async_batch.rb

# frozen_string_literal: true

# A folder of long PDFs, submitted to the queue and collected as they finish.
#
# Two things make this different from a loop of analyze calls:
#
# 1. Nothing is held open for 60 seconds. A 50-page contract can take minutes, and a
#    synchronous request would be killed at 60 s with a 504 sync_timeout.
# 2. A stable idempotency_key per file means re-running the script after a crash replays
#    the stored responses instead of paying for the batch twice.
#
#   ruby examples/async_batch.rb ./contracts 4

require "date"
require "vision_api"

folder = ARGV[0] || "./contracts"
workers = (ARGV[1] || 4).to_i

vision = VisionAPI.new
batch = Date.today.iso8601 # one idempotency namespace per day

queue = Queue.new
Dir.glob(File.join(folder, "*.pdf")).sort.each { |path| queue << path }
results = Queue.new

# The rate limit is 60 requests/minute per key, and polling counts. Four concurrent files
# at one poll every 2 s is comfortably inside it.
threads = Array.new([workers, queue.size].min) do
  Thread.new do
    while (path = queue.pop(true) rescue nil)
      name = File.basename(path)
      begin
        ref = vision.analyze_async(
          file: path,
          preset: "contract",
          pages: "1-50",
          # Derived from the file, not random: a re-run of the same batch must not re-charge.
          idempotency_key: "#{batch}:#{name}"
        )
        task = vision.wait_for_task(ref["task_id"], poll_interval: 2, max_wait: 20 * 60)
        results << { name: name, credits: task["credits_used"] }
      rescue VisionAPI::TaskFailedError => e
        # A failed task costs 0 credits — the reservation is released in full.
        results << { name: name, error: e.code }
      rescue VisionAPI::APIError => e
        results << { name: name, error: e.code }
      end
    end
  end
end

threads.each(&:join)
results.close

spent = 0
count = 0
until results.empty?
  r = results.pop
  count += 1
  if r[:error]
    puts "✗ #{r[:name]}: #{r[:error]}"
  else
    spent += r[:credits].to_i
    puts "✓ #{r[:name]}: #{r[:credits]} credits"
  end
end

puts "\n#{count} files, #{spent} credits"