debshots/lib/tasks/vectorize_packages.rake
Christoph Haas 2b79470feb Simplify vectorization task to sequential processing
Drop the threaded worker pool (queue, mutexes, connection pooling).
The task now processes packages one by one via find_each - plenty
fast since the external embedding service dominates latency, and
much easier to reason about.
2026-08-23 00:13:50 +02:00

53 lines
1.7 KiB
Ruby

# frozen_string_literal: true
# Long rake task bodies are normal; the block-length cop is meant for
# application code.
# rubocop:disable Metrics/BlockLength
namespace :debshots do
desc 'Compute embedding vectors for all packages based on their description ' \
'and long description (missing ones only unless FORCE=1)'
task compute_vectors: :environment do
logger = Logger.new($stdout)
logger.level = Logger::INFO
force = ENV['FORCE'].present?
scope = force ? Package.all : Package.where(embedding: nil)
total = scope.count
if total.zero?
logger.info 'No packages need vector computation.'
next
end
logger.info "Computing vectors for #{total} packages " \
"(#{force ? 'recomputing' : 'missing only'}, " \
"service: #{Vectorizer.service_url})"
done = 0
failed = 0
scope.find_each do |package|
begin
raise ArgumentError, 'no description text' if package.embedding_text.blank?
package.update_embedding!
done += 1
rescue StandardError => e
# A single broken package must never stop the whole run - log
# enough detail to diagnose it and move on.
failed += 1
logger.error "Failed for package #{package.name}: #{e.class}: #{e.message}"
logger.error e.backtrace.first(3).join("\n") if e.backtrace
end
processed = done + failed
next unless (processed % 100).zero? || processed == total
percent = (processed.to_f / total * 100).round(1)
logger.info "Progress: #{processed}/#{total} (#{percent}%)"
end
logger.info "Done. #{done} vectors computed, #{failed} failures."
end
end
# rubocop:enable Metrics/BlockLength