lib/Frey/ScrapBook.pm

package Frey::ScrapBook;
use Moose;

extends 'Frey';
with 'Frey::Web', 'Frey::File';

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 ],
        }
}

__PACKAGE__->meta->make_immutable;
no Moose;

1;
1	dpavlin	994	package Frey::ScrapBook;
2			use Moose;
3
4			extends 'Frey';
5	dpavlin	1133	with 'Frey::Web', 'Frey::File';
6	dpavlin	994
7			use XML::Simple;
8
9			=head1 DESCRIPTION
10
11			import data from ScrapBook Firefox plugin from L<http://amb.vis.ne.jp/mozilla/scrapbook/>
12
13			=cut
14
15			has scrapbook_path => (
16			is => 'rw',
17			isa => 'Str',
18			required => 1,
19			);
20
21			sub as_sponge {
22			my ($self) = @_;
23
24			my $xml = $self->scrapbook_path . '/scrapbook.rdf';
25			die "missing $xml" unless -e $xml;
26
27			warn "# xml ",$self->path_size($xml);
28			my $rdf = XMLin( $xml ) \|\| die "can't open $xml: $!";
29
30			my @rows;
31			my @name;
32			my $name_pos;
33
34			foreach my $item ( @{ $rdf->{'RDF:Description'} } ) {
35			warn "## item ", $self->dump( $item ) if $self->debug;
36
37			my @row;
38
39			foreach my $k ( keys %$item ) {
40			next if $k =~ m{^RDF:}i; # skip keys from RDF: namespace
41
42			my $name = $k;
43			$name =~ s{^\w+:}{}g; # strip namespace:
44
45			if ( ! defined $name_pos->{$name} ) {
46			push @name, $name;
47			$name_pos->{$name} = $#name;
48			warn "## name_pos: ", $self->dump( $name_pos ) if $self->debug;
49			}
50
51			my $pos = $name_pos->{$name};
52			$row[$pos] = $item->{$k};
53
54			}
55
56			# add generated columns
57			map {
58			push @name, $_;
59			$name_pos->{$_} = $#name;
60			warn "## additional name_pos: ", $self->dump( $name_pos ) if $self->debug;
61			} grep { ! defined $name_pos->{$_} } ( 'date', 'html_path' );
62
63			my $id = $row[ $name_pos->{id} ] \|\| die "can't find id in ", $self->dump( $name_pos );
64			my $date = $id;
65			$date =~ s{^(\d{4})(\d\d)(\d\d)(\d\d)(\d\d)(\d\d)$}{$1-$2-$3 $4:$5:$6};
66			push @row, $date;
67
68			my $html_path = $self->scrapbook_path . '/data/' . $id . '/index.html';
69			push @row, $html_path;
70
71			warn "## row ", $self->dump( @row ) if $self->debug;
72
73			push @rows, [ @row ];
74
75			}
76
77			warn "## name ", $self->dump( @name ) if $self->debug;
78
79			return {
80			rows => [ @rows ],
81			NAME => [ @name ],
82			}
83			}
84
85	dpavlin	1133	__PACKAGE__->meta->make_immutable;
86			no Moose;
87
88	dpavlin	994	1;