🐛 Correct search count

A funky bug was observed when on a parent work of a PDF, the count was
off by one.  It registers OCR hits as both an OCR hit and a metadata
hit.  This is likely because of adding snippets in the search since all
the file sets' texts need to be indexed on the parent work as well.  The
parent work essentially would double all the texts found.  This commit
is a bit hacky but it removes that extra hit while keeping functionality
for both OCR hits and metadata hits.  This is a  bit of future proofing
since it only would happen in applications with snippets enabled.

app/models/iiif_print/iiif_search_response_decorator.rb

@@ -6,6 +6,10 @@ def annotation_list
  json_results = super
  resources = json_results&.[]('resources')
 
+ resources.delete_if do |resource|
+ resource["on"].include?(IiifPrint::BlacklightIiifSearch::AnnotationDecorator::INVALID_MATCH_TEXT)
+ end
+
  resources&.each do |result_hit|
  next if result_hit['resource'].present?
  result_hit['resource'] = {

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:
@@ -28,16 +29,23 @@ def canvas_uri_for_annotation
  # @return [String]
  def coordinates
  return default_coords if query.blank?
- 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
+ coords_json = fetch_and_parse_coords
+
+ coords_check_result = check_coords_json_and_properties(coords_json, sanitized_query)
+ return coords_check_result if coords_check_result
+
  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
 
@@ -54,6 +62,20 @@ 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 check_coords_json_and_properties(coords_json, sanitized_query)
+ return if coords_json && coords_json['coords']
+
+ properties = @document.keys.select { |key| key.ends_with? "_tesim" }
+ properties.each { |property| return default_coords if @document[property].join.downcase.include?(sanitized_query) }
+
+ INVALID_MATCH_TEXT
+ end
+
  ##
  # a default set of coordinates
  # @return [String]