save results to yaml

consultation_ipred/answer.rb

+class Answer < Struct.new(:name, :files, :annexes, :languages)
+  def initialize(*args)
+    super
+    self.annexes ||= []
+    self.files ||= []
+    self.languages ||= []
+  end
+end
+

consultation_ipred/parser.rb

@@ -3,54 +3,53 @@ require "rubygems"
 require "bundler/setup"
 require 'cgi'
 require 'iconv'
-
-files = %w(organisations.links public_authorities.links)
-
-file = files.first
-
-links = File.read(file).split("\n")
-
-class Answer < Struct.new(:name, :files, :annexes, :languages)
-  def initialize(*args)
-    super
-    self.annexes ||= []
-    self.files ||= []
-    self.languages ||= []
+require 'answer'
+require 'yaml'
+
+files = %w(organisations public_authorities)
+
+files.each do |type|
+  file = type + ".links"
+
+  links = File.read(file).split("\n")
+
+  answers = {}
+
+  links.each do |url|
+    name = File.basename(URI.parse(url).path, ".pdf")
+    name = CGI::unescape(name)
+    #name = Iconv.iconv("utf-8", "iso8859-15", name).first
+    names = name.split("_")
+    language = names.pop
+    
+    annex = names.index { |part| part =~ /^annex/ }
+    if annex
+      names[annex..-1] = []
+    end
+    
+    name = names.map(&:capitalize).join(" ")
+    #name = "#{name} (#{language})"
+    
+    answer = answers[name] ||= Answer.new(name)
+    if annex
+      answer.annexes << url
+    else
+      answer.files << url
+    end
+    answer.languages << language
+    answer.languages.uniq!
   end
-end
 
-answers = {}
-
-links.each do |url|
-  name = File.basename(URI.parse(url).path, ".pdf")
-  name = CGI::unescape(name)
-  #name = Iconv.iconv("utf-8", "iso8859-15", name).first
-  names = name.split("_")
-  language = names.pop
-  
-  annex = names.index { |part| part =~ /^annex/ }
-  if annex
-    names[annex..-1] = []
-  end
-  
-  name = names.map(&:capitalize).join(" ")
-  #name = "#{name} (#{language})"
-  
-  answer = answers[name] ||= Answer.new(name)
-  if annex
-    answer.annexes << url
-  else
-    answer.files << url
+  answers.sort.each do |name, answer|
+    puts "#{answer.name} (#{answer.languages.join(",")})"
+    (answer.files + answer.annexes).each do |url|
+      # p answer
+      name = File.basename(URI.parse(url).path)
+      puts "  " + name
+    end
   end
-  answer.languages << language
-  answer.languages.uniq!
-end
 
-answers.sort.each do |name, answer|
-  puts "#{answer.name} (#{answer.languages.join(",")})"
-  (answer.files + answer.annexes).each do |url|
-    # p answer
-    name = File.basename(URI.parse(url).path)
-    puts "  " + name
+  File.open(type + ".yaml", "w") do |f|
+    f.puts answers.to_yaml
   end
 end

consultation_ipred/public_authorities.yaml

+--- 
+Uk: !ruby/struct:Answer 
+  name: Uk
+  files: 
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/6436e9a1-1ba6-414e-af8c-d50fa3ac2c52/uk_governement.pdf
+  annexes: []
+
+  languages: 
+  - governement
+"": !ruby/struct:Answer 
+  name: ""
+  files: 
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/4edaf82a-d143-4a07-9880-e59529afc595/france.pdf
+  annexes: []
+
+  languages: 
+  - france
+Portugal: !ruby/struct:Answer 
+  name: Portugal
+  files: 
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/fcf00a0f-7d50-4423-8101-2deb638cd16e/portugal_pt.pdf
+  annexes: []
+
+  languages: 
+  - pt
+Spain Ministry Of Justice: !ruby/struct:Answer 
+  name: Spain Ministry Of Justice
+  files: 
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/8c1ac475-47f7-486f-adc6-d68fddd7a141/spain_ministry_of_justice_es.pdf
+  annexes: []
+
+  languages: 
+  - es
+Italy Ministry Of Agrifood: !ruby/struct:Answer 
+  name: Italy Ministry Of Agrifood
+  files: 
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/0c470e92-6585-442f-9d26-e76046744085/italy_ministry_of_agrifood_en.pdf
+  annexes: []
+
+  languages: 
+  - en
+Hungary Ministry Public  administration And Justice: !ruby/struct:Answer 
+  name: Hungary Ministry Public  administration And Justice
+  files: 
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/7668b8f8-3080-4de2-8a90-62c7d6be0e60/hungary_ministry_public_%20administration_and_justice_en.pdf
+  annexes: []
+
+  languages: 
+  - en
+Finland Ministry Of Empl And Economy: !ruby/struct:Answer 
+  name: Finland Ministry Of Empl And Economy
+  files: 
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/cce241c1-0810-4d19-b288-5c5f6d8e26d4/finland_ministry_of_empl_and_economy_en.pdf
+  annexes: []
+
+  languages: 
+  - en
+Romania: !ruby/struct:Answer 
+  name: Romania
+  files: 
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/919f019f-bb07-443d-801c-f6db3a668134/romania_en.pdf
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/046d1325-6cc6-4f8e-aee5-25e354b60f69/romania_ro.pdf
+  annexes: []
+
+  languages: 
+  - en
+  - ro
+Bulgaria: !ruby/struct:Answer 
+  name: Bulgaria
+  files: 
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/3346b4d7-2171-428d-a9af-fe31a3dd67c9/bulgaria_en.pdf
+  annexes: []
+
+  languages: 
+  - en
+Ak Oesterreich: !ruby/struct:Answer 
+  name: Ak Oesterreich
+  files: 
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/39852257-57a4-4002-a6df-8b799661c3ea/ak_oesterreich_de.pdf
+  annexes: []
+
+  languages: 
+  - de
+Lithuania Ministry Culture: !ruby/struct:Answer 
+  name: Lithuania Ministry Culture
+  files: 
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/d941e7ac-0ad0-4934-b05d-711f816b95e7/lithuania_ministry_culture_lt.pdf
+  annexes: []
+
+  languages: 
+  - lt
+Wko Austria: !ruby/struct:Answer 
+  name: Wko Austria
+  files: 
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/df9bf458-a928-43c2-931e-fe722ca1e317/wko_austria_de.pdf
+  annexes: []
+
+  languages: 
+  - de
+Italy Office Of The Prime Minister: !ruby/struct:Answer 
+  name: Italy Office Of The Prime Minister
+  files: 
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/71f09778-f7f6-4b49-af15-3dd02d03c42e/italy_office_of_the_prime_minister_it.pdf
+  annexes: []
+
+  languages: 
+  - it
+Finnish Commerce Federation: !ruby/struct:Answer 
+  name: Finnish Commerce Federation
+  files: 
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/13b94e78-5719-4892-af14-469f47d8cf2d/finnish_commerce_federation_fi.pdf
+  annexes: []
+
+  languages: 
+  - fi
+Belgium: !ruby/struct:Answer 
+  name: Belgium
+  files: 
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/32fe9217-6c5a-4b3b-a954-88e491337f7f/belgium_en.pdf
+  annexes: []
+
+  languages: 
+  - en
+Slovakia: !ruby/struct:Answer 
+  name: Slovakia
+  files: 
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/05142029-2e42-43d6-b4b8-0bb42d5e836a/slovakia_sk.pdf
+  annexes: []
+
+  languages: 
+  - sk
+Latvia: !ruby/struct:Answer 
+  name: Latvia
+  files: 
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/0f084a74-d6ca-41e2-afe5-97b74f539ea3/latvia_en.pdf
+  annexes: []
+
+  languages: 
+  - en
+Danish Chamber Of Commerce: !ruby/struct:Answer 
+  name: Danish Chamber Of Commerce
+  files: 
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/295fb523-cbec-4f7f-b9f1-55ec2715652b/danish_chamber_of_commerce_en.pdf
+  annexes: []
+
+  languages: 
+  - en
+Poland Ministry Culture: !ruby/struct:Answer 
+  name: Poland Ministry Culture
+  files: 
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/e4d28106-09a7-4989-8aad-cfbe36fdfc02/poland_ministry_culture_en.pdf
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/b32505cb-63b5-4828-a4dd-81ba03fe8a43/poland_ministry_culture_pl.pdf
+  annexes: []
+
+  languages: 
+  - en
+  - pl
+Parti Pirate: !ruby/struct:Answer 
+  name: Parti Pirate
+  files: 
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/54f046e5-c3b3-4b75-8394-0f84d204ae97/parti_pirate_fr.pdf
+  annexes: []
+
+  languages: 
+  - fr
+Malta: !ruby/struct:Answer 
+  name: Malta
+  files: 
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/c8884ac2-d625-44e5-a3a7-7e7b98b111ac/malta_en.pdf
+  annexes: []
+
+  languages: 
+  - en
+Denmark: !ruby/struct:Answer 
+  name: Denmark
+  files: 
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/f37efa48-d5c4-4b22-8d87-893d50170a70/denmark_en.pdf
+  annexes: []
+
+  languages: 
+  - en
+Netherlands Ministry Security Justice: !ruby/struct:Answer 
+  name: Netherlands Ministry Security Justice
+  files: 
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/11bc6214-9b28-4d89-a1f6-58bf0a4e1abd/netherlands_ministry_security_justice_nl.pdf
+  annexes: 
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/607ebe73-a7ae-489e-8f14-2dff48521320/netherlands_ministry_security_justice_annex1_nl.pdf
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/36934912-7ec6-4f14-aa80-e8780fa56cb0/netherlands_ministry_security_justice_annex2_nl.pdf
+  languages: 
+  - nl
+Ireland: !ruby/struct:Answer 
+  name: Ireland
+  files: 
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/ddc21a57-b983-4ede-a02f-59dd3f2bc8a6/ireland_en.pdf
+  annexes: []
+
+  languages: 
+  - en
+Hadopi: !ruby/struct:Answer 
+  name: Hadopi
+  files: 
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/c2cdc614-5073-4306-8293-2c40c9fa84da/hadopi_fr.pdf
+  annexes: []
+
+  languages: 
+  - fr
+Czech Republic: !ruby/struct:Answer 
+  name: Czech Republic
+  files: 
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/2c8a3544-c150-4fc0-941b-6496375682b8/czech_republic_en.pdf
+  annexes: []
+
+  languages: 
+  - en
+Germany: !ruby/struct:Answer 
+  name: Germany
+  files: 
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/0ef00dda-f4be-4950-bfad-f48baf11cb0c/germany_de.pdf
+  annexes: []
+
+  languages: 
+  - de
+European Parliament Committee On Legal Affairs: !ruby/struct:Answer 
+  name: European Parliament Committee On Legal Affairs
+  files: 
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/fa293e2c-cfae-43ed-abcb-59ac2d31a4fb/european_parliament_committee_on_legal_affairs_en.pdf
+  annexes: []
+
+  languages: 
+  - en
+Deutscher Bundestag: !ruby/struct:Answer 
+  name: Deutscher Bundestag
+  files: 
+  - https://circabc.europa.eu/d/d/workspace/SpacesStore/82d024ea-ce88-4974-9122-2bf368f91906/deutscher_bundestag_de.pdf
+  annexes: []
+
+  languages: 
+  - de