Convert AP template Excel file to SHACL (RDF/Turtle) format.

template2shacl.rb

#!/usr/bin/env ruby

require "roo"

xlsx = Roo::Spreadsheet.open(ARGV[0])

sheets = xlsx.sheets

puts <<-EOF
@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#>.
@prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>.
@prefix owl: <http://www.w3.org/2002/07/owl#>.
@prefix dct: <http://purl.org/dc/terms/>.
@prefix sh: <http://www.w3.org/ns/shacl#>.
@prefix xsd: <http://www.w3.org/2001/XMLSchema#>.
@prefix ex: <http://example.org/>.

EOF

prev_entity_id = nil
sheets.each do |sheet_name|
  props = []
  xlsx.sheet(sheet_name).parse(headers: true) do |row|
    # cardinality property        description     name    value kind
    #p row
    min, max = nil
    shapes = []
    if row["property"] =~ /:/
      shapes << "sh:path #{row["property"]}"
    else
      shapes << "sh:path ex:#{row["property"]}"
    end
    shapes << "sh:name #{row["name"].inspect}" if row["name"]
    shapes << "sh:description #{row["description"].inspect}" if row["description"]
    case row["value kind"]
    when "string"
      shapes << "sh:datatype xsd:string"
    when "URI"
      shapes << "sh:nodeKind sh:IRI"
    when /^EntityRef:(\w+)$/
      shapes << "sh:class ex:#$1"
    end
    if row[:cardinality]
    min, max, = row[:cardinality].split(/-/)
    min = min.to_i if min
    max = nil if max == "n"
    end
    props << shapes
  end
  puts "ex:#{sheet_name.capitalize}Shape a sh:NodeShape;"
  puts "  sh:property ["
  puts "    " + props.map{|e| e.join(";\n    ") }.join("\n  ];\n  sh:property [\n    ")
  puts  "  ]."
end