dor-services 4.8.3 → 4.9.0

data/lib/dor/datastreams/content_metadata_ds.rb

@@ -6,6 +6,7 @@ module Dor
     set_terminology do |t|
       t.root :path => 'contentMetadata', :index_as => [:not_searchable]
       t.contentType :path => '/contentMetadata/@type', :index_as => [:not_searchable]
+      t.stacks :path=> '/contentMetadata/@stacks', :index_as => [:not_searchable]
       t.resource(:index_as => [:not_searchable]) do
         t.id_ :path => { :attribute => 'id' }
         t.sequence :path => { :attribute => 'sequence' }#, :data_type => :integer

data/lib/dor/models/shelvable.rb

@@ -7,21 +7,58 @@ module Dor
 
     # Push file changes for shelve-able files into the stacks
     def shelve
+      # retrieve the differences between the current contentMetadata and the previously ingested version
+      shelve_diff = get_shelve_diff
+      stacks_object_pathname = get_stacks_location
+      # determine the location of the object's files in the stacks area
+      stacks_druid = DruidTools::StacksDruid.new id, stacks_object_pathname
+      stacks_object_pathname = Pathname(stacks_druid.path)
+      # determine the location of the object's content files in the workspace area
+      workspace_druid = DruidTools::Druid.new(id,Config.stacks.local_workspace_root)
+      workspace_content_pathname = workspace_content_dir(shelve_diff, workspace_druid)
+      # delete, rename, or copy files to the stacks area
+      DigitalStacksService.remove_from_stacks(stacks_object_pathname, shelve_diff)
+      DigitalStacksService.rename_in_stacks(stacks_object_pathname, shelve_diff)
+      DigitalStacksService.shelve_to_stacks(workspace_content_pathname, stacks_object_pathname, shelve_diff)
+    end
+
+    # retrieve the differences between the current contentMetadata and the previously ingested version
+    # (filtering to select only the files that should be shelved to stacks)
+    def get_shelve_diff
       inventory_diff_xml = self.get_content_diff(:shelve)
       inventory_diff = Moab::FileInventoryDifference.parse(inventory_diff_xml)
-      content_group_diff = inventory_diff.group_difference("content")
-      deltas = content_group_diff.file_deltas
-
-      if content_group_diff.rename_require_temp_files(deltas[:renamed])
-        triplets = content_group_diff.rename_tempfile_triplets(deltas[:renamed])
-        DigitalStacksService.rename_in_stacks self.pid, triplets.collect{|old,new,temp| [old,temp]}
-        DigitalStacksService.rename_in_stacks self.pid, triplets.collect{|old,new,temp| [temp,new]}
-      else
-        DigitalStacksService.rename_in_stacks self.pid, deltas[:renamed]
-      end
-      DigitalStacksService.shelve_to_stacks   self.pid, deltas[:modified] + deltas[:added] + deltas[:copyadded].collect{|old,new| new}
-      DigitalStacksService.remove_from_stacks self.pid, deltas[:deleted] + deltas[:copydeleted]
+      shelve_diff = inventory_diff.group_difference("content")
+      shelve_diff
     end
 
+    # Find the location of the object's content files in the workspace area
+    # @param [Moab::FileGroupDifference] content_diff The differences between the current contentMetadata and the previously ingested version
+    # @param [DruidTools::Druid] workspace_druid the location of the object's files in the workspace area
+    # @return [Pathname] The location of the object's content files in the workspace area
+    def workspace_content_dir (content_diff, workspace_druid)
+      deltas = content_diff.file_deltas
+      filelist = deltas[:modified] + deltas[:added] + deltas[:copyadded].collect{|old,new| new}
+      return nil if filelist.empty?
+      content_pathname = Pathname(workspace_druid.find_filelist_parent('content', filelist))
+      content_pathname
+    end
+    
+    
+    # get the stack location based on the contentMetadata stacks attribute 
+    # or using the default value from the config file if it doesn't exist
+    def get_stacks_location
+      
+      contentMetadataDS = self.datastreams['contentMetadata']
+      unless contentMetadataDS.nil? or contentMetadataDS.stacks.length == 0
+        stacks_location = contentMetadataDS.stacks[0]        
+        if stacks_location.start_with?"/"  #Absolute stacks path
+          return stacks_location
+        else        
+          raise "stacks attribute for item: "+self.id+ " contentMetadata should start with /. The current value is "+stacks_location
+        end
+      end      
+      return Config.stacks.local_stacks_root #Default stacks
+       
+    end
   end
 end

data/lib/dor/services/digital_stacks_service.rb

@@ -2,42 +2,128 @@ require 'net/ssh'
 require 'net/sftp'
 
 module Dor
+
   class DigitalStacksService
 
-    def self.transfer_to_document_store(id, content, filename)
-      druid = DruidTools::PurlDruid.new id, Config.stacks.local_document_cache_root
-      druid.content_dir # create the druid tree if it doesn't exist yet
-      File.open(File.join(druid.content_dir, filename), 'w') {|f| f.write content }
+    # Delete files from stacks that have change type 'deleted', 'copydeleted', or 'modified'
+    # @param [Pathname] stacks_object_pathname the stacks location of the digital object
+    # @param [Moab::FileGroupDifference] content_diff the content file version differences report
+    def self.remove_from_stacks(stacks_object_pathname, content_diff)
+      [:deleted, :copydeleted, :modified].each do |change_type|
+        subset = content_diff.subset(change_type) # {Moab::FileGroupDifferenceSubset}
+        subset.files.each do |moab_file| # {Moab::FileInstanceDifference}
+          moab_signature = moab_file.signatures.first # {Moab::FileSignature}
+          file_pathname = stacks_object_pathname.join(moab_file.basis_path)
+          self.delete_file(file_pathname, moab_signature)
+        end
+      end
+    end
+
+    # Delete a file, but only if it exists and matches the expected signature
+    # @param [Pathname] file_pathname The location of the file to be deleted
+    # @param [Moab::FileSignature] moab_signature The fixity values of the file
+    # @return [Boolean] true if file deleted, false otherwise
+    def self.delete_file(file_pathname, moab_signature)
+      if file_pathname.exist? and (file_pathname.size == moab_signature.size)
+        file_signature = Moab::FileSignature.new.signature_from_file(file_pathname)
+        if (file_signature == moab_signature)
+          file_pathname.delete
+          return true
+        end
+      end
+      return false
+    end
+
+    # Rename files from stacks that have change type 'renamed' using an intermediate temp filename.
+    # The 2-step renaming allows chained or cyclic renames to occur without file collisions.
+    # @param [Pathname] stacks_object_pathname the stacks location of the digital object
+    # @param [Moab::FileGroupDifference] content_diff the content file version differences report
+    def self.rename_in_stacks(stacks_object_pathname, content_diff)
+      subset = content_diff.subset(:renamed) # {Moab::FileGroupDifferenceSubset
+
+      # 1st Pass - rename files from original name to checksum-based name
+      subset.files.each do |moab_file| # {Moab::FileInstanceDifference}
+        moab_signature = moab_file.signatures.first # {Moab::FileSignature}
+        original_pathname = stacks_object_pathname.join(moab_file.basis_path)
+        temp_pathname = stacks_object_pathname.join(moab_signature.checksums.values.last)
+        self.rename_file(original_pathname, temp_pathname, moab_signature)
+      end
+
+      # 2nd Pass - rename files from checksum-based name to new name
+      subset.files.each do |moab_file| # {Moab::FileInstanceDifference}
+        moab_signature = moab_file.signatures.first # {Moab::FileSignature}
+        temp_pathname = stacks_object_pathname.join(moab_signature.checksums.values.last)
+        new_pathname = stacks_object_pathname.join(moab_file.other_path)
+        self.rename_file(temp_pathname, new_pathname, moab_signature)
+      end
+
     end
 
-    def self.remove_from_stacks(id, files)
-      files.each do |file|
-        dr = DruidTools::StacksDruid.new id, Config.stacks.local_stacks_root
-        content = dr.find_content file
-        FileUtils.rm content if content
+    # Rename a file, but only if it exists and has the expected signature
+    # @param [Pathname] old_pathname The original location/name of the file being renamed
+    # @param [Pathname] new_pathname The new location/name of the file
+    # @param [Moab::FileSignature] moab_signature The fixity values of the file
+    # @return [Boolean] true if file renamed, false otherwise
+    def self.rename_file(old_pathname, new_pathname, moab_signature)
+      if old_pathname.exist? and (old_pathname.size == moab_signature.size)
+        file_signature = Moab::FileSignature.new.signature_from_file(old_pathname)
+        if (file_signature == moab_signature)
+          new_pathname.parent.mkpath
+          old_pathname.rename(new_pathname)
+          return true
+        end
       end
+      return false
     end
 
-    # @param [String] id object pid
-    # @param [Array<Array<String>>] file_map an array of two string arrays.  Each inner array represents old-file/new-file mappings.  First string is the old file name, second string is the new file name. e.g:
-    #   [ ['src1.file', 'dest1.file'], ['src2.file', 'dest2.file'] ]
-    def self.rename_in_stacks(id, file_map)
-      return if file_map.nil? or file_map.empty?
-      dr = DruidTools::StacksDruid.new id, Config.stacks.local_stacks_root
-      content_dir = dr.find_filelist_parent('content', file_map.first.first)
-      file_map.each do |src, dest|
-        File.rename(File.join(content_dir, src), File.join(content_dir, dest))
+    # Add files to stacks that have change type 'added', 'copyadded' or 'modified'.
+    # @param [Pathname] workspace_content_pathname The dor workspace location of the digital object's content fies
+    # @param [Pathname] stacks_object_pathname the stacks location of the digital object's shelved files
+    # @param [Moab::FileGroupDifference] content_diff the content file version differences report
+    def self.shelve_to_stacks(workspace_content_pathname, stacks_object_pathname, content_diff)
+      return false if workspace_content_pathname.nil?
+      [:added, :copyadded, :modified,].each do |change_type|
+        subset = content_diff.subset(change_type) # {Moab::FileGroupDifferenceSubset
+        subset.files.each do |moab_file| # {Moab::FileInstanceDifference}
+          moab_signature = moab_file.signatures.last # {Moab::FileSignature}
+          filename = (change_type == :modified) ? moab_file.basis_path : moab_file.other_path
+          workspace_pathname = workspace_content_pathname.join(filename)
+          stacks_pathname = stacks_object_pathname.join(filename)
+          self.copy_file(workspace_pathname, stacks_pathname, moab_signature)
+        end
       end
+      true
     end
 
-    def self.shelve_to_stacks(id, files)
-      workspace_druid = DruidTools::Druid.new(id,Config.stacks.local_workspace_root)
-      stacks_druid = DruidTools::StacksDruid.new(id,Config.stacks.local_stacks_root)
-      files.each do |file|
-        stacks_druid.content_dir
-        workspace_file = workspace_druid.find_content(file)
-        FileUtils.cp workspace_file, stacks_druid.content_dir
+    # Copy a file to stacks, but only if it does not yet exist with the expected signature
+    # @param [Pathname] workspace_pathname The location of the file in the DOR workspace
+    # @param [Pathname] stacks_pathname The location of the file in the stacks
+    # @param [Moab::FileSignature] moab_signature The fixity values of the file
+    # @return [Boolean] true if file copied, false otherwise
+    def self.copy_file(workspace_pathname, stacks_pathname, moab_signature)
+      if stacks_pathname.exist?
+        file_signature = Moab::FileSignature.new.signature_from_file(stacks_pathname)
+        stacks_pathname.delete if (file_signature != moab_signature)
+      end
+      unless stacks_pathname.exist?
+        stacks_pathname.parent.mkpath
+        FileUtils.cp workspace_pathname.to_s, stacks_pathname.to_s
+        return true
       end
+      return false
+    end
+
+    ### depricated ???
+
+    # Create a file inside the content directory under the stacks.local_document_cache_root
+    # @param [String] id The druid identifier for the object
+    # @param [String] content The contents of the file to be created
+    # @param [String] filename The name of the file to be created
+    # @return [void]
+    def self.transfer_to_document_store(id, content, filename)
+      druid = DruidTools::PurlDruid.new id, Config.stacks.local_document_cache_root
+      druid.content_dir # create the druid tree if it doesn't exist yet
+      File.open(File.join(druid.content_dir, filename), 'w') { |f| f.write content }
     end
 
     # Assumes the digital stacks storage root is mounted to the local file system

data/lib/dor/version.rb

@@ -1,3 +1,3 @@
 module Dor
-  VERSION = '4.8.3'
+  VERSION = '4.9.0'
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: dor-services
 version: !ruby/object:Gem::Version
-  version: 4.8.3
+  version: 4.9.0
   prerelease: 
 platform: ruby
 authors:
@@ -13,7 +13,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2014-06-24 00:00:00.000000000 Z
+date: 2014-07-01 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: active-fedora