cul_hydra 1.0.0

data/lib/cul_hydra/fedora.rb

@@ -0,0 +1,41 @@
+module Cul
+  module Hydra
+    module Fedora
+      autoload :DummyObject, 'cul_hydra/fedora/dummy_object'
+      autoload :RubydoraPatch, 'cul_hydra/fedora/rubydora_patch'
+      autoload :UrlHelperBehavior, 'cul_hydra/fedora/url_helper_behavior'
+      module ClassMethods
+        def config_path
+          File.join(Rails.root.to_s, 'config', 'fedora.yml')
+        end
+        def config
+          ActiveFedora.fedora_config.credentials
+        end
+        def connection
+          @connection ||= ActiveFedora::RubydoraConnection.new(ActiveFedora.fedora_config.credentials)
+        end
+
+        def repository
+          @repository ||= begin
+            repo = connection.connection
+            repo.extend(RubydoraPatch)
+            repo
+          end
+        end
+
+        def ds_for_uri(fedora_uri, fake_obj=nil)
+          return nil unless fedora_uri =~ /info\:fedora\/.*/
+          p = fedora_uri.split('/')
+          return ds_for_opts({pid: p[1], dsid: p[2]})
+        end
+
+        def ds_for_opts(opts={}, fake_obj=nil)
+          return nil unless opts[:pid] and opts[:dsid]
+          fake_obj = fake_obj.nil? ? DummyObject.new(opts[:pid]) : fake_obj.spawn(opts[:pid])
+          return (opts[:class] || ::Rubydora::Datastream).new(fake_obj, opts[:dsid])
+        end
+      end
+      extend ClassMethods
+    end
+  end
+end

data/lib/cul_hydra/fedora/dummy_object.rb

@@ -0,0 +1,37 @@
+module Cul
+  module Hydra
+    module Fedora
+      class DummyObject
+        attr_accessor :pid
+        def initialize(pid, isNew=false)
+          @pid = pid
+          @isNew = isNew
+        end
+        def new_record?
+          @isNew
+        end
+        def new_record=(val)
+          @isNew = val
+        end
+        def internal_uri
+          @uri ||= "info:fedora/#{@pid}"
+        end
+        def connection
+          Cul::Hydra::Fedora.connection
+        end    
+        def repository
+          Cul::Hydra::Fedora.repository
+        end
+        def spawn(pid)
+          s = DummyObject.new(pid, @isNew)
+          s.connection= connection
+          s.repository= repository
+          s
+        end
+        protected
+        def connection=(connection); @connection = connection; end
+        def repository=(repo); @repository = repo; end
+      end
+    end
+  end
+end

data/lib/cul_hydra/fedora/rubydora_patch.rb

@@ -0,0 +1,16 @@
+module Cul
+  module Hydra
+    module Fedora
+      module RubydoraPatch
+        def find_by_itql query, options = {}
+          begin
+            self.risearch(query, {:lang => 'itql'}.merge(options))
+          rescue Exception => e
+            logger.error e if defined?(logger)
+            "{\"results\":[]}"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/cul_hydra/fedora/url_helper_behavior.rb

@@ -0,0 +1,32 @@
+module Cul
+  module Hydra
+    module Fedora
+      module UrlHelperBehavior
+
+        def fedora_url
+          @fedora_url ||= ActiveFedora.config.credentials[:url]
+        end
+
+        def pid_for_url(pid)
+          pid.gsub(/^\//,'').gsub(/info:fedora\//,'')
+        end
+
+        def fedora_object_url(pid)
+          fedora_url + '/objects/' + pid_for_url(pid)
+        end
+
+        def fedora_ds_url(pid, dsid)
+          fedora_object_url(pid) + '/datastreams/' + dsid
+        end
+
+        def fedora_method_url(pid, method)
+          fedora_object_url(pid) + '/methods/' + method
+        end
+
+        def fedora_risearch_url
+          fedora_url + '/risearch'
+        end
+      end
+    end
+  end
+end

data/lib/cul_hydra/indexer.rb

@@ -0,0 +1,102 @@
+module Cul::Hydra::Indexer
+
+  NUM_FEDORA_RETRY_ATTEMPTS = 3
+  DELAY_BETWEEN_FEDORA_RETRY_ATTEMPTS = 5.seconds
+
+  def self.descend_from(pid, pids_to_omit=nil, verbose_output=false)
+    if pid.blank?
+      raise 'Please supply a pid (e.g. rake recursively_index_fedora_objects pid=ldpd:123)'
+    end
+
+    begin
+
+      unless ActiveFedora::Base.exists?(pid)
+        raise 'Could not find Fedora object with pid: ' + pid
+      end
+
+      if pids_to_omit.present? && pids_to_omit.include?(pid)
+        puts 'Skipping topmost object in this set (' + pid + ') because it has been intentionally omitted...' if verbose_output
+      else
+        puts 'Indexing topmost object in this set (' + pid + ')...' if verbose_output
+        puts 'If this is a BagAggregator with a lot of members, this may take a while...' if verbose_output
+
+        yield pid
+
+      end
+
+      puts 'Recursively retreieving and indexing all members of ' + pid + '...'
+
+      unique_pids = Cul::Hydra::RisearchMembers.get_recursive_member_pids(pid, true)
+
+      total_number_of_members = unique_pids.length
+      puts 'Recursive search found ' + total_number_of_members.to_s + ' members.' if verbose_output
+
+      if pids_to_omit.present?
+        unique_pids = unique_pids - pids_to_omit
+        total_number_of_members = unique_pids.length
+        puts 'After checking against the list of omitted pids, the total number of objects to index will be: ' + total_number_of_members.to_s if verbose_output
+      end
+
+      i = 1
+      if total_number_of_members > 0
+        unique_pids.each {|pid|
+
+          puts 'Recursing on ' + i.to_s + ' of ' + total_number_of_members.to_s + ' members (' + pid + ')...' if verbose_output
+
+          yield pid
+
+          i += 1
+        }
+      end
+
+    rescue RestClient::Unauthorized => e
+      error_message = "Skipping #{pid} due to error: " + e.message + '.  Problem with Fedora object?'
+      puts error_message
+      logger.error error_message if defined?(logger)
+    end
+
+    puts 'Recursion complete!'
+
+  end
+  def self.recursively_index_fedora_objects(top_pid, pids_to_omit=nil, skip_generic_resources=false, verbose_output=false)
+
+    descend_from(top_pid, pids_to_omit, verbose_output) do |pid|
+      self.index_pid(pid, skip_generic_resources, verbose_output)
+    end
+
+  end
+
+  def self.index_pid(pid, skip_generic_resources=false, verbose_output=false)
+    # We found an object with the desired PID. Let's reindex it
+    begin
+      active_fedora_object = nil
+
+      NUM_FEDORA_RETRY_ATTEMPTS.times { |i|
+        begin
+          active_fedora_object = ActiveFedora::Base.find(pid, :cast => true)
+          break
+        rescue RestClient::RequestTimeout, Errno::EHOSTUNREACH => e
+          remaining_attempts = (NUM_FEDORA_RETRY_ATTEMPTS-1) - i
+          if remaining_attempts == 0
+            raise e
+          else
+            Rails.logger.error "Error: Could not connect to fedora. (#{e.class.to_s + ': ' + e.message}).  Will retry #{remaining_attempts} more #{remaining_attempts == 1 ? 'time' : 'times'} (after a #{DELAY_BETWEEN_FEDORA_RETRY_ATTEMPTS} second delay)."
+            sleep DELAY_BETWEEN_FEDORA_RETRY_ATTEMPTS
+          end
+        end
+      }
+
+      if skip_generic_resources && active_fedora_object.is_a?(GenericResource)
+        puts 'Object was skipped because GenericResources are being skipped and it is a GenericResource.'
+      else
+        active_fedora_object.update_index
+        puts 'done.' if verbose_output
+      end
+    rescue SystemExit, Interrupt => e
+      # Allow system interrupt (ctrl+c)
+      raise e
+    rescue Exception => e
+      puts "Encountered problem with #{pid}.  Skipping record.  Exception: #{e.message}"
+    end
+  end
+end

data/lib/cul_hydra/om.rb

@@ -0,0 +1,7 @@
+module Cul
+  module Hydra
+    module Om
+      autoload :StandardMods, "cul_hydra/om/standard_mods"
+    end
+  end
+end

data/lib/cul_hydra/om/standard_mods.rb

@@ -0,0 +1,115 @@
+require 'active-fedora'
+require 'solrizer'
+require 'cul_hydra/solrizer_patch'
+module Cul
+module Hydra
+module Om
+  class StandardMods < ::ActiveFedora::OmDatastream
+
+    set_terminology do |t|
+      t.root(:path=>"mods",
+             :xmlns=>"http://www.loc.gov/mods/v3",
+             :schema=>"http://www.loc.gov/standards/mods/v3/mods-3-4.xsd")
+      t.identifier(:path=>"identifier", :attributes=>{:type=>"local"}, :data_type=>:symbol)
+      t.clio(:path=>"identifier", :attributes=>{:type=>"CLIO"}, :data_type=>:symbol)
+      t.title_info(:path=>"titleInfo", :index_as=>[:not_searchable]) {
+        t.main_title(:path=>"title", :index_as=>[:not_searchable])
+      }
+      t.title(:path=>'mods/oxns:titleInfo/oxns:title', :index_as=>[:searchable,:displayable, :sortable])
+      t.abstract
+      t.subject {
+        t.topic
+      }
+      t.type_of_resource(:path=>"typeOfResource", :index_as=>[:not_searchable])
+      t.physical_description(:path=>"physicalDescription", :index_as=>[:not_searchable]){
+        t.form_marc(:path=>"form", :attributes=>{:authority=>"marcform"}, :index_as=>[:not_searchable])
+        t.form_nomarc(:path=>"form[@authority !='marcform']", :index_as=>[:not_searchable, :displayable, :facetable, :textable])
+        t.extent(:path=>"extent", :index_as=>[:not_searchable])
+        t.reformatting_quality(:path=>"reformattingQuality", :index_as=>[:not_searchable])
+        t.internet_media_type(:path=>"internetMediaType", :index_as=>[:not_searchable])
+        t.digital_origin(:path=>"digitalOrigin", :index_as=>[:not_searchable])
+      }
+      t.lib_format(proxy: [:physical_description, :form_nomarc] )
+      t.location(:path=>"location", :index_as=>[:not_searchable]){
+        t.repo_text(:path=>"physicalLocation",:attributes=>{:authority=>:none},  :index_as=>[:not_searchable])
+        t.repo_code(:path=>"physicalLocation",:attributes=>{:authority=>"marcorg"}, :index_as=>[:not_searchable])
+      }
+      t.lib_repo_text(:ref=>[:location, :repo_text], :label=>"lib_repo", :index_as=>[:searchable])
+      t.lib_repo(:ref=>[:location, :repo_code], :index_as=>[:not_searchable,:facetable, :displayable])
+      t.project_host(:path=>"relatedItem", :attributes=>{:type=>"host", :displayLabel=>"Project"}, :index_as=>[:not_searchable]){
+        t.p_title(:path=>'titleInfo',:index_as=>[:not_searchable])
+      }
+      t.lib_project(:proxy=>[:project_host, :p_title],:index_as=>[:facetable,:displayable, :not_searchable])
+      t.collection_host(:path=>"relatedItem", :attributes=>{:type=>"host", :displayLabel=>"Collection"}, :index_as=>[:not_searchable]){
+        t.c_title(:path=>'titleInfo',:index_as=>[:not_searchable])
+      }
+      t.lib_project(:path=>"mods/oxns:relatedItem[@type='host'][@displayLabel='Project']/oxns:titleInfo/oxns:title",:index_as=>[:facetable,:displayable, :not_searchable])
+      t.lib_collection(:path=>"mods/oxns:relatedItem[@type='host'][@displayLabel='Collection']/oxns:titleInfo/oxns:title",:index_as=>[:facetable,:displayable, :not_searchable])
+      t.note(:path=>"note")
+      t.access_condition(:path=>"accessCondition", :attributes=>{:type=>"useAndReproduction"}, :index_as => [:searchable], :data_type => :symbol)
+      t.record_info(:path=>"recordInfo", :index_as=>[:not_searchable]) {
+        t.record_creation_date(:path=>"recordCreationDate",:attributes=>{:encoding=>"w3cdtf"}, :index_as=>[:not_searchable])
+        t.record_content_source(:path=>"recordContentSource",:attributes=>{:authority=>"marcorg"}, :index_as=>[:not_searchable])
+        t.language_of_cataloging(:path=>"languageOfCataloging", :index_as=>[:not_searchable]){
+          t.language_term(:path=>"languageTerm", :index_as=>[:not_searchable], :attributes=>{:type=>:none})
+          t.language_code(:path=>"languageTerm",:attributes=>{:type=>'code',:authority=>"iso639-2b"}, :index_as=>[:not_searchable])
+        }
+        t.record_origin(:path=>"recordOrigin", :index_as=>[:not_searchable])
+      }
+
+      t.origin_info(:path=>"originInfo", :index_as=>[:not_searchable]){
+        t.date(:path=>"dateIssued", :attributes=>{:encoding=>'w3cdtf'}, :index_as=>[:not_searchable])
+        t.key_date(:path=>"dateIssued", :attributes=>{:encoding=>'w3cdtf',:keyDate=>'yes'}, :index_as=>[:not_searchable])
+        t.start_date(:path=>"dateIssued", :attributes=>{:encoding=>'w3cdtf',:keyDate=>'yes',:point=>'start'}, :index_as=>[:not_searchable])
+        t.end_date(:path=>"dateIssued", :attributes=>{:encoding=>'w3cdtf',:point=>'end'}, :index_as=>[:not_searchable])
+      }
+    end
+
+    def self.xml_template
+      builder = Nokogiri::XML::Builder.new do |xml|
+        xml.mods(:version=>"3.4",
+           "xmlns"=>"http://www.loc.gov/mods/v3",
+           "xmlns:xsi"=>"http://www.w3.org/2001/XMLSchema-instance"){
+        }
+      end
+      builder.doc.encoding = 'UTF-8'
+      builder.doc.root["xsi:schemaLocation"] = 'http://www.loc.gov/mods/v3 http://www.loc.gov/standards/mods/v3/mods-3-4.xsd'
+      return builder.doc
+    end
+
+    def prefix
+      #if ::ActiveFedora::VERSION >= '8'
+      #  Rails.logger.warn("the prefix method of #{self.class.name} was overriden to maintain backwards compatibility")
+      #end
+      ''
+    end
+
+    def method_missing method, *args
+      query = false
+      _mname = method.id2name
+      if _mname[-1,1] == '?'
+        query = true
+        _mname = _mname[0,_mname.length-1]
+      end
+      _msym = _mname.to_sym
+      begin
+        has_term = self.class.terminology.has_term?(_msym)
+
+        _r = (has_term)? find_by_terms(_msym, *args) : nil
+        if query
+          return !( _r.nil? || _r.size()==0)
+        else
+          return _r
+        end
+      rescue
+        super
+      end
+    end
+    def update_values(params)
+      super
+      self.dirty = true
+    end
+  end
+end
+end
+end

data/lib/cul_hydra/risearch_members.rb

@@ -0,0 +1,92 @@
+module Cul::Hydra::RisearchMembers
+module ClassMethods
+  def get_recursive_member_pids(pid, verbose_output=false, cmodel_type='all')
+
+    recursive_member_query =
+      'select $child $parent from <#ri>
+      where
+      walk($child <http://purl.oclc.org/NET/CUL/memberOf> <fedora:' + pid + '> and $child <http://purl.oclc.org/NET/CUL/memberOf> $parent)'      
+
+    unless cmodel_type == 'all'
+      recursive_member_query += ' and $child <fedora-model:hasModel> $cmodel'
+      recursive_member_query += ' and $cmodel <mulgara:is> <info:fedora/ldpd:' + cmodel_type + '>'
+    end
+
+    puts 'Performing query:' if verbose_output
+    puts recursive_member_query if verbose_output
+
+    search_response = JSON(Cul::Hydra::Fedora.repository.find_by_itql(recursive_member_query, {
+      :type => 'tuples',
+      :format => 'json',
+      :limit => '',
+      :stream => 'on'
+    }))
+
+    unique_pids = search_response['results'].map{|result| result['child'].gsub('info:fedora/', '') }.uniq
+
+    return unique_pids
+
+  end
+
+  def get_direct_member_results(pid, verbose_output=false, format='json')
+
+    direct_member_query =
+      'select $pid from <#ri>
+      where $pid <http://purl.oclc.org/NET/CUL/memberOf> <fedora:' + pid + '>'
+
+    puts 'Performing query:' if verbose_output
+    puts direct_member_query if verbose_output
+
+    search_response = JSON(Cul::Hydra::Fedora.repository.find_by_itql(direct_member_query, {
+      :type => 'tuples',
+      :format => format,
+      :limit => '',
+      :stream => 'on'
+    }))
+
+    return search_response['results']
+  end
+
+  def get_direct_member_pids(pid, verbose_output=false)
+    unique_pids = get_direct_member_results(pid,verbose_output,'json')
+    unique_pids.map{|result| result['pid'].gsub('info:fedora/', '') }.uniq
+  end
+
+  def get_direct_member_count(pid, verbose_output=false)
+    count = get_direct_member_results(pid,verbose_output,'count/json')
+    return count.blank? ? 0 : count[0]['count'].to_i
+  end
+
+  #Project constituents
+
+  def get_project_constituent_results(pid, verbose_output=false, format='json')
+
+    project_constituent_query =
+      'select $pid from <#ri>
+      where $pid <info:fedora/fedora-system:def/relations-external#isConstituentOf> <fedora:' + pid + '>'
+
+    puts 'Performing query:' if verbose_output
+    puts project_constituent_query if verbose_output
+
+    search_response = JSON(Cul::Hydra::Fedora.repository.find_by_itql(project_constituent_query, {
+      :type => 'tuples',
+      :format => format,
+      :limit => '',
+      :stream => 'on'
+    }))
+
+    return search_response['results']
+  end
+
+  def get_project_constituent_pids(pid, verbose_output=false)
+    unique_pids = get_project_constituent_results(pid,verbose_output,'json')
+    unique_pids.map{|result| result['pid'].gsub('info:fedora/', '') }.uniq
+  end
+
+  def get_project_constituent_count(pid, verbose_output=false)
+    count = get_project_constituent_results(pid,verbose_output,'count/json')
+    return count.blank? ? 0 : count[0]['count'].to_i
+  end
+end
+extend ClassMethods
+end