ddr-core 0.2.1

data/lib/ddr/index/csv_query_result.rb

@@ -0,0 +1,84 @@
+require "csv"
+
+module Ddr::Index
+  class CSVQueryResult < AbstractQueryResult
+
+    MAX_ROWS         = 10**8 # Just set to a really high number :)
+    CSV_MV_SEPARATOR = ";"
+
+    delegate :headers, :to_s, :to_csv, to: :table
+
+    def delete_empty_columns!
+      table.by_col!.delete_if { |c, vals| vals.all?(&:nil?) }
+    end
+
+    def each(&block)
+      table.by_row!.each(&block)
+    end
+
+    def [](index_or_header)
+      table.by_col_or_row![index_or_header]
+    end
+
+    def table
+      @table ||= CSV.parse(data, csv_opts)
+    end
+
+    def csv_opts
+      { headers:        csv_headers,
+        converters:     [convert_semicolons, convert_escaped_newlines],
+        return_headers: false,
+        write_headers:  true,
+      }
+    end
+
+    def solr_csv_opts
+      { "csv.mv.separator" => CSV_MV_SEPARATOR,
+        "csv.header"       => solr_csv_header?,
+        "rows"             => solr_csv_rows,
+        "wt"               => "csv",
+      }
+    end
+
+    def query_field_headings
+      query.fields.map { |f| f.respond_to?(:heading) ? f.heading : f.to_s }
+    end
+
+    def csv_headers
+      :first_row
+    end
+
+    def solr_csv_header?
+      query.fields.empty?
+    end
+
+    def solr_csv_rows
+      query.rows || MAX_ROWS
+    end
+
+    def solr_csv_params
+      params.merge(solr_csv_opts)
+    end
+
+    def data
+      if solr_csv_header?
+        solr_data
+      else
+        [ query_field_headings.join(","), solr_data ].join("\n")
+      end
+    end
+
+    def solr_data
+      Connection.get("select", params: solr_csv_params)
+    end
+
+    def convert_semicolons
+      lambda { |f| f.gsub(/\\#{CSV_MV_SEPARATOR}/, CSV_MV_SEPARATOR) rescue f }
+    end
+
+    def convert_escaped_newlines
+      lambda { |f| f.gsub(/\\r/, "\r").gsub(/\\n/, "\n") rescue f }
+    end
+
+  end
+end

data/lib/ddr/index/document_builder.rb

@@ -0,0 +1,9 @@
+module Ddr::Index
+  class DocumentBuilder
+
+    def self.build(doc)
+      ::SolrDocument.new(doc)
+    end
+
+  end
+end

data/lib/ddr/index/field.rb

@@ -0,0 +1,35 @@
+require 'solrizer'
+
+module Ddr::Index
+  class Field < SimpleDelegator
+
+    attr_reader :base
+
+    def initialize(base, *args)
+      @base = base.to_s
+      name = if args.empty?
+               @base
+             elsif args.last.is_a?(Hash) && args.last[:solr_name]
+               args.last[:solr_name]
+             else
+               Solrizer.solr_name(base, *args)
+             end
+      super(name)
+    end
+
+    def label
+      I18n.t "#{i18n_base}.label", default: base.titleize
+    end
+
+    def heading
+      I18n.t "#{i18n_base}.heading", default: base
+    end
+
+    private
+
+    def i18n_base
+      "ddr.index.fields.#{base}"
+    end
+
+  end
+end

data/lib/ddr/index/field_attribute.rb

@@ -0,0 +1,22 @@
+require "virtus"
+
+module Ddr::Index
+  class FieldAttribute < Virtus::Attribute
+
+    def self.coerce(value)
+      case value
+      when Field
+        value
+      when String
+        Field.new(value)
+      when Symbol
+        Fields.get(value)
+      end
+    end
+
+    def coerce(value)
+      self.class.coerce(value)
+    end
+
+  end
+end

data/lib/ddr/index/fields.rb

@@ -0,0 +1,154 @@
+module Ddr::Index
+  module Fields
+
+    ID = UniqueKeyField.instance
+
+    ACCESS_ROLE                   = Field.new :access_role, :stored_sortable
+    ADMIN_POLICY_ID               = Field.new :admin_policy_id, :symbol
+    ADMIN_SET                     = Field.new :admin_set, :stored_sortable
+    ADMIN_SET_FACET               = Field.new :admin_set_facet, :facetable
+    ADMIN_SET_TITLE               = Field.new :admin_set_title, :stored_sortable
+    AFFILIATION                   = Field.new :affiliation, :stored_searchable
+    AFFILIATION_FACET             = Field.new :affiliation_facet, :facetable
+    ALEPH_ID                      = Field.new :aleph_id, :stored_sortable
+    ALL_TEXT                      = Field.new :all_text, solr_name: "all_text_timv"
+    ARRANGER_FACET                = Field.new :arranger_facet, :facetable
+    ASPACE_ID                     = Field.new :aspace_id, :stored_sortable
+    ATTACHED_FILES_HAVING_CONTENT = Field.new :attached_files_having_content, :symbol
+    ATTACHED_TO_ID                = Field.new :attached_to_id, :symbol
+    BIBLICAL_BOOK_FACET           = Field.new :biblical_book_facet, :facetable
+    BOX_NUMBER_FACET              = Field.new :box_number_facet, :facetable
+    CATEGORY_FACET                = Field.new :category_facet, :facetable
+    CHAPTER_AND_VERSE_FACET       = Field.new :chapter_and_verse_facet, :facetable
+    COLLECTION_FACET              = Field.new :collection_facet, :facetable
+    COLLECTION_ID                 = Field.new :collection_id, :symbol
+    COLLECTION_TITLE              = Field.new :collection_title, :stored_sortable
+    COMMON_MODEL_NAME             = Field.new :common_model_name, :stored_sortable
+    COMPANY_FACET                 = Field.new :company_facet, :facetable
+    COMPOSER_FACET                = Field.new :composer_facet, :facetable
+    CONTENT_CREATE_DATE           = Field.new :content_create_date, :stored_sortable, type: :date
+    CONTENT_SIZE                  = Field.new :content_size, solr_name: "content_size_lsi"
+    CONTENT_SIZE_HUMAN            = Field.new :content_size_human, :symbol
+    CONTRIBUTOR_FACET             = Field.new :contributor_facet, :facetable
+    CREATOR_FACET                 = Field.new :creator_facet, :facetable
+    DATE_FACET                    = Field.new :date_facet, :facetable
+    DATE_SORT                     = Field.new :date_sort, :sortable
+    DC_IS_PART_OF                 = Field.new :isPartOf, :symbol
+    DEPOSITOR                     = Field.new :depositor, :stored_sortable
+    DISPLAY_FORMAT                = Field.new :display_format, :stored_sortable
+    DOI                           = Field.new :doi, :symbol
+    EAD_ID                        = Field.new :ead_id, :stored_sortable
+    ENGRAVER_FACET                = Field.new :engraver_facet, :facetable
+    EXTRACTED_TEXT                = Field.new :extracted_text, solr_name: "extracted_text_tsm"
+    FOLDER_FACET                  = Field.new :folder_facet, :facetable
+    FOR_COLLECTION_ID             = Field.new :for_collection_id, :symbol
+    FORMAT_FACET                  = Field.new :format_facet, :facetable
+    GENRE_FACET                   = Field.new :genre_facet, :facetable
+    IDENTIFIER_ALL                = Field.new :identifier_all, :symbol
+    ILLUSTRATED_FACET             = Field.new :illustrated_facet, :facetable
+    ILLUSTRATOR_FACET             = Field.new :illustrator_facet, :facetable
+    INGESTED_BY                   = Field.new :ingested_by, :stored_sortable
+    INGESTION_DATE                = Field.new :ingestion_date, :stored_sortable, type: :date
+    INSTRUMENTATION_FACET         = Field.new :instrumentation_facet, :facetable
+    INTERVIEWER_NAME_FACET        = Field.new :interviewer_name_facet, :facetable
+    IS_FORMAT_OF                  = Field.new :isFormatOf, :symbol
+    IS_LOCKED                     = Field.new :is_locked, :stored_sortable
+    # Keeping IS_MEMBER_OF_COLLECTION and IS_PART_OF for now for backwards compatibility
+    IS_MEMBER_OF_COLLECTION       = Field.new :is_member_of_collection, :symbol
+    IS_PART_OF                    = Field.new :is_part_of, :symbol
+    LANGUAGE_FACET                = Field.new :language_facet, :facetable
+    LANGUAGE_NAME                 = Field.new :language_name, :stored_searchable
+    LAST_FIXITY_CHECK_ON          = Field.new :last_fixity_check_on, :stored_sortable, type: :date
+    LAST_FIXITY_CHECK_OUTCOME     = Field.new :last_fixity_check_outcome, :symbol
+    LAST_VIRUS_CHECK_ON           = Field.new :last_virus_check_on, :stored_sortable, type: :date
+    LAST_VIRUS_CHECK_OUTCOME      = Field.new :last_virus_check_outcome, :symbol
+    LICENSE                       = Field.new :license, :stored_sortable
+    LITHOGRAPHER_FACET            = Field.new :lithographer_facet, :facetable
+    LOCAL_ID                      = Field.new :local_id, :stored_sortable
+    LYRICIST_FACET                = Field.new :lyricist_facet, :facetable
+    MEDIA_SUB_TYPE                = Field.new :content_media_sub_type, :facetable
+    MEDIA_MAJOR_TYPE              = Field.new :content_media_major_type, :facetable
+    MEDIA_TYPE                    = Field.new :content_media_type, :symbol
+    MEDIUM_FACET                  = Field.new :medium_facet, :facetable
+    MULTIRES_IMAGE_FILE_PATH      = Field.new :multires_image_file_path, :stored_sortable
+    NESTED_PATH                   = Field.new :nested_path, :stored_sortable
+    NESTED_PATH_TEXT              = Field.new :nested_path_text, :searchable
+    ORIGINAL_FILENAME             = Field.new :original_filename, :stored_sortable
+    PARENT_ID                     = Field.new :parent_id, :symbol
+    PERFORMER_FACET               = Field.new :performer_facet, :facetable
+    PERMANENT_ID                  = Field.new :permanent_id, :stored_sortable
+    PERMANENT_URL                 = Field.new :permanent_url, :stored_sortable
+    PLACEMENT_COMPANY_FACET       = Field.new :placement_company_facet, :facetable
+    POLICY_ROLE                   = Field.new :policy_role, :symbol
+    PRODUCER_FACET                = Field.new :producer_facet, :facetable
+    PRODUCT_FACET                 = Field.new :product_facet, :facetable
+    PUBLICATION_FACET             = Field.new :publication_facet, :facetable
+    PUBLISHER_FACET               = Field.new :publisher_facet, :facetable
+    RESEARCH_HELP_CONTACT         = Field.new :research_help_contact, :stored_sortable
+    RESOURCE_CREATE_DATE          = Field.new :created_at, :stored_sortable, type: :date
+    RESOURCE_MODEL                = Field.new :resource_model, solr_name: "internal_resource_ssim"
+    RESOURCE_MODIFIED_DATE        = Field.new :updated_at, :stored_sortable, type: :date
+    RESOURCE_ROLE                 = Field.new :resource_role, :symbol
+    RIGHTS_NOTE                   = Field.new :rights_note, :stored_searchable
+    ROLL_NUMBER_FACET             = Field.new :roll_number_facet, :facetable
+    SERIES_FACET                  = Field.new :series_facet, :facetable
+    SETTING_FACET                 = Field.new :setting_facet, :facetable
+    SPATIAL_FACET                 = Field.new :spatial_facet, :facetable
+    STREAMABLE_MEDIA_TYPE         = Field.new :streamable_media_type, :stored_sortable
+    STRUCTURE                     = Field.new :structure, solr_name: "structure_ss"
+    STRUCTURE_SOURCE              = Field.new :structure_source, :stored_sortable
+    SUBJECT_FACET                 = Field.new :subject_facet, :facetable
+    SUBSERIES_FACET               = Field.new :subseries_facet, :facetable
+    TARGET_ID                     = Field.new :target_id, :symbol
+    TECHMD_COLOR_SPACE            = Field.new :techmd_color_space, :symbol
+    TECHMD_CREATING_APPLICATION   = Field.new :techmd_creating_application, :symbol
+    TECHMD_CREATION_TIME          = Field.new :techmd_creation_time, :stored_searchable, type: :date
+    TECHMD_FILE_SIZE              = Field.new :techmd_file_size, solr_name: "techmd_file_size_lsi"
+    TECHMD_FITS_VERSION           = Field.new :techmd_fits_version, :stored_sortable
+    TECHMD_FITS_DATETIME          = Field.new :techmd_fits_datetime, :stored_sortable, type: :date
+    TECHMD_FORMAT_LABEL           = Field.new :techmd_format_label, :symbol
+    TECHMD_FORMAT_VERSION         = Field.new :techmd_format_version, :symbol
+    TECHMD_ICC_PROFILE_NAME       = Field.new :techmd_icc_profile_name, :symbol
+    TECHMD_ICC_PROFILE_VERSION    = Field.new :techmd_icc_profile_version, :symbol
+    TECHMD_IMAGE_HEIGHT           = Field.new :techmd_image_height, :stored_searchable, type: :integer
+    TECHMD_IMAGE_WIDTH            = Field.new :techmd_image_width, :stored_searchable, type: :integer
+    TECHMD_MD5                    = Field.new :techmd_md5, :stored_sortable
+    TECHMD_MEDIA_TYPE             = Field.new :techmd_media_type, :symbol
+    TECHMD_MODIFICATION_TIME      = Field.new :techmd_modification_time, :stored_searchable, type: :date
+    TECHMD_PRONOM_IDENTIFIER      = Field.new :techmd_pronom_identifier, :symbol
+    TECHMD_VALID                  = Field.new :techmd_valid, :symbol
+    TECHMD_WELL_FORMED            = Field.new :techmd_well_formed, :symbol
+    TEMPORAL_FACET                = Field.new :temporal_facet, :facetable
+    TITLE                         = Field.new :title, :stored_sortable
+    TONE_FACET                    = Field.new :tone_facet, :facetable
+    TYPE_FACET                    = Field.new :type_facet, :facetable
+    VOLUME_FACET                  = Field.new :volume_facet, :facetable
+    WORKFLOW_STATE                = Field.new :workflow_state, :stored_sortable
+    YEAR_FACET                    = Field.new :year_facet, solr_name: "year_facet_iim"
+
+    def self.get(name)
+      const_get(name.to_s.upcase, false)
+    end
+
+    def self.techmd
+      @techmd ||= constants(false).select { |c| c =~ /\ATECHMD_/ }.map { |c| const_get(c) }.freeze
+    end
+
+    def self.descmd
+      @descmd ||= Ddr::Describable.term_names.map do |term_name|
+        Field.new term_name, *Ddr::Describable.indexers_for(term_name)
+      end.freeze
+    end
+
+    def self.adminmd
+      @adminmd ||= Ddr::HasAdminMetadata.term_names.map do |term_name|
+        begin
+          get(term_name)
+        rescue NameError
+          Field.new term_name
+        end
+      end.freeze
+    end
+
+  end
+end

data/lib/ddr/index/filter.rb

@@ -0,0 +1,139 @@
+require "forwardable"
+require "virtus"
+
+module Ddr::Index
+  class Filter
+    include Virtus.model
+
+    attribute :clauses, Array, default: [ ]
+
+    def ==(other)
+      other.instance_of?(self.class) && (other.clauses == self.clauses)
+    end
+
+    module Api
+      def raw(*clauses)
+        self.clauses += clauses
+        self
+      end
+
+      def term(conditions)
+        self.clauses += conditions.map { |f, v| QueryClause.term(f, v) }
+        self
+      end
+
+      def where(conditions)
+        self.clauses += conditions.map { |f, v| QueryClause.where(f, v) }
+        self
+      end
+
+      def where_not(conditions)
+        self.clauses += conditions.map do |field, v|
+          Array(v).map { |value| QueryClause.negative(field, value) }
+        end.flatten
+        self
+      end
+
+      def absent(field)
+        self.clauses << QueryClause.absent(field)
+        self
+      end
+
+      def present(field)
+        self.clauses << QueryClause.present(field)
+        self
+      end
+
+      def negative(field, value)
+        self.clauses << QueryClause.negative(field, value)
+        self
+      end
+
+      def before(field, value)
+        self.clauses << QueryClause.before(field, value)
+        self
+      end
+
+      def before_days(field, value)
+        self.clauses << QueryClause.before_days(field, value)
+        self
+      end
+
+      def join(**args)
+        self.clauses << QueryClause.join(**args)
+        self
+      end
+
+      def regexp(field, value)
+        self.clauses << QueryClause.regexp(field, value)
+        self
+      end
+    end
+
+    module ClassMethods
+      extend Forwardable
+
+      delegate Api.public_instance_methods => :new_filter
+
+      def has_content
+        model "Ddr::Component", "Ddr::Attachment", "Ddr::Target"
+      end
+
+      def is_governed_by(arg)
+        term admin_policy_id: get_join_id(arg)
+      end
+
+      def is_member_of_collection(arg)
+        term is_member_of_collection: get_id(arg)
+      end
+
+      def is_part_of(arg)
+        term is_part_of: get_id(arg)
+      end
+
+      def model(*models)
+        namespaced_models = models.map{ |model| model.starts_with?("Ddr::") ? model : "Ddr::#{model}" }
+        where resource_model: namespaced_models
+      end
+
+      private
+
+      # def get_uri(arg)
+      #   case arg
+      #   when ActiveFedora::Base
+      #     arg.internal_uri
+      #   when String
+      #     arg.start_with?("info:fedora/") ? arg : "info:fedora/#{arg}"
+      #   else
+      #     raise TypeError, "Argument must be a String or ActiveFedora::Base instance: #{arg.class}"
+      #   end
+      # end
+
+      def get_id(arg)
+        case arg
+          when Ddr::Resource
+            arg.id.to_s
+          when Valkyrie::ID
+            arg.to_s
+          when String
+            arg
+          else
+            raise TypeError, "Argument must be a String or Ddr::Resource or Valkyrie::ID instance: #{arg.class}"
+        end
+      end
+
+      def get_join_id(arg)
+        "id-#{get_id(arg)}"
+      end
+
+      def new_filter
+        Filter.new
+      end
+
+    end
+
+    include Api
+    extend ClassMethods
+
+  end
+end