ctan

#!bash

# $Id: ctan 52 2008-05-24 06:10:06Z kyanh $

export PATH=$PATH:$HOME/bin

source vnmik.configuration

CTAN_ARCHIVE=http://ctan.org/tex-archive
CTAN_CONTRIB=$CTAN_ARCHIVE/macros/latex/contrib/
KYANHNET_REPOS=http://kyanh.net/ctan.tools/
DESTDIR=$PKGDIR/ctan
TOOLS="lynx wget tar bzip2 sed sort gawk"
HAVE_LZMA=0

# functions

tar()
{
	z $*
}

msg()
{
	echo -e ":: $*"
}

# check for executable tool

ctan_check()
{
	type $1 2>/dev/null 1>&2
	if [[ $? -gt 0 ]]; then
		ERROR=1
		msg "not found '$1'"
		return 1
	fi	
}

# check for all tools

ctan_check_all()
{
	# msg "checking dependencies..."
	ERROR=0
	for tool in $TOOLS; do
		ctan_check $tool
	done
	if [[ $ERROR == 1 ]]; then
		msg "error occured. exit 1"
		exit 1
	fi

	# msg "found $TOOLS"

	# optional packages

	ctan_check lzma
	if [[ $? == 0 ]]; then
		HAVE_LZMA=1
		# msg "found lzma"
	fi
	
	return 0
}

# show information after update

ctan_info()
{
		msg
		msg "* `wc -l $DESTDIR/packages.txt |gawk '{print $1}'` package(s) found"		
		msg "* `wc -l $DESTDIR/files.txt |gawk '{print $1}'` file(s) found"		
}

# update database

ctan_update()
{
	if [[ "x$1" == "x--first-time" ]]; then
		shift
		msg "This maybe the first time you use the tool."
		msg "The directory ~/.ctan will be created"
	fi

	mkdir -pv $DESTDIR

	# using the cache file http://kyanh.net/ctan.tools/cache.tbz

	if [[ ( "x$1" == "x" ) || ( "x$1" == "x--use-cache" ) ]]; then
		shift
		msg "update by using cache file..."

		if [[ $HAVE_LZMA == 1 ]]; then
			msg "getting $KYANHNET_REPOS/ctan.tools/cache.tlzma..."
			wget $KYANHNET_REPOS/cache.tlzma -O $DESTDIR/cache.tlzma

			msg "uncompressing $DESTDIR/cache.tlzma..."
			tar xvf $DESTDIR/cache.tlzma --use-compress-program lzma -C $DESTDIR
		else
			msg "getting $KYANHNET_REPOS/ctan.tools/cache.tbz..."
			wget $KYANHNET_REPOS/cache.tbz -O $DESTDIR/cache.tbz

			msg "uncompressing $DESTDIR/cache.tbz..."
			tar xfjv $DESTDIR/cache.tbz -C $DESTDIR
		fi

		msg "removing {dirs,files,dump_files}.txt as it is too big..."
		msg rm -fv $DESTDIR/{dump_files.txt,cache.tbz,cache.tlzma}

		ctan_info

		return 0
	fi

	# donot use cache. download directly from CTAN (15MB)

	msg "update will get about 15MB from CTAN and may take very long time."
	read -n1 -p":: Are you sure you wanna do this (y/N)?" reply
	msg
	if [[ ! "x$reply" == "xy" ]]; then
		msg "Update was cancelled."
		return 0
	fi

	# create list of packages ####################################################

	msg "dumping $CTAN_CONTRIB..."
	lynx -dump $CTAN_CONTRIB > $DESTDIR/contrib.txt
	grep -E '^[ ]*\[[0-9]+\][^ ]+/' $DESTDIR/contrib.txt \
		|gawk '{printf("%s, %s\n", $1,$2);}' \
		|sed \
			-e 's/\/, /\t/g' \
			-e 's#\[[0-9]\+\]##g'	\
		> $DESTDIR/packages.txt

	# create list of files #######################################################

	msg "searching all files..."

	lynx -dump 'http://ctan.org/cgi-bin/filenameSearch.py?filename=.' \
		> $DESTDIR/dump_files.txt

	if [[ $? -gt 0 ]]; then
		msg "failed to download from CTAN. exit 1"
		rm -fv $DESTDIR/dump_files.txt
		exit 1
	fi

	msg "creating list of directories..."

	grep tex-archive/ $DESTDIR/dump_files.txt \
		|sed \
			-e 's/#/ /g' \
			-e 's#http://ctan.org/tex-archive/# #g' \
		|gawk '{print $2}' \
		|sort -u \
		> $DESTDIR/dirs.txt

	msg "creating list of files..."

	grep -E '^\[[0-9]+\][^ ]+' $DESTDIR/dump_files.txt \
		|gawk -F']' '{print $2}'\
		|sort -u \
		> $DESTDIR/files.txt

	msg "compressing the results..."
	cd $DESTDIR
	
	msg "creating tbz archive..."
	tar -cvj -f .cache.tbz {dirs,files,packages}.txt
	mv .cache.tbz cache.tbz
	
	if [[ $HAVE_LZMA == 1 ]]; then
		msg "creating lzma archive..."
		tar -cv --use-compress-program lzma -f .cache.tlzma {dirs,files,packages}.txt
		mv .cache.tlzma cache.tlzma
	fi

	msg "removing dump_files.txt as it's too big..."
	rm -fv $DESTDIR/dump_files.txt

	ctan_info

	if [[ "x$1" == "x--keep-only-cache" ]]; then
		msg "removing all temporary files. Only cache is kept"
		rm -fv $DESTDIR/*.txt
	fi
		
}

check_first_time()
{
	[[ -d $DESTDIR/ ]] || ctan_update --first-time --use-cache
}

# search something

ctan_search()
{
	check_fist_time
	if [[ "x$1" == "x-file" ]]; then
		shift
		grep $* $DESTDIR/files.txt | grep -v obsolete
	else # search package name	
		grep $* $DESTDIR/packages.txt | grep -v obsolete
	fi	
	if [[ $? -gt 0 ]]; then
		msg "no package/file found"
		return 1
	fi
}

# download files

ctan_get()
{
	check_first_time
	ctan_search $* > $DESTDIR/tmp.txt
	if [[ $? -gt 0 ]]; then
		msg "no package matches '$*'"
		return
	fi
	if [[ "x$1" == "x-file" ]]; then
		shift
		ctan_grep_type='file'
	else
		ctan_grep_type='pkg'
	fi
	if [[ "$ctan_grep_type" == "file" ]]; then		
		for pkg in `cat $DESTDIR/tmp.txt`; do
			msg "$CTAN_ARCHIVE/$pkg";			
		done
		msg
		read -n 1 -p ":: Are you sure you want to download (y/N)?" download
		if [[ "x$download" == "xy" ]]; then
			msg
			for pkg in `cat $DESTDIR/tmp.txt`; do
				msg "downloading $pkg..."
				wget --directory-prefix=$PKGDIR/ctan -c "$CTAN_ARCHIVE/$pkg";
			done				
		fi		
	else
		gawk '{print $1}' $DESTDIR/tmp.txt > $DESTDIR/tmp2.txt
		for pkg in `cat $DESTDIR/tmp2.txt`; do
			msg "$pkg\t $CTAN_CONTRIB/$pkg.zip";
		done
		msg
		read -n 1 -p "Are you sure you want to download (y/N)?" download
		if [[ "x$download" == "xy" ]]; then
			msg
			for f in `cat $DESTDIR/tmp2.txt`; do
				msg "downloading $pkg..."
				wget --directory-prefix=$PKGDIR/ctan -c "$CTAN_CONTRIB/$pkg.zip";
			done	
		fi
	fi	
}

ctan_about()
{
	msg "ABOUT"
	msg
	msg "\tThis small tool is used to seach/download some LaTeX packages that were listed in CTAN:"
	msg "\t\t$CTAN_CONTRIB"
	msg "\tor search/download any files in"
	msg "\t\t$CTAN_ARCHIVE/."
	msg ""
	msg "\tSince version 1.0.1 the tool uses the cache file"
	msg "\t\t$KYANHNET_REPOS/cache.tbz"
	msg "\tThis file contains the list of all files in CTAN. It's updated daily by a cron job hosted"
	msg "\tby http://kyanh.net/. The size of file is about 400k."
	msg
	msg "\tThe tool requires wget, lynx, gawk, wc, tar/bzip and sed program to run correctly."
	msg "\tSome linux distributions donot contain lynx by default and you must install lynx manually"
	msg "\tbefore running this script. So this tool requires a *nix-liked environment to work."
	msg "\ttexer <VietTUG> tested and reported that the tool worked fine in Cygwin environment."
	msg
	msg "\tThis tool was written by kyanh <xkyanh@gmail.com>"
	msg "\tkyanh is a member of Vietnamese TeX Users Group <http://viettug.org>."
	msg
	msg "VERSION"
	msg
	msg	"\t1.0.0 2008/05/11: first version"
	msg	"\t1.1.0 2008/05/15: search files (use cache file)"
	msg	"\t1.1.1 2008/05/15: fix bug (ctan_update)"
	msg "\t1.2.0 2008/05/16: smaller cache. Thanks to Karl Berry <karl@freefriends.org>"
	msg "\t1.3.0 2008/05/23: everything is cached. Faster. New options. Thanks to texer at VietTUG"
	msg "\t1.3.1 2008/05/24: fix typo. lzma support"
	msg
	ctan_usage
	msg
	msg "TODO"
	msg
	msg "\t* package information supported"
	msg "\t* ability to select FTP mirror"
	msg "\t* search by package description,..."
	msg "\t* package build script"
	msg
	msg "THANKS"
	msg
	msg "\t* Karl Berry <karl@freefriends.org>"
	msg "\t* Nguyen Van Hanh <hanhnguyenvan@gmail.com>"
	msg
	msg "LICENSE"
	msg
	msg "\tThis tool is published under LPPL."
	msg
	msg "BUGS"
	msg "\tPlease report to kyanh <xkyanh@gmail.com>."
	msg
	
}

ctan_usage()
{
	msg "USAGE"
	msg
	msg "\tctan about :\tshow all information about this tool"
	msg "\tctan usage :\tshow usage"
	msg "\tctan version :\tshow script version"
	msg
	msg "\tctan update :\tupdate using $KYANHNET_REPOS/cache.tbz (450k)"
	msg "\tctan update --direct :\tupdate directly from CTAN. You are going to download 15MB"
	msg "\tctan grep <string> :\tsearch packages match <string>. grep ability is supported"
	msg "\tctan get <string> :\tdownload packages match <string> to working directory"
	msg
	msg "\tIf you want to search files:"
	msg
	msg "\tctan fgrep <string> :\tsearch files."
	msg "\tctan fget <string> :\tdownload files match <string> to working directory"
	msg "\tctan grep -file <string>"
	msg "\tctan get -file <string>"
	msg
	msg "EXAMPLES"
	msg
	msg "\tctan grep theorem\t# search packages match 'theorem'"
	msg "\tctan grep ^n\t# search packages srated by 'n'"
	msg "\tctan get ^n\t# download packages started by 'n'"
	msg "\tctan get -file contrib/ntheorem.zip"
	msg "\tctan fget contrib/ntheorem.zip"
	msg
	msg	"\tThe result by '-file' option should be filtered"
	msg
	msg "\tctan fgrep ntheorem | grep zip"
	msg 	
}

ctan_arg()
{
	if [[ "x$1" == "x-file" ]]; then
		shift
	fi	
	if [[ "x$1" == "x" ]]; then
		msg "missing parameter"
		exit 1
	fi
}

# check for requirements or not

case "x$1" in
"xusage");;
"xhelp");;
"x");;
"xabout");;
"xdoc");;
"xversion");;
"xcheck");;
*) ctan_check_all;;
esac

# main program

case "x$1" in
"xupdate")
	shift
	ctan_update $*
	;;
"xfgrep")
	shift
	ctan_arg $*
	ctan_search -file $*
	;;
"xgrep")
	shift
	ctan_arg $*
	ctan_search $*
	;;
"xget")
	shift
	ctan_arg $*
	ctan_get $*
	;;
"xfget")
	shift
	ctan_arg $*
	ctan_get -file $*
	;;
"x")	ctan_usage;;
"xusage") ctan_usage;;
"xhelp") ctan_usage;;
"xdoc") ctan_about;;
"xabout") ctan_about;;
"xversion") grep '^# $Id' $0;;
"xcheck") ctan_check_all;;
*)
	msg "wrong parameter. please try 'ctan usage'"
	;;
esac