: ksh regression test harness :

USAGE_LICENSE="[-author?David Korn <dgk@research.att.com>][-author?Glenn Fowler <gsf@research.att.com>][-copyright?Copyright (c) 2000-2012 AT&T Intellectual Property][-license?http://www.eclipse.org/org/documents/epl-v10.html]"

command=shtests

setslocale='*@(locale).sh'
timesensitive='*@(options|sigchld|subshell).sh'
valgrindflags='--xml=yes --log-file=/dev/null --track-origins=yes --read-var-info=yes'

USAGE=$'
[-s8?
@(#)$Id: shtests (AT&T Research/ksh93) 2020-08-11 $
]
'$USAGE_LICENSE$'
[+NAME?shtests - ksh regression test harness]
[+DESCRIPTION?\bshtests\b is the \bksh\b(1) regression test harness for
    \b$SHELL\b or \bksh\b if \bSHELL\b is not defined and exported. If
    none of the \b--posix --utf8 --compile\b options are specified then
    all three are enabled.]
[+INPUT FILES?\bshtests\b regression test files are shell scripts that
    run in an environment controlled by \bshtests\b. An identification
    message is printed before and after each test on the standard output.
    The default environment settings are:]
    {
        [+unset LANG]
        [+unset LC_ALL]
        [+LC_NUMERIC=C?\b.\b radix point assumed by all test scripts.]
        [+VMALLOC_OPTIONS=abort?\bvmalloc\b(1) arena checking enabled
            with \babort(2)\b on error.]
    }
[c:compile?Run test scripts using \bshcomp\b(1).]
[d:debug?Enable \bshtests\b execution trace.]
[k:keep?Keep temporary files after test run; shtests will report the location.]
[l:locale?Disable \b--utf8\b and run the \b--posix\b and \b--compile\b
    tests, if enabled, in the locale of the caller. However, for locales
    where \b.\b is not the radix point, \bLC_NUMERIC\b is set to \bC\b
    to avoid invalid regressions.]
[p:posix?Run the test scripts in the posix/C locale.]
[t!:time?Include the current date/time in the test identification
    messages.]
[u:utf8?Run the test scripts in the ast-specific C.UTF-8 locale.]
[v!:vmalloc_options?Run tests with \bVMALLOC_OPTIONS=abort\b. Test
    script names matching \b'$timesensitive$'\b are run with
    \bVMALLOC_OPTIONS\b unset.]
[V:valgrind?Set \b--novmalloc_options\b and run the test scripts with
    \bvalgrind\b(1) on \bksh\b. If \b$SHELL-g\b exists and is executable,
    then it is used instead of \b$SHELL\b.]
[x:trace?Enable script execution trace.]

[ test.sh ... ] [ name=value ... ]

[+SEE ALSO?\bksh\b(1), \bregress\b(1), \brt\b(1)]
'

function usage
{
	OPTIND=0
	getopts -a $command "$USAGE" OPT '--??long'
	exit 2
}

function valxml
{
	typeset state=INIT data dir file fn line what
	integer errors=0

	#print === $1 ===; cat $1; print === ===
	while	read data
	do	case $state in
		INIT)	case $data in
			'<error>')
				state=ERROR
				;;
			esac
			;;
		ERROR)	case $data in
			'<kind>'Leak*'</kind>')
				state=SKIP
				;;
			'<kind>'*'</kind>')
				state=KEEP
				what=UNKNOWN
				;;
			esac
			;;
		FRAME)	case $data in
			'<dir>'*'</dir>')
				dir=${data#'<dir>'}
				dir=${dir%'</dir>'}
				;;
			'<file>'*'</file>')
				file=${data#'<file>'}
				file=${file%'</file>'}
				;;
			'<fn>'*'</fn>')
				fn=${data#'<fn>'}
				fn=${fn%'</fn>'}
				;;
			'<line>'*'</line>')
				line=${data#'<line>'}
				line=${line%'</line>'}
				;;
			'</frame>')
				[[ $dir ]] && dir+=/
				dir+=$file
				[[ $dir ]] && dir+=:
				[[ $line ]] && dir+=$line:
				[[ $fn ]] && dir+=$fn
				[[ $dir ]] && echo $'\t    '$dir
				state=KEEP
				;;
			esac
			;;
		KEEP)	case $data in
			'<auxwhat>'*'</auxwhat>')
				what=${data#'<auxwhat>'}
				what=${what%'</auxwhat>'}
				echo $'\t'"$what"
				;;
			'<frame>')
				state=FRAME
				dir=
				file=
				fn=
				line=
				;;
			'<what>Syscall param mount(type) points to unaddressable byte(s)</what>')
				state=SKIP
				;;
			'<what>'*'</what>')
				(( errors++ ))
				what=${data#'<what>'}
				what=${what%'</what>'}
				echo $'\n\t'"$what"
				;;
			'<xwhat>')
				state=WHAT
				;;
			'</error>')
				state=INIT
				;;
			esac
			;;
		SKIP)	case $data in
			'</error>')
				state=INIT
				;;
			esac
			;;
		WHAT)	case $data in
			'<text>'*'</text>')
				(( errors++ ))
				what=${data#'<text>'}
				what=${what%'</text>'}
				echo $'\n\t'"$what"
				;;
			'</xwhat>')
				state=KEEP
				;;
			esac
			;;
		esac
	done < "$1"
	(( errors )) && echo
	return $errors
}

unset DISPLAY FIGNORE HISTFILE
export ENV=/./dev/null
trap + PIPE # unadvertized -- set SIGPIPE to SIG_DFL #

integer compile=-1 posix=-1 utf8=-1
integer debug=0 keep=0 locale=0 time=1
typeset vmalloc_options=abort trace= valgrind=
vmalloc_options= #XXX# until multi-region vmalloc trace fixed #XXX#

while	getopts -a $command "$USAGE" OPT
do	case $OPT in
	c)	if	(( $OPTARG ))
		then	compile=2
		else	compile=0
		fi
		;;
	d)	debug=$OPTARG
		;;
	k)	keep=$OPTARG
		;;
	l)	locale=$OPTARG
		;;
	p)	posix=$OPTARG
		;;
	t)	time=$OPTARG
		;;
	u)	utf8=$OPTARG
		;;
	v)	if	(( OPTARG ))
		then	vmalloc_options=abort
		else	vmalloc_options=
		fi
		;;
	V)	valgrind="${VALGRIND:-valgrind} ${VALGRINDFLAGS:-$valgrindflags}"
		vmalloc_options=
		;;
	x)	trace=-x
		;;
	*)	usage
		;;
	esac
done
shift $OPTIND-1

if	(( debug )) || [[ $trace ]]
then	export PS4='+ [${SECONDS:+${SECONDS%????}s|}${.sh.subshell:+S${.sh.subshell},}${.sh.file:+${.sh.file#${.sh.file%/*/*}/},}${.sh.fun:+${.sh.fun},}${LINENO:+L$LINENO,}e$?] '
	if	(( debug ))
	then	set -x
	fi
fi

while	[[ $1 == *=* ]]
do	eval export "$1"
	shift
done

if	(( compile <= 0 && posix <= 0 && utf8 <= 0 ))
then	(( compile )) && compile=1
	(( posix )) && posix=1
	(( utf8 )) && utf8=1
fi
(( compile < 0 )) && compile=0
(( posix < 0 )) && posix=0
(( utf8 < 0 )) && utf8=0
if	(( locale ))
then	utf8=0
	if	[[ $LC_ALL ]]
	then	export LANG=$LC_ALL
		unset ${!LC_*}
	fi
	if	! let 1.0 2>/dev/null
	then	export LC_NUMERIC=C
	fi
else	unset LANG LC_ALL
	export LC_NUMERIC=C
fi
if	[[ $VMALLOC_OPTIONS ]]
then	vmalloc_options=$VMALLOC_OPTIONS
else	VMALLOC_OPTIONS=$vmalloc_options
fi
[[ $VMALLOC_OPTIONS ]] || timesensitive=.
export PATH PWD SHCOMP SHELL VMALLOC_OPTIONS
PWD=$(pwd)
SHELL=${SHELL-ksh}
case $0 in
/*)	d=$(dirname $0);;
*/*)	d=$PWD/$(dirname $0);;
*)	d=$PWD;;
esac
case $SHELL in
/*)	;;
*/*)	SHELL=$d/$SHELL;;
*)	SHELL=$(whence $SHELL);;
esac
PATH=$(
	PATH=/run/current-system/sw/bin:/usr/xpg7/bin:/usr/xpg6/bin:/usr/xpg4/bin:/bin:/usr/bin:$PATH
	getconf PATH 2>/dev/null || { builtin getconf 2>/dev/null && getconf PATH; }
) || PATH=/bin:/usr/bin
if	[[ -d /usr/ucb ]]
then	PATH=$PATH:/usr/ucb
fi
PATH=$PATH:$d
if	[[ $INSTALLROOT && -r $INSTALLROOT/bin/.paths ]]
then	PATH=$INSTALLROOT/bin:$PATH
fi
if	[[ ${SHELL%/*} != $INSTALLROOT/bin ]]
then	PATH=${SHELL%/*}:$PATH
fi
if	[[ ! $SHCOMP ]]
then	s=${SHELL:##*sh}
	s=${SHELL:%/*}/shcomp$s
	if	[[ -x $s ]]
	then	SHCOMP=$s
	elif	[[ -x ${s%-g} ]]
	then	SHCOMP=${s%-g}
	else	SHCOMP=shcomp
	fi
fi

tmp=$(
	d=${TMPDIR:-/tmp}/ksh93.shtests.$$.${RANDOM:-0}
	mkdir -m700 -- "$d" && CDPATH= cd -P -- "$d" && pwd
) || {
	echo 'mkdir failed' >&2
	exit 1
}
if	(( keep ))
then	trap 'printf "\nTemporary files left in: %s\n" "$tmp"' EXIT
else	trap 'cd / && rm -rf "$tmp"' EXIT
fi

# for interactive shell ('ksh -i') tests: avoid affecting ~/.sh_history
export HISTFILE=$tmp/sh_history

if	(( compile ))
then	if	whence $SHCOMP > /dev/null
	then	:
	elif	(( compile > 1 ))
	then	echo $0: --compile: $SHCOMP not found >&2
		exit 1
	else	compile=0
	fi
fi
if	[[ $valgrind ]]
then	if	[[ -x $SHELL-g ]]
	then	SHELL=$SHELL-g
	fi
	valxml=$tmp/valgrind.xml
	valgrind+=" --xml-file=$valxml"
fi
typeset -A tests
typeset -i total_e=0
for i in ${*-*.sh}
do	[[ $i == *.sh ]] || i+='.sh'
	if	[[ ! -r $i ]]
	then	echo $0: $i: not found >&2
		(( ++total_e ))
		continue
	fi
	t=$(grep -c err_exit $i)
	if	(( t > 2 ))
	then	(( t = t - 2 ))
	fi
	tests[$i]=$t
	T=test
	if	(( t != 1 ))
	then	T=${T}s
	fi
	u=${i##*/}
	u=${u%.sh}
	if	[[ $i == $timesensitive ]]
	then	VMALLOC_OPTIONS=
	fi
	if	(( posix || utf8 ))
	then	locales=
		(( posix )) && locales+=" ${LANG:-C}"
		[[ $utf8 == 0 || $i == $setslocale ]] || locales+=" C.UTF-8"
		for lang in $locales
		do	o=$u
			tmp_s=$tmp/$u.$lang
			mkdir -m700 "$tmp_s" || exit
			if	[[ $lang == C ]]
			then	lang=
			else	o="$o($lang)"
				lang=LANG=$lang
			fi
			echo test $o begins ${time:+"at $(date +%Y-%m-%d+%H:%M:%S)"}
			(
				export ${lang:+"$lang"} "tmp=$tmp_s"
				$valgrind $SHELL $trace $i
			)
			e=$?
			if	(( !keep ))
			then	rm -rf "$tmp_s"
			fi
			if	[[ $valgrind ]]
			then	valxml $valxml
				(( e += $? ))
			fi
			if	(( e > 128 && ++total_e ))
			then	E="(killed by SIG$(kill -l "$e"))"
			elif	(( e == 1 && ++total_e ))
			then	E="1 error"
			else	E="$e errors"
				(( total_e += e ))
			fi
			if	(( e == 0 ))
			then	echo test $o passed ${time:+"at $(date +%Y-%m-%d+%H:%M:%S)"} "[ $t $T $E ]"
			else	echo test $o failed ${time:+"at $(date +%Y-%m-%d+%H:%M:%S)"} with exit code $e "[ $t $T $E ]"
			fi
			case	$E in
			*INT\))	kill -s INT $$ ;;  # if test was ^C'd, don't keep going but pass down SIGINT
			esac
		done
	fi
	if	(( compile ))
	then	c=$tmp/shcomp-$u.ksh
		o="$u(shcomp)"
		echo test $o begins ${time:+"at $(date +%Y-%m-%d+%H:%M:%S)"}
		tmp_s=$tmp/$u.shcomp
		mkdir -m700 "$tmp_s" || exit
		if	$SHCOMP $i > $c
		then	if	tmp=$tmp_s $valgrind $SHELL $trace $c
			then	echo test $o passed ${time:+"at $(date +%Y-%m-%d+%H:%M:%S)"} "[ $t $T 0 errors ]"
			else	e=$?
				if	(( e > 128 && ++total_e ))
				then	E="(killed by SIG$(kill -l "$e"))"
				elif	(( e == 1 && ++total_e ))
				then	E="1 error"
				else	E="$e errors"
					(( total_e += e ))
				fi
				echo test $o failed ${time:+"at $(date +%Y-%m-%d+%H:%M:%S)"} with exit code $e "[ $t $T $E ]"
				case	$E in
				*INT\))	kill -s INT $$ ;;  # if test was ^C'd, don't keep going but pass down SIGINT
				esac
			fi
		else	e=$?
			(( ++total_e ))
			echo test $o failed to compile ${time:+"at $(date +%Y-%m-%d+%H:%M:%S)"} with exit code $e "[ 1 test 1 error ]"
		fi
		if	(( !keep ))
		then	rm -rf "$tmp_s" "$c"
		fi
		if	[[ $i == $timesensitive ]]
		then	VMALLOC_OPTIONS=$vmalloc_options
		fi
	fi
done
print "Total errors: $total_e"

# This test harness should continue to work with old and buggy versions of ksh93, so check
# if we have the 'times' builtin and can use process substitution with output redirection.
# See:	https://github.com/ksh93/ksh/commit/65d363fd
#	https://github.com/ksh93/ksh/issues/2
if 	(	ulimit -t unlimited	# fork to circumvent old ksh bugs
		unalias times
		PATH=/dev/null whence -q times || exit
		cd "$tmp" || exit
		echo ok > >(cat > procsubst_test)
		wait
		[[ $(< procsubst_test) == ok ]]
	) 2>/dev/null
then	# Transforming 'times' output requires a process substitution.
	# A regular 'times | ...' pipeline would cause 'times' to be
	# run in a subshell, which would reset all the times to zero.
	times > >(
		t[0]='CPU time'	t[1]='user:'	t[2]='system:'
		t[3]='main:'; 	read t[4]	t[5]
		t[6]='tests:';	read t[7]	t[8]
		printf '%-8s%12s %12s\n' "${t[@]}"
	)
	wait "$!"  # process substitutions are executed asynchronously
else	# No pretty-printing.
	times
fi

exit $((total_e > 125 ? 125 : total_e))