Ghostscript to convert PDF 2 Back to PDF version 1.7 - pdf

Question

I need to build a PDF serverside by reading in a number of PDFs and inserting each page into a new multipage PDF. The problem is that the PDFs are provided in version 2.0 format, but my application can only read version 1.7. I would like to convert the version 2 files back into a version 1.7 file so that my application can read it.
I am using ghostscript version 9.27, and have tried several commands, but each time I end up with an empty PDF. Example:
/usr/local/bin/gs \
-q -dNOPROMPT \
-dBATCH \
-dDEVICEWIDTH=595 \
-dDEVICEHEIGHT=842 \
-sDEVICE=pdfwrite \
-dCompatibilityLevel=1.7 \
-sFileName=pdf-version-2.pdf \
-sOutputFile=fileout.pdf
There is no error, just an empty PDF. The "file" command does give the expected output "PDF document, version 1.7" but that's not much good when the file is blank. Any help greatly appreciated!

Answer 1

OK so I think the problem is your command line (as pointed out to me by one of my colleagues). You've specified -sFileName=pdf-version-2.pdf, which looks like you're trying to specify the input file.
There is no Ghostscript switch -sFileName, you specify the input filename(s) by simply putting the name on the command line. So you really want:
/usr/local/bin/gs \
-q -dNOPROMPT \
-dBATCH \
-dDEVICEWIDTH=595 \
-dDEVICEHEIGHT=842 \
-sDEVICE=pdfwrite \
-dCompatibilityLevel=1.7 \
-sOutputFile=fileout.pdf \
pdf-version-2.pdf
For reasons that were good 30 years ago, the Ghostscript command line switches are copied into the PostScript environment, where they can be accessed by PostScript programs. So while its true (and possibly the szource of your confusion) that some of the utility programs shipped with Ghostscript do use -sFileName, Ghostscript itself doesn't, it just defines a PostScript variable using that name, so that programs can read it.
Because you've specified BATCH and NOPROMPT, but haven't specified an input file, the interpreter starts up, erases the current page to white, then exits. Closing the pdfwrite device causes it to write out the current content of the page, which is, well, white, resulting in your empty PDF file.
The slightly modified command line above worked well for me, but as I noted in my comments specifying DEVICEWIDTHPOINTS and DEVICEHEIGHTPOINTS won't actually do anything.