Provides caching for citations to speed up and reduce load on browse …

…view generation. Disabled by default. Copy default_zero/cfg.d/citations.pl to local archive and set enabled=1 and run epadmin update
eprints · May 31, 2019 · 6aedd1c · 6aedd1c
1 parent 37209b0
commit 6aedd1c
Show file tree

Hide file tree

Showing 7 changed files with 409 additions and 7 deletions.
diff --git a/bin/epadmin b/bin/epadmin
@@ -61,6 +61,8 @@ Where I<command> is one of:
 
 =item refresh_abstracts
 
+=item refresh_citations
+
 =item refresh_views
 
 =item rehash
@@ -169,6 +171,10 @@ Tell the webserver that all views pages must be regenerated. The webserver will
 
 Tell the webserver that all abstract summary pages must be regenerated. The webserver will update them next , but won't update them again unless something on the EPrint changes or you re-run refresh abstracts. Also causes config to be reloaded.
 
+=item B<epadmin> refresh_citations I<repository_id>
+
+Tell the webserver that all citations must be regenerated. The webserver will update them next , but won't update them again unless something on the citationable object changes or you re-run refresh citation.
+
 =item B<epadmin> redo_mime_type I<repository_id> dataset [ objectid, ... ]
 
 Re-run the file format identification. Dataset may be one of 'document' or
@@ -322,6 +328,7 @@ else
 	elsif( $action eq "erase_fulltext_index" ) { erase_fulltext_index( $repoid ); }
 	elsif( $action eq "reload" ) { reload( $repoid ); }
 	elsif( $action eq "refresh_abstracts" ) { refresh_abstracts( $repoid ); }
+	elsif( $action eq "refresh_citations" ) { refresh_citations( $repoid ); }
 	elsif( $action eq "refresh_views" ) { refresh_views( $repoid ); }
 	elsif( $action eq "redo_mime_type" ) { redo_mime_type( $repoid, @ARGV ); }
 	elsif( $action eq "redo_thumbnails" ) { redo_thumbnails( $repoid, @ARGV ); }
@@ -1425,7 +1432,32 @@ END
 	}
 
 	reload( $repoid );
-}	
+}
+
+
+sub refresh_citations
+{
+        my( $repoid ) = @_;
+
+        my $repo = &repository( $repoid );
+
+        my $file = $repo->config( "variables_path" )."/citations.timestamp";
+        unless( open( CHANGEDFILE, ">$file" ) )
+        {
+                EPrints::abort( "Cannot write to file $file" );
+        }
+        print CHANGEDFILE "This file last poked at: ".EPrints::Time::human_time()."\n";
+        close CHANGEDFILE;
+        if( $noise > 0 )
+        {
+                print <<END;
+Citations will be updated when they are next requested.  You may wish to run
+refresh_abstracts and/or refresh_views as citations will only be updated on 
+non-cached pages.
+END
+        }
+}
+
 
 sub reload
 {

diff --git a/flavours/pub_lib/defaultcfg/cfg.d/citations.pl b/flavours/pub_lib/defaultcfg/cfg.d/citations.pl
@@ -0,0 +1,7 @@
+##List Dataset and Fields##
+$c->{datasets}->{citation} = {
+        class => "EPrints::DataObj::Citation",
+        sqlname => "citation",
+        index => 0,
+};
+$c->{citation_caching}->{enabled} = 0;
diff --git a/lib/defaultcfg_zero/cfg.d/citations.pl b/lib/defaultcfg_zero/cfg.d/citations.pl
@@ -0,0 +1,7 @@
+##List Dataset and Fields##
+$c->{datasets}->{citation} = {
+        class => "EPrints::DataObj::Citation",
+        sqlname => "citation",
+        index => 0,
+};
+$c->{citation_caching}->{enabled} = 0;
diff --git a/perl_lib/EPrints.pm b/perl_lib/EPrints.pm
@@ -284,6 +284,7 @@ use EPrints::DataObj;
 use EPrints::DataObj::SubObject;
 use EPrints::DataObj::Access;
 use EPrints::DataObj::Cachemap;
+use EPrints::DataObj::Citation;
 use EPrints::DataObj::Document;
 use EPrints::DataObj::EPrint;
 use EPrints::DataObj::File;

diff --git a/perl_lib/EPrints/DataObj.pm b/perl_lib/EPrints/DataObj.pm
@@ -619,6 +619,34 @@ sub clear_changed
 ######################################################################
 =pod
 
+=begin InternalDoc
+
+=item $dataobj->clear_citations( )
+
+Clear any citations associated with the DataObj.
+
+=end InternalDoc
+
+=cut
+######################################################################
+
+sub clear_citations
+{
+	my( $self ) = @_;
+
+	my $db = $self->{session}->get_database();
+	my $citations_sql = "SELECT `citationid` WHERE `datasetid` = '" . $self->{dataset}->confid . "' AND `objectid` = '" . $self->id . "'";
+	my $statement = $db->prepare( $citations_sql );
+	$db->execute($statement, $citations_sql);
+	while ( my $res = $statement->fetchrow_hashref )
+	{
+		$self->{session}->dataset( "citation" )->dataobj( $res->{citationid} )->delete();
+	}
+}
+
+######################################################################
+=pod
+
 =item $success = $dataobj->commit( [$force] )
 
 Write this object to the database and reset the changed fields.
@@ -675,6 +703,9 @@ sub commit
 	# clear changed fields
 	$self->clear_changed();
 
+	# clear citations unless this is a citation
+	$self->clear_citations() if defined $self->{session}->config( "citation_caching", "enabled" ) && $self->{session}->config( "citation_caching", "enabled" ) && $self->{dataset}->confid != "citation";
+
 	return $success;
 }
 
@@ -1146,9 +1177,56 @@ sub render_citation
 		);
 	}
 
-	return $citation->render( $self,
-		in=>"citation ".$self->{dataset}->confid."/".$style, 
-		%params );
+	my $citation_htmlobj = undef; 
+	my $res = undef;
+	my $citation_caching_enabled = defined $self->{session}->config( "citation_caching", "enabled" ) && $self->{session}->config( "citation_caching", "enabled" );
+
+	if ( $citation_caching_enabled )
+	{
+		my $db = $self->{session}->get_database();
+        	my $citations_sql = "SELECT `citationid`, `citation_text`, `timestamp_year` AS `year`, `timestamp_month` AS `month`, `timestamp_day` AS `day`, `timestamp_hour` AS `hour`, `timestamp_minute` AS `minute`, `timestamp_second` AS `second` FROM `citation` WHERE `datasetid` = '" . $self->{dataset}->confid . "' AND `objectid` = '" . $self->id . "' AND `style` = '" . $style . "' ORDER BY `timestamp_year`, `timestamp_month`, `timestamp_day`, `timestamp_hour`, `timestamp_minute`, `timestamp_second` DESC LIMIT 1";
+		my $statement = $db->prepare($citations_sql);
+        	$db->execute($statement, $citations_sql);
+	        my $res = $statement->fetchrow_hashref;
+	}
+        if (EPrints::Utils::is_set($res))
+        {
+		my $citation_unixtime = EPrints::Time::datetime_utc( $res->{year}, $res->{month}, $res->{day}, $res->{hour}, $res->{minute}, $res->{second} );
+
+                my $timestampfile = $self->{session}->get_conf( "variables_path" )."/citations.timestamp";
+                my $refresh_unixtime = ( -e $timestampfile ) ? (stat( $timestampfile ))[9] : 0;
+
+                if ( $citation_unixtime > $refresh_unixtime ) 
+		{
+                        my $citation_text = "<citation>" . $res->{ 'citation_text' } . "</citation>";
+                        return EPrints::XML->contents_of( EPrints::XML->parse_string( $citation_text )->documentElement() );  # convert to HTML Object
+                }
+                else
+                {
+                        my $citation_dataobj = $self->{session}->dataset( "citation" )->dataobj( $res->{citationid} );
+                        $citation_htmlobj = $citation->render( $self, in=>"citation ".$self->{dataset}->confid."/".$style, %params );
+                        $citation_dataobj->set_value( 'citation_text', $self->{session}->xml->to_string( $citation_htmlobj ) );
+                        $citation_dataobj->set_value( 'timestamp', EPrints::Time::get_iso_timestamp() );
+                        $citation_dataobj->commit;
+
+                }
+        }
+	else
+	{
+		$citation_htmlobj = $citation->render( $self, in=>"citation ".$self->{dataset}->confid."/".$style, %params );
+		if ( $citation_caching_enabled )
+		{
+			my $citation_data = {
+				datasetid => $self->{dataset}->confid,
+				objectid => $self->id,
+				style => $style,
+				citation_text => $self->{session}->xml->to_string( $citation_htmlobj ),
+				timestamp => EPrints::Time::get_iso_timestamp(),
+			};
+			my $citation_dataobj = $self->{session}->dataset('citation')->create_object( $self->{session}, $citation_data );
+		}
+	}
+	return $citation_htmlobj;
 }
 
 
@@ -1404,18 +1482,18 @@ sub uri
 	my( $self ) = @_;
 
 	return undef if !EPrints::Utils::is_set( $self->get_id );
-
+	
 	my $ds_id = $self->get_dataset->confid;
 	if( $self->get_session->get_repository->can_call( "dataobj_uri", $ds_id ) )
 	{
 		return $self->get_session->get_repository->call( [ "dataobj_uri", $ds_id ], $self );
 	}
-		
+
 	if ( EPrints::Utils::is_set( $self->get_session->get_repository->get_conf( "uri_url" ) ) )
         {
                 return $self->get_session->get_repository->get_conf( "uri_url" ).$self->internal_uri;
         }
-	
+
 	return $self->get_session->get_repository->get_conf( "base_url" ).$self->internal_uri;
 }