Determine the transitive closure of dependencies of a TeX document

tlclosure.sh

#!/usr/bin/env bash

set -eo pipefail

program="${0##*/}"
scheme="scheme-infraonly"

opts="$(getopt -o s: -l scheme: -n "$0" -- "$@")"
err="$?"
eval set -- "$opts"
while true; do case $1 in
	-s|--scheme) scheme="$2"; shift 2 ;;
	--) shift; break ;;
esac done
flsfile="$1"

if [ $err -ne 0 ] || [ $# -lt 1 ]; then
	cat >&2 <<-EOF
	Usage: $program [OPTIONS]... FILE.fls

	OPTIONS

	  -s, --scheme=SET    TeX Live scheme to base on (default: scheme-infraonly)

	EXAMPLES

	  $ pdflatex -recorder test.tex
	  $ tlclosure -s scheme-basic test.fls

	EOF
	exit 1
fi

tmpdir=$(mktemp -d)
arch=$(tlmgr print-platform)

echo "Listing packages defined by $scheme"
cachedir="${XDG_CACHE_HOME:-$HOME/.cache}/tlclosure"
scheme_cache="$cachedir/$scheme.txt"
recurse() {
	while read pkgs; do
		[ -z "$pkgs" ] && break
		echo "$pkgs" | tr ' ' '\n' | tee -a "$tmpdir/scheme.txt"
		tlmgr info --data depends $pkgs | sed "s/ARCH/$arch/g" | tr ':' ' ' | tr '\n' ' ' | xargs | recurse
	done
}

# TODO: expire cache
# compare mtime of $scheme_cache and tlpdb
if [ -f "$scheme_cache" ]; then
	echo "=> Cache exists"
else
	echo "$scheme" | recurse
	mkdir -p "$cachedir"
	sort "$tmpdir/scheme.txt" | uniq > "$scheme_cache"
fi

echo "Looking up the corresponding packages in TeX Live"
touch "$tmpdir/pkgs.txt"
grep '^INPUT ' "$flsfile" | xargs -i -P $(nproc) bash -c "
line=\"{}\"
file=\"\${line##INPUT }\"
[ -f \"./\$file\" ] && exit
pkg=\"\$(tlmgr search --file \"/\${file##*/}\\\$\" | head -n 1 | tr -d ':')\"
{
	flock -x 9
	if [ -n \"\$pkg\" ] &&
		! grep -qFx \"\$pkg\" '$scheme_cache' &&
		! grep -qFx \"\$pkg\" '$tmpdir/pkgs.txt'
	then
		echo \"\$pkg\" | tee -a '$tmpdir/pkgs.txt'
	fi
} 9> '$tmpdir/pkgs.lock'
"

echo "$scheme" > "./pkgs.txt"
sort "$tmpdir/pkgs.txt" | uniq >> "./pkgs.txt"