Merge pull request #259 from scientist-softserv/ocr-count-correction

🐛 Correct search count

app/models/iiif_print/iiif_search_response_decorator.rb

@@ -4,16 +4,37 @@ module IiifSearchResponseDecorator
  # @see https://github.com/scientist-softserv/louisville-hyku/commit/67467e5cf9fdb755f54419f17d3c24c87032d0af
  def annotation_list
  json_results = super
- resources = json_results&.[]('resources')
 
- resources&.each do |result_hit|
+ # Check and process invalid hit
+ if json_results&.[]('resources')
+ remove_invalid_hit(json_results)
+ add_metadata_match(json_results)
+ end
+
+ json_results
+ end
+
+ def remove_invalid_hit(json_results)
+ resources = json_results['resources']
+ invalid_hit = resources.detect { |resource| resource["on"].include?(IiifPrint::BlacklightIiifSearch::AnnotationDecorator::INVALID_MATCH_TEXT) }
+ return unless invalid_hit
+
+ # Delete invalid hit from resources, remove first hit (which is from the invalid hit), decrement total within
+ resources.delete(invalid_hit)
+ json_results['hits'].shift
+ json_results['within']['total'] -= 1
+ end
+
+ def add_metadata_match(json_results)
+ json_results['resources'].each do |result_hit|
  next if result_hit['resource'].present?
+
+ # Add resource details if not present
  result_hit['resource'] = {
  "@type": "cnt:ContentAsText",
  "chars": "Metadata match, see sidebar for details"
  }
  end
- json_results
  end
  end
 end

lib/iiif_print/blacklight_iiif_search/annotation_decorator.rb

@@ -2,6 +2,7 @@
 module IiifPrint
  module BlacklightIiifSearch
  module AnnotationDecorator
+ INVALID_MATCH_TEXT = "#xywh=INVALID,INVALID,INVALID,INVALID".freeze
  ##
  # Create a URL for the annotation
  # use a Hyrax-y URL syntax:
@@ -28,19 +29,28 @@ def canvas_uri_for_annotation
  # @return [String]
  def coordinates
  return default_coords if query.blank?
+
+ sanitized_query = sanitize_query
  coords_json = fetch_and_parse_coords
- return default_coords unless coords_json && coords_json['coords']
- sanitized_query = query.match(additional_query_terms_regex)[1].strip
+ return derived_coords_json_and_properties(sanitized_query) unless coords_json && coords_json['coords']
+
  query_terms = sanitized_query.split(' ').map(&:downcase)
+
  matches = coords_json['coords'].select do |k, _v|
  k.downcase =~ /(#{query_terms.join('|')})/
  end
  return default_coords if matches.blank?
+
  coords_array = matches.values.flatten(1)[hl_index]
  return default_coords unless coords_array
+
  "#xywh=#{coords_array.join(',')}"
  end
 
+ def sanitize_query
+ query.match(additional_query_terms_regex)[1].strip
+ end
+
  ##
  # return the JSON word-coordinates file contents
  # @return [JSON]
@@ -54,6 +64,23 @@ def fetch_and_parse_coords
  end
  end
 
+ # This is a bit hacky but it is checking if any of the properties contain the query term
+ # if there are no coords and there is a metadata property match
+ # then we return the default coords
+ # else we insert a invalid match text to be stripped out at a later point
+ # @see IiifPrint::IiifSearchResponseDecorator#annotation_list
+ def derived_coords_json_and_properties(sanitized_query)
+ property = @document.keys.detect do |key|
+ (key.ends_with?("_tesim") || key.ends_with?("_tsim")) && property_includes_sanitized_query?(key, sanitized_query)
+ end
+
+ property ? default_coords : INVALID_MATCH_TEXT
+ end
+
+ def property_includes_sanitized_query?(property, sanitized_query)
+ @document[property].join.downcase.include?(sanitized_query)
+ end
+
  ##
  # a default set of coordinates
  # @return [String]
@@ -97,6 +124,13 @@ def file_set_id
  def additional_query_terms_regex
  /(.*)(?= AND (\(.+\)|\w+)$)/
  end
+
+ ##
+ # @return [IIIF::Presentation::Resource]
+ def text_resource_for_annotation
+ IIIF::Presentation::Resource.new('@type' => 'cnt:ContentAsText',
+ 'chars' => sanitize_query)
+ end
  end
  end
 end