17 |
|
|
18 |
or modify L<ScrapBookDir> path (relative to Grep installation static root). |
or modify L<ScrapBookDir> path (relative to Grep installation static root). |
19 |
|
|
|
=head1 SEE ALSO |
|
|
|
|
|
L<http://amb.vis.ne.jp/mozilla/scrapbook/> - ScrapBook FireFox extension |
|
|
|
|
20 |
=cut |
=cut |
21 |
|
|
22 |
use XML::Simple; |
use XML::Simple; |
23 |
use File::Slurp; |
use File::Slurp; |
24 |
|
use HTML::ResolveLink; |
25 |
|
use HTML::TreeBuilder; |
26 |
use Data::Dump qw/dump/; |
use Data::Dump qw/dump/; |
27 |
|
|
28 |
sub import { |
sub import { |
29 |
my $self = shift; |
my $self = shift; |
30 |
my $search = shift or die "need search"; |
|
31 |
die "search is ", ref($search), " and not Grep::Search" unless ($search->isa('Grep::Search')); |
my $config = Jifty->config->app('Import')->{'ScrapBook'}; |
32 |
|
|
33 |
|
if (! $config ) { |
34 |
|
Jifty->log->warn("skipping ScrapBook importer, no application->Import->ScrapBook config"); |
35 |
|
return; |
36 |
|
} |
37 |
|
|
38 |
|
# required parametars in config.yml |
39 |
|
foreach my $param ( qw/Dir OwnerEmail/ ) { |
40 |
|
Jifty->log->die("can't find $param in Scrapbook config") unless defined ( $config->{$param} ); |
41 |
|
}; |
42 |
|
|
43 |
my $dir = |
my $dir = |
44 |
Jifty::Util->app_root . '/' . |
Jifty::Util->app_root . '/' . |
45 |
Jifty->config->framework('Web')->{'StaticRoot'} . '/' . |
Jifty->config->framework('Web')->{'StaticRoot'} . '/' . $config->{'Dir'}; |
|
Jifty->config->app('ScrapBookDir'); |
|
46 |
|
|
47 |
my $path = $dir . '/scrapbook.rdf'; |
my $path = $dir . '/scrapbook.rdf'; |
48 |
$path =~ s!//+!/!g; |
$path =~ s!//+!/!g; |
59 |
|
|
60 |
# warn "## original rdf -> ", dump( $rdf ); |
# warn "## original rdf -> ", dump( $rdf ); |
61 |
|
|
62 |
my $feed = Grep::Model::Feed->new(); |
my $owner = Grep::Model::User->new(); |
63 |
|
$owner->load_by_cols( email => $config->{OwnerEmail} ); |
64 |
|
die "can't find ScrapBookOwner ", $config->{OwnerEmail} unless ( $owner->id ); |
65 |
|
|
66 |
|
Jifty->log->info( "Using user ", $owner->id, " from ", $owner->email, " for import" ); |
67 |
|
|
68 |
|
my $feed = Grep::Model::Feed->new( current_user => $owner ); |
69 |
|
|
70 |
$feed->load_or_create( |
$feed->load_or_create( |
71 |
uri => 'file://' . $path, |
uri => 'file://' . $path, |
72 |
title => 'ScrapBook', |
title => 'ScrapBook', |
73 |
#source => 'Grep::Source', |
#source => 'Grep::Source', |
74 |
|
owner => $owner, |
75 |
); |
); |
76 |
|
|
77 |
|
my $search = Grep::Search->new; |
78 |
|
|
79 |
|
my $stats; |
80 |
|
|
81 |
foreach my $item ( @{ $rdf->{'RDF:Description'} } ) { |
foreach my $item ( @{ $rdf->{'RDF:Description'} } ) { |
82 |
|
|
83 |
warn "## item = ",dump( $item ); |
$stats->{total}++; |
84 |
|
|
85 |
|
#warn "## item = ",dump( $item ); |
86 |
|
|
87 |
my $hash; |
my $hash; |
88 |
foreach my $k ( keys %$item ) { |
foreach my $k ( keys %$item ) { |
93 |
$hash->{$n} = $item->{$k}; |
$hash->{$n} = $item->{$k}; |
94 |
} |
} |
95 |
|
|
96 |
warn "## hash = ", dump( $hash ); |
#warn "## hash = ", dump( $hash ); |
97 |
|
|
98 |
|
|
99 |
# fetch full-text content and import it |
# fetch full-text content and import it |
100 |
|
|
101 |
my $content_path = $dir . '/data/' . $hash->{id} . '/index.html'; |
my $rel_path = '/data/' . $hash->{id} . '/index.html'; |
102 |
|
|
103 |
|
my $content_path = $dir . $rel_path; |
104 |
if ( ! -r $content_path ) { |
if ( ! -r $content_path ) { |
105 |
Jifty->log->warn("can't import $content_path: $!"); |
Jifty->log->warn("can't import $content_path: $!"); |
106 |
|
$stats->{failure}++; |
107 |
next; |
next; |
108 |
} |
} |
|
my $content = read_file( $content_path ) or |
|
|
die "can't read $content_path: $!"; |
|
|
|
|
109 |
|
|
110 |
# create date from id |
# create date from id |
111 |
|
|
112 |
my $dt; |
my $dt; |
113 |
if ( $hash->{id} =~ m/^(\d{4})(\d\d)(\d\d)(\d\d)(\d\d)(\d\d)$/ ) { |
if ( $hash->{id} =~ m/^(\d{4})(\d\d)(\d\d)(\d\d)(\d\d)(\d\d)$/ ) { |
114 |
$dt = DateTime->new( |
$dt = Jifty::DateTime->new( |
115 |
year => $1, |
year => $1, |
116 |
month => $2, |
month => $2, |
117 |
day => $3, |
day => $3, |
121 |
#time_zone => 'UTC', |
#time_zone => 'UTC', |
122 |
); |
); |
123 |
} else { |
} else { |
124 |
warn "can't parse date from ", $hash->{id}; |
die "can't parse date from ", $hash->{id}; |
125 |
} |
} |
126 |
|
|
127 |
|
my $i = Grep::Model::Item->new( current_user => $owner ); |
128 |
|
|
129 |
my $i = Grep::Model::Item->new(); |
if ( $i->load_by_cols( |
|
my ($id,$msg) = $i->load_or_create( |
|
130 |
in_feed => $feed, |
in_feed => $feed, |
131 |
title => $hash->{title}, |
title => $hash->{title}, |
132 |
link => $hash->{source}, |
link => $hash->{source}, |
133 |
content => $content, |
) ) { |
134 |
issued => $hash->{id}, |
$stats->{old}++; |
135 |
); |
Jifty->log->info("existing ", $i->id ," ", $i->link); |
136 |
|
} else { |
137 |
|
|
138 |
|
my $content = read_file( $content_path ) or |
139 |
|
die "can't read $content_path: $!"; |
140 |
|
|
141 |
|
my $tree = HTML::TreeBuilder->new or die "can't create html tree"; |
142 |
|
$tree->parse( $content ) or die "can't parse fetched content"; |
143 |
|
|
144 |
|
my $body = $tree->look_down( '_tag', 'body' ); |
145 |
|
|
146 |
|
my $resolver = HTML::ResolveLink->new( base => '/static/' . $config->{Dir} . $rel_path ); |
147 |
|
$content = $resolver->resolve( $body->as_HTML ); |
148 |
|
|
149 |
|
my ($ok,$msg) = $i->create( |
150 |
|
in_feed => $feed, |
151 |
|
created_on => $dt, |
152 |
|
title => $hash->{title}, |
153 |
|
link => $hash->{source}, |
154 |
|
content => $content, |
155 |
|
); |
156 |
|
|
157 |
warn ">> item $id $msg\n"; |
# $i->set_created_on( $dt ); |
158 |
|
|
159 |
|
if ( ! $ok ) { |
160 |
|
Jifty->log->error( $msg ); |
161 |
|
$stats->{failure}++; |
162 |
|
next; |
163 |
|
} |
164 |
|
|
165 |
|
if ( $msg && $msg !~ m/^Found/ ) { |
166 |
|
$stats->{new}++; |
167 |
|
Jifty->log->info("created ", $i->id ," ", $i->link, " ", length( $content ), " bytes"); |
168 |
|
$search->add( $i, $owner->id ); |
169 |
|
} else { |
170 |
|
Jifty->log->info("skipped ", $i->id ," ", $i->link, " ", length( $content ), " bytes"); |
171 |
|
} |
172 |
|
} |
173 |
} |
} |
174 |
|
|
175 |
return 1; |
$search->finish; |
176 |
|
|
177 |
|
return $stats; |
178 |
} |
} |
179 |
|
|
180 |
|
=head1 SEE ALSO |
181 |
|
|
182 |
|
L<http://amb.vis.ne.jp/mozilla/scrapbook/> - ScrapBook FireFox extension |
183 |
|
|
184 |
|
=cut |
185 |
|
|
186 |
1; |
1; |