migrants/src/map_countries.rb

65 lines
1.8 KiB
Ruby
Executable file

#!/usr/bin/env ruby
require 'sparql/client'
require 'rdf'
require 'pry'
found_one = File.open('found_countries_one.ttl', 'w')
found_zero = File.open('found_countries_zero.ttl', 'w')
found_many = File.open('found_countries_many.ttl', 'w')
wikidata = SPARQL::Client.new('https://query.wikidata.org/sparql')
# Vocabularies
wd = RDF::Vocabulary.new('http://www.wikidata.org/entity/')
wdt = RDF::Vocabulary.new('http://www.wikidata.org/prop/direct/')
rdfs = RDF::Vocabulary.new('http://www.w3.org/2000/01/rdf-schema#')
# binding.pry
# query = wikidata.select.where([:s, :p, :o]).limit(10)
# query.each_solution do |solution|
# p solution
# end
require_relative 'database'
COUNTRIES = {}
DB[:location].each do |location|
unless location[:Country].nil? or COUNTRIES.include? location[:Country]
COUNTRIES[location[:Country]] = {
id: "region:#{location[:country]}"
}
end
end
def country_definition(country_name_literal, wd_solutions)
wd_ids = wd_solutions.map do |solution|
country_uri = solution[:country].to_s
get_wd_name(country_uri)
end
"country:#{toName(country_name_literal.to_s)} a tm:Country ;\n" \
" rdfs:label #{country_name_literal.to_ntriples} ;\n" \
" tm:wikidataID #{wd_ids.join(' , ')} .\n"
end
COUNTRIES.each do |country_name, country_attrs|
p country_name
country_name_literal = RDF::Literal.new(country_name, language: :en)
query = wikidata.select
.where([:country, rdfs.label, country_name_literal])
.where([:country, wdt.P31, wd.Q6256])
solutions = query.solutions
case solutions.size
when 0
found_zero << "#{country_name}\n"
when 1
found_one << country_definition(country_name_literal, solutions)
else
found_many << country_definition(country_name_literal, solutions)
end
sleep(1)
end