dawarich/app/services/points/raw_data/archiver.rb

185 lines
6 KiB
Ruby
Raw Normal View History

0.36.3 (#2013) * fix: move foreman to global gems to fix startup crash (#1971) * Update exporting code to stream points data to file in batches to red… (#1980) * Update exporting code to stream points data to file in batches to reduce memory usage * Update changelog * Update changelog * Feature/maplibre frontend (#1953) * Add a plan to use MapLibre GL JS for the frontend map rendering, replacing Leaflet * Implement phase 1 * Phases 1-3 + part of 4 * Fix e2e tests * Phase 6 * Implement fog of war * Phase 7 * Next step: fix specs, phase 7 done * Use our own map tiles * Extract v2 map logic to separate manager classes * Update settings panel on v2 map * Update v2 e2e tests structure * Reimplement location search in maps v2 * Update speed routes * Implement visits and places creation in v2 * Fix last failing test * Implement visits merging * Fix a routes e2e test and simplify the routes layer styling. * Extract js to modules from maps_v2_controller.js * Implement area creation * Fix spec problem * Fix some e2e tests * Implement live mode in v2 map * Update icons and panel * Extract some styles * Remove unused file * Start adding dark theme to popups on MapLibre maps * Make popups respect dark theme * Move v2 maps to maplibre namespace * Update v2 references to maplibre * Put place, area and visit info into side panel * Update API to use safe settings config method * Fix specs * Fix method name to config in SafeSettings and update usages accordingly * Add missing public files * Add handling for real time points * Fix remembering enabled/disabled layers of the v2 map * Fix lots of e2e tests * Add settings to select map version * Use maps/v2 as main path for MapLibre maps * Update routing * Update live mode * Update maplibre controller * Update changelog * Remove some console.log statements * Pull only necessary data for map v2 points * Feature/raw data archive (#2009) * 0.36.2 (#2007) * fix: move foreman to global gems to fix startup crash (#1971) * Update exporting code to stream points data to file in batches to red… (#1980) * Update exporting code to stream points data to file in batches to reduce memory usage * Update changelog * Update changelog * Feature/maplibre frontend (#1953) * Add a plan to use MapLibre GL JS for the frontend map rendering, replacing Leaflet * Implement phase 1 * Phases 1-3 + part of 4 * Fix e2e tests * Phase 6 * Implement fog of war * Phase 7 * Next step: fix specs, phase 7 done * Use our own map tiles * Extract v2 map logic to separate manager classes * Update settings panel on v2 map * Update v2 e2e tests structure * Reimplement location search in maps v2 * Update speed routes * Implement visits and places creation in v2 * Fix last failing test * Implement visits merging * Fix a routes e2e test and simplify the routes layer styling. * Extract js to modules from maps_v2_controller.js * Implement area creation * Fix spec problem * Fix some e2e tests * Implement live mode in v2 map * Update icons and panel * Extract some styles * Remove unused file * Start adding dark theme to popups on MapLibre maps * Make popups respect dark theme * Move v2 maps to maplibre namespace * Update v2 references to maplibre * Put place, area and visit info into side panel * Update API to use safe settings config method * Fix specs * Fix method name to config in SafeSettings and update usages accordingly * Add missing public files * Add handling for real time points * Fix remembering enabled/disabled layers of the v2 map * Fix lots of e2e tests * Add settings to select map version * Use maps/v2 as main path for MapLibre maps * Update routing * Update live mode * Update maplibre controller * Update changelog * Remove some console.log statements --------- Co-authored-by: Robin Tuszik <mail@robin.gg> * Remove esbuild scripts from package.json * Remove sideEffects field from package.json * Raw data archivation * Add tests * Fix tests * Fix tests * Update ExceptionReporter * Add schedule to run raw data archival job monthly * Change file structure for raw data archival feature * Update changelog and version for raw data archival feature --------- Co-authored-by: Robin Tuszik <mail@robin.gg> * Set raw_data to an empty hash instead of nil when archiving * Fix storage configuration and file extraction * Consider MIN_MINUTES_SPENT_IN_CITY during stats calculation (#2018) * Consider MIN_MINUTES_SPENT_IN_CITY during stats calculation * Remove raw data from visited cities api endpoint * Use user timezone to show dates on maps (#2020) * Fix/pre epoch time (#2019) * Use user timezone to show dates on maps * Limit timestamps to valid range to prevent database errors when users enter pre-epoch dates. * Limit timestamps to valid range to prevent database errors when users enter pre-epoch dates. * Fix tests failing due to new index on stats table * Fix failing specs * Update redis client configuration to support unix socket connection * Update changelog * Fix kml kmz import issues (#2023) * Fix kml kmz import issues * Refactor KML importer to improve readability and maintainability * Implement moving points in map v2 and fix route rendering logic to ma… (#2027) * Implement moving points in map v2 and fix route rendering logic to match map v1. * Fix route spec * fix(maplibre): update date format to ISO 8601 (#2029) * Add verification step to raw data archival process (#2028) * Add verification step to raw data archival process * Add actual verification of raw data archives after creation, and only clear raw_data for verified archives. * Fix failing specs * Eliminate zip-bomb risk * Fix potential memory leak in js * Return .keep files * Use Toast instead of alert for notifications * Add help section to navbar dropdown * Update changelog * Remove raw_data_archival_job * Ensure file is being closed properly after reading in Archivable concern --------- Co-authored-by: Robin Tuszik <mail@robin.gg>
2025-12-14 06:05:59 -05:00
# frozen_string_literal: true
module Points
module RawData
class Archiver
SAFE_ARCHIVE_LAG = 2.months
def initialize
@stats = { processed: 0, archived: 0, failed: 0 }
end
def call
unless archival_enabled?
Rails.logger.info('Raw data archival disabled (ARCHIVE_RAW_DATA != "true")')
return @stats
end
Rails.logger.info('Starting points raw_data archival...')
archivable_months.each do |month_data|
process_month(month_data)
end
Rails.logger.info("Archival complete: #{@stats}")
@stats
end
def archive_specific_month(user_id, year, month)
month_data = {
'user_id' => user_id,
'year' => year,
'month' => month
}
process_month(month_data)
end
private
def archival_enabled?
ENV['ARCHIVE_RAW_DATA'] == 'true'
end
def archivable_months
# Only months 2+ months old with unarchived points
safe_cutoff = Date.current.beginning_of_month - SAFE_ARCHIVE_LAG
# Use raw SQL to avoid GROUP BY issues with ActiveRecord
# Use AT TIME ZONE 'UTC' to ensure consistent timezone handling
sql = <<-SQL.squish
SELECT user_id,
EXTRACT(YEAR FROM (to_timestamp(timestamp) AT TIME ZONE 'UTC'))::int as year,
EXTRACT(MONTH FROM (to_timestamp(timestamp) AT TIME ZONE 'UTC'))::int as month,
COUNT(*) as unarchived_count
FROM points
WHERE raw_data_archived = false
AND raw_data IS NOT NULL
AND raw_data != '{}'
AND to_timestamp(timestamp) < ?
GROUP BY user_id,
EXTRACT(YEAR FROM (to_timestamp(timestamp) AT TIME ZONE 'UTC')),
EXTRACT(MONTH FROM (to_timestamp(timestamp) AT TIME ZONE 'UTC'))
SQL
ActiveRecord::Base.connection.exec_query(
ActiveRecord::Base.sanitize_sql_array([sql, safe_cutoff])
)
end
def process_month(month_data)
user_id = month_data['user_id']
year = month_data['year']
month = month_data['month']
lock_key = "archive_points:#{user_id}:#{year}:#{month}"
# Advisory lock prevents duplicate processing
# Returns false if lock couldn't be acquired (already locked)
lock_acquired = ActiveRecord::Base.with_advisory_lock(lock_key, timeout_seconds: 0) do
archive_month(user_id, year, month)
@stats[:processed] += 1
true
end
Rails.logger.info("Skipping #{lock_key} - already locked") unless lock_acquired
rescue StandardError => e
ExceptionReporter.call(e, "Failed to archive points for user #{user_id}, #{year}-#{month}")
@stats[:failed] += 1
end
def archive_month(user_id, year, month)
points = find_archivable_points(user_id, year, month)
return if points.empty?
point_ids = points.pluck(:id)
log_archival_start(user_id, year, month, point_ids.count)
archive = create_archive_chunk(user_id, year, month, points, point_ids)
mark_points_as_archived(point_ids, archive.id)
update_stats(point_ids.count)
log_archival_success(archive)
end
def find_archivable_points(user_id, year, month)
timestamp_range = month_timestamp_range(year, month)
Point.where(user_id: user_id, raw_data_archived: false)
.where(timestamp: timestamp_range)
.where.not(raw_data: nil)
.where.not(raw_data: '{}')
end
def month_timestamp_range(year, month)
start_of_month = Time.utc(year, month, 1).to_i
end_of_month = (Time.utc(year, month, 1) + 1.month).to_i
start_of_month...end_of_month
end
def mark_points_as_archived(point_ids, archive_id)
Point.transaction do
Point.where(id: point_ids).update_all(
raw_data_archived: true,
raw_data_archive_id: archive_id
)
end
end
def update_stats(archived_count)
@stats[:archived] += archived_count
end
def log_archival_start(user_id, year, month, count)
Rails.logger.info("Archiving #{count} points for user #{user_id}, #{year}-#{format('%02d', month)}")
end
def log_archival_success(archive)
Rails.logger.info("✓ Archived chunk #{archive.chunk_number} (#{archive.size_mb} MB)")
end
def create_archive_chunk(user_id, year, month, points, point_ids)
# Determine chunk number (append-only)
chunk_number = Points::RawDataArchive
.where(user_id: user_id, year: year, month: month)
.maximum(:chunk_number).to_i + 1
# Compress points data
compressed_data = Points::RawData::ChunkCompressor.new(points).compress
# Create archive record
archive = Points::RawDataArchive.create!(
user_id: user_id,
year: year,
month: month,
chunk_number: chunk_number,
point_count: point_ids.count,
point_ids_checksum: calculate_checksum(point_ids),
archived_at: Time.current,
metadata: {
format_version: 1,
compression: 'gzip',
archived_by: 'Points::RawData::Archiver'
}
)
# Attach compressed file via ActiveStorage
# Uses directory structure: raw_data_archives/:user_id/:year/:month/:chunk.jsonl.gz
# The key parameter controls the actual storage path
archive.file.attach(
io: StringIO.new(compressed_data),
filename: "#{format('%03d', chunk_number)}.jsonl.gz",
content_type: 'application/gzip',
key: "raw_data_archives/#{user_id}/#{year}/#{format('%02d', month)}/#{format('%03d', chunk_number)}.jsonl.gz"
)
archive
end
def calculate_checksum(point_ids)
Digest::SHA256.hexdigest(point_ids.sort.join(','))
end
end
end
end