#!/bin/bash
#
# A script to create an apt sources.list file automatically by downloading
# the list of Debian mirrors and choosing the fastest using netselect.
#
# Author: Avery Pennarun <apenwarr@debian.org>
# Enhancements: Filippo Giunchedi <filippo@esaurito.net>
#               Javier Fernandez-Sanguino <jfs@debian.org>
#
# License: public domain.  Please feel free to improve this script.  It
# doesn't really belong in the netselect package, particularly since the
# netselect package doesn't depend on wget and this script does.  If you
# feel like merging this into another package, or creating a new package,
# please do.  Then tell me, so I can remove it from the netselect package.
#
# TO DO:
#   - have some way to pass options (especially -t) to netselect.
#   - test the generated files automatically.  Some mirrors in the list
#       are broken.  We should at least verify that the Packages and Sources
#       files exist and aren't ancient.
#   - maybe generate redundant entries, in case one server is missing files?

# our defaults
# RATIONALE for WANT_SOURCES and WANT_NONFREE environmental variables:
# both variables can be system wide and can be useful to not always specify
# them on the commandline

want_sources=${WANT_SOURCES:-0}
want_nonfree=${WANT_NONFREE:-0}
want_security=1
infile=""
tmpinfile="1"
outfile="sources.list"
distro="stable"
protocol="HTTP"
url="http://www.debian.org/mirror/mirrors_full"
arch=$(/usr/bin/dpkg-architecture -qDEB_HOST_ARCH)

options="-o a:si:o:nfhu -l arch:,sources,infile:,outfile:,nonfree,ftp,help"

trap '[ "$tmpinfile" = "1" ] && rm -f "$infile"' TERM INT EXIT

# misc functions
log()
{
	echo "$@" >&2
}

run_netselect()
{
	SEARCH="$1"
	PROTO="$2"
        hosts=$(cat "$infile" \
		| perl -n -e '
			$/="<br><br>";
			while(<>){
				next if $_ !~ /Site:/;
				if( ( /Includes architectures:.+'"$arch"'.+/i ||
						$_ !~ /Includes architectures:/ ) &&
						m@<br>'"$SEARCH"':.*<a.*?href="('"$PROTO"'://.*?)">@i
					){
					print("$1\n");
				}
			}')
        [ -z "$hosts" ] && return 200
	out=`netselect -v -s 1 $hosts`
	rv=$?
	echo $out | awk '{print $2}'
	return $rv
}

netselect_generic_error()
{
	log "netselect was unable to find a mirror, this probably means that"
	log "you are behind a firewall and it is blocking traceroute."
}

netselect_permission_error()
{
	log "netselect was unable to operate successfully, please check the errors,"
	log "most likely you don't have enough permission."
}

netselect_parse_error()
{
        log "netselect-apt was unable to obtain a list of valid hosts from"
        if [ "$tmpinfile" = "0" ] ;then
            log "the file provided ($infile)."
        else
            log "the file downloaded from the url '$url'."
        fi
        log "This might happen because of any of the following reasons: "
        log "   - there was an error in the file "
        log "   - the file is not in the format netselect-apt expected "
        log "   - there is a bug in netselect-apt "
        log "Please manually check the file. If you believe its contents are correct, file "
        log "a bug (hint: use 'reportbug') against netselect-apt and provide the file as "
        log "well as the output generated by the program (hint: use 'script')."
}

usage()
{
	log "Usage: netselect-apt [OPTIONS] [ debian_release ]"
	log "       debian_release is one of stable, testing, unstable, experimental"
	log "       or a codename etch, lenny, squeeze, sid"
	log "Options:"
	log "   -a, --arch ARCH        Use mirrors containing arch ($arch)"
	log "   -s, --sources          Include deb-src lines in generated file (no)"
	log "   -i, --infile INFILE    Use INFILE as the input file (temp file)"
	log "   -o, --outfile OUTFILE  Use OUTFILE as the output file (sources.list)"
	log "   -n, --nonfree          Use also non-free packages in OUTFILE (no)"
	log "   -f, --ftp              Use FTP as the protocol for OUTFILE (HTTP)"
}


# commandline parsing
temp=$(getopt $options -n 'netselect-apt' -- "$@")
if [ $? != 0 ]; then echo "Terminating..." >&2; exit 2; fi
eval set -- "$temp"
while true; do
	case "$1" in
		-a|--arch) arch=$2; shift 2 ;;
		-s|--sources) want_sources=1; shift ;;
		-i|--infile) infile="$2"; tmpinfile="0"; shift 2 ;;
		-o|--outfile) outfile="$2"; shift 2 ;;
		-n|--nonfree) want_nonfree=1; shift ;;
		-f|--ftp) protocol="FTP"; shift ;;
		-h|--help) usage; exit 0;;
		--) shift; break;;
		*) echo "Internal Error!"; echo "args: $@"; exit 1;;
	esac
done

# distro is a non-option for backward compatibility
case "$1" in
	# don't forget to update the usage with new codenames
	stable|testing|unstable|experimental|etch|lenny|squeeze|sid) distro="$1" ;;
	'') ;;
	*) log "Invalid distribution: $1"; exit 1 ;;
esac

if [ "$distro" != "stable" ]; then
	want_security=0
fi

# netselect starting
log "Using distribution $distro."

if [ "$tmpinfile" = "0" -a ! -f "$infile" -a ! -x /usr/bin/wget ]; then
	log "Sorry, this script requires the 'wget' package in order to run."
	log "You can also download the mirrors list yourself and pass it"
	log "with -i option, consult the manpage for further details:"
	log "        $url"
	exit 2
fi

if [ ! -f "$infile" ]; then
	if [ "$tmpinfile" = "1" ]; then
		infile=$(mktemp -t netselect-apt.XXXXXX) ||
			{ log "unable to create tempfile"; exit 2; }
	fi

	log "Retrieving the list of mirrors from www.debian.org..."
	log

	if ! /usr/bin/wget -O "$infile" "$url"; then
		log "$0: wget failed.  Please try to correct the problem"
		log "by reading the wget messages printed above."
		exit 2
	fi
else
	log "$infile has been found."
	log "I'll use that, rather than downloading it again."
	log
fi

log "Choosing a main Debian mirror using netselect."
main=$(run_netselect "Packages over $protocol" $protocol)
netselect_rv=$?
if [ $netselect_rv -eq 0 ]; then
	log "The fastest server seems to be:"
	log "        $main"
	log
elif [ $netselect_rv -eq 6 ]; then
	netselect_permission_error
	exit 2
elif [ $netselect_rv -eq 200 ]; then
        netselect_parse_error
        exit 2
else
	netselect_generic_error
	exit 2
fi

log "Writing $outfile."

if [ -f "$outfile" ]; then
	destfile="$outfile.$(date +%s)"
	log "$outfile exists, moving to $destfile"
	mv $outfile $destfile
fi

sections="main contrib"
if [ "$want_nonfree" -eq 1 ]; then sections="$sections non-free"; fi

(
	echo "# Debian packages for $distro"
	echo "deb $main $distro $sections"
	echo "# Uncomment the deb-src line if you want 'apt-get source'"
	echo "# to work with most packages."
	if [ "$want_sources" -eq 0 ]; then
		echo -n "# "
	fi
	echo "deb-src $main $distro $sections"

	echo
	echo "# Security updates for stable"
	if [ "$want_security" -eq 0 ]; then
		echo -n "# "
	fi
	echo "deb http://security.debian.org/ stable/updates $sections"

) > $outfile

echo "Done."