jst-csv2nt.rb

#!/usr/bin/env ruby-2.3

=begin
# SPARQL endpoint
https://stirdf.jglobal.jst.go.jp/sparql

# SPARQL query
select *
where {
  ?term a skos:Concept ;
            rdfs:seeAlso ?mesh .
  FILTER strstarts(str(?mesh), "http://id.nlm.nih.gov/mesh/")
  ?term rdfs:label ?label .
  OPTIONAL {
    ?term dc:subject ?subject ;
         dct:subject ?subject_uri .
  }
  OPTIONAL { ?term skos:narrower ?narrower . }
  OPTIONAL { ?term skos:broader ?broader . }
  OPTIONAL { ?term skos:related ?related . }
}

# Download into CSV
OFFSET 0 LIMIT 40000
OFFSET 40000 LIMIT 40000
=end

require 'csv'

header = CSV.parse(ARGF.gets).shift

# ["term", "mesh", "label", "subject", "subject_uri", "narrower", "broader", "related"]

# wrap uri
def u(str)
  "<#{str}>"
end

# quote literal
def q(str)
  %Q("#{str}")
end

# expand uri
def x(predicate)
  pref, term = predicate.split(':')
  case pref
  when 'rdfs'
    "<http://www.w3.org/2000/01/rdf-schema#" + term + ">"
  when 'skos'
    "<http://www.w3.org/2004/02/skos/core#" + term + ">"
  when 'dc'
    "<http://purl.org/dc/elements/1.1/" + term + ">"
  when 'dct'
    "<http://purl.org/dc/terms/" + term + ">"
  end
end

# triple
def t(s, p, o)
  puts [s, p, o, '.'].join("\t")
end

CSV.parse(ARGF.read).each do |row|
  term, mesh, label, subject, subject_uri, narrower, broader, related = *row
  t(u(term), x('rdfs:seeAlso'), u(mesh))
  t(u(term), x('rdfs:label'), q(label))
  t(u(term), x('dc:subject'), q(subject))
  t(u(term), x('dct:subject'), u(subject_uri))
  t(u(term), x('skos:narrower'), u(narrower))
  t(u(term), x('skos:broader'), u(broader))
  t(u(term), x('skos:related'), u(related))
end