including xhtml questions

[Tim Arnold <jtim.arnold@gmail.com>]

Hi,
I create DocBook 5 XML and use the html stylesheets to produce html.
I  also use the dbhtml file= processing instruction to insert some
xhtml files that have been externally generated. Unfortunately I have
no control over the external process and the xhtml contains  
entities and duplicate IDs.
I have been postprocessing the files to remove these before running
the DocBook html stylesheets (using a Python script).

I'd rather not do that because
(a) it adds another external step to the build process and
(b) I have to include these modified xhtml files if I need to deliver
the DocBook content to others.

My questions:
Is there any method I can use to create a valid DocBook XML file that
doesn't require me to ship the external xhtml files?
If not, then would it be better (maintenance, simplicity) to create a
custom processing instruction to include these files instead of using
Python?

thanks,
--Tim Arnold