sufia 4.0.0.beta2 → 4.0.0.beta3

data/sufia-models/app/actors/sufia/generic_file/actor.rb

@@ -0,0 +1,98 @@
+module Sufia::GenericFile
+  # Actions are decoupled from controller logic so that they may be called from a controller or a background job.
+  class Actor
+
+    attr_reader :generic_file, :user
+
+    def initialize(generic_file, user)
+      @generic_file = generic_file
+      @user = user
+    end
+
+    # in order to avoid two saves in a row, create_metadata does not save the file by default.
+    # it is typically used in conjunction with create_content, which does do a save.
+    # If you want to save when using create_metadata, you can do this:
+    #   create_metadata(batch_id) { |gf| gf.save }
+    def create_metadata(batch_id)
+      generic_file.apply_depositor_metadata(user)
+      generic_file.date_uploaded = Date.today
+      generic_file.date_modified = Date.today
+      generic_file.creator = user.name
+
+      if batch_id
+        generic_file.add_relationship("isPartOf", "info:fedora/#{Sufia::Noid.namespaceize(batch_id)}")
+      else
+        logger.warn "unable to find batch to attach to"
+      end
+      yield(generic_file) if block_given?
+    end
+
+    def create_content(file, file_name, dsid)
+      generic_file.add_file(file, dsid, file_name)
+      save_characterize_and_record_committer do
+        if Sufia.config.respond_to?(:after_create_content)
+          Sufia.config.after_create_content.call(generic_file, user)
+        end
+      end
+    end
+
+    def revert_content(revision_id, datastream_id)
+      revision = generic_file.content.get_version(revision_id)
+      generic_file.add_file(revision.content, datastream_id, revision.label)
+      save_characterize_and_record_committer
+    end
+
+    def update_content(file, datastream_id)
+      generic_file.add_file(file, datastream_id, file.original_filename)
+      save_characterize_and_record_committer
+    end
+
+    def update_metadata(attributes, visibility)
+      generic_file.attributes = generic_file.sanitize_attributes(attributes)
+      generic_file.visibility = visibility
+      generic_file.date_modified = DateTime.now
+      save_and_record_committer
+    end
+
+    # Takes an optional block and executes the block if the save was successful.
+    def save_characterize_and_record_committer
+      save_and_record_committer { push_characterize_job }.tap do |val|
+        yield if block_given? && val
+      end
+    end
+
+    # Takes an optional block and executes the block if the save was successful.
+    def save_and_record_committer
+      save_tries = 0
+      begin
+        return false unless generic_file.save
+      rescue RSolr::Error::Http => error
+        logger.warn "Sufia::GenericFile::Actor::save_and_record_committer Caught RSOLR error #{error.inspect}"
+        save_tries+=1
+        # fail for good if the tries is greater than 3
+        raise error if save_tries >=3
+        sleep 0.01
+        retry
+      end
+      yield if block_given?
+      generic_file.record_version_committer(user)
+      true
+    end
+
+    def push_characterize_job
+      Sufia.queue.push(CharacterizeJob.new(@generic_file.pid))
+    end
+
+    class << self
+      def virus_check(file)
+        path = file.is_a?(String) ? file : file.path
+        unless defined?(ClamAV)
+          logger.warn "Virus checking disabled, #{path} not checked"
+          return
+        end
+        scan_result = ClamAV.instance.scanfile(path)
+        raise Sufia::VirusFoundError.new("A virus was found in #{path}: #{scan_result}") unless scan_result == 0
+      end
+    end
+  end
+end

data/sufia-models/app/models/batch.rb

@@ -3,10 +3,10 @@ class Batch < ActiveFedora::Base
   include Sufia::ModelMethods
   include Sufia::Noid
 
-  has_metadata :name => "descMetadata", :type => BatchRdfDatastream
+  has_metadata name: "descMetadata", type: BatchRdfDatastream
 
-  belongs_to :user, :property => "creator"
-  has_many :generic_files, :property => :is_part_of
+  belongs_to :user, property: "creator"
+  has_many :generic_files, property: :is_part_of
 
   has_attributes :title, :creator, :part, :status, datastream: :descMetadata, multiple: true
 

data/sufia-models/app/models/checksum_audit_log.rb

@@ -1,9 +1,9 @@
 class ChecksumAuditLog < ActiveRecord::Base
 
   def ChecksumAuditLog.get_audit_log(version)
-    ChecksumAuditLog.find_or_create_by_pid_and_dsid_and_version(:pid => version.pid,
-                                                                :dsid => version.dsid,
-                                                                :version => version.versionID)
+    ChecksumAuditLog.find_or_create_by_pid_and_dsid_and_version(pid: version.pid,
+                                                                dsid: version.dsid,
+                                                                version: version.versionID)
   end
 
   def ChecksumAuditLog.prune_history(version)

data/sufia-models/app/models/collection.rb

@@ -0,0 +1,41 @@
+class Collection < ActiveFedora::Base
+  include Hydra::Collection
+  include Sufia::ModelMethods
+  include Sufia::Noid
+  include Sufia::GenericFile::Permissions
+  include Sufia::GenericFile::WebForm # provides initialize_fields method
+
+  before_save :update_permissions
+  validates :title, presence: true
+
+  has_metadata "properties", type: PropertiesDatastream
+
+  def terms_for_display
+    [:title, :creator, :description, :date_modified, :date_uploaded]
+  end
+  
+  def terms_for_editing
+    terms_for_display - [:date_modified, :date_uploaded]
+  end
+  
+  # Test to see if the given field is required
+  # @param [Symbol] key a field
+  # @return [Boolean] is it required or not
+  def required?(key)
+    self.class.validators_on(key).any?{|v| v.kind_of? ActiveModel::Validations::PresenceValidator}
+  end
+  
+  def to_param
+    noid
+  end
+
+  def to_solr(solr_doc={}, opts={})
+    super(solr_doc, opts)
+    solr_doc[Solrizer.solr_name("noid", Sufia::GenericFile.noid_indexer)] = noid
+    return solr_doc
+  end
+
+  def update_permissions
+    self.visibility = "open"
+  end
+end

data/sufia-models/{lib/sufia/models → app/models/concerns/sufia}/generic_file.rb

@@ -3,21 +3,9 @@ module Sufia
 
   module GenericFile
     extend ActiveSupport::Concern
-    extend ActiveSupport::Autoload
-    autoload :Actions, 'sufia/models/generic_file/actions'
-    autoload :Permissions, 'sufia/models/generic_file/permissions'
-    autoload :WebForm, 'sufia/models/generic_file/web_form'
-    autoload :AccessibleAttributes, 'sufia/models/generic_file/accessible_attributes'
-    autoload :Trophies, 'sufia/models/generic_file/trophies'
-    autoload :Featured, 'sufia/models/generic_file/featured'
-    autoload :Metadata, 'sufia/models/generic_file/metadata'
-    autoload :Versions, 'sufia/models/generic_file/versions'
-    autoload :VirusCheck, 'sufia/models/generic_file/virus_check'
-    autoload :ReloadOnSave, 'sufia/models/generic_file/reload_on_save'
     include Sufia::ModelMethods
     include Sufia::Noid
     include Sufia::GenericFile::MimeTypes
-    include Sufia::GenericFile::Thumbnail
     include Sufia::GenericFile::Export
     include Sufia::GenericFile::Characterization
     include Sufia::GenericFile::Audit
@@ -30,13 +18,14 @@ module Sufia
     include Sufia::GenericFile::Versions
     include Sufia::GenericFile::VirusCheck
     include Sufia::GenericFile::ReloadOnSave
+    include Hydra::Collections::Collectible
 
     included do
-      belongs_to :batch, :property => :is_part_of
+      belongs_to :batch, property: :is_part_of
 
-      around_save :characterize_if_changed, :retry_warming
+      around_save :retry_warming
 
-      attr_accessible *(ds_specs['descMetadata'][:type].fields + [:permissions])
+      attr_accessible *(terms_for_display + [:part_of, :permissions])
     end
 
     def persistent_url
@@ -44,34 +33,34 @@ module Sufia
     end
 
     def retry_warming
-        save_tries = 0
-        conflict_tries = 0
-        begin
-          yield
-        rescue RSolr::Error::Http => error
-          save_tries += 1
-          logger.warn "Retry Solr caught RSOLR error on #{self.pid}: #{error.inspect}"
-          # fail for good if the tries is greater than 3
-          raise if save_tries >=3
-          sleep 0.01
-          retry
-        rescue  ActiveResource::ResourceConflict => error
+      save_tries = 0
+      conflict_tries = 0
+      begin
+        yield
+      rescue RSolr::Error::Http => error
+        save_tries += 1
+        logger.warn "Retry Solr caught RSOLR error on #{self.pid}: #{error.inspect}"
+        # fail for good if the tries is greater than 3
+        raise if save_tries >=3
+        sleep 0.01
+        retry
+      rescue  ActiveResource::ResourceConflict => error
+        conflict_tries += 1
+        logger.warn "Retry caught Active Resource Conflict #{self.pid}: #{error.inspect}"
+        raise if conflict_tries >=10
+        sleep 0.01
+        retry
+      rescue => error
+        if error.to_s.downcase.include? "conflict"
           conflict_tries += 1
           logger.warn "Retry caught Active Resource Conflict #{self.pid}: #{error.inspect}"
           raise if conflict_tries >=10
           sleep 0.01
           retry
-        rescue =>error
-          if (error.to_s.downcase.include? "conflict")
-            conflict_tries += 1
-            logger.warn "Retry caught Active Resource Conflict #{self.pid}: #{error.inspect}"
-            raise if conflict_tries >=10
-            sleep 0.01
-            retry
-          else
-            raise
-          end
+        else
+          raise
         end
+      end
     end
 
     # Get the files with a sibling relationship (belongs_to :batch)
@@ -99,12 +88,13 @@ module Sufia
     end
 
     def to_solr(solr_doc={}, opts={})
-      solr_doc = super(solr_doc, opts)
-      solr_doc[Solrizer.solr_name('label')] = self.label
-      solr_doc[Solrizer.solr_name('noid', Sufia::GenericFile.noid_indexer)] = noid
-      solr_doc[Solrizer.solr_name('file_format')] = file_format
-      solr_doc[Solrizer.solr_name('file_format', :facetable)] = file_format
-      return solr_doc
+      super(solr_doc, opts).tap do |solr_doc|
+        solr_doc[Solrizer.solr_name('label')] = self.label
+        solr_doc[Solrizer.solr_name('noid', Sufia::GenericFile.noid_indexer)] = noid
+        solr_doc[Solrizer.solr_name('file_format')] = file_format
+        solr_doc[Solrizer.solr_name('file_format', :facetable)] = file_format
+        solr_doc = index_collection_pids(solr_doc)
+      end
     end
 
     # Redefine this for more intuitive keys in Redis

data/sufia-models/{lib/sufia/models → app/models/concerns/sufia}/generic_file/accessible_attributes.rb

@@ -41,7 +41,7 @@ module Sufia
         #     attr_accessor :name, :credit_rating
         #
         #     attr_accessible :name
-        #     attr_accessible :name, :credit_rating, :as => :admin
+        #     attr_accessible :name, :credit_rating, as: :admin
         #
         #     def assign_attributes(values, options = {})
         #       sanitize_for_mass_assignment(values, options[:as]).each do |k, v|
@@ -53,7 +53,7 @@ module Sufia
         # When using the :default role:
         #
         #   customer = Customer.new
-        #   customer.assign_attributes({ "name" => "David", "credit_rating" => "Excellent", :last_login => 1.day.ago }, :as => :default)
+        #   customer.assign_attributes({ "name" => "David", "credit_rating" => "Excellent", last_login: 1.day.ago }, as: :default)
         #   customer.name          # => "David"
         #   customer.credit_rating # => nil
         #
@@ -63,7 +63,7 @@ module Sufia
         # And using the :admin role:
         #
         #   customer = Customer.new
-        #   customer.assign_attributes({ "name" => "David", "credit_rating" => "Excellent", :last_login => 1.day.ago }, :as => :admin)
+        #   customer.assign_attributes({ "name" => "David", "credit_rating" => "Excellent", last_login: 1.day.ago }, as: :admin)
         #   customer.name          # => "David"
         #   customer.credit_rating # => "Excellent"
         #

data/sufia-models/{lib/sufia/models → app/models/concerns/sufia}/generic_file/audit.rb

@@ -24,7 +24,7 @@ module Sufia
       end
 
       def logs(dsid)
-        ChecksumAuditLog.where(:dsid=>dsid, :pid=>self.pid).order('created_at desc, id desc')
+        ChecksumAuditLog.where(dsid: dsid, pid: self.pid).order('created_at desc, id desc')
       end
 
       def audit!
@@ -40,19 +40,18 @@ module Sufia
         audit_results = logs.collect { |result| result["pass"] }
 
         # check how many non runs we had
-        non_runs =audit_results.reduce(0) { |sum, value| (value == NO_RUNS) ? sum = sum+1 : sum }
-        if (non_runs == 0)
-          result =audit_results.reduce(true) { |sum, value| sum && value }
+        non_runs = audit_results.reduce(0) { |sum, value| value == NO_RUNS ? sum += 1 : sum }
+        if non_runs == 0
+          result = audit_results.reduce(true) { |sum, value| sum && value }
           return result
-        elsif (non_runs < audit_results.length)
-          result =audit_results.reduce(true) { |sum, value| (value == NO_RUNS) ? sum : sum && value }
+        elsif non_runs < audit_results.length
+          result = audit_results.reduce(true) { |sum, value| value == NO_RUNS ? sum : sum && value }
           return 'Some audits have not been run, but the ones run were '+ ((result)? 'passing' : 'failing') + '.'
         else
           return 'Audits have not yet been run on this file.'
         end
       end
 
-
       module ClassMethods
         def audit!(version)
           ::GenericFile.audit(version, true)
@@ -68,24 +67,20 @@ module Sufia
 
           # run the find just incase the job has finished already
           latest_audit = self.find(version.pid).logs(version.dsid).first
-          latest_audit = ChecksumAuditLog.new(:pass=>NO_RUNS, :pid=>version.pid, :dsid=>version.dsid, :version=>version.versionID) unless latest_audit
-          return latest_audit
+          latest_audit = ChecksumAuditLog.new(pass: NO_RUNS, pid: version.pid, dsid: version.dsid, version: version.versionID) unless latest_audit
+          latest_audit
         end
 
         def needs_audit?(version, latest_audit)
           if latest_audit and latest_audit.updated_at
-            #logger.debug "***AUDIT*** last audit = #{latest_audit.updated_at.to_date}"
             days_since_last_audit = (DateTime.now - latest_audit.updated_at.to_date).to_i
-            #logger.debug "***AUDIT*** days since last audit: #{days_since_last_audit}"
             if days_since_last_audit < Sufia.config.max_days_between_audits
-              #logger.debug "***AUDIT*** No audit needed for #{version.pid} #{version.versionID} (#{latest_audit.updated_at})"
               return false
             end
           else
             logger.warn "***AUDIT*** problem with audit log!  Latest Audit is not nil, but updated_at is not set #{latest_audit}"  unless latest_audit.nil?
           end
-          #logger.info "***AUDIT*** Audit needed for #{version.pid} #{version.versionID}"
-          return true
+          true
         end
 
         def audit_everything(force = false)
@@ -102,16 +97,15 @@ module Sufia
 
         def run_audit(version)
           if version.dsChecksumValid
-            #logger.info "***AUDIT*** Audit passed for #{version.pid} #{version.versionID}"
             passing = 1
             ChecksumAuditLog.prune_history(version)
           else
             logger.warn "***AUDIT*** Audit failed for #{version.pid} #{version.versionID}"
             passing = 0
           end
-          check = ChecksumAuditLog.create!(:pass=>passing, :pid=>version.pid,
-                                           :dsid=>version.dsid, :version=>version.versionID)
-          return check
+          check = ChecksumAuditLog.create!(pass: passing, pid: version.pid,
+                                           dsid: version.dsid, version: version.versionID)
+          check
         end
       end
     end

data/sufia-models/{lib/sufia/models → app/models/concerns/sufia}/generic_file/characterization.rb

@@ -3,7 +3,7 @@ module Sufia
     module Characterization
       extend ActiveSupport::Concern
       included do
-        has_metadata "characterization", :type => FitsDatastream
+        has_metadata "characterization", type: FitsDatastream
         has_attributes :mime_type, datastream: :characterization, multiple: false
         has_attributes :format_label, :file_size, :last_modified,
                                         :filename, :original_checksum, :rights_basis,
@@ -40,12 +40,6 @@ module Sufia
         characterization.sample_rate.blank? ? characterization.video_sample_rate : characterization.sample_rate
       end
 
-      def characterize_if_changed
-        content_changed = self.content.changed?
-        yield
-        Sufia.queue.push(CharacterizeJob.new(self.pid)) if content_changed
-      end
-
       ## Extract the metadata from the content datastream and record it in the characterization datastream
       def characterize
         self.characterization.ng_xml = self.content.extract_metadata
@@ -89,7 +83,6 @@ module Sufia
         h
       end
 
-
     end
   end
 end

data/sufia-models/app/models/concerns/sufia/generic_file/derivatives.rb

@@ -0,0 +1,26 @@
+module Sufia
+  module GenericFile
+    module Derivatives
+      extend ActiveSupport::Concern
+
+      included do
+        include Hydra::Derivatives
+
+        makes_derivatives do |obj|
+          case obj.mime_type
+          when *pdf_mime_types
+            obj.transform_datastream :content, { thumbnail: { format: 'jpg', size: '338x493', datastream: 'thumbnail' } }
+          when *office_document_mime_types
+            obj.transform_datastream :content, { thumbnail: { format: 'jpg', size: '200x150>', datastream: 'thumbnail' } }, processor: :document
+          when *audio_mime_types
+            obj.transform_datastream :content, { mp3: { format: 'mp3', datastream: 'mp3' }, ogg: { format: 'ogg', datastream: 'ogg' } }, processor: :audio
+          when *video_mime_types
+            obj.transform_datastream :content, { webm: { format: 'webm', datastream: 'webm' }, mp4: { format: 'mp4', datastream: 'mp4' }, thumbnail: { format: 'jpg', datastream: 'thumbnail' } }, processor: :video
+          when *image_mime_types
+            obj.transform_datastream :content, { thumbnail: { format: 'jpg', size: '200x150>', datastream: 'thumbnail' } }
+          end
+        end
+      end
+    end
+  end
+end