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 |
$feed->load_or_create( |
$feed->load_or_create( |
70 |
uri => 'file://' . $path, |
uri => 'file://' . $path, |
71 |
title => 'ScrapBook', |
title => 'ScrapBook', |
72 |
#source => 'Grep::Source', |
#source => 'Grep::Source', |
73 |
|
owner => $owner, |
74 |
); |
); |
75 |
|
|
76 |
|
my $search = Grep::Search->new; |
77 |
|
|
78 |
my $stats; |
my $stats; |
79 |
|
|
80 |
foreach my $item ( @{ $rdf->{'RDF:Description'} } ) { |
foreach my $item ( @{ $rdf->{'RDF:Description'} } ) { |
97 |
|
|
98 |
# fetch full-text content and import it |
# fetch full-text content and import it |
99 |
|
|
100 |
my $content_path = $dir . '/data/' . $hash->{id} . '/index.html'; |
my $rel_path = '/data/' . $hash->{id} . '/index.html'; |
101 |
|
|
102 |
|
my $content_path = $dir . $rel_path; |
103 |
if ( ! -r $content_path ) { |
if ( ! -r $content_path ) { |
104 |
Jifty->log->warn("can't import $content_path: $!"); |
Jifty->log->warn("can't import $content_path: $!"); |
105 |
$stats->{failure}++; |
$stats->{failure}++; |
108 |
my $content = read_file( $content_path ) or |
my $content = read_file( $content_path ) or |
109 |
die "can't read $content_path: $!"; |
die "can't read $content_path: $!"; |
110 |
|
|
111 |
|
my $tree = HTML::TreeBuilder->new or die "can't create html tree"; |
112 |
|
$tree->parse( $content ) or die "can't parse fetched content"; |
113 |
|
|
114 |
|
my $body = $tree->look_down( '_tag', 'body' ); |
115 |
|
|
116 |
|
my $resolver = HTML::ResolveLink->new( base => '/static/' . $config->{Dir} . $rel_path ); |
117 |
|
$content = $resolver->resolve( $body->as_HTML ); |
118 |
|
|
119 |
# create date from id |
# create date from id |
120 |
|
|
130 |
#time_zone => 'UTC', |
#time_zone => 'UTC', |
131 |
); |
); |
132 |
} else { |
} else { |
133 |
warn "can't parse date from ", $hash->{id}; |
die "can't parse date from ", $hash->{id}; |
134 |
} |
} |
135 |
|
|
136 |
my $i = Grep::Model::Item->new(); |
my $i = Grep::Model::Item->new( current_user => $owner ); |
137 |
my ($ok,$msg) = $i->load_or_create( |
my ($ok,$msg) = $i->load_or_create( |
138 |
in_feed => $feed, |
in_feed => $feed, |
139 |
title => $hash->{title}, |
title => $hash->{title}, |
140 |
link => $hash->{source}, |
link => $hash->{source}, |
141 |
content => $content, |
content => $content, |
142 |
issued => $hash->{id}, |
# created_on => $dt, |
143 |
); |
); |
144 |
|
|
145 |
|
$i->set_created_on( $dt ); |
146 |
|
|
147 |
if ( ! $ok ) { |
if ( ! $ok ) { |
148 |
Jifty->log->error( $msg ); |
Jifty->log->error( $msg ); |
149 |
$stats->{failure}++; |
$stats->{failure}++; |
154 |
$stats->{old}++; |
$stats->{old}++; |
155 |
} else { |
} else { |
156 |
$stats->{new}++; |
$stats->{new}++; |
157 |
Jifty->log->info("imported ", $i->id ," ", $i->link, " ", length( $content ), " bytes"); |
Jifty->log->info("created ", $i->id ," ", $i->link, " ", length( $content ), " bytes"); |
158 |
$search->add( $i, $i->in_feed->owner->id ); |
$search->add( $i, $owner->id ); |
159 |
} |
} |
160 |
|
|
161 |
} |
} |
162 |
|
|
163 |
|
$search->finish; |
164 |
|
|
165 |
return $stats; |
return $stats; |
166 |
} |
} |
167 |
|
|