wiki:Z3950_sru_improvements

Version 2 (modified by simonp, 14 years ago) ( diff )

--

Proposal title

Date 2010/04/128
Contact(s) Timo Proescholdt <timo_AT_proescholdt.de>, Simon Pigot
Last edited Timestamp
Status complete
Assigned to release 2.5.0
Resources available

Overview

Currently Geonetwork has only basic Z39.50 support and no support for SRU. The current Z39.50 implementation is based on the JZKit library version 1, dating from 2002, which is no longer maintained (sourcecode is not even longer officially available). The current version of JZKit is 3. It offers support for SRU and addresses many of the issues that have only basic support in version 1. Among others these are:

  • configurability (e.g error behaviour if server does not support a requested index, mapping between context-sets)
  • session and request-set caching (important for performant web-based SRU).
  • result-record and query transformation
  • possibility to mix local and remote targets
  • more mature Z39.50 support

Moreover, the current Z39.50 interface hasn't been maintained: no date search support, improved geospacial indexing not applied to this part of the software, the Z3950 server in GeoNetwork does not produce HTML or other formats and harvest from Z3950 sources is not supported.

The proposal is to undertake an overhaul of the Z3950 part of the code, to update the JZKit library to version 3, implement an SRU webinterface, Z3950 harvester and enable the Z3950 server provide at least HTML (as well as the usual XML). Support for remote search will also be returned to GeoNetwork but changes to the current search interface to integrate it with local search will be in a later proposal.

Proposal Type

  • Type: Core Change, Module Change
  • App: GeoNetwork

Voting History

  • Vote proposed 2010/04/28.

Motivations

  • More complete Z39.50 interface support, especially concerning error conditions.
  • Support for web-based Z39.50 (SRU) adds an additional search interface.
  • Clean up of the Z39.50 code.
  • Add Z3950 harvester
  • Return Z39.50 remote search capability to trunk (but no integrated search interface till later)
  • GeoNetwork Z3950 server will provide HTML (in addition to XML) if requested by a Z client.

Proposal

Update the JZKit library to version 3, adding SRU support in the process.

Backward-compatibility is important. The traditional Z39.50 interface should not be affected, although using the new library.

JZKit3 port, implementation of SRU, server production of HTML as well as XML, Z3950 harvesting and a tab based remote search client exist in the BlueNetMEST sandbox. Bringing all of these (apart from the tab based remote search client) into the trunk for the 2.5 Unstable release will allow them to be evaluated, discussed and enhanced by the community in time for the next stable release (2.6).

SRU interface implemented as jeeves service. (http://127.0.0.1:8080/geonetwork/srv/en/portal.sru?operation=searchretrieve&version=1.2&query=title%20any%20WMO ).

The spring application-context had to be loosely integrated into Geonetwork, since JZKit makes extensive use of it. A ApplicationContext reference was added to the GeonetworkContext to make the Spring ApplicationContext available. Initialization in Geonetwork.java uppon Z39.50 initialization.

JZKit configuration files need a directory in the classpath and have to be tweaked to support a "Geonetwork" context-set. This is currently being done by creating a "virtual" context-set "geonet", which defines the indices that are searchable. Queries that contain unsupported indices (and relations) are rejected (this can be configured). Currently this is hard-wired in a configuration file, but this could be dynamically configured by Geonetwork if the set of searchable indices was available.

JZKit is dependent on the following libraries

1) JZkit libraries (remove old ki-*.jar)

  • a2j-2.0.4.jar
  • jzkit_core-3.0.0.jar
  • jzkit_service-3.0.0.jar
  • jzkit_z3950_server-3.0.0.jar
  • jzkit_z3950_plugin-3.0.0.jar

2) spring

  • spring-beans-2.5.3.jar
  • spring-context-2.5.3.jar
  • spring-core-2.5.3.jar

3) apache commons

  • commons-digester-1.6.jar
  • commons-beanutils-1.7.0.jar
  • persistence-api-1.0.jar
  • hibernate-annotations.jar

4) SRU query language parser

  • cql-java-0.0.7.jar

5) update of libs contained in GN (remove old versions from classpath)

  • marc4j-2.3.1.jar
  • jdbm-1.0.jar

Backwards Compatibility Issues

The z39.50 remote search and server depends on JZKit3. Since JZKit has no backwards compatibility these facilities have had to be re-implemented.

Risks

JZKit is not well documented at the moment, although documentation will be provided in the course of a World Meteorological Organization project on Z39.50.

Participants

  • Timo Proescholdt (developer)
  • Simon Pigot (developer)
Note: See TracWiki for help on using the wiki.