From fa85b6b3b061b1b48a46e6b6d7892f1e27e24a55 Mon Sep 17 00:00:00 2001
From: Mohamed El-Sawy <melsawy.be7eri@gmail.com>
Date: Mon, 9 Sep 2024 11:23:47 +0300
Subject: [PATCH] CSV-5225: ignore nested documents that exceeds
 nested_objects_limit (#2025)

* CSV-5225: ignore nested documents that exceeds nested_objects_limit

* CV2-5225: add tests

* CV2-5225: cleanup

* CV2-5225: apply nested_objects_limit on tags and requests
---
 app/lib/check_elastic_search.rb           |  1 +
 app/models/annotations/tag.rb             |  7 +++++++
 app/models/tipline_request.rb             |  5 +++++
 test/controllers/elastic_search_9_test.rb | 23 ++++++++++++++++++++++-
 4 files changed, 35 insertions(+), 1 deletion(-)

diff --git a/app/lib/check_elastic_search.rb b/app/lib/check_elastic_search.rb
index 00cb5681ab..35268006bb 100644
--- a/app/lib/check_elastic_search.rb
+++ b/app/lib/check_elastic_search.rb
@@ -82,6 +82,7 @@ def get_fresh_value(data)
 
   def add_update_nested_obj(options)
     return if self.disable_es_callbacks || RequestStore.store[:disable_es_callbacks]
+    return if options[:op] == 'create' && self.respond_to?(:hit_nested_objects_limit?) && self.hit_nested_objects_limit?
     model = { klass: self.class.name, id: self.id }
     ElasticSearchWorker.perform_in(1.second, YAML::dump(model), YAML::dump(options), 'create_update_doc_nested')
   end
diff --git a/app/models/annotations/tag.rb b/app/models/annotations/tag.rb
index d365976f03..4755b39d62 100644
--- a/app/models/annotations/tag.rb
+++ b/app/models/annotations/tag.rb
@@ -84,6 +84,13 @@ def self.run_bulk_create_callbacks(ids_json, pmids_json)
     ProjectMedia.where(id: pmids).find_each { |pm| pm.tags_as_sentence }
   end
 
+  def hit_nested_objects_limit?
+    ret = false
+    pm = self.project_media
+    ret = pm.get_annotations(self.annotation_type).count > CheckConfig.get('nested_objects_limit', 10000, :integer) unless pm.nil?
+    ret
+  end
+
   private
 
   def get_tag_text_reference
diff --git a/app/models/tipline_request.rb b/app/models/tipline_request.rb
index 385f64d258..ad0fb75147 100644
--- a/app/models/tipline_request.rb
+++ b/app/models/tipline_request.rb
@@ -64,6 +64,11 @@ def associated_graphql_id
     Base64.encode64("#{self.associated_type}/#{self.associated_id}")
   end
 
+  def hit_nested_objects_limit?
+    associated = self.associated
+    associated.tipline_requests.count > CheckConfig.get('nested_objects_limit', 10000, :integer)
+  end
+
   private
 
   def set_team_and_user
diff --git a/test/controllers/elastic_search_9_test.rb b/test/controllers/elastic_search_9_test.rb
index a015628a52..6493472a94 100644
--- a/test/controllers/elastic_search_9_test.rb
+++ b/test/controllers/elastic_search_9_test.rb
@@ -206,7 +206,7 @@ def setup
   test "shoud add team filter by default" do
     t = create_team
     t2 = create_team
-    pm1 = create_project_media team: t, quote: 'test',  disable_es_callbacks: false
+    pm1 = create_project_media team: t, quote: 'test', disable_es_callbacks: false
     pm2 = create_project_media team: t2, quote: 'test', disable_es_callbacks: false
     ProjectMedia.where(id: [pm1.id, pm2.id]).update_all(project_id: nil)
     options = {
@@ -230,5 +230,26 @@ def setup
     Team.unstub(:current)
   end
 
+  test "should ignore index document that exceeds nested objects limit" do
+    team = create_team
+    pm = create_project_media team: team
+    stub_configs({ 'nested_objects_limit' => 2 }) do
+      tr = create_tipline_request associated: pm, disable_es_callbacks: false
+      tr2 = create_tipline_request associated: pm, disable_es_callbacks: false
+      tr3 = create_tipline_request associated: pm, disable_es_callbacks: false
+      t = create_tag annotated: pm, disable_es_callbacks: false
+      t2 = create_tag annotated: pm, disable_es_callbacks: false
+      t3 = create_tag annotated: pm, disable_es_callbacks: false
+      sleep 2
+      es = $repository.find(pm.get_es_doc_id)
+      requests = es['requests']
+      assert_equal 2, requests.size
+      assert_equal [tr.id, tr2.id], requests.collect{|r| r['id']}.sort
+      tags = es['tags']
+      assert_equal 2, tags.size
+      assert_equal [t.id, t2.id], tags.collect{|i| i['id']}.sort
+    end
+  end
+
   # Please add new tests to test/controllers/elastic_search_10_test.rb
 end