From fd7467e63b981dea5fac2badf03b63dcdd776aca Mon Sep 17 00:00:00 2001
From: Eugene Burmakin <iamfrey@gmail.com>
Date: Sun, 7 Dec 2025 14:02:48 +0100
Subject: [PATCH] Add schedule to run raw data archival job monthly

---
 app/jobs/family/invitations/cleanup_job.rb     |  2 ++
 app/jobs/points/raw_data/archive_job.rb        |  2 ++
 app/services/points/raw_data/archiver.rb       |  4 ++--
 config/schedule.yml                            |  5 +++++
 spec/services/points/raw_data/archiver_spec.rb | 13 ++++++++-----
 5 files changed, 19 insertions(+), 7 deletions(-)

diff --git a/app/jobs/family/invitations/cleanup_job.rb b/app/jobs/family/invitations/cleanup_job.rb
index a80ad443..7100938a 100644
--- a/app/jobs/family/invitations/cleanup_job.rb
+++ b/app/jobs/family/invitations/cleanup_job.rb
@@ -4,6 +4,8 @@ class Family::Invitations::CleanupJob < ApplicationJob
   queue_as :families
 
   def perform
+    return unless DawarichSettings.family_feature_enabled?
+
     Rails.logger.info 'Starting family invitations cleanup'
 
     expired_count = Family::Invitation.where(status: :pending)
diff --git a/app/jobs/points/raw_data/archive_job.rb b/app/jobs/points/raw_data/archive_job.rb
index 4ea6dfdc..de788361 100644
--- a/app/jobs/points/raw_data/archive_job.rb
+++ b/app/jobs/points/raw_data/archive_job.rb
@@ -6,6 +6,8 @@ module Points
       queue_as :archival
 
       def perform
+        return unless ENV['ARCHIVE_RAW_DATA'] == 'true'
+
         stats = Points::RawData::Archiver.new.call
 
         Rails.logger.info("Archive job complete: #{stats}")
diff --git a/app/services/points/raw_data/archiver.rb b/app/services/points/raw_data/archiver.rb
index f12d3783..76a13feb 100644
--- a/app/services/points/raw_data/archiver.rb
+++ b/app/services/points/raw_data/archiver.rb
@@ -84,8 +84,8 @@ module Points
 
         Rails.logger.info("Skipping #{lock_key} - already locked") unless lock_acquired
       rescue StandardError => e
-        Rails.logger.error("Archive failed for #{user_id}/#{year}/#{month}: #{e.message}")
-        Sentry.capture_exception(e) if defined?(Sentry)
+        ExceptionReporter.call(e, "Failed to archive points for user #{user_id}, #{year}-#{month}")
+
         @stats[:failed] += 1
       end
 
diff --git a/config/schedule.yml b/config/schedule.yml
index ae920927..11f81d24 100644
--- a/config/schedule.yml
+++ b/config/schedule.yml
@@ -49,3 +49,8 @@ nightly_family_invitations_cleanup_job:
   cron: "30 2 * * *" # every day at 02:30
   class: "Family::Invitations::CleanupJob"
   queue: family
+
+raw_data_archival_job:
+  cron: "0 2 1 * *"  # Monthly on the 1st at 2 AM
+  class: "Points::RawData::ArchiveJob"
+  queue: archival
diff --git a/spec/services/points/raw_data/archiver_spec.rb b/spec/services/points/raw_data/archiver_spec.rb
index 97911b0c..c0bc71e6 100644
--- a/spec/services/points/raw_data/archiver_spec.rb
+++ b/spec/services/points/raw_data/archiver_spec.rb
@@ -7,7 +7,6 @@ RSpec.describe Points::RawData::Archiver do
   let(:archiver) { described_class.new }
 
   before do
-    # Stub broadcasting to avoid ActionCable issues in tests
     allow(PointsChannel).to receive(:broadcast_to)
   end
 
@@ -20,6 +19,7 @@ RSpec.describe Points::RawData::Archiver do
 
       it 'returns early without processing' do
         result = archiver.call
+
         expect(result).to eq({ processed: 0, archived: 0, failed: 0 })
       end
     end
@@ -44,18 +44,20 @@ RSpec.describe Points::RawData::Archiver do
 
       it 'marks points as archived' do
         archiver.call
+
         expect(Point.where(raw_data_archived: true).count).to eq(5)
       end
 
       it 'nullifies raw_data column' do
         archiver.call
-        Point.where(user: user).each do |point|
+        Point.where(user: user).find_each do |point|
           expect(point.raw_data).to be_nil
         end
       end
 
       it 'returns correct stats' do
         result = archiver.call
+
         expect(result[:processed]).to eq(1)
         expect(result[:archived]).to eq(5)
         expect(result[:failed]).to eq(0)
@@ -115,7 +117,8 @@ RSpec.describe Points::RawData::Archiver do
     it 'creates archive with correct metadata' do
       archiver.archive_specific_month(user.id, test_date.year, test_date.month)
 
-      archive = Points::RawDataArchive.last
+      archive = user.raw_data_archives.last
+
       expect(archive.user_id).to eq(user.id)
       expect(archive.year).to eq(test_date.year)
       expect(archive.month).to eq(test_date.month)
@@ -126,7 +129,7 @@ RSpec.describe Points::RawData::Archiver do
     it 'attaches compressed file' do
       archiver.archive_specific_month(user.id, test_date.year, test_date.month)
 
-      archive = Points::RawDataArchive.last
+      archive = user.raw_data_archives.last
       expect(archive.file).to be_attached
       expect(archive.file.filename.to_s).to match(/raw_data_.*_chunk001\.jsonl\.gz/)
     end
@@ -154,7 +157,7 @@ RSpec.describe Points::RawData::Archiver do
 
       # Verify first batch is archived
       june_points_batch1.each(&:reload)
-      expect(june_points_batch1.all? { |p| p.raw_data_archived }).to be true
+      expect(june_points_batch1.all?(&:raw_data_archived)).to be true
 
       # Add more points for same month (retrospective import)
       # Use unique timestamps to avoid uniqueness validation errors