krikri 0.4.0 → 0.5.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 8ba2b247b82e5f0d8a3f52adb0e3242d41deff3b
-  data.tar.gz: 429288310bd3531d4fbb20881d7a17983c69a015
+  metadata.gz: 59195bbb3fbde31f549ca473ae9941c28a294947
+  data.tar.gz: dca46026cd0cf2c5cc51970aa0d69ae7a9ba3275
 SHA512:
-  metadata.gz: 783c38573ddf5cac189f56220981aeba2768c1eba0a5e4d873d36642f4e633ca9e3ef449ba14ceae49d58b0bc05e426c9b5b84708f0ed7e9788ec305f3416464
-  data.tar.gz: 817df4eee63756335abc562d8badc40160d32d83a92aa5fde93e32ba17ca2938b2b311eed93cf8469756939f6551afd3b0bca25ca2b269f504d6b21b258457b8
+  metadata.gz: 97d5fae3cf1dd0860ee9d66b800e0f1bde373367ec82bffaf3d48b5e73f4f7d2ba0976b06c6a43e0a2d0b9102e6830d5083819d7cbbcbc01995b656fd4220bb3
+  data.tar.gz: 767df8d2522301a580cabf47320347932ddfae28a3fd1e136d406d343cae6726827d26447c26541372d9debf9858933e79fe3efd332efba16da022ad22e252b4

data/app/assets/stylesheets/krikri/application.css

@@ -17,3 +17,18 @@
  #document_list_table td {
   padding-right: 3em;
  }
+
+ h1 {
+  font-size: 2.2em;
+  margin-bottom: 0px;
+ }
+
+ h2 {
+  font-size: 1.5em;
+  color: #888;
+  margin-top: 2px;
+ }
+
+ h3 {
+  font-size: 1.3em;
+ }

data/app/controllers/krikri/providers_controller.rb

@@ -0,0 +1,27 @@
+module Krikri
+  ##
+  # Handles requests for provider dashboards for 'All Providers' and
+  # individual providers by ID.
+  #
+  # @see Krikri::Provider
+  class ProvidersController < ApplicationController
+    ##
+    # Renders the index view with `@providers` as an Array of {Krikri::Provider}s.
+    def index
+      @providers = Krikri::Provider.all
+    end
+
+    ##
+    # Renders the show view for the provider given by `id`.
+    def show
+      if params[:set_session]
+        session[:current_provider] = params[:id]
+        redirect_to :back, provider: params[:id]
+      elsif params[:clear_session]
+        session.delete :current_provider
+        redirect_to providers_path
+      end
+      @current_provider = Krikri::Provider.find(params[:id])
+    end
+  end
+end

data/app/controllers/krikri/qa_reports_controller.rb

@@ -0,0 +1,28 @@
+module Krikri
+  ##
+  # Handles HTTP requests for QA Reports
+  #
+  # @see Krikri::QAReport
+  class QaReportsController < ApplicationController
+    ##
+    # Renders a list of current reports
+    def index
+      @reports = Krikri::QAReport.all
+    end
+
+    ##
+    # Rendering the report as either a full `field` report or a `count` report.
+    #
+    # Responds to format of `text/csv` with a CSV rendering of the requested
+    # report type.
+    def show
+      @report = Krikri::QAReport.find(params[:id])
+      @type = params[:type] == 'count' ? :count : :field
+
+      respond_to do |format|
+        format.html
+        format.csv { render text: @report.send("#{@type}_csv".to_sym).to_csv }
+      end
+    end
+  end
+end

data/app/controllers/krikri/records_controller.rb

@@ -7,11 +7,14 @@ module Krikri
   # ApplicationController.  It does not interit from Krikri's
   # ApplicationController.
   class RecordsController < CatalogController
-    before_action :authenticate_user!
+    before_action :authenticate_user!, :set_provider
+
+    self.solr_search_params_logic += [:records_by_provider]
+
     ##
-    # RecordsReportsController has access to views in the following
+    # RecordsController has access to views in the following
     # directories:
-    #   krikri/validation_reports
+    #   krikri/records
     #   catalog (defined in Blacklight)
     # It inherits view templates from the host application's
     # ApplicationController.  It uses krikri's application layout:
@@ -81,6 +84,8 @@ module Krikri
       config.solr_document_model = Krikri::SearchIndexDocument
     end
 
+    private
+
     ##
     # Construct a valid item URI from a local name, and use it to fetch a single
     # document from the search index.
@@ -90,9 +95,28 @@ module Krikri
     # with Blacklight v5.10.
     # @param String id is a local name.
     def get_solr_response_for_doc_id(id=nil, extra_controller_params={})
-      id_uri = Krikri::Settings.marmotta.item_container << '/' << id
-      solr_response = solr_repository.find(id_uri, extra_controller_params)
+      id = (RDF::URI(Krikri::Settings.marmotta.item_container) / id).to_s if id
+      solr_response = solr_repository.find(id, extra_controller_params)
       [solr_response, solr_response.documents.first]
     end
+
+    ##
+    # Limit the records returned by a Solr request to those belonging to the
+    # current provider.
+    # @param [Hash] solr_parameters a hash of parameters to be sent to Solr.
+    # @param [Hash] user_parameters a hash of user-supplied parameters.
+    def records_by_provider(solr_params, user_params)
+      if @provider_id.present?
+        provider = Krikri::Provider.find(@provider_id)
+        solr_params[:fq] ||= []
+        solr_params[:fq] << "provider_id:\"#{provider.rdf_subject}\""
+      end
+    end
+
+    ##
+    # Sets the provider id for use as a search filter/view
+    def set_provider
+      @provider_id = params[:provider]
+    end
   end
 end

data/app/controllers/krikri/reports_controller.rb

@@ -0,0 +1,24 @@
+module Krikri
+  ##
+  # Handles HTTP requests for the Reports dashboard, presenting all types of
+  # reports, filtered by {Krikri::Provider} if given.
+  class ReportsController < ApplicationController
+    layout 'krikri/application'
+
+    ##
+    # Renders the index view, giving `@validation_reports` and `@qa_reports`
+    # for the specified provider.
+    def index
+      @current_provider = params[:provider]
+      @validation_reports = Krikri::ValidationReport.new.all do
+        self.provider_id = @current_provider
+      end
+
+      if @current_provider
+        @qa_reports = Array(Krikri::QAReport.find_by(provider: @current_provider))
+      else
+        @qa_reports = Krikri::QAReport.all
+      end
+    end
+  end
+end

data/app/controllers/krikri/validation_reports_controller.rb

@@ -1,14 +1,12 @@
 module Krikri
   ##
-  # Marshals SolrDocuments for views.
-  # Sets default Solr request params for Validation Reports.
+  # Marshals validation reports for views.
   #
   # ValidationReportsController inherits from the host application's
   # ApplicationController.  It does not interit from Krikri's
   # ApplicationController.
   class ValidationReportsController < CatalogController
     before_action :authenticate_user!
-    before_filter :valid_params, :only => :index
 
     ##
     # ValidationReportsController has access to views in the following
@@ -19,32 +17,16 @@ module Krikri
     # ApplicationController.  It uses krikri's application layout:
     layout 'krikri/application'
 
-    configure_blacklight do |config|
-
-      # Default parameters to send to solr for all search-like requests.
-      config.default_solr_params = {
-        :qt => 'standard',
-        :rows => 100
-      }
-
-      # solr fields to be displayed in the index (search results) view
-      #   The ordering of the field names is the order of the display
-      config.add_index_field 'sourceResource_title', :label => 'Title',
-                             helper_method: 'link_to_show'
-      config.add_index_field 'id', :label => 'ID', helper_method: 'link_to_show'
-      config.add_index_field 'isShownAt_id', :label => 'Is Shown At',
-                             helper_method: 'make_this_a_link'
-
-      config.show.route = { controller: 'records' }
-
-      config.solr_document_model = Krikri::SearchIndexDocument
+    def show
+      provider_id = params[:provider]
+      page = params[:page]
+      per_page = params[:per_page]
+      @response = ValidationReport.new.find(params[:id]) do
+        self.provider_id = provider_id
+        self.start = page
+        self.rows = per_page
+      end
+      @documents = @response.documents
     end
-
-    private
-
-    def valid_params
-      redirect_to :report_lists if !params['report_name'] || !params['q']
-    end
-
   end
 end

data/app/helpers/krikri/application_helper.rb

@@ -1,4 +1,66 @@
 module Krikri
   module ApplicationHelper
+    ##
+    # @return [Array<Blacklight::SolrResponse::Facets::FacetItem>]
+    def available_providers
+      Krikri::Provider.all(&:reload)
+    end
+
+    ##
+    # @param provider [String, nil]
+    # @return [String]
+    def provider_name(provider)
+      provider = Krikri::Provider.find(provider) if provider.is_a? String
+      provider.present? ? provider.provider_name : "All Providers"
+    end
+
+    ##
+    # Gives the last path fragment for a given URI string in HTML escaped form
+    #
+    # @example
+    #   local_name('http://my_domain/blah/0123') => '0123'
+    #
+    # @param uri [#to_s] a URL formatted string to split
+    # @return [String] the escaped fragment
+    def local_name(uri)
+      uri.to_s.split('/').last.html_safe
+    end
+
+    ##
+    # Link to the current page, changing the session provider the given
+    # value.
+    # @param provider [String, nil]
+    def link_to_current_page_by_provider(provider)
+      provider = Krikri::Provider.find(provider) if provider.is_a? String
+
+      return link_to_provider_page(provider) if params[:controller] ==
+                                                'krikri/providers'
+      params[:provider] = provider.id
+      params[:session_provider] = provider.id
+      link_to provider_name(provider), params
+    end
+
+    def set_session_provider(provider)
+      provider = Krikri::Provider.find(provider) if provider.is_a? String
+      link_to provider_name(provider),
+              krikri.provider_path(provider.id, set_session: 1),
+              rel: 'nofollow'
+    end
+
+    def remove_session_provider
+      link_to "All Providers",
+              krikri.provider_path('clear', clear_session: 1),
+              rel: 'nofollow'
+    end
+
+    ##
+    # Link to the ProviderController's :index or :show route, using the given
+    # provider id in the case of :show.
+    # @param provider [String, nil]
+    def link_to_provider_page(provider)
+      return link_to(provider_name(provider), providers_path) unless provider
+      return link_to provider_name(provider),
+              provider_path(provider.id, set_session: provider.id)
+    end
   end
 end

data/app/helpers/krikri/records_helper.rb

@@ -1,5 +1,19 @@
 module Krikri
   module RecordsHelper
     include Krikri::SearchResultsHelperBehavior
+
+    ##
+    # Return the id of a document randomly selected from the search index.
+    # @param provider_id [String, nil] the id of the provider that the randomly
+    # selected document will belong to.
+    # @return [String, nil] the id of the randomly selected document. If none are
+    #   available, gives `nil`
+    def random_record_id(provider_id)
+      doc = Krikri::RandomSearchIndexDocumentBuilder.new do
+        self.provider_id = provider_id
+      end.document
+
+      doc.present? ? local_name(doc.id) : nil
+    end
   end
 end

data/app/helpers/krikri/reports_helper.rb

@@ -0,0 +1,16 @@
+module Krikri
+  module ReportsHelper
+    def render_validation_report(report)
+      return validation_report_title(report) if report.items.first.hits == 0
+      link_to validation_report_title(report), { :controller =>
+        'validation_reports', :action => 'show', :id => report.name,
+        :provider => @current_provider }, :method => :get
+    end
+
+    def validation_report_title(report)
+      return "#{report.name} missing (#{report.items.first.hits} record)" if
+        report.items.first.hits == 1
+      "#{report.name} missing (#{report.items.first.hits} records)"
+    end
+  end
+end

data/app/models/krikri/activity.rb

@@ -83,5 +83,32 @@ module Krikri
       RDF::URI(Krikri::Settings['marmotta']['ldp']) /
         Krikri::Settings['prov']['activity'] / id.to_s
     end
+
+    ##
+    # Return an Enumerator of URI strings of entities (e.g. aggregations or
+    # original records) that pertain to this activity
+    #
+    # @return [Enumerator] URI strings
+    def entity_uris
+      activity_uri = RDF::URI(rdf_subject)  # This activity's LDP URI
+      query = Krikri::ProvenanceQueryClient.find_by_activity(activity_uri)
+      query.each_solution.lazy.map do |s|
+        s.record.to_s
+      end
+    end
+
+    ##
+    # Return an Enumerator of entities (e.g. aggregations or original records)
+    # that have been affected by this activity.
+    #
+    # The kind of object that is returned depends on the EntityBehavior class
+    # that is associated with the SoftwareAgent that is represented by the
+    # Activity's `agent' field.
+    #
+    # @return [Enumerator] Objects
+    def entities
+      agent_instance.entity_behavior.entities(self)
+    end
+
   end
 end

data/app/models/krikri/provider.rb

@@ -0,0 +1,80 @@
+module Krikri
+  ##
+  # This models provider data from the {Krikri::Repository}.
+  #
+  # @todo create an appropriate `ActiveTriples::PersistenceStrategy' for
+  #   handling providers as data from a repository alongside LDP data.
+  # @see ActiveTriples::Resource
+  class Provider < DPLA::MAP::Agent
+    configure :base_uri => Krikri::Settings.prov.provider_base
+
+    ##
+    # Gives a list of all providers, ignoring those with no URI (bnodes).
+    #
+    # @return [Array<Krikri::Provider>]
+    def self.all
+      query_params = { :rows => 0,
+                       :id => '*:*',
+                       'facet.field' => 'provider_id' }
+      response = Krikri::SolrResponseBuilder.new(query_params).response
+      response.facets.first.items.map do |item|
+        provider = new(item.value)
+        provider.node? ? nil : provider
+      end.compact
+    end
+
+    ##
+    # @param id [#to_s] the identifier (local name or URI) of the provider
+    #   to find
+    #
+    # @return [Krikri::Provider]
+    # @todo Use {ActiveTriples}/{RDF::Repository} to populate the object
+    def self.find(id)
+      return nil if id.nil?
+      provider = new(id)
+      provider.reload
+      return nil if provider.empty?
+      provider
+    end
+
+    ##
+    # Loads the provider's data from the repository.
+    #
+    # @return [Krikri::Provider] self
+    def reload
+      query = Krikri::Repository.query_client.select.where([self, :p, :o])
+      query.each_solution do |solution|
+        set_value(solution.p, solution.o)
+      end
+
+      self
+    end
+
+    ##
+    # A list of records of type ore:Aggregation associated with this provider
+    #
+    # @return [Array<DPLA::MAP::Aggregation>] the Aggregations with this
+    #   provider
+    def records
+      query = Krikri::Repository.query_client.select
+              .where([:record, RDF::EDM.provider, self],
+                     [:record, RDF.type, RDF::ORE.Aggregation])
+
+      query.execute.map do |solution|
+        DPLA::MAP::Aggregation.new(solution.record)
+      end
+    end
+
+    ##
+    # @return [String] the local name (last uri segment) for the provider
+    def id
+      node? ? nil : rdf_subject.to_s.gsub(base_uri.to_s, '')
+    end
+
+    ##
+    # @return [String] the name of the provider
+    def provider_name
+      label.first || providedLabel.first || id
+    end
+  end
+end

data/app/models/krikri/qa_report.rb

@@ -0,0 +1,238 @@
+module Krikri
+  ##
+  # Represents a QA Report, giving details about a set of records associated
+  # with a given provider. Reports are given as structured hashes, containing
+  # keys for each registered property in the {DPLA::MAP::Aggregation} structure.
+  #
+  # Reports include two report types, a `#field_report` and a `#count_report`.
+  #
+  # The `#field_report` includes values for each property with the URIs for the
+  # `ore:Aggregation` and `edm:isShownAt` (`edm:WebResource`) associated with
+  # that field.
+  #
+  # The `#count_report` includes values for each field, with the count occurances
+  # for that value across all records for the provider.
+  #
+  # `QAReports` are saved to the database (with timestamp, etc... via
+  # {ActiveRecord}). More than one report can be associated with each provider.
+  #
+  # Reports can also serialize themselves as `csv`, via `#field_csv` and
+  # `#count_csv`.
+  #
+  # @example:
+  #
+  #   report = QAReport.create(provider: 'moomin_valley')
+  #   report.generate_field_report!
+  #   report.generate_count_report!
+  #
+  #   report.field_report['edm:aggregatedCHO->dc:alternative']
+  #   => {"Stonewall Inn Graffiti"=>
+  #      [{:aggregation=>"http://localhost:8983/marmotta/ldp/items/krikri_sample",
+  #        :isShownAt=> "http://digitalcollections.nypl.org/items/12345"}]}
+  #
+  #   report.count_report['edm:aggregatedCHO->dc:alternative']
+  #   => {"Stonewall Inn Graffiti"=>1}
+  #
+  # @see Krikri::QAQueryClient
+  class QAReport < ActiveRecord::Base
+    serialize :field_report, Hash
+    serialize :count_report, Hash
+
+    ##
+    # Generates and saves the field report for the provider, sending SPARQL
+    # queries as necessary.
+    #
+    # @return [Hash]
+    def generate_field_report!
+      report = field_queries.inject({}) do |report_hash, (k, v)|
+        report_hash[k] = solutions_to_hash(v.execute)
+        report_hash
+      end
+      update(field_report: report)
+    end
+
+    ##
+    # Generates and saves the coun treport for the provider, sending SPARQL
+    # queries as necessary.
+    #
+    # @return [Hash]
+    def generate_count_report!
+      report = count_queries.inject({}) do |report_hash, (k, v)|
+        report_hash[k] = solutions_to_counts(v.execute)
+        report_hash
+      end
+
+      update(count_report: report)
+    end
+
+    ##
+    # @param include_fields
+    #
+    # @return [CSV::Table] a table containing values, aggregations, and
+    #   isShownAt URLs for each included field
+    def field_csv(*include_fields)
+      fields = field_report.keys
+      fields.select! { |f| include_fields.include? f } unless
+        include_fields.empty?
+
+      variables = [:value, :aggregation, :isShownAt]
+      headers = fields.product(variables).map { |header| header.join(' ') }
+
+      table = CSV::Table.new([CSV::Row.new(headers, [], true)])
+      return table if field_report.nil? || field_report.empty?
+
+      rows = []
+
+      field_report.each do |field, values|
+        values.each do |value, agg_list|
+          agg_list.each_with_index do |agg_hash, i|
+            rows[i] ||= CSV::Row.new(headers, [])
+            rows[i]["#{field} value"] = value.to_s
+            rows[i]["#{field} aggregation"] = agg_hash[:aggregation].to_s
+            rows[i]["#{field} isShownAt"] = agg_hash[:isShownAt].to_s
+          end
+        end
+      end
+
+      rows.each { |r| table << r }
+      table
+    end
+
+    ##
+    # @param include_fields
+    #
+    # @return [CSV::Table] a table containing values and their counts
+    #   for each included field
+    def count_csv(*include_fields)
+      fields = count_report.keys
+      fields.select! { |f| include_fields.include? f } unless
+        include_fields.empty?
+
+      variables = [:value, :count]
+      headers = fields.product(variables).map { |header| header.join(' ') }
+
+      table = CSV::Table.new([CSV::Row.new(headers, [], true)])
+      return table if count_report.nil? || count_report.empty?
+
+      rows = []
+
+      count_report.each do |field, hash|
+        hash.to_a.each_with_index do |value, i|
+          rows[i] ||= CSV::Row.new(headers, [])
+          rows[i]["#{field} value"] = value.first.to_s
+          rows[i]["#{field} count"] = value.last
+        end
+      end
+
+      rows.each { |r| table << r }
+      table
+    end
+
+    ##
+    # Retrieves the provider as an object
+    #
+    # @return [Krikri::Provider] the provider as a populated
+    #   {ActiveTriples::RDFSource}
+    # @todo figure out a better relations pattern between {ActiveRecord} objects
+    #   and {ActiveTriples}
+    def build_provider
+      Krikri::Provider.find(provider)
+    end
+
+    private
+
+    ##
+    # @return [Hash<SPARQL::Client::Query]
+    def count_queries
+      queries = {}
+      each_property(DPLA::MAP::Aggregation).each do |properties|
+        queries[property_name(properties)] =
+          QAQueryClient.counts_for_predicate(properties,
+                                             RDF::URI(build_provider.rdf_subject))
+      end
+      queries
+    end
+
+    ##
+    # @return [Array<SPARQL::Client::Query]
+    def field_queries
+      queries = {}
+      each_property(DPLA::MAP::Aggregation).each do |properties|
+        queries[property_name(properties)] =
+          QAQueryClient.values_for_predicate(properties,
+                                             RDF::URI(build_provider.rdf_subject))
+      end
+      queries
+    end
+
+    ##
+    # @param properties [Array<RDF::URI>] an array of URIs to join
+    #
+    # @return [String] a string representing the property array as qualified
+    #   names
+    def property_name(properties)
+      properties.map(&:pname).join('->')
+    end
+
+    ##
+    # @param solutions [RDF::Query::Solutions]
+    # @return [Hash] a hash of values, aggregations, and isShownAt URIs
+    def solutions_to_hash(solutions)
+      matches = {}
+      solutions.each do |solution|
+        key = solution.value.to_s
+        matches[key] ||= []
+        matches[key] <<  { aggregation: solution.aggregation.to_s,
+                           isShownAt: solution.isShownAt.to_s }
+
+      end
+      matches
+    end
+
+    ##
+    # @param solutions [RDF::Query::Solutions]
+    # @return [Hash] a hash of values and counts
+    def solutions_to_counts(solutions)
+      matches = {}
+      solutions.each do |solution|
+        count = solution[:count].to_i
+        matches[solution.value.to_s] = count unless count == 0
+      end
+      matches
+    end
+
+    ##
+    # Gives an enumerator for the properties on an {ActiveTriples::Resource}
+    # class; #each/#next give an Array representing the property chain for each
+    # property config that does not have a `class_name` (i.e. is configured as
+    # a literal).
+    #
+    # @return [Enumerator] an deep enumerator over all registered
+    #   properties
+    # @todo move to {ActiveTriples}?
+    def each_property(klass)
+      nested_properties_list(klass).flatten(1).to_enum
+    end
+
+    ##
+    # @param klass [#properties]
+    #
+    # @return [Array<Array<RDF::Term>>] An array of property predicates
+    # @todo move to {ActiveTriples}?
+    def nested_properties_list(klass)
+      klass.properties.map do |_, config|
+        properties = []
+        if config.class_name.nil? || config.class_name == klass
+          properties << Array(config.predicate)
+        else
+          nested_properties_list(config.class_name).each do |prop|
+            prop.each do |p|
+              properties << Array(config.predicate).append(p).flatten
+            end
+          end
+        end
+        properties
+      end
+    end
+  end
+end