migrants/Rakefile

RUBY = RbConfig.ruby

def run(script)
  sh "#{RUBY} #{script}"
end

# ── Individual file tasks ────────────────────────────────────────────────────

file 'data/countries.ttl' => 'src/map_countries_from_location.rb' do
  run 'src/map_countries_from_location.rb'
end

# Generates data/countries_wikidata_review.ttl as a side-effect
file 'data/countries_wikidata.ttl' => ['data/countries.ttl', 'src/link_countries_to_wikidata.rb'] do
  run 'src/link_countries_to_wikidata.rb'
end

file 'data/locations.ttl' => 'src/map_locations.rb' do
  run 'src/map_locations.rb'
end

file 'data/migrations.ttl' => 'src/map_migration_table.rb' do
  run 'src/map_migration_table.rb'
end

file 'data/occupations.ttl' => 'src/map_occupations_from_person_profession.rb' do
  run 'src/map_occupations_from_person_profession.rb'
end

file 'data/organisations.ttl' => 'src/map_organisation.rb' do
  run 'src/map_organisation.rb'
end

file 'data/personnames.ttl' => 'src/map_personnames.rb' do
  run 'src/map_personnames.rb'
end

# Generates data/persons_occupations_review.ttl as a side-effect
file 'data/persons_occupations.ttl' => ['data/occupations.ttl', 'src/map_person_profession.rb'] do
  run 'src/map_person_profession.rb'
end

file 'data/persons.ttl' => 'src/map_person.rb' do
  run 'src/map_person.rb'
end

file 'data/relationships.ttl' => 'src/map_relationship.rb' do
  run 'src/map_relationship.rb'
end

file 'data/persons_religions.ttl' => 'src/map_religions.rb' do
  run 'src/map_religions.rb'
end

# Generates data/religions_wikidata_review.ttl as a side-effect
file 'data/religions_wikidata.ttl' => ['data/religions.ttl', 'src/link_religions_to_wikidata.rb'] do
  run 'src/link_religions_to_wikidata.rb'
end

file 'data/works.ttl' => 'src/map_work.rb' do
  run 'src/map_work.rb'
end

# ── Pipeline steps ───────────────────────────────────────────────────────────

file 'data/graph-01.ttl' => 'map/step-01.rb' do
  run 'map/step-01.rb'
end

UPDATE_QUERIES = FileList['updates/*.rq']

file 'data/graph-02.ttl' => ['data/graph-01.ttl', 'map/step-02.rb'] + UPDATE_QUERIES do
  run 'map/step-02.rb'
end

UPDATE_QUERIES_STEP03 = FileList['updates_step03/*.rq']

file 'data/graph-03.ttl' => ['data/graph-02.ttl'] + UPDATE_QUERIES_STEP03 do
  sh 'step-03'
end

UPDATE_QUERIES_STEP04 = FileList['updates_step04/*.rq']

file 'data/graph-04.ttl' => ['data/graph-03.ttl'] + UPDATE_QUERIES_STEP04 do
  sh 'step-04'
end

UPDATE_QUERIES_STEP05 = FileList['updates_step05/*.rq']

file 'data/graph-05.ttl' => ['data/graph-04.ttl'] + UPDATE_QUERIES_STEP05 do
  sh 'step-05'
end

# ── Examples ─────────────────────────────────────────────────────────────────

SPARQL = File.expand_path('~/.cargo/bin/sparql')

file 'data_examples/step_01_1hop.ttl' => ['data/graph-01.ttl', 'queries/step_01_1hop_example.rq'] do
  sh "#{SPARQL} queries/step_01_1hop_example.rq --graph data/graph-01.ttl --prettify > data_examples/step_01_1hop.ttl"
end

file 'data_examples/step_01_2hop.ttl' => ['data/graph-02.ttl', 'queries/step_01_2hop_example.rq'] do
  sh "#{SPARQL} queries/step_01_2hop_example.rq --graph data/graph-02.ttl --prettify > data_examples/step_01_2hop.ttl"
end

file 'data_examples/step_02_1hop.ttl' => ['data/graph-02.ttl', 'queries/step_02_1hop_example.rq'] do
  sh "#{SPARQL} queries/step_02_1hop_example.rq --graph data/graph-02.ttl --prettify > data_examples/step_02_1hop.ttl"
end

file 'data_examples/step_02_2hop.ttl' => ['data/graph-02.ttl', 'queries/step_02_2hop_example.rq'] do
  sh "#{SPARQL} queries/step_02_2hop_example.rq --graph data/graph-02.ttl --prettify > data_examples/step_02_2hop.ttl"
end

file 'data_examples/step_03_1hop.ttl' => ['data/graph-03.ttl', 'queries/step_03_1hop_example.rq'] do
  sh "#{SPARQL} queries/step_03_1hop_example.rq --graph data/graph-03.ttl --prettify > data_examples/step_03_1hop.ttl"
end

file 'data_examples/step_03_2hop.ttl' => ['data/graph-03.ttl', 'queries/step_03_2hop_example.rq'] do
  sh "#{SPARQL} queries/step_03_2hop_example.rq --graph data/graph-03.ttl --prettify > data_examples/step_03_2hop.ttl"
end

file 'data_examples/step_04_1hop.ttl' => ['data/graph-04.ttl', 'queries/step_04_1hop_example.rq'] do
  sh "#{SPARQL} queries/step_04_1hop_example.rq --graph data/graph-04.ttl --prettify > data_examples/step_04_1hop.ttl"
end

file 'data_examples/step_04_2hop.ttl' => ['data/graph-04.ttl', 'queries/step_04_2hop_example.rq'] do
  sh "#{SPARQL} queries/step_04_2hop_example.rq --graph data/graph-04.ttl --prettify > data_examples/step_04_2hop.ttl"
end

file 'data_examples/step_05_1hop.ttl' => ['data/graph-05.ttl', 'queries/step_05_1hop_example.rq'] do
  sh "#{SPARQL} queries/step_05_1hop_example.rq --graph data/graph-05.ttl --prettify > data_examples/step_05_1hop.ttl"
end

file 'data_examples/step_05_2hop.ttl' => ['data/graph-05.ttl', 'queries/step_05_2hop_example.rq'] do
  sh "#{SPARQL} queries/step_05_2hop_example.rq --graph data/graph-05.ttl --prettify > data_examples/step_05_2hop.ttl"
end

# ── Aggregate tasks ──────────────────────────────────────────────────────────

GENERATED = %w[
  data/countries.ttl
  data/countries_wikidata.ttl
  data/locations.ttl
  data/migrations.ttl
  data/occupations.ttl
  data/organisations.ttl
  data/personnames.ttl
  data/persons_occupations.ttl
  data/persons.ttl
  data/relationships.ttl
  data/persons_religions.ttl
  data/religions_wikidata.ttl
  data/works.ttl
].freeze

GRAPHS = %w[
  data/graph-01.ttl
  data/graph-02.ttl
  data/graph-03.ttl
  data/graph-04.ttl
  data/graph-05.ttl
].freeze

EXAMPLES = %w[
  data_examples/step_01_1hop.ttl
  data_examples/step_01_2hop.ttl
  data_examples/step_02_1hop.ttl
  data_examples/step_02_2hop.ttl
  data_examples/step_03_1hop.ttl
  data_examples/step_03_2hop.ttl
  data_examples/step_04_1hop.ttl
  data_examples/step_04_2hop.ttl
  data_examples/step_05_1hop.ttl
  data_examples/step_05_2hop.ttl
].freeze

task examples: EXAMPLES

task default: GENERATED + GRAPHS + EXAMPLES

task :clean do
  review_files = %w[
    data/countries_wikidata_review.ttl
    data/religions_wikidata_review.ttl
  ]
  rm_f GENERATED + GRAPHS + EXAMPLES + review_files
end