dor-services 5.16.0 → 5.17.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 6f1d00053e0c4619d64ea02c747efffeae0ba54e
-  data.tar.gz: 46e5cc9321777766ac8a0e37766c5737b8d3bbd8
+  metadata.gz: 64af1946ede8a667403407ffbcc4fb150c9926bc
+  data.tar.gz: cd676eaee2571de373db7e780b1da2e0f5c18892
 SHA512:
-  metadata.gz: 957be3c6e10cf559b4266d52abb5ef0d2de3457d4ffe7459eb62f3c60a8aee37e47c857e5391d3fd9a97f94f5d06abd74cedd06d4fc66746fe22be7783cea018
-  data.tar.gz: 5d2ae7bb68ab836139fafa2120678980edcc7cb1d49e66e2683ac739cb8dddde3b09a7e955dbd2a03f091feae159895953b39ff64f15c4ebe82cfe6e03a8dd07
+  metadata.gz: 33733ccfdc025b1ccf0eee580ac4c1f73e3a9a17ecb0e1ecd9ab5755d755609ab02546629723e0831d3d920945338ac1a928608fc9de477a09b313c8f9acab6a
+  data.tar.gz: 8c8f067f129b21bf9cdd3fa59a3110748b9b057408a7f3dc2feb0c1276b1e47037ae53751b29f909c297dbef0c6fe0af910f97c7dfe8e623ac722b2b80d060b2

data/config/config_defaults.yml

@@ -47,7 +47,7 @@
   :document_cache_host:
   :document_cache_user:
   :local_stacks_root: /stacks
-  :local_document_cache_root: /purl
+  :local_document_cache_root: /purl/document_cache
   :url:
   :iiif_profile:
 :suri:
@@ -63,6 +63,8 @@
   :shift_age: weekly
 :dor_services:
   :url:
+:dor_indexing_app:
+  :url:
 :indexing_svc:
   :log: 'log/indexing_svc.log'
   :log_date_format_str: '%Y-%m-%d %H:%M:%S.%L'

data/lib/dor-services.rb

@@ -21,9 +21,7 @@ module Dor
     # index is missing the objectType property.
     # @param [String] pid The object's PID
     def load_instance(pid)
-      obj = Dor::Abstract.find pid, cast: false
-      return nil if obj.new_record?
-      obj.adapt_to_cmodel
+      Dor::Abstract.find pid, cast: true
     end
 
     # Get objectType information from solr and load the correct class the first time,
@@ -32,29 +30,17 @@ module Dor
     # @param [String] pid The object's PID
     # @return [Object] the ActiveFedora-modeled object
     def find(pid, opts = {})
-      opts[:rows] = 1 # we know we are going to just pay attention to the .first doc anyway
-      find_all(%(id:"#{pid}"), opts).first || load_instance(pid)
+      load_instance(pid)
     end
 
     # TODO: return enumerable and lazy load_instance
     # TODO: restrict fieldlist (fl) for non-:lightweight queries
     def find_all(query, opts = {})
+      ActiveSupport::Deprecation.warn 'Dor.ensure_models_loaded! is unnecessary and has been deprecated.'
+
       resp = SearchService.query query, opts
       resp['response']['docs'].collect do |solr_doc|
-        doc_version = solr_doc[INDEX_VERSION_FIELD].first rescue '0.0.0'
-        doc_version = Gem::Version.new(doc_version)
-        object_type = Array(solr_doc[ActiveFedora::SolrService.solr_name('objectType', :symbol)]).first
-        object_class = registered_classes[object_type] || ActiveFedora::Base
-        if opts[:lightweight] && doc_version >= Gem::Version.new('3.1.0')
-          begin
-            object_class.load_instance_from_solr solr_doc['id'], solr_doc
-          rescue Exception => e
-            Dor.logger.warn("Exception: '#{e.message}' trying to load #{solr_doc['id']} from solr. Loading from Fedora")
-            load_instance(solr_doc['id'])
-          end
-        else
-          load_instance solr_doc['id']
-        end
+        find solr_doc['id']
       end
     end
 

data/lib/dor/datastreams/content_metadata_ds.rb

@@ -62,7 +62,7 @@ module Dor
 
         # always use title regardless of whether a child label is present
         src_label = doc.create_element('label')
-        src_label.content = src_item.datastreams['DC'].title.first
+        src_label.content = src_item.full_title
 
         # add the extracted label and imageData
         externalFile.add_previous_sibling(src_label)

data/lib/dor/datastreams/desc_metadata_ds.rb

@@ -29,6 +29,10 @@ module Dor
       t.scale       :index_as => [:symbol]
       t.topic       :index_as => [:symbol, :stored_searchable]
       t.abstract    :index_as => [:stored_searchable]
+
+      # 'identifier' conflicts with identityMetadata indexing. Explicitly namespace this one value
+      # until we use #prefix to automatically namespace them for us.
+      t.mods_identifier path: 'identifier', :index_as => [:symbol, :stored_searchable]
     end
 
     def self.xml_template

data/lib/dor/models/abstract.rb

@@ -1,5 +1,11 @@
 module Dor
   class Abstract < ::ActiveFedora::Base
     include Identifiable
+    include Eventable
+    include Governable
+    include Rightsable
+    include Describable
+    include Versionable
+    include Processable
   end
 end

data/lib/dor/models/admin_policy_object.rb

@@ -1,11 +1,6 @@
 module Dor
-  class AdminPolicyObject < ::ActiveFedora::Base
-    include Identifiable
-    include Governable
+  class AdminPolicyObject < Dor::Abstract
     include Editable
-    include Describable
-    include Processable
-    include Versionable
 
     has_many :things, :property => :is_governed_by, :class_name => 'ActiveFedora::Base'
     has_object_type 'adminPolicy'

data/lib/dor/models/collection.rb

@@ -1,14 +1,7 @@
 module Dor
-  class Collection < ::ActiveFedora::Base
-    include Identifiable
-    include Processable
-    include Governable
-    include Describable
-    include Publishable
-    include Versionable
+  class Collection < Dor::Set
     include Releaseable
 
-    has_many :members, :property => :is_member_of_collection, :class_name => 'ActiveFedora::Base'
     has_object_type 'collection'
   end
 end

data/lib/dor/models/describable.rb

@@ -3,6 +3,8 @@ module Dor
     extend ActiveSupport::Concern
 
     MODS_TO_DC_XSLT = Nokogiri::XSLT(File.new(File.expand_path(File.dirname(__FILE__) + "/mods2dc.xslt")))
+    XMLNS_OAI_DC = 'http://www.openarchives.org/OAI/2.0/oai_dc/'.freeze
+    XMLNS_DC = 'http://purl.org/dc/elements/1.1/'.freeze
 
     class CrosswalkError < Exception; end
 
@@ -45,7 +47,7 @@ module Dor
       desc_md = descMetadata.ng_xml.dup(1)
       add_collection_reference(desc_md) if include_collection_as_related_item
       dc_doc = MODS_TO_DC_XSLT.transform(desc_md)
-      dc_doc.xpath('/oai_dc:dc/*[count(text()) = 0]').remove # Remove empty nodes
+      dc_doc.xpath('/oai_dc:dc/*[count(text()) = 0]', oai_dc: XMLNS_OAI_DC).remove # Remove empty nodes
       raise CrosswalkError, "Dor::Item#generate_dublin_core produced incorrect xml (no root):\n#{dc_doc.to_xml}" if dc_doc.root.nil?
       raise CrosswalkError, "Dor::Item#generate_dublin_core produced incorrect xml (no children):\n#{dc_doc.to_xml}" if dc_doc.root.children.size == 0
       dc_doc
@@ -178,7 +180,7 @@ module Dor
         # load the title from the parent's DC.title
         titleInfo = doc.create_element 'titleInfo'
         title = doc.create_element 'title'
-        title.content = parent_item.datastreams['DC'].title.first
+        title.content = Dor::Describable.get_collection_title(parent_item)
         titleInfo << title
         relatedItem << titleInfo
 
@@ -197,12 +199,18 @@ module Dor
     def to_solr(solr_doc = {}, *args)
       solr_doc = super solr_doc, *args
       mods_sources = {
+        'sw_display_title_tesim'      => :sw_title_display,
+        'sw_author_ssim'              => :main_author_w_date,
+        'sw_author_tesim'             => :main_author_w_date,
+        'sw_author_sort_ssi'          => :sw_sort_author,
         'sw_language_ssim'            => :sw_language_facet,
         'sw_language_tesim'           => :sw_language_facet,
         'sw_genre_ssim'               => :sw_genre,
         'sw_genre_tesim'              => :sw_genre,
         'sw_format_ssim'              => :format_main,   # basically sw_typeOfResource_ssim
         'sw_format_tesim'             => :format_main,   # basically sw_typeOfResource_tesim
+        'sw_topic_ssim'               => :topic_facet,
+        'sw_topic_tesim'              => :topic_facet,
         'sw_subject_temporal_ssim'    => :era_facet,
         'sw_subject_temporal_tesim'   => :era_facet,
         'sw_subject_geographic_ssim'  => :geographic_facet,
@@ -220,15 +228,15 @@ module Dor
         dc_doc = generate_dublin_core(include_collection_as_related_item: false)
         # we excluding the generated collection relation here; we instead get the collection
         # title from Dor::Identifiable.
-        dc_doc.xpath('/oai_dc:dc/*').each do |node|
+        dc_doc.xpath('/oai_dc:dc/*', oai_dc: XMLNS_OAI_DC).each do |node|
           add_solr_value(solr_doc, "public_dc_#{node.name}", node.text, :string, [:stored_searchable])
         end
         creator = ''
-        dc_doc.xpath('//dc:creator').each do |node|
+        dc_doc.xpath('//dc:creator', dc: XMLNS_DC).each do |node|
           creator = node.text
         end
         title = ''
-        dc_doc.xpath('//dc:title').each do |node|
+        dc_doc.xpath('//dc:title', dc: XMLNS_DC).each do |node|
           title = node.text
         end
         creator_title = creator + title
@@ -275,15 +283,11 @@ module Dor
     end
 
     def self.get_collection_title(obj)
-      xml = obj.descMetadata.ng_xml
-      title = ''
-      title_node = xml.at_xpath('//mods:mods/mods:titleInfo/mods:title', 'mods' => 'http://www.loc.gov/mods/v3')
-      if title_node
-        title = title_node.content
-        subtitle = xml.at_xpath('//mods:mods/mods:titleInfo/mods:subTitle', 'mods' => 'http://www.loc.gov/mods/v3')
-        title += " (#{subtitle.content})" if subtitle
-      end
-      title
+      obj.full_title
+    end
+
+    def full_title
+      stanford_mods.sw_title_display
     end
 
     private

data/lib/dor/models/embargoable.rb

@@ -1,7 +1,6 @@
 module Dor
   module Embargoable
     extend ActiveSupport::Concern
-    include Dor::Publishable
 
     included do
       has_metadata :name => 'embargoMetadata', :type => Dor::EmbargoMetadataDS, :label => 'Embargo metadata'

data/lib/dor/models/governable.rb

@@ -1,7 +1,6 @@
 module Dor
   module Governable
     extend ActiveSupport::Concern
-    include Rightsable
 
     included do
       belongs_to :admin_policy_object, :property => :is_governed_by, :class_name => 'Dor::AdminPolicyObject'

data/lib/dor/models/identifiable.rb

@@ -2,7 +2,6 @@ module Dor
   module Identifiable
     extend ActiveSupport::Concern
     include SolrDocHelper
-    include Eventable
 
     included do
       has_metadata :name => 'DC', :type => SimpleDublinCoreDs, :label => 'Dublin Core Record for self object'
@@ -184,9 +183,7 @@ module Dor
     def get_related_obj_display_title(related_obj, default_title)
       return default_title unless related_obj
 
-      desc_md_ds = related_obj.datastreams['descMetadata']
-      desc_md_ds_title = desc_md_ds ? desc_md_ds.title_info.main_title.first : nil
-      desc_md_ds_title.present? ? desc_md_ds_title : default_title
+      related_obj.full_title || default_title
     end
 
     # a regex that can be used to identify the last part of a druid (e.g. oo000oo0001)

data/lib/dor/models/item.rb

@@ -1,11 +1,11 @@
 module Dor
-  class Item < ::ActiveFedora::Base
-    include Processable
+  class Item < Dor::Abstract
     include Shelvable
-    include Embargoable # implies Publishable implies Identifiable, Describable, Governable, Rightsable ...
+    include Embargoable
+    include Publishable
+    include Itemizable
     include Preservable
     include Assembleable
-    include Versionable
     include Contentable
     include Geoable
     include Releaseable

data/lib/dor/models/publishable.rb

@@ -4,11 +4,6 @@ require 'fileutils'
 module Dor
   module Publishable
     extend ActiveSupport::Concern
-    include Identifiable
-    include Governable
-    include Describable
-    include Itemizable
-    include Rightsable
 
     # Compute the thumbnail for this object following the rules at https://consul.stanford.edu/display/chimera/The+Rules+of+Thumb
     # @return [String] the computed thumb filename, with the druid prefix and a slash in front of it, e.g. oo000oo0001/filenamewith space.jp2

data/lib/dor/models/releaseable.rb

@@ -4,7 +4,6 @@ require 'retries'
 module Dor
   module Releaseable
     extend ActiveSupport::Concern
-    include Itemizable
 
     # Add release tags to an item and initialize the item release workflow
     # Each tag should be of the form !{:tag => 'Fitch : Batch2', :what => 'self', :to => 'Searchworks', :who => 'petucket', :release => true}
@@ -80,7 +79,7 @@ module Dor
       return_tags = release_nodes || {}
       collections.each do |collection|
         next if collection.id == id  # recursive, so parents of parents are found, but we need to avoid an infinite loop if the collection references itself (i.e. bad data)
-        return_tags = combine_two_release_tag_hashes(return_tags, Dor.find(collection.id).get_release_tags_for_item_and_all_governing_sets)
+        return_tags = combine_two_release_tag_hashes(return_tags, collection.get_release_tags_for_item_and_all_governing_sets)
       end
       return_tags
     end

data/lib/dor/models/set.rb

@@ -1,11 +1,6 @@
 module Dor
-  class Set < ::ActiveFedora::Base
-    include Identifiable
-    include Processable
-    include Governable
-    include Describable
+  class Set < Dor::Abstract
     include Publishable
-    include Versionable
 
     has_many :members, :property => :is_member_of_collection, :class_name => 'ActiveFedora::Base'
     has_object_type 'set'

data/lib/dor/models/shelvable.rb

@@ -3,7 +3,6 @@ require 'moab/stanford'
 module Dor
   module Shelvable
     extend ActiveSupport::Concern
-    include Itemizable
 
     # Push file changes for shelve-able files into the stacks
     def shelve

data/lib/dor/models/versionable.rb

@@ -1,7 +1,6 @@
 module Dor
   module Versionable
     extend ActiveSupport::Concern
-    include Processable
 
     included do
       has_metadata :name => 'versionMetadata', :type => Dor::VersionMetadataDS, :label => 'Version Metadata', :autocreate => true

data/lib/dor/models/workflow_object.rb

@@ -1,9 +1,7 @@
 require 'dor/datastreams/workflow_definition_ds'
 
 module Dor
-  class WorkflowObject < ::ActiveFedora::Base
-    include Identifiable
-    include Governable
+  class WorkflowObject < Dor::Abstract
     @@xml_cache  = {}
     @@repo_cache = {}
 

data/lib/dor/services/indexing_service.rb

@@ -2,6 +2,8 @@ require 'benchmark'
 
 module Dor
   class IndexingService
+    class ReindexError < RuntimeError; end
+
     ##
     # Returns a Logger instance for recording info about indexing attempts
     # @yield attempt to execute 'entry_id_block' and use the result as an extra identifier for the log
@@ -30,6 +32,26 @@ module Dor
       solr_doc
     end
 
+    # Use the dor-indexing-app service to reindex a pid
+    # @param [String] `pid` the druid
+    # @raise [ReindexError] on failure
+    def self.reindex_pid_remotely(pid)
+      pid = "druid:#{pid}" unless pid =~ /^druid:/
+      realtime = Benchmark.realtime do
+        with_retries(max_tries: 3, rescue: [ RestClient::Exception, Errno::ECONNREFUSED ]) do
+          RestClient.post(URI.join(Config.dor_indexing_app.url, "reindex/#{pid}"), '')
+        end
+      end
+      default_index_logger.info "successfully updated index for #{pid} in #{'%.3f' % realtime}s"
+    rescue RestClient::Exception, Errno::ECONNREFUSED => e
+      msg = "failed to reindex #{pid}: #{e}"
+      default_index_logger.error msg
+      raise ReindexError.new(msg)
+    rescue => e
+      default_index_logger.error "failed to reindex #{pid}: #{e}"
+      raise
+    end
+
     # retrieves a single Dor object by pid, indexes the object to solr, does some logging
     # (will use a default logger if one is not provided).  doesn't commit automatically.
     #

data/lib/dor/version.rb

@@ -1,3 +1,3 @@
 module Dor
-  VERSION = '5.16.0'.freeze
+  VERSION = '5.17.1'.freeze
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: dor-services
 version: !ruby/object:Gem::Version
-  version: 5.16.0
+  version: 5.17.1
 platform: ruby
 authors:
 - Michael Klein
@@ -14,7 +14,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2016-11-15 00:00:00.000000000 Z
+date: 2016-11-18 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: active-fedora