ddr-models 2.0.1 → 2.1.0.rc1

data/lib/ddr/derivatives/png_generator.rb

@@ -1,3 +1,5 @@
+require 'open3'
+
 module Ddr
   module Derivatives
     class PngGenerator < Generator

data/lib/ddr/derivatives/ptif_generator.rb

@@ -1,3 +1,5 @@
+require 'open3'
+
 module Ddr
   module Derivatives
     class PtifGenerator < Generator

data/lib/ddr/events/fixity_check_event.rb

@@ -28,8 +28,8 @@ module Ddr
       end
 
       def to_solr
-        { Ddr::IndexFields::LAST_FIXITY_CHECK_ON => event_date_time_s,
-          Ddr::IndexFields::LAST_FIXITY_CHECK_OUTCOME => outcome }
+        { Ddr::Index::Fields::LAST_FIXITY_CHECK_ON => event_date_time_s,
+          Ddr::Index::Fields::LAST_FIXITY_CHECK_OUTCOME => outcome }
       end
 
       protected

data/lib/ddr/events/virus_check_event.rb

@@ -9,20 +9,8 @@ module Ddr
       self.description = "Content file scanned for viruses"
 
       def to_solr
-        { Ddr::IndexFields::LAST_VIRUS_CHECK_ON => event_date_time_s,
-          Ddr::IndexFields::LAST_VIRUS_CHECK_OUTCOME => outcome }
-      end
-
-      # Message sent by ActiveSupport::Notifications
-      def self.call(*args)
-        notification = ActiveSupport::Notifications::Event.new(*args)
-        result = notification.payload[:result] # Ddr::Antivirus::ScanResult instance
-        create(pid: notification.payload[:pid],
-               event_date_time: result.scanned_at,
-               outcome: result.ok? ? SUCCESS : FAILURE,
-               software: result.version,
-               detail: result.to_s
-               )
+        { Ddr::Index::Fields::LAST_VIRUS_CHECK_ON => event_date_time_s,
+          Ddr::Index::Fields::LAST_VIRUS_CHECK_OUTCOME => outcome }
       end
 
     end

data/lib/ddr/index.rb

@@ -0,0 +1,29 @@
+module Ddr
+  module Index
+    extend ActiveSupport::Autoload
+
+    autoload :AbstractQueryResult
+    autoload :Connection
+    autoload :CSVQueryResult
+    autoload :DocumentBuilder
+    autoload :Field
+    autoload :Fields
+    autoload :Filter
+    autoload :Filters
+    autoload :LegacyLicenseFields
+    autoload :Query
+    autoload :QueryBuilder
+    autoload :QueryClause
+    autoload :QueryResult
+    autoload :QueryValue
+    autoload :Response
+    autoload :UniqueKeyField
+
+    def self.pids
+      builder = QueryBuilder.new
+      query = builder.query
+      query.pids
+    end
+
+  end
+end

data/lib/ddr/index/abstract_query_result.rb

@@ -0,0 +1,23 @@
+module Ddr::Index
+  class AbstractQueryResult
+    include Enumerable
+
+    attr_reader :query, :conn
+    delegate :params, to: :query
+
+    def initialize(query)
+      @query = query
+      @conn = Connection.new
+    end
+
+    def count
+      response = conn.select(params, rows: 0)
+      response.num_found
+    end
+
+    def each
+      raise NotImplementedError, "Subclasses must implement `#each`."
+    end
+
+  end
+end

data/lib/ddr/index/connection.rb

@@ -0,0 +1,17 @@
+module Ddr::Index
+  class Connection < SimpleDelegator
+
+    def initialize
+      super RSolr.connect(ActiveFedora.solr_config)
+    end
+
+    def select(params, extra={})
+      Response.new get("select", params: params.merge(extra))
+    end
+
+    def page(*args)
+      Response.new paginate(*args)
+    end
+
+  end
+end

data/lib/ddr/index/csv_query_result.rb

@@ -0,0 +1,61 @@
+require "csv"
+
+module Ddr::Index
+  class CSVQueryResult < AbstractQueryResult
+
+    MAX_ROWS = 10**8
+
+    COL_SEP    = CSV::DEFAULT_OPTIONS[:col_sep].freeze
+    QUOTE_CHAR = CSV::DEFAULT_OPTIONS[:quote_char].freeze
+
+    SOLR_CSV_OPTS = {
+      "csv.header"       => "false",
+      "csv.mv.separator" => "|",
+      "wt"               => "csv",
+    }.freeze
+
+    CSV_OPTS = {
+      return_headers: true,
+      write_headers:  true,
+    }.freeze
+
+    attr_reader :csv_opts, :solr_csv_opts
+
+    def initialize(query, **opts)
+      super(query)
+
+      @solr_csv_opts = SOLR_CSV_OPTS.dup
+      @solr_csv_opts[:rows] ||= MAX_ROWS
+
+      @csv_opts = CSV_OPTS.dup
+      @csv_opts[:headers] = query.fields.map(&:label)
+
+      # Set column separator and quote character consistently
+      @csv_opts[:col_sep]    = @solr_csv_opts["csv.separator"]    = opts.fetch(:col_sep, COL_SEP)
+      @csv_opts[:quote_char] = @solr_csv_opts["csv.encapsulator"] = opts.fetch(:quote_char, QUOTE_CHAR)
+    end
+
+    def csv
+      CSV.new(data, csv_opts)
+    end
+
+    def each
+      csv.each
+    end
+
+    def to_s
+      csv.string
+    end
+
+    private
+
+    def csv_params
+      params.merge(solr_csv_opts)
+    end
+
+    def data
+      conn.get "select", params: csv_params
+    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 +1,23 @@
+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 "ddr.index.fields.#{base}", default: base.titleize
+    end
+
+  end
+end

data/lib/ddr/index/fields.rb

@@ -0,0 +1,83 @@
+module Ddr::Index
+  module Fields
+
+    include LegacyLicenseFields
+
+    def self.get(name)
+      const_get(name.to_s.upcase, false)
+    end
+
+    ID = UniqueKeyField.instance
+    PID = UniqueKeyField.instance
+
+    ACCESS_ROLE                 = Field.new :access_role, :stored_sortable
+    ACTIVE_FEDORA_MODEL         = Field.new :active_fedora_model, :stored_sortable
+    ADMIN_SET                   = Field.new :admin_set, :stored_sortable
+    ADMIN_SET_FACET             = Field.new :admin_set_facet, :facetable
+    BOX_NUMBER_FACET            = Field.new :box_number_facet, :facetable
+    COLLECTION_FACET            = Field.new :collection_facet, :facetable
+    COLLECTION_URI              = Field.new :collection_uri, :symbol
+    CONTENT_CONTROL_GROUP       = Field.new :content_control_group, :searchable, type: :string
+    CONTENT_SIZE                = Field.new :content_size, :stored_sortable, type: :integer
+    CONTENT_SIZE_HUMAN          = Field.new :content_size_human, :symbol
+    CREATOR_FACET               = Field.new :creator_facet, :facetable
+    DATE_FACET                  = Field.new :date_facet, :facetable
+    DATE_SORT                   = Field.new :date_sort, :sortable
+    DEPOSITOR                   = Field.new :depositor, :stored_sortable
+    DISPLAY_FORMAT              = Field.new :display_format, :stored_sortable
+    DOI                         = Field.new :doi, :symbol
+    EXTRACTED_TEXT              = Field.new :extracted_text, :searchable, type: :text
+    HAS_MODEL                   = Field.new :has_model, :symbol
+    IDENTIFIER_ALL              = Field.new :identifier_all, :symbol
+    INTERNAL_URI                = Field.new :internal_uri, :stored_sortable
+    IS_ATTACHED_TO              = Field.new :is_attached_to, :symbol
+    IS_EXTERNAL_TARGET_FOR      = Field.new :is_external_target_for, :symbol
+    IS_GOVERNED_BY              = Field.new :is_governed_by, :symbol
+    IS_MEMBER_OF                = Field.new :is_member_of, :symbol
+    IS_MEMBER_OF_COLLECTION     = Field.new :is_member_of_collection, :symbol
+    IS_PART_OF                  = Field.new :is_part_of, :symbol
+    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
+    LOCAL_ID                    = Field.new :local_id, :stored_sortable
+    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
+    MULTIRES_IMAGE_FILE_PATH    = Field.new :multires_image_file_path, :stored_sortable
+    OBJECT_PROFILE              = Field.new :object_profile, :displayable
+    OBJECT_STATE                = Field.new :object_state, :stored_sortable
+    OBJECT_CREATE_DATE          = Field.new :system_create, :stored_sortable, type: :date
+    OBJECT_MODIFIED_DATE        = Field.new :system_modified, :stored_sortable, type: :date
+    PERMANENT_ID                = Field.new :permanent_id, :stored_sortable, type: :string
+    PERMANENT_URL               = Field.new :permanent_url, :stored_sortable, type: :string
+    POLICY_ROLE                 = Field.new :policy_role, :symbol
+    PUBLISHER_FACET             = Field.new :publisher_facet, :facetable
+    RESEARCH_HELP_CONTACT       = Field.new :research_help_contact, :stored_sortable
+    RESOURCE_ROLE               = Field.new :resource_role, :symbol
+    SERIES_FACET                = Field.new :series_facet, :facetable
+    SPATIAL_FACET               = Field.new :spatial_facet, :facetable
+    STRUCT_MAPS                 = Field.new :struct_maps, :stored_sortable
+    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, :stored_searchable, type: :integer
+    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_IMAGE_HEIGHT         = Field.new :techmd_image_height, :stored_searchable, type: :integer
+    TECHMD_IMAGE_WIDTH          = Field.new :techmd_image_width, :stored_searchable, type: :integer
+    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
+    TITLE                       = Field.new :title, :stored_sortable
+    TYPE_FACET                  = Field.new :type_facet, :facetable
+    WORKFLOW_STATE              = Field.new :workflow_state, :stored_sortable
+    YEAR_FACET                  = Field.new :year_facet, solr_name: "year_facet_iim"
+
+  end
+end

data/lib/ddr/index/filter.rb

@@ -0,0 +1,48 @@
+module Ddr::Index
+  class Filter
+
+    class << self
+      delegate :where, :raw, :before_days, :before, :present, :absent, to: :new
+    end
+
+    attr_accessor :clauses
+
+    def initialize
+      @clauses = [ ]
+    end
+
+    def where(conditions)
+      clauses = conditions.map do |field, value|
+        if value.respond_to?(:each)
+          QueryClause.or_values(field, *value)
+        else
+          QueryClause.term(field, value)
+        end
+      end
+      raw *clauses
+    end
+
+    # Adds clause (String) w/o escaping
+    def raw(*clauses)
+      self.clauses += clauses
+      self
+    end
+
+    def present(field)
+      raw QueryClause.present(field)
+    end
+
+    def absent(field)
+      raw QueryClause.absent(field)
+    end
+
+    def before(field, date_time)
+      raw QueryClause.before(field, date_time)
+    end
+
+    def before_days(field, days)
+      raw QueryClause.before_days(field, days)
+    end
+
+  end
+end

data/lib/ddr/index/filters.rb

@@ -0,0 +1,19 @@
+module Ddr::Index
+  module Filters
+
+    HAS_CONTENT = Filter.where(Fields::ACTIVE_FEDORA_MODEL => ["Component", "Attachment", "Target"])
+
+    class << self
+      def is_governed_by(pid)
+        Filter.where(Fields::IS_GOVERNED_BY => internal_uri(pid))
+      end
+
+      def internal_uri(pid)
+        ActiveFedora::Base.internal_uri(pid)
+      end
+    end
+
+    private_class_method :internal_uri
+
+  end
+end

data/lib/ddr/index/legacy_license_fields.rb

@@ -0,0 +1,14 @@
+module Ddr::Index
+  module LegacyLicenseFields
+
+    warn "[DEPRECATION] `Ddr::Index::LegacyLicenseFields` is deprecated and will be removed from ddr-models 3.0."
+
+    DEFAULT_LICENSE_DESCRIPTION = Field.new :default_license_description, type: :string
+    DEFAULT_LICENSE_TITLE       = Field.new :default_license_title, type: :string
+    DEFAULT_LICENSE_URL         = Field.new :default_license_url, type: :string
+    LICENSE_DESCRIPTION         = Field.new :license_description, type: :string
+    LICENSE_TITLE               = Field.new :license_title, type: :string
+    LICENSE_URL                 = Field.new :license_url, type: :string
+
+  end
+end

data/lib/ddr/index/query.rb

@@ -0,0 +1,35 @@
+module Ddr::Index
+  class Query
+
+    attr_reader :q, :fields, :filters, :sort, :rows
+
+    delegate :count, :docs, :pids, :each_pid, :all, to: :result
+
+    def inspect
+      "#<#{self.class.name} q=#{q.inspect}, filters=#{filters.inspect}," \
+      " sort=#{sort.inspect}, rows=#{rows.inspect}, fields=#{fields.inspect}>"
+    end
+
+    def to_s
+      URI.encode_www_form(params)
+    end
+
+    def params
+      { q:    q,
+        fq:   filters.map(&:clauses).flatten,
+        fl:   fields.join(","),
+        sort: sort.join(","),
+        rows: rows,
+      }.select { |k, v| v.present? }
+    end
+
+    def result
+      QueryResult.new(self)
+    end
+
+    def csv(**opts)
+      CSVQueryResult.new(self, **opts)
+    end
+
+  end
+end

data/lib/ddr/index/query_builder.rb

@@ -0,0 +1,74 @@
+module Ddr::Index
+  class QueryBuilder
+
+    def self.build
+      builder = new
+      yield builder
+      builder.query
+    end
+
+    def initialize
+      @q       = nil
+      @fields  = [ ]
+      @filters = [ ]
+      @sort    = [ ]
+      @rows    = nil
+    end
+
+    def query
+      Query.new.tap do |qry|
+        instance_variables.each do |var|
+          qry.instance_variable_set(var, instance_variable_get(var))
+        end
+      end
+    end
+
+    def id(pid)
+      q QueryClause.id(pid)
+      limit 1
+    end
+
+    def filter(*fltrs)
+      @filters.push *fltrs
+      self
+    end
+
+    def fields(*flds)
+      @fields.push *flds
+      self
+    end
+
+    def limit(num)
+      @rows = num
+      self
+    end
+
+    def order_by(field, order)
+      @sort << [field, order].join(" ")
+      self
+    end
+
+    def asc(field)
+      order_by field, "asc"
+    end
+
+    def desc(field)
+      order_by field, "desc"
+    end
+
+    def q(q)
+      @q = q
+      self
+    end
+
+    protected
+
+    def method_missing(name, *args, &block)
+      if Filter.respond_to? name
+        return filter Filter.send(name, *args, &block)
+      end
+      super
+    end
+
+  end
+end