dor-services 2.2.4 → 4.4.10

data/lib/datastreams/content_metadata_ds.rb

@@ -1,12 +0,0 @@
-class ContentMetadataDS < ActiveFedora::NokogiriDatastream 
-  
-  def public_xml
-    result = self.ng_xml.clone
-    result.xpath('/contentMetadata/resource[not(file[(@deliver="yes" or @publish="yes")])]').each { |n| n.remove }
-    result.xpath('/contentMetadata/resource/file[not(@deliver="yes" or @publish="yes")]').each { |n| n.remove }
-    result.xpath('/contentMetadata/resource/file').xpath('@preserve|@shelve|@publish|@deliver').each { |n| n.remove }
-    result.xpath('/contentMetadata/resource/file/checksum').each { |n| n.remove }
-    result
-  end
-  
-end

data/lib/datastreams/identity_metadata_ds.rb

@@ -1,28 +0,0 @@
-class IdentityMetadataDS < ActiveFedora::NokogiriDatastream 
-  
-  set_terminology do |t|
-    t.root(:path=>"identityMetadata", :xmlns => '')
-    t.objectId(:index_as=>[:searchable, :displayable, :facetable, :sortable],  :required=>:true, :type=>:string, :namespace_prefix => nil )
-    t.objectType(:index_as=>[:searchable, :displayable, :facetable, :sortable],  :required=>:true, :type=>:string, :namespace_prefix => nil )  
-    t.objectLabel(:index_as=>[:searchable, :displayable, :facetable, :sortable],  :required=>:true, :type=>:string, :namespace_prefix => nil )  
-    t.citationCreator(:index_as=>[:searchable, :displayable, :facetable, :sortable],  :required=>:true, :type=>:string, :namespace_prefix => nil )  
-    t.sourceId(:index_as=>[:searchable,  :displayable, :facetable, :sortable], :attributes=>{:type=>"source"},  :required=>:true, :type=>:string, :namespace_prefix => nil )  
-    t.otherId(:index_as=>[:searchable,  :displayable, :facetable, :sortable], :attributes=>{:type=>"name"},  :required=>:true, :type=>:string, :namespace_prefix => nil )  
-    t.agreementId(:index_as=>[:searchable, :displayable, :facetable, :sortable],  :required=>:true, :type=>:string, :namespace_prefix => nil )  
-    t.tag(:index_as=>[:searchable, :displayable, :facetable, :sortable],  :required=>:true, :type=>:string, :namespace_prefix => nil )  
-    t.citationTitle(:index_as=>[:searchable, :displayable, :facetable, :sortable],  :required=>:true, :type=>:string, :namespace_prefix => nil )  
-    t.objectCreator(:index_as=>[:searchable, :displayable, :facetable, :sortable], :required=>:true, :type=>:string, :namespace_prefix => nil )
-    t.adminPolicy(:index_as=>[:searchable, :displayable, :facetable, :sortable], :required=>:true, :type=>:string, :namespace_prefix => nil )
-  end
-  
-  def self.xml_template
-    builder = Nokogiri::XML::Builder.new do |xml|
-      xml.identityMetadata {
-        xml.citationTitle
-        xml.objectCreator
-      }
-    end
-    return builder.doc
-  end #self.xml_template
-  
-end #class

data/lib/datastreams/ng_tidy.rb

@@ -1,19 +0,0 @@
-class Nokogiri::XML::Text
-  
-  def normalize
-    self.content =~ /\S/ ? self.content.gsub(/\s+/,' ').strip : self.content
-  end
-  
-  def normalize!
-    self.content = self.normalize
-  end
-  
-end
-
-class Nokogiri::XML::Node
-
-  def normalize_text!
-    self.xpath('//text()').each { |t| t.normalize! }
-  end
-  
-end

data/lib/datastreams/simple_dublin_core_ds.rb

@@ -1,23 +0,0 @@
-class SimpleDublinCoreDs < ActiveFedora::NokogiriDatastream 
-  
-  set_terminology do |t|
-    t.root(:path=>"dc", :xmlns=>"http://www.openarchives.org/OAI/2.0/oai_dc/", :schema=>"http://cosimo.stanford.edu/standards/oai_dc/v2/oai_dc.xsd", :namespace_prefix => 'oai_dc')
-    t.title(:index_as=>[:searchable, :displayable, :facetable, :sortable], :xmlns => "http://purl.org/dc/elements/1.1/", :namespace_prefix => 'dc')
-    t.creator(:index_as=>[:searchable, :displayable, :facetable, :sortable], :xmlns => "http://purl.org/dc/elements/1.1/", :namespace_prefix => 'dc')
-    t.identifier(:index_as=>[:searchable, :displayable, :sortable], :xmlns => "http://purl.org/dc/elements/1.1/", :namespace_prefix => 'dc')
-  end
-  
-  def self.xml_template
-    builder = Nokogiri::XML::Builder.new do |xml|
-      xml.dc(:xmlns=>"http://www.openarchives.org/OAI/2.0/oai_dc/", 
-        'xmlns:dc'=>'http://purl.org/dc/elements/1.1/') {
-          xml['dc'].title
-          xml['dc'].creator
-          xml['dc'].identifier
-      }   
-    end
-
-    return builder.doc
-  end
-  
-end

data/lib/datastreams/workflow_definition_ds.rb

@@ -1,105 +0,0 @@
-class WorkflowProcess
-
-  def initialize(workflow, node)
-    @workflow = workflow
-    @node = node
-  end
-  
-  def name
-    @node['name']
-  end
-  
-  def sequence
-    @node['sequence']
-  end
-  
-  def lifecycle
-    @node['lifecycle']
-  end
-  
-  def label
-    @node.at_xpath('label/text()').to_s
-  end
-  
-  def prerequisites
-    @node.xpath('prereq').collect do |p| 
-      if (p['repository'].nil? and p['workflow'].nil?) or (p['repository'] == workflow.repository and p['workflow'] == workflow.name)
-        p.text.to_s
-      else
-        [(p['repository'] or workflow.repository),(p['workflow'] or workflow.name),p.text.to_s].join(':')
-      end
-    end
-  end
-  
-end
-
-class WorkflowDefinitionDs < ActiveFedora::NokogiriDatastream 
-  
-  define_template :process do |builder,workflow,name,seq,label,lifecycle,prereqs|
-    attrs = {:name => name}
-    attrs[:sequence] = seq unless seq.nil?
-    attrs[:lifecycle] = lifecycle unless lifecycle.nil?
-    builder.process(attrs) do |node|
-      prereqs.each do |prereq|
-        (repo,wf,prereq_name) = prereq.split(/:/)
-        if prereq_name.nil?
-          prereq_name = repo
-          repo = nil
-        end
-        if (repo == workflow.repository and wf = workflow.name)
-          repo = nil
-          wf = nil
-        end
-        attrs = (repo.nil? and wf.nil?) ? {} : { :repository => repo, :workflow => wf }
-        node.prereq(attrs) { node.text prereq_name }
-      end
-    end
-  end
-
-  def add_process(name, seq, label, lifecycle, prereqs)
-    add_child_node(ng_xml.at_xpath('/workflow'), :process, self, name, seq, label, lifecycle, prereqs)
-  end
-  
-  def processes
-    ng_xml.xpath('/workflow/process').collect do |node|
-      WorkflowProcess.new(self, node)
-    end
-  end
-
-  def name
-    ng_xml.at_xpath('/workflow/@id').to_s
-  end
-  
-  def repository
-    ng_xml.at_xpath('/workflow/@repository').to_s
-  end
-  
-  def configuration
-    result = {
-      'repository' => repository,
-      'name' => name
-    }
-    processes.each_pair do |process_name,process|
-      result[process_name] = {
-        'prerequisites' => process.prerequisites.collect { |p| p.name }
-      }
-    end
-  end
-  
-  def configuration=(hash)
-    ng_xml = Nokogiri::XML(%{<workflow id="#{hash['name']}" repository="#{hash['repository']}"/>})
-    i = 0
-    hash.each_pair do |k,v| 
-      if v.is_a?(Hash)
-        add_process(k,i+=1,nil,nil,v['prerequisite'])
-      end
-    end
-  end
-  
-  def to_yaml
-    s = StringIO.new('')
-    YAML.dump(self.configuration, s)
-    s.string
-  end
-  
-end

data/lib/datastreams/workflow_ds.rb

@@ -1,16 +0,0 @@
-class WorkflowDs < ActiveFedora::NokogiriDatastream 
-  
-  set_terminology do |t|
-    t.root(:path=>"workflow", :xmlns => '', :namespace_prefix => nil)
-    t.workflowId(:path=>{:attribute => "id"}, :index_as => [:displayable, :facetable])
-    t.process(:path=>'process', :namespace_prefix => nil) {
-      t._name(:path=>{:attribute=>"name"}, :index_as => [:displayable, :facetable, :sortable])
-      t.status(:path=>{:attribute=>"status"}, :index_as => [:displayable, :facetable, :sortable])
-      t.timestamp(:path=>{:attribute=>"datetime"}, :index_as => [:searchable, :sortable])
-      t.elapsed(:path=>{:attribute=>"elapsed"})
-      t.lifecycle(:path=>{:attribute=>"lifecycle"}, :index_as => [:displayable, :facetable, :sortable])
-      t.attempts(:path=>{:attribute=>"attempts"})
-    }
-  end
-    
-end

data/lib/dor/admin_policy_object.rb

@@ -1,11 +0,0 @@
-module Dor
-  
-  class AdminPolicyObject < Base
-    
-    has_metadata :name => "administrativeMetadata", :type => ActiveFedora::NokogiriDatastream
-    has_metadata :name => "roleMetadata", :type => ActiveFedora::NokogiriDatastream
-    has_metadata :name => "defaultObjectRights", :type => ActiveFedora::NokogiriDatastream
-    
-  end
-  
-end

data/lib/dor/base.rb

@@ -1,81 +0,0 @@
-require 'active_fedora'
-require 'datastreams/identity_metadata_ds'
-require 'datastreams/simple_dublin_core_ds'
-require 'datastreams/workflow_ds'
-require 'dor/suri_service'
-
-module Dor
-
-  class Base < ::ActiveFedora::Base
-    
-    attr_reader :workflows
-
-    has_metadata :name => "DC", :type => SimpleDublinCoreDs
-    has_metadata :name => "RELS-EXT", :type => ActiveFedora::RelsExtDatastream
-    has_metadata :name => "identityMetadata", :type => IdentityMetadataDS
-
-    # Make a random (and harmless) API-M call to get gsearch to reindex the object
-    def self.touch(*pids)
-      client = Dor::Config.fedora.client
-      pids.collect { |pid|
-        response = client["objects/#{pid}/datastreams/DC?dsState=A&ignoreContent=true"].put('', :content_type => 'text/xml')
-        response.code
-      }
-    end
-    
-    def self.get_foxml(pid, interpolate_refs = false)
-      foxml = Nokogiri::XML(Dor::Config.fedora.client["objects/#{pid}/objectXML"].get)
-      if interpolate_refs
-        external_refs = foxml.xpath('//foxml:contentLocation[contains(@REF,"/workflows/")]')
-        external_refs.each do |ref|
-          begin
-            external_doc = Nokogiri::XML(RestClient.get(ref['REF']))
-            external_root = external_doc.root
-            ref.replace('<foxml:xmlContent/>').first.add_child(external_doc.root)
-            external_root.traverse { |node| node.namespace = nil }
-          rescue
-            ref.remove
-          end
-        end
-      end
-      return foxml
-    end
-    
-    def initialize(attrs = {})
-      unless attrs[:pid]
-        attrs = attrs.merge!({:pid=>Dor::SuriService.mint_id})  
-        @new_object=true
-      else
-        @new_object = attrs[:new_object] == false ? false : true
-      end
-      @inner_object = Fedora::FedoraObject.new(attrs)
-      @datastreams = {}
-      @workflows = {}
-      configure_defined_datastreams
-    end  
-
-    def identity_metadata
-      if self.datastreams.has_key?('identityMetadata')
-        IdentityMetadata.from_xml(self.datastreams['identityMetadata'].content)
-      else
-        nil
-      end
-    end
-    
-    # Self-aware datastream builders
-    def build_datastream(datastream, force = false)
-      ds = datastreams[datastream]
-      if force or ds.new_object? or (ds.content.to_s.empty?)
-        proc = "build_#{datastream}_datastream".to_sym
-        content = self.send(proc, ds)
-        ds.save
-      end
-      return ds
-    end
-    
-    def reindex
-      Dor::SearchService.reindex(self.pid)
-    end
-
-  end
-end

data/lib/dor/cleanup_service.rb

@@ -1,32 +0,0 @@
-require 'fileutils'
-require 'lyber-utils'
-
-module Dor
-  class CleanupService
-    Config.declare(:cleanup) do
-  	  local_workspace_root '/dor/workspace'
-      local_export_home '/dor/export'
-    end
-
-    # Delete all workspace and export entities for the druid
-    # @param [LyberCore::Robots::WorkItem]
-    def self.cleanup(dor_item)
-      druid = dor_item.pid
-      workspace_dir = Druid.new(druid).path(Config.cleanup.local_workspace_root)
-      self.remove_entry(workspace_dir)
-      bag_dir = File.join(Config.cleanup.local_export_home, druid)
-      self.remove_entry(bag_dir)
-      tarfile = "#{bag_dir}.tar"
-      self.remove_entry(tarfile)
-    end
-
-    # Deleta a filesystem entry
-    # @param [String]
-    def self.remove_entry(entry)
-      FileUtils.remove_entry(entry) if File.exist?(entry)
-    end
-
-
-  end
-
-end

data/lib/dor/digital_stacks_service.rb

@@ -1,82 +0,0 @@
-require 'tempfile'
-require 'systemu'
-
-module Dor
-  class DigitalStacksService
-    
-    Config.declare(:stacks) do
-  	  document_cache_storage_root nil
-  	  document_cache_host nil
-  	  document_cache_user nil
-  	  
-  	  storage_root '/stacks'
-  	  host nil
-  	  user nil
-  	  
-  	  local_workspace_root '/dor'
-    end
-    
-    # TODO copied from lyber-core, but didn't want to create circular dependency for between gems for this one method
-    # Executes a system command in a subprocess. 
-    # The method will return stdout from the command if execution was successful.
-    # The method will raise an exception if if execution fails. 
-    # The exception's message will contain the explaination of the failure.
-    # @param [String] command the command to be executed
-    # @return [String] stdout from the command if execution was successful
-    def self.execute(command)
-      status, stdout, stderr = systemu(command)
-      if (status.exitstatus != 0)
-        raise stderr
-      end
-      return stdout
-    rescue
-      msg = "Command failed to execute: [#{command}] caused by <STDERR =\n#{stderr.split($/).join("\n")}>"
-      msg << "\nSTDOUT =\n#{stdout.split($/).join("\n")}" if (stdout && (stdout.length > 0))
-      raise msg
-    end
-    
-    def self.druid_tree(druid)
-      Druid.new(druid).path
-    rescue
-      nil
-    end
-        
-    def self.transfer_to_document_store(id, content, filename)
-      path = self.druid_tree(id)
-      raise "Invalid druid: #{id}" if(path.nil?)
-      
-      # create the remote directory in the document cache
-      remote_document_cache_dir = File.join(Config.stacks.document_cache_storage_root, path)
-      command = "ssh #{Config.stacks.document_cache_user}@#{Config.stacks.document_cache_host} mkdir -p #{remote_document_cache_dir}"
-      self.execute(command)
-
-      # create a temp file containing the content and copy the contents to the remote document cache
-      Tempfile.open(filename) do |tf| 
-        tf.write(content) 
-        tf.flush
-        command = "scp \"#{tf.path}\" #{Config.stacks.document_cache_user}@#{Config.stacks.document_cache_host}:#{remote_document_cache_dir}/#{filename}"
-        self.execute(command)
-      end
-    end
-    
-    def self.shelve_to_stacks(id, files)
-      path = self.druid_tree(id)
-      raise "Invalid druid: #{id}" if(path.nil?)
-      
-      # create the remote directory on the digital stacks
-      remote_storage_dir = File.join(Config.stacks.storage_root, path)
-      command = "ssh #{Config.stacks.user}@#{Config.stacks.host} mkdir -p #{remote_storage_dir}"
-      self.execute(command)
-
-      # copy the contents for the given object from the local workspace directory to the remote directory
-      local_storage_dir = File.join(Config.stacks.local_workspace_root, path)
-      files.each do |file|
-        command = "scp \"#{local_storage_dir}/#{file}\" #{Config.stacks.user}@#{Config.stacks.host}:#{remote_storage_dir}"
-        self.execute(command)
-      end
-    end
-
-  end
-
-end
-

data/lib/dor/druid_utils.rb

@@ -1,41 +0,0 @@
-class Druid
-  attr_accessor :druid
-  
-  DRUID_PATTERN = /^(?:druid:)?([a-z]{2})(\d{3})([a-z]{2})(\d{4})$/
-  def initialize(druid)
-    if druid !~ DRUID_PATTERN
-      raise ArgumentError, "Invalid DRUID: #{druid}"
-    end
-    @druid = druid
-  end
-  
-  def id
-    @druid.scan(/^(?:druid:)?(.+)$/).flatten.last
-  end
-  
-  def tree
-    @druid.scan(DRUID_PATTERN).flatten
-  end
-  
-  def path(base=nil)
-    File.join(*([base,tree].compact))
-  end
-  
-  def mkdir(base)
-    FileUtils.mkdir_p(path(base))
-  end
-  
-  def rmdir(base)
-    parts = tree
-    while parts.length > 0
-      dir = File.join(base, *parts)
-      begin
-        FileUtils.rm(File.join(dir,'.DS_Store'), :force => true)
-        FileUtils.rmdir(dir)
-      rescue Errno::ENOTEMPTY
-        break
-      end
-      parts.pop
-    end
-  end
-end