hydra-works 0.2.0 → 0.3.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 28a2824769ab44dca4138961476455d25db0d016
-  data.tar.gz: 7e0c7dae7cc6873680eb07bf882a1cc1b008e13e
+  metadata.gz: 1f59d9e4057f3221480712f84c8c1eca414d64c1
+  data.tar.gz: 9aaa3a1afbcb5e27e5998822959d988f34d01101
 SHA512:
-  metadata.gz: 0ea7c389ceaeffe31367dffe19490162932afa1abeb5f5016ea2842003c7f36cc00bbe25a7b6160766f684fcb6009817074532d0bfc936928c6b0f570f9767be
-  data.tar.gz: cc6825f4d60eed260531714eb646d512debcdb004feba0c81f80b253c7ba935289a614202efa225966f178d677ea99898b135cd8619f911d7401aaf0f9b802e0
+  metadata.gz: 539c2c4c03e59c2e4d1ee667ad087bf0aa28551b7862acf32df477d42a137e03f1024b7f94fad1bb1fc12bbab6a82338f443b568762470c0f092c8d6c86934bf
+  data.tar.gz: 43ae1663d7070c525b6be37a1cb03ded81c4a7fa135ecb243d8fe639806414325ff4401c5d5fcde35415a153479daa7a3c6185b55281328dc705ad6f4fe32b42

data/.rubocop.yml

@@ -15,23 +15,28 @@ Metrics/LineLength:
 
 Metrics/CyclomaticComplexity:
   Exclude:
-    - lib/hydra/works/services/generic_file/add_file_to_generic_file.rb
+    - lib/hydra/works/services/add_file_to_file_set.rb
 
 Metrics/PerceivedComplexity:
   Exclude:
-    - lib/hydra/works/services/generic_file/add_file_to_generic_file.rb
+    - lib/hydra/works/services/add_file_to_file_set.rb
 
 Metrics/MethodLength:
   Exclude:
-    - lib/hydra/works/services/generic_file/add_file_to_generic_file.rb
-    - lib/hydra/works/models/concerns/generic_file/virus_check.rb
+    - lib/hydra/works/services/add_file_to_file_set.rb
+    - lib/hydra/works/models/concerns/file_set/virus_check.rb
+    - lib/hydra/works/models/characterization/fits_datastream.rb
+
+Metrics/ClassLength:
+  Exclude:
+    - lib/hydra/works/models/characterization/fits_datastream.rb
 
 Metrics/AbcSize:
   Exclude:
-    - lib/hydra/works/services/generic_file/add_file_to_generic_file.rb
-    - lib/hydra/works/services/generic_file/full_text_extraction_service.rb	
-    - lib/hydra/works/models/concerns/generic_file/mime_types.rb
-    - lib/hydra/works/models/concerns/generic_file/virus_check.rb
+    - lib/hydra/works/services/add_file_to_file_set.rb
+    - lib/hydra/works/services/full_text_extraction_service.rb
+    - lib/hydra/works/models/concerns/file_set/mime_types.rb
+    - lib/hydra/works/models/concerns/file_set/virus_check.rb
 
 Style/CollectionMethods:
   PreferredMethods:
@@ -70,3 +75,7 @@ RSpec/FilePath:
 
 RSpec/InstanceVariable:
   Enabled: false
+
+RSpec/DescribedClass:
+  Exclude:
+    - spec/hydra/works/models/concerns/generic_file/characterization_spec.rb

data/README.md

@@ -11,9 +11,9 @@
 
 The Hydra::Works gem provides a set of [Portland Common Data Model](https://github.com/duraspace/pcdm/wiki)-compliant ActiveFedora models and associated behaviors around the broad concept of multi-file "works", the need for which was expressed by a variety of [community use cases](https://github.com/projecthydra-labs/hydra-works/tree/master/use-cases). The Hydra::Works domain model includes:
 
- * **GenericFile**: a *Hydra::PCDM::Object* that encapsulates one or more directly related *Hydra::PCDM::File*s, such as a PDF document, its derivatives, and extracted full-text
- * **GenericWork**: a *Hydra::PCDM::Object* that holds zero or more **GenericFile**s and zero or more **GenericWork**s
- * **Collection**: a *Hydra::PCDM::Collection* that indirectly contains zero or more **GenericWork**s and zero or more **Collection**s
+ * **FileSet**: a *Hydra::PCDM::Object* that encapsulates one or more directly related *Hydra::PCDM::File*s, such as a PDF document, its derivatives, and extracted full-text
+ * **GenericWork**: a *Hydra::PCDM::Object* that holds zero or more **FileSets**s and zero or more **Work**s (often you won't use GenericWork, but instead write your own Work class)
+ * **Collection**: a *Hydra::PCDM::Collection* that indirectly contains zero or more **Work**s and zero or more **Collection**s
 
 View [a diagram of the domain model](https://docs.google.com/drawings/d/1-NkkRPpGpZGoTimEpYTaGM1uUPRaT0SamuWDITvtG_8/edit).
 
@@ -43,18 +43,18 @@ class Collection < ActiveFedora::Base
 end
 
 class Book < ActiveFedora::Base
-  include Hydra::Works::GenericWorkBehavior
+  include Hydra::Works::WorkBehavior
 end
 
 class Page < ActiveFedora::Base
-  include Hydra::Works::GenericFileBehavior
+  include Hydra::Works::FileSetBehavior
 end
 
 collection = Collection.create
 book = BookWork.create
 page = Page.create
 
-collection.generic_works << book
+collection.works << book
 collection.save
 book.generic_files << page
 book.save
@@ -70,12 +70,12 @@ To turn on virus detection, install clamav on your system and add the `clamav` g
 
     gem 'clamav'
 
-Then include the `VirusCheck` module in your `GenericFile` class:
+Then include the `VirusCheck` module in your `FileSet` class:
 
 ```ruby
 class BookFiles < ActiveFedora::Base
-  include Hydra::Works::GenericFileBehavior
-  include Hydra::Works::GenericFile::VirusCheck
+  include Hydra::Works::FileSetBehavior
+  include Hydra::Works::VirusCheck
 end
 ```
 

data/hydra-works.gemspec

@@ -18,10 +18,9 @@ Gem::Specification.new do |spec|
   spec.test_files    = spec.files.grep(%r{^(test|spec|features)/})
   spec.require_paths = ['lib']
 
-  spec.add_dependency 'hydra-pcdm', '~> 0.2'
-  spec.add_dependency 'hydra-derivatives', '~> 2.0'
-  spec.add_dependency 'active-fedora', '>= 9.4.1'
-  spec.add_dependency 'activefedora-aggregation', '~> 0.4'
+  spec.add_dependency 'hydra-pcdm', '~> 0.3'
+  spec.add_dependency 'hydra-derivatives', '~> 3.0'
+  spec.add_dependency 'hydra-file_characterization', '~> 0.3', '>= 0.3.3'
 
   spec.add_development_dependency 'bundler', '~> 1.7'
   spec.add_development_dependency 'rake', '~> 10.0'

data/lib/hydra/works.rb

@@ -14,29 +14,38 @@ module Hydra
       end
     end
 
+    autoload_under 'models/concerns/file_set' do
+      autoload :Derivatives
+      autoload :MimeTypes
+      autoload :ContainedFiles
+      autoload :VersionedContent
+      autoload :VirusCheck
+    end
+
+    autoload :Characterization
+
     autoload_under 'models' do
       autoload :Collection
+      autoload :FileSet
       autoload :GenericWork
-      autoload :GenericFile
+      autoload :GenericFile # Deprecated. Remove in 0.4.0
     end
 
     autoload_under 'models/concerns' do
       autoload :CollectionBehavior
-      autoload :GenericWorkBehavior
-      autoload :GenericFileBehavior
-      autoload :BlockChildObjects
+      autoload :FileSetBehavior
+      autoload :GenericWorkBehavior # Deprecated. Remove in 0.4.0
+      autoload :GenericFileBehavior # Deprecated. Remove in 0.4.0
+      autoload :WorkBehavior
     end
 
-    autoload_under 'services/generic_file' do
-      autoload :AddFileToGenericFile
-      autoload :GenerateThumbnail
-      autoload :UploadFileToGenericFile
+    autoload_under 'services' do
+      autoload :AddFileToFileSet
+      autoload :AddFileToGenericFile # Deprecated. Remove in 0.4.0
+      autoload :UploadFileToFileSet
+      autoload :UploadFileToGenericFile # Deprecated. Remove in 0.4.0
       autoload :PersistDerivative
-      autoload :FullTextExtractionService
-    end
-
-    autoload_under 'errors' do
-      autoload :FullTextExtractionError
+      autoload :CharacterizationService
     end
   end
 end

data/lib/hydra/works/characterization.rb

@@ -0,0 +1,31 @@
+module Hydra::Works
+  module Characterization
+    extend ActiveSupport::Concern
+    extend ActiveSupport::Autoload
+
+    autoload :FitsDatastream, 'hydra/works/models/characterization/fits_datastream.rb'
+    autoload :AlreadyThereStrategy, 'hydra/works/models/characterization/already_there_strategy.rb'
+
+    autoload_under 'schema' do
+      autoload :AudioSchema
+      autoload :BaseSchema
+      autoload :DocumentSchema
+      autoload :ImageSchema
+      autoload :VideoSchema
+    end
+
+    autoload :Base, 'hydra/works/models/concerns/file_set/characterization/base.rb'
+    autoload :Image, 'hydra/works/models/concerns/file_set/characterization/image.rb'
+    autoload :Document, 'hydra/works/models/concerns/file_set/characterization/document.rb'
+    autoload :Video, 'hydra/works/models/concerns/file_set/characterization/video.rb'
+    autoload :Audio, 'hydra/works/models/concerns/file_set/characterization/audio.rb'
+
+    included do
+      include Base
+      include Image
+      include Audio
+      include Video
+      include Document
+    end
+  end
+end

data/lib/hydra/works/characterization/schema/audio_schema.rb

@@ -0,0 +1,12 @@
+module Hydra::Works::Characterization
+  class AudioSchema < ActiveTriples::Schema
+    property :bit_depth, predicate: RDF::Vocab::NFO.bitDepth
+    property :channels, predicate: RDF::Vocab::NFO.channels
+    property :data_format, predicate: RDF::Vocab::EBUCore.hasDataFormat
+    property :frame_rate, predicate: RDF::Vocab::NFO.frameRate
+    property :duration, predicate: RDF::Vocab::NFO.duration
+    property :sample_rate, predicate: RDF::Vocab::EBUCore.sampleRate
+    # properties without cannonical URIs
+    property :offset, predicate: RDF::URI('http://projecthydra.org/ns/audio/offset')
+  end
+end

data/lib/hydra/works/characterization/schema/base_schema.rb

@@ -0,0 +1,17 @@
+module Hydra::Works::Characterization
+  class BaseSchema < ActiveTriples::Schema
+    property :filename, predicate: RDF::Vocab::EBUCore.filename, multiple: false
+    property :format_label, predicate: RDF::Vocab::PREMIS.hasFormatName
+    property :file_size, predicate: RDF::Vocab::EBUCore.fileSize
+    property :well_formed, predicate: RDF::URI.new("http://projecthydra.org/ns/fits/wellFormed")
+    property :valid, predicate: RDF::URI.new("http://projecthydra.org/ns/fits/valid")
+    property :date_created, predicate: RDF::Vocab::EBUCore.dateCreated
+    property :last_modified, predicate: RDF::Vocab::EBUCore.dateModified
+    property :fits_version, predicate: RDF::Vocab::PREMIS.hasCreatingApplicationVersion
+    property :exif_version, predicate: RDF::Vocab::EXIF.exifVersion
+    property :original_checksum, predicate: RDF::Vocab::NFO.hashValue
+    property :mime_type, predicate: RDF::Vocab::EBUCore.hasMimeType, multiple: false do |index|
+      index.as :stored_searchable
+    end
+  end
+end

data/lib/hydra/works/characterization/schema/document_schema.rb

@@ -0,0 +1,18 @@
+module Hydra::Works::Characterization
+  class DocumentSchema < ActiveTriples::Schema
+    property :file_title, predicate: RDF::DC11.title
+    property :file_author, predicate: RDF::DC11.creator
+    property :page_count, predicate: RDF::Vocab::NFO.pageCount
+    property :file_language, predicate: RDF::DC11.language
+    property :word_count, predicate: RDF::Vocab::NFO.wordCount
+    property :character_count, predicate: RDF::Vocab::NFO.characterCount
+    property :line_count, predicate: RDF::Vocab::NFO.lineCount
+    property :character_set, predicate: RDF::URI.new("http://www.w3.org/2011/content#characterEncoding")
+    property :markup_basis, predicate: RDF::URI.new("http://www.w3.org/2011/content#doctypeName")
+    property :markup_language, predicate: RDF::URI.new("http://www.w3.org/2011/content#systemId")
+    # properties without canonical URIs
+    property :paragraph_count, predicate: RDF::URI.new("http://projecthydra.org/ns/odf/paragraphCount")
+    property :table_count, predicate: RDF::URI.new("http://projecthydra.org/ns/odf/tableCount")
+    property :graphics_count, predicate: RDF::URI.new("http://projecthydra.org/ns/odf/graphicsCount")
+  end
+end

data/lib/hydra/works/characterization/schema/image_schema.rb

@@ -0,0 +1,22 @@
+module Hydra::Works::Characterization
+  class ImageSchema < ActiveTriples::Schema
+    property :byte_order, predicate: RDF::URI('http://sweet.jpl.nasa.gov/2.2/reprDataFormat.owl#byteOrder'), multiple: false
+    property :compression, predicate: RDF::URI('http://projecthydra.org/ns/mix/compressionScheme'), multiple: false
+    property :height, predicate: RDF::Vocab::EBUCore.height
+    property :width, predicate: RDF::Vocab::EBUCore.width
+    # property :width, predicate: RDF::Vocab::EXIF.imageWidth, multiple: false
+    # property :height, predicate: RDF::Vocab::EXIF.imageLength, multiple: false
+    property :color_space, predicate: RDF::Vocab::EXIF.colorSpace, multiple: false
+    property :profile_name, predicate: RDF::URI('http://projecthydra.org/ns/mix/colorProfileName')
+    property :profile_version, predicate: RDF::URI('http://projecthydra.org/ns/mix/colorProfileVersion')
+    property :orientation, predicate: RDF::Vocab::EXIF.orientation, multiple: false
+    property :color_map, predicate: RDF::URI('http://projecthydra.org/ns/mix/colorMap')
+    property :image_producer, predicate: RDF::Vocab::MARCRelators.pht
+    property :capture_device, predicate: RDF::Vocab::EXIF.model
+    property :scanning_software, predicate: RDF::Vocab::EXIF.software
+    # exif_version is in core set of tehcnical metadata
+    property :gps_timestamp, predicate: RDF::Vocab::EXIF.gpsTimeStamp
+    property :latitude, predicate: RDF::Vocab::EXIF.gpsLatitude
+    property :longitude, predicate: RDF::Vocab::EXIF.gpsLongitude
+  end
+end

data/lib/hydra/works/characterization/schema/video_schema.rb

@@ -0,0 +1,9 @@
+module Hydra::Works::Characterization
+  class VideoSchema < ActiveTriples::Schema
+    property :height, predicate: RDF::Vocab::EBUCore.height
+    property :width, predicate: RDF::Vocab::EBUCore.width
+    property :frame_rate, predicate: RDF::Vocab::NFO.frameRate
+    property :duration, predicate: RDF::Vocab::NFO.duration
+    property :sample_rate, predicate: RDF::Vocab::EBUCore.sampleRate
+  end
+end

data/lib/hydra/works/models/characterization/already_there_strategy.rb

@@ -0,0 +1,12 @@
+module Hydra::Works::Characterization
+  class AlreadyThereStrategy < ActiveTriples::ExtensionStrategy
+    # override apply method to check if property already exists or reciever already has predicate defined.
+    # Do not add property if the rdf_resource already responds to the property name
+    # Do not add property if the rdf_resource already has a property with the same predicate.
+    def self.apply(resource, property)
+      return if resource.respond_to?(property.name)
+      return if resource.properties.any? { |p| p[1].predicate == property.predicate }
+      resource.property property.name, property.to_h
+    end
+  end
+end

data/lib/hydra/works/models/characterization/fits_datastream.rb

@@ -0,0 +1,162 @@
+module Hydra::Works::Characterization
+  class FitsDatastream < ActiveFedora::OmDatastream
+    include OM::XML::Document
+
+    set_terminology do |t|
+      t.root(path: 'fits',
+             xmlns: 'http://hul.harvard.edu/ois/xml/ns/fits/fits_output',
+             schema: 'http://hul.harvard.edu/ois/xml/xsd/fits/fits_output.xsd')
+      t.fits_v(path: { attribute: 'version' })
+      t.identification do
+        t.identity do
+          t.format_label(path: { attribute: 'format' })
+          t.mime_type(path: { attribute: 'mimetype' })
+          t.tool(attributes: { toolname: "Exiftool" }) do
+            t.exif_tool_version(path: { attribute: 'toolversion' })
+          end
+        end
+      end
+      t.fileinfo do
+        t.file_size(path: 'size')
+        t.last_modified(path: 'lastmodified', attributes: { toolname: "Exiftool" })
+        t.filename(path: 'filename')
+        t.original_checksum(path: 'md5checksum')
+        t.date_created(path: 'created')
+        t.rights_basis(path: 'rightsBasis')
+        t.copyright_basis(path: 'copyrightBasis')
+        t.copyright_note(path: 'copyrightNote')
+      end
+      t.filestatus do
+        t.well_formed(path: 'well-formed')
+        t.valid(path: 'valid')
+        t.status_message(path: 'message')
+      end
+      t.metadata do
+        t.document do
+          t.file_title(path: 'title')
+          t.file_author(path: 'author')
+          t.file_language(path: 'language')
+          t.page_count(path: 'pageCount')
+          t.word_count(path: 'wordCount')
+          t.character_count(path: 'characterCount')
+          t.paragraph_count(path: 'paragraphCount')
+          t.line_count(path: 'lineCount')
+          t.table_count(path: 'tableCount')
+          t.graphics_count(path: 'graphicsCount')
+        end
+        t.image do
+          t.byte_order(path: 'byteOrder')
+          t.compression(path: 'compressionScheme')
+          t.width(path: 'imageWidth')
+          t.height(path: 'imageHeight')
+          t.color_space(path: 'colorSpace')
+          t.profile_name(path: 'iccProfileName')
+          t.profile_version(path: 'iccProfileVersion')
+          t.orientation(path: 'orientation')
+          t.color_map(path: 'colorMap')
+          t.image_producer(path: 'imageProducer')
+          t.capture_device(path: 'captureDevice')
+          t.scanning_software(path: 'scanningSoftwareName')
+          t.exif_version(path: 'exifVersion')
+          t.gps_timestamp(path: 'gpsTimeStamp')
+          t.latitude(path: 'gpsDestLatitude')
+          t.longitude(path: 'gpsDestLongitude')
+        end
+        t.text do
+          t.character_set(path: 'charset')
+          t.markup_basis(path: 'markupBasis')
+          t.markup_language(path: 'markupLanguage')
+        end
+        t.audio do
+          t.duration(path: 'duration')
+          t.bit_depth(path: 'bitDepth')
+          t.sample_rate(path: 'sampleRate')
+          t.channels(path: 'channels')
+          t.data_format(path: 'dataFormatType')
+          t.offset(path: 'offset')
+        end
+        t.video do
+          t.width(path: 'imageWidth')
+          t.height(path: 'imageHeight')
+          t.duration(path: 'duration')
+          t.sample_rate(path: 'sampleRate')
+          t.audio_sample_rate(path: 'audioSampleRate')
+          t.frame_rate(path: 'frameRate')
+        end
+      end
+      # fits_version needs a different name than it's target node since they're at the same level
+      t.fits_version(proxy: [:fits, :fits_v])
+      t.format_label(proxy: [:identification, :identity, :format_label])
+      # Can't use .mime_type because it's already defined for datastream so method missing won't work.
+      t.file_mime_type(proxy: [:identification, :identity, :mime_type])
+      t.exif_tool_version(proxy: [:identification, :identity, :tool, :exif_tool_version])
+      t.file_size(proxy: [:fileinfo, :file_size])
+      t.last_modified(proxy: [:fileinfo, :last_modified])
+      t.filename(proxy: [:fileinfo, :filename])
+      t.original_checksum(proxy: [:fileinfo, :original_checksum])
+      t.date_created(proxy: [:fileinfo, :date_created])
+      t.rights_basis(proxy: [:fileinfo, :rights_basis])
+      t.copyright_basis(proxy: [:fileinfo, :copyright_basis])
+      t.copyright_note(proxy: [:fileinfo, :copyright_note])
+      t.well_formed(proxy: [:filestatus, :well_formed])
+      t.valid(proxy: [:filestatus, :valid])
+      t.status_message(proxy: [:filestatus, :status_message])
+      t.file_title(proxy: [:metadata, :document, :file_title])
+      t.file_author(proxy: [:metadata, :document, :file_author])
+      t.page_count(proxy: [:metadata, :document, :page_count])
+      t.file_language(proxy: [:metadata, :document, :file_language])
+      t.word_count(proxy: [:metadata, :document, :word_count])
+      t.character_count(proxy: [:metadata, :document, :character_count])
+      t.paragraph_count(proxy: [:metadata, :document, :paragraph_count])
+      t.line_count(proxy: [:metadata, :document, :line_count])
+      t.table_count(proxy: [:metadata, :document, :table_count])
+      t.graphics_count(proxy: [:metadata, :document, :graphics_count])
+      t.byte_order(proxy: [:metadata, :image, :byte_order])
+      t.compression(proxy: [:metadata, :image, :compression])
+      t.width(proxy: [:metadata, :image, :width])
+      t.video_width(proxy: [:metadata, :video, :width])
+      t.height(proxy: [:metadata, :image, :height])
+      t.video_height(proxy: [:metadata, :video, :height])
+      t.color_space(proxy: [:metadata, :image, :color_space])
+      t.profile_name(proxy: [:metadata, :image, :profile_name])
+      t.profile_version(proxy: [:metadata, :image, :profile_version])
+      t.orientation(proxy: [:metadata, :image, :orientation])
+      t.color_map(proxy: [:metadata, :image, :color_map])
+      t.image_producer(proxy: [:metadata, :image, :image_producer])
+      t.capture_device(proxy: [:metadata, :image, :capture_device])
+      t.scanning_software(proxy: [:metadata, :image, :scanning_software])
+      t.exif_version(proxy: [:metadata, :image, :exif_version])
+      t.gps_timestamp(proxy: [:metadata, :image, :gps_timestamp])
+      t.latitude(proxy: [:metadata, :image, :latitude])
+      t.longitude(proxy: [:metadata, :image, :longitude])
+      t.character_set(proxy: [:metadata, :text, :character_set])
+      t.markup_basis(proxy: [:metadata, :text, :markup_basis])
+      t.markup_language(proxy: [:metadata, :text, :markup_language])
+      t.audio_duration(proxy: [:metadata, :audio, :duration])
+      t.video_duration(proxy: [:metadata, :video, :duration])
+      t.bit_depth(proxy: [:metadata, :audio, :bit_depth])
+      t.audio_sample_rate(proxy: [:metadata, :audio, :sample_rate])
+      t.video_sample_rate(proxy: [:metadata, :video, :sample_rate])
+      t.video_audio_sample_rate(proxy: [:metadata, :video, :audio_sample_rate])
+      t.channels(proxy: [:metadata, :audio, :channels])
+      t.data_format(proxy: [:metadata, :audio, :data_format])
+      t.offset(proxy: [:metadata, :audio, :offset])
+      t.frame_rate(proxy: [:metadata, :video, :frame_rate])
+    end
+
+    def self.xml_template
+      builder = Nokogiri::XML::Builder.new do |xml|
+        xml.fits(xmlns: 'http://hul.harvard.edu/ois/xml/ns/fits/fits_output',
+                 'xmlns:xsi' => 'http://www.w3.org/2001/XMLSchema-instance',
+                 'xsi:schemaLocation' =>
+      "http://hul.harvard.edu/ois/xml/ns/fits/fits_output
+      http://hul.harvard.edu/ois/xml/xsd/fits/fits_output.xsd",
+                 version: '0.6.0',
+                 timestamp: '1/25/12 11:04 AM') do
+          xml.identification { xml.identity(toolname: 'FITS') }
+        end
+      end
+      builder.doc
+    end
+  end
+end