krikri 0.5.4 → 0.5.5

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 9b2d8957a0e4e4129ae3a818d8bbd972a14bc591
-  data.tar.gz: 79217f0cc7464b0cec61013d7b0fc10cb6c53f99
+  metadata.gz: f571b90beb068972a81dbbe4488941c52ff650a4
+  data.tar.gz: f660257e1db2d1f5eb21419963db13cfecf4cfa3
 SHA512:
-  metadata.gz: 788f41ca87a996bc7ee6bfcec0b02f298020f73a0075df7154479b4cfc3ce065d45c38bc957d8e9c99bd10de4a1a344d512a4dae17f35cc6c3f12063d821a264
-  data.tar.gz: 6d25298019ee75a91147d6c6f2f973311bd46e9db7c0a967486515c1b95490c90c357d0d39aa7b1b8df97778d1172b8e469e41e4bb176f7c560fa7120a34dd88
+  metadata.gz: d328c8375ab9596ff0fc9110a1808bd83779b531d69d19e0b9628dd50b4e5bc429e5ae9b6ea2ac019e8c4473fd70f2ce0b1c6285ca5827a2536c1b69af7c7007
+  data.tar.gz: 75f8dccf0cfa12d5b5e7233924264f07e93638f8377b14bbb82ee4af4ddb9c03a8f7c752e5d01ed3a20973db535c0c2bcc87d30efa6002f97fa368dc45244eb1

data/app/controllers/krikri/_qa_report_list.html.erb~

@@ -0,0 +1,10 @@
+<h3>Validation Reports</h3>
+<% if @validation_reports %>
+  <ul>
+    <% @validation_reports.each do |validation_report| %>
+      <li><%= render_validation_report(validation_report) %></li>
+    <% end %>
+  </ul>
+<% else %>
+  <div>There was an error generating the QA reports.</div>
+<% end %>

data/app/controllers/krikri/concerns/provider_context.rb~

@@ -0,0 +1,7 @@
+module Krikri::Concerns
+  class ProviderContext
+    def set_current_provider(provider_id = params[:provider])
+      @current_provider = Krikri::Provider.find(provider_id)
+    end
+  end
+end

data/app/controllers/krikri/qa_reports_controller.rb~

@@ -0,0 +1,18 @@
+module Krikri
+  ##
+  # Handles HTTP requests for QA Reports
+  class QAReportsController < ApplicationController
+    include Concerns::ProviderContext
+
+    before_action :authenticate_user!, :set_current_provider
+
+    def show
+      require 'pry'
+      binding.pry
+      respond_to do |format|
+        format.html
+        format.csv { render csv: 'test'  }
+      end
+    end
+  end
+end

data/app/models/krikri/harvesters/oai_harvester.rb~

@@ -0,0 +1,10 @@
+module Krikri::Harvesters
+  class OAIHarvester < Krikri::Harvester
+
+    attr_accessor :client
+
+    def initialize(opts={})
+      @client = OAI::Client.new(opts[:endpoint])
+    end
+  end
+end

data/app/models/krikri/original_record_desc.rb~

@@ -0,0 +1,6 @@
+module Krikri
+  ##
+  # Descriptive Metadata for OriginalRecord
+  class OriginalRecordMetadata < ActiveTriples::Resource
+  end
+end

data/app/models/krikri/original_record_metadata.rb~

@@ -0,0 +1,9 @@
+module Krikri
+  ##
+  # Descriptive Metadata for OriginalRecord
+  class OriginalRecordMetadata < ActiveTriples::Resource
+    include Krikri::Ldp::RdfSource
+
+    property :wasGeneratedBy, predicate: RDF::PROV.wasGeneratedBy
+  end
+end

data/app/models/krikri/parser.rb~

@@ -0,0 +1,5 @@
+module Krikri
+  class Parser
+    
+  end
+end

data/app/models/krikri/parsers/xml_parser.rb~

@@ -0,0 +1,5 @@
+module Krikri::Parsers
+
+  class XmlParser
+  end
+end

data/app/models/krikri/qa_report.rb~

@@ -0,0 +1,7 @@
+module Krikri
+  class QAReport
+    class << self
+      def all
+    end
+  end
+end

data/app/views/krikri/qa_reports/index.html.erb~

@@ -0,0 +1,3 @@
+<% # krikri/qa_records/index %>
+
+<% content_for :title do %>Institutions<% end %>

data/app/views/krikri/qa_reports/show.html.erb~

@@ -0,0 +1 @@
+<% # krikri/institutions/show %>

data/config/initializers/config.rb~

@@ -0,0 +1,19 @@
+setting_files = [self.root.join('config', 'settings.yml'),
+                 self.root.join('config', 'settings', "#{Rails.env}.yml"),
+                 self.root.join('config', 'environments', "#{Rails.env}.yml"),
+                 self.root.join('config', 'settings.local.yml'),
+                 self.root.join('config', 'settings', "#{Rails.env}.local.yml"),
+                 self.root.join('config', 'environments', 
+                                "#{Rails.env}.local.yml")
+                ].map { |path| path.to_s }
+
+settings_const = Kernel.const_get(RailsConfig.const_name)
+
+source_paths = settings_const.add_source!('nil')[0..-2].map do |source| 
+  source.path 
+end
+
+source_paths = setting_files + source_paths
+
+RailsConfig.load_and_set_settings(source_paths)
+Krikri::Settings = Kernel.const_get(RailsConfig.const_name)

data/config/settings/test.yml~

@@ -0,0 +1,3 @@
+---
+marmotta:
+  base: 'http://localhost:8080/marmotta/'

data/db/migrate/20150328035550_add_qa_report.rb~

@@ -0,0 +1,11 @@
+class AddQaReport < ActiveRecord::Migration
+  def change
+    create_table :qa_report do |t|
+      t.text :field_report, limit: nil
+      t.text :count_report, limit: nil
+      t.string :provider, primary_key: true
+
+      t.timestamps
+    end
+  end
+end

data/lib/aggregation.rb~

@@ -0,0 +1,4 @@
+class DPLA::MAP::Aggregation
+  include Krikri::LDP::RdfSource
+  configure :base_uri => Krikri::Settings['marmotta']['item_container']
+end

data/lib/blacklight.rb~

@@ -0,0 +1,8 @@
+module Blacklight
+  def self.connection_config
+    @connection_config ||= begin
+      raise "The #{::Rails.env} environment settings were not found in the blacklight.yml config" unless blacklight_yml[::Rails.env]
+      blacklight_yml[::Rails.env].symbolize_keys
+    end
+  end
+end

data/lib/krikri/engine.rb

@@ -1,3 +1,4 @@
+require 'rdf/isomorphic'
 require 'active_support'
 require 'rails_config'
 require 'krikri/ldp'
@@ -8,6 +9,7 @@ require 'rdf/marmotta'
 require 'oai/client'
 require 'rest-client'
 require 'edtf'
+require 'text'
 require 'csv'
 
 require 'resque'

data/lib/krikri/enrichment.rb~

@@ -0,0 +1,5 @@
+module Krikri
+  class Enrichment
+    extend SoftwareAgent
+  end
+end

data/lib/krikri/enrichment_agent.rb~

@@ -0,0 +1,11 @@
+module Krikri
+  ##
+  # EnrichmentAgent
+  class EnrichmentAgent
+    include SoftwareAgent
+
+    def initialize(opts = {})
+      @generator_uri = RDF::URI(opts.fetch(:generator_uri))
+    end
+  end
+end

data/lib/krikri/enrichment_agent_spec.rb~

@@ -0,0 +1,2 @@
+describe Krikri::Enrichment::Agent do
+  it_behaves_like 'a software agent', opts

data/lib/krikri/enrichment_job.rb~

@@ -0,0 +1,13 @@
+module Krikri
+  ##
+  # MappingJob: perform enqueued Krikri::Mapper::Agent
+  #
+  # @see: Krikri::Mapper, Krikri::Mapper::Agent
+  class EnrichmentJob < Krikri::Job
+    @queue = :enrichment
+
+    def self.run(mapper, activity_uri = nil)
+      mapper.run(activity_uri)
+    end
+  end
+end

data/lib/krikri/enrichments.rb~

@@ -0,0 +1,4 @@
+module Krikri
+  module Enrichments
+  end
+end

data/lib/krikri/enrichments/contvert_to_sentence_case.rb~

@@ -0,0 +1,13 @@
+module Krikri::Enrichments
+  ##
+  #
+  class ConvertToSentenceCase
+    include Krikri::FieldEnrichment
+
+    def enrich_value(value)
+      return value unless value.is_a? String
+      require 'pry'
+      binding.pry
+    end
+  end
+end

data/lib/krikri/enrichments/convert_to_sentence_case.rb

@@ -0,0 +1,20 @@
+module Krikri::Enrichments
+  ##
+  # Converts a string to sentence case.
+  #
+  # @example
+  #
+  #   string = 'this is a sentence about Moomins. this is another about Snorks.'
+  #   Krikri::Enrichments::ConvertToSentenceCase.enrich_value(string)
+  #   # => 'This is a sentence about moomins. This is another about snorks.'
+  class ConvertToSentenceCase
+    include Krikri::FieldEnrichment
+
+    def enrich_value(value)
+      return value unless value.is_a? String
+      value.gsub(/([a-z])((?:[^.?!]|\.(?=[a-z]))*)/i) do
+        $1.upcase + $2.downcase.rstrip
+      end
+    end
+  end
+end

data/lib/krikri/enrichments/convert_to_sentence_case.rb~

@@ -0,0 +1,13 @@
+module Krikri::Enrichments
+  ##
+  #
+  class ConvertToSentenceCase
+    include Krikri::FieldEnrichment
+
+    def enrich_value(value)
+      return value unless value.is_a? String
+      require 'pry'
+      binding.pry
+    end
+  end
+end

data/lib/krikri/enrichments/dcmi_enforcer.rb

@@ -0,0 +1,17 @@
+module Krikri::Enrichments
+  ##
+  # Removes non-DCMI Type values from a field
+  class DcmiEnforcer
+    include Krikri::FieldEnrichment
+
+    ##
+    # @param value [Object] the value to enrich
+    #
+    # @return [DPLA::MAP::Controlled::DCMIType, nil] the original value or `nil`
+    def enrich_value(value)
+      return nil unless value.is_a? DPLA::MAP::Controlled::DCMIType
+      return nil unless value.valid?
+      value
+    end
+  end
+end

data/lib/krikri/enrichments/dcmi_enforcer.rb~

@@ -0,0 +1,5 @@
+module Krikri::Enrichments
+  class DcmiEnforcer
+
+  end
+end

data/lib/krikri/enrichments/dcmi_type_map.rb

@@ -0,0 +1,128 @@
+# coding: utf-8
+module Krikri::Enrichments
+  ##
+  # Maps string values to DCMI Type vocabulary terms.
+  #
+  # Mapping is performed by comparing a hash of string keys to the pre-enriched
+  # value, using White Similarity comparisons to determine the closest match. If
+  # a suitable match is found, a {DLPA::MAP::Controlled::DCMIType} object is
+  # built from the appropriate hash value and set as the new value.
+  #
+  # This enrichment ignores strings without a DCMI Type match.
+  #
+  # @example
+  #
+  #   type_mapper = DcmiTypeMap.new
+  #   type_mapper.enrich_value('image') # finds RDF::DCMITYPE.Image
+  #   type_mapper.enrich_value('a book') # finds RDF::DCMITYPE.Text
+  #   type_mapper.enrich_value('a really cool book') # => 'a really cool book'
+  #
+  # @example
+  #
+  #   type_mapper = DcmiTypeMap.new('poloriod' => RDF::DCMIType.Image)
+  #   type_mapper.enrich_value('poloroid.') # finds RDF::DCMITYPE.Image
+  #
+  class DcmiTypeMap
+    include Krikri::FieldEnrichment
+
+    DEFAULT_MAP = { 'image' => RDF::DCMITYPE.Image,
+                    'photograph' => RDF::DCMITYPE.Image,
+                    'sample book' => RDF::DCMITYPE.Image,
+                    'specimen' => RDF::DCMITYPE.Image,
+                    'textile' => RDF::DCMITYPE.Image,
+                    'frame' => RDF::DCMITYPE.Image,
+                    'costume' => RDF::DCMITYPE.Image,
+                    'statue' => RDF::DCMITYPE.Image,
+                    'sculpture' => RDF::DCMITYPE.Image,
+                    'container' => RDF::DCMITYPE.Image,
+                    'jewelry' => RDF::DCMITYPE.Image,
+                    'furnishing' => RDF::DCMITYPE.Image,
+                    'furniture' => RDF::DCMITYPE.Image,
+                    'drawing' => RDF::DCMITYPE.Image,
+                    'print' => RDF::DCMITYPE.Image,
+                    'painting' => RDF::DCMITYPE.Image,
+                    'illumination' => RDF::DCMITYPE.Image,
+                    'poster' => RDF::DCMITYPE.Image,
+                    'appliance' => RDF::DCMITYPE.Image,
+                    'tool' => RDF::DCMITYPE.Image,
+                    'electronic component' => RDF::DCMITYPE.Image,
+                    'postcard' => RDF::DCMITYPE.Image,
+                    'equipment' => RDF::DCMITYPE.Image,
+                    'cartographic' => RDF::DCMITYPE.Image,
+                    'notated music' => RDF::DCMITYPE.Image,
+                    'mixed material' => RDF::DCMITYPE.Image,
+                    'text' => RDF::DCMITYPE.Text,
+                    'book' => RDF::DCMITYPE.Text,
+                    'publication' => RDF::DCMITYPE.Text,
+                    'magazine' => RDF::DCMITYPE.Text,
+                    'journal' => RDF::DCMITYPE.Text,
+                    'correspondence' => RDF::DCMITYPE.Text,
+                    'writing' => RDF::DCMITYPE.Text,
+                    'written' => RDF::DCMITYPE.Text,
+                    'manuscript' => RDF::DCMITYPE.Text,
+                    'audio' => RDF::DCMITYPE.Sound,
+                    'sound' => RDF::DCMITYPE.Sound,
+                    'oral history recording' => RDF::DCMITYPE.Sound,
+                    'finding aid' => RDF::DCMITYPE.Collection,
+                    'online collection' => RDF::DCMITYPE.Collection,
+                    'electronic resource' => RDF::DCMITYPE.InteractiveResource,
+                    'video game' => RDF::DCMITYPE.InteractiveResource,
+                    'online exhibit' => RDF::DCMITYPE.InteractiveResource,
+                    'moving image' => RDF::DCMITYPE.MovingImage,
+                    'movingimage' => RDF::DCMITYPE.MovingImage,
+                    'motion picture' => RDF::DCMITYPE.MovingImage,
+                    'film' => RDF::DCMITYPE.MovingImage,
+                    'video' => RDF::DCMITYPE.MovingImage,
+                    'object' => RDF::DCMITYPE.PhysicalObject
+                  }
+
+    ##
+    # @param map [Hash<String, RDF::Vocabulary::Term>]
+    def initialize(map = nil)
+      @map = map || DEFAULT_MAP
+    end
+
+    ##
+    # @param value [Object] the value to enrich
+    #
+    # @return [DPLA::MAP::Controlled::DCMIType, nil] the matching DCMI Type
+    #   term. `nil` if no matches are found.
+    def enrich_value(value)
+      return value unless value.is_a? String
+
+      match = @map.fetch(value.downcase) { most_similar(value) }
+      return value if match.nil?
+      dcmi = DPLA::MAP::Controlled::DCMIType.new(match)
+      dcmi.prefLabel = match.label
+      dcmi
+    end
+
+
+    private
+
+    ##
+    # Performs White Similarity comparison against the keys, and gives the
+    # value of the closest match.
+    #
+    # @param value [String] a string value to compare to the hash map keys.
+    # @param threshold [Float] the value at which a string is considered to
+    #   be a match
+    #
+    # @return [RDF::Vocabulary::Term, nil] the closest DCMI type match, or `nil`
+    #   if none is sufficiently close
+    #
+    # @see Text::WhiteSimilarity
+    # @see http://www.catalysoft.com/articles/strikeamatch.html article defining
+    #   the White Similarity algorithm
+    #
+    # @todo consider text similarity algorithms/strategies and move text
+    #   matching to a utility and behind a Facade interface.
+    def most_similar(value, threshold = 0.5)
+      @white ||= Text::WhiteSimilarity.new
+      result = @map.max_by { |str, _| @white.similarity(value, str) }
+
+      return result[1] if @white.similarity(value, result.first) > threshold
+      nil
+    end
+  end
+end

data/lib/krikri/enrichments/dcmi_type_map.rb~

@@ -0,0 +1,8 @@
+module Krikri::Enrichments
+  ##
+  #
+  #
+  class DcmiTypeMap
+    include Krikri::FieldEnrichment
+  end
+end

data/lib/krikri/enrichments/dedup_nodes.rb

@@ -0,0 +1,72 @@
+module Krikri::Enrichments
+  ##
+  # Enrichment to remove duplicate blank nodes from an
+  # {ActiveTriples::Resource}, where "duplicate" means having the same
+  # `providedLabel`. URIs and Literal values are retained.
+  #
+  # @example
+  #   # given a SourceResource
+  #   sourceResource.creator
+  #   # => [#<DPLA::MAP::Agent:0x3fa8828e08e0(default)>,
+  #         #<DPLA::MAP::Agent:0x3fa882910220(default)>,
+  #         #<DPLA::MAP::Agent:0x3fa882942ce8(default)>]
+  #
+  #   sourceResource.creator.map(&:rdf_subject)
+  #   # => [#<RDF::Node:0x3fa8828e0674(_:g69992977401460)>,
+  #         #<RDF::Node:0x3fa882913d08(_:g69992977612040)>,
+  #         #<RDF::URI:0x3fa882942748 URI:http://example.org/moomin>]
+  #   sourceResource.creator.map(&:providedLabel)
+  #   # => [["moomin"], ["moomin"], ["moomin"]]
+  #
+  #   enrich = Krikri::Enrichments::DedupNodes.new
+  #   new_sr = enrich.enrich_value(sourceResource)
+  #
+  #   new_sr.creator
+  #   # => [#<DPLA::MAP::Agent:0x3fa8828e08e0(default)>,
+  #         #<DPLA::MAP::Agent:0x3fa882942ce8(default)>]
+  #
+  #   sourceResource.creator.map(&:rdf_subject)
+  #   # => [#<RDF::Node:0x3fa8828e0674(_:g69992977401460)>,
+  #         #<RDF::URI:0x3fa882942748 URI:http://example.org/moomin>]
+  #
+  class DedupNodes
+    include Krikri::FieldEnrichment
+
+    ##
+    # @param value [Object]
+    #
+    # @return [Object] the original value altered to remove nodes with the same
+    #   `providedLabel`, if any
+    def enrich_value(value)
+      return value unless value.is_a? ActiveTriples::Resource
+      deduplicate_resource(value)
+      value
+    end
+
+    private
+
+    ##
+    # @param value [ActiveTriples::Resource]
+    #
+    # @return [ActiveTriples::Resource] returns the node after running `#uniq`
+    #   against the provided labels of an nodes.
+    def deduplicate_resource(value)
+      value.class.properties.values.map(&:term).map do |property|
+        unique = value.send(property).uniq { |v| providedLabel_or_value(v) }
+        value.send("#{property}=".to_sym, unique)
+      end
+    end
+
+    ##
+    # @param value [Object]
+    #
+    # @return [Object] if `value` is an RDF::Node, the first result of its
+    #   `#providedLabel`, if any; otherwise the original `value`.
+    def providedLabel_or_value(value)
+      return value unless value.respond_to? :providedLabel
+      return value unless value.node?
+      return value.providedLabel.first if value.providedLabel.any?
+      value
+    end
+  end
+end