#!/bin/ksh
# @(#) pareacct.ksh 1.1 97/06/29
# 97/06/22 john h. dubois iii (john@armory.com)
# 97/06/29 Added pax options.
# todo: clean up core/temp/versioned files first;
# todo: allow du -k as alternative to quota

# This utility requires version 1.7 or later of the 'stat' program,
# available at ftp.armory.com.

### start of doy-date lib
# 97/06/29 john@armory.com
typeset -i MStart
function mstart_setup {
    if [[ $(( `date +%y` % 4 )) = 0 ]]; then
	set -A MStart 0 31 60 91 121 152 182 213 244 274 305 335 366
    else
	set -A MStart 0 31 59 90 120 151 181 212 243 273 304 334 365
    fi
}

function initMonths {
    set -A mNum2Month "" jan feb mar apr may jun jul aug sep oct nov dec
}

function month2num {
    typeset -lL3 month=$1
    typeset -i i=1
    [ ${#mNum2Month} -eq 0 ] && initMonths
    while [ i -le 12 ]; do
	[ ${mNum2Month[i]} = "$month" ] && return $i
	let i+=1
    done
    return 0
}

# doy2date day-of-year
# Returns the day-of-year converted to month & day (separated by a space)
# in global doy2date
# assumes conversion is for the current year
function doy2date {
    typeset -i M
    M=$1/32+1
    [ $1 -gt MStart[M] ] && M=M+1
    doy2date="$M $(($1 - MStart[M-1]))"
}

# date2doy month day-of-month
# Returns the month & day-of-month converted to day-of-year in global date2doy
# assumes conversion is for the current year
function date2doy {
    typeset -i month
    [[ $1 = +([0-9]) ]] && month=$1 || {
	month2num "$1" && return 0
	month=$?
    }
    date2doy=$((MStart[month-1] + $2))
}

# date2days year month day-of-month
# Returns the number of complete days that passed from 1900 Jan 1 to the start
# of the given date in global date2days.
# The month may be given in numeric form (Jan=1) or by name, in which case at
# least the first 3 characters must be passed (case is ignored).
# Works from 1901 to 2099.
# If year < 100, it is assumed to be part of the 1900 century
typeset -i date2days
function date2days {
    typeset -i year=$1 day=$3 leap_days MDays month
    [ year -ge 100 ] && let year-=1900
    let leap_days=year/4+1
    [[ $2 = +([0-9]) ]] && month=$2 || {
	month2num "$2" && return -1
	month=$?
    }
    [[ month -le 2 && $((year%4)) = 0 ]] && let leap_days-=1
    [ ${#MDays[*]} -eq 0 ] &&
    set -A MDays 0 0 31 59 90 120 151 181 212 243 273 304 334 365
    date2days="year*365+MDays[month]+day-1+leap_days"
}

# unixdays year month day-of-month
# Returns the number of complete days that passed from 1970 Jan 1
# to the start of the given date in global unixdays
typeset -i unixdays
function unixdays {
    date2days "$@"
    unixdays=date2days-25568
}

# diffdays year1 month1 day-of-month1 year2 month2 day-of-month2 {
# prints the number of complete days that passed from date 1 to date 2
typeset -i diffdays
function diffdays {
    date2days $4 $5 $6
    diffdays=date2days
    date2days $1 $2 $3
    diffdays=diffdays-date2days
}

# Returns the UNIX epoch time in global curtime
# Depends on SECONDS not being messed with
# If all date utilities had %s this wouldn't be neccessary...
typeset -i curtime _shell_start=-1
function curtime {
    typeset -i Y m d H M S
    if [ _shell_start -eq -1 ]; then
	TZ=0 date '+%Y %m %d %H %M %S' | read Y m d H M S
	unixdays $Y $m $d
	curtime="unixdays*86400+H*3600+M*60+S"
	_shell_start=curtime-SECONDS
    else
	curtime=_shell_start+SECONDS
    fi
}
### end of doy-date lib

typeset -i daysold
function daysold {
    daysold="(curtime-$1)/86400"
}
### start of main program

name=${0##*/}
Usage=\
"Usage: $name [-dhpr] [-a<min-age] [-s<min-size>] [-k<reduction>] user ..."
typeset -i defMinSize=100 reduction=-1 i
typeset -i minSize=defMinSize
remove=false
nodot=true
unset amtime sd
byProduct=false
debug=false
sort="sort -n"

while getopts a:hs:rk:dpx opt; do
    case $opt in
    h)
	print -r -- \
"$name: Reduce usage of users to their disk quota by removing large files,
in order from oldest to newest.  In sorting the files by age, whichever of
the access and modification time for each file is later (more recent) is used.
By default, files are listed but not removed.
$Usage
Options:
-d: Do not skip dotfiles (files whose name begins with '.').  By default, these
    files are not considered candidates for removal.
-k<reduction>: Set the reduction quota for all users to <reduction> KB.
    By default, the reduction quota is set to the difference between a user's
    usage and quota.
-r: Remove files.
-s<min-size>: Set the minimum size file to remove, in KB.
    The default is ${defMinSize} KB.
-a<min-age>: Set the minimum age a file to be removed, in days.  The default is
    to remove files all the way to the current time if neccessary.
-p: Remove files in order of age,filesize product.  For each file, the age
    (as described above) and filesize are multiplied, and files are removed
    in order of largest to smallest of this product.  
-h: Print this help."
	exit 0
	;;
    k)
	reduction=$OPTARG || exit 1
	;;
    d)
	nodot=false
	;;
    s)
	minSize=$OPTARG || exit 1
	;;
    r)
	remove=true
	;;
    a)
	i=$OPTARG || exit 1	# make sure it's an integer
	# reduce i by 1 because -xtime +i means > i days
	let i-=1
	amtime="-atime +$i -mtime +$i"
	;;
    p)
	byProduct=true
	typeset -i s_xtime s_size
	# Make the age-size product be in days*KB, to reduce the chance that
	# it will be > max_int
	sort='
while read s_xtime s_size s_fname; do
print -r "$(((curtime-s_xtime)/86400*(s_size/1024))):$s_xtime $s_size $s_fname";
done | sort -rn'
	;;
    x)
	debug=true
	;;
    ?)
	print -r -u2 "Use -h for help."
	exit 1
	;;
    esac
done

# remove args that were options
let OPTIND=OPTIND-1
shift $OPTIND

if [ $# -lt 1 ]; then
    print -r -u2 -- "$Usage
Use -h for help."
    exit 1
fi

$remove && xargs rm -f 2>/dev/null |&

curtime	# sets $curtime
typeset -i size usage quota over total num
# Don't typeset -i xtime because it may be set to sortkey:time
for user; do
    cd ~$user || {
	print -ru2 -- "Could not change working directory to ~$user"
	continue
    }
    if [ reduction -ne -1 ]; then
	over=reduction
    else
	quota -p $user | read u fs usage quota
	if [ quota -le 0 -o usage -le 0 ]; then
	    print -ru2 -- \
"Bad quota for user $user? -- \"$u $fs $usage $quota\""
	    continue
	fi
	over=usage-quota 
    fi
    total=0 num=0
    if [ over -le 0 ]; then
	print -ru2 "User $user is not over quota."
	continue
    fi
    # Skip files with embedded whitespace, for simplicity.
    find . -type f ! -name '*[
 	]*' -size +$((minSize*2)) $amtime -print|stat -r -Lysn|eval $sort |
	while read xtime size filename; do
	    $debug && $byProduct &&
	    print "size-age product of $filename: ${xtime%:*}"
	    $nodot && [[ "$filename" = */.+([!/]) ]] && continue
	    if $remove; then
		$remove && print -p "$PWD/$filename"
		print -r -- "Removing: $filename"
	    else
		print -r -- "$filename"
	    fi
	    let total+=size/1024 num+=1
	    if [ total -ge over ]; then
		print -ru2 -- \
"Reduction quota of ${over} KB reached for user $user after $num file(s)."
		break
	    fi
	done
	if [ total -lt over ]; then
	    print -ru2 -- \
"Reduction quota of ${over} KB not reached for user $user after $num file(s)."
	    [ total -gt 0 ] && print -ru2 "Got only ${total} KB."
	fi
	if [ "$filename" != "" ]; then
	    daysold ${xtime#*:}
	    $byProduct && sd=" (size-age product: ${xtime%:*})"
	    print -ru2 -- "Most recent file is $daysold days old$sd: $filename"
	fi
done