use JSON;
use Data::Dump qw(dump);
use RT::Client::REST;
+use URI::Escape;
use CouchDB;
use Digest::MD5 qw(md5_hex);
$db->put("zotero_$UserID/_design/zotero" => decode_json $view)
};
+eval { $db->put("rt") }; # create RT database
+
my @urls = map { "https://api.zotero.org/users/$UserID/$_?format=atom&content=json&order=dateModified&sort=desc" } qw( collections items );
# we don't need to fetch tags since we can generate using CouchDB views
my $file = $UserID . '.' . md5_hex($url) . '.atom';
$FETCH = 1 if ! -e $file;
-warn "# $url -> $file\n";
+warn "# mirror $FETCH $url -> $file\n";
if ( $FETCH && mirror( $url => $file ) == RC_NOT_MODIFIED ) {
warn "not modified";
-# exit 0;
}
-my $feed = XMLin( $file );
-#warn "# feed ",dump($feed);
+my $xml = XML::Simple->new(ForceArray => [ qw( entry ) ]);
+my $feed = eval { $xml->XMLin( $file ) };
+if ( $! ) {
+ warn "ERROR $file $!\n";
+ goto skip_url;
+}
+warn "# feed ",dump($feed);
sub link_to_id {
my $link = shift;
return $link;
}
+my @collection_items;
+
foreach my $entry ( keys %{ $feed->{entry} } ) {
warn "# entry $entry ",dump($entry);
my $id = link_to_id $entry;
+ push @collection_items, $id if $url =~ m{/collections/(\w+)/items};
+
my $item = $feed->{entry}->{$entry};
- warn "# entry $entry ",dump($item);
+ warn "# item $id $entry ",dump($item),$/;
foreach my $i ( 0 .. $#{ $item->{link} } ) {
my $link = $item->{link}->[$i];
if ( $link->{rel} eq 'up' ) {
push @{ $tree->{$key} }, $id;
+ } elsif ( $link->{rel} eq 'self' && $link->{href} =~ m{/collections/} ) {
+ warn "# get items in this collection";
+ push @urls, "$link->{href}/items?content=json";
}
}
$items->{$id} = $item;
- my $json_md5 = md5_hex encode_json $item;
- $item->{zapi}->{json_md5} = $json_md5;
+ $db->modify( "zotero_$UserID/$id" => $item );
- if ( my $old_item = eval { $db->get( "zotero_$UserID/$id" ) } ) {
- warn "# old_item ", $old_item->{_rev}; #dump($old_item);
+}
- if ( $old_item->{zapi}->{etag} ne $item->{zapi}->{etag} || $json_md5 ne $old_item->{zapi}->{json_md5} ) {
- $item->{_rev} = $old_item->{_rev};
- warn :"# update $id";
- $db->put( "zotero_$UserID/$id" => $item );
- } else {
- warn "# unchanged";
- }
- } else {
- $db->put( "zotero_$UserID/$id" => $item );
- warn "# insert $id ", dump($item);
- }
+if ( @collection_items ) {
+ my $id = $1 if $url =~ m{/collections/(\w+)/items};
+ $db->modify( "zotero_$UserID/$id" => sub {
+ my $doc = shift;
+ $doc->{x_meta}->{collection_items} = [ @collection_items ];
+ return $doc;
+ });
}
delete $feed->{entry};
goto restart;
}
+skip_url:
+
if ( $url = shift @urls ) {
warn "## next url $url";
goto restart;
foreach my $nr ( keys %$ticket_items ) {
- my $ticket = $rt->show(type => 'ticket', id => $nr);
+ my $ticket = eval { $rt->show(type => 'ticket', id => $nr) };
warn "# ticket $nr ",dump($ticket);
+ next unless $ticket;
+
+ $ticket->{zotero_items} = $ticket_items->{$nr};
+
+ my $modified = $db->modify( "rt/$nr" => sub {
+ my $doc = shift;
+ $doc->{$_} = $ticket->{$_} foreach keys %$ticket;
+ return $doc;
+ });
+
+ warn "# modified ",dump($modified);
+
+ # copy attachments to CouchDB (they never change, so do it just once
+ if ( my @attachment_ids = $rt->get_attachment_ids( id => $nr ) ) {
+
+ warn "# get_attachment_ids = ",dump( @attachment_ids );
+ my $doc = $db->get("rt/$nr");
+ my @attachments;
+
+ foreach my $attachment_id ( @attachment_ids ) {
+ my $attachment = $rt->get_attachment( parent_id => $nr, id => $attachment_id );
+ if ( $attachment->{Filename} && $attachment->{ContentEncoding} eq 'base64' ) {
+ #$attachment->{Filename} ||= $attachment_id;
+ my $content = delete $attachment->{Content};
+ if ( ! exists $doc->{_attachments}->{ $attachment->{Filename} } ) {
+ utf8::encode($content) || warn "utf8::encode error!";
+ warn "# extracted ",length( $content ), " bytes";
+ warn "## attachment ",dump( $attachment );
+ my $url = sprintf 'rt/%d/%s?rev=%s', $nr, uri_escape($attachment->{Filename}), $modified->{rev};
+# $modified = $db->request( PUT => $url, $content, $attachment->{ContentType} );
+ }
+ }
+ push @attachments, $attachment;
+ }
+
+
+ $db->modify( "rt/$nr" => sub {
+ my $doc = shift;
+ $doc->{attachments} = [ @attachments ];
+ warn "## attachments on $nr = ", $#attachments + 1;
+ return $doc;
+ }) if @attachments;
+
+ }
+
if ( $ticket->{Queue} !~ m/ILL/i ) {
warn "SKIP $ticket not in ILL queue!";
next;
# $rt->comment( ticket_id => $nr, message => dump( $items->{$id} ) );
- last; # FIXME just first
-
}
}
__DATA__
-{"_id":"_design/zotero","language":"javascript","views":{"itemType":{"map":"function(doc) {\n emit(doc.zapi.itemType,1);\n}","reduce":"_count"},"updated":{"map":"function(doc) {\n emit(doc.updated,1);\n}","reduce":"_count"},"tags":{"map":"function(doc) {\n \n doc.content.tags.forEach( function(v) {\n emit(v, doc._id);\n });\n}","reduce":"_count"},"link_up":{"map":"function(doc) {\n if ( doc.link[1].rel == 'up' )\n emit( doc.link[1].key, doc._id );\n}","reduce":"_count"},"year,publisher":{"map":"function(doc) {\n if ( doc.zapi.year )\n emit([doc.zapi.year, doc.content.publisher], 1);\n}","reduce":"_count"}}}
+{"_id":"_design/zotero","views":{"link_up":{"map":"function(doc) {\n if ( doc.link[1].rel == 'up' )\n emit( doc.link[1].key, doc._id );\n}","reduce":"_count"},"year,publisher":{"map":"function(doc) {\n if ( doc.zapi.year )\n emit([doc.zapi.year, doc.content.publisher], 1);\n}","reduce":"_count"},"updated":{"map":"function(doc) {\n emit(doc.updated,1);\n}","reduce":"_count"},"itemType":{"map":"function(doc) {\n emit(doc.zapi.itemType,1);\n}","reduce":"_count"},"tags":{"map":"function(doc) {\n \n doc.content.tags.forEach( function(v) {\n emit(v, doc._id);\n });\n}","reduce":"_count"},"collection_items":{"map":"function(doc) {\n if ( doc.x_meta ) {\n doc.x_meta.collection_items.forEach( function(id) {\n emit(doc.content.name, id);\n });\n }\n}","reduce":"_count"}},"language":"javascript"}