You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@cocoon.apache.org by Jonas Petersen <pe...@pixelpark.com> on 2002/06/12 19:20:42 UTC
AcceptSelector
Hi,
I am working on a selector that analyses the Accept header of the
browser and provides a way to (for example) response with the best
fitting content(-type). It's comparable with the apache module
'mod_negotiation'.
The main class is named 'AcceptSelector'. It is based on the abstract
class 'WeightedTypesSelector'.
The (sitemap) configuration of the AcceptSelector is made up of a list
of possible variants to check against. Each variant has a name, a
mime-type and a weight.
Example of a Accept header sent by the browser:
---------------------------------------------------------------------
Accept: text/xml, application/xml, application/xhtml+xml,
text/html;q=0.9, image
/png, image/jpeg, image/gif;q=0.2, text/plain;q=0.8, text/css, */*;q=0.1
---------------------------------------------------------------------
Example of a sitemap configuration:
---------------------------------------------------------------------
<map:selector logger="sitemap.selector.accept" name="accept"
src="org.apache.cocoon.selection.AcceptSelector">
<variant name="wml" mime-type="application/vnd.wap.wmlc" qs="0.9"/>
<variant name="html" mime-type="text/html" qs="0.8"/>
<variant name="wildcard" mime-type="*/*" qs="1.0"/>
</map:selector>
---------------------------------------------------------------------
The selector multiplies the weights of the mime-types in the Accept
header (q=...) with the weights of the variants in the sitemap (qs=...)
(if no weight is given a value of 1.0 is used). The variant with the
highest value gets selected. If multiple variants have the same result,
the variant that appears first gets selected.
In the above example the variant 'html' would make it, because the
mime-type 'text/html' is found in the header and the config and yields a
weight of 0.72 (0.9 * 0.8). 'wml' matches with '*/*' (0.1 * 0.9 = 0.09).
Using the mime-type '*/*' as a variant helps in providing default
variants (it behaves differently than <map:otherwise/>).
Example of a <map:select/> using the AcceptSelector:
---------------------------------------------------------------------
<map:match pattern="foo">
<map:select type="accept">
<map:when test="wml">
<map:generate src="foo.wml"/>
<map:serialize type="wml"/>
</map:when>
<map:when test="html">
<map:generate src="foo.html"/>
<map:serialize type="html"/>
</map:when>
<map:otherwise>
<map:generate src="foo-baz.html"/>
<map:serialize type="html"/>
</map:otherwise>
</map:select>
</map:match>
---------------------------------------------------------------------
Now a problem that I encountered:
For each <map:when/> the select() function gets called. There I do the
weight calculations with the current Accept header against the
configuration. This produces some overhead because this really needs to
happen only once (in the beginning of the <map:select/>.
I'm thinking about caching the results for each different Accept header.
Maybe it would be good to do these calculations at the beginning of the
<map:select/> block. But it seems there is no temporary context/instance
within the block or a way to initialize a <map:select/>. Or is there? To
me this seems to be a general problem with the selectors, because if
there is lots of <map:when/>s then there might be lots of unnecessarily
repeated code.
The best would probably be to combine the caching and the precalculation
at the beginning of the block.
So what do you think about it? Is it useful? Would it make sense to
include it into cocoon?
How should I provide the source code so you can check it out? (I'm new
to cocoon-dev.)
regards,
Jonas
---------------------------------------------------------------------
To unsubscribe, e-mail: cocoon-dev-unsubscribe@xml.apache.org
For additional commands, email: cocoon-dev-help@xml.apache.org
Re: AcceptSelector
Posted by Peter Royal <pr...@apache.org>.
On Wednesday 12 June 2002 01:20 pm, Jonas Petersen wrote:
> I am working on a selector that analyses the Accept header of the
> browser and provides a way to (for example) response with the best
> fitting content(-type). It's comparable with the apache module
> 'mod_negotiation'.
awesome!
> Now a problem that I encountered:
>
> For each <map:when/> the select() function gets called. There I do the
> weight calculations with the current Accept header against the
> configuration. This produces some overhead because this really needs to
> happen only once (in the beginning of the <map:select/>.
Already encountered. Look at the Switch Selector:
http://cvs.apache.org/viewcvs/xml-cocoon2/src/java/org/apache/cocoon/selection/SwitchSelector.java
-pete
--
peter royal -> proyal@apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: cocoon-dev-unsubscribe@xml.apache.org
For additional commands, email: cocoon-dev-help@xml.apache.org
Re: AcceptSelector
Posted by Diana Shannon <sh...@apache.org>.
On Wednesday, June 12, 2002, at 01:30 PM, Michael Hartle wrote:
>> How should I provide the source code so you can check it out? (I'm new
>> to cocoon-dev.)
>>
> This sounds interesting, so just go to Bugzilla and add it as a "bug"
> for Cocoon 2, prefixing your Summary with "[PATCH]" and adding your
> code as a zipped attachment; that way, anyone is able to check out your
> code, it will be listed in the regular Patch mail to the cocoon-dev
> list and if there are no objections, your addition can be added to the
> CVS by a committer who has the time to do so.
>
> For an example of such a [PATCH] "bug" for Bugzilla, see
> http://nagoya.apache.org/bugzilla/show_bug.cgi?id=9404. To file your
> patch, go to
> http://nagoya.apache.org/bugzilla/enter_bug.cgi?product=Cocoon%202. You
> will have to create an account, but this goes fast and doesn't hurt ;)
Please check out Cocoon's fine How-To, written by our very own David
Crossley:
http://xml.apache.org/cocoon/howto/howto-bugzilla.html
Diana
---------------------------------------------------------------------
To unsubscribe, e-mail: cocoon-dev-unsubscribe@xml.apache.org
For additional commands, email: cocoon-dev-help@xml.apache.org
Re: AcceptSelector
Posted by Michael Hartle <mh...@hartle-klug.com>.
Jonas Petersen wrote:
<snip/>
>The best would probably be to combine the caching and the precalculation
>at the beginning of the block.
>
>So what do you think about it? Is it useful? Would it make sense to
>include it into cocoon?
>
>How should I provide the source code so you can check it out? (I'm new
>to cocoon-dev.)
>
This sounds interesting, so just go to Bugzilla and add it as a "bug"
for Cocoon 2, prefixing your Summary with "[PATCH]" and adding your code
as a zipped attachment; that way, anyone is able to check out your code,
it will be listed in the regular Patch mail to the cocoon-dev list and
if there are no objections, your addition can be added to the CVS by a
committer who has the time to do so.
For an example of such a [PATCH] "bug" for Bugzilla, see
http://nagoya.apache.org/bugzilla/show_bug.cgi?id=9404. To file your
patch, go to
http://nagoya.apache.org/bugzilla/enter_bug.cgi?product=Cocoon%202. You
will have to create an account, but this goes fast and doesn't hurt ;)
Best regards,
Michael Hartle,
Hartle & Klug GbR
---------------------------------------------------------------------
To unsubscribe, e-mail: cocoon-dev-unsubscribe@xml.apache.org
For additional commands, email: cocoon-dev-help@xml.apache.org