#!/bin/bash
#
#    prtwash - a simple bash script for cleaning the port tree
#              of the CRUX Linux distribution.
#
#    Copyright (c) 2003 by Simone Rota                  <sip@varlock.com>
#    Revised 2021 by John McQuah			<jmcquah@disroot.org>
#
#    *************************************************************************
#    *                                                                       *
#    * This program is free software; you can redistribute it and/or modify  *
#    * it under the terms of the GNU General Public License as published by  *
#    * the Free Software Foundation; either version 2 of the License, or     *
#    * (at your option) any later version.                                   *
#    *                                                                       *
#    *************************************************************************
#
#   Took some inspiration and code from Martin Opel's prtsweep v1.6 script,
#     and comments by therealfun on FlySpray bug 1851.
#
#   **** USE AT YOUR OWN RISK ****
#
#   TODO  
#       - a summary of files / bytes removed after the
#         execution would be useful.
#	- verify signatures before sourcing each Pkgfile.
#

VERSION="0.9"

usage() {
    echo "Usage: prtwash [-p] [-s] [-d] [-t] [-a] <path> [<path> ...]"
    exit 1
}

showversion() {
    echo "prtwash" $VERSION
    echo "(c) 2003 by Simone Rota, revised 2021 by John McQuah"
    echo "This program is distributed under the GNU GPL license"
}

interrupted() {
    echo "=======> operation interrupted."
    exit 1
}

checkparams() {
# Do some test on given parameters.
    if [ "$auto" = "0" ]; then

        if [ -z "$prtdirs" ]; then
            usage
            exit -1
        fi

        for (( p=0 ; p<=${#prtdirs[@]} ; p++ )); do
        
            if [ -z "${prtdirs[$p]}" ]; then
                unset prtdirs[$p]
                continue
            fi
            
            if [ ! -d "${prtdirs[$p]}" ]; then
                echo "WARN: '${prtdirs[$p]}' is not the directory for a port. Skipping."
                unset prtdirs[$p]
                continue
            fi

            if [ ! -f "${prtdirs[$p]}"/Pkgfile ]; then
                echo "WARN: no Pkgfile found in '${prtdirs[$p]}'. Skipping."
                unset prtdirs[$p]
            fi
        done
    fi
}

getoptions () {
# Name says it all.
    while getopts bpsdathv opt; do
        case "$opt" in
	    b) removeoldbuilds=1;;
            p) removepackage=1;;
            s) removesources=1;;
            d) removeaddonfiles=1;;
            t) test=1;;
            a) auto=1;;
            h) usage
                exit 0
                ;;
            v) showversion
                exit 0
                ;;
            \?) usage
                exit -1
                ;;
        esac
    done
    shift $(($OPTIND - 1))
    prtdirs=("$@")
}

getdirs() {
# scans /etc/prt-get.conf for port dirs to process
    current=""
    for s in `sed 's/^[ \t]*//;s/[ \t]*$//' \
             $CONF_PGET|grep '^prtdir.*:'|sed 's/:/ /;s/,/ /;s/prtdir//;s/#.*//'`
    do
        if [ "`echo $s|grep '/'`" != "" ]; then
            current=$s
        else
            if [ "$current" != "" ]; then
                singledirs=(${singledirs[*]} $current/$s)
            fi    
        fi
    done

    for s in `sed 's/^[ \t]*//;s/[ \t]*$//;/:/d' \
             $CONF_PGET|grep '^prtdir.*'|sed 's/prtdir//;s/#.*//'`
    do
        basedirs=(${basedirs[*]} $s)
    done

}


remove() {
# Removes a file/directory
    if [ "$test" = 1 ]; then
        echo "+ (t) removing" $1
    else
        echo "+ removing" $1
        rm -r $1
    fi
}

swash() {
# There are faster and safer ways of cleaning a directory shared among all ports than 
# to iterate through the ports collection itself.
# Inform the admin that this script is ill-suited for their configuration.
	echo "It looks like you have defined common directories for downloads or packages." >&2
	echo "For such a configuration, this script is slower and less reliable" >&2
	echo "than a custom tool would be." >&2
	echo "If you intended to clean out your ports repositories instead," >&2
	echo "consider the companion tool prtsweep." >&2
	echo "Proceed anyway? (y/n)" >&2
	read CONFIRM
	if [ "$CONFIRM" != "y" ]; then
		exit 0
	fi
}

finalwarning() {
	echo "Last chance to bail out!" >&2
	echo "This script will remove from the common download directories more files " >&2
	echo "than perhaps you intended." >&2
	echo "Proceed anyway? (y/n)" >&2
	read CONFIRM
	if [ "$CONFIRM" != "y" ]; then
		exit 0
	fi
}

nwash() {
    if [ "$auto" = 0 ]; then

        for p in ${prtdirs[@]}; do
	    echo "Washing the directory of port $p:"
            wash $p
        done

        exit 0
    else

    if [ ! -f "$CONF_PGET" ]; then
        echo "ERROR: cannot find configuration file '$CONF_PGET'"
        exit -1
    fi
    echo "Analyzing port tree..."

    getdirs
    
    # wash base dirs
    for d in ${basedirs[*]}; do
        if [ -d "$d" ]; then
            for p in "$d"/*; do
                if [ -d "$p" ]; then
                    wash $p
                fi  
            done
        else
            echo "ERROR: directory '$d' not found, check your configuration file!" 
        fi 
    done 

    # wash custom dirs (see prt-get.conf)
    for p in ${singledirs[*]}; do
         if [ -d "$p" ]; then
             wash $p
         else
             echo "ERROR: directory '$p' not found, check your configuration file!"
         fi      
    done
    exit 0
    fi
    exit 0
}

wash() {
# Does the actual removal work.
    local tosave name version source renames pkgfiles packagename portname SAVELOC PKGLOC
    local dir=$1
    if [ ! -f "$dir"/Pkgfile ]; then
    # tested before, the test here is for
    # auto (-a) command
        echo "WARN: no Pkgfile found in $dir. Skipping."
    else
        declare -a tosave # array of files not to be deleted
	pkgfiles=(`( . $dir/Pkgfile; PKGLOC=$(eval "printf '%s' $PKGGLOB"); \
		PKGFILE=$(eval "printf '%s' $BLTGLOB"); \
		SAVELOC=$(eval "printf '%s' $SRCGLOB"); \
	    for (( p=0; p<${#source[@]}; p++ )); do \
	    [ -n "${renames[$p]}" -a "${renames[$p]}" != "SKIP" ] && source[$p]="ftp://host/${renames[$p]}"; \
	    done; \
	printf '%s ' "${SAVELOC:-$dir}"; \
	printf '%s ' "${PKGLOC:-$dir}/$PKGFILE"; \
       	printf '%s ' "${source[@]}")`)

	SAVELOC=${pkgfiles[0]}
        packagename=`basename ${pkgfiles[1]}`
	PKGLOC=`dirname ${pkgfiles[1]}`

        # keep sources
	if [ "$removesources" = 0 ]; then
	for (( p=2; p<${#pkgfiles[@]}; p++ )) ; do 
		tosave=( "${tosave[@]}" `get_filename "${pkgfiles[$p]}"` )
	done
	fi

        # new in version 0.3 we (try to) always save files that are
        # retrieved by rsync/httpup (non http(s) or ftp in source)
        # ie patches, etc. These can be deleted with -d (extra files)
        # option
	if [ "$removesources" = 1 ] && [ "$removeaddonfiles" = 0 ]; then
		for (( p=2; p<${#pkgfiles[@]}; p++ )) ; do 
		     src="${pkgfiles[$p]}"
		     pkgfiles[$p]=$(get_filename $src)
		     [ "${src:0:7}" = "ftp://" \
                     -o "${src:0:8}" = "http://" \
                     -o "${src:0:6}" = "https://" ] ||
		     tosave=( "${tosave[@]}" "${pkgfiles[$p]}" )
		done
	fi

	# Now that the distinction between files obtained via rsync/httpup
	# and files downloaded during `pkgmk -d` has been encoded in the
	# array $tosave, clean up the filenames in the $pkgfiles array.
	if [ "$removesources" = 0 ] || [ "$removeaddonfiles" = 1 ]; then
		for (( p=2; p<${#pkgfiles[@]}; p++ )) ; do 
			pkgfiles[$p]=$(get_filename ${pkgfiles[$p]})
		done
	fi

        if [ ! "$removepackage" = 1 ]; then
        # keep package
               tosave=( "${tosave[@]}" "$packagename" )
        fi

        if [ ! "$removeaddonfiles" = 1 ]; then
        # keep additional and dot files
            tosave=( "${tosave[@]}" \
			".signature" \
			".32bit" \
			".footprint" \
			".nostrip" \
			"README" \
			"FAQ" \
			"pre-install" \
			"post-install" )
        fi

        # keep Pkgfile. We always want to keep this nice funny file.
        tosave=( "${tosave[@]}" "Pkgfile" )

	unset pkgfiles[0] pkgfiles[1]

	# trust that our parsing of the Pkgfile has not set the wrong source directory.
	#  But if an absolute path is specified, restrict the `ls -A` output
	#  to avoid overzealous deletions.
	cd $SAVELOC
	if [ "${SAVELOC:0:1}" != "/" ]; then
	    for f in `ls -A`; do
		keep $f || remove $f
            done
	elif [ ${#pkgfiles[@]} -gt 0 ]; then
	    for f in ${pkgfiles[@]}; do
		for match in `ls -A ${f}* 2>/dev/null`; do
		    keep $match || remove $match
	        done
            done
	fi
	cd - &>/dev/null

	# same trust as to the location of the built packages, if different from SAVELOC.
	# Older versions can be discarded too, if $removeoldbuilds = 1.
	if [ "$removepackage" = 1 ]; then
	(cd $PKGLOC; [ -f $packagename ] && keep "$packagename" || remove "$packagename")
	fi

	if [ "$removeoldbuilds" = 1 ]; then
	portname=$(echo "$packagename" | sed 's/#.*//')
	(cd $PKGLOC; for f in `ls ${portname}*.pkg.tar.[bgx]z* 2>/dev/null`; do keep $f || remove $f; done)
	fi
    fi
}

keep() {
# Checks if filename item is in tosave[] array
# return 0 if present, 1 if not.
    item=$1
    local excode
    excode=1 # assume not in the array until proven otherwise

    for i in "${tosave[@]}"; do
	if [ "$i" = "$item" ]; then
	    [ "$test" = "1" ] && echo "= (t) ... keeping $item" || echo "= ... keeping $item"
            excode=0
	    break
        fi
    done

    return $excode
}

get_filename() {
	if [[ $1 =~ ^(http|https|ftp|file)://.*/(.+) ]]; then
		echo "${BASH_REMATCH[2]}"
	else
		echo $1
	fi
}

printparams() {
    echo "removepackage: $removepackage"
    echo "removeoldbuilds: $removeoldbuilds"
    echo "removesources: $removesources"
    echo "removeaddonfiles: $removeaddonfiles"
    echo "auto: $auto"
    echo "test: $test"
}


###################################################
# Main
###################################################

trap "interrupted" SIGHUP SIGINT SIGQUIT SIGTERM

if [ $# -lt 1 ]; then
    usage
    exit -1
fi

removepackage=0
removesources=0
removeaddonfiles=0
removeoldbuilds=0
test=0
auto=0
basedirs=()
singledirs=()
prtdirs=()

CONF_PGET="/etc/prt-get.conf"
CONF_PMK="/etc/pkgmk.conf"

getoptions $@
checkparams $@

COMPRESSION_MODE="gz"
if [ -f $PKGCONFIG ]; then
	COMPRESS_CFG=$(grep "PKGMK_COMPRESSION_MODE" $CONF_PMK | \
	sed 's/[\t w]*#.*//;s/.*=//;s/\"//g')
	[ -n "$COMPRESS_CFG" ] && COMPRESSION_MODE=$COMPRESS_CFG
	SRCGLOB=$(grep "PKGMK_SOURCE_DIR" $CONF_PMK | sed 's/[\t w]*#.*//; s/.*=//; s/\"//g')
	PKGGLOB=$(grep "PKGMK_PACKAGE_DIR" $CONF_PMK | sed 's/[\t w]*#.*//; s/.*=//; s/\"//g')
fi

BLTGLOB='$name#$version-$release.pkg.tar.$COMPRESSION_MODE'

# First determine the type of fs layout the admin configured for source downloads.
# Layouts 1 and 2: at least one directory is shared among all ports (for sources or built packages),
#   or the admin has organized downloads in some other way outside of $PORTS_ROOT.

if [ -n "${SRCGLOB}" ] && echo "${SRCGLOB}" | grep -q -v '\$name'; then
     swash
     nwash
elif [ -n "${PKGGLOB}" ] && echo "${PKGGLOB}" | grep -q -v '\$name'; then
     swash
     nwash
fi

# Layout 3: separate directories, named according to port, are potentially cluttered with downloads.
# Unlike prtsweep, actually try to follow these dynamically-defined paths and clean them up.
# $PORTS_ROOT might still remain cluttered after this "wash", but the admin can just run prtsweep.
# For such filesystem layouts, the two scripts complement each other nicely.
if [ -z "${SRCGLOB}" -o -z "${PKGGLOB}" ]; then
	nwash
elif [ -n "${SRCGLOB}" ] && echo "$SRCGLOB" | grep -q '\$name'; then
	nwash
elif [ -n "${PKGGLOB}" ] && echo "$PKGGLOB" | grep -q '\$name'; then
	nwash
else
	# Should have exited by now, but if not,
	# offer another chance to abort.
	finalwarning
	nwash
fi