chore: Contact import improvements (CW-1362) (#6747)

Fixes: https://linear.app/chatwoot/issue/CW-1362/csv-imports-are-not-working-properly
Fixes: #3462

---------

Co-authored-by: Sojan <sojan@pepalo.com>

app/jobs/data_import_job.rb

@@ -5,53 +5,123 @@ class DataImportJob < ApplicationJob
   queue_as :low
 
   def perform(data_import)
-    contacts = []
-    data_import.update!(status: :processing)
-    csv = CSV.parse(data_import.import_file.download, headers: true)
-    csv.each { |row| contacts << build_contact(row.to_h.with_indifferent_access, data_import.account) }
-    result = Contact.import contacts, on_duplicate_key_update: :all, batch_size: 1000
-    data_import.update!(status: :completed, processed_records: csv.length - result.failed_instances.length, total_records: csv.length)
+    @data_import = data_import
+    process_import_file
+    send_failed_records_to_admin
   end
 
   private
 
-  def build_contact(params, account)
-    # TODO: rather than doing the find or initialize individually lets fetch objects in bulk and update them in memory
-    contact = init_contact(params, account)
+  def process_import_file
+    @data_import.update!(status: :processing)
 
+    contacts, rejected_contacts = parse_csv_and_build_contacts
+
+    import_contacts(contacts)
+    update_data_import_status(contacts.length, rejected_contacts.length)
+    save_failed_records_csv(rejected_contacts)
+  end
+
+  def parse_csv_and_build_contacts
+    contacts = []
+    rejected_contacts = []
+    csv = CSV.parse(@data_import.import_file.download, headers: true)
+
+    csv.each do |row|
+      current_contact = build_contact(row.to_h.with_indifferent_access, @data_import.account)
+      if current_contact.valid?
+        contacts << current_contact
+      else
+        row['errors'] = current_contact.errors.full_messages.join(', ')
+        rejected_contacts << row
+      end
+    end
+
+    [contacts, rejected_contacts]
+  end
+
+  def import_contacts(contacts)
+    # <struct ActiveRecord::Import::Result failed_instances=[], num_inserts=1, ids=[444, 445], results=[]>
+    Contact.import(contacts, synchronize: contacts, on_duplicate_key_ignore: true, track_validation_failures: true, validate: true, batch_size: 1000)
+  end
+
+  def update_data_import_status(processed_records, rejected_records)
+    @data_import.update!(status: :completed, processed_records: processed_records, total_records: processed_records + rejected_records)
+  end
+
+  def build_contact(params, account)
+    contact = find_or_initialize_contact(params, account)
     contact.name = params[:name] if params[:name].present?
+    contact.additional_attributes ||= {}
+    contact.additional_attributes[:company] = params[:company] if params[:company].present?
+    contact.additional_attributes[:city] = params[:city] if params[:city].present?
     contact.assign_attributes(custom_attributes: contact.custom_attributes.merge(params.except(:identifier, :email, :name, :phone_number)))
     contact
   end
 
-  def get_identified_contacts(params, account)
-    identifier_contact = account.contacts.find_by(identifier: params[:identifier]) if params[:identifier]
-    email_contact = account.contacts.find_by(email: params[:email]) if params[:email]
-    phone_number_contact = account.contacts.find_by(phone_number: params[:phone_number]) if params[:phone_number]
-    contact = merge_identified_contact_attributes(params, [identifier_contact, email_contact, phone_number_contact])
-    # intiating the new contact / contact attributes only by ensuring the identifier, email or phone_number duplication errors won't occur
-    contact ||= merge_contact(email_contact, phone_number_contact)
-    contact
-  end
-
-  def merge_contact(email_contact, phone_number_contact)
-    contact ||= email_contact
-    contact ||= phone_number_contact
-    contact
-  end
-
-  def merge_identified_contact_attributes(params, available_contacts)
-    identifier_contact, email_contact, phone_number_contact = available_contacts
-
-    contact = identifier_contact
-    contact&.email = params[:email] if params[:email].present? && email_contact.blank?
-    contact&.phone_number = params[:phone_number] if params[:phone_number].present? && phone_number_contact.blank?
-    contact
-  end
-
-  def init_contact(params, account)
-    contact = get_identified_contacts(params, account)
+  def find_or_initialize_contact(params, account)
+    contact = find_existing_contact(params, account)
     contact ||= account.contacts.new(params.slice(:email, :identifier, :phone_number))
     contact
   end
+
+  def find_existing_contact(params, account)
+    contact = find_contact_by_identifier(params, account)
+    contact ||= find_contact_by_email(params, account)
+    contact ||= find_contact_by_phone_number(params, account)
+
+    update_contact_with_merged_attributes(params, contact) if contact.present? && contact.valid?
+    contact
+  end
+
+  def find_contact_by_identifier(params, account)
+    return unless params[:identifier]
+
+    account.contacts.find_by(identifier: params[:identifier])
+  end
+
+  def find_contact_by_email(params, account)
+    return unless params[:email]
+
+    account.contacts.find_by(email: params[:email])
+  end
+
+  def find_contact_by_phone_number(params, account)
+    return unless params[:phone_number]
+
+    account.contacts.find_by(phone_number: params[:phone_number])
+  end
+
+  def update_contact_with_merged_attributes(params, contact)
+    contact.email = params[:email] if params[:email].present?
+    contact.phone_number = params[:phone_number] if params[:phone_number].present?
+    contact.save
+  end
+
+  def save_failed_records_csv(rejected_contacts)
+    csv_data = generate_csv_data(rejected_contacts)
+
+    return if csv_data.blank?
+
+    @data_import.failed_records.attach(io: StringIO.new(csv_data), filename: "#{Time.zone.today.strftime('%Y%m%d')}_contacts.csv",
+                                       content_type: 'text/csv')
+    send_failed_records_to_admin
+  end
+
+  def generate_csv_data(rejected_contacts)
+    headers = CSV.parse(@data_import.import_file.download, headers: true).headers
+    headers << 'errors'
+    return if rejected_contacts.blank?
+
+    CSV.generate do |csv|
+      csv << headers
+      rejected_contacts.each do |record|
+        csv << record
+      end
+    end
+  end
+
+  def send_failed_records_to_admin
+    AdministratorNotifications::ChannelNotificationsMailer.with(account: @data_import.account).failed_records(@data_import).deliver_later
+  end
 end