iiif_print 3.2.2 → 3.2.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 6273bfe83a864471ade7c0735cffc768d80d15054c34d479ae00f34993703329
-  data.tar.gz: 17651696ee25b9bedad6c415f8ab804d99adf13bc943e08f1f0fa910c3cbb70d
+  metadata.gz: 63e30799cf45be6cb719486414c6ce6ffaec6742823fad504b37777891bec6a7
+  data.tar.gz: 5c33a0f762a1ee6d74b160704f244bc0447fdabce3670878a3973e802dc01446
 SHA512:
-  metadata.gz: 269d3160d6d7dcade6c1455f336fd8d0d68401f2e895fd7c99f3b9f31ad662eb1ced1320265eb2b6c21d98c32f49c8e806695a29fe5042024826eed9d1f846d6
-  data.tar.gz: a8a38e1bdd90ccb00a294548b4f9bb2c0cd1fd2b0c0e5f45a457c0a738e097c24f907e8349267ab1f3de4d4e50228001629e72ede031effb0de0792926b341c3
+  metadata.gz: 977f1e11d02f47a6ee5fd31bf42f4574179529c71e534ee175cf2d947a0f3d6684a553da7f2f68ea205f681ce8fc116b377a2de51f39c138218f6c7f268d423b
+  data.tar.gz: 87c6d0397b7eb0c96f678e1023ce22820b82f3c543b3126b18778f1027c80b27d6a0c065a64c3dfaa3a70490af039d4d760c1f3eb1b0ab673c2af7bd8afd0790

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: /app/samvera/hyrax-webapp
   specs:
-    iiif_print (3.2.2)
+    iiif_print (3.2.3)
       blacklight_iiif_search (>= 1.0, < 3.0)
       derivative-rodeo (~> 0.5)
       hyrax (>= 2.5, < 6)

data/lib/iiif_print/persistence_layer/active_fedora_adapter.rb

@@ -181,6 +181,14 @@ module IiifPrint
         true
       end
 
+      ##
+      # does nothing for ActiveFedora; it keeps no records for extracted text.
+      #
+      # @return [TrueClass]
+      def self.remove_extracted_text_for(*)
+        true
+      end
+
       ##
       # Extract text from the derivatives
       #

data/lib/iiif_print/persistence_layer/valkyrie_adapter.rb

@@ -207,6 +207,29 @@ module IiifPrint
         Hyrax::ValkyriePersistDerivatives.call(stream, directives)
       end
 
+      ##
+      # Remove the fileset's extracted text records, so that re-extracting
+      # text replaces them rather than adding to them.  The stored files are
+      # kept: the new records are written to the same derivative paths.
+      #
+      # @param [Hyrax::FileSet] a Valkyrie fileset
+      # @return [TrueClass, nil] nil when there is nothing to remove
+      def self.remove_extracted_text_for(file_set:)
+        # re-read: earlier derivative plugins in the same job (e.g. the thumbnail)
+        # have saved the file set since the caller loaded it
+        file_set = Hyrax.query_service.find_by(id: file_set.id)
+        extracted = Hyrax.custom_queries.find_many_file_metadata_by_use(resource: file_set,
+        use: Hyrax::FileMetadata::Use.uri_for(use: :extracted_file)).to_a
+        return if extracted.empty?
+
+        # file_ids can hold Strings as well as Valkyrie::IDs, so compare as strings
+        extracted_ids = extracted.map { |file_metadata| file_metadata.id.to_s }
+        file_set.file_ids = file_set.file_ids.reject { |id| extracted_ids.include?(id.to_s) }
+        Hyrax.persister.save(resource: file_set)
+        extracted.each { |file_metadata| Hyrax.persister.delete(resource: file_metadata) }
+        true
+      end
+
       ##
       # Extract text from the derivatives
       #

data/lib/iiif_print/persistence_layer.rb

@@ -117,6 +117,10 @@ module IiifPrint
       def self.pdf_path_for(file_set:)
         raise NotImplementedError, "#{self}.{__method__}"
       end
+
+      def self.remove_extracted_text_for(file_set:)
+        raise NotImplementedError, "#{self}.{__method__}"
+      end
     end
   end
 end

data/lib/iiif_print/text_extraction/page_ocr.rb

@@ -75,10 +75,11 @@ module IiifPrint
 
       private
 
-      # transform the image into a one-bit TIFF for OCR
+      # transform the image into a one-bit TIFF for OCR; a one-bit JP2 is
+      # converted too, since Leptonica cannot read JP2 at one bit per sample
       def preprocess_image
         tool = IiifPrint::ImageTool.new(@path)
-        return if tool.metadata[:color] == 'monochrome'
+        return if tool.metadata[:color] == 'monochrome' && tool.metadata[:content_type] != 'image/jp2'
         intermediate_path = File.join(Dir.mktmpdir, 'monochrome-interim.tif')
         tool.convert(intermediate_path, true)
         @path = intermediate_path

data/lib/iiif_print/text_extraction_derivative_service.rb

@@ -28,13 +28,16 @@ module IiifPrint
       # TODO: Do we need this source_path instance variable?
       @source_path = filename
       ocr = page_ocr_service_class.new(filename)
+      # run all OCR first, so a failed or empty run leaves existing text in place
+      contents = ocr_derivatives.transform_values { |method_name| ocr.public_send(method_name) }
+      return if contents[:txt].blank?
 
-      ocr_derivatives.each do |extension, method_name|
-        path = prepare_path(extension.to_s)
-        content = ocr.public_send(method_name)
+      IiifPrint.remove_extracted_text_for(file_set: file_set)
+
+      contents.each do |extension, content|
         next if content.blank?
 
-        write(content: content, path: path, extension: extension)
+        write(content: content, path: prepare_path(extension.to_s), extension: extension)
       end
     end
 
@@ -47,7 +50,7 @@ module IiifPrint
     end
 
     def cleanup_derivatives(*)
-      ocr_derivatives.keys do |extension|
+      ocr_derivatives.each_key do |extension|
         super(extension.to_s)
       end
     end

data/lib/iiif_print/text_formats_from_alto_service.rb

@@ -18,7 +18,7 @@ module IiifPrint
       # Write data as UTF-8 encoded text
       File.open(save_path, "w:UTF-8") do |f|
         f.write(data)
-        IiifPrint.copy_derivatives_from_data_store(stream: data, directives: { url: file_set.id.to_s, container: 'extracted_text', mime_type: mime_type })
+        IiifPrint.copy_derivatives_from_data_store(stream: data, directives: { url: save_path, container: 'extracted_text', mime_type: mime_type })
       end
     end
 
@@ -28,6 +28,10 @@ module IiifPrint
       !File.size(path).zero?
     end
 
+    def alto_with_words?(path)
+      nonempty_file?(path) && IiifPrint::TextExtraction::AltoReader.new(path).text.present?
+    end
+
     # if there was no derivative yet, there might be one in-transit from
     #   an ingest, so check for that, and use its source if applicable:
     def incoming_alto_path
@@ -35,16 +39,17 @@ module IiifPrint
         fileset_id: @file_set.id,
         destination_name: 'xml'
       ).pluck(:path).uniq.first
-      path if nonempty_file?(path)
+      path if alto_with_words?(path)
     end
 
     def alto_path
-      # check first for existing, non-empty derivative data:
+      # check first for existing derivative data with words; an ALTO without
+      # words (e.g. from a failed OCR run) should not prevent OCR:
       path = derivative_path_factory.derivative_path_for_reference(
         @file_set,
         'xml'
       )
-      return path if nonempty_file?(path)
+      return path if alto_with_words?(path)
       incoming_alto_path
     end
 
@@ -67,6 +72,8 @@ module IiifPrint
         width,
         height
       )
+      IiifPrint.remove_extracted_text_for(file_set: @file_set)
+      save_derivative('xml', source_file)
       save_derivative('json', reader.json)
       save_derivative('txt', reader.text)
     end

data/lib/iiif_print/version.rb

@@ -1,3 +1,3 @@
 module IiifPrint
-  VERSION = '3.2.2'.freeze
+  VERSION = '3.2.3'.freeze
 end

data/lib/iiif_print.rb

@@ -80,6 +80,7 @@ module IiifPrint
       :solr_name,
       :solr_query,
       :pdf_path_for,
+      :remove_extracted_text_for,
       to: :persistence_adapter
     )
   end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: iiif_print
 version: !ruby/object:Gem::Version
-  version: 3.2.2
+  version: 3.2.3
 platform: ruby
 authors:
 - Sean Upton
@@ -11,10 +11,9 @@ authors:
 - Kirk Wang
 - LaRita Robinson
 - Jeremy Friesen
-autorequire:
 bindir: bin
 cert_chain: []
-date: 2026-10-21 00:00:00.000000000 Z
+date: 1980-01-02 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: blacklight_iiif_search
@@ -502,7 +501,6 @@ homepage: https://github.com/notch8/iiif_print/
 licenses:
 - Apache-2.0
 metadata: {}
-post_install_message:
 rdoc_options: []
 require_paths:
 - lib
@@ -517,8 +515,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.3.20
-signing_key:
+rubygems_version: 3.6.7
 specification_version: 4
 summary: IiifPrint is a gem (Rails "engine") for Hyrax-based digital repository applications
   to support displaying parent/child works in the same viewer (Universal Viewer) and