#!/bin/bash
# Copyright 2011-2019 Gentoo Authors; Distributed under the GPL v2
# might be earlier copyright, no history available

# 1) Create the tarball
# 2) Sanity check the tarball size and bail out if it appears abnormal.
# 3) Compress tarball(bzip2,xz)
# 4) sign
# 5) delta generation
# 6) create some symlinks
# 7) tidy up
# 8) clean up old deltas

umask 022
RSYNC_VARS=${RSYNC_VARS:=/usr/local/bin/mastermirror/rsync-gen.vars}
source "${RSYNC_VARS}" || exit 3

# locations (if used in multiple files, please move to rsync-gen.vars file)
MASTER="${FINALDIR_repo_gentoo}" # the master gentoo-x86 copy
TEMP="${BASE}/snapshots-tmp/" # working directory
HISTORY="7" #number in dsays of previous snapshots to keep
DELTA_UPLOAD="${UPLOAD}/deltas/"

[[ -d ${TEMP} ]]			|| mkdir ${TEMP}
[[ -d ${UPLOAD} ]]			|| mkdir ${UPLOAD}
[[ -d ${DELTA_UPLOAD} ]]		|| mkdir ${DELTA_UPLOAD}

write_time_log "---------------------------------------------------------"
write_time_log "START ENTIRE SCRIPT	$(date -u)"

# used to name the file
DELTA_BASE=$(/bin/date -d '-2 day' +%Y%m%d)
DELTA_FILENAME="portage-${DELTA_BASE}.tar.bz2"
YESTERDAY=$(/bin/date -d yesterday +%Y%m%d)
FILENAME="portage-${YESTERDAY}.tar.bz2"

FILENAME_NEW="gentoo-${YESTERDAY}.tar.xz"

# Parallel compressors can use a LOT of cpu, be nice about it
NICE="nice -n 10"

SIGNKEYID=${SIGNKEYID_snapshot}

# Verify signing key is available
if ! gpg --list-keys "${SIGNKEYID}" | grep -sq "${SIGNKEYID}"; then
	echo "${SIGNKEYID} not imported! exiting"
	exit 1
fi

# Use yesterdays snapshot, if its unavailable, use the 'latest' snapshot.
# If both are missing, bail out.
POTENTIAL_SNAPSHOTS=(
	${UPLOAD}/${DELTA_FILENAME}
	${UPLOAD}/portage-latest.tar.bz2
	FAIL
)

for previous_snapshot in ${POTENTIAL_SNAPSHOTS[*]}; do
	[[ -e ${previous_snapshot} ]] && break
done

if [[ ${previous_snapshot} == FAIL ]]; then
	# Cut FAIL out of the printed list, to ease debugging.
	echo "Previous snapshots do not exist: ${POTENTIAL_SNAPSHOTS[*]:0:2}"
	exit 1
fi

# Find fastest BZIP2
# lbzip2,pbzip2 default to multiple threads
# emerge-delta-webrsync relies on verifying signature of tarball after
# re-compressing it with bzip2, therefore failing if the signature
# was done on lbzip2 or pbzip2 compressed tarball, #573908
for BZIP2_PROG in bzip2 FAIL ; do
	[ -n "$(type $BZIP2_PROG 2>/dev/null)" ] && break
done
if [ $BZIP2_PROG == FAIL ];  then
	echo "Could not find any BZIP2" 1>&2
	exit 1
fi
# Find fastest XZ
# pixz appends some data and leads to SIGPIPE, #573642
for XZ_PROG in xz FAIL ; do
	[ -n "$(type $XZ_PROG 2>/dev/null)" ] && break
done
if [ $XZ_PROG == FAIL ];  then
	echo "Could not find any xz" 1>&2
	exit 1
fi
# Newer 'xz' supports threads as well, but defaults to single-threaded
if $XZ_PROG --help |grep -sq threads=NUM ; then
	XZ_PROG="${XZ_PROG} -T 0"
fi

# working dir
cd ${TEMP}

# Tarball generation
write_time_log "START TARBALL		$(date -u)"

# 1a) Tarball prep:
write_time_log "START TARBALL(prep) $(date -u)"
# Build exclusion list
EXCLUSION_LIST="$(mktemp -p ${TEMP} snapshot-exclude.XXXXXXXXXX)"
/usr/local/bin/mastermirror/print-exclusion-list.sh "${MASTER}" >"${EXCLUSION_LIST}"
write_time_log "END TARBALL(prep)		$(date -u)"

# 1b) Create the tarball
COMMON_TAR_OPTIONS=(
	# Force a small block size
	'--blocking-factor=1'
	'--record-size=512'
	# Tar format rationale:
	# ---------------------
	# Longest directory prefix, 94 chars:
	# gentoo-YYYYMMDD/profiles/default/linux/powerpc/ppc64/17.0/64bit-userland/little-endian/systemd/
	# Longest path, 140 chars:
	# gentoo-YYYYMMDD/gnome-extra/gnome-shell-extension-applications-overview-tooltip/gnome-shell-extension-applications-overview-tooltip-6.ebuild
	# Longest filename, 91 chars:
	# spirv-tools-2019.10_pre20191027-Respect-CMAKE_INSTALL_LIBDIR-in-installed-CMake-files.patch
	#
	# These length require that a tar format with unlimited length is used, the
	# split-255 layout of ustar is not gaurenteed to be long enough.
	#
	# The tar formats with unlimited length are: gnutar, posix
	#
	# The posix tar format has additional 1K overhead per entry, and the Gentoo
	# repo has ~160k entries (files & dirs), which adds up to 160M extra in the
	# tarball.
	# The differ/bdelta toolchain is also confirmed to work on posix tar
	# format.
	'--format=gnu'
	# Sorting by name produces consistent ordering and helps compression of
	# related content. Custom ordering might further improve ordering in future
	# (eg all metadata.xml first)
	'--sort=name'
	# Force ownership of content:
	'--owner=portage'
	'--group=portage'
	# Excluded content:
	'--no-wildcards'
	"--exclude-from=${EXCLUSION_LIST}"
	# Do not capture any xattr/acl info at all.
	'--no-acls'
	'--no-xattrs'
	'--no-selinux'
	# Include a volume ID for tracing
	# volume header is NOT supported by:
	# - Docker https://bugs.gentoo.org/631644
	# - tarsync https://bugs.gentoo.org/631616
	#"--label=${FILENAME%.bz2}"
	# do everything relative to the destination
	"--directory=${MASTER}"
)
# create the tarball and move it to the right location
write_time_log "START TARBALL(tar,old)		$(date -u)"
if [ ! -f "${FILENAME%.bz2}" ]; then
	OLD_TARBALL_OPTIONS=(
		# The . needs to match the file argument
		"--transform=s,^\\.,portage,g"
		# The operation, destination, source arguments
		'--create'
		"--file=${FILENAME%.bz2}"
		.
	)
	tar \
		"${COMMON_TAR_OPTIONS[@]}" \
		"${OLD_TARBALL_OPTIONS[@]}"
	rc=$?
	if [ $rc -ne 0 ]; then
		echo "TARBALL(tar,old) failed!"
		echo "TARBALL(tar,old) failed!" 1>&2
		exit 1
	fi
fi
write_time_log "END TARBALL(tar,old)		$(date -u)"
write_time_log "START TARBALL(umd5,old)		$(date -u)"
if [ ! -f " ${FILENAME}.umd5sum" ]; then
	md5sum ${FILENAME%.bz2} | tee ${FILENAME%.bz2}.bz2.umd5sum ${FILENAME%.bz2}.xz.umd5sum >/dev/null
fi
write_time_log "END TARBALL(umd5,old)		$(date -u)"
# end 1b)

# 1c) Also create new-style tarball,
# but do it via transformation of old tarball
write_time_log "START TARBALL(tar,new) $(date -u)"
if [ ! -f "${FILENAME_NEW%.xz}" ]; then


	# The newer tarball differs in the following ways:
	# Filename 'portage-YYYYMMDD' -> 'gentoo-YYYYMMDD'
	# Path prefex 'portage' -> 'gentoo-YYYYMMDD'
	#
	# Earlier code used tar-transform-names.pl as a wrapper around Perl
	# Archive::Tar::Stream, but it was found that the Archive::Tar::Stream
	# codebase did not handle gnutar format correctly.
	# https://bugs.gentoo.org/703460
	#
	# While the stream processing was MUCH faster (because it didn't traverse
	# the filesystem at all), it broke on very long filenames that ARE present
	# in the Gentoo repo:
	#/usr/local/bin/mastermirror//tar-transform-names.pl \
	#	--input-file "${FILENAME%.bz2}" \
	#	--output-file "${FILENAME_NEW%.xz}" \
	#	--regex "s/^portage/${FILENAME_NEW%%.*}/"

	NEW_TARBALL_OPTIONS=(
		# The . needs to match the file argument
		"--transform=s,^\\.,${FILENAME_NEW%%.*},g"
		# The operation, destination, source arguments
		'--create'
		"--file=${FILENAME_NEW%.xz}"
		.
	)
	tar \
		"${COMMON_TAR_OPTIONS[@]}" \
		"${NEW_TARBALL_OPTIONS[@]}"
	rc=$?
	if [ $rc -ne 0 ]; then
		echo "TARBALL(tar,new) failed!"
		echo "TARBALL(tar,new) failed!" 1>&2
		exit 1
	fi
fi
write_time_log "END TARBALL(tar,new) $(date -u)"
write_time_log "START TARBALL(umd5,new)		$(date -u)"
if [ ! -f " ${FILENAME_NEW}.umd5sum" ]; then
	md5sum ${FILENAME_NEW%.xz} > ${FILENAME_NEW}.umd5sum
fi
write_time_log "END TARBALL(umd5,new)		$(date -u)"
# end 1c)

# end 1)
write_time_log "END TARBALL		$(date -u)"

# 2) Sanity check the tarball size and bail out if it appears abnormal.
write_time_log "START SIZE SANITY	$(date -u)"
current_size=$(stat -c '%s' "${FILENAME%.bz2}")
if [ "${previous_snapshot}" != "${previous_snapshot%.bz2}" ] ; then
	previous_size=$($BZIP2_PROG -k -d -c "${previous_snapshot}" |wc -c)
elif [ "${previous_snapshot}" != "${previous_snapshot%.xz}" ] ; then
	previous_size=$($XZ_PROG -k -d -c "${previous_snapshot}" |wc -c)
elif [ "${previous_snapshot}" != "${previous_snapshot%.tar}" ] ; then
	previous_size=$(stat -c '%s' "${previous_snapshot}")
else
	previous_size=0
fi
if [ -z "$current_size" ] || [ "${current_size}" -lt "${previous_size}" ]; then
	size_difference=$(expr ${previous_size} - ${current_size})
	difference_ratio=$(expr ${previous_size} / ${size_difference})
	if [ ${difference_ratio} -lt 2 ]; then
		echo "Snapshot size has decreased by more than 50% in one day!!!"
		echo "${FILENAME} ${current_size} bytes"
		echo "${previous_snapshot} ${previous_size} bytes"
		exit 1
	elif [ ${difference_ratio} -lt 5 ]; then
		echo "Snapshot size has decreased by more than 20% in one day!!!"
		echo "${FILENAME} ${current_size} bytes"
		echo "${previous_snapshot} ${previous_size} bytes"
		# Make this non-fatal while we recover more space.
		#exit 1
	fi
fi
write_time_log "END SIZE SANITY		$(date -u)"
# end 2)

# 3) create xz tarball
write_time_log "START COMPRESS		$(date -u)"
# This happens only AFTER the sanity check
# 3a) bzip2 of old tarball
write_time_log "START COMPRESS(bzip2,old)		$(date -u)"
[ ! -f "${FILENAME%.bz2}.bz2" ] && ${NICE} $BZIP2_PROG -k9 ${FILENAME%.bz2}
write_time_log "END COMPRESS(bzip2,old)		$(date -u)"
# 3b) xz of old tarball
write_time_log "START COMPRESS(xz,old) $(date -u)"
if [ ! -f "${FILENAME%.*}.xz" ] ; then
	# pixz, pxz, xz all differ in filename generation
	# xz: .tar -> .tar.xz
	# pixz: .tar -> .tpxz
	# pxz: .tar -> .txz
	#
	# To avoid this, be explicit by using IO.
	${NICE} ${XZ_PROG} -9 -e <"${FILENAME%.*}" >"${FILENAME%.*}.xz" || exit $?
fi
write_time_log "END COMPRESS(xz,old)			$(date -u)"
# 3c) xz of new tarball
write_time_log "START COMPRESS(xz,new) $(date -u)"
if [ ! -f "${FILENAME_NEW%.*}.xz" ] ; then
	# pixz, pxz, xz all differ in filename generation
	# xz: .tar -> .tar.xz
	# pixz: .tar -> .tpxz
	# pxz: .tar -> .txz
	#
	# To avoid this, be explicit by using IO.
	${NICE} ${XZ_PROG} -9 -e <"${FILENAME_NEW%.*}" >"${FILENAME_NEW%.*}.xz" || exit $?
fi
write_time_log "END COMPRESS(xz,new)			$(date -u)"
write_time_log "END COMPRESS		$(date -u)"
# end 3)

# 4) sign
write_time_log "START SIGN		$(date -u)"
for f in "${FILENAME}" "${FILENAME%.*}".xz "${FILENAME_NEW}"; do
	if [ ! -f "${UPLOAD}${f}".umd5sum ]; then
		cp "${f}".umd5sum "${UPLOAD}${f}".umd5sum || exit $?
		md5sum "$f" > "$f".md5sum || exit $?
	fi
	if [ ! -f "$f".gpgsig ]; then
		gpg --batch -u "${SIGNKEYID}" --armor --detach-sign \
			--output "$f".gpgsig "$f" || exit $?
	fi
	mv "$f" "$f".md5sum "$f".gpgsig "${UPLOAD}"/ || exit $?
done
write_time_log "END SIGN		$(date -u)"
# end 4)

# 5) delta generation
# Delta is not generated for new filename at this time
write_time_log "START DELTA		$(date -u)"
PATCH=snapshot-${DELTA_BASE}-${YESTERDAY}.patch.bz2
if [ ! -f "${PATCH}" ]; then
	SNAPSHOT_BASE="${TEMP}"/$(basename "${previous_snapshot%.bz2}")-snapshotbase
	${NICE} ${BZIP2_PROG} -dkc "${previous_snapshot}" > "${SNAPSHOT_BASE}"
	/usr/bin/differ -f bdelta \
		"${SNAPSHOT_BASE}" \
		"${FILENAME%.bz2}" \
		"${PATCH%.bz2}"
	${NICE} ${BZIP2_PROG} -9 "${PATCH%.bz2}"
	md5sum "${PATCH}" > "${PATCH}.md5sum"
	chmod 644 "${PATCH}"{,.md5sum}
	mv "${PATCH}"{,.md5sum} "${DELTA_UPLOAD}"
fi
write_time_log "END DELTA		$(date -u)"
# end 5)

# 6) create some symlinks
write_time_log "START SYMLINK		$(date -u)"
cd ${UPLOAD}
for f in "${FILENAME}" "${FILENAME%.*}".xz ; do
	ext=${f##*.}
	ln -sf "$f" "${UPLOAD}"portage-latest.tar.${ext} || exit $?
	rm -f "${UPLOAD}"portage-latest.tar.${ext}.md5sum || exit $?
	sed "s/${f}\$/portage-latest.tar.${ext}/" "${UPLOAD}"${f}.md5sum > \
		"${UPLOAD}"portage-latest.tar.${ext}.md5sum || exit $?
	ln -sf "${f}".gpgsig "${UPLOAD}"portage-latest.tar.${ext}.gpgsig || exit $?
done
# shellcheck disable=SC2066
for f in "${FILENAME_NEW}" ; do
	ext=${f##*.}
	ln -sf "$f" "${UPLOAD}"gentoo-latest.tar.${ext} || exit $?
	rm -f "${UPLOAD}"gentoo-latest.tar.${ext}.md5sum || exit $?
	sed "s/${f}\$/gentoo-latest.tar.${ext}/" "${UPLOAD}"${f}.md5sum > \
		"${UPLOAD}"gentoo-latest.tar.${ext}.md5sum || exit $?
	ln -sf "${f}".gpgsig "${UPLOAD}"gentoo-latest.tar.${ext}.gpgsig || exit $?
done
write_time_log "END SYMLINK		$(date -u)"
# end 6)

# 7) tidy up
write_time_log "START CLEANUP		$(date -u)"

# Cleanup immediate stuff
rm -v -f \
	"${TEMP}"/orig \
	"${TEMP}"/"${FILENAME%.bz2}"{.bz2,.xz,}{,.umd5sum} \
	"${TEMP}"/"${FILENAME_NEW%.xz}"{.bz2,.xz,}{,.umd5sum} \
	"${TEMP}"/"${EXCLUSION_LIST}" \
	|| exit $?
# Cleanup older stuff that might have leaked
/usr/bin/find \
	"${TEMP}" \
	-regextype egrep \
	-maxdepth 1 \
	-type f \
	-mtime +0 \
	\( \
		-regex '.*/(portage|gentoo)-[0-9]{6}.*' \
		-o \
		-name 'snapshot-exclude.*' \
		-o \
		-name orig \
	\) \
	-delete
# Cleanup old snapshots
/usr/bin/find \
	"${UPLOAD}" \
	-maxdepth 1 \
	-type f \
	-mtime +${HISTORY} \
	-delete

write_time_log "END CLEANUP		$(date -u)"
# end 7)

# 8) clean up old deltas
# Delta is not generated for new filename at this time
write_time_log "START CLEANUP DELTA	$(date -u)"
/usr/local/bin/mastermirror/clean-old-deltas.py "${DELTA_UPLOAD}" "${YESTERDAY}" $(stat -c '%s' "${UPLOAD}/${FILENAME}") > /dev/null
write_time_log "END CLEANUP DELTA	$(date -u)"
# end 8)
write_time_log "END ENTIRE SCRIPT	$(date -u)"