#!/bin/ksh
# @(#) unarch.ksh 2.8 97/07/27
# 92/10/08 john h. dubois iii (john@armory.com)
# 92/10/31 make it actually work if archive isn't in current dir!
# 93/11/10 Added pack and gzip archive support
# 94/10/22 Added -h option.
# 95/01/29 Work with cpio too (generalized from untar)
# 95/08/19 Understand packed extensions (.tgz, etc.).  More verbose messages.
#          If everything in the archive is under one dir, make that be the
#          main dir.  Added f option.
# 95/12/20 v5 mvdir does not recognize --; work around it.
# 96/03/29 Added arc.
# 96/04/19 Catch files with unknown archive type.  Added zip.
# 96/04/27 Added t option.
# 96/04/30 Added c1 options.
# 96/09/13 Added e option.
# 96/11/01 Added hzip and bzip support.
# 96/11/27 Added d option.
# 96/12/27 Added arj support.
# 97/01/17 Added shar support.
# 97/03/11 Added ar support.
# 97/07/27 Do not give -c to uncompressors

name=${0##*/}
Usage=\
"Usage: $name [-cfht1] [-e<ext>] [-d<directory>]
       archive[.[tar|a|cpio|shar][.[Z|[ghb]z]]|t[ghb]z|tz|arc|zip|arj] ..."
Force=false
Tell=false
NoArchDir=false
One=false
pExt=
fixedDir=
typeset -i debug=0
function debug {
    if [ debug -ge $1 ]; then
	shift
	print -ru2 -- "$*"
    fi
}

while getopts :cfht1x:e:d: opt; do
    case $opt in
    h)
	echo \
"$name: extract archives into directories, uncompressing if required.
$Usage
If an archive name given does not end in a recognized extension, it is searched
for with each possible extensions in turn.  The real filename must end with one
of them.  A directory with the name of the archive is created in the current
directory (not necessarily the directory that the archive is in), and the
contents of the archive are extracted into it.  Absolute pathnames in tar and
cpio archives are suppressed.
Options:
-c: Extract files relative to current directory (don't create archive dirs).
-d<directory>: Extract files into <directory> instead of a directory derived
    from the archive name.  Only one archive name should be given.
-f: Extract an archive even if the directory it would be extracted into already
    exists.
-e<ext>: For the purpose of determining archive type, treat all filenames as
    though <ext> was tacked onto them.  For the purpose of reading the files,
    the file names are left alone, so they must be the real names of the input
    files.  This allows processing of files that do not include their
    compression or compression & archive type extensions.
    Examples: \"$name -etar.Z foo\" and \"$name -eZ foo.tar\" will both treat
    the file foo as though it was foo.tar.Z.  
-h: Print this help.
-t: Tell what is in an archive, without writing any files.
    This does not work for sharfiles.
-1: Treat only the first non-option argument as an archive name.  Any arguments
    given after it are the names of files to extract; if any are given, only
    those files are extracted.  This does not work for sharfiles."
	exit 0
	;;
    c)
	NoArchDir=true
	;;
    d)
	fixedDir=$OPTARG
	;;
    e)
	pExt=.$OPTARG
	;;
    1)
	One=true
	;;
    f)
	Force=true
	;;
    t)
	Tell=true
	NoArchDir=true
	;;
    x)
	if [[ "$OPTARG" != +([0-9]) ]]; then
	    print -ru2 -- "$name: Option -x must be given an integer argument."
	    exit 1
	fi
	debug=$OPTARG || exit 1
	;;
    +?)
	print -u2 "$name: options should not be preceded by a '+'."
	exit 1
	;;
    :) 
	print -r -u2 -- \
	"$name: Option '$OPTARG' requires a value.  Use -h for help."
	exit 1
	;;
    ?)
	print -u2 "$name: $OPTARG: bad option.  Use -h for help."
	exit 1
	;;
    esac
done

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

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

OWD=$PWD

# @(#) glob.ksh 1.0 95/07/29
# 95/07/29 john h. dubois iii
# Usage: Glob <Pattern> ...
# Glob() sets the global Glob_ret[] to contain the results of globbing on
# the given filename patterns.  The difference between this and using the
# patterns directly is that Glob_ret[] will not contain any entries for
# patterns that did not match any filenames.
# Note that only 1024 filenames can be stored in Glob_ret[].  If the
# globbing produces more than 1024 matches, Glob() will return 2.
# Unfortunately, it isn't possible to set the global positional parameters
# (which can have more than 1024 elements) from within a function.
# Example: Glob ".[!.]*" "*"
# This will set Glob_ret[] to all filenames in the current directory except
# .  and ..
# Return value: 1 if any of the patterns did not match; 2 if more than 1024
# matched resulted; else 0.
function Glob {
    set +o noglob	# Turn on globbing; local to this function
    # In order for a pattern contained in a shell var to be globbed, the shell
    # var must not be quoted.  But, if the pattern contains any whitespace, the
    # pattern will be split into words on the whitespace before globbing.
    # To get around this, we set IFS to the null string so no word splitting
    # will occur.  But, this causes ${array[*]} to be expanded as a single
    # word, so we have to glob each element of the array individually.
    typeset Patterns OIFS=$IFS IFS
    typeset -i NumPats=$# NoMatch=0 PatNum=0
    set -A Patterns -- "$@"
    unset Glob_ret[*]
    while [ PatNum -lt NumPats ]; do
	IFS=
	set -- ${Patterns[PatNum]}
	IFS=$OIFS
	if [ $# = 1 -a "$1" = "${Patterns[PatNum]}" -a ! -a "$1" ]; then
	    NoMatch=1
	else
	    set -A Glob_ret -- "${Glob_ret[@]}" "$@" || return 2
	fi
	let PatNum+=1
    done
    return $NoMatch
}

if $One; then
    # Move everything except 1st arg to file list
    Arch=$1
    shift
    set -A FileList -- "$@"
    set -- "$Arch"
else
    unset FileList
fi

for file in "$@"; do
    debug 1 "Processing: $file"
    cd -- "$OWD"

    tName="$file$pExt"
    ## Determine what the real archive full & base names are
    if [[ "$tName" = \
    *.@(@(tar|a|cpio|shar)?(.@(z|Z|[ghb]z))|[ct]@(Z|z|[ghb]z)|ar[cj]|zip) ]]
    then
	# If file was given with a known extension, the archive name
	# is the filename without the extension.
	ArchName=\
${tName%%.@(@(tar|a|cpio|shar)?(.@(z|Z|[ghb]z))|[ct]@(Z|z|[ghb]z)|ar[cj]|zip)}
	debug 1 "$tName has known extension; archive name = $ArchName"
    elif [ -n "$pExt" ]; then
	print -u2 "$name: Unknown archive type: $tName"
	continue
    else
        ArchName=$file
	file=
	# The filename did not have a known extension; try attaching all
	# known extensions and see if the archive exists.
	for ext in \
	{tar,a,cpio,shar}{,.Z,.z,.{g,h,b}z} {t,c}{z,Z,{g,h,b}z} arc arj zip; do
	    debug 3 "Searching for file as: $ArchName.$ext"
	    if [ -f "$ArchName.$ext" ]; then
		file="$ArchName.$ext"
		break
	    fi
	done
	if [ -z "$file" ]; then
	    print -u2 -- \
"$name: Unknown archive type (not found with any known extension):
$ArchName"
	    continue
	fi
	debug 1 "No known extension in $ArchName; found archive file as $file"
	tName=$file
    fi
    if [ ! -r "$file" ]; then
	print -u2 "$name: Cannot read archive file: $file"
	continue
    fi

    [[ "$file" != /* ]] && file="$OWD/$file"
    if $Tell; then
	print "Contents of archive $file:"
    elif $NoArchDir; then
	print "Extracting archive $file into current directory ($PWD)..."
    else
	if [ -n "$fixedDir" ]; then
	    DirName=$fixedDir
	else
	    DirName=${ArchName##*/}
	fi
	# If the directory doesn't already exist in the local dir, make it.
	if [ -a "$DirName" ]; then
	    if [ ! -d "$DirName" ]; then
		print -u2 \
"$name: $DirName already exists in the current directory 
and is not a directory.  It must be removed in order to
extract the archive \"$file\"."
		continue
	    fi
	    $Force || {
		print -u2 \
"$name: The directory $DirName already exists.  It must be removed,
or the -f option must be used to force extraction into it."
		continue
	    }
	else
	    mkdir "$DirName" || {
		print -u2 "$name: Could not make archive directory: $DirName"
		continue
	    }
	    debug 1 "Created directory $DirName"
	fi
	cd -- "$DirName"
	print "Extracting archive $file into directory $DirName"
    fi
    case "$tName" in
    *.?([ct])Z)
	uncomp=zcat
	compression="compressed "
	;;
    *.?([ct])z)
	uncomp=pcat
	compression="packed "
	;;
    *.?([ct])[ghb]z)
	l=${tName##*.}
	l=${l#[ct]}
	l=${l%z}
	# Do not give -c to uncompressor, because input is always redirected
	# into it and some do not like being given -c if not given a filename
	uncomp="${l}zip -d"
	compression="${l}zipped "
	;;
    *) uncomp=
	compression=
	;;
    esac
    debug 1 "Uncompressor=$uncomp"
    case "$tName" in
    *.tar?(.@(z|Z|[ghb]z))|*.t@(Z|z|[ghb]z))
	$Tell && x=t || x=x
	# Redirect input to uncompress routines because file might not end
	# with the extension they expect
	[ -z "$uncomp" ] && tar nvAf$x "$file" "${FileList[@]}" ||
	$uncomp < "$file" | tar vAf$x - "${FileList[@]}"
	arch=tar
	;;
    *.a?(.@(z|Z|[ghb]z))|*.t@(Z|z|[ghb]z))
	$Tell && x=t || x=x
	# Redirect input to uncompress routines because file might not end
	# with the extension they expect
	[ -z "$uncomp" ] && ar v$x "$file" "${FileList[@]}" ||
	$uncomp < "$file" | ar v$x /dev/stdin "${FileList[@]}"
	arch=ar
	;;
    *.cpio?(.@(z|Z|[ghb]z))|*.c@(Z|z|[ghb]z))
	$Tell && x=t || x=
	[ -z "$uncomp" ] && cpio -${x}idmvAI "$file" "${FileList[@]}" ||
	$uncomp < "$file" | cpio -${x}idmvA "${FileList[@]}"
	arch=cpio
	;;
    *.shar?(.@(z|Z|[ghb]z))|*.c@(Z|z|[ghb]z))
	if $One; then
	    print -u2 -- "$name: Cannot use -1 with a sharfile.  Skipping."
	    continue
	fi
	if $Tell; then
	    print -u2 -- "$name: Cannot use -t with a sharfile.  Skipping."
	    continue
	fi
	[ -z "$uncomp" ] && unshar -f "$file" "${FileList[@]}" ||
	$uncomp < "$file" | unshar -f "${FileList[@]}"
	arch=shell
	;;
    *.arj)
	$Tell && x=l || x=e
	unarj $x "$file" "${FileList[@]}"
	arch=arj
	;;
    *.arc)
	$Tell && x=v || x=e
	arc $x "$file" "${FileList[@]}"
	arch=arc
	;;
    *.zip)
	$Tell && x=v || x=e
	unzip -$x "$file" "${FileList[@]}"
	arch=zip
	;;
    *)
	# Shouldn't be able to get here!
	print -u2 "Fatal error: unknown extension in $tName"
	exit 1
	;;
    esac
    $Tell && continue
    print \
"Done extracting $compression$arch archive $file
into directory: $DirName"
    $NoArchDir && continue
    Glob ".[!.]*" "*"
    if [ ${#Glob_ret[*]} = 1 -a -d "${Glob_ret[0]}" ]; then
	root=${Glob_ret[0]}
	print \
"Archive root contains one directory, \"$root\".
Moving $DirName/$root to: $DirName"
	# mvdir does not recognize -- to end option list, so use ./
	cd .. &&
	mvdir "./$DirName" "$DirName.$$" &&
	mvdir "./$DirName.$$/$root" "$DirName" &&
	rmdir -- "$DirName.$$" &&
	print "Directory move completed successfully." ||
	print "Directory move failed."
    fi
done