Re: [Pagekit-users] page postprocessors other than FOP?

[Boris Z. <bz...@2b...>]

Hi Shimon,

Am Dienstag, 16. M=E4rz 2004 06:26 schrieb Shimon Rura:
> I'd like to produce PDF views of some of my web pages, but rather than
> using Apache FOP I'd like to use LaTeX.  I would produce LaTeX code usi=
ng
> template files, and then feed that into pdflatex, which would output PD=
F
> for the client.
>
> Is there a way I can setup pdflatex (or probably a wrapper shell script=
 for
> it) as a postprocessor for a certain pagekit view?  Does PDF generation
> with FOP have to be a special case or is there a generic way to setup a
> postprocessor for a certain view?
>

The pdf generation with FOP is currently somewhat special ( and ugly ). I=
f you=20
have special requirement let us know and we can change the behavior.

> I think I could hack this using a specialized (fake) fop_command settin=
g,
> but that is kinda dirty.  I was thinking I could set fop_command to poi=
nt
> at a postprocessor-wrapper which would look at the input file and decid=
e
> based on that what actual processor it should feed the input to.
>
> Suggestions?

Currently it looks best to me to fake the fop_command setting. Just outpu=
t=20
'[ERROR]:' at the beginning of a line in error case. End the script with =
exit=20
0. Thats all.

A largely better idea is to use the pkit_output_filter method for that ki=
nd of=20
things. But this is currently _not_ working for pdf's. Sure you can, for=20
testing purpose, just do a cat infile >outfile.

For a better general API for that kind of things I suggest a view to mime=
type=20
mapping followed by a pkit_output_filter. That looks easy to me and gives=
=20
full power back to the user.

The drawback is that all pages within a view are handled with the same=20
pkit_output_filter.=20

>
> shimon.
>

--=20
Boris