WebPAC/Input/XML.pm

package WebPAC::Input::XML;

use warnings;
use strict;

use WebPAC::Input;
use base qw/WebPAC::Common/;

use XML::Simple;
use File::Find;

use Data::Dump qw/dump/;

=head1 NAME

WebPAC::Input::XML - support for reading XML files

=cut

our $VERSION = '0.01';

=head1 FUNCTIONS

=head2 new

Returns new low-level input API object

  my $input = new WebPAC::Input::XML(
        path => '/path/to/XML/records.txt'
        filter => sub {
                my ($l,$field_nr) = @_;
                # do something with $l which is line of input file
                return $l;
        },
  }

Options:

=over 4

=item path

path to directory with xml files ending in C<.xml>

=back

=cut

sub new {
        my $class = shift;
        my $self = {@_};
        bless($self, $class);

        my $arg = {@_};

        my $log = $self->_get_logger();

        $log->logdie("can't find path ", $arg->{path}, ": $!\n") unless -d $arg->{path};

        $log->info("collecting xml files from ", $arg->{path});

        my @files;

        find({
                wanted => sub {
                        my $path = $File::Find::name;
                        return unless -f $path && $path =~ m/\.xml$/i;
                        push @files, $path;
                },
                follow => 1,
        }, $arg->{path} );

        $log->info("found ", $#files + 1, " XML files in ", $arg->{path});

        $self->{_files} = \@files;

        $self ? return $self : return undef;
}

=head2 fetch_rec

Return record with ID C<$mfn> from database

  my $rec = $input->fetch_rec( $mfn, $filter_coderef );

=cut

sub fetch_rec {
        my $self = shift;

        my ( $mfn, $filter_coderef ) = @_;

        my $path = $self->{_files}->[ $mfn - 1 ] || return;

        my $log = $self->_get_logger();

        my $xml = XMLin( $path, ForceArray => 1 ) || $log->logdie("can't open $path: $!");

        warn "### fetch_rec( $mfn ) => $path => xml is ",dump($xml);

        # add mfn
        $xml->{'000'} = [ $mfn ];

        return $xml;
}


=head2 size

Return number of records in database

  my $size = $input->size;

=cut

sub size {
        my $self = shift;
        return $#{$self->{_files}} + 1;
}

=head1 AUTHOR

Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>

=head1 COPYRIGHT & LICENSE

Copyright 2007 Dobrica Pavlinusic, All Rights Reserved.

This program is free software; you can redistribute it and/or modify it
under the same terms as Perl itself.

=cut

1;
1	package WebPAC::Input::XML;
2
3	use warnings;
4	use strict;
5
6	use WebPAC::Input;
7	use base qw/WebPAC::Common/;
8
9	use XML::Simple;
10	use File::Find;
11
12	use Data::Dump qw/dump/;
13
14	=head1 NAME
15
16	WebPAC::Input::XML - support for reading XML files
17
18	=cut
19
20	our $VERSION = '0.01';
21
22	=head1 FUNCTIONS
23
24	=head2 new
25
26	Returns new low-level input API object
27
28	my $input = new WebPAC::Input::XML(
29	path => '/path/to/XML/records.txt'
30	filter => sub {
31	my ($l,$field_nr) = @_;
32	# do something with $l which is line of input file
33	return $l;
34	},
35	}
36
37	Options:
38
39	=over 4
40
41	=item path
42
43	path to directory with xml files ending in C<.xml>
44
45	=back
46
47	=cut
48
49	sub new {
50	my $class = shift;
51	my $self = {@_};
52	bless($self, $class);
53
54	my $arg = {@_};
55
56	my $log = $self->_get_logger();
57
58	$log->logdie("can't find path ", $arg->{path}, ": $!\n") unless -d $arg->{path};
59
60	$log->info("collecting xml files from ", $arg->{path});
61
62	my @files;
63
64	find({
65	wanted => sub {
66	my $path = $File::Find::name;
67	return unless -f $path && $path =~ m/\.xml$/i;
68	push @files, $path;
69	},
70	follow => 1,
71	}, $arg->{path} );
72
73	$log->info("found ", $#files + 1, " XML files in ", $arg->{path});
74
75	$self->{_files} = \@files;
76
77	$self ? return $self : return undef;
78	}
79
80	=head2 fetch_rec
81
82	Return record with ID C<$mfn> from database
83
84	my $rec = $input->fetch_rec( $mfn, $filter_coderef );
85
86	=cut
87
88	sub fetch_rec {
89	my $self = shift;
90
91	my ( $mfn, $filter_coderef ) = @_;
92
93	my $path = $self->{_files}->[ $mfn - 1 ] \|\| return;
94
95	my $log = $self->_get_logger();
96
97	my $xml = XMLin( $path, ForceArray => 1 ) \|\| $log->logdie("can't open $path: $!");
98
99	warn "### fetch_rec( $mfn ) => $path => xml is ",dump($xml);
100
101	# add mfn
102	$xml->{'000'} = [ $mfn ];
103
104	return $xml;
105	}
106
107
108	=head2 size
109
110	Return number of records in database
111
112	my $size = $input->size;
113
114	=cut
115
116	sub size {
117	my $self = shift;
118	return $#{$self->{_files}} + 1;
119	}
120
121	=head1 AUTHOR
122
123	Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>
124
125	=head1 COPYRIGHT & LICENSE
126
127	Copyright 2007 Dobrica Pavlinusic, All Rights Reserved.
128
129	This program is free software; you can redistribute it and/or modify it
130	under the same terms as Perl itself.
131
132	=cut
133
134	1;