#!/bin/sh
#
# typical usage:
# test-from-commits -b main -t pcp -r 6bad98e4b0537c2e99b3906ca7a36f4998ea3419..HEAD
#

tmp=/var/tmp/$$
trap "rm -f $tmp.*; exit 0" 0 1 2 3 15

# find QA tests containing a particular qa application (qa/src/foo) or
# archive (qa/archives/foo) or qa PMDA (qa/pmda/foo) or linux fake
# proc file (qa/linux/foo) or similar ...
#
_hunt_qa()
{

    cd $topdir/qa

    # use the file name to produce an grep -E pattern ($_pat) to be
    # used to search in the QA tests ... to skip the file name, don't
    # set $_pat
    # - use .meta as the representative for each archive, skip the
    #   .index and .N files
    # for some special cases, e.g. archives, they are used but not
    # named in QA tests, so grep -E won't help ... set $_tests to a
    # list of QA tests in this case
    #
    _pat=''
    _tests=''
    qa_file=`echo "$1" | sed -e 's/^qa\///'`
    #debug# echo >&2 "qa_file=$qa_file"
    case "$qa_file"
    in
	# not implicated in any particular QA test
	#
	GNUmakefile*\
       |Makefile\
       |all-by-group\
       |bad-by-group\
       |buildbot/*\
       |check*\
       |chk*\
       |cull-pmlogger-config\
       |disk-work\
       |findmetric\
       |getpmcdhosts\
       |helgrind-suppress\
       |mk.*\
       |new\
       |new-seqs\
       |qa_hosts.main\
       |recheck\
       |remake\
       |sanity.coverage\
       |setup.051\
       |show-me\
       |sniff4dodgey\
       |xlate_2_new_pmns)		# not used in any QA test scripts
	    ;;
	[0-9][0-9][0-9][-.]*\
       |[0-9][0-9][0-9][0-9][-.]*)	# not used in any QA test scripts
	    _tests="`echo $qa_file | sed -e 's/[-.].*//'`"
	    ;;
	archives/*.[0-9]*\
       |archives/*.index*\
       |archives/*.pmns\
       |archives/config.*\
       |archives/*.config\
       |archives/GNUmakefile*\
       |archives/idtohex\
       |archives/mk.*\
       |archives/multi*/GNUmakefile*)	# not used in any QA test scripts
	    ;;
	# these files are not used in any QA test as far as we can tell
	# ...
	archives/cputime.meta\
       |archives/cputime.mpstat\
       |archives/cputime.sar\
       |archives/pcp-atop-log.folio)	# not used in any QA test scripts
	    ;;
	# search full filename
	#
	archives/collectl-qa\
       |archives/crontab-sar-report.tar.xz\
       |archives/ds390-access.gz\
       |archives/pcp-atop.folio\
       |archives/pcp-hotatop.folio)
	    _pat="$qa_file"
	    ;;
	# the QA test scripts
	#
	[0-9][0-9][0-9]|[0-9][0-9][0-9][0-9])
	    _tests="$qa_file"
	    ;;
	# special cases ...
	#
	20071023-08-06-36.*)
	    _pat='20071023-08-06-36\.'
	    ;;
	archives/archive-goodred-20150417.meta*\
       |archives/cputime.meta*\
       |archives/new_mysql.meta*\
       |archives/new_snmp.meta*\
       |archives/new_tcp.meta*\
       |archives/oview-short.meta*\
       |archives/pmatop-log.meta*\
       |archives/reduce-2.meta*\
       |archives/spotinstance.meta*)
	    _tests='661
1225'
	    ;;
	archives/src-vldb-disks.meta*\
       |archives/viewqa?.meta*)
	    _tests='661'
	    ;;
	archives/mark_no_mark*.meta*)
	    _tests='661
1225'
	    _pat=archives/mark_no_mark_
	    ;;
	archives/nfsclient_??.meta*)
	    _tests='661
1225'
	    _pat=archives/nfsclient_
	    ;;

	# default archive pattern
	#
	archives/*.meta*)
	    _pat=`echo "$qa_file" | sed -e 's/\.bz$//' -e 's/\.xz$//' -e 's/\.[^.]*$//' -e 's/$/( |\.|$)/'`
	    ;;
	badarchives/*.[0-9]*\
       |badarchives/*.index*\
       |badarchives/config.*\
       |badarchives/*.config\
       |badarchives/GNUmakefile*\
       |badarchives/mk.*)		# not used in any QA test scripts
	    ;;
	badarchives/541380.meta*\
       |badarchives/ace.meta*\
       |badarchives/babylon.percpu.meta*\
       |badarchives/babylon.pmview.meta*\
       |badarchives/bigace.meta*\
       |badarchives/bug.meta*)
	    _tests='661
663'
	    ;;
	badarchives/badlabel-*.meta*\
       |badarchives/badlen-*.meta*\
       |badarchives/badlog-*.meta*\
       |badarchives/badmeta-*.meta*\
       |badarchives/badti-*.meta*)
	    _pat=`echo "$qa_file" | sed -e 's/\-.*/-/'`
	    ;;
	badarchives/*.meta*)
	    _pat=`echo "$qa_file" | sed -e 's/\.bz$//' -e 's/\.xz$//' -e 's/\.[^.]*$//' -e 's/$/( |\.|$)/'`
	    ;;
	cifs/*)
	    _tests='-g pmda.cifs'
	    ;;
	cisco/*)
	    _tests=383
	    ;;
	collectl/*)
	    _tests='-g collectl'
	    ;;
	common.bcc)
	    _tests='-g pmda.bcc'
	    ;;
	common.bpf)
	    _tests='-g pmda.bpf'
	    ;;
	ganglia/*)
	    _tests='-g ganglia2pcp'
	    ;;
	gluster/*)
	    _tests='-g pmda.gluster'
	    ;;
	gpfs/*)
	    _tests='-g pmda.gpfs'
	    ;;
	haproxy/*)
	    _tests='-g pmda.haproxy'
	    ;;
	interact/*)			# not used in any QA test scripts
	    ;;
	java/*)				# not used?
	    ;;
	json/*)
	    _tests='884
1101'
	    ;;
	# not used in any QA test
	linux/GNUmakefile*\
       |linux/bigsys-root-hpbl920gen8.tgz\
       |linux/mk.procpid)		# not used in any QA test scripts
	    ;;
	linux/cgroups-*)
	    _tests='-g cgroups'
	    ;;
	linux/containers-docker-*\
       |linux/containers-lxc-*)
	    _tests='761'
	    ;;
	linux/cpuinfo-[0-9]*)
	    _tests='885'
	    ;;
	linux/interrupts-[0-9]*\
       |linux/softirqs-[0-9]*)
	    _tests='886'
	    ;;
	linux/nfsrpc-*)
	    _tests='732'
	    ;;
	linux/proc_serial_*)
	    _tests='665'
	    ;;
	linux/procnet-[0-9]*)
	    _tests='1164'
	    ;;
	linux/procnet-snmp*)
	    _tests='872'
	    ;;
	linux/procpid-[0-9]*)
	    _tests='735
852
974'
	    ;;
	linux/procsys-root-*)
	    _tests='1073'
	    ;;
	linux/sysdev-root-*)
	    _tests='747'
	    ;;
	linux/xfs-root-*)
	    _tests='987
988'
	    ;;
	linux/*)	# all over the shop ...
	    _pat=`echo "$qa_file" | sed -e 's/\.[^.]*$//' -e 's/-[0-9][0-9][0-9]$//'`
	    ;;
	lio/*)
	    _tests='-g pmda.lio'
	    ;;
	lustre/*)
	    _tests='-g pmda.lustre'
	    ;;
	mic/*)
	    _tests='-g pmda.mic'
	    ;;
	mmv/*)
	    _tests='-g pmda.mmv'
	    ;;
	named/*)
	    _tests='-g pmda.named'
	    ;;
	nfsclient/*)
	    _tests='-g pmda.nfsclient'
	    ;;
	pconf/*)
	    _tests='-g pmieconf'
	    ;;
	perfevent/*)
	    _tests='-g pmda.perfevent'
	    ;;
	perl/*)		# not used?
	    ;;
	pmdas/GNUmakefile*\
       |pmdas/*/GNUmakefile*\
       |pmdas/*/Install\
       |pmdas/*/Remove)			# not used in any QA test scripts
	    ;;
	pmdas/bigun/*\
       |pmdas/schizo/*)			# not used in any QA test scripts
	    ;;
	pmdas/*)
	    _pat=`echo "$qa_file" | sed -e 's/\(.*pmdas\/[^/]*\)\/.*/\1/' -e 's/$/( |\/|$)/'`
	    ;;
	pmlogconv/*)			# not used in any QA test scripts
	    ;;
	postfix/*)
	    _tests='-g pmda.postfix'
	    ;;
	prometheus/*)
	    _tests='-g pmda.prometheus'
	    ;;
	qt/*)
	    _tests='-g libpcp_qmc'
	    ;;
	sadist/*\
       |sheet/*)
	    _tests='-g pmimport'
	    ;;
	secure/*)
	    _tests='550
551'
	    ;;
	slurm/*)
	    _tests='-g pmda.slurm'
	    ;;
	smart/*)
	    _tests='-g pmda.smart'
	    ;;
	src/*.index\
       |src/*.meta\
       |src/*.[0-9]\
       |src/*.config\
       |src/*.h\
       |src/GNUlocaldefs\
       |src/GNUmakefile*\
       |src/err_v1.dump\
       |src/fixhosts\
       |src/grind-tools\
       |src/ipcs_clear\
       |src/mkpermslist\
       |src/mv-me\
       |src/permslist\
       |src/root_*\
       |src/show-args)			# not used in any QA test scripts
	    ;;
	src/anon-sa.c\
       |src/chkacc4.c\
       |src/hp-mib.c\
       |src/indom2int.c\
       |src/int2indom.c\
       |src/int2pmid.c\
       |src/iohack.c\
       |src/mmv_ondisk.c\
       |src/pcp_lite_crash.c\
       |src/pmid2int.c\
       |src/qa_test.c\
       |src/qa_timezone.c\
       |src/rootclient.c\
       |src/sha1int2ext.c\
       |src/statvfs.c\
       |src/test_pmda_memleak.python\
       |src/versiondefs)	# not used in any QA test scripts
	    ;;
	src/nvidia-ml.c)
	    _pat='src/libnvidia-ml'
	    ;;
	src/sgi.bf)
	    _tests='377'
	    ;;
	src/tiny.pmns)
	    _tests='052'
	    ;;
	# default QA app
	#
	src/*.c)
	    _pat=`echo "$qa_file" | sed -e 's/\.[^.]*$//' -e 's/$/( |;|&|>|<|\||$)/'`
	    ;;
	# Perl and Python scripts
	#
	src/*.p[ely]*)
	    _pat=`echo "$qa_file" | sed -e 's/\.[^.]*$/\\./'`
	    ;;
	tmparch/*)			# made on the fly, not checked in
	    ;;
	unbound/*)
	    _tests='-g pmda.unbound'
	    ;;
	valgrind-*)
	    _tests='-g valgrind'
	    ;;
	views/*)
	    _tests='1006'
	    ;;
	*)
	    echo >&2 "Warning: $1: no QA test search pattern produced"
    esac

    if [ -n "$_pat" ]
    then
	grep -E -l "$_pat" [0-9][0-9][0-9] [0-9][0-9][0-9][0-9] >$tmp.tmp
	if [ -s $tmp.tmp ]
	then
	    cat $tmp.tmp
	else
	    echo >&2 "Warning: $1: no QA test matches pattern: \"$_pat\""
	fi
    fi
    if [ -n "$_tests" ]
    then
	echo "$_tests"
    fi

    cd $here
}

# remove files that don't drive QA
#
_cull()
{
    sed \
	-e '/^aclocal.m4$/d' \
	-e '/^CHANGELOG$/d' \
	-e '/^configure$/d' \
	-e '/^configure\.in$/d' \
	-e '/^config\./d' \
	-e '/^COPYING$/d' \
	-e '/^\.dir-locals\.el$/d' \
	-e '/^Makefile$/d' \
	-e '/^install-sh$/d' \
	-e '/^pcp\.lsm\.in$/d' \
	-e '/^\.pylintrc$/d' \
	-e '/^README/d' \
	-e '/\/README/d' \
	-e '/^\.travis\.yml$/d' \
	-e '/^Vagrantfile/d' \
	-e '/^VERSION/d' \
	-e '/^[^/]*\.md$/d' \
	-e '/^books\//d' \
	-e '/^debian\/copyright$/d' \
	-e '/^doc\//d' \
	-e '/^images\//d' \
	-e '/^m4\//d' \
	-e '/^provisioning\//d' \
	-e '/^qa\/admin\//d' \
	-e '/^qa\/group$/d' \
	-e '/^qa\/group-stats$/d' \
	-e '/^qa\/common/d' \
	-e '/^src\/libpcp_qwt/d' \
	-e '/^.gitignore$/d' \
	-e '/\/.gitignore$/d' \
	-e '/\.pro$/d' \
	-e '/\.ui$/d' \
	-e '/\.ui\.h$/d' \
	-e '/\/Makefile$/d' \
    # end
}

# hard work happens here ...
#
# map a pathname in the git tree onto QA group(s)
_map()
{
    while read f
    do
	case "$f"
	in
	    Makepkgs\
	   |configure.ac\
	   |src/include/*\
	   |build/*\
	   |debian/*\
	   |GNUmakefile)
		# not much we can do here ... the scope is potentially huge
		# but if it builds OK then check packaging
		#
		echo '-g verify'
		;;
	    man/*\
	   |*.1)
		echo '-g man'
		;;
	    qa/*)
		_hunt_qa "$f"
		;;
	    scripts/*)			# not used in any QA test scripts
		;;
	    src/GNUmakefile\
	   |src/Makefile)		# not used in any QA test scripts
		;;
	    src/autofsd-probe/*)	# no direct QA tests
		;;
	    src/bashrc/*)
		echo '-g bash'
		;;
	    src/chart/*) 		# long gone -> pmchart
		echo '-g pmchart'
		;;
	    src/derived/*)
		echo '-g derive'
		;;
	    src/discover/*)
		echo '-g pmfind'
		;;
	    src/external/*)		# other people's code, no explicit QA tests
		;;
	    src/genpmda/*)		# no longer useful nor tested
		;;
	    src/libpcp/src/*derive*)
		echo '-g derive'
		;;
	    src/libpcp/src/p_*.c\
	   |src/libpcp/src/*pdu*.c)
		echo '-g pdu'
		;;
	    src/libpcp/src/*fetch*.c)
		echo '-g fetch'
		;;
	    src/libpcp/src/log*.c\
	   |src/libpcp/src/interp.c)
		echo '-g archive'
		;;
	    src/libpcp/src/context.c)
		echo '-g context'
		;;
	    src/libpcp/src/connectlocal.c\
	   |src/libpcp/src/fetchlocal.c)
		echo '-g context_local'
		;;
	    src/libpcp/src/events.c)
		echo '-g event'
		;;
	    src/libpcp/src/getopt.c)
		echo '-g getopt'
		;;
	    src/libpcp/src/instance.c\
	   |src/libpcp/src/sortinst.c)
		echo '-g indom'
		;;
	    src/libpcp/src/io_xz.c)
		echo '-g decompress-xz'
		;;
	    src/libpcp/src/io*.c)
		echo '-g multi-archive'
		;;
	    src/libpcp/src/secure*.c)
		echo '-g secure'
		;;
	    src/libpcp/src/store.c)
		echo '-g pmstore'
		;;
	    src/libpcp/src/lock.c)
		echo '-g threads'
		;;
	    src/libpcp/*)
		echo '-g libpcp'
		;;
	    src/libpcp_fault/*)
		echo '-g fault'
		;;
	    src/libpcp_import/*)
		echo '-g pmimport'
		;;
	    src/libpcp_pmda/*)
		echo '-g pmda'
		;;
	    src/libpcp_trace/*)
		echo '-g trace'
		;;
	    src/libpcp_*)
		echo "$f" | sed -e 's/src\///' -e 's/\/.*//' -e 's/^/-g /'
		;;
	    src/newhelp/*)
		echo '-g help'
		;;
	    src/pcp2*)
		echo "$f" | sed -e 's/src\///' -e 's/\/.*//' -e 's/^/-g /'
		;;
	    src/*2pcp/*)
		echo '-g pmimport'
		;;
	    src/pcp/atop/*)
		echo '-g atop'
		;;
	    src/pcp/collectl/*)
		echo '-g collectl'
		;;
	    src/pcp/dstat/*)
		echo '-g dstat'
		;;
	    src/pcp/iostat/*)
		echo '-g pmiostat'
		;;
	    src/pcp/pidstat/*)
		echo '-g pidstat'
		;;
	    src/pmafm/*)
		echo '-g folio'
		;;
	    src/pmconfig/*)		# no direct QA tests, used in a few for _notrun guards
		;;
	    src/pmdbg/*)
		echo '219'
		;;
	    src/pmerr/*)
		echo '006
610
636'
		;;
	    src/pmdas/aix/*\
	   |src/pmdas/darwin/*\
	   |src/pmdas/freebsd/*\
	   |src/pmdas/linux/*\
	   |src/pmdas/netbsd/*\
	   |src/pmdas/openbsd/*\
	   |src/pmdas/solaris/*\
	   |src/pmdas/windows/*)
		# same as other PMDAs plus -g kernel
		echo "$f" | sed -e 's/src\/pmdas\//pmda./' -e 's/\/.*//' -e 's/^/-g /'
		echo '-g kernel'
		;;
	    src/pmdas/*)
		echo "$f" | sed -e 's/src\/pmdas\//pmda./' -e 's/\/.*//' -e 's/^/-g /'
		;;
	    src/pmgenmap/*)
	    	echo '021'
		;;
	    src/pmgetopt/*)
		echo '-g getopt'
		;;
	    src/pmhostname/*)		# no direct QA tests, used in some
		echo '054
116
215'
		;;
	    src/pmie/control*\
	   |src/pmie/crontab*\
	   |src/pmie/pmie_*\
	   |src/pmie.service*)
		echo '-g pmieutil'
		;;
	    src/pmiestatus/*)		# not used in any QA test scripts
		;;
	    src/pmjson/*)
		echo '1234'
		;;
	    src/pmgenmap/*)
	    	echo '021'
		;;
	    src/pmlock/*)		# no direct QA tests, used in some
		echo '638
639'
		;;
	    src/pmgetopt/*)
		echo '-g getopt'
		;;
	    src/pmlogger/control*\
	   |rc/pmlogger/crontab*\
	   |rc/pmlogger/pmlogger_*\
	   |rc/pmlogger.service*)
		echo '-g logutil'
		;;
	    src/pmlogger/pmnewlog.sh)
		echo '-g pmnewlog'
		;;
	    src/pmlogger/pmlogmv.sh)
		echo '-g pmlogmv'
		;;
	    src/pmlogsummary/pmdiff.sh)
		echo '-g pmdiff'
		;;
	    src/pmpython/*)
		echo '863'
		;;
	    src/pmsignal/*)		# used everywhere, this is a specific QA test
		echo '881'
		;;
	    src/pmsleep/*)		# used everywhere, no specific QA test
		;;
	    src/pmsnap/*)		# not used in any QA test scripts
		;;
	    src/pmtrace/*)
		echo '601'
		;;
	    src/procmemstat/*)		# not used in any QA test scripts
		;;
	    src/telnet-probe/*)
		echo '589'
		;;
	    src/time/*) 		# long gone -> pmtime
		echo '-g pmtime'
		;;
	    src/win32ctl/*)		# not used in any QA test scripts
		;;
	    src/zabbix-agent/*)
		echo '-g zbxpcp'
		;;
	    src/zshrc/*)
		echo '-g zsh'
		;;

	    # for all of these, the QA group is the same as the directory
	    # below src
	    #
	    src/dbpmda/*\
	   |src/pcp/*\
	   |src/pmclient/*\
	   |src/pmchart/*\
	   |src/pmcd/*\
	   |src/pmcd_wait/*\
	   |src/pmcpp/*\
	   |src/pmdate/*\
	   |src/pmdumplog/*\
	   |src/pmdumptext/*\
	   |src/pmfind/*\
	   |src/pmgadgets/*\
	   |src/pminfo/*\
	   |src/pmie/*\
	   |src/pmieconf/*\
	   |src/pmlc/*\
	   |src/pmlogger/*\
	   |src/pmlogcheck/*\
	   |src/pmlogconf/*\
	   |src/pmlogextract/*\
	   |src/pmloglabel/*\
	   |src/pmlogreduce/*\
	   |src/pmlogrewrite/*\
	   |src/pmlogsize/*\
	   |src/pmlogsummary/*\
	   |src/pmns/*\
	   |src/*pmpost/*\
	   |src/pmprobe/*\
	   |src/pmproxy/*\
	   |src/pmseries/*\
	   |src/pmsocks/*\
	   |src/pmstat/*\
	   |src/pmstore/*\
	   |src/pmtime/*\
	   |src/pmview/*\
	   |src/pmquery/*\
	   |src/pmrep/*\
	   |src/pmval/*\
	   |src/perl/*\
	   |src/python/*\
	   |src/selinux/*)
		echo "$f" | sed -e 's/src\///' -e 's/\/.*//' -e 's/^/-g /'
		;;

	    tools/*)			# not used in any QA test scripts
		;;
	    *)
		echo >&2 "Warning: $f: no matching QA group from _map"
		;;
	esac
    done
}

# expand -g groupname into the corresponding list of tests
#
_expand()
{

    cd $topdir/qa

    while read line
    do
	case "$line"
	in
	    [0-9][0-9][0-9]|[0-9][0-9][0-9][0-9])
		echo "$line"
		;;

	    -g\ *)
		./check -n $line
		;;

	    *)
		echo >&2 "_expand: botch: $line: unexpected"
		;;
	esac
    done

    cd $here
}


_usage()
{
    echo >&2 "Usage: groups-to-test [options]"
    echo >&2
    echo >&2 "options:"
    echo >&2 "  -b branch           [defaults to main]"
    echo >&2 "  -r range            [defaults to \$(cat tested.sha)..HEAD)]"
    exit 1
}

# find the base of the git tree
#
here=`pwd`
rm -f $tmp.tmp
while true
do
    if [ -d .git ]
    then
	topdir=`pwd`/
	break
    fi
    if [ `pwd` = "/" ]
    then
	echo >&2 "Error: no .git between / and $here, do you know what you're doing?"
	exit 1
    fi
    cd ..
done

if [ -f $topdir/tested.sha ]
then
    range=$(cat $topdir/tested.sha)..HEAD
else
    range=''
fi
branch=main
while getopts "b:r:?" c
do
    case $c
    in
	b)
	    branch="$OPTARG"
	    ;;
	r)
	    range="$OPTARG"
	    ;;
	*)
	    _usage
	    # NOTREACHED
    esac
done
shift `expr $OPTIND - 1`

[ $# -eq 0 ] || _usage

if [ -z "$range" ]
then
    echo "Error: no range from tested.sha, so need -r range"
    exit 1
fi

unset GIT_EXTERNAL_DIFF

git diff --name-only $range \
| _cull \
| _map \
| sort \
| uniq \
| _expand \
| sort -n \
| uniq

exit