Tools in Wayback Module

edit

dk.netarkivet.wayback.NetarchiveCacheResourceStore

Wayback is a tool for browsing in webarchives. It can be downloaded from http://archive-access.sourceforge.net/projects/wayback/. The NetarchiveSuite plugin for wayback is a class NetarchiveCacheResourceStore which implements org.archive.wayback.ResourceStore. NetarchiveCacheResourceStore instantiates a connection to a NetarchiveSuite ArcRepository and retrieves archive data from it via NetarchiveSuite.

In order to make use of the plugin, it is necessary to

  1. Copy the required jar files into the lib-directory of your wayback installation
  2. Ensure that wayback has access to a NetarchiveSuite settings file with the necessary connection information

  3. Configure wayback to use NetarchiveResourceStore

The lib directory for wayback will be under wayback/WEB-INF/lib in your tomcat webapps directory. Copy into the lib directory all the jar files in netarchivesuite/lib except

 dk.netarkivet.deploy.jar
 dk.netarkivet.harvester.jar
 dk.netarkivet.viewerproxy.jar
 dk.netarkivet.monitor.jar

and the jar files for the packages wayback, je, jericho, jetty, junit, poi and libidn. These are either not required or aleady included in the wayback distribution.

The NetarchiveSuite settings file location can be specified in the catalina.sh file of your tomcat with a line like CATALINA_OPTS="-Ddk.netarkivet.settings.file=/home/user/settings_for_my_repository.xml".

NetarchiveCacheResourceStore has been tested with a wayback localcdxcollection using settings like:

<bean id="localcdxcollection" class="org.archive.wayback.webapp.WaybackCollection">

     <property name="resourceStore">
       <bean class="dk.netarkivet.wayback.NetarchiveResourceStore">
       </bean>
     </property>


    <property name="resourceIndex">
          <bean class="org.archive.wayback.resourceindex.LocalResourceIndex">
          <property name="source">
          <bean class="org.archive.wayback.resourceindex.CompositeSearchResultSource">
            <property name="CDXSources">
              <list>
                 <value>index1.cdx</value>
                 <value>index2.cdex</value>
              </list>
            </property>
          </bean>

        </property>
        <property name="maxRecords" value="40000" />
      </bean>

    </property>
  </bean>

but should work with other types of wayback collection.

There is an ant build file which can be used to repack the wayback war-file with the addition of the netarchivesuite plugin. Ant tasks to unpack and repack the wayback war-file are in wayback.build.xml and there are samples settings in examples/wayback/*.

dk.netarkivet.wayback.batch.ExtractWaybackCDXBatchJob

This batch job is a wrapper for the parts of the wayback API which generate CDX index files for use in wayback. The job can be called with a script like

java \
   -Ddk.netarkivet.settings.file=../../settings_wayback.xml \
   -Dsettings.common.applicationInstanceId=CDX_BATCH \
   -cp ../lib/dk.netarkivet.archive.jar \
   dk.netarkivet.archive.tools.RunBatch \
   -Ndk.netarkivet.wayback.ExtractWaybackCDXBatchJob \
   -J../lib/dk.netarkivet.wayback.jar,../lib/wayback-core-1.4.2.jar \
   -R1042-.*(?<!metadata-[0-9]).arc \
   -BMYREPLICA \
   -Oout.cdx

Note the syntax of the regular expression which selects all arcfiles generated by job 1042 except for metadata arcfiles. The cdx files generated are unsorted. For use in wayback they must be sorted and merged e.g. using unix sort:

export LC_ALL=C; sort --temporary-directory=/tmp 1.cdx 2.cdx 3.cdx 4.cdx > sorted.cdx

Our experience is that sorting and merging of files with total size up to 100GB can be accomplished in a few hours on a moderately powerful server machine.

dk.netarkivet.wayback.batch.ExtractDeduplicateCDXBatchJob

In netarchivesuite, duplicate objects which are not harvested are recorded as extra metadata in the heritrix crawl log. In order to be able to browse these items, these deduplication records need to be indexed. Each deduplication record will generate a cdx record showing the harvested time as the time when the duplicate record was discovered, but pointing to the archive location here the original records is stored. The batch job to execute this indexing is invoked in exactly the same way as that described above for indexing the archived data, except that in this case we would use a regular expression which matches only metadata files, rather than one which matches everything except metadata files. For example

-R1042-'.*'metadata'.*'arc

As in the above case, the returned cdx files are unsorted.

dk.netarkivet.wayback.DeduplicateToCDXApplication

This is a command line interface to the same code for generating CDX indexes from deduplication records in crawl log files (not metadata arcfiles). It can be invoked by

java -cp dk.netarkivet.wayback.jar dk.netarkivet.wayback.DeduplicateToCDXApplication crawl1.log crawl2.log crawl3.log > out.cdx

Additional Tools Manual 3.16/Tools in Wayback Module (last edited 2011-05-03 14:55:47 by SoerenCarlsen)