#
# Copyright (C) 1998, 1999 Ken MacLeod
# XML::Grove::Path is free software; you can redistribute it
# and/or modify it under the same terms as Perl itself.
#
# $Id: Path.pm,v 1.2 1999/08/17 15:01:28 kmacleod Exp $
#
package XML::Grove::Path;
use XML::Grove;
use XML::Grove::XPointer;
use UNIVERSAL;
sub at_path {
my $element = shift; # or Grove
my $path = shift;
$path =~ s|^/*||;
my @path = split('/', $path);
return (_at_path ($element, [@path]));
}
sub _at_path {
my $element = shift; # or Grove
my $path = shift;
my $segment = shift @$path;
# segment := [ type ] [ '[' index ']' ]
#
# strip off the first segment, finding the type and index
$segment =~ m|^
([^\[]+)? # - look for an optional type
# by matching anything but '['
(?: # - don't backreference the literals
\[ # - literal '['
([^\]]+) # - index, any non-']' chars
\] # - literal ']'
)? # - the whole index is optional
|x;
my ($node_type, $instance, $match) = ($1, $2, $&);
# issues:
# - should assert that no chars come after index and before next
# segment or the end of the query string
$instance = 1 if !defined $instance;
my $object = $element->xp_child ($instance, $node_type);
if ($#$path eq -1) {
return $object;
} elsif (!$object->isa('XML::Grove::Element')) {
# FIXME a location would be nice.
die "\`$match' doesn't exist or is not an element\n";
} else {
return (_at_path($object, $path));
}
}
package XML::Grove::Document;
sub at_path {
goto &XML::Grove::Path::at_path;
}
package XML::Grove::Element;
sub at_path {
goto &XML::Grove::Path::at_path;
}
1;
__END__
=head1 NAME
XML::Grove::Path - return the object at a path
=head1 SYNOPSIS
use XML::Grove::Path;
# Using at_path method on XML::Grove::Document or XML::Grove::Element:
$xml_obj = $grove_object->at_path("/some/path");
# Using an XML::Grove::Path instance:
$pather = XML::Grove::Path->new();
$xml_obj = $pather->at_path($grove_object);
=head1 DESCRIPTION
C<XML::Grove::Path> returns XML objects located at paths. Paths are
strings of element names or XML object types seperated by slash ("/")
characters. Paths must always start at the grove object passed to
`C<at_path()>'. C<XML::Grove::Path> is B<not> XPath, but it should
become obsolete when an XPath implementation is available.
Paths are like URLs
/html/body/ul/li[4]
/html/body/#pi[2]
The path segments can be element names or object types, the objects
types are named using:
#element
#pi
#comment
#text
#cdata
#any
The `C<#any>' object type matches any type of object, it is
essentially an index into the contents of the parent object.
The `C<#text>' object type treats text objects as if they are not
normalized. Two consecutive text objects are seperate text objects.
=head1 AUTHOR
Ken MacLeod, ken@bitsko.slc.ut.us
=head1 SEE ALSO
perl(1), XML::Grove(3)
Extensible Markup Language (XML) <http://www.w3c.org/XML>
=cut