bplmodels 0.0.91 → 0.0.93

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 45fa0e10e32adebfefb0eb641cf428a2e6f91fa8
-  data.tar.gz: 14f75699119c3990c4f8e18ceed8253deca191a7
+  metadata.gz: 1d0d4c255a889d52e7b90473c99af4462e38e539
+  data.tar.gz: 6421e9f724413858bab031e78c459f1c80c0ab09
 SHA512:
-  metadata.gz: 4e63652810c8d068c3eb2f91bb7cefcfe42ca29beae7f04f00cdd3864b6c076a7b6e7e963209675578d0828035d5e1832c0dca8154672c86ff706f62a66636a8
-  data.tar.gz: c0b1657ed6b96678b9dfb3237c28b561b5ae66da2ef676f55df6b5e12164c27ffc1d9279773482780eb022f05b83c35931318bd1baa9695b925ae0e531fa51ad
+  metadata.gz: d09b6154408234cbe5e06018b79fb98c7463716d4e335bdb563ee3d201475ae0a6644fd310fe0e1dcd688ec8dcd58b5c8bf37641d855c097172fe5a890f611ab
+  data.tar.gz: 09a2f2086cb22716f088c431caaa5d1de31e96757b4c34353ea4166b3c025ca0d369bccfa78d88f4751e9ee3c9eeb292fcfb5b217c4a850e3e50fa1761c832fd

data/app/models/bplmodels/characterization.rb

@@ -24,7 +24,9 @@ module Bplmodels
     end
 
     def fits_config_values
-      return {:file_title => :title, :file_author => :creator }
+      #This is to add that to the end metadata record from FITS...
+      #return {:file_title => :title, :file_author => :creator }
+      return {}
     end
 
     def width

data/app/models/bplmodels/collection.rb

@@ -8,7 +8,7 @@ module Bplmodels
 
     belongs_to :institutions, :class_name => 'Bplmodels::Institution', :property => :is_member_of
 
-    has_many :exemplary_image, :class_name => "ActiveFedora::Base", :property=> :is_exemplary_image_of
+    #has_many :exemplary_image, :class_name => "ActiveFedora::Base", :property=> :is_exemplary_image_of
 
     # Uses the Hydra modsCollection profile for collection list
     #has_metadata :name => "members", :type => Hydra::ModsCollectionMembers
@@ -52,25 +52,45 @@ module Bplmodels
     def to_solr(doc = {} )
       doc = super(doc)
 
+      basic_genre_array = ['Collections']
+      Bplmodels::ObjectBase.find_in_batches('is_member_of_collection_ssim'=>"info:fedora/#{self.pid}") do |group|
+        group.each { |object_id|
+          #object_id_array << Bplmodels::ObjectBase.find(object_id['id']).adapt_to_cmodel
+          basic_genre_array += object_id['genre_basic_ssim'] if object_id['genre_basic_ssim'].present?
+        }
+      end
+      doc['genre_basic_ssim'] = basic_genre_array.uniq!
+      doc['genre_basic_tsim'] = basic_genre_array.uniq!
+
       # description
       doc['abstract_tsim'] = self.descMetadata.abstract
 
+      # basic genre
+=begin
+      basic_genre = 'Collections'
+      doc['genre_basic_ssim'] = basic_genre
+      doc['genre_basic_tsim'] = basic_genre
+=end
+
       # institution
       if self.institutions
         collex_location = self.institutions.label.to_s
         doc['physical_location_ssim'] = collex_location
         doc['physical_location_tsim'] = collex_location
-        doc['institution_pid_si'] = self.institutions.pid
-        # TODO: need to remove the 3 above and refactor apps as necessary
-        # collections (AKA sets) have institutions, not locations
         doc['institution_name_ssim'] = collex_location
         doc['institution_name_tsim'] = collex_location
         doc['institution_pid_ssi'] = self.institutions.pid
       end
 
+      exemplary_check = Bplmodels::ImageFile.find_with_conditions({"is_exemplary_image_of_ssim"=>"info:fedora/#{self.pid}"}, rows: '1', fl: 'id' )
+      if exemplary_check.present?
+        doc['exemplary_image_ssi'] = exemplary_check.first["id"]
+      end
+=begin
       if self.exemplary_image.first != nil
         doc['exemplary_image_ssi'] = self.exemplary_image.first.pid
       end
+=end
 
       doc
 
@@ -88,7 +108,11 @@ module Bplmodels
       args[:namespace_id] ||= ARK_CONFIG_GLOBAL['namespace_commonwealth_pid']
 
       response = Typhoeus::Request.post(ARK_CONFIG_GLOBAL['url'] + "/arks.json", :params => {:ark=>{:parent_pid=>args[:parent_pid], :namespace_ark => ARK_CONFIG_GLOBAL['namespace_commonwealth_ark'], :namespace_id=>args[:namespace_id], :url_base => ARK_CONFIG_GLOBAL['ark_commonwealth_base'], :model_type => self.name, :local_original_identifier=>args[:local_id], :local_original_identifier_type=>args[:local_id_type]}})
-      as_json = JSON.parse(response.body)
+      begin
+        as_json = JSON.parse(response.body)
+      rescue => ex
+        raise('Error in JSON response for minting a collection pid.')
+      end
 
       Bplmodels::Collection.find_in_batches('id'=>as_json["pid"]) do |group|
         group.each { |solr_result|

data/app/models/bplmodels/ereader_file.rb

@@ -0,0 +1,14 @@
+module Bplmodels
+  class EreaderFile < Bplmodels::File
+
+    # Use a callback method to declare which derivatives you want
+    makes_derivatives :generate_derivatives
+
+    def fedora_name
+      'ereader_file'
+    end
+
+
+
+  end
+end

data/app/models/bplmodels/file.rb

@@ -18,6 +18,10 @@ module Bplmodels
 
     has_file_datastream 'access800', versionable: false,  label:'access800 datastream'
 
+    has_file_datastream 'ocrMaster', versionable: false,  label:'OCR master datastream'
+    has_file_datastream 'djvuCoords', versionable: false,  label:'djvu coordinate json datastream'
+
+
     belongs_to :object, :class_name => "Bplmodels::ObjectBase", :property => :is_image_of
 
     belongs_to :exemplary, :class_name => "Bplmodels::ObjectBase", :property => :is_exemplary_image_of
@@ -29,6 +33,16 @@ module Bplmodels
 
     has_metadata :name => "workflowMetadata", :type => WorkflowMetadata
 
+    has_metadata :name => "pageMetadata", :type => PageMetadata
+
+    def delete
+      #FIXME: This should just not be hardcoded in AVI Processor
+      if ActiveFedora.config.credentials[:url].match('fedora.digitalcommonwealth.org/fedora')
+        #cache_invalidate
+      end
+      super()
+    end
+
     def apply_default_permissions
       self.datastreams["rightsMetadata"].update_permissions( "group"=>{"Repository Administrators"=>"edit"} )
       self.save
@@ -46,6 +60,30 @@ module Bplmodels
         doc['marked_for_deletion_reason_ssi']  =  self.workflowMetadata.marked_for_deletion.reason.first
       end
 
+      if ['image/tiff', 'image/png', 'image/jpg', 'image/jp2'].include?(self.productionMaster.mimeType) and self.accessMaster.blank?
+        doc['derivative_processsed_ssi'] = 'false'
+      else
+        doc['derivative_processsed_ssi'] = 'true'
+      end
+
+      if self.ocrMaster.present?
+        doc['ocr_tsiv'] = Bplmodels::DatastreamInputFuncs.strip_value(self.ocrMaster.content.squish)
+        doc['has_ocr_text_bsi'] = true
+      end
+
+      if self.pageMetadata.present?
+        doc['page_type_ssi'] = self.pageMetadata.pageData.page.page_type.first
+        doc['hand_side_ssi'] = self.pageMetadata.pageData.page.hand_side.first
+        doc['page_num_label_ssi'] = self.pageMetadata.pageData.page.page_number.first if self.pageMetadata.pageData.page.page_number.present?
+        doc['page_num_label_type_ssi'] = self.pageMetadata.pageData.page.page_number.sequence.first if self.pageMetadata.pageData.page.page_number.present?
+        #doc['has_ocr_master_ssi'] = self.bookMetadata.book.page_data.page.has_ocrMaster.first
+        #doc['has_djvu_json_ssi'] = self.bookMetadata.book.page_data.page.has_djvu.first
+
+        if self.djvuCoords.present?
+          doc['has_djvu_json_ssi'] = 'true'
+        end
+      end
+
       doc['checksum_file_md5_ssi'] = self.productionMaster.checksum
 
       doc
@@ -53,23 +91,59 @@ module Bplmodels
     end
 
     def generate_derivatives
+      #Sample object on prod: https://fedora.digitalcommonwealth.org/fedora/objects/commonwealth:9w032896b/datastreams/productionMaster/content
+      #Sample object from test book object: https://fedoratest.bpl.org/fedora/objects/bpl-test:rf55zp77f/datastreams/productionMaster/content
       case self.productionMaster.mimeType
         when 'application/pdf'
           #transform_datastream :productionMaster, { :thumb => "100x100>" }
+
+          pdffile = Tempfile.new(['derivative','.pdf'])
+          pdffile.binmode
+          pdffile.write(self.productionMaster.content) #ActiveFedora.config.credentials[:url] + '/objects/' + self.pid + '/datastreams/productionMaster/content'
+          pdffile.close
+
+          current_page = 0
+          total_colors = 0
+          until total_colors > 1 do
+            #Won't work... asks for login with the brackets...
+            img = Magick::Image.read(pdffile.path + '[' + current_page.to_s + ']'){
+              self.quality = 100
+              self.density = 200
+            }.first
+            total_colors = img.total_colors
+            current_page = current_page + 1
+          end
+
+          #This is horrible. But if you don't do this, some PDF files won't come out right at all.
+          #Multiple attempts have failed to fix this but perhaps the bug will be patched in ImageMagick.
+          #To duplicate, one can use the PDF files at: http://libspace.uml.edu/omeka/files/original/7ecb4dc9579b11e2b53ccc2040e58d36.pdf
+          img = Magick::Image.from_blob( img.to_blob { self.format = "jpg" } ).first
+
+          thumb = img.resize_to_fit(300,300)
+
+          self.thumbnail300.content = thumb.to_blob { self.format = "jpg" }
+          self.thumbnail300.mimeType = 'image/jpeg'
+          self.thumbnail300.dsLabel = self.productionMaster.label
+
+          pdffile.delete
         when 'audio/wav'
           #transform_datastream :productionMaster, { :mp3 => {format: 'mp3'}, :ogg => {format: 'ogg'} }, processor: :audio
         when 'video/avi'
           #transform_datastream :productionMaster, { :mp4 => {format: 'mp4'}, :webm => {format: 'webm'} }, processor: :video
-        when 'image/tiff', 'image/png', 'image/jpg'
+        when 'image/tiff', 'image/png', 'image/jpg', 'image/png'
           begin
             transform_datastream :productionMaster, { :testJP2k => { recipe: :default, datastream: 'accessMaster'  } }, processor: 'jpeg2k_image'
           rescue => error
-            if error.message.include?('compressed TIFF files')
-              Magick::limit_resource(:memory, 5500000000)
-              Magick::limit_resource(:map, 5500000000)
-              #jp2_img =  Magick::Image.read("#{self.fedora_connection[0].options[:url]}/objects/#{self.pid}datastreams/productionMaster/content").first
-              jp2_img =  Magick::Image.from_blob(self.productionMaster.content).first
-              self.accessMaster.content = jp2_img.to_blob { self.format = "jp2" }
+            # First one is from Blue Books collection. Second one is from commonwealth:xd07m887b
+            if error.message.include?('compressed TIFF files') || error.message.include?("The number of colours associated with the colour space specified using")
+=begin
+              jp2_img = MiniMagick::Image.read(self.productionMaster.content) do |b|
+                b.format "jp2"
+              end
+=end
+              jp2_img = MiniMagick::Image.read(self.productionMaster.content)
+              jp2_img.format 'jp2'
+              self.accessMaster.content = jp2_img.to_blob
               self.accessMaster.mimeType = 'image/jp2'
               jp2_img.destroy!
             else
@@ -82,13 +156,62 @@ module Bplmodels
           self.thumbnail300.dsLabel = self.productionMaster.label
           self.access800.dsLabel = self.productionMaster.label
         when 'image/jpeg' #FIXME
-          Magick::limit_resource(:memory, 5500000000)
-          Magick::limit_resource(:map, 5500000000)
-          jp2_img =  Magick::Image.from_blob(self.productionMaster.content).first
-          self.accessMaster.content = jp2_img.to_blob { self.format = "jp2" }
+=begin
+          jp2_img = MiniMagick::Image.read(self.productionMaster.content) do |b|
+            b.format "jp2"
+          end
+=end
+          jp2_img = MiniMagick::Image.read(self.productionMaster.content)
+          jp2_img.format 'jp2'
+
+          self.accessMaster.content = jp2_img.to_blob
           self.accessMaster.mimeType = 'image/jp2'
           jp2_img.destroy!
+
+          transform_datastream :productionMaster, { :thumb => {size: "300x300>", datastream: 'thumbnail300', format: 'jpg'} }
+          transform_datastream :productionMaster, { :thumb => {size: "x800>", datastream: 'access800', format: 'jpg'} }
+          self.accessMaster.dsLabel = self.productionMaster.label
+          self.thumbnail300.dsLabel = self.productionMaster.label
+          self.access800.dsLabel = self.productionMaster.label
+        when 'image/jp2'
+          self.accessMaster.content = self.productionMaster.content
+          self.accessMaster.mimeType = 'image/jp2'
+
+
+=begin
+          source_path = "#{ActiveFedora.config.credentials[:url]}/#{self.productionMaster.url}"
+          image = MiniMagick::Image.open(source_path)
+          quality = image['%[channels]'] == 'gray' ? 'gray' : 'color'
+          long_dim = Hydra::Derivatives::Jpeg2kImage.long_dim(image)
+          directives = { recipe: :default, datastream: 'accessMaster'  }
+          recipe = Hydra::Derivatives::Jpeg2kImage.kdu_compress_recipe(directives, quality, long_dim)
+
+          file_path = Hydra::Derivatives::Jpeg2kImage.tmp_file('.jp2')
+          image.write file_path
+
+
+          output_file = Hydra::Derivatives::Jpeg2kImage.tmp_file('.jp2')
+          Hydra::Derivatives::Jpeg2kImage.encode(file_path, recipe, output_file)
+          self.accessMaster.content = ::File.open(output_file)
+          self.accessMaster.mimeType = 'image/jp2'
+          #End wonky code
+
+          #TODO: Move this to a function in Hydra Derivatives...
+          self.save
+          image.destroy!
+          File.unlink(output_file)
+          File.unlink(file_path)
+=end
+
+          transform_datastream :productionMaster, { :thumb => {size: "300x300>", datastream: 'thumbnail300', format: 'jpg'} }
+          transform_datastream :productionMaster, { :thumb => {size: "x800>", datastream: 'access800', format: 'jpg'} }
+          self.accessMaster.dsLabel = self.productionMaster.label
+          self.thumbnail300.dsLabel = self.productionMaster.label
+          self.access800.dsLabel = self.productionMaster.label
+
+
       end
+
     end
 
     def derivative_service(is_new)
@@ -117,7 +240,11 @@ module Bplmodels
       args[:namespace_id] ||= ARK_CONFIG_GLOBAL['namespace_commonwealth_pid']
 
       response = Typhoeus::Request.post(ARK_CONFIG_GLOBAL['url'] + "/arks.json", :params => {:ark=>{:parent_pid=>args[:parent_pid], :namespace_ark => ARK_CONFIG_GLOBAL['namespace_commonwealth_ark'], :namespace_id=>args[:namespace_id], :url_base => ARK_CONFIG_GLOBAL['ark_commonwealth_base'], :model_type => self.name, :local_original_identifier=>args[:local_id], :local_original_identifier_type=>args[:local_id_type]}})
-      as_json = JSON.parse(response.body)
+      begin
+        as_json = JSON.parse(response.body)
+      rescue => ex
+        raise('Error in JSON response for minting a file pid.')
+      end
 
       #Below stopped working suddenly?
 =begin
@@ -147,5 +274,18 @@ module Bplmodels
       return object
     end
 
+
+    def cache_invalidate
+      response = Typhoeus::Request.post(DERIVATIVE_CONFIG_GLOBAL['url'] + "/processor/objectcacheinvalidation.json", :params => {:file_pid=>self.pid, :environment=>Bplmodels.environment})
+      as_json = JSON.parse(response.body)
+
+      if as_json['result'] == "false"
+        raise "Error Deleting the Cache! Server error!"
+      end
+
+      return true
+    end
+
+
   end
 end

data/app/models/bplmodels/institution.rb

@@ -5,6 +5,8 @@ module Bplmodels
 
     has_metadata :name => "workflowMetadata", :type => WorkflowMetadata
 
+    has_many :exemplary_image, :class_name => "Bplmodels::File", :property=> :is_exemplary_image_of
+
     #A collection can have another collection as a member, or an image
     def insert_member(fedora_object)
       if (fedora_object.instance_of?(Bplmodels::Collection))
@@ -27,9 +29,59 @@ module Bplmodels
       'institution'
     end
 
+    def new_examplary_image(file_path, file_name)
+
+      #Delete any old images
+      ActiveFedora::Base.find_in_batches('is_exemplary_image_of_ssim'=>"info:fedora/#{self.pid}") do |group|
+        group.each { |image_solr|
+            current_image_file = ActiveFedora::Base.find(image_solr['id']).adapt_to_cmodel
+            current_image_file.delete
+        }
+      end
+
+
+      image_file = Bplmodels::ImageFile.mint(:parent_pid=>self.pid, :local_id=>file_name, :local_id_type=>'File Name', :label=>file_name, :institution_pid=>self.pid)
+
+      datastream = 'productionMaster'
+      image_file.send(datastream).content = ::File.open(file_path)
+
+      if file_name.split('.').last.downcase == 'tif'
+        image_file.send(datastream).mimeType = 'image/tiff'
+      elsif file_name.split('.').last.downcase == 'jpg'
+        image_file.send(datastream).mimeType = 'image/jpeg'
+      elsif file_name.split('.').last.downcase == 'jp2'
+        image_file.send(datastream).mimeType = 'image/jp2'
+      elsif file_name.split('.').last.downcase == 'png'
+        image_file.send(datastream).mimeType = 'image/png'
+      elsif file_name.split('.').last.downcase == 'txt'
+        image_file.send(datastream).mimeType = 'text/plain'
+      else
+        #image_file.send(datastream).mimeType = 'image/jpeg'
+        raise "Could not find a mimeType for #{file_name.split('.').last.downcase}"
+      end
+
+      image_file.send(datastream).dsLabel = file_name.gsub(/\.(tif|TIF|jpg|JPG|jpeg|JPEG|jp2|JP2|png|PNG|txt|TXT)$/, '')
+
+      #FIXME!!!
+      original_file_location = "Uploaded file of #{file_path}"
+      image_file.workflowMetadata.insert_file_source(original_file_location,file_name,datastream)
+      image_file.workflowMetadata.item_status.state = "published"
+      image_file.workflowMetadata.item_status.state_comment = "Added via the institution admin form using Bplmodels new_exemplary_image method on " + Time.new.year.to_s + "/" + Time.new.month.to_s + "/" + Time.new.day.to_s
+
+      image_file.add_relationship(:is_image_of, "info:fedora/" + self.pid)
+      image_file.add_relationship(:is_file_of, "info:fedora/" + self.pid)
+      image_file.add_relationship(:is_exemplary_image_of, "info:fedora/" + self.pid)
+
+
+      image_file.save
+      image_file
+    end
+
     def to_solr(doc = {} )
       doc = super(doc)
 
+
+
       # description
       doc['abstract_tsim'] = self.descMetadata.abstract
 
@@ -118,6 +170,22 @@ module Bplmodels
       doc['institution_pid_si'] = self.pid
       doc['institution_pid_ssi'] = self.pid
 
+      # basic genre
+      basic_genre = 'Institutions'
+      doc['genre_basic_ssim'] = basic_genre
+      doc['genre_basic_tsim'] = basic_genre
+
+      # physical location
+      # slightly redundant, but needed for faceting and A-Z filtering
+      institution_name = self.descMetadata.mods(0).title.first
+      doc['physical_location_ssim'] = institution_name
+      doc['physical_location_tsim'] = institution_name
+
+      exemplary_check = Bplmodels::ImageFile.find_with_conditions({"is_exemplary_image_of_ssim"=>"info:fedora/#{self.pid}"}, rows: '1', fl: 'id' )
+      if exemplary_check.present?
+        doc['exemplary_image_ssi'] = exemplary_check.first["id"]
+      end
+
 
       doc
 
@@ -135,7 +203,11 @@ module Bplmodels
       #TODO: Duplication check here to prevent over-writes?
 
       response = Typhoeus::Request.post(ARK_CONFIG_GLOBAL['url'] + "/arks.json", :params => {:ark=>{:namespace_ark => ARK_CONFIG_GLOBAL['namespace_commonwealth_ark'], :namespace_id=>ARK_CONFIG_GLOBAL['namespace_commonwealth_pid'], :url_base => ARK_CONFIG_GLOBAL['ark_commonwealth_base'], :model_type => self.name, :local_original_identifier=>args[:local_id], :local_original_identifier_type=>args[:local_id_type]}})
-      as_json = JSON.parse(response.body)
+      begin
+        as_json = JSON.parse(response.body)
+      rescue => ex
+        raise('Error in JSON response for minting an institution pid.')
+      end
 
       Bplmodels::Institution.find_in_batches('id'=>as_json["pid"]) do |group|
         group.each { |solr_result|

data/app/models/bplmodels/marc_file.rb

@@ -0,0 +1,14 @@
+module Bplmodels
+  class MarcFile < Bplmodels::File
+
+    # Use a callback method to declare which derivatives you want
+    makes_derivatives :generate_derivatives
+
+    def fedora_name
+      'marc_file'
+    end
+
+
+
+  end
+end