Re: PDFPIC misbehaves if called two times in a row

[joerg van den hoff]

ok, I did what you've proposed. *none* of the commands in that pipe fails. even 
the complete pipe
including `sed` yields exist status 0 but outputs

Page size:       510 x 318 pts

i.e. the unmodified output from the initial grep. this is on OSX rather than 
linux.

as already stated in previous mail to deri: it seems that "my" sed does want 
single backslashes
for defining the parenthesis around the backreference patterns (if there is not 
a further level
of protection necessary when using the script in the `.sy' call in pdfpic.tmac).
on the shell prompt, I get the correct result here, when using

pdfinfo yourfile.pdf | tr -d '\000' | grep "Page *size" \
  | sed -e 's/Page *size: *\([[:digit:].]*\) *x *\([[:digit:].]*\).*$/\
.nr pdfpic*width (p;\\1)\\n\
.nr pdfpic*height  (p;\\2)/' || echo failed, status $?

(note `\(...\)' instead of `\\(...\\)').

no idea, why exactly it does not work for me but seemingly does work for 
everybody else out of the box?

as said, for the time being I have settled at replacing that grep|sed thingy by 
a simple(r) awk call:

awk '/Page[ ]+size:/ {\
   print ".nr pdfpic*width (p;" $3 ")";\
   print ".nr pdfpic*height (p;" $5 ")";\
}' > @*[pdfpic*temporary-file]

so problem solved temporarily (until next update run of the package manager ;)).

best,
joerg

...
> I would try running the command that the `PDFPIC` macro is running.
> Build up the pipeline step by step and check the exit status at every
> point.  (I have the exit status of the previous command in my shell
> prompt for this reason.)
>
> pdfinfo yourfile.pdf || echo failed, status $?
> pdfinfo yourfile.pdf | tr -d '\000' || echo failed, status $?
> pdfinfo yourfile.pdf | tr -d '\000' | grep "Page *size" \
>    || echo failed, status $?
> pdfinfo yourfile.pdf | tr -d '\000' | grep "Page *size" \
>    | sed -e 's/Page *size: *\\([[:digit:].]*\\) *x *\\([[:digit:].]*\\).*$/\
> .nr pdfpic*width (p;\\1)\\n\
> .nr pdfpic*height  (p;\\2)/' || echo failed, status $?
>
> > so no inclusion of pictures at all...?
>
> Since the `PDFPIC` macro is failing to perform an essential
> operation--it has to know how big the image is--that's right.
>
> However it should be possible to sort this out with the procedure above.
> It is a combination of Bernd Warken's code and mine; a true
> Frankenstein's monster, in other words.  But without a PDF interpreter
> built in to the GNU troff formatter or requiring the user to do
> preparatory work to create PDF or image file descriptions out-of-band, I
> can't imagine any other way this work could get done.
>
> Some details may need to be fiddled, of course.
>
> Regards,
> Branden
>
> [1] ...or string, or diversion...