how it would work in practice,
it seems that I'd end up with macros nested up to 3 times, at which
point my brain explodes :-)
It might help if I show the content of my current macro here:
{{groovy}}
import java.security.*
// Calculate SHA1 of content - the idea is that we can avoid
regenerating if we've calculated this output before
def content = xcontext.macro.content
def messageDigest = MessageDigest.getInstance("SHA1")
messageDigest.update(content.getBytes())
def contentSHA1 = new BigInteger(1,
messageDigest.digest()).toString(16).padLeft(40, '0')
// Generate the output - ideally, only if we don't have saved output
for this SHA1
def cmd = 'generate-png.exe'
def proc = cmd.execute()
proc << content
proc.out.close()
def png = proc.in.bytes
def out = png.encodeBase64().toString()
println "{{html}}"
println "<img src=\"data:image/png;base64,$out\"/>"
println "{{/html}}"
{{/groovy}}
It's running that executable that I want to skip - it will always give
the same output for the same input. The output should be persistent
(over restarts of XWiki, ideally). As a concrete example of
generate-png.exe, think of the Graphviz "dot" command. For any given
graph, the PNG won't change. Essentially, that's what I'm implementing
here (longer term, the idea would be that the precise command to
generate the image would be configurable).
I've looked at the PlantUML and Graphviz extensions, but I don't
really follow how they work. On the other hand, they seem pretty
complex compared to the code above, so I'm not sure how much
understanding them would help.
Paul.
27 May
27 May
5:11 a.m.
On 05/26/2012 08:52 AM, Paul Moore wrote:
I am writing a macro that performs a relatively
expensive calculation
to generate content for display (a PNG image) based on user input. The
basic macro works fine, but I'd like to cache the generated content to
save recalculating it when I don't need to.
My idea is that I generate a SHA1 digest of the macro content, and
persistently cache the output generated for that input. I'm thinking
of using an LRU cache to avoid keeping massive amounts of data that'll
never get used again. The question I have is where I can put the cache
so that it is saved with the page, and I can access and update it from
the macro code. I suspect that the answer is to attach an object to
the page somehow, but I've been looking round the documentation and
can't find any examples of how I might do this (TBH, I'm struggling to
find my way round the documentation, so it may be that I've missed
something obvious - if so, then by all means point me to the right
place :-))
As the object is a PNG image, I'm currently base64-encoding it, and
using a data URL to display it. But if it's possible to cache the data
in such a way that the raw data is (or can be made) directly
addressible via a URL, that would be even better.
Can anyone help?
Thanks,
Paul.
Some resources that you can use:
https://github.com/xwiki/xwiki-platform/blob/master/xwiki-platform-core/xwi…
is the proper way of creating caches. You declare an @Inject dependency
on that, then use the createNewCache method (if you don't plan on
running a cluster, or don't think that the cached resources should be
shared inside a cluster, then it's slightly better for performance to
call createNewLocalCache). You can then use this cache to store some
entries.
https://github.com/xwiki/xwiki-commons/blob/master/xwiki-commons-core/xwiki…
is going to give you the location of the temporary storage directory.
Combine this with
https://github.com/xwiki/xwiki-platform/blob/master/xwiki-platform-core/xwi…
and you have a nice way of serving image files stored on the disk to the
HTTP client. Unfortunately, this way you have to implement your own LRU
functionality and make sure you clean up files, otherwise you'll keep
filling up the disk with temporary files.
--
Sergiu Dumitriu
http://purl.org/net/sergiu/
9:28 a.m.
On 27 May 2012 04:11, Sergiu Dumitriu <sergiu(a)xwiki.com> wrote:
Some resources that you can use:
https://github.com/xwiki/xwiki-platform/blob/master/xwiki-platform-core/xwi…
is the proper way of creating caches. You declare an @Inject dependency on
that, then use the createNewCache method (if you don't plan on running a
cluster, or don't think that the cached resources should be shared inside a
cluster, then it's slightly better for performance to call
createNewLocalCache). You can then use this cache to store some entries.
https://github.com/xwiki/xwiki-commons/blob/master/xwiki-commons-core/xwiki…
is going to give you the location of the temporary storage directory.
Combine this with
https://github.com/xwiki/xwiki-platform/blob/master/xwiki-platform-core/xwi…
and you have a nice way of serving image files stored on the disk to the
HTTP client. Unfortunately, this way you have to implement your own LRU
functionality and make sure you clean up files, otherwise you'll keep
filling up the disk with temporary files.
That's brilliant - thanks for those, I'll go and do some reading now.
Paul.
29 May
29 May
11:48 p.m.
On 27 May 2012 08:28, Paul Moore <p.f.moore(a)gmail.com> wrote:
One quick question. I gather from what I read that this uses the "old
way" of creating actions which involves registering them in the struts
xml file. (I hope I got that right, I don't really understand all of
this stuff very well). I saw a few mentions of the possibility of
new-style components being able to register custom actions - but I got
the impression that this was a plan rather than a reality. One of the
problems I've found with looking for information about XWiki on the
web is that a lot of things are somewhat out of date - so my question,
is there a new-style way of creating a component that includes an
action similar to the TempResourceAction code here?
But that's just to save having to write messy tempfile-maintenance
code, and is really very much a "nice to have".
Thanks,
Paul.
https://github.com/xwiki/xwiki-platform/blob/master/xwiki-platform-core/xwi…
and you have a nice way of serving image files stored on the disk to the
HTTP client. Unfortunately, this way you have to implement your own LRU
functionality and make sure you clean up files, otherwise you'll keep
filling up the disk with temporary files.
That's brilliant - thanks for those, I'll go and do some reading now.
4493
days inactive
4496
days old
5 comments
3 participants
participants (3)
-
Ludovic Dubost -
Paul Moore -
Sergiu Dumitriu