ds-rss/gen.sh

#!/usr/bin/env sh
assert_tools () {
	err=0
	while test $# -gt 0; do
		which $1  >/dev/null 2>/dev/null || {
			>&2 echo "tool missing: "$1
			err=$(( $err + 1 ))
		}
		shift
	done
	test $err -eq 0 || exit $err
}
test $# -eq 0 && {
	dependencies="xsltproc"
	assert_tools ${dependencies}
	xsltproc --encoding utf-8 -o ds-feed.xml rss.xsl data.xml
	xsltproc --encoding iso-8859-1 -o download.html download.xsl data.xml
	xsltproc --encoding utf-8 -o index.html index.xsl data.xml
	exit 0
}
test "$1" = "sort" && {
	dependencies="xsltproc"
	assert_tools ${dependencies}
	xsltproc --encoding utf-8 -o sorted.xml sort.xsl data.xml
	mv -i sorted.xml data.xml
}
linkcheck () {
	dependencies="curl uuid"
	assert_tools ${dependencies}
	tmpfile="/tmp/"$(uuid); curl https://ds.ccc.de/download.html > $tmpfile
	for f in $(echo "download.html "$tmpfile); do
		test -e $f || continue
		for e in $(cat ${f}|sed 's/<a /\n<a /g'|grep -i " href="|cut -d"=" -f2|cut -d'"' -f2); do
			status=$(curl --max-time 1 --head --silent --output /dev/null --write-out "%{http_code}" $e )
			test $status -eq 200 && >&2 echo ${e}" – "${status}" OK" || >&2 echo ${e}" – "${status}" FAILED"  
		done
	done
}
mirror_website () {
	dependencies="wget"
	assert_tools ${dependencies}
	local url=$@
	local domain=`expr "$url" : '^http[s]*://\([^/?]*\)'`
	wget \
		--recursive \
		--no-clobber \
		--page-requisites \
		--html-extension \
		--convert-links \
		--restrict-file-names=windows \
		--domains $domain \
		--no-parent \
		$url 
}
newdataentry () {
	dependencies="cat grep head cut date printf"
	assert_tools ${dependencies}
	err=0
	fl=""
	test $# -eq 0 && {
		>&2 echo "determining entry based on data.xml.\n" 
		lastentry=$(cat data.xml |grep -i "<schleuder id="|head -1|cut -d'"' -f2)
		newentry=$(( $lastentry + 1 ))
	} || {
		test "$1" = "help" && {
			echo "prints a new data entry for a release of datenschleuder\n"
			echo "release [options]\n"
			echo "option\tdescription\n"
			echo "number\tprint entry for release with custom number, default is the increment of the top entry in data.xml\n"
			echo "date\tuse a unix timestamp for a custom date\n"
			echo "teaser\tinclude teaser xml-message as last argument(s)\n"
			echo "full example: `./gen.sh release 99 1568505600 Die Ausgabe 99 ist fertig.`"
			exit 1
		}
		test $(( $1 * 1 )) -eq $1 && {
			>&2 echo "custom entry for data.xml.\n" 
			newentry=$1; shift;
		} || { echo "argument not a number."; exit 1; }
		test $(( $1 * 1 )) -eq $1 && {
			>&2 echo "custom datetime.\n" 
			datetime=$1; shift;
		} || { datetime=0; } # remaining arguments should be teaser xml-message.
	}
	echo "<schleuder id=\"$newentry\">"
	test $datetime -eq 0 && {
		#~ datetime=$(LANG=en_US.utf8 date -u +"%a %d %b %Y %H:%M") # manual copy of known entries
		datetime=$(LANG=en_US.utf8 date -u -R |rev|cut -d":" -f2-|rev) # RFC-5322-Format without seconds and TZ
	} || {
		datetime=$(LANG=en_US.utf8 date -d @${datetime} -u -R|rev|cut -d":" -f2-|rev)
	}
	echo "\t<date>"${datetime}" UT</date>" # custom format universal time
	fn="ds"$(printf "%03d" $newentry)
	for d in $(echo "covers pdfs epubs"); do
		case "$d" in
			covers) ffn=${d}"/"${fn}".jpg"; cover=$ffn ;;
			pdfs)   ffn=${d}"/"${fn}".pdf"  ;;
			epubs)  ffn=${d}"/"${fn}".epub" ;;
			#~ *) echo "file "${ffn}" not found, in "${d}"." ;;
		esac
		#~ echo $ffn
		test -e ${ffn} && {
			test "$d" = "covers" && {
				echo "\t<image>"${cover}"</image>"
			} || {
				fs=$(stat -c %s ${ffn})
				echo "\t<link filesize=\""${fs}"\">"${ffn}"</link>"
			}
		} || {
			mfl=${ffn}"\n"
			#~ echo "file "${ffn}" not found."
			err=3
		}
	done
    #~ echo "\t<preface></preface>" # not used on server yet
	test $# -eq 0 && { teaser="" ; } || {
		teaser=$@ ;
		echo "\t<teaser>"${teaser}"</teaser>"
	}
	echo "</schleuder>"
	test "$teaser" = "" && { >&2 echo "\nmissing teaser message."; }
	test -e $cover || { >&2 echo "\nmissing cover."; }
	test $err -gt 0 && { >&2 echo "\nmissing files:\n"${mfl}; }
}
newinfoentry () {
	echo "not implemented yet.\n\nwill be similar to `release`."
}
# a mirror is useful for local testing with files
test "$1" = "mirror" && {
	shift
	>&2 echo "downloading the website may take quite some time."
	mirror_website http://ds.ccc.de/
	
}
test "$1" = "linkcheck" && {
	shift
	linkcheck http://ds.ccc.de/
}
test "$1" = "release" && {
	shift
	newdataentry $@
}
test "$1" = "info" && {
	shift
	newinfoentry $@
}
#gs -sDEVICE=pdfwrite -dPDFSETTINGS=/ebook -q -o ds100-smaller.pdf ds100.pdf