CV2-6038: add rake task to migrate existing FactCheck (#2221)

* CV2-6038: add rake task to migrate existing FactCheck

* CV2-6038: add ProjectMedia join

app/models/claim_description.rb

@@ -95,7 +95,7 @@ def update_report
       # clear claim_description fields
       data = { 'claim_description_content' => nil, 'claim_description_context' => nil }
       # clear fact-check values
-      data.merge!({ 'fact_check_title' => '', 'fact_check_summary' => '', 'fact_check_url' => '', 'fact_check_languages' => [] }) unless self.fact_check.nil?
+      data.merge!({ 'fact_check_title' => nil, 'fact_check_summary' => nil, 'fact_check_url' => nil, 'fact_check_languages' => [] }) unless self.fact_check.nil?
       self.index_in_elasticsearch(pm.id, data)
     end
   end

app/models/concerns/team_associations.rb

@@ -22,6 +22,7 @@ module TeamAssociations
     has_many :tipline_newsletters
     has_many :tipline_requests, as: :associated
     has_many :explainers, dependent: :destroy
+    has_many :claim_descriptions
     has_many :api_keys
 
     has_annotations

app/models/fact_check.rb

@@ -154,9 +154,9 @@ def update_report
   def article_elasticsearch_data(action = 'create_or_update')
     return if self.project_media.nil? || self.disable_es_callbacks || RequestStore.store[:disable_es_callbacks]
     data = action == 'destroy' ? {
-        'fact_check_title' => '',
-        'fact_check_summary' => '',
-        'fact_check_url' => '',
+        'fact_check_title' => nil,
+        'fact_check_summary' => nil,
+        'fact_check_url' => nil,
         'fact_check_languages' => []
       } : {
         'fact_check_title' => self.title,

lib/tasks/migrate/20230130074014_add_mapping_for_keyword_search_fields.rake

@@ -1,32 +1,38 @@
 namespace :check do
   namespace :migrate do
-    task index_fc_url_and_cd_context_search_fields: :environment do
+    task index_fact_check_fields: :environment do
       # This rake task to index the following fields
-      # 1) claim_description_context
-      # 2) fact_check_url
+      # 1) claim_description [content, context]
+      # 2) fact_check [title, summary, url, language]
       started = Time.now.to_i
       index_alias = CheckElasticSearchModel.get_index_alias
       client = $repository.client
-      last_team_id = Rails.cache.read('check:migrate:index_new_search_fields:team_id') || 0
+      last_team_id = Rails.cache.read('check:migrate:index_fact_check_fields:team_id') || 0
       Team.where('id > ?', last_team_id).find_each do |team|
-        team.project_medias.find_in_batches(:batch_size => 1000) do |pms|
+        team.claim_descriptions.joins(:project_media).find_in_batches(:batch_size => 1000) do |cds|
           es_body = []
-          ids = pms.map(&:id)
-          ProjectMedia.select('project_medias.id as id, fc.url as url, cd.context as context')
+          ids = cds.map(&:id)
+          ClaimDescription.select('claim_descriptions.project_media_id as pm_id, claim_descriptions.description, claim_descriptions.context, fact_checks.*')
           .where(id: ids)
-          .joins("INNER JOIN claim_descriptions cd ON project_medias.id = cd.project_media_id")
-          .joins("INNER JOIN fact_checks fc ON cd.id = fc.claim_description_id")
+          .joins(:fact_check)
           .find_in_batches(:batch_size => 1000) do |items|
             print '.'
             items.each do |item|
-              doc_id = Base64.encode64("ProjectMedia/#{item['id']}")
-              fields = { 'fact_check_url' => item['url'], 'claim_description_context' => item['context'] }
+              doc_id = Base64.encode64("ProjectMedia/#{item['pm_id']}")
+              fields = {
+                'claim_description_content' => item['description'],
+                'claim_description_context' => item['context'],
+                'fact_check_title' => item['title'],
+                'fact_check_summary' => item['summary'],
+                'fact_check_url' => item['url'],
+                'fact_check_languages' => [item['language']]
+              }
               es_body << { update: { _index: index_alias, _id: doc_id, retry_on_conflict: 3, data: { doc: fields } } }
             end
           end
           client.bulk body: es_body unless es_body.blank?
         end
-        Rails.cache.write('check:migrate:index_new_search_fields:team_id', team.id)
+        Rails.cache.write('check:migrate:index_fact_check_fields:team_id', team.id)
       end
       minutes = ((Time.now.to_i - started) / 60).to_i
       puts "[#{Time.now}] Done in #{minutes} minutes."

lib/tasks/migrate/20250205224319_add_mapping_for_explainer_title_field.rake

@@ -17,5 +17,33 @@ namespace :check do
       minutes = ((Time.now.to_i - started) / 60).to_i
       puts "[#{Time.now}] Done in #{minutes} minutes."
     end
+
+    task project_media_initiate_fact_check: :environment do
+      started = Time.now.to_i
+      Team.find_each do |team|
+        params = {
+          claim_description_content:nil,
+          claim_description_context:nil,
+          fact_check_title:nil,
+          fact_check_summary:nil,
+          fact_check_url:nil,
+          fact_check_languages:[]
+        }
+        options = {
+          index: CheckElasticSearchModel.get_index_alias,
+          conflicts: 'proceed',
+          body: {
+            script: {
+              source: "ctx._source.claim_description_content = params.claim_description_content;ctx._source.claim_description_context = params.claim_description_context;ctx._source.fact_check_title=params.fact_check_title;ctx._source.fact_check_summary=params.fact_check_summary;ctx._source.fact_check_url=params.fact_check_url;ctx._source.fact_check_languages=params.fact_check_languages",
+              params: params
+            },
+            query: { term: { team_id: { value: team.id } } }
+          }
+        }
+        $repository.client.update_by_query options
+      end
+      minutes = ((Time.now.to_i - started) / 60).to_i
+      puts "[#{Time.now}] Done in #{minutes} minutes."
+    end
   end
 end