#!/usr/bin/env bash

# SPDX-License-Identifier: GPL-2.0-only
# SPDX-FileCopyrightText: 2024 Eddie Chapman <eddie@ehuk.net>

# WARNING: this script is currently not a full replacement for xz, it just mimicks 
# some of the decompression functionality of xz. It is only designed at the
# moment to be called by Portage and even then it does not yet cover all cases of 
# that.

# Some places in portage where xz is called:
# - /usr/lib/portage/python3.11/phase-helpers.sh
#      This is where 99% of calls to xz happen, from the line:
#      __unpack_tar "xz -T$(___makeopts_jobs) -d"
#      in the unpack phase.
#      This results in a call to xz inside __unpack_tar() where the -c arg is added (for stdout)
#      and the filename is added as an argument.
# - /usr/bin/deb2targz
#      Some packages e.g. google-chrome have deb distfiles which can contain a data.tar.xz
#      file so deb2targz launches xz -dc to decompress that.
# - /usr/bin/rpm2tar
#      Some packages e.g. libreoffice have rpm distfiles compressed with xz so rpm2tar launches 
#      xz -dc to decompress them
# - /usr/portage/eclass/llvm.org.eclass
#      xz is not called directly but tar -x -J is run (and tar then runs "xz -d" with piped 
#      in/out, with no file as argument)

LOGGER=$(command -v logger)

if [ ! -x "${LOGGER}" ]; then
echo "(wrapper): Fatal error: logger command does not appear to exist!"
exit 1
fi

LOG_PREFIX="(wrapper):"

# /usr/bin/7za is just a wrapper that executes this
SEVEN_ZA="/usr/lib64/p7zip/7za"

DATE_CMD=$(command -v date)
MKTEMP_CMD=$(command -v mktemp)
PS_CMD=$(command -v ps)
CAT_CMD=$(command -v cat)
CHMOD=$(command -v chmod)
WHOAMI=$(command -v whoami)
GREP=$(command -v grep)
FILE_CMD=$(command -v file)
READLINK=$(command -v readlink)

for EXE_F in ${SEVEN_ZA} ${DATE_CMD} ${MKTEMP_CMD} ${PS_CMD} ${CAT_CMD} ${CHMOD} \
${WHOAMI} ${GREP} ${FILE_CMD} ${READLINK}; do

	if [ ! -x "${EXE_F}" ]; then
	MSG="${LOG_PREFIX} Fatal Error: ${EXE_F} does not exist or is not an exe!"
	${LOGGER} -p syslog.err -t "${0}" "${MSG}"
	exit 1
	fi

done

DECOMPRESS_REQUESTED=N
STDOUT_REQUESTED=N

for myarg in "${@}"; do

	# Look for the 3 forms of xz's decompress argument when it is by itself.
	# TO-DO: collapse these into one grep command, improve the horrible regex.
	echo "${myarg}" | ${GREP} -Eq '^[-]d$' 
	retA=$?
	echo "${myarg}" | ${GREP} -Eq '^[-][-]decompress$'
	retB=$?
	echo "${myarg}" | ${GREP} -Eq '^[-][-]uncompress$'
	retC=$?

	if [ ${retA} -eq 0 ] || [ ${retB} -eq 0 ] || [ ${retC} -eq 0 ]; then
	DECOMPRESS_REQUESTED=Y
	fi

	# Look for the 3 forms of xz's stdout argument when it is by itself.
	# TO-DO: collapse these into one grep command, improve the horrible regex.
	echo "${myarg}" | ${GREP} -Eq '^[-]c$' 
	retA=$?
	echo "${myarg}" | ${GREP} -Eq '^[-][-]to[-]stdout$'
	retB=$?
	echo "${myarg}" | ${GREP} -Eq '^[-][-]stdout$'
	retC=$?

	if [ ${retA} -eq 0 ] || [ ${retB} -eq 0 ] || [ ${retC} -eq 0 ]; then
	STDOUT_REQUESTED=Y
	fi

	# and look for both together as -dc or -cd
	# TO-DO: collapse these into one grep command, improve the horrible regex.
	echo "${myarg}" | ${GREP} -Eq '^[-]dc$' 
	retA=$?
	echo "${myarg}" | ${GREP} -Eq '^[-]cd$' 
	retB=$?

	if [ ${retA} -eq 0 ] || [ ${retB} -eq 0 ]; then
	DECOMPRESS_REQUESTED=Y
	STDOUT_REQUESTED=Y
	fi

done

# This script only tries to decompress. No compress functionaility at all
# at this stage in its development.
if [ "${DECOMPRESS_REQUESTED}" = "N" ]; then

	MSG="${LOG_PREFIX} Fatal Error: no (d|decompress|uncompress) option on the command line. Sorry, this wrapper script only supports decompression."
	#echo "$MSG"
	${LOGGER} -p syslog.err -t "${0}" "${MSG}"
	exit 1

fi

# DEBUG
#MSG="${LOG_PREFIX} (DEBUG) stdout requested? ${STDOUT_REQUESTED}"
#${LOGGER} -p syslog.info -t "${0}" "${MSG}"

WHO_CALLED=$(${WHOAMI})

# get the parent command, very useful for debugging
PARENT_CMD=$(${PS_CMD} -o args= ${PPID})

# DEBUG, avoid leaving enabled long term as potential for future security problem, 
# due to unescaped attacker controlled info being passed to logger
#MSG="${LOG_PREFIX} (DEBUG) U: ${WHO_CALLED}, PARENT: ${PARENT_CMD}, ARGS: ${@}"
#${LOGGER} -p syslog.info -t "${0}" "${MSG}"

f_passed_to_script=

# loop over args again to see if any file has been passed as an arg
# TO-DO there will be a better way of doing this.
for myarg in "${@}"; do

	# TO-DO, are there other possible extensions for xz. Also theoretically possible we
	# could be passed one with extension in caps.
	echo "${myarg}" | ${GREP} -Eq '[.]xz$' 
	r=$?

	if [ ${r} -eq 0 ]; then
	f_passed_to_script="${myarg}"
	break
	fi

done


function do_uncompress {

	# remember return numbers can only be btw 0 - 255

	# some sanity checks follow ...

	if [ -z "${1}" ]; then
	MSG_TO_SHOW="function requires 1 argument; a filename with our without leading path"
	return 184
	fi

	realf=$(${READLINK} -e "${1}" 2>/dev/null)

	if [ ! -f "${realf}" ]; then
	MSG_TO_SHOW="argument supplied either is not a file or, if it is a file, I cannot find it."
	return 194
	fi

	if ! test -s "${realf}"; then
	MSG_TO_SHOW="file supplied is empty!"
	return 204
	fi

	if [ ! -r "${realf}" ]; then
	MSG_TO_SHOW="file supplied cannot be read!"
	return 214
	fi

	# 7z does not like the file path being passed to it inside quotes. So lets just make 
	# sure not to pass it anything that contains any characters NOT in our sane list 
	# in our regex below (so nothing needs quoting as no shell metachars).
	# TO-DO: there will be a better way of dealing with this, prob by converting to an 
	# escaped string. But for now (2024) haven't come across any distfiles with weird chars 
	# in them thankfully.
	echo "${realf}" | ${GREP} -Evq '[A-Za-z0-9_:@%+/.-]'
	r=$?

	if [ ${r} -eq 0 ]; then
	MSG_TO_SHOW="found unsupported characters in the (real) file path."
	return 224
	fi

	# Make sure we have been given an xz file.
	# the -e arguments exclude tests we're not interested in, hopefully some tiny perf gain
	# but more importantly reduce attack surface.
	# Also we hae it output the mime type rather than a human readable string, more reliable
	${FILE_CMD} -e ascii -e cdf -e apptype -e csv -e elf -e json -e simh -e tar --mime ${realf} 2>/dev/null | ${GREP} -q 'application/x-xz;'
	r=$?

	if [ ${r} -ne 0 ]; then
	MSG_TO_SHOW="file supplied does not appear to be an xz file, according to the file command"
	return 234
	fi

	# initialise this string for the 7za stdout option (-so). Empty (no stdout) by default.
	STDOUT_OPT_STR=''
	# and this string to, by default, redirect 7za output to /dev/null, as it is somewhat
	# chatty when decompressing.
	STDOUT_REDIR_STR='>/dev/null'

	# If the uncompressed data should be sent to stdout then the above vars need to be changed.
	if [ "${STDOUT_REQUESTED}" = "Y" ]; then
	STDOUT_OPT_STR='-so'
	STDOUT_REDIR_STR=''
	fi

	# we currently set stderr to redirect to /dev/null always.
	STDERR_REDIR_STR='2>/dev/null'

	SEVEN_ZA_FULL_CMD="${SEVEN_ZA} e ${realf} ${STDOUT_OPT_STR} -bd ${STDOUT_REDIR_STR} ${STDERR_REDIR_STR}"

	MSG="${LOG_PREFIX} About to run: ${SEVEN_ZA_FULL_CMD}"
	${LOGGER} -p syslog.info -t "${0}" "${MSG}"

	# In 99% of cases we will redirect 7za decompressed output to stdout (-so).
	# So make really sure nothing gets output to stdout by this script after this point!
	# Log messages all only to logger.
	eval "${SEVEN_ZA_FULL_CMD}"
	return $?

}

last_r=0
LAST_ERR_MSG=

# if no file was detected we assume we will get compressed data via stdin
if [ -z "${f_passed_to_script}" ]; then

	# We create a temp file to save the stdin compressed data into as the current p7zip provided 
	# 7za does not work properly if fed data via stdin, though according to docs it
	# should work, so probably a bug.
	# Also note within sandbox tmp space is not the system /tmp AFAICT.
	# Unfortunately this means we need to make sure we have enough space for the uncompressed
	# data both in /tmp as well as whichever filesystem we set portage to do builds on.

	mytf=`${MKTEMP_CMD}`
	sleep 0.3

	# sanity
	if [ -e "${mytf}" ]; then

		${CHMOD} 0600 "${mytf}"

	else

		MSG="${LOG_PREFIX} Fatal Error: something has gone very wrong, no temp file exists!"
		${LOGGER} -p syslog.err -t "${0}" "${MSG}"
		exit 1

	fi

	# save stdin to our tmpfile. quotes shld not be needed but, what the hell, might as well.
	cat > "${mytf}"
	r=$?

	# sanity
	if [ ${r} -ne 0 ]; then

		MSG="${LOG_PREFIX} Fatal Error: cat returned non-zero code of ${r} when redirecting to ${mytf}!"
		${LOGGER} -p syslog.err -t "${0}" "${MSG}"
		exit 1

	else

		# Even if stdout was NOT requested using a command line argument (and thus STDOUT_REQUESTED will be set to N),
		# xz assumes that you *do* want the uncompressed stream to go to stdout if no file was given on the command line (naturally).
		# So we need to force this to Y here to make sure that happens.
		STDOUT_REQUESTED=Y

		# Having saved stdin to the tmp file above we can now have 7za decompress said tmp file.
		# Remember from here on we run 7za which will in most cases output binary data to stdout.
		# So make REALLY sure nothing else gets output after this!
		# Log messages all only to logger.
		do_uncompress "${mytf}"
		last_r=$?

		# error message numbers inside do_uncompress(), hopefully none clash with those used by 7za
		if [ ${last_r} = 184 ] || [ ${last_r} = 194 ] || [ ${last_r} = 204 ] || [ ${last_r} = 214 ] || [ ${last_r} = 224 ] || [ ${last_r} = 234 ]; then

			LAST_ERR_MSG="do_uncompress(): ${MSG_TO_SHOW}"

		elif [ ${last_r} -ne 0 ]; then

			LAST_ERR_MSG="7za returned non-zero code of ${last_r} when trying to decompress stdin!"

		fi

	fi
	
	# this is our created temp file rather than one supplied to the script so shld be deleted
	rm -f "${mytf}"

elif [ -e "${f_passed_to_script}" ]; then

	# Remember from here on we run 7za which will in most cases output binary data to stdout.
	# So make REALLY sure nothing else gets output after this!
	# Log messages all only to logger.
	do_uncompress "${f_passed_to_script}"
	last_r=$?

	# error message numbers inside do_uncompress(), hopefully none clash with those used by 7za
	if [ ${last_r} = 184 ] || [ ${last_r} = 194 ] || [ ${last_r} = 204 ] || [ ${last_r} = 214 ] || [ ${last_r} = 224 ] || [ ${last_r} = 234 ]; then

		LAST_ERR_MSG="do_uncompress(): ${MSG_TO_SHOW}"

	elif [ ${last_r} -ne 0 ]; then

		LAST_ERR_MSG="7za returned non-zero code of ${last_r} when trying to decompress the file!"

	fi

	# 7za does not delete the original file by default but xz does
	# If stdout was not requested then they will be expecting us to delete it so do that.
	# TO-DO: catch --keep argument and do not delete if it is passed
	if [ "${STDOUT_REQUESTED}" = "N" ]; then
	rm -f "${f_passed_to_script}"
	fi

else

	last_r=1
	LAST_ERR_MSG="no valid file was found in supplied arguments and stdin was not an xz stream!"

fi

if [ ${last_r} -ne 0 ]; then

	${LOGGER} -p syslog.err -t "${0}" "${LOG_PREFIX} Fatal Error: ${LAST_ERR_MSG}" >/dev/null 2>&1
	exit 1

else

	exit 0

fi