#!/usr/bin/env bash

# xhtml2metadata.sh - a front-end to xhtml2metadata.pl

# Eric Lease Morgan <emorgan@nd.edu>
# (c) University of Notre Dame; distributed under a GNU Public License

# April   25, 2024 - first documentation but probably written about a month ago
# October 25, 2024 - using locate to index file system


# configure
INDEX='./etc/carrels.idx'
PATTERN='index.html'
XHTML2METADATA='./bin/xhtml2metadata.pl'
HEADER='carrel\tidentifier\tcreator\ttitle\tdate\tdescription\tsubjects\tprovenance'

# initialize output
echo -e $HEADER

# locate all XHTML files and process each
INDEXES=( $( locate -d $INDEX $PATTERN  ) )
for HTML in ${INDEXES[@]}; do

	echo $HTML >&2
	$XHTML2METADATA $HTML
	
done

# fini
exit