#!/usr/bin/perl -w
# Blosxom Plugin: rss_to_wp
# Author(s): jps
# Version: n/a
# Blosxom Home/Docs/Licensing: http://www.raelity.org/apps/blosxom/
#
# A number of fixes for the RSS2.0 -> WordPress import
#
#	Requires:
#		interpolate_fancy
#		changes to rss20 flavour templates as described below

package rss_to_wp;

sub start { 1; }

# Wordpress converts newlines to <br/>, which can break linewrapped
# tags e.g. <a title="..." [newline] href="...">
sub story {
  	my($pkg, $path, $filename, $story_ref, $title_ref, $body_ref) = @_;
  	${$body_ref} =~ s/[\n\s]+/ /gis;
}

# Turn a Blosxom path - /category/subcategory/subsubcategory - into
# a list delimited by "</category><category>" instead
#
# Usage: 
#	include interpolate_fancy package
#	then in flavour templates, write e.g:
#		<title><@rss_to_wp.unescapeAmp text="$title" output="yes" /></title>
sub pathtocats {
	# Package
	my $arg1 = shift;
	# Hash ref
	my $arg2 = shift;

	# Get path attribute of quasi-tag
	my $path = $arg2->{path};

	# Strip off any initial whitespace or a slash
	$path =~ s!^\s*/?!!;
	# All other slashes become tag close/open
	$path =~ s!/!</category>\n\n<category>!g;

	return $path;
}

# Some fields get incorrectly escaped, so unescape any Unicode entities
# we find in these
#
# Usage: 
#	include interpolate_fancy package
#	then in flavour templates, write e.g:
#		<category><@rss_to_wp.pathtocats path="$path" output="yes" /></category>
sub unescapeAmp {
	# Package
	my $arg1 = shift;
	# Hash ref
	my $arg2 = shift;

	# Escape all entities of the forms
	#	#999999...
	#	#x999999...
	#	abcdef...
	$arg2->{text} =~ s/&amp;(#x?[0-9a-f]+|[a-z]+);/&\1;/gi;

	return $arg2->{text};
}

# Blosxom needs this to know the package is valid
1;