Convert B3Kat NTriples to several Beacon files

b3kat-isbn-and-isil.awk

#!/usr/bin/awk -f

#
# Given B3Kat in NTriples format this script extracts:
#
# 1. mappings from ISBN to B3Kat title URI
#    into file isbn2b3kat.beacon
# 2. mappings from B3Kat title URI to ISIL of holding library
#    into file b3kat2isil.beacon
#
# Given the actual dumps from <http://lod.b3kat.de/download> one can invoke:
#
#   zcat lod.b3kat.de.part??.rdf.gz | \
#   rapper -I http://lod.b3kat.de/ -i rdfxml -o ntriples - | \
#   ./b3kat-isbn-and-isil.awk
#
# See <http://gbv.github.com/beaconspec/beacon.html> for a specification of
# the resulting Beacon text format.
#

BEGIN {
    I2B = "isbn2b3kat.beacon"
    B2I = "b3kat2isil.beacon"

    print "#FORMAT: BEACON" > I2B
    print "#PREFIX: urn:isbn:" >> I2B
    print "#RELATION: http://purl.org/ontology/bibo/isbn" >> I2B
    print "#TARGET: http://lod.b3kat.de/title/" >> I2B
    print "#DESCRIPTION: Mapping of ISBN to title URI in B3Kat" >> I2B
    print "#TIMESTAMP:", `date -u +%FT%TZ` >> I2B
    print >> I2B

    print "#FORMAT: BEACON" > B2I
    print "#PREFIX: http://lod.b3kat.de/title/" >> B2I
    print "#RELATION: http://purl.org/ontology/daia/heldBy" >> B2I
    print "#TARGET: http://lobid.org/organisation/" >> B2I
    print "#DESCRIPTION: Mapping of B3Kat title URI to ISIL" >> B2I
    print "#TIMESTAMP:", `date -u +%FT%TZ` >> I2B
    print >> B2I
}

# Subject must look like a title URI 
! $1 ~ /^<http:\/\/lod\.b3kat\.de\/title\// { }

# B3Kat title URI to ISBN
$2 == "<http://purl.org/ontology/bibo/isbn>" {
    title = substr($1,28,length($1)-28)
    isbn  = substr($3, 2, length($3)-2)
    OFS="||"
    print isbn, title >> I2B
}

# B3Kat title URI to item URI
$2 == "<http://purl.org/vocab/frbr/core#exemplar>" &&
$3 ~ /^<http:\/\/lod\.b3kat\.de\/bib\/[A-Z0-9:-]+\/item\// {
    # FIXME: ISIL allows '/' (but not German ISIL)
    title = substr($1,28,length($1)-28)
    split($3,a,"/")
    isil  = a[5]
    OFS="||"
    print title, isil >> B2I
}