#!/bin/sh
#
# syspkgdeps [-a arch] [-m machine] [-s setsdir] [-p prefix] sets
#
# Compute naive package dependencies based on file & directory
# nesting. E.g., if pkg P contains /foo/bar and Q contains /foo,
# then Q is considered a dependency of P.
#
# Each line of output contains two syspkg names,
# where the first syspkg depends on the second syspkg.
#

#set -u

prog="${0##*/}"
rundir="$(dirname "$0")" # ${0%/*} isn't good enough when there's no "/"
. "${rundir}/sets.subr"

#
# set defaults
#
prefix=/

usage()
{
	cat 1>&2 <<USAGE
Usage: ${0##*/} [-a arch] [-m machine] [-s setsdir] [-p prefix] setname [...]
	-a arch		set arch (e.g, m68k, mips, powerpc)	[${MACHINE_ARCH}]
	-m machine	set machine (e.g, amiga, i386, macppc)	[${MACHINE}]
	-s setsdir	directory to find sets			[${setsdir}]
	-p prefix	prefix for created plist		[${prefix}]
	setname [...]	sets to find dependencies for
USAGE
	exit 1
}

# parse arguments
while getopts a:m:p:s: ch; do
	case ${ch} in
	a)
		MACHINE_ARCH="${OPTARG}"
		MACHINE_CPU="$(arch_to_cpu "${OPTARG}")"
		;;
	m)
		MACHINE="${OPTARG}"
		;;
	p)
		prefix="${OPTARG}"
		;;
	s)
		setsdir="${OPTARG}"
		;;
	*)
		usage
		;;
	esac
done
shift $((${OPTIND} - 1))
if [ $# -lt 1 ]; then
	usage
fi

sets="$*"
case "${sets}" in
all)	sets="${nlists}" ;;
esac

# TBD clean up
SCRATCH="$(${MKTEMP} -d "/var/tmp/${prog}.XXXXXX")"

if [ $? -ne 0 ]; then
	echo >&2 "${prog}: Could not create scratch directory."
	exit 1
fi

PATH_MEMBERSHIP="${SCRATCH}/path-membership"
PATH_TO_PKGNAME="${SCRATCH}/pathpkg.db"
PARENT_PKGNAMES="${SCRATCH}/parent-pkgnames"
PARENT_PATHNAMES="${SCRATCH}/parent-pathnames"

echo >&2 "${prog}: indexing packages by pathnames"

list_set_files ${sets} | ${SED} 's/^\.\///' | \
${ENV_CMD} PREFIX="${prefix}" ${AWK} '{
	if ($1 == ".") {
		print ENVIRON["PREFIX"] " " $2;
	} else {
		print ENVIRON["PREFIX"] $1 " " $2;
	}
}' | ${SORT} -k 1 -u > "${PATH_MEMBERSHIP}"

${DB} -q -w -f - btree "${PATH_TO_PKGNAME}" < "${PATH_MEMBERSHIP}"

if [ $? -ne 0 ]; then
	echo >&2 "${prog}: error creating database, aborting"
	exit 1
fi

echo >&2 "${prog}: computing parent pathnames"

while read pathname pkgname; do
	# print parent pathname.
	# (This uses a cheap implementation of dirname from sets.subr.)
	dirname "${pathname}"
done < "${PATH_MEMBERSHIP}" > "${PARENT_PATHNAMES}"

echo >&2 "${prog}: selecting parent packages using parent pathnames"

${DB} -q -f - btree "${PATH_TO_PKGNAME}" < "${PARENT_PATHNAMES}" | \
	${PASTE} "${PATH_MEMBERSHIP}" - | \
	${AWK} '{ if ($2 != $4) print $2 " " $4; }' | \
	${SORT} -u > "${SCRATCH}/alldeps"

if [ $? -ne 0 ]; then
	echo >&2 "${prog}: error in parent-directory lookup, aborting"
	exit 1
fi

echo >&2 "${prog}: checking for cyclic dependencies"

tsort_errors="$(${TSORT} < "${SCRATCH}/alldeps" 2>&1 >/dev/null)"

if [ -n "${tsort_errors}" ]; then
	# Errors from tsort are usually to do with cyclic dependencies.
	# The most likely underlying cause is that /foo and /foo/bar/baz
	# are in syspkg A, but /foo/bar is in syspkg B.
	echo >&2 "${tsort_errors}" # this is likely to be multiple lines
	echo >&2 "${prog}: Above messages probably indicate an error in the lists"
	exit 1
fi

echo >&2 "${prog}: removing redundant dependencies"

${HOST_SH} "${rundir}/culldeps" < "${SCRATCH}/alldeps"

if [ $? -ne 0 ]; then
	echo >&2 "${prog}: error in culldeps, aborting"
	exit 1
fi