Re: ps2pdf: Embedded files are removed

Hi Ken

Thanks for this feedback which explains the behavior, much appreciated.

I guess the solutionn is to check for aspects that are not supported by the interpreter before sending it to ghostscript as gs will just silently ignore non-supported non-marking aspects. Is there a reference on what aspects are (not) supported by the interpreter?

Best, Andreas

> Am 18.03.2016 um 09:40 schrieb Ken Sharp <[email protected]>:
> 
> At 23:07 17/03/2016 +0100, Andreas Nef wrote:
> 
> 
>> sDEVICE=pdfwrite -sOutputFile=out.pdf PDFA_def.ps input.pdf The thing that I would expect is that depending on the -dPDFACompatibilityPolicy switch I would receive different results. However, the resulting documents are currently always the same, regardless of this switch, and none of the output files contain the originally embedded office file anymore. How can I get the expected behaviour for -dPDFACompatibilityPolicy=2 (i.e. an error message) and -dPDFACompatibilityPolicy=0 (i.e. output is a regular pdf file still containing the embedded file)?
> 
> Bluntly, you can't.
> 
> This is, I'm afraid, a misapprehension  about the way that pdfwrite works. Many people are under the illusion that pdfwrite 'edits' or 'manipulate's the original PDF file, it does not.
> 
> Ghostscript fully interprets the input (in what ever format, PostScript, PDF, etc) into marking operations, or graphic primitives (eg fill, stroke, image etc) which it sends to the selected device. In general devices call upon the graphics library to render these operations to a bitmap and at the end of the page write the bitmap to a file. The high level, or vector, devices such as pdfwrite instead convert the primitive into an equivalent operation in a different Page Description Language, such as PDF, and emit that. Thus the output file bears no relation, internally, to the input file even if they are of the same format. All that is expected is that the appearance should be the same.
> 
> Your embedded files do not make any marks on the pages, so they are not converted to graphics primitives, are not passed to the device and so are not present in the output. This is not the same as 'removing' embedded files, those embedded files are not there, ever.
> 
> Now, the PDF interpreter and pdfwrite devices do make some effort to preserve certain non-marking aspects of the input, however embedded files are not one of those and therefore there is currently no way to take an input PDF file with embedded files and produce an output file from pdfwrite which includes those same embedded files, no matter what command line you use.
> 
> 
>                    Ken
>