#!/bin/ksh
# @(#) bigfiles.ksh 1.1 97/05/21
# 93/04/13 john h. dubois iii (john@armory.com)
# 93/06/15 Added -d and -v options, fixed sorting so it works with -n,
#          changed xargs command to avoid running l -d w/o args
# 93/08/02 Added -b, -s, and -i options.
# 94/06/26 Added jpg and mpg to compressed extentions
# 96/01/05 5.0 port: different block size for find
# 97/03/31 Do not list executable files by default.  Added X option.
# 97/04/05 Use gfind if available to make ext comparisons case insensitive
# 97/05/21 Added x option.  Increased defK to 10.

alias istrue="test 0 -ne"
alias isfalse="test 0 -eq"

Name=${0##*/}
comp_exts="Z,z,gz,hz,bz,arc,zip,zoo,gif,jpg,mpg,avi,arj,tz,tgz,thz,tbz,"
bin_exts="exe,lib,com,ovl,obj,bin,sys,o,a,wav"
typeset -i days=0 defK=10 verbose=0
Usage=\
"Usage: $Name [-abhnsv] [-i<file>] [-k<size>] [-eE<exts>] [-d<days>] [dir ..]"
cut_cmd="cut -c 34-"
sort_cmd="sort +4 -5 -n -r"
grep_cmd=cat
xmode="! -perm -100"
Debug=false

while getopts :ahnvsbk:e:E:d:i:x opt; do
    case $opt in
    a)
	comp_exts=
	;;
    h) echo \
"$Name: find large uncompressed files and list them sorted by size.
$Usage
The named directories and all subdirectories of them are searched for files
larger than ${defK}K.  Any such ordinary non-executable files that do not have
an extension indicating that they are compressed are listed.  The compression
extensions are:" {$comp_exts} "
If the 'gfind' utility is available, extensions will be matched regardless of
their case.  If no directory names are given, the current directory is used.
Options:
-h: Print this help.
-n: Print filenames only.  By default the size, date, and name are printed.
-e<ext[,ext...]>: Do not list files with any of the given extensions.
   The given list replaces the internal list.  Extensions should be
   separated by commas and should not be preceded by a '.'  Example: -ez,Z,gz
-E<ext[,ext...]>: Like -e, except that the given extensions are added to
   the internal list instead of replacing it.
-b: Add the following binary file extensions to the default list of ignored
   extensions:" {$bin_exts} "
-i<file>: Ignore the files whose names are given in <file>, one per line.
   The filenames must be given with the same path as $Name would print.
-a: List all big files regardless of extension.
-X: Check all regular files.  Normally files executable by the invoking user
    are skipped.
-d<days>: List only files that have not been accessed in at least <days> days.
   This changes the default for <size> to 0; that is, all files at least <days>
   old will be listed, regardless of size.  The size can still be set with -k.
-k<size>: List only files larger than <size> kilobytes.
-s: Sort by filename instead of size.
-v: Print the command issued to find files."
       exit 0;;
    X)
	xmode=;;
    x)
	Debug=true;;
    n)
	cut_cmd="awk '{ print \$9; }'"
	;;
    k) 
	typeset -i K=$OPTARG || exit 1
	;;
    d) 
	days=$OPTARG || exit 1
	defK=0
	;;
    v)
	verbose=1
	;;
    e) 
	comp_exts=$OPTARG
	;;
    E) 
	comp_exts="$comp_exts,$OPTARG"
	;;
    b)
	comp_exts="$comp_exts,$bin_exts"
	;;
    s)
	sort_cmd="sort +8"
	;;
    i)
	grep_cmd="fgrep -xvf $OPTARG"
	;;
    +?) 
	echo "$Name: options should not be preceded by a '+'."
	exit 1
	;;
    :) 
	print -u2 "$Name: Option '$OPTARG' requires a value.  Use -h for help."
	exit 1
	;;
    ?) 
	echo "$Name: $OPTARG: bad option.  Use -h for help."
	exit 1
	;;
    esac
done
 
# Like the ! that later ksh have
function not { "$@"; return $((!$?)); }

# Sets _OSRelease to the entire OS release (e.g. 3.2v5.0.0}
# Sets _OSVersion to the version part (e.g. 5.0.0)
# Returns the major part of the version (e.g. 5)
# If an OSVersion value is given as an arg, returns 0 if the system OS version
# is lexicographically less than it; 1 if they are equal; 2 if the system OS
# version is greater.
function OSVersion {
    typeset arg=$1
    if [ -z "$_OSRelease" ]; then
	# Name of release field is different in different langs
	set -- $(LANG=english_us.ascii uname -X)
	while [ "$1" != Release -a $# -ge 3 ]; do
	    shift 3
	done
	[ $# -lt 3 ] && return 1
	_OSRelease=$3
	_OSVersion=${3##*v}
    fi
    [ -z "$arg" ] && return ${_OSVersion%%.*}
    [[ "$_OSVersion" > "$arg" ]] && return 2
    [[ "$_OSVersion" < "$arg" ]] && return 0
    return 1
}

# Increase the liklihood that we can find gfind
[[ "$0" = */* ]] && PATH=$PATH:${0%/*} || PATH=$PATH:.

if type gfind > /dev/null; then
    find=gfind
    nam=iname
    $Debug && print -ru2 -- "Found gfind"
else
    find=find
    nam=name
fi

[ -z "$K" ] && typeset -i K=$defK
# In 5.0.0 the units for find changed to 1/2 K blocks
if not OSVersion 5.0.0 || [ $find = gfind ]; then
    K=$((K*2))
fi

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

[ $# -eq 0 ] && set -- .

set +o noglob

ExtPred=
if [ -n "$comp_exts" ]; then
    ExtPred='( '`eval echo "-$nam \*."{$comp_exts}" -o"`' -name "" )'
fi

set -o noglob

# Use val & +val instead of +(val-1) so that they will work if val is 0,
# w/o having to remove expression for special case 0.
istrue verbose && set -x

for dir
do
    $find "$dir" -size +$K -type f $xmode \
    \( -atime $days -o -atime +$days \)\
    \( -mtime $days -o -mtime +$days \) ! $ExtPred -print |
    $grep_cmd | 
    ( xargs sh -c '[ $# -gt 0 ] && l -d "$@"' ) |
    $sort_cmd |
    eval $cut_cmd
done