#!/bin/sh
#
# See LICENSE for license details.
#
# This is based on:
#     https://github.com/revan/pptx2md
#
# The algorithm was simplified to the essence of what the ugly
# pptx format is capable of. Microsoft amateurs are unable to use
# XML properly.
#
# Requires: xml2tsv (git://bitreich.org/xml2tsv)
#

if [ $# -lt 1 ];
then
	printf "usage: %s file.pptx\n" "$(basename "$0")" >&2
	exit 1
fi

pptxfile="$1"
mdfile="$1.md"
tmpdir="$(mktemp -u)"

unzip -oq -d "${tmpdir}" "${pptxfile}"
if [ $? -ne 0 ];
then
	printf "Failed to extract %s.\n" "${pptxfile}"
	exit 1
fi

for slidefile in ${tmpdir}/ppt/slides/*.xml;
do
	linenum=0
	cat "${slidefile}" \
	| xml2tsv 2>/dev/null \
	| grep a:r/a:t \
	| cut -s -f 2 \
	| while read -r line;
	do
		if [ $linenum -eq 0 ];
		then
			printf "## %s\n" "${line}" >> "${mdfile}"
		else
			printf "%s\n" "${line}" >> "${mdfile}"
		fi
		linenum=1
	done
	printf "\n" >> "${mdfile}"
done


rm -r "$tmpdir"
exit 0