lib/Frey/ScrapBook.pm

package Frey::ScrapBook;
use Moose;

extends 'Frey';
with 'Frey::Web';
with 'Frey::File';
#with 'Frey::Storage';

use XML::Simple;

=head1 DESCRIPTION

import data from ScrapBook Firefox plugin from L<http://amb.vis.ne.jp/mozilla/scrapbook/>

=cut

has scrapbook_path => (
        is => 'rw',
        isa => 'Str',
        required => 1,
);

sub as_sponge {
        my ($self) = @_;

        my $xml = $self->scrapbook_path . '/scrapbook.rdf';
        die "missing $xml" unless -e $xml;

        warn "# xml ",$self->path_size($xml);
        my $rdf = XMLin( $xml ) || die "can't open $xml: $!";

        my @rows;
        my @name;
        my $name_pos;

        foreach my $item ( @{ $rdf->{'RDF:Description'} } ) {
                warn "## item ", $self->dump( $item ) if $self->debug;

                my @row;

                foreach my $k ( keys %$item ) {
                        next if $k =~ m{^RDF:}i; # skip keys from RDF: namespace

                        my $name = $k;
                        $name =~ s{^\w+:}{}g; # strip namespace:

                        if ( ! defined $name_pos->{$name} ) {
                                push @name, $name;
                                $name_pos->{$name} = $#name;
                                warn "## name_pos: ", $self->dump( $name_pos ) if $self->debug;
                        }

                        my $pos = $name_pos->{$name};
                        $row[$pos] = $item->{$k};

                }

                # add generated columns
                map {
                        push @name, $_;
                        $name_pos->{$_} = $#name;
                        warn "## additional name_pos: ", $self->dump( $name_pos ) if $self->debug;
                } grep { ! defined $name_pos->{$_} } ( 'date', 'html_path' );

                my $id = $row[ $name_pos->{id} ] || die "can't find id in ", $self->dump( $name_pos );
                my $date = $id;
                $date =~ s{^(\d{4})(\d\d)(\d\d)(\d\d)(\d\d)(\d\d)$}{$1-$2-$3 $4:$5:$6};
                push @row, $date;
                
                my $html_path = $self->scrapbook_path . '/data/' . $id . '/index.html';
                push @row, $html_path;

                warn "## row ", $self->dump( @row )  if $self->debug;

                push @rows, [ @row ];

        }

        warn "## name ", $self->dump( @name ) if $self->debug;

        return {
                rows => [ @rows ],
                NAME => [ @name ],
        }
}

1;
1	package Frey::ScrapBook;
2	use Moose;
3
4	extends 'Frey';
5	with 'Frey::Web';
6	with 'Frey::File';
7	#with 'Frey::Storage';
8
9	use XML::Simple;
10
11	=head1 DESCRIPTION
12
13	import data from ScrapBook Firefox plugin from L<http://amb.vis.ne.jp/mozilla/scrapbook/>
14
15	=cut
16
17	has scrapbook_path => (
18	is => 'rw',
19	isa => 'Str',
20	required => 1,
21	);
22
23	sub as_sponge {
24	my ($self) = @_;
25
26	my $xml = $self->scrapbook_path . '/scrapbook.rdf';
27	die "missing $xml" unless -e $xml;
28
29	warn "# xml ",$self->path_size($xml);
30	my $rdf = XMLin( $xml ) \|\| die "can't open $xml: $!";
31
32	my @rows;
33	my @name;
34	my $name_pos;
35
36	foreach my $item ( @{ $rdf->{'RDF:Description'} } ) {
37	warn "## item ", $self->dump( $item ) if $self->debug;
38
39	my @row;
40
41	foreach my $k ( keys %$item ) {
42	next if $k =~ m{^RDF:}i; # skip keys from RDF: namespace
43
44	my $name = $k;
45	$name =~ s{^\w+:}{}g; # strip namespace:
46
47	if ( ! defined $name_pos->{$name} ) {
48	push @name, $name;
49	$name_pos->{$name} = $#name;
50	warn "## name_pos: ", $self->dump( $name_pos ) if $self->debug;
51	}
52
53	my $pos = $name_pos->{$name};
54	$row[$pos] = $item->{$k};
55
56	}
57
58	# add generated columns
59	map {
60	push @name, $_;
61	$name_pos->{$_} = $#name;
62	warn "## additional name_pos: ", $self->dump( $name_pos ) if $self->debug;
63	} grep { ! defined $name_pos->{$_} } ( 'date', 'html_path' );
64
65	my $id = $row[ $name_pos->{id} ] \|\| die "can't find id in ", $self->dump( $name_pos );
66	my $date = $id;
67	$date =~ s{^(\d{4})(\d\d)(\d\d)(\d\d)(\d\d)(\d\d)$}{$1-$2-$3 $4:$5:$6};
68	push @row, $date;
69
70	my $html_path = $self->scrapbook_path . '/data/' . $id . '/index.html';
71	push @row, $html_path;
72
73	warn "## row ", $self->dump( @row ) if $self->debug;
74
75	push @rows, [ @row ];
76
77	}
78
79	warn "## name ", $self->dump( @name ) if $self->debug;
80
81	return {
82	rows => [ @rows ],
83	NAME => [ @name ],
84	}
85	}
86
87	1;