ammocpr/lib/ammoprices/scraping/scrape_job.ex
Graham McIntire 77ff082ebb
Switch to self-scheduling scrapes with longer intervals
Replace cron-based scheduling (every 4h) with self-scheduling jobs
that randomly pick 7-12 hours between runs. Also increase the
inter-request delay from 5-15 minutes to 10-20 minutes.
2026-03-12 13:26:46 -05:00

56 lines
1.5 KiB
Elixir

defmodule Ammoprices.Scraping.ScrapeJob do
@moduledoc false
use Oban.Worker,
queue: :scraping,
max_attempts: 3,
unique: [period: :infinity, states: [:available, :scheduled, :executing, :retryable]]
alias Ammoprices.Catalog
alias Ammoprices.Scraping.Runner
@scrapers [
Ammoprices.Scraping.Retailers.LuckyGunner,
Ammoprices.Scraping.Retailers.SgAmmo,
Ammoprices.Scraping.Retailers.TargetSportsUsa,
Ammoprices.Scraping.Retailers.TrueShotAmmo,
Ammoprices.Scraping.Retailers.PalmettoStateArmory,
Ammoprices.Scraping.Retailers.BulkAmmo,
Ammoprices.Scraping.Retailers.Natchez
]
@impl Oban.Worker
def perform(_job) do
calibers = Catalog.list_calibers()
for scraper <- @scrapers, caliber <- calibers do
scrape_delay()
Runner.run(scraper, caliber)
end
Phoenix.PubSub.broadcast(Ammoprices.PubSub, "prices:updated", {:prices_updated, %{}})
schedule_next()
:ok
end
@doc false
def schedule_next do
{min, max} = Application.get_env(:ammoprices, :scrape_interval_seconds, {25_200, 43_200})
delay_seconds = Enum.random(min..max)
%{}
|> new(scheduled_at: DateTime.add(DateTime.utc_now(), delay_seconds, :second))
|> Oban.insert()
end
# Random delay between requests: 10-20 minutes in prod, 0 in test
defp scrape_delay do
{min, max} = Application.get_env(:ammoprices, :scrape_delay_ms, {600_000, 1_200_000})
delay = Enum.random(min..max)
if delay > 0 do
Process.sleep(delay)
end
end
end