From ad08f2707bfb097c32f60792a9210efaf5de5f18 Mon Sep 17 00:00:00 2001
From: Caio Almeida <117518+caiosba@users.noreply.github.com>
Date: Wed, 4 Sep 2024 10:15:30 -0300
Subject: [PATCH] Include blank media when iterating through items for
 clusterization. (#2021)

This is to restore the behavior of before we pushed the changes for not including blank items by default.

Fixes CV2-5158.
---
 lib/tasks/check_khousheh.rake | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/lib/tasks/check_khousheh.rake b/lib/tasks/check_khousheh.rake
index 098d9f3c0..98f467258 100644
--- a/lib/tasks/check_khousheh.rake
+++ b/lib/tasks/check_khousheh.rake
@@ -20,12 +20,13 @@ namespace :check do
       FileUtils.mkdir_p(File.join(Rails.root, 'tmp', 'feed-clusters-input'))
       started = Time.now.to_i
       sort = [{ annotated_id: { order: :asc } }]
+      all_types = CheckSearch::MEDIA_TYPES + ['blank']
       Feed.find_each do |feed|
         # Only feeds that are sharing media
         if feed.data_points.to_a.include?(2)
           output = { call_id: "#{TIMESTAMP}-#{feed.uuid}", nodes: [], edges: [] }
           Team.current = feed.team
-          query = { feed_id: feed.id, feed_view: 'media', show_similar: true }
+          query = { feed_id: feed.id, feed_view: 'media', show_similar: true, show: all_types }
           es_query = CheckSearch.new(query.to_json).medias_query
           total = CheckSearch.new(query.to_json, nil, feed.team.id).number_of_results
           pages = (total / PER_PAGE.to_f).ceil
@@ -211,7 +212,8 @@ namespace :check do
               end
               # Add items to clusters
               Team.current = feed.team
-              query = { feed_id: feed.id, feed_view: 'media', show_similar: true }
+              all_types = CheckSearch::MEDIA_TYPES + ['blank']
+              query = { feed_id: feed.id, feed_view: 'media', show_similar: true, show: all_types }
               es_query = CheckSearch.new(query.to_json).medias_query
               total = CheckSearch.new(query.to_json, nil, feed.team.id).number_of_results
               pages = (total / PER_PAGE.to_f).ceil