Adds BlacklightIiifSearch to UniversalViewer. (#2296)

* Adds BlacklightIiifSearc to UniversalViewer.

* Correct error.

* Adds and corrects rspec and adds hostname to URL helper.

* adds require.

* Moving classes to a better distributor file.

* Switch concat method.

* Remove invalid UTF-8 text.

* Try a different tactic.

* trying a different service encapsulation.

* Another tactic change.

* add search to config.

* changes activation location.

* puts search service on work level.

* revert changes to config.

* delivers iiif search url the most convenient way.

* rubo

* Adds transcript_text_tesi to default search.

* Improves matching capabilities.

* Try to make highlighting work.

* Sanitizes solr document url.

* match id structure to our manifest.

* Reverts possibly superfluous config.

* Reverts possibly superfluous config part 2.

* Adds documentation and license information.

* adds rspec.

Gemfile

@@ -10,6 +10,7 @@ git_source(:github) do |repo_name|
 end
 
 gem 'archivesspace-client'
+gem 'blacklight_iiif_search'
 gem 'bootsnap', require: false
 gem 'bootstrap-sass', '~> 3.0'
 gem 'bulkrax'

Gemfile.lock

@@ -1412,6 +1412,10 @@ GEM
       bootstrap-sass (~> 3.0)
       openseadragon (>= 0.2.0)
       rails
+    blacklight_iiif_search (1.0.0)
+      blacklight (~> 6.0)
+      iiif-presentation
+      rails (>= 4.2, < 6)
     bootsnap (1.13.0)
       msgpack (~> 1.2)
     bootstrap-sass (3.4.1)
@@ -1792,6 +1796,10 @@ GEM
     ice_nine (0.11.2)
     iiif-image-api (0.2.0)
       activesupport
+    iiif-presentation (1.1.0)
+      activesupport (>= 3.2.18)
+      faraday (>= 0.9)
+      json
     iiif_manifest (1.1.1)
       activesupport (>= 4)
     iso8601 (0.9.1)
@@ -2373,6 +2381,7 @@ PLATFORMS
 DEPENDENCIES
   archivesspace-client
   bixby (~> 3.0.1)
+  blacklight_iiif_search
   bootsnap
   bootstrap-sass (~> 3.0)
   bulkrax

app/controllers/catalog_controller.rb

@@ -17,7 +17,19 @@ def self.modified_field
     solr_name('system_modified', :stored_sortable, type: :date)
   end
 
+  # CatalogController-scope behavior and configuration for BlacklightIiifSearch
+  include BlacklightIiifSearch::Controller
+
   configure_blacklight do |config|
+    # configuration for Blacklight IIIF Content Search
+    config.iiif_search = {
+      full_text_field:       'transcript_text_tesi', # FileSet field
+      object_relation_field: 'is_page_of_ssi', # FileSet field
+      supported_params:      %w[q page],
+      autocomplete_handler:  'iiif_suggest',
+      suggester_name:        'iiifSuggester'
+    }
+
     config.view.gallery.partials = [:index_header, :index]
     config.view.masonry.partials = [:index]
     config.view.slideshow.partials = [:index]
@@ -35,10 +47,12 @@ def self.modified_field
     config.http_method = :post
 
     ## Default parameters to send to solr for all search-like requests. See also SolrHelper#solr_search_params
+    # NOTE: transcript_text_tesi is needed here because the `iiif_search` path utilizes the default `search` qt to
+    #   match terms in Full-Text search-enabled FileSets.
     config.default_solr_params = {
       qt:   "search",
       rows: 10,
-      qf:   "title_tesim description_tesim creator_tesim keyword_tesim"
+      qf:   "title_tesim description_tesim creator_tesim keyword_tesim transcript_text_tesi"
     }
 
     # solr field configuration for document/show views

app/indexers/curate/file_set_indexer.rb

@@ -1,5 +1,4 @@
 # frozen_string_literal: true
-
 module Curate
   class FileSetIndexer < Hyrax::FileSetIndexer
     def generate_solr_document
@@ -93,9 +92,11 @@ def preservation_event_value(preservation_event)
         preservation_event.pluck(:value).first
       end
 
+      # All fields assigned here are utilized by BlacklightIiifSearch.
       def full_text_fields(solr_doc)
-        solr_doc['alto_xml_tesi'] = object.alto_xml if object.alto_xml.present?
+        solr_doc['alto_xml_tesi'] = Curate::TextExtraction::AltoReader.new(object.alto_xml).json if object.alto_xml.present?
         solr_doc['transcript_text_tesi'] = object.transcript_text if object.transcript_text.present?
+        solr_doc['is_page_of_ssi'] = object.parent.id if object.parent.present?
       end
   end
 end

app/lib/curate/text_extraction/alto_reader.rb

@@ -0,0 +1,131 @@
+# frozen_string_literal: true
+require 'active_support/core_ext/module/delegation'
+require 'json'
+require 'nokogiri'
+
+# NOTE: This model is largely derived from IiifPrint's (v3.0.1)
+#   IiifPrint::TextExtraction::AltoReader class. Minor changes have been made to bring
+#   the code into Rubocop compliancy. The IiifPrint Gem application is licensed under the
+#   Apache License 2.0. At the time of adopting this licensed work into this application,
+#   Commercial use, Modification, and Private use were listed under this Gem's Permissions.
+#   The referenced License can be found here:
+#   https://github.com/scientist-softserv/iiif_print/blob/v3.0.1/LICENSE
+module Curate
+  # Module for text extraction
+  module TextExtraction
+    # Class to obtain plain text and JSON word-coordinates from ALTO source
+    class AltoReader
+      attr_accessor :source, :doc_stream
+      delegate :text, to: :doc_stream
+
+      # SAX Document Stream class to gather text and word tokens from ALTO
+      class AltoDocStream < Nokogiri::XML::SAX::Document
+        attr_accessor :text, :words
+
+        def initialize(image_width = nil)
+          super()
+          # scaling matters:
+          @image_width = image_width
+          @scaling = 1.0 # pt to px, if ALTO using points
+          # plain text buffer:
+          @text = ''
+          # list of word hash, containing word+coord:
+          @words = []
+        end
+
+        # Return coordinates from String element attribute hash
+        #
+        # @param attrs [Hash] hash containing ALTO `String` element attributes.
+        # @return [Array] Array of position x, y, width, height in px.
+        def s_coords(attrs)
+          height = scale_value((attrs['HEIGHT'] || 0).to_i)
+          width = scale_value((attrs['WIDTH'] || 0).to_i)
+          hpos = scale_value((attrs['HPOS'] || 0).to_i)
+          vpos = scale_value((attrs['VPOS'] || 0).to_i)
+          [hpos, vpos, width, height]
+        end
+
+        def compute_scaling(attrs)
+          return if @image_width.nil?
+          match = attrs.find { |e| e[0].casecmp?('WIDTH') }
+          return if match.empty?
+          page_width = match[1].to_i
+          return if @image_width == page_width
+          @scaling = page_width / @image_width.to_f
+        end
+
+        def scale_value(v)
+          (v / @scaling).to_i
+        end
+
+        # Callback for element start, implementation of which ignores
+        #   non-String elements.
+        #
+        # @param name [String] element name.
+        # @param attrs [Array] Array of key, value pair Arrays.
+        def start_element(name, attrs = [])
+          values = attrs.to_h
+          compute_scaling(attrs) if name == 'Page'
+          return if name != 'String'
+          token = values['CONTENT']
+          @text += token
+          @words << {
+            word:        token,
+            coordinates: s_coords(values)
+          }
+        end
+
+        # Callback for element end, used here to manage endings of lines and
+        #   blocks.
+        #
+        # @param name [String] element name.
+        def end_element(name)
+          @text += " " if name == 'String'
+          @text += "\n" if name == 'TextBlock'
+          @text += "\n" if name == 'TextLine'
+        end
+
+        # Callback for completion of parsing ALTO, used to normalize generated
+        #   text content (strip unneeded whitespace incidental to output).
+        def end_document
+          # postprocess @text to remove trailing spaces on lines
+          @text = @text.split("\n").map(&:strip).join("\n")
+          # remove trailing whitespace at end of buffer
+          @text.strip!
+        end
+      end
+
+      # Construct with either path
+      #
+      # @param xml [String], and process document
+      def initialize(xml, image_width = nil, image_height = nil)
+        @source = isxml?(xml) ? xml : File.read(xml)
+        @image_width = image_width
+        @image_height = image_height
+        @doc_stream = AltoDocStream.new(image_width)
+        parser = Nokogiri::XML::SAX::Parser.new(doc_stream)
+        parser.parse(@source)
+      end
+
+      # Determine if source parameter is path or xml
+      #
+      # @param xml [String] either path to xml file or xml source
+      # @return [true, false] true if string appears to be XML source, not path
+      def isxml?(xml)
+        xml.lstrip.start_with?('<')
+      end
+
+      # Output JSON flattened word coordinates
+      #
+      # @return [String] JSON serialization of flattened word coordinates
+      def json
+        words = @doc_stream.words
+        Curate::TextExtraction::WordCoordsBuilder.json_coordinates_for(
+          words:  words,
+          width:  @image_width,
+          height: @image_height
+        )
+      end
+    end
+  end
+end

app/lib/curate/text_extraction/word_coords_builder.rb

@@ -0,0 +1,47 @@
+# frozen_string_literal: true
+
+# NOTE: This model is largely derived from IiifPrint's (v3.0.1)
+#   IiifPrint::TextExtraction::WordCoordsBuilder class. Minor changes have been made to bring
+#   the code into Rubocop compliancy. The IiifPrint Gem application is licensed under the
+#   Apache License 2.0. At the time of adopting this licensed work into this application,
+#   Commercial use, Modification, and Private use were listed under this Gem's Permissions.
+#   The referenced License can be found here:
+#   https://github.com/scientist-softserv/iiif_print/blob/v3.0.1/LICENSE
+module Curate
+  # Module for text extraction (OCR or otherwise)
+  module TextExtraction
+    class WordCoordsBuilder
+      # @params words [Array<Hash>] an array of hash objects that have the keys `:word` and `:coordinates`.
+      # @params width [Integer] the width of the "canvas" on which the words appear.
+      # @params height [Integer] the height of the "canvas" on which the words appear.
+      # @return [String] a JSON encoded string.
+      def self.json_coordinates_for(words:, width: nil, height: nil)
+        new(words, width, height).to_json
+      end
+
+      def initialize(words, width = nil, height = nil)
+        @words = words
+        @width = width
+        @height = height
+      end
+
+      # Output JSON flattened word coordinates
+      #
+      # @return [String] JSON serialization of flattened word coordinates
+      def to_json
+        coordinates = {}
+        @words.each do |w|
+          word_chars = w[:word]
+          word_coords = w[:coordinates]
+          if coordinates[word_chars]
+            coordinates[word_chars] << word_coords
+          else
+            coordinates[word_chars] = [word_coords]
+          end
+        end
+        payload = { width: @width, height: @height, coords: coordinates }
+        JSON.generate(payload)
+      end
+    end
+  end
+end

app/models/concerns/blacklight_iiif_search/iiif_search_annotation_behavior.rb

@@ -0,0 +1,77 @@
+# frozen_string_literal: true
+
+# Blacklight IIIF Search v1.0.0 Override: per this application's instructions,
+#   this module must be overridden if coordinates will be provided within the results
+#   of this Gem's search API. It was also necessary to override #annotation_id and
+#   #canvas_uri_for_annotation so that we can match the format of each canvas' @id
+#   url value.
+
+# customizable behavior for IiifSearchAnnotation
+module BlacklightIiifSearch
+  module AnnotationBehavior
+    ##
+    # Create a URL for the annotation
+    # @return [String]
+    def annotation_id
+      "#{emory_iiif_id_url}/canvas/#{document[:id]}/annotation/#{hl_index}"
+    end
+
+    ##
+    # Create a URL for the canvas that the annotation refers to
+    # @return [String]
+    def canvas_uri_for_annotation
+      "#{emory_iiif_id_url}/canvas/#{document[:id]}" + coordinates
+    end
+
+    # NOTE: The methods #coordinates, #fetch_and_parse_coords, and #default_coords below are largely derived
+    #   from IiifPrint's (v3.0.1) IiifPrint::BlacklightIiifSearch::AnnotationDecorator module methods of the
+    #   same name. The methods have been refactored to function according to our expectations. The IiifPrint Gem
+    #   application is licensed under the Apache License 2.0. At the time of adopting this licensed work into
+    #   this application, Commercial use, Modification, and Private use were listed under this Gem's Permissions.
+    #   The referenced License can be found here:
+    #   https://github.com/scientist-softserv/iiif_print/blob/v3.0.1/LICENSE
+
+    ##
+    # return a string like "#xywh=100,100,250,20"
+    # corresponding to coordinates of query term on image
+    # @return [String]
+    def coordinates
+      coords_json = fetch_and_parse_coords
+      return default_coords unless coords_json.present? && coords_json['coords'].present? && query.present?
+
+      query_terms = query.split(' ').map(&:downcase)
+      matches = coords_json['coords'].select do |k, _v|
+        k.downcase =~ /(#{query_terms.join('|')})/
+      end
+      coords_array = matches&.values&.flatten(1)&.[](hl_index)
+
+      coords_array.present? ? "#xywh=#{coords_array.join(',')}" : default_coords
+    end
+
+    private
+
+      ##
+      # a default set of coordinates
+      # @return [String]
+      def default_coords
+        '#xywh=0,0,0,0'
+      end
+
+      ##
+      # return the JSON word-coordinates file contents
+      # @return [JSON]
+      def fetch_and_parse_coords
+        coords = document['alto_xml_tesi']
+        return nil if coords.blank?
+        begin
+          JSON.parse(coords)
+        rescue JSON::ParserError
+          nil
+        end
+      end
+
+      def emory_iiif_id_url
+        "http://#{ENV['HOSTNAME'] || 'localhost:3000'}/iiif/#{parent_document[:id]}/manifest"
+      end
+  end
+end

app/models/file_set.rb

@@ -93,7 +93,7 @@ def alto_xml
   end
 
   def transcript_text
-    transcript_file&.content&.to_s if transcript_file&.file_name&.first&.include?('.txt')
+    transcript_file&.content&.force_encoding('UTF-8') if transcript_file&.file_name&.first&.include?('.txt')
   end
 
   private

app/models/iiif_search_builder.rb

@@ -0,0 +1,17 @@
+# frozen_string_literal: true
+
+# SearchBuilder for full-text searches with highlighting and snippets
+class IiifSearchBuilder < Blacklight::SearchBuilder
+  include Blacklight::Solr::SearchBuilderBehavior
+
+  self.default_processor_chain += [:ocr_search_params]
+
+  # set params for ocr field searching
+  def ocr_search_params(solr_parameters = {})
+    solr_parameters[:facet] = false
+    solr_parameters[:hl] = true
+    solr_parameters[:'hl.fl'] = blacklight_config.iiif_search[:full_text_field]
+    solr_parameters[:'hl.fragsize'] = 100
+    solr_parameters[:'hl.snippets'] = 10
+  end
+end

app/models/solr_document.rb

@@ -162,4 +162,13 @@ def human_readable_visibility
   def source_collection_title
     self['source_collection_title_ssim']
   end
+
+  # Added here since the SolrDocument is easily available within app/views/manifest/manifest.json.jbuilder partial.
+  def work_iiif_search_url
+    return ('http://localhost:3000/catalog/' + self['id'] + '/iiif_search') if ENV['IIIF_SERVER_URL'].blank?
+    parsed_iiif_url = URI.parse(ENV['IIIF_SERVER_URL'])
+    base_path = parsed_iiif_url.to_s[/\A.*(?=#{parsed_iiif_url.path}\z)/]
+
+    base_path + '/catalog/' + self['id'] + '/iiif_search'
+  end
 end