Chris@23: #!/bin/bash
Chris@23: 
Chris@23: # Rebuild the RDF database of composers, works, etc.
Chris@23: 
Chris@23: # Do not use this script.  Once the database is built, we need to be
Chris@23: # able to carry out manual editing, disambiguation etc without risking
Chris@23: # losing our changes in a future rebuild.  Following the initial build
Chris@23: # and publication, we really need to ensure that updates can be made
Chris@23: # without a complete rebuild.
Chris@23: 
Chris@23: # Perhaps we will need to make partial rebuild scripts (importing some
Chris@23: # new type of data that was not in the database at all before, for
Chris@23: # example) based on this.
Chris@23: 
Chris@23: # All that said, this script is provided anyway for purposes of review
Chris@23: # and reproducability.
Chris@23: 
Chris@23: echo "Running importer, log is written to importer.log"
Chris@23: 
Chris@25: #./importer 2>importer.log || exit 1
Chris@23: 
Chris@23: echo "Assembling additional sources"
Chris@23: 
Chris@23: rm -f ready.ntriples
Chris@23: 
Chris@23: for ttl in \
Chris@23:     imported.ttl \
Chris@23:     extra/cmn.ttl \
Chris@23:     extra/composer-mappings.ttl \
Chris@23:     extra/conductors.ttl \
Chris@23:     extra/new-names.ttl \
Chris@23:     extra/pianists-dbpedia.ttl \
Chris@23:     extra/styles.ttl ; do
Chris@23:     cat extra/prefixes.ttl "$ttl" | rapper -i turtle -o ntriples - http://dbtune.org/classical/resource/ >> ready.ntriples
Chris@23: done
Chris@23: 
Chris@25: sort ready.ntriples | uniq > ready.2.ntriples && mv ready.2.ntriples ready.ntriples
Chris@24: 
Chris@23: grep composer ready.ntriples | fgrep -v .html | sed 's/^.*composer\///' | \
Chris@24:     sed 's/>.*//' | grep -v http | sort | uniq > check/new-composer-uris
Chris@23: 
Chris@23: diff -u check/composer-uris check/new-composer-uris | grep -v '^---' | grep -v '^+++' > /tmp/$$
Chris@23: 
Chris@23: added=`grep '^+' /tmp/$$ | wc -l | awk '{ print $1; }'`
Chris@23: removed=`grep '^-' /tmp/$$ | wc -l | awk '{ print $1; }'`
Chris@23: 
Chris@23: echo "Done, result is in ready.ntriples"
Chris@23: echo
Chris@23: echo "Composer URI comparison:"
Chris@23: echo "Added: $added"
Chris@23: echo "Removed: $removed"
Chris@23: 
Chris@23: rm /tmp/$$