dor-services 2.2.4 → 4.4.10

data/lib/dor/models/identifiable.rb

@@ -0,0 +1,211 @@
+module Dor
+  module Identifiable
+    extend ActiveSupport::Concern
+    include SolrDocHelper
+    include Eventable
+    include Upgradable
+
+    included do
+      has_metadata :name => "DC", :type => SimpleDublinCoreDs, :label => 'Dublin Core Record for self object'
+      has_metadata :name => "identityMetadata", :type => Dor::IdentityMetadataDS, :label => 'Identity Metadata'
+    end
+
+    module ClassMethods
+      attr_reader :object_type
+      def has_object_type str
+        @object_type = str
+        Dor.registered_classes[str] = self
+      end
+    end
+
+    def initialize attrs={}
+      if Dor::Config.suri.mint_ids
+        unless attrs[:pid]
+          attrs = attrs.merge!({:pid=>Dor::SuriService.mint_id, :new_object => true})
+        end
+      end
+      super
+    end
+
+    # helper method to get the tags as an array
+    def tags
+      self.identityMetadata.tag
+    end
+
+    # helper method to get just the content type tag
+    def content_type_tag
+      content_tag=tags.select {|tag| tag.include?('Process : Content Type')}
+      content_tag.size == 1 ? content_tag[0].split(':').last.strip : ""
+    end
+
+    # Syntactic sugar for identifying applied DOR Concerns
+    # e.g., obj.is_identifiable? is the same as obj.is_a?(Dor::Identifiable)
+    def method_missing sym, *args
+      if sym.to_s =~ /^is_(.+)\?$/
+        begin
+          klass = Dor.const_get $1.capitalize.to_sym
+          return self.is_a?(klass)
+        rescue NameError
+          return false
+        end
+      else
+        super
+      end
+    end
+
+    @@collection_hash={}
+    @@apo_hash={}
+    @@hydrus_apo_hash={}
+    @@hydrus_collection_hash={}
+    def to_solr(solr_doc=Hash.new, *args)
+      self.assert_content_model
+      super(solr_doc)
+      solr_doc[Dor::INDEX_VERSION_FIELD] = Dor::VERSION
+      solr_doc[solr_name('indexed_at',:date)] = Time.now.utc.xmlschema
+      add_solr_value(solr_doc, 'indexed_day', Time.now.beginning_of_day.utc.xmlschema, :string, [:searchable, :facetable])
+      datastreams.values.each do |ds|
+        unless ds.new?
+          add_solr_value(solr_doc,'ds_specs',ds.datastream_spec_string,:string,[:displayable])
+        end
+      end
+      add_solr_value(solr_doc, 'title_sort', self.label, :string, [:sortable])
+      rels_doc = Nokogiri::XML(self.datastreams['RELS-EXT'].content)
+      apos=rels_doc.search('//rdf:RDF/rdf:Description/hydra:isGovernedBy','hydra' => 'http://projecthydra.org/ns/relations#', 'fedora' => 'info:fedora/fedora-system:def/relations-external#', 'rdf' => 'http://www.w3.org/1999/02/22-rdf-syntax-ns#' 	)
+      apos.each do |apo_node|
+        druid=apo_node['rdf:resource']
+        if druid
+          druid=druid.gsub('info:fedora/','')
+          if @@apo_hash.has_key? druid or @@hydrus_apo_hash.has_key? druid
+            add_solr_value(solr_doc, "hydrus_apo_title", @@hydrus_apo_hash[druid], :string, [:searchable, :facetable]) if @@hydrus_apo_hash.has_key? druid
+            add_solr_value(solr_doc, "apo_title", @@apo_hash[druid] , :string, [:searchable, :facetable]) if @@apo_hash.has_key? druid 
+          else
+            begin
+              apo_object=Dor.find(druid)
+              if apo_object.tags.include? 'Project : Hydrus'
+                add_solr_value(solr_doc, "hydrus_apo_title", apo_object.label, :string, [:searchable, :facetable])
+                @@hydrus_apo_hash[druid]=apo_object.label
+              else
+                add_solr_value(solr_doc, "apo_title", apo_object.label, :string, [:searchable, :facetable])
+                @@apo_hash[druid]=apo_object.label
+              end
+            rescue
+              add_solr_value(solr_doc, "apo_title", druid, :string, [:searchable, :facetable])
+            end
+          end
+        end
+      end
+      collections=rels_doc.search('//rdf:RDF/rdf:Description/fedora:isMemberOfCollection','fedora' => 'info:fedora/fedora-system:def/relations-external#', 'rdf' => 'http://www.w3.org/1999/02/22-rdf-syntax-ns#' 	)
+      collections.each do |collection_node| 
+        druid=collection_node['rdf:resource']
+        if(druid)
+          druid=druid.gsub('info:fedora/','')
+          if @@collection_hash.has_key? druid or @@hydrus_collection_hash.has_key? druid
+            add_solr_value(solr_doc, "hydrus_collection_title", @@hydrus_collection_hash[druid], :string, [:searchable, :facetable]) if @@hydrus_collection_hash.has_key? druid
+            add_solr_value(solr_doc, "collection_title", @@collection_hash[druid], :string, [:searchable, :facetable]) if @@collection_hash.has_key? druid
+          else
+            begin
+              collection_object=Dor.find(druid)
+              if collection_object.tags.include? 'Project : Hydrus'
+                add_solr_value(solr_doc, "hydrus_collection_title", collection_object.label, :string, [:searchable, :facetable])
+                @@hydrus_collection_hash[druid]=collection_object.label
+              else
+                add_solr_value(solr_doc, "collection_title", collection_object.label, :string, [:searchable, :facetable])
+                @@collection_hash[druid]=collection_object.label
+              end
+            rescue
+              add_solr_value(solr_doc, "collection_title", druid, :string, [:searchable, :facetable])
+            end
+          end
+        end
+      end 
+      # Fix for ActiveFedora 3.3 to ensure all date fields are properly formatted as UTC XML Schema datetime strings
+      solr_doc.each_pair { |k,v| 
+        if k =~ /_dt|_date$/
+          if v.is_a?(Array)
+            solr_doc[k] = v.collect { |t| Time.parse(t.to_s).utc.xmlschema }
+          else
+            solr_doc[k] = Time.parse(v.to_s).utc.xmlschema
+          end
+        end
+      }
+
+      solr_doc
+    end
+    def set_source_id(source_id)
+      self.identityMetadata.sourceId = source_id
+    end
+
+    def add_other_Id(type,val)
+      if self.identityMetadata.otherId(type).length>0        
+        raise 'There is an existing entry for '+node_name+', consider using update_other_identifier.'
+      end
+      identity_metadata_ds = self.identityMetadata
+      identity_metadata_ds.add_otherId(type+':'+val)
+    end
+
+    def update_other_Id(type,new_val, val=nil)
+      identity_metadata_ds = self.identityMetadata
+      ds_xml=identity_metadata_ds.ng_xml
+      #split the thing they sent in to find the node name
+      updated=false
+      ds_xml.search('//otherId[@name=\''+type+'\']').each do |node|
+        if node.content==val or val==nil
+          node.content=new_val
+          updated=true
+        end
+      end
+      return updated
+    end
+
+    def remove_other_Id(type,val=nil)
+      ds_xml=self.identityMetadata.ng_xml
+      #split the thing they sent in to find the node name
+      removed=false
+
+      ds_xml.search('//otherId[@name=\''+type+'\']').each do |node|
+        if node.content===val or val==nil
+          node.remove
+          removed=true
+        end
+      end
+      return removed
+    end
+
+    def add_tag(tag)
+      identity_metadata_ds = self.identityMetadata
+      prefix=tag.split(/:/).first
+      identity_metadata_ds.tags.each do |existing_tag|
+        if existing_tag.split(/:/).first ==prefix 
+          raise 'An existing tag ('+existing_tag+') has the same prefix, consider using update_tag?'
+        end
+      end
+      identity_metadata_ds.add_value(:tag,tag)
+    end
+
+    def remove_tag(tag)
+      identity_metadata_ds = self.identityMetadata
+      ds_xml=identity_metadata_ds.ng_xml
+      removed=false
+      ds_xml.search('//tag').each do |node|
+        if node.content===tag
+          node.remove
+          removed=true
+        end
+      end
+      return removed
+    end
+
+    def update_tag(old_tag,new_tag)
+      identity_metadata_ds = self.identityMetadata
+      ds_xml=identity_metadata_ds.ng_xml
+      updated=false
+      ds_xml.search('//tag').each do |node|
+        if node.content==old_tag
+          node.content=new_tag
+          updated = true
+        end
+      end
+      return updated 
+    end
+  end
+end

data/lib/dor/models/item.rb

@@ -0,0 +1,44 @@
+module Dor
+  module BasicItem
+    extend ActiveSupport::Concern
+    
+    include Identifiable
+    include Processable
+    include Governable
+    include Describable
+    include Publishable
+    include Shelvable
+    include Embargoable
+    include Preservable
+    include Assembleable
+    include Versionable
+    include Contentable
+    include Discoverable
+    
+  end
+  
+  class Abstract < ::ActiveFedora::Base
+    include Identifiable
+  end
+
+  class Item < ::ActiveFedora::Base
+    include BasicItem
+    has_object_type 'item'
+  end
+end
+
+# Describable  = Descriptive metadata.
+# Embargoable  = Time limits and processes for embargoed materials.
+# Governable   = Relationships to collections and codified administrative policies.
+# Identifiable = Object identity and source metadata.
+# Itemizable   = Hierarchical content metadata.
+# Preservable  = Provenance and technical metadata; preservation repository transfer.
+# Processable  = Workflow.
+# Publishable  = Transfer of metadata to discovery and access systems.
+# Shelvable    = Transfer of content to digital stacks.
+# Upgradable   = Remediation of existing objects when content standards change.
+
+# Required for all DOR objects:
+#   - Identifiable
+#   - Governable
+#   - Describable

data/lib/dor/models/itemizable.rb

@@ -0,0 +1,66 @@
+require 'fileutils'
+require 'uri'
+
+module Dor
+  module Itemizable
+    extend ActiveSupport::Concern
+
+    included do
+      has_metadata :name => "contentMetadata", :type => Dor::ContentMetadataDS, :label => 'Content Metadata', :control_group => 'M'
+    end
+    
+    DIFF_FILENAME = 'cm_inv_diff'
+    DIFF_QUERY = DIFF_FILENAME.gsub('_', '-')
+
+    # Deletes all cm_inv_diff files in the workspace for the Item
+    def clear_diff_cache
+      if Dor::Config.stacks.local_workspace_root.nil?
+        raise ArgumentError, 'Missing Dor::Config.stacks.local_workspace_root' 
+      end
+      druid = DruidTools::Druid.new(self.pid, Dor::Config.stacks.local_workspace_root)
+      diff_pattern = File.join(druid.temp_dir, DIFF_FILENAME + '.*')
+      FileUtils.rm_f Dir.glob(diff_pattern)
+    end
+    
+    # Retrieves file difference manifest for contentMetadata from SDR
+    #
+    # @param [String] subset keyword for file attributes :shelve, :preserve, :publish. Default is :all.
+    # @param [String] version
+    # @return [String] XML contents of cm_inv_diff manifest
+    def get_content_diff(subset = :all, version = nil)
+      if Dor::Config.stacks.local_workspace_root.nil?
+        raise Dor::ParameterError, 'Missing Dor::Config.stacks.local_workspace_root' 
+      end
+      unless %w{all shelve preserve publish}.include?(subset.to_s)
+        raise Dor::ParameterError, "Invalid subset value: #{subset}"
+      end
+
+      basename = version.nil? ? "#{DIFF_FILENAME}.#{subset}.xml" : "#{DIFF_FILENAME}.#{subset}.#{version}.xml"
+      druid = DruidTools::Druid.new(self.pid, Dor::Config.stacks.local_workspace_root)
+      diff_cache = File.join(druid.temp_dir, basename)
+      # check for cached copy before contacting SDR
+      if File.exists? diff_cache
+        File.read(diff_cache)
+      else
+        # fetch content metadata inventory difference from SDR
+        if Dor::Config.sdr.rest_client.nil?
+          raise Dor::ParameterError, 'Missing Dor::Config.sdr.rest_client'
+        end
+        sdr_client = Dor::Config.sdr.rest_client
+        current_content = self.datastreams['contentMetadata'].content
+        if current_content.nil?
+          raise Dor::Exception, "Missing contentMetadata datastream"
+        end
+        query_string = { :subset => subset.to_s }
+        query_string[:version] = version.to_s unless version.nil?
+        query_string = URI.encode_www_form(query_string)
+        sdr_query = "objects/#{self.pid}/#{DIFF_QUERY}?#{query_string}"
+        response = sdr_client[sdr_query].post(current_content, :content_type => 'application/xml')
+        # cache response
+        File.open(diff_cache, 'w') { |f| f << response }
+        response
+      end
+    end
+    
+  end
+end

data/lib/dor/{mods2dc.xslt → models/mods2dc.xslt}

@@ -42,7 +42,7 @@ Version 1.0	2007-05-04 Tracy Meehleib <tmee@loc.gov>
 	<xsl:template match="/">
 		<xsl:choose>
 		<xsl:when test="//mods:modsCollection">			
-			<srw_dc:dcCollection xsi:schemaLocation="info:srw/schema/1/dc-schema http://www.loc.gov/standards/sru/dc-schema.xsd">
+			<srw_dc:dcCollection xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:srw_dc="info:srw/schema/1/dc-schema" xsi:schemaLocation="info:srw/schema/1/dc-schema http://www.loc.gov/standards/sru/dc-schema.xsd">
 				<xsl:apply-templates/>
 			<xsl:for-each select="mods:modsCollection/mods:mods">			
 				<srw_dc:dc xsi:schemaLocation="info:srw/schema/1/dc-schema http://www.loc.gov/standards/sru/dc-schema.xsd">
@@ -54,7 +54,7 @@ Version 1.0	2007-05-04 Tracy Meehleib <tmee@loc.gov>
 		</xsl:when>
 		<xsl:otherwise>
 			<xsl:for-each select="mods:mods">
-			<oai_dc:dc xsi:schemaLocation="http://www.openarchives.org/OAI/2.0/oai_dc/ http://www.openarchives.org/OAI/2.0/oai_dc.xsd">
+			<oai_dc:dc xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:srw_dc="info:srw/schema/1/dc-schema" xsi:schemaLocation="http://www.openarchives.org/OAI/2.0/oai_dc/ http://www.openarchives.org/OAI/2.0/oai_dc.xsd">
 				<xsl:apply-templates/>
 
 			</oai_dc:dc>
@@ -184,6 +184,8 @@ Version 1.0	2007-05-04 Tracy Meehleib <tmee@loc.gov>
 
 	<xsl:template match="mods:abstract | mods:tableOfContents | mods:note">
 		<dc:description>
+			<xsl:copy-of select="@type"/>
+			<xsl:copy-of select="@displayLabel"/>
 			<xsl:value-of select="."/>
 		</dc:description>
 	</xsl:template>
@@ -191,7 +193,7 @@ Version 1.0	2007-05-04 Tracy Meehleib <tmee@loc.gov>
 	<xsl:template match="mods:originInfo">
 		<xsl:apply-templates select="*[@point='start']"/>
 		<xsl:for-each
-			select="mods:dateIssued[@point!='start' and @point!='end'] |mods:dateCreated[@point!='start' and @point!='end'] | mods:dateCaptured[@point!='start' and @point!='end'] | mods:dateOther[@point!='start' and @point!='end']">
+			select="mods:dateIssued[(count(@point) = 0) or (@point!='start' and @point!='end')] |mods:dateCreated[(count(@point) = 0) or (@point!='start' and @point!='end')] | mods:dateCaptured[(count(@point) = 0) or (@point!='start' and @point!='end')] | mods:dateOther[(count(@point) = 0) or (@point!='start' and @point!='end')]">
 			<dc:date>
 				<xsl:value-of select="."/>
 			</dc:date>
@@ -335,7 +337,7 @@ Version 1.0	2007-05-04 Tracy Meehleib <tmee@loc.gov>
 		</xsl:choose>
 	</xsl:template>
 
-	<xsl:template match="mods:location">
+	<xsl:template match="mods:location[mods:url]">
 		<dc:identifier>
 			<xsl:for-each select="mods:url">
 				<xsl:value-of select="."/>
@@ -368,8 +370,12 @@ Version 1.0	2007-05-04 Tracy Meehleib <tmee@loc.gov>
 			<xsl:when test="@type='series'"/>
 			<xsl:otherwise>
 				<dc:relation>
+					<xsl:if test="mods:location/mods:url">
+						<xsl:attribute name="type">url</xsl:attribute>
+						<xsl:attribute name="href"><xsl:value-of select="mods:location/mods:url"/></xsl:attribute>
+					</xsl:if>
 					<xsl:for-each
-						select="mods:titleInfo/mods:title | mods:identifier | mods:location/mods:url">
+						select="mods:titleInfo/mods:title | mods:identifier">
 						<xsl:if test="normalize-space(.)!= ''">
 							<xsl:value-of select="."/>
 
@@ -381,6 +387,27 @@ Version 1.0	2007-05-04 Tracy Meehleib <tmee@loc.gov>
 		</xsl:choose>
 	</xsl:template>
 
+
+    <xsl:template match="mods:location[mods:physicalLocation[@type='repository']]">
+        <dc:relation type="repository">
+            <xsl:value-of select="normalize-space(.)"/>
+        </dc:relation>
+    </xsl:template>
+	<xsl:template match="mods:relatedItem[@type='host']">
+      <xsl:choose>
+        <xsl:when test="mods:location/mods:physicalLocation[@type='location']">
+          <dc:relation type="location">
+              <xsl:value-of select="normalize-space(.)"/>
+          </dc:relation>
+        </xsl:when>
+        <xsl:when test="mods:typeOfResource[@collection='yes']">
+          <dc:relation type="collection">
+              <xsl:value-of select="normalize-space(mods:titleInfo/mods:title)"/>
+          </dc:relation>
+        </xsl:when>
+      </xsl:choose>
+    </xsl:template>
+
 	<xsl:template match="mods:accessCondition">
 
 		<dc:rights>
@@ -391,17 +418,18 @@ Version 1.0	2007-05-04 Tracy Meehleib <tmee@loc.gov>
 	<xsl:template name="name">
 		<xsl:variable name="name">
 			<xsl:for-each select="mods:namePart[not(@type)]">
-				<xsl:value-of select="."/>
-
-				<xsl:text> </xsl:text>
+                <xsl:if test="normalize-space(.)!= ''">
+                    <xsl:text> </xsl:text>
+                    <xsl:value-of select="normalize-space(.)"/>
+                    <xsl:text></xsl:text>
+                </xsl:if>
 			</xsl:for-each>
 			<xsl:value-of select="mods:namePart[@type='family']"/>
 			<xsl:if test="mods:namePart[@type='given']">
 				<xsl:text>, </xsl:text>
-				<xsl:value-of select="mods:namePart[@type='given']"/>
+				<xsl:value-of select="normalize-space(mods:namePart[@type='given'])"/>
 			</xsl:if>
 			<xsl:if test="mods:namePart[@type='date']">
-
 				<xsl:text>, </xsl:text>
 				<xsl:value-of select="mods:namePart[@type='date']"/>
 				<xsl:text/>
@@ -409,7 +437,6 @@ Version 1.0	2007-05-04 Tracy Meehleib <tmee@loc.gov>
 			<xsl:if test="mods:displayForm">
 				<xsl:text> (</xsl:text>
 				<xsl:value-of select="mods:displayForm"/>
-
 				<xsl:text>) </xsl:text>
 			</xsl:if>
 			<xsl:for-each select="mods:role[mods:roleTerm[@type='text']!='creator']">
@@ -434,7 +461,7 @@ Version 1.0	2007-05-04 Tracy Meehleib <tmee@loc.gov>
 		<xsl:value-of select="."/>-<xsl:value-of select="../mods:temporal[@point='end']"/>
 	</xsl:template>
 	
-	<xsl:template match="mods:temporal[@point!='start' and @point!='end']  ">
+	<xsl:template match="mods:temporal[(count(@point) = 0) or (@point!='start' and @point!='end')]  ">
 		<xsl:value-of select="."/>
 	</xsl:template>
 	

data/lib/dor/models/preservable.rb

@@ -0,0 +1,50 @@
+require 'tmpdir'
+
+module Dor
+  module Preservable
+    extend ActiveSupport::Concern
+
+    included do
+      has_metadata :name => "provenanceMetadata", :type => ActiveFedora::OmDatastream, :label => 'Provenance Metadata'
+      has_metadata :name => "technicalMetadata", :type => ActiveFedora::OmDatastream, :label => 'Technical Metadata', :control_group => 'M'
+    end
+
+    def build_provenanceMetadata_datastream(workflow_id, event_text)
+      workflow_provenance = create_workflow_provenance(workflow_id, event_text)
+      dsname = 'provenanceMetadata'
+      ds = datastreams[dsname]
+      unless datastreams.keys.include?(dsname)
+        ds.label = 'Provenance Metadata'
+      end
+      ds.ng_xml = workflow_provenance
+			ds.content=ds.ng_xml.to_s
+      ds.save
+    end
+
+    def build_technicalMetadata_datastream(ds=nil)
+      TechnicalMetadataService.add_update_technical_metadata(self)
+    end
+
+    def sdr_ingest_transfer(agreement_id)
+      SdrIngestService.transfer(self,agreement_id)
+    end
+
+
+    # @return [Nokogiri::Document]
+    def create_workflow_provenance(workflow_id, event_text)
+      builder = Nokogiri::XML::Builder.new do |xml|
+        xml.provenanceMetadata(:objectId => self.pid) {
+          xml.agent(:name => 'DOR') {
+            xml.what(:object => self.pid) {
+              xml.event(:who => "DOR-#{workflow_id}", :when => Time.new.iso8601) {
+                xml.text(event_text)
+              }
+            }
+          }
+        }
+      end
+      builder.doc
+    end
+
+  end
+end