#!/bin/bash
#
echo

if [ -z "$1" ]; then
   cat <<EOF
$0: [-d directory] pathname...
   Import MakeStuff blog entries or lj posts into a Jekyll _posts directory.
   The entries are of the form ./yy/mm/dd--name.ext; -d specifies the
   directory they are contained in, and defaults to ~/.ljarchive.  

   Filenames are converted from yyyy/mm/dd--name to yyyy-mm-dd-name format,
   and names ending in _dddd have that replaced by SLUG.html.
   The front matter is put into YAML format no matter what format it
   had originally.

   The "Subject" header is converted to "title", tags get surrounded by
   brackets to make a list, and tags of "curmudgeon" are removed from the
   list.  (That last could be generalized, but it can wait.)

   Files that have already been imported are skipped.
EOF
   exit 1
fi

case $1 in
    -d) shift; DIR=$1/; shift
	;;
esac
if [ -z $DIR ]; then
    DIR=~/.ljarchive
fi

shopt -s extglob # Required to trim whitespace; see below
read_headers () {
    REST=
    while IFS=':' read key value; do
	# trim whitespace in "value"
	value=${value##+([[:space:]])}; value=${value%%+([[:space:]])}

	[ -z "$key" ] && break	# the separator line between headers and body

	# values with colons or brackets need to be escaped/quoted
	#        Start by just double quoting everything except tags, which are safe
	#        and need special treatment anyway.

	case "$key" in
	    [sS]ubject|[tT]itle) SUBJECT="\"$value\""
		    ;;
	    [tT]ags) TAGS="${value//[\[\]]/}" # remove brackets from tag list, if any
		     ;;
	    ---)
		[ ! -z $yaml_started ] && break;
		yaml_started=1
		;;
	    "--text follows this line--") break  # Handle emacs's separator
					  ;;
	    *) if [ -z "$REST" ]; then
		   REST="${key,,}: \"$value\""
	       else
		   REST="$REST\n${key,,}: \"$value\""
	       fi
	       ;;
	esac
    done
}

slug () {
     echo "$1" | sed -e 's/Subject: *//' -e 's/"//g' -e 's/\[.*\]//' \
     	  -e 's/[ ]\+/-/g' -e 's/^-*//'  -e s/-*$// \
	  -e 's/[^-a-zA-Z0-9]//g' | tr '[A-Z]' '[a-z]'
}

for ff in $*; do
    if [ -e $ff ]; then
	f=`basename $ff`
	d=`dirname $ff`
    else
	f=$ff
	d=$DIR
    fi
    g=`echo $f | sed -e 's/\\//-/g' -e s/--/-/ -e s/[.]-// -e s/_[0-9]*$//`
    if echo $f | grep -se '_[0-9]*$'; then
       # the filename ends in a sequence number, so it must be an archived LJ post
       # extract the subject and slugify it.
       subj="`sed -n -e '/Subject:/p' -e '/Subject:/q' $DIR/$f`"
       g=$g-`slug "$subj"`.html
    fi
    if [ -e $g ]; then
	echo $g exists, skipping $f
    else
	echo $d/$f '->' $g
	{
	    echo ---
	    read_headers
	    echo title: $SUBJECT
	    echo tags: \[ `echo $TAGS | sed -e 's/, curmudgeon/,/' -e 's/curmudgeon,//'` \]
	    echo -e "$REST"
	    echo layout: post # needed because Jekyll's default mechanism appears to be broken
	    echo ---
	    cat
	    # FIXME:  take out any "another fine post" colophon.  Possibly not needed
	    #         now that we've imported all the original curmudgeon posts.
	    #         sed -i -e '/Another fine/,/The Computer Curmudgeon/d'
	} < $d/$f > $g
    fi
done