#!/usr/bin/env perl

# prt-notes.pl
# outputs text only version of notes from a Libreoffice Impress file
# It will also work on the ODP save of a Microsoft Office Powerpoint presentation

# (c) Richard Roth - rroth@on-the-net.com
# Released under Mozilla Public License V2.0 http://mozilla.org/MPL/2.0/

$deb = 0;

use Data::Dumper;
use XML::LibXML::Reader;
use XML::LibXML::PrettyPrint;
use Text::Unidecode;

#use XML::XPath;
#use XML::XPath::XMLParser;

$in = $ARGV[0];

binmode STDOUT, ":utf8";

#qx!cp -av $in /tmp/present.zip!;
qx!unzip -p -- '$in' content.xml >/tmp/content.xml!;

my $document = XML::LibXML->new->parse_file("/tmp/content.xml");


#my $reader = XML::LibXML::Reader->new(IO => $zipin);
#$reader->finish;
#$document = $reader->document;

 my $pp = XML::LibXML::PrettyPrint->new(indent_string => "  ");
# $pp->pretty_print($document); # modified in-place
$notes = $document->toString;

$pages = $document->find("//draw:page");

sub pageSort {
   my($a, $b) = @_;
   $pa = $a->getAttribute("draw:name");
   $pa =~ s/Slide//;

   $pb = $b->getAttribute("draw:name");
   $pb =~ s/Slide//;

   return $pa <=> $pb;
}

for $page (sort { pageSort($a, $b) }  @$pages) {
	print "=1111==============\n" if($deb);
	print __LINE__ . ":".Dumper($page) if($deb);
	print $pp->pretty_print($page) if($deb);

	$pageNum = $page->getAttribute("draw:name");
	print "---- PAGE=$pageNum\n";

	print "--- Title: ----\n";
	$titleElem = $page->find('./draw:frame[@presentation:class="title"]');
	print __LINE__.":".Dumper($titleElem) if($deb);
	if($titleElem) {
	    print $pp->pretty_print($titleElem->[0]) if($deb);
	    $realtexts = $titleElem->[0]->find(".//text:span/text()");
	    for $rtext (@$realtexts) {
			print __LINE__.":".Dumper($rtext) if($deb);
#			print $pp->pretty_print($rtext)->toString if($deb);
			print unidecode($rtext->data) . "\n";
	    }
	}


	print ("=222" x 20) ."\n"  if($deb);
	$notes = $page->find("./presentation:notes");
	print __LINE__.":".Dumper($notes) if($deb);
	for $note (@$notes) {
	   print __LINE__.":".Dumper($note) if($deb);
	   print $pp->pretty_print($note) if($deb);


	   print "---- NOTES: ---------------\n";
	   $texts = $note->find(".//draw:text-box");
	   for $text (@$texts) {
		print __LINE__.":".Dumper($text) if($deb);
		print $pp->pretty_print($text)->toString if($deb);

		$realtexts = $text->find(".//text:span/text()");
		print __LINE__.":".Dumper($realtexts) if($deb);
		for $rtext (@$realtexts) {
#			print __LINE__.":".Dumper($rtext) if($deb);
#			print $pp->pretty_print($rtext)->toString;
			$body = unidecode($rtext->data);
			print $body . "\n" if($body);
		}
	   }
	}

	print "\n\n";
}