[DRAFT] Publish collection membership data to purl

app/services/publish_metadata_service.rb

@@ -20,6 +20,7 @@ def publish
     release_tags = ReleaseTags.for(item: item)
 
     transfer_metadata(release_tags)
+    bookkeep_collections
     publish_notify_on_success
   end
 
@@ -37,8 +38,61 @@ def transfer_metadata(release_tags)
     transfer_to_document_store(PublicDescMetadataService.new(item).to_xml, 'mods')
   end
 
+  # Maintain bidirectional symlinks from:
+  #  - an item to the collections it belongs to
+  #  - a collection to the items within
+  def bookkeep_collections
+    FileUtils.mkdir_p(item_collections_dir)
+    existing_collections = Dir.children(item_collections_dir)
+
+    # Write bidirectional symlinks for collection membership
+    item.collections.each do |coll|
+      Rails.logger.debug("[Publish][#{item.pid}] Adding collection association with #{coll.pid}")
+      collection_items_dir = collection_member_dir(coll.pid)
+      FileUtils.mkdir_p(collection_items_dir)
+      FileUtils.ln_s(item.content_dir, File.join(collection_items_dir, local_part(item.pid)), force: true)
+      FileUtils.ln_s(coll.content_dir, File.join(item_collections_dir, local_part(coll.pid)), force: true)
+    end
+
+    # Remove bidirectional collection membership for collections no longer asserted
+    (existing_collections - item.collections.map { |coll| local_part(coll.pid) }).each do |coll_pid|
+      Rails.logger.debug("[Publish][#{item.pid}] Removing collection association with #{coll_pid}")
+      FileUtils.rm(File.join(item_collections_dir, coll_pid), force: true)
+
+      collection_items_dir = collection_member_dir(coll_pid)
+      next unless Dir.exist? collection_items_dir
+
+      FileUtils.rm(File.join(collection_items_dir, local_part(item.pid)), force: true)
+    end
+  end
+
+  # Remove all collection membership symlinks
+  def unbookkeep_collections
+    return unless Dir.exist? item_collections_dir
+
+    existing_collections = Dir.children(item_collections_dir)
+    existing_collections.each do |coll_pid|
+      collection_items_dir = collection_member_dir(coll_pid)
+      next unless Dir.exist? collection_items_dir
+
+      FileUtils.rm(File.join(collection_items_dir, local_part(item.pid)))
+    end
+
+    members_dir = collection_member_dir(item.pid)
+    if Dir.exist? members_dir
+      existing_members = Dir.children(members_dir)
+      existing_members.each do |item_pid|
+        item_dir = item_collections_dir(item_pid)
+        next unless Dir.exist? item_dir
+
+        FileUtils.rm(File.join(item_dir, local_part(item.pid)))
+      end
+    end
+  end
+
   # Clear out the document cache for this item
   def unpublish
+    unbookkeep_collections
     PruneService.new(druid: purl_druid).prune!
     publish_delete_on_success
   end
@@ -62,6 +116,24 @@ def purl_druid
     @purl_druid ||= DruidTools::PurlDruid.new item.pid, Settings.stacks.local_document_cache_root
   end
 
+  # Get the collection membership directory for an item
+  def item_collections_dir(item_pid = nil)
+    item_druid = if item_pid
+      DruidTools::PurlDruid.new item_pid, Settings.stacks.local_document_cache_root
+    else
+      purl_druid
+    end
+
+    File.join(item_druid.content_dir, 'is_member_of_collection')
+  end
+
+  # Get the collection members directory for a collection
+  def collection_member_dir(collection_pid)
+    collection_druid = DruidTools::PurlDruid.new collection_pid, Settings.stacks.local_document_cache_root
+    File.join(collection_druid.content_dir, 'has_member_of_collection')
+  end
+
+
   ##
   # When publishing a PURL, we notify purl-fetcher of changes.
   #
@@ -77,10 +149,12 @@ def publish_delete_on_success
   end
 
   def purl_services_url
-    id = item.pid.gsub(/^druid:/, '')
-
     raise 'You have not configured perl-fetcher (Settings.purl_services_url).' unless Settings.purl_services_url
 
-    "#{Settings.purl_services_url}/purls/#{id}"
+    "#{Settings.purl_services_url}/purls/#{local_part(item.pid)}"
+  end
+
+  def local_part(pid)
+    Dor::PidUtils.remove_druid_prefix(pid)
   end
 end

spec/services/publish_metadata_service_spec.rb

@@ -67,12 +67,44 @@
         druid1 = DruidTools::Druid.new item.pid, purl_root
         druid1.mkdir
         File.open(File.join(druid1.path, 'tmpfile'), 'w') { |f| f.write 'junk' }
+        expect(service).to receive(:unbookkeep_collections)
+
         service.publish
         expect(File).not_to exist(druid1.path) # it should now be gone
         expect(WebMock).to have_requested(:delete, 'example.com/purl/purls/ab123cd4567')
       end
     end
 
+    let(:release_tags) do
+      { 'Searchworks' => { 'release' => true }, 'Some_special_place' => { 'release' => true } }
+    end
+
+    # the individual steps are tested below
+    context 'a public item' do
+      before do
+        allow(ReleaseTags).to receive(:for).and_return(release_tags)
+        item.rightsMetadata.content = "<rightsMetadata><access type='discover'><machine><world/></machine></access></rightsMetadata>"
+      end
+
+      it 'calls the appropriate subfunctions' do
+        allow(service).to receive(:transfer_metadata)
+        allow(service).to receive(:bookkeep_collections)
+        allow(service).to receive(:publish_notify_on_success)
+
+        service.publish
+
+        expect(service).to have_received(:transfer_metadata).with(release_tags)
+        expect(service).to have_received(:bookkeep_collections)
+        expect(service).to have_received(:publish_notify_on_success)
+      end
+    end
+  end
+
+  describe '#transfer_metadata' do
+    before do
+      allow(OpenURI).to receive(:open_uri).with('https://purl-test.stanford.edu/ab123cd4567.xml').and_return('<xml/>')
+    end
+
     context 'copies to the document cache' do
       let(:mods) do
         <<-EOXML
@@ -102,7 +134,6 @@
           expect_any_instance_of(described_class).to receive(:transfer_to_document_store).with(/<oai_dc:dc/, 'dc')
           expect_any_instance_of(described_class).to receive(:transfer_to_document_store).with(/<publicObject/, 'public')
           expect_any_instance_of(described_class).to receive(:transfer_to_document_store).with(/<mods:mods/, 'mods')
-          expect_any_instance_of(described_class).to receive(:publish_notify_on_success).with(no_args)
         end
 
         let(:release_tags) do
@@ -111,7 +142,7 @@
 
         it 'identityMetadta, contentMetadata, rightsMetadata, generated dublin core, and public xml' do
           item.rightsMetadata.content = "<rightsMetadata><access type='discover'><machine><world/></machine></access></rightsMetadata>"
-          service.publish
+          service.send(:transfer_metadata, release_tags)
           expect(DublinCoreService).to have_received(:new).with(item)
           expect(PublicXmlService).to have_received(:new).with(item, released_for: release_tags)
           expect(PublicDescMetadataService).to have_received(:new).with(item)
@@ -120,7 +151,7 @@
         it 'even when rightsMetadata uses xml namespaces' do
           item.rightsMetadata.content = %q(<rightsMetadata xmlns="http://hydra-collab.stanford.edu/schemas/rightsMetadata/v1">
             <access type='discover'><machine><world/></machine></access></rightsMetadata>)
-          service.publish
+          service.send(:transfer_metadata, release_tags)
         end
       end
 
@@ -140,16 +171,34 @@
           expect_any_instance_of(described_class).to receive(:transfer_to_document_store).with(/<oai_dc:dc/, 'dc')
           expect_any_instance_of(described_class).to receive(:transfer_to_document_store).with(/<publicObject/, 'public')
           expect_any_instance_of(described_class).to receive(:transfer_to_document_store).with(/<mods:mods/, 'mods')
-          expect_any_instance_of(described_class).to receive(:publish_notify_on_success).with(no_args)
         end
 
         it 'ignores missing data' do
-          service.publish
+          service.send(:transfer_metadata, {})
         end
       end
     end
   end
 
+  describe '#bookkeep_collections' do
+    it 'adds a link from the item to the collection' do
+    end
+
+    it 'adds a link from the collection to the item' do
+    end
+
+    it 'cleans up links that are not expressed in the item' do
+    end
+  end
+
+  describe '#unbookkeep_collections' do
+    it 'cleans up links from collections to this item' do
+    end
+
+    it 'cleans up links from items to this collection' do
+    end
+  end
+
   describe '#publish_notify_on_success' do
     subject(:notify) { service.send(:publish_notify_on_success) }