#!/bin/ksh
# @(#) cleannews.ksh 1.1 96/01/20
# 94/11/19 john h. dubois iii (john@armory.com)
# 94/11/21 Remove dirs.  Print running group count.
# 94/12/12 Let group-list be given as 2nd arg.  Allow spool dirs in group-list.
# 96/01/20 Create temp file more carefully

Active=/usr/lib/news/active

Help="Usage: cleannews save-file [group-list-file]
save-file is a list of groups to save.  If group-list-file is given, it should
contain a list of groups to consider for article removal.  Only the first word
of each line is used.  The word can be a newsgroup name or a spool directory
(relative to /usr/spool/news).  If a group-list-file is not given, the file
$Active is used.  All articles in all groups in the
group-list-file that are not in save-file are removed."

case $# in
1)
    if [ "$1" = -h ]; then
	print "$Help"
	exit 0
    fi
    ;;
2)
    Active=$2
    ;;
*)
    print -u2 "Wrong number of arguments.  Use -h for help."
    exit 1
    ;;
esac

# @(#) mktempfile 1.0 96/05/22
# 96/01/20 jhdiii

# Usage: mkfiles name ...
# Creates the named files with some attempt at security.
# This will be more reliable if user do not have chown authority.
# Any file that contains no / characters is created in the user's tempdir.
# If TMP was not set, it is set by this function.
# Returns 0 on success; prints a diagnostic message & returns 1 on failure.
function mkfiles {
    typeset file files
    typeset -i i=0

    : ${TMP:=$TMPDIR}
    : ${TMP:=/tmp}
    for file; do
	[[ "$file" != */* ]] && file="$TMP/$file"
	files[i]=$file
	let i+=1
    done
    set -- "${files[@]}"

    rm -f "$@" || {
	# hopefully rm will have printed a more specific message.
	print -u2 "Could not remove pre-existing files."
	return 1
    }
    for file; do
	# Use >> to avoid 0'ing the file in case a symlink was just made from
	# the filename to something we don't want to truncate
	>> "$file" || {
	    print -u2 "Could not create temp file $file"
	    return 1
	}
	# These are very suspicious
	[ -s "$file" ] && {
	    print -u2 "New tempfile is not empty!!!"
	    return 1
	}
	[ -O "$file" ] || {
	    print -u2 "Do not own $file!!!"
	    return 1
	}
    done
    # Could do some more stuff here, but anyone concerned with security should
    # have chown authorization off for most users.
    return 0
}

# Usage: mktempfile name
# Creates a tempfile name tempdir/#name$$, and sets the global mktempfile_ret
# to that name.  tempdir is a temp directory in $TMP, $TMPDIR, or /tmp, and $$
# is the PID of the current process.
# name should be 8 characters or less, so that the resulting filename will
# not be more than 14 characters long (a limit on some machines).
# Returns 0 on success, prints a diagnostic message & returns 1 on failure.
function mktempfile {
    typeset file="#$1$$"
    mkfiles "$file"
    mktempfile_ret="$TMP/#$1$$"
}

name=${0##/}
ClearToEOL=$(tput el)
newsdir=/usr/spool/news
xargsCmd="env - /usr/local/bin/gxargs -n 2000 -P 2 /bin/rm --"
typeset -Z4 groupNum=0
mktempfile clnnews. || {
    print -u2 "$name: Exiting."
    exit 1
}
dirFile=$mktempfile_ret

cd $newsdir

# start coprocess to keep track of how many files in the undesired groups
# were not cross-posted (had only one link).
# make stat emit filename after link count just so we can detect /dev/null
# when it comes through.
stat -rnfln | awk '
$2 == "/dev/null" {
    printf "%d of %d files had only one link.\n",count,NR | "cat 1>&2"
    exit
}
$1 == 1 {
    count++
}
' |&

# Find all groups in active file that are not in save-groups file and print
# their spool directories
awk -v "active=$Active" '
{
    SaveGroups[$1]
}
END {
    if (NR < 10) {
	print "Only " NR " groups in save file?!" | "cat 1>&2"
	exit 1
    }
    printf "%d groups to be saved.\n",NR | "cat 1>&2"
    close("cat 1>&2")	# flush output
    while ((ret = (getline < active)) == 1) {
	group = $1
	gsub("/",".",group)	# convert paths to newsgroup names
	gsub(/^\.+/,"",group)	# in case paths begin with e.g. ./ or .
	if (!(group in SaveGroups)) {
	    gsub(/\./,"/",group)
	    print group
	}
    }
    if (ret != 0)
	print "Error reading group-list-file" | "cat 1>&2"
}
' "$1" | while read dir; do
    let groupNum+=1
    # Echoing filenames w/full group path will require many more 
    if [ -d $newsdir/$dir ] && cd $newsdir/$dir; then
	print $dir >> $dirFile
	print -u2 -n "${ClearToEOL}$groupNum Checking $dir\r"
	# Pattern ?(+([0-9])|.thread) does not work
	set -- +([0-9])
	[[ $1 = \+* ]] && set --
	[ -f .thread ] && set -- "$@" .thread
	if [ $# -gt 0 ]; then
	    print -u2 "Removing $# file(s) from $dir"
	    # If removing these files with full group path would require
	    # more than two invokations of rm, do a special gxargs/rm
	    # from the group dir for it.
	    if [ $(( (${#dir}+1)*$# )) -gt 10000 ]; then
		print -- "$*" | $xargsCmd
		for file; do
		    print -p $dir/$file
		done
	    else
		for file; do
		    print -- $dir/$file
		    print -p $dir/$file
		done
	    fi
	fi
    fi
    # jobs > /dev/null 2>&1	# work around accumulating jobs ksh bug
done | $xargsCmd
# use gxargs because it takes arbitrarily long input lines and uses max command
# line space.  -P 2 tells it to start up to 2 processes at a time.
# Get rid of environment to leave more room for cmd line args.

# Doesn't seem to be any way of closing pipe to coprocess.
# So, use /dev/null as the eof marker.
print -p /dev/null

print -u2 "\nRemoving directories..."
cd $newsdir
# Reverse sort to ensure that lowest dirs come first
sort -r $dirFile |
env - /usr/local/bin/gxargs -n 2000 -P 2 /bin/rmdir -- 2>/dev/null