extract update to CouchDB to force update document

[ILL-Zotero-RT] / zotero.pl

diff --git a/zotero.pl b/zotero.pl
index 2308e2f..afee591 100755 (executable)
--- a/zotero.pl
+++ b/zotero.pl
@@ -16,10 +16,21 @@ my $key    = $ENV{key}    || die "key required";
 
 my $FETCH  = $ENV{FETCH}  || 0;
 
-my $db = CouchDB->new('10.60.0.95', 5984);
-eval { $db->put("z_$UserID") }; # create user database
+my $db = CouchDB->new('10.60.0.92', 5984);
+eval { $db->put("zotero_$UserID") }; # create user database
+eval {
+       local $/ = undef;
+       my $view = <DATA>;
+       warn "# create $view";
+       $db->put("zotero_$UserID/_design/zotero" => decode_json $view)
+};
 
-my $url = "https://api.zotero.org/users/$UserID/items?format=atom&content=json&order=dateModified&sort=desc";
+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 $url = shift @urls;
 
 my $tree;
 my $ticket_items;
@@ -41,10 +52,16 @@ if ( $FETCH && mirror( $url => $file ) == RC_NOT_MODIFIED ) {
 my $feed = XMLin( $file );
 #warn "# feed ",dump($feed);
 
+sub link_to_id {
+       my $link = shift;
+       $link =~ s{.+/(items|collections)/}{}; # leave just ID
+       $link =~ s{\?.+}{};
+       return $link;
+}
+
 foreach my $entry ( keys %{ $feed->{entry} } ) {
        warn "# entry $entry ",dump($entry);
-       my $id = $entry;
-       $id =~ s{.+/items/}{}; # leave just ID
+       my $id = link_to_id $entry;
 
        my $item = $feed->{entry}->{$entry};
        warn "# entry $entry ",dump($item);
@@ -53,49 +70,50 @@ foreach my $entry ( keys %{ $feed->{entry} } ) {
                my $link = $item->{link}->[$i];
                warn "# link $id $i:",dump($link);
 
-               my $key = $link->{href};
-               $key =~ s{.+/items/}{};
-               $key =~ s{\?.+}{};
-               $item->{link}->[$i]->{key} = $key;
+               $item->{link}->[$i]->{key} = link_to_id $link->{href};
 
                if ( $link->{rel} eq 'up' ) {
                        push @{ $tree->{$key} }, $id;
                }
        }
 
-       $item->{zapi_etag} = $item->{content}->{'zapi:etag'};
+       if ( exists $item->{content} ) {
+               my $type = ( grep { exists $item->{content}->{$_} } qw(zapi:type type) )[0];
+               warn "# content has $type";
 
-       if ( $item->{content}->{'zapi:type'} eq 'json' ) {
-               my $json = $item->{content}->{content};
-               warn "# $json\n";
-               $json = $item->{content} = decode_json $json;
-               warn "# json $id ", dump $json;
+               $item->{zapi}->{etag} = $item->{content}->{'zapi:etag'} if exists $item->{content}->{'zapi:etag'};
 
-               foreach my $tag ( @{ $json->{tags} } ) {
-                       $tag = $tag->{tag};
-                       warn "# tag $id $tag\n";
-                       next unless $tag =~ m/#(\d+)/;
-                       push @{ $ticket_items->{$1} }, $id;
-               }
-       }
+               $type = $item->{zapi}->{type} = $item->{content}->{$type};
 
-       $items->{$id} = $item;
+               if ( $type =~ m/json/ ) {
 
-       my $json_md5 = md5_hex encode_json $item;
-       $item->{json_md5} = $json_md5;
+                       my $json = $item->{content}->{content};
+                       warn "# $json\n";
+                       $json = $item->{content} = decode_json $json;
+                       warn "# json $id ", dump $json;
 
-       if ( my $old_item = eval { $db->get( "z_$UserID/$id" ) } ) {
-               warn "# old_item ",dump($old_item);
+                       foreach my $tag ( @{ $json->{tags} } ) {
+                               $tag = $tag->{tag};
+                               warn "# tag $id $tag\n";
+                               next unless $tag =~ m/#(\d+)/; # XXX RT number in tag
+                               push @{ $ticket_items->{$1} }, $id;
+                       }
 
-               if ( $old_item->{zapi_etag} ne $item->{zapi_etag} || $json_md5 ne $old_item->{json_md5} ) {
-                       $item->{_rev} = $old_item->{_rev};
-                       $db->put( "z_$UserID/$id" => $item );
                } else {
-                       warn "# unchanged";
+                       warn "ERROR: $type not decoded!";
                }
-       } else {
-               $db->put( "z_$UserID/$id" => $item );
        }
+
+       foreach my $zapi ( grep { m/^zapi:/ } keys %$item ) {
+               my $name = $zapi;
+               $name =~ s/^zapi://;
+               $item->{zapi}->{$name} = delete $item->{$zapi};
+       }
+
+       $items->{$id} = $item;
+
+       $db->update( "zotero_$UserID/$id" => $item );
+
 }
 
 delete $feed->{entry};
@@ -107,6 +125,11 @@ if ( my @next = map { $_->{href} } grep { $_->{rel} eq 'next' && $_->{type} eq '
        goto restart;
 }
 
+if ( $url = shift @urls ) {
+       warn "## next url $url";
+       goto restart;
+}
+
 warn "# tree ",dump( $tree );
 
 warn "# ticket_items ",dump( $ticket_items );
@@ -124,6 +147,8 @@ foreach my $nr ( keys %$ticket_items ) {
        my $ticket = $rt->show(type => 'ticket', id => $nr);
        warn "# ticket $nr ",dump($ticket);
 
+       $db->update( "rt/$nr" => $ticket );
+
        if ( $ticket->{Queue} !~ m/ILL/i ) {
                warn "SKIP $ticket not in ILL queue!";
                next;
@@ -139,3 +164,6 @@ foreach my $nr ( keys %$ticket_items ) {
        }
 
 }
+
+__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"}}}