#!/bin/sh
set -e

# usage: 
#	./md.sh <target dir>
#	./md.sh <input md> <output html>
#
# target dir: input `src.md`, output `index.html`
# dependencies: lowdown, fzf, python (for now)


# i/o variables

IN="$1/src.md"
OUT="$1/index.html"

if [ "$2" ] ; then
	IN="$1" ;
	OUT="$2" ;
fi

# confirm source exists before continuing

[ -f $IN ] || { echo "---- [FATAL] $IN not found ----" ; exit 1 ; }


### initial write

cat md-header > $OUT
lowdown -Thtml $IN >> $OUT
cat md-footer >> $OUT
echo '[ok] initial assembly'


### useful functions

sedesc() {
	sed -e 's;&;\\\&;g' -e 's/;/\\\;/g'
}

unesc() {
	sed -e 's;\\\&;\&;g' -e 's/\\\;/;/g'
}

encode() ( # TODO: python't
	python -c "import urllib, sys; print urllib.quote(sys.argv[1])" "$1"
)

decode() {
	/bin/printf '%b' "$(echo "$1" | sed 's;%;\\x;g')"
}


### header variables extracted from metadata section

TITLE=$(grep -e '^title: ' -m 1 $IN | cut -f 2- -d ' ' | sedesc)
sed "s/<title>#/<title>${TITLE}/1" -i'' $OUT
printf '[ok] title: %s\n' "$(echo $TITLE | unesc)"

BOMBER=$(grep -e '^bomber: ' -m 1 $IN | cut -f 2- -d ' ' | sedesc)
sed "s/<div class=\"bomber\">#/<div class=\"bomber\">${BOMBER}/1" -i'' $OUT
printf '[ok] bomber: %s\n' "$(echo $BOMBER | unesc)"

MDATE=$(stat -c '%y' $IN | sed -e 's/\.[0-9]* //' -e 's/\(..\)\(..\)$/\1:\2/')	# format: "yyyy-mm-dd hh:mm:ss±zz:zz" (ISO 8601)
DATE=$(echo $MDATE | cut -f 1 -d ' ' | sed -e 's/-//g' -e 's/^..//')		# format: "yymmdd" (for humans)
sed "s;<div class=\"ident\"><time>#;<div class=\"ident\"><time datetime=\"${MDATE}\">${DATE};1" -i'' $OUT
sed "/^<\/head>/i <meta http-equiv=\"last-modified\" content=\"${MDATE}\" />" -i'' $OUT
printf '[ok] date: %s (%s)\n' "$DATE" "$MDATE"

LEADING=$(grep -e '^leading: ' -m 1 $IN | cut -f 2- -d ' ' | sedesc)
sed "s;<div class=\"leading\">#;<div class=\"leading\">${LEADING};1" -i'' $OUT
printf '[ok] leading: "%s"\n' "$(echo $LEADING | unesc)"

METADESC="\&laquo\;${BOMBER}\&raquo\; \&mdash\; ${LEADING}"
sed "s;<meta name=\"description\" content=\"#;<meta name=\"description\" content=\"${METADESC};1" -i'' $OUT
printf '[ok] meta description\n'


### image handling

grep -n '<p><img' $OUT | \
while IFS='' read -r data ; do
	LINE=$(echo $data | cut -f 1 -d :) ;	# extract line number
	FILE=$(echo $data | cut -f 2 -d \") ;	# extract file path
	CAPT=$(echo $data | cut -f 4 -d \") ;	# extract image caption

	if [ "$CAPT" = "#tex" ] ; then 		# for tex figures: remove caption and containers
		sed "${LINE}s;.*;<figure><img src=\"${FILE}\" alt=\"\"></figure>;" -i'' $OUT ;
		printf '[ok] tex figure at line %s\n' "$LINE" ;
	else					# use new image format
		sed "${LINE}s;.*;<figure><a href=\"${FILE}\"><img src=\"${FILE}\" alt=\"\"></a><figcaption><p>${CAPT}</p></figcaption></figure>;" -i'' $OUT ;
		printf '[ok] image at line %s\n' "$LINE" ;
	fi
done


### table of contents generation

tocgen() (

tmp=$(mktemp -p /tmp) # using a tempfile, more convenient for storage and operations

TOCLIST=$(grep -e '^#' $IN | sed -e 's; ;- ;' -e 's;^#;;' -e 's;#;	;g') # headers from source as md list

echo "$TOCLIST" | while IFS='' read -r data ; do # encode headers as urls
	HEADER=$(echo "$data" | sed 's/^[[:space:]-]*//' | encode "$(cat /dev/stdin)") ;
	echo "$data" | sed "s/^\([[:space:]-]*\)\(.*\)$/\1${HEADER}/" >> $tmp ;
done

buffer=$(cat $tmp) ; echo "$buffer" | lowdown -Thtml > $tmp # convert to html, write to tempfile (reset)

sed -e '1i <nav class="toc">' -e '/^$/d' -i'' $tmp # prepend starting tag, remove empty lines

cat $tmp | while IFS='' read -r data ; do

	# get header (both encoded and raw), write as link
	ID=$(echo $data | sed "s;^<li>\([^</>]*\)\(</li>\)*$;\1;") ;
	HEADER=$(decode "$ID" | sedesc) ;
	sed -e "s;^<li>\(${ID}\);<li><a href=\"#\1\">${HEADER};" -e "s;\([[:alnum:]]\)</li>$;\1</a></li>;" -i'' $tmp ;

	# find old id in output, replace with new id
	IDLINE=$(grep -ne "^<h[[:digit:]] id=" $OUT | fzf -f "$(echo $HEADER | unesc)" | head -n 1 | cut -f 1 -d :) ;
	if [ "$IDLINE" ] ; then
		sed -e "${IDLINE}s;id=\".*\">.*</h;id=\"${ID}\">${HEADER}</h;" -i'' $OUT ;
		printf '     assembled header "%s"\n' "$(echo $HEADER | unesc)" ;
	# else
	#	printf '[warning] could not match header id: "%s"\n' "$HEADER" ;
 	#	echo "---- [FATAL] could not find header id (possible escapement issue) ----" ;
 	#	rm $tmp ;
 	#	exit 1 ; 
 	fi ;
done

# close link tags on open lines, append closing tag

sed "s;[[:alnum:]]$;&</a>;" -i'' $tmp
echo '</nav>' >> $tmp

# get toc line number in output, insert data from tempfile

TOCLINE=$(grep -n 'end of header' $OUT | cut -f 1 -d :)
sed "${TOCLINE}r $tmp" -i'' $OUT

rm $tmp # cleanup

echo '[ok] table of contents'

)


### metadata option handling

opts=$(grep -e '^opts: ' -m 1 $IN | cut -f 2- -d ' ')

while [ "$opts" ]; do
case "$opts" in
	(*[Cc]*)
		tocgen
		opts=$(echo $opts | sed 's/[Cc]//g')
	;;

	(*[Hh]*)
		sed "/^<header>/,/^<\/header>/d" -i'' $OUT
		sed "/bg.js/d" -i'' $OUT
		echo "[ok] headerless"
		opts=$(echo $opts | sed 's/[Hh]//g')
	;;

	(*)
		echo "[warning] metadata: unrecognized option(s)" 
	;;
esac
done


echo "---- build complete: ${OUT} ----"