Commit c69c6c7cde3b77385451a2c7366f2e4a75688cf3

Authored by Victor Costa
1 parent b1499fb9

virtuoso: noosfero triplification script

plugins/virtuoso/lib/virtuoso_plugin/noosfero_harvest.rb 0 → 100644
... ... @@ -0,0 +1,110 @@
  1 +class VirtuosoPlugin::NoosferoHarvest
  2 +
  3 + COMMON_MAPPING = {
  4 + :type => {:predicate => "http://purl.org/dc/terms/type", :value => lambda {|s, t| t.class.name}},
  5 + :created_at => {:predicate => "http://purl.org/dc/terms/created"},
  6 + :updated_at => {:predicate => "http://purl.org/dc/terms/modified"},
  7 + }
  8 +
  9 + ARTICLE_MAPPING = {
  10 + :title => {:predicate => "http://purl.org/dc/terms/title"},
  11 + :abstract => {:predicate => "http://purl.org/dc/terms/abstract"},
  12 + :body => {:predicate => "http://purl.org/dc/terms/description"},
  13 + :part_of => {:predicate => "http://purl.org/dc/terms/isPartOf", :value => lambda {|s, t| url_for(s.url)} },
  14 + :published_at => {:predicate => "http://purl.org/dc/terms/issued"},
  15 + :author => {:predicate => "http://purl.org/dc/terms/creator", :value => lambda {|s, t| url_for(t.author_url) if t.author_url} },
  16 + }
  17 + PROFILE_MAPPING = {
  18 + :name => {:predicate => "http://purl.org/dc/terms/title"},
  19 + :public? => {:predicate => "http://purl.org/socialparticipation/opa#publicProfile"},
  20 + }
  21 + COMMENT_MAPPING = {
  22 + :title => {:predicate => "http://purl.org/dc/terms/title"},
  23 + :body => {:predicate => "http://purl.org/dc/terms/description"},
  24 + :part_of => {:predicate => "http://purl.org/dc/terms/isPartOf", :value => lambda {|s, t| url_for(s.url)} },
  25 + :author => {:predicate => "http://purl.org/dc/terms/creator", :value => lambda {|s, t| url_for(t.author_url) if t.author_url} },
  26 + }
  27 + FRIENDSHIP_MAPPING = {
  28 + :knows => {:predicate => "http://xmlns.com/foaf/0.1/knows", :value => lambda {|s, t| url_for(t.url)} },
  29 + :type => nil, :created_at => nil, :updated_at => nil,
  30 + }
  31 +
  32 + def initialize(environment)
  33 + @environment = environment
  34 + @graph = environment.top_url
  35 + end
  36 +
  37 + attr_reader :environment
  38 +
  39 + def plugin
  40 + @plugin ||= VirtuosoPlugin.new(self)
  41 + end
  42 +
  43 + delegate :settings, :to => :plugin
  44 +
  45 + include Rails.application.routes.url_helpers
  46 +
  47 + def triplify_comments(article)
  48 + article.comments.each do |comment|
  49 + subject_identifier = url_for(comment.url)
  50 + puts "triplify #{subject_identifier} comment"
  51 + triplify_mappings(COMMENT_MAPPING, subject_identifier, article, comment)
  52 + end
  53 + end
  54 +
  55 + def triplify_articles(profile)
  56 + profile.articles.public.each do |article|
  57 + subject_identifier = url_for(article.url)
  58 + puts "triplify #{subject_identifier} article"
  59 + triplify_mappings(ARTICLE_MAPPING, subject_identifier, profile, article)
  60 + triplify_comments(article)
  61 + end
  62 + end
  63 +
  64 + def triplify_friendship(person)
  65 + person.friends.each do |friend|
  66 + subject_identifier = url_for(person.url)
  67 + puts "triplify #{subject_identifier} friendship"
  68 + triplify_mappings(FRIENDSHIP_MAPPING, subject_identifier, person, friend)
  69 + end
  70 + end
  71 +
  72 + def triplify_profiles
  73 + environment.profiles.each do |profile|
  74 + subject_identifier = url_for(profile.url)
  75 + puts "triplify #{subject_identifier} profile"
  76 + triplify_mappings(PROFILE_MAPPING, subject_identifier, environment, profile)
  77 + triplify_articles(profile) if profile.public?
  78 + triplify_friendship(profile) if profile.person?
  79 + end
  80 + end
  81 +
  82 + def run
  83 + triplify_profiles
  84 + end
  85 +
  86 + protected
  87 +
  88 + def triplify_mappings(mapping, subject_identifier, source, target)
  89 + COMMON_MAPPING.merge(mapping).each do |k, v|
  90 + next unless v
  91 + value = nil
  92 + if v[:value]
  93 + value = v[:value].kind_of?(Proc) ? v[:value].call(source, target) : v[:value]
  94 + elsif target.respond_to?(k)
  95 + value = target.send(k)
  96 + end
  97 + insert_triple(RDF::URI.new(subject_identifier), RDF::URI.new(v[:predicate]), value) if value.present?
  98 + end
  99 + end
  100 +
  101 + def insert_triple(subject, predicate, value)
  102 + value = RDF::URI.new(value) if value.kind_of?(String) && /https?:\/\//.match(value)
  103 + value = RDF::Literal::DateTime.new(value) if value.kind_of?(ActiveSupport::TimeWithZone)
  104 + value = RDF::Literal::Boolean.new(value) if !!value == value
  105 +
  106 + query = RDF::Virtuoso::Query.insert_data([RDF::URI.new(subject), RDF::URI.new(predicate), value]).graph(RDF::URI.new(@graph))
  107 + plugin.virtuoso_client.insert(query)
  108 + end
  109 +
  110 +end
... ...