mirabilos authored Jan 20, 2022 and mirabilos committed Jan 20, 2022

see commit 636879066fa0e31f129ca3c1054953a07633041c in verein.git
for the rationale; the method used was…

$ git find -gitfiles \*.pdf -print0 | xargs -0r mksh ~/pdfstrippagegroup.ksh

… with the following script:
$ cat ~/pdfstrippagegroup.ksh
#!/bin/mksh

unset LANGUAGE
export LC_ALL=C

die() {
	print -ru2 -- "E: ${0##*/}: $dst: $*"
	exit 1
}

for dst in "$@"; do
	tpf=$dst\~.tmp
	tq1=$tpf.1.qdf
	tq2=$tpf.2.qdf
	tq3=$tpf.3.qdf
	tmp=$tpf.pdf
	qpdf --stream-data=uncompress --normalize-content=n --qdf \
	    "$dst" "$tq1" || die 'qpdf error converting to QDF'
	# remove PDF page groups (required)
	# also remove Producer info (size improvement)
	perl -e '
		use strict;
		use warnings;

		my $s = "0";
		my $ispage = 0;
		my $gt = "<ERR>";
		while (my $line = <>) {
			my $skip = 0;
			chomp(my $l = $line);
			if ($l eq "<<") {
				$ispage = 0;
				$s = 1 if $s eq 0;
			} elsif ($l eq ">>") {
				if ($s eq 3) {
					# only omit page groups
					print $gt unless $ispage;
				}
				$s = 0;
			} elsif ($l eq "  /Type /Page") {
				if (($s == 1) || ($s == 3)) {
					$ispage = 1;
				}
			} elsif (($s == 1) && ($l eq "  /Group <<")) {
				$gt = $line;
				$s = 2;
				$skip = 1;
			} elsif (($s == 1) && ($l =~ qr`^  /Group `)) {
				$gt = $line;
				$s = 3;
				$skip = 1;
			} elsif ($s == 2) {
				$gt .= $line;
				$s = 3 if $l eq "  >>";
				$skip = 1;
			} elsif ($l eq "trailer <<") {
				$s = 4;
			} elsif ($s == 4) {
				# size optimisation hack
				# remove CreationDate, Producer, etc.
				$skip = 1 if $l =~ qr`^  /Info `;
			}
			print $line unless $skip;
		}
	' <"$tq1" >"$tq2" || die 'error during perl QDF filtering'
	fix-qdf <"$tq2" >"$tq3" || die 'error during fix-qdf'
	# bullseye first then buster which lacks options
	if ! (set -x; qpdf --stream-data=compress --recompress-flate \
	    --compression-level=9 --normalize-content=n \
	    --object-streams=disable --deterministic-id \
	    "$tq3" "$tmp") 2>"$tq1" && ! (set -x; qpdf \
	    --stream-data=compress --normalize-content=n \
	    --object-streams=disable --deterministic-id \
	    "$tq3" "$tmp") 2>>"$tq1"; then
		cat "$tq1" >&2
		die 'qpdf error converting from QDF'
	fi
	mv "$tmp" "$dst" || die 'could not create target file'
	rm "$tpf"* || die 'error cleaning up tmp QDF files'
	print -ru2 -- "I: fixed up $dst"
done
print -ru2 -- "I: done"