Editing New Features in EPrints 3.3 (section)

==New Features in EPrints 3.3.x==

Note: 3.3.5 is the first stable release of 3.3. Versions 3.3.1 to 3.3.4 can be accessed via [http://files.eprints.org/view/type/beta=5Frelease.html Beta releases].

===EPrints Bazaar===

"1-click" extensions and add-ons, see http://bazaar.eprints.org/. To install add-ons go to Admin &rarr; System Tools &rarr; EPrints Bazaar.

EPrints includes GUI tools for creating new Bazaar packages and selecting the source files.

As part of this change a new system-level <code>lib/plugins</code> directory has been added allowing extensions to be installed globally. This reflects how mod_perl works, which has a single namespace for all repositories. By default plugins installed in lib/plugins will be disabled and hence can be enabled on a per-repository basis.

===Search engine plugins===

Searches are now executed through the plugins layer with a new ''Search'' plugin type. EPrints 3.3 comes with support for one new search engine: Xapian. [http://www.xapian.org/download Xapian] is a probabilistic search engine that supports boolean queries. To use Xapian you must install the appropriate version of [http://search.cpan.org/search?query=search+xapian&mode=all Search::Xapian Perl library] and perform a full re-index. The Xapian index will be written to archives/[repoid]/var/xapian/.

The (default) Internal search engine plugin has support for prefixed terms in simple search e.g. "title:(american eagle) birds". This allows users to specify more complex search queries in a similar fashion to e.g. Google. The available prefixes are based on the field id listed in the simple search configuration.

===XSLT-based Imports, Exports and Citation Styles===

If XML::LibXSLT is installed import and export plugins and citation styles can now be written using the XSLT language (XML stylesheets). XSLT export plugins support "templating" to add headers and footers to the export.

===Flexible object support===

EPrints already has a very flexible metadata scheme basic on primitive types (ints, dates etc.). In 3.3 this flexibility is enhanced by allowing the easy creation of user-defined dataset classes. Objects in these flexible datasets can be browsed, searched and viewed in a similar way to eprints or users. They may also be referred to by existing objects (item-referencing) or behave as a child-parent (as documents are to eprints).

===SWORD 2.0/CRUD support===

The EPrints <code>/id/</code> URIs are now CRUD-aware, including content-negotiation. To content-negotiate for JSON do:

 curl -H 'Accept:application/json' http://myrepo.example/id/eprint/10

Authorised users can PUT and DELETE EPrint URIs to update and delete records:

 curl -X DELETE http://myrepo.example/id/eprint/10

===Improved workflow and document management===

AJAX-based upload, deletion and updates of documents. Files now upload automatically and have a progress bar that works in all current browsers.

Document "actions" are now individual plugins allowing extensions to add actions to individual documents in the workflow. Supplied actions include conversion, unpacking (.zip and .tar.gz), additional files and metadata extraction.

===Import workflows===

Import plugins can now have a sibling Screen plugin that provides the interface for using that importer. This is demonstrated by the ISI Web of Knowledge plugin that (if SOAP::ISIWoK is available) provides a search query and tool for importing items.

===Flexible Login and Registration===

User authentication and registration are now controlled through sub-classes of [[API:EPrints/Plugin/Screen/Login|Screen::Login]] and [[API:EPrints/Plugin/Screen/Register|Screen::Register]]. A Bazaar package is provided to enable OpenID single sign-on support. Internal (default) login and registration can be trivially disabled by disabling the ''Internal'' sub-classes.

===Scheduled tasks===

Support for scheduled tasks has been added to the EPrints indexer. See [[API:EPrints/Plugin/Event|Plugin::Event]] for an example. A <code>tools/schedule</code> tool has been added to allow easy creation of indexer events (including scheduled tasks).

Tasks can be scheduled similarly to ''cron'', down to a per-minute resolution. The indexer provides no guarantees for when tasks may actually happen, just that they will not occur more frequently than the scheduled times.

===Misc.===

* Multipart field type, same behaviour as Compound but for multiples stores all parts in a single table (Name is now a sub-class)
* System:: classes for abstracting different system behaviours (i.e. making MSWin32 work)
* Support for importing OpenXML bibliography format