On 23/11/06, Volker Kuhlmann <[EMAIL PROTECTED]> wrote:
> It's not american size paper that is the problem. The problem is that
> the generated pdf somehow gets the wrong size - it gets the size of
> the .eps that is embedded in the page, instead of the size of the page
> itself.
That very much looks like screwy eps to me. I think you'll find that the
eps is actually illegal - it contains instructions for page construction
which shouldn't be used for a contained graphic.
The eps was created by dot from the graphviz package. There are plenty
of other dot-created eps graphics included as well - and those pages
don't cause problems. I think you are right though - something in that
eps is screwy. The only thing that is on the problem page is the eps
and a caption and a page number.
The weird thing is that, as a postscript page, the viewers I've tried
render it correctly. But when converted to pdf *by the same software
that sees it correctly* (i.e. ghostscript) the top and bottom of the
page gets truncated. How can ghostscript know the correct page
boundaries to show it as a ps, but then pass the boundaries of the
included eps file as the boundaries of the page when it creates pdf?
Even if I tell it to create a4?
Also, if I don't split up the document into each page, every single
page of the generated pdf has the truncated boundaries, not just the
two pages that are the real problem.
I found the two problem pages by splitting the ps file into separate
files for each page, then converting each file to pdf. (I'm so glad I
put the effort in way back when to learn the unix command line! Try
doing that in a GUI for a large document).
You can usually find the eps inside a ps easily. Copy it into a new
file, and run ps2epsi (ghostscript) and/or ps2eps (some perl script I
have in /usr/local/bin, by Roland Bless, v 1.47) on it. Or fix it
yourself if you can. Then copy it back into the ps.
What was the creation path of your ps? Does that lend an angle of
attack?
latex. But the latex source is not easily reconstructible, I think.
> Haven't tried any others yet. I think most are fnt ends to distill or
> ps2pdf anyway.
Possibly, but there are other pdf libraries in use as well.
Will have a look this evening when I get home.
Cheers,
Carl.
