ammocpr/lib/ammoprices/scraping/runner.ex
Graham McIntire ae7e117b23
Prune listings that disappear from retailer scrapes
When a product no longer appears in a successful scrape for its
retailer+caliber, mark it out of stock. After each scrape cycle, hard
delete any product whose last_seen_at is older than 30 days (snapshots
cascade via FK). A zero-result scrape leaves existing listings alone so
a parse failure or transient site glitch can't nuke the catalog.
2026-04-22 15:41:40 -05:00

94 lines
2.6 KiB
Elixir

defmodule Ammoprices.Scraping.Runner do
@moduledoc false
alias Ammoprices.Catalog
alias Ammoprices.Prices
alias Ammoprices.Scraping.HttpClient
def run(scraper, caliber) do
retailer = Catalog.get_retailer_by_slug!(scraper.retailer_slug())
if function_exported?(scraper, :fetch, 2) do
do_fetch(scraper, retailer, caliber)
else
case scraper.category_url(caliber) do
nil ->
{:ok, %{products_count: 0, snapshots_count: 0}}
path ->
url = retailer.base_url <> path
do_scrape(scraper, retailer, caliber, url)
end
end
end
defp do_fetch(scraper, retailer, caliber) do
case scraper.fetch(retailer, caliber) do
{:ok, parsed} ->
process_results(parsed, retailer, caliber)
{:error, reason} ->
{:error, reason}
end
end
defp do_scrape(scraper, retailer, caliber, url) do
case HttpClient.get(url) do
{:ok, %{status: 200, body: body}} ->
parsed = scraper.parse_products(body)
process_results(parsed, retailer, caliber)
{:ok, %{status: status}} ->
{:error, "HTTP #{status}"}
{:error, reason} ->
{:error, reason}
end
end
defp process_results(parsed, retailer, caliber) do
now = DateTime.truncate(DateTime.utc_now(), :second)
results =
Enum.map(parsed, fn product_data ->
product_attrs = %{
title: product_data.title,
url: product_data.url,
brand: product_data.brand,
grain_weight: product_data.grain_weight,
round_count: product_data.round_count,
casing: product_data.casing,
condition: "new",
in_stock: product_data.in_stock,
subsonic: Map.get(product_data, :subsonic, false),
last_seen_at: now
}
case Catalog.upsert_product(retailer.id, caliber.id, product_attrs) do
{:ok, product} ->
snapshot_attrs = %{
price_cents: product_data.price_cents,
price_per_round_cents: product_data.price_per_round_cents,
in_stock: product_data.in_stock,
recorded_at: now
}
{:ok, _snapshot} = Prices.create_snapshot(product.id, snapshot_attrs)
:ok
{:error, _changeset} ->
:error
end
end)
successful = Enum.count(results, &(&1 == :ok))
if successful > 0 do
Catalog.mark_unseen_products_out_of_stock(retailer.id, caliber.id, now)
end
Catalog.update_retailer(retailer, %{last_scraped_at: now})
{:ok, %{products_count: successful, snapshots_count: successful}}
end
end