opensde-nopast/bin/sde-download

#!/bin/bash
# --- SDE-COPYRIGHT-NOTE-BEGIN ---
# This copyright note is auto-generated by ./scripts/Create-CopyPatch.
# 
# Filename: bin/sde-download
# Copyright (C) 2006 - 2007 The OpenSDE Project
# Copyright (C) 2004 - 2006 The T2 SDE Project
# Copyright (C) 1998 - 2003 Clifford Wolf
# 
# More information can be found in the files COPYING and README.
# 
# This program is free software; you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation; version 2 of the License. A copy of the
# GNU General Public License can be found in the file COPYING.
# --- SDE-COPYRIGHT-NOTE-END ---

#Description: Download sources
#Alias: get

set -e

[ -n "$SDEROOT" ] ||
	export SDEROOT=$( cd "${0%/*}/.."; pwd -P )

. $SDEROOT/lib/libsde.in
. $SDEROOT/lib/functions.in

download_usage() {
cat <<EOT
Usage:
	sde download <options> [ Package(s) ]
	sde download <options> [ Desc file(s) ]	
	sde download <options> -repository Repositories
	sde download <options> [ -all | -required ]

	Options is an alias for:
	[ -cfg <config> ] [ -nock ] [ -alt-dir <AlternativeDirectory> ]
	[ -mirror <URL> | -check ]  [ -try-questionable ]  [ -notimeout ]
	[ -longtimeout ]  [ -curl-opt <curl-option>[:<curl-option>[:..]] ]
	[ -copy ] [ -move ]

	On default, this script auto-detects the best OpenSDE mirror.
	
	Mirrors can also be a local directories in the form of 'file:///<dir>'.

	sde download [ -list | -list-missing | -list-cksums ]

	See '-mirror none' output for help on bypassing the official mirrors.
EOT
}

umask 022

cd "$SDEROOT"

# Handle options passed on the command line
#
mkdir -p tmp/ download/

# Load system wide configuration for this tool
#
config=default mirror= altdir= proxy= proxyauth=

if [ -s "$SDESETTINGS" ]; then
	eval $( $SDEROOT/bin/sde-config-ini -F "$SDESETTINGS" download )
	eval $( $SDEROOT/bin/sde-config-ini -F "$SDESETTINGS" download-$sdever )
fi

this_is_the_2nd_run=0
checkonly=0 tryques=0 nocheck=0 notimeout=0
options='-this_is_the_2nd_run '
curl_options='-A opensde-downloader --disable-epsv --location -f'
altcopy=link ; verbose=1
downloaderror=0

#
while [ $# -gt 0 ]; do
	case "$1" in

	    -this_is_the_2nd_run)
		this_is_the_2nd_run=1 
		;;

	    -cfg)
		options="$options -cfg $2"
		config="$2" ; shift ;;

	    -q)
		options="$options -q"
		verbose=0 ;;
		
	    -nock)
		# -nock skips checksum checking (don't use lightly)
		options="$options -nock"
		nocheck=1 ; shift ;;

	    -mirror)
		# -mirror uses a mirror for finding source files
		if [ "$2" = none ]; then
			echo
			echo "The option '-mirror none' is not supported anymore!"
			echo
			echo "You may edit \$HOME/.sde/settings if you really"
			echo "want to use the original download resources. However, this"
			echo "is not supported and if such a download fails, this is not"
			echo "a bug in the OpenSDE and doesn't neccessarily needs fixing."
			echo
			exit 1;
		elif [ "$2" = auto ]; then
			mirror=
		else
			options="$options -mirror $2"
			mirror="$2"
			$SDEROOT/bin/sde-config-ini -F "$SDESETTINGS" "download-$sdever.mirror=$mirror"
		fi
		shift ;;

	    -check)
		# -check just validates the file using the checksum
		options="$options -check"
		checkonly=1 ;;

	    -notimeout)
		# don't add timeout curl options
		options="$options -notimeout"
		notimeout=2 ;;

	    -longtimeout)
		# don't add timeout curl options
		options="$options -longtimeout"
		notimeout=1 ;;

	    -curl-opt)
		# additional curl options
		options="$options -curl-opt $2"
		curl_options="$curl_options `echo $2 | tr : ' '`"
		shift ;;

	    -alt-dir)
		# check for an alternative directory where to search for
		# package source tarballs
		altdir=$( cd $2; pwd -P )
		options="$options -alt-dir $2"
		shift ;;

	    -try-questionable)
		# also try to download questionable URLs
		options="$options -try-questionable"
		tryques=1 ;;
	
	    -move) altcopy=move ;;
	    -copy) altcopy=copy ;;

	    *)	break	;;
	esac
	shift
done

if [ $notimeout -eq 0 ] ; then
	curl_options="$curl_options -y 10 -Y 10 --connect-timeout 60"
fi
if [ $notimeout -eq 1 ] ; then
	curl_options="$curl_options -y 60 -Y 1  --connect-timeout 300"
fi

# proxy (server[:port])
if [ -n "$proxy" ]; then
	curl_options="$curl_options --proxy $proxy"

	# proxy needs auth (username[:password])
	[ -z "$proxyauth" ] || curl_options="$curl_options --proxy-user $proxyauth"

	# only show once
	[ $this_is_the_2nd_run = 1 ] || echo_info "Using <$proxy> as ${proxyauth:+authenticated }http proxy."
fi

#Disable checking for certificates on https downloads 
curl_options="$curl_options -k"

# Autodetect best Mirror and safe url in $mirror
#
detect_mirror() {
	if [ "$mirror" = "none" ] ; then
		echo_info "Using original download locations only."
		return
	elif [ "$mirror" = "broken" ]; then
		echo_warning "Previous detection of the mirror failed, trying again."
	elif [ -n "$mirror" ]; then
		echo_info "Using mirror <$mirror>."
		return
	fi

	echo_warning "Auto-detecting best mirror ..."

	echo_info "Downloading mirror-list from opensde.net."
	curl -s -S $curl_options -o tmp/Download-Mirror-List \
		"http://opensde.net/opensde-download-mirrors/$sdever"

	if [ -r tmp/Download-Mirror-List ]; then
		bash lib/sde-download/mirror-test.sh < tmp/Download-Mirror-List
	fi 2>&1 | echo_info

	# read new mirror info
	mirror=
	eval $( $SDEROOT/bin/sde-config-ini -F "$SDESETTINGS" download-$sdever )

	if [ -z "$mirror" ]; then
		echo_error "Mirror detection loop hit a bug!"
	elif [ "$mirror" == "broken" ]; then
		echo_warning "No Mirror Found!"
	else
		echo_info "Using mirror <$mirror>."
	fi
}

# download_file local-filename download-location cksum repo pkg
#
# This function decides if download directly or from a mirror,
# validates checksum, etc.
# Calls download_file_now to do the actual download.
#
download_file() {

	# Init
	#
	local gzfile="$1" location="$2" cksum="$3" repo="$4" pkg="$5"
	# Make src directory for creating tar balls
	mkdir -p tmp/
	# Tarball file name:
	bzfile="`bz2filename "$gzfile"`"
	# Remove optional '-' prefix from $location
	[ "${location:0:1}" == '-' ] && location="${location:1}"
	# Lock file name:
	lkfile="tmp/down.lockfile.`echo $bzfile | tr / -`"

	# Check if it's already there
	#
	[ -s "$bzfile" -a $checkonly != 1 ] && return 0

	# Make locking
	#
	if [ -s "$lkfile" ]; then
		echo "Found $lkfile -> skip download."
		return 0
	fi
	trap 'rm -f "$lkfile"' INT
	echo $$ > "$lkfile"

	# Check if we only like to test the cksum(s)
	#
	if [ $checkonly = 1 ] ; then
		gzfile="$bzfile"
		if [ ! -f "$bzfile" ] ; then
			echo "File missing: $bzfile"
			rm -f "$lkfile" ; trap INT ; return 1
		fi
		if [ -z "${cksum##X*}" ] ; then
			echo "No checksum (ignore): $bzfile"
			rm -f "$lkfile" ; trap INT ; return 1
		fi
		if [ "$cksum" -eq 0 ] ; then
			echo "No checksum (missing): $bzfile"
			rm -f "$lkfile" ; trap INT ; return 1
		fi

	elif [ -s "$gzfile" ] ; then

		echo ; echo "Already downloaded $pkg:$gzfile ..."

	else

		echo ; echo "Downloading $pkg:$gzfile ..."

		# Existing *.cksum-err
		#
		if [ -s "$gzfile.cksum-err" ] ; then
			# cksum-err file alread exists:
			echo "ERROR: Found $gzfile.cksum-err."
			echo "ERROR: That means that we downloaded the" \
			     "file already and it had an"
			echo "ERROR: incorrect checksum. Remove the" \
			     "*.cksum-err file to force a"
			echo "ERROR: new download of that file."
			rm -f "$lkfile" ; trap INT ; return 1
		fi

		# Existing *.extck-err
		#
		if [ -s "$gzfile.extck-err" ] ; then
			# extck-err file alread exists:
			echo "ERROR: Found $gzfile.extck-err."
			echo "ERROR: That means that we downloaded the" \
			     "file already and it's content"
			echo "ERROR: did not match it's filename extension." \
			     "Remove the *.extck-err file"
			echo "ERROR: to force a new download of that file."
			rm -f "$lkfile" ; trap INT ; return 1
		fi

		# Questionable URL
		#
		if [ "$location" != "${location#\?}" ] ; then
			if [ "$tryques" = 0 ] ; then
				echo "ERROR: URL is marked as questionable." \
					"Not downloading this file."
				rm -f "$lkfile" ; trap INT ; return 1
			else
				echo "WARNING: URL is marked as questionable." \
					"Downloading it anyways."
				location="${location#\?}"
			fi
		fi

		# Make directory (if required)
		#
		if [ ! -d `dirname "$bzfile"` ] ; then
			mkdir -p `dirname "$bzfile"`
		fi
		
		# Alternative Directory
		#
		if [ "$altdir" ] ; then
		    altfile=$(find $altdir/ -name `basename $bzfile` | head -n 1)
		else
		    altfile=""
		fi

		#FIXME: compatibility, can be removed sooner or later...
		# Check old download dir layout
		if [ -z "$altfile" ]; then
			if [ -f "download/$repo${pkg:+/}$pkg/`basename $bzfile`" ]; then
				altfile="download/$repo${pkg:+/}$pkg/`basename $bzfile`"
			fi
		fi

		if [ "$altfile" ] ; then

			echo "Found `basename $bzfile` as $altfile."
			if [ "$altcopy" = 'link' ]; then
				cp -lv $altfile $bzfile
			elif [ "$altcopy" = 'copy' ]; then
				cp -v $altfile $bzfile
			elif [ "$altcopy" = 'move' ]; then
				mv -v $altfile $bzfile
			fi
			gzfile="$bzfile"

		else

		    # Mirroring
		    #
		    if [ -n "$mirror" -a "$mirror" != "none" -a "$mirror" != "broken" -a -z "${bzfile##download/mirror/*}" ] ; then
			# try to use mirror

			if ! download_file_now "!$mirror/${bzfile#download/mirror/}" $bzfile $bzfile; then
				echo "INFO: download from mirror failed, trying original URL."
				download_file_now "$location" $gzfile $bzfile \
					|| downloaderror=1
	
			else
				gzfile="$bzfile"
			fi
		    else
			# don't want to use mirror
			download_file_now "$location" $gzfile $bzfile \
				downloaderror=1
		    fi
		fi

		if [ ! -s "$gzfile" ]; then
			rm -f "$lkfile" ; trap INT ; return 1
		fi
	fi

	# unsign .gpg file
	if [[ $gzfile = *.gpg ]]; then
		gzfile=${gzfile%.gpg}
		if [ -f $gzfile.gpg ]; then
			echo "unsigning GnuPG file: $gzfile.gpg"
			gpg $gzfile.gpg
		fi
		if [ ! -f $gzfile ]; then
			echo "unsigning failed"
			rm -f "$lkfile" ; trap INT ; return 1
		fi
	fi

	sh ./lib/sde-download/validate.sh "$gzfile" "$bzfile" "$cksum" || downloaderror=1

	# Free Lock and finish
	#
	rm -f "$lkfile" ; trap INT ; return 0
}

# download_file_now location remote_filename local_filename
#
# This function executes the actual download using curl.
#
download_file_now() {
	local location="$1" gzfile="$2" bzfile="$3" curlret=0

	# Create URL
	#
	case "$location" in
		manual://*) url="$location"	;;
		!*)	url="${location#!}"	;;
		*)	url="${location%/*}/${gzfile##*/}"	;;
	esac

	# Download
	#
	case "$url" in
	manual://*)
		# Determine if the file has already been downloaded
		# manually. For this we first look in $HOME then in
		# download/manual. 
		downloadpath=${altdir:-$HOME}
		downloadfile="${gzfile##*/}"
		if [ -e $downloadpath/$downloadfile ]; then
			location="file://$downloadpath/"
		else
			location="http://${url#manual://}"
			# No manual download has taken place yet.
			# So inform the user to do so.
			cat <<-EOT
			The file $downloadfile can not be fetched automatically
			please visit: $location
			and download it manually into $HOME or somewhere else using -alt-dir
			EOT
			return 1;
		fi

		# I am to lazy to do the copy and conversion myself,
		# so I use this function again with a modified
		# download location.
		download_file_now "$location" $gzfile $bzfile
		return "$?"
		;;
	http://*|https://*|ftp://*|file://*)
		if [ -s "$gzfile.incomplete" ] ; then
		    echo "INFO: Trying to resume previous download .."
		    resume="-C -"
		else
		    resume=""
		fi

		[ -s download/translations.sed ] &&
			trfile=download/translations.sh ||
			trfile=etc/download.sed
		trurl="$( echo "$url" | sed -f $trfile )"
		if [ -n "$trurl" -a "$trurl" != "$url" ]; then
			echo "INFO: url translated."
			url="$trurl"
		fi
		unset trurl trfile

		curl -w '\rFinished downloading %{size_download} bytes in %{time_total} seconds (%{speed_download} bytes/sec). \n' --progress-bar $resume $curl_options "$url" -o "$gzfile.incomplete"
		curlret="$?"

		if [ "$resume" ] && \
		   [ $curlret -eq 33 -o $curlret -eq 36 ] ; then
		    echo "INFO: Resuming download not possible. ->" \
		         "Overwriting old file."
		    rm -f "$gzfile.incomplete"
		    curl -w '\rFinished downloading %{size_download} bytes in %{time_total} seconds (%{speed_download} bytes/sec). \n' --progress-bar $curl_options "$url" -o "$gzfile.incomplete"
		    curlret="$?"
		fi

		if [ $curlret -ne 0 ] ; then
		    case "$curlret" in
		      18)
		      echo "WARNING: Got only some of the" \
			   "file. A re-run of $0"
		      echo "WARNING: is required to complete" \
			   "the download." ;;
		      130)
		      echo -e '\rWARNING: CURL got a SIGINT' \
			   "(someone pressed Ctrl-C). A re-run of"
		      echo "WARNING: $0 is required to complete" \
			   "the download." ; sleep 1 ;;
		      *)
		      echo "$curlret $gzfile $url" \
					>> tmp/Download-Errors
		      echo -e '\rERROR: CURL Returned Error' \
			   "$curlret. Please read" \
			   "the curl manpage." ;;
		    esac
		    return 1
		elif [ ! -s "$gzfile.incomplete" ] ; then
		    echo "0 $gzfile $url" >> tmp/Download-Errors
		    echo "ERROR: CURL returned success but" \
		         "we have no data!"
		    curlret=1
		else
		    case "$gzfile" in
		      *.gz|*.tgz)
			  typeexpr="gzip compressed data" ;;
		      *.bz2|*.tbz2|*.tbz)
			  typeexpr="bzip2 compressed data" ;;
		      *.Z|*.tZ)
			  typeexpr="compress'd data" ;;
		      *.zip|*.jar)
			  typeexpr="Zip archive data" ;;
		      *.tar)
			  typeexpr="tar archive" ;;
		      *)
			  echo "WARNING: Unkown file extension: $gzfile"
			  typeexpr="." ;;
		    esac
		    if file "$gzfile.incomplete" | grep -v "$typeexpr"
		    then
			echo "ERROR: File type does not match" \
			     "filename ($typeexpr)!"
			mv "$gzfile.incomplete" "$gzfile.extck-err"
		    else
			mv "$gzfile.incomplete" "$gzfile"
		    fi
		fi
		;;
	*)
		protocol="${url%%://*}"

		# we need to use $location - $url is already mangled above -ReneR
		# $protocol://$url $options
		url="`echo "$location" | sed "s,$protocol://\([^ ]*\).*,\1,"`"
		options="`echo "$location" | cut -d' ' -f2-`"

		case "$protocol" in
			cvs)
				# the first option is the module name
				module="${options%% *}"
				options="${options#* }"
				cmdline="cvs -z4 -Q -d $url co -P $options $module"
		  
		  		# sometimes cvs wants to read ~/.cvspass just for fun ..
				touch $HOME/.cvspass
				;;
			svn|svn\+http)
				if [ "$protocol" == "svn+http" ]; then
					url="http://$url"
				else
					url="svn://$url"
				fi
				
				if [ "${options:0:1}" == "-" ]; then
					# the module is the last dir of $url
					module="${url##*/}"
				else
					# the first option is the module name
					module="${options%% *}"
					options="${options#* }"
				fi
				cmdline="svn co $options $url $module"
				;;
			*)
				echo "$cmdclient unrecognized!"
				return 1
				;;
		esac

		cvsdir="tmp/down.${protocol}dir.`echo $bzfile | tr / -`"
		saved_pwd=$PWD ; mkdir -p $cvsdir ; cd $cvsdir

		echo "$cmdline"
		{
		  $cmdline || touch .cvs_error
		} &> .cvs_output &

		while fuser .cvs_output &> /dev/null ; do
			echo -ne `nice du -sh 2> /dev/null | \
				cut -f1` 'downloaded from archive so far...\r'
			sleep 3
		done

		if [ -f .cvs_error ] ; then
			cd $saved_pwd ; rm -rf $cvsdir
			echo -e "\nError during checkout."
			return 1
		fi

		echo `du -sh 2> /dev/null | \
			cut -f1` 'downloaded from archive (download finished).'


		if [ `echo * | wc -w` -gt 1 ]; then
			# multi-module module
			echo "Multi-module package detected, relocating..."
			mkdir t2-module.$$
			for x in *; do
				[ "$x" != "t2-module.$$" ] && mv -f $x t2-module.$$/
			done
			mkdir -p "$module"
			mv -f t2-module.$$/* "$module"
			rm -f t2-module.$$
		fi

		cd `dirname $module`
		tarname="`basename $bzfile`"
		echo "Preparing files for final tarball ..."
		find -type d \( -name CVS -o -name .svn \) | xargs rm -rf

		if [ `find -type f | wc -l` -gt 4 ]; then
			find `basename $module` | xargs touch -t 200001010000
			tar --owner root --group root \
			    --use-compress-program=bzip2 \
			    -cf $tarname `basename $module`
			mv $tarname $saved_pwd/$bzfile
		else
			echo "Too few files - assuming checkout failure."
			curlret=1
		fi

		cd $saved_pwd ; rm -rf $cvsdir
		;;
	esac
	return $curlret
}

list_dtags() {
	{ 
		grep -H '^\[D\] ' package/*/*/*.desc
		grep -H '^[X0-9]' target/*/download.txt 2> /dev/null | sed 's,:,:[D] ,'
	} | column_clean
}

list_cksums() {
	trap '' INT

	# we know we only have single spaces due to list_dtags' column_clean
	list_dtags | sed -n \
	  -e 's,[^ ]* \([X0-9]*\) \(.\)\([^ ]*\) -.*,\1 download/local/\2/\2\3,p' \
	  -e 's,[^ ]* \([X0-9]*\) \(.\)\([^ ]*\) [^-].*,\1 download/mirror/\2/\2\3,p'

	trap INT
}

list() {
	trap '' INT
	list_cksums | cut -f2- -d' '
	trap INT
}

list_missing() {
	trap '' INT
	list | bz2filename | \
	while read fn ; do
		[ -f "$fn" ] || echo "$fn"
	done
	trap INT
}

repository() {
	for repository ; do
		packages `echo package/$repository/*/*.desc`
	done
}

required() {
    # Choosen config must exist
    #
    if [ ! -f config/$config/packages ]; then
        echo "ERROR: Config $config doesn't exist."
        echo "ERROR: try ./scripts/Config -cfg $config first."
        exit 1
    fi

    while read on a b repo pkg c ; do
	if [ "$on" = "X" ] ; then
	    grep -H '^\[D\] ' package/$repo/$pkg/$pkg.desc > tmp/down.$$.lst
	    while read tag cksum file url ; do
		download_file "`source_file cksum $file "$url"`" "$url" "$cksum" "$repo" "$pkg"
	    done < tmp/down.$$.lst ; rm -f tmp/down.$$.lst
	fi
    done < config/$config/packages

    # NOTE: ROCKCFG -> SDECFG automatized convertion
    grep -q 'ROCKCFG' ./config/$config/config && 
	sed -i -e 's,ROCKCFG,SDECFG,g' ./config/$config/config

    target=`grep '^export SDECFG_TARGET=' config/$config/config | \
						cut -f2 -d= | tr -d "'"`
    targetchain="$target"; x="$target"
    while [ -f "target/$x/extends" ]; do
	x="`cat target/$x/extends`"
	targetchain="$targetchain $x"
    done

    for target in $targetchain; do
    if [ -f target/$target/download.txt ] ; then
	while read cksum file url ; do
	    download_file "`source_file cksum $file "$url"`" "$url" "$cksum" "$target"
	done < target/$target/download.txt
    fi
    done
}

all() {
    local each repo pkg

    for repo in $( cd package; ls -1 ); do
	for each in package/$repo/*/*.desc; do
		pkg="`echo $each | cut -f3 -d/`"

		while read tag cksum file url ; do
		    download_file "`source_file cksum $file "$url"`" "$url" "$cksum" "$repo" "$pkg"
		done < <(grep -H '^\[D\] ' package/$repo/$pkg/$pkg.desc)
	done
    done
    
    for each in $( ls -1 target/*/download.txt 2> /dev/null ); do
	target="`echo $each | cut -f2 -d/`"

	while read cksum file url ; do
	    download_file "`source_file cksum $file "$url"`" "$url" "$cksum" "$target"
	done < <(cat $each)
    done
}

package() {
	descfile="`echo package/*/$1/$1.desc`"

	if [ ! -f $descfile ]; then
		echo "Skipping \"$1\" (not found)!"
		return
	fi

	pkg="`echo $descfile | cut -f3 -d/`"
	repo="`echo $descfile | cut -f2 -d/`"

	while read tag cksum file url ; do
		download_file "`source_file cksum $file "$url"`" "$url" "$cksum" "$repo" "$pkg"
	done < <(grep -H '^\[D\] ' package/$repo/$pkg/$pkg.desc)
}

packages() {
	local descfile
	for arg; do
		case "$arg" in
		target/*)
			if [ ! -f $arg ]; then
			    echo "Skipping \"$arg\" (not found)!"
			    continue
			fi

			target="`echo $arg | cut -f2 -d/`"

			while read cksum file url ; do
			    download_file "`source_file cksum $file "$url"`" "$url" "$cksum" "$target"
			done < <(cat $arg)
			;;
		*)
			if [ "${arg%.desc}" != "$arg" ]; then
			arg="`echo $arg | cut -f3 -d/`"; fi


			# active extensions
			local extender=

			# pkg_*_{pre,post}.conf is only activated if extender
			# is enabled on $config/packages, so we will only
			# download files of those extenders
			#
			for extender in `ls -1 package/*/*/pkg_${arg}_{pre,post}.conf 2> /dev/null |
			                 cut -d/ -f3 | sort -u`; do
				if grep -q "^X .* $extender " \
				   config/$config/packages; then
					echo_info "Also downloading $extender ..."
					package $extender
				fi
			done
			package $arg
			;;
		esac
	done
}

set +e 

# Things to do only for downloading
#
if [ "${1:0:5}" != "-list" -a $checkonly = 0 ]; then
	# we need curl
	if [ -z "`type -p curl`" ]; then
		echo_abort 2 "we need \`curl\` installed and available on \$PATH to proceed."
	fi

	# do mirror detection, only once
	[ $this_is_the_2nd_run = 1 ] || detect_mirror
fi

case "$1" in
	-list)		list ;;
	-list-missing)	list_missing ;;
	-list-cksums)	list_cksums ;;

	-required)	required ;;
	-all)		all ;;

	-repository)	shift ; repository "$@" ;;

	-*|"")		download_usage
			exit 1;;

	*)		packages "$@" ;;
esac

exit $downloaderr