#!/bin/sh
set -eu

[ "$#" -lt 2 ] && {
	echo "Use: [workdir] [basepath] #items on stdin"
	exit;
}

readonly mepath="$(readlink -m $(dirname $0)/..)"

starttime="$(date +%s)"
ddate() { date -u "-d@$starttime" "$1"; }
ftime() { faketime -f "$(ddate '+%Y-%m-%d %H:%M:%SZ') x0" "$@"; }
base_dir="$1"; shift
base_url="$1"; shift

# hacky thing to make sure url ends in /
if ! echo "$base_url" | grep -q '/$'; then
	base_url="${base_url}/"
fi

safeurl() { echo "$1" | sed 's,/,_,g;s,_$,,'; }

[ -d "$base_dir" ] || mkdir "$base_dir"

base_url_hostonly="$(echo "$base_url" | sed -e 's,^http,,;s,^s,,;s,^://,,')"
item_id="$(safeurl "$base_url_hostonly")__bundles_$(ddate '+%Y%m%d')${1:-}"
work_dir="${base_dir}/${item_id}"

echo "-- making $item_id folder --"
mkdir "$work_dir"

while IFS='' read -r url; do
	urlp="$(safeurl "$url")"
	urlwp="$work_dir/$urlp"
	{
		echo "$ fossil clone --unversioned -v --once --no-open $url $urlp.fossil"
		USER=archiveteam-codearchiver-runner ftime fossil clone --unversioned -v --once --no-open --user archiveteam-codearchiver-runner "$url" "$urlwp.fossil" || true
		echo "$ sqlite3 $urlp.fossil \"UPDATE user SET pw='archiveteam' WHERE login='anonymous';\""
		ftime sqlite3 "$urlwp.fossil" "UPDATE user SET pw='archiveteam' WHERE login='anonymous';" || true
		echo "$ sqlite3 $urlp.fossil \"UPDATE user SET pw='archiveteam' WHERE login='archiveteam-codearchiver-runner';\""
		ftime sqlite3 "$urlwp.fossil" "UPDATE user SET pw='archiveteam' WHERE login='archiveteam-codearchiver-runner';" || true
		echo "$ sqlite3 $urlp.fossil \"VACUUM;\""
		ftime sqlite3 "$urlwp.fossil" "VACUUM;" || true
	} 2>&1 | tee "$urlwp.log"
done

echo "-- finished job for $base_url --"
echo "-- generating dir-to-ia.conf (check it and write as .config) --"

cat >"${work_dir}/.dir-to-ia.conf" <<EoC
exit 1 # make user edit check this
sha256=yes
rm=no
rmwait=yes
sha1check=yes
clobber=no
derive=no

# Custom options for ia-upload-stream, as an array of args; this can be used to choose the part size and concurrency, for example, using (--part-size 1G --concurrency 4).
iauploadstreamopts=()

# Item metadata (array with 'key:value' elements); the only mandatory variable with no default
metadata=(
	"collection:open_source_software"
	"date:$(ddate '+%Y-%m-%d')"
	"description:Fossil repos for repositories under $base_url as of $(ddate '+%Y-%m-%d %H:%M:%S') UTC. Retrieved with $(fossil version|sed -e 's,^This is ,,' -e 's,^,(,;s,$,),'), see corresponding log files for command(s) ran"
	"mediatype:software"
	"title:Fossil $base_url_hostonly repository clones ($(ddate '+%Y-%m-%d'))"
)
EoC