Files
cobblemon-ui/lib/cobblemon_ui/tier_list_scraper.ex
Alex Mickelson 37602fc78a
All checks were successful
Build and Deploy / Build & Push Image (push) Successful in 37s
scraper
2026-03-16 20:44:47 -06:00

91 lines
2.7 KiB
Elixir

defmodule CobblemonUi.TierListScraper do
require Logger
@url "https://rankedboost.com/pokemon/tier-list/"
@filename "pokemon_tier_list.json"
def output_file do
dir = System.get_env("CACHE_DIR", ".")
Path.join(dir, @filename)
end
def run do
Logger.info("[TierListScraper] Starting scrape from #{@url}")
with {:ok, html} <- fetch_page(),
{:ok, pokemon} <- parse(html),
:ok <- write_json(pokemon) do
Logger.info("[TierListScraper] Successfully scraped and saved #{length(pokemon)} pokemon")
{:ok, pokemon}
else
{:error, reason} = err ->
Logger.error("[TierListScraper] Scrape failed: #{inspect(reason)}")
err
end
end
def fetch_page do
Logger.debug("[TierListScraper] Fetching #{@url}")
case Req.get(@url, headers: [{"user-agent", "Mozilla/5.0 (compatible; CobblemonUI/1.0)"}]) do
{:ok, %Req.Response{status: 200, body: body}} ->
Logger.debug("[TierListScraper] Fetch OK, body size: #{byte_size(body)} bytes")
{:ok, body}
{:ok, %Req.Response{status: status}} ->
Logger.warning("[TierListScraper] Unexpected HTTP status: #{status}")
{:error, {:http_error, status}}
{:error, err} ->
Logger.error("[TierListScraper] HTTP request failed: #{inspect(err)}")
{:error, err}
end
end
def parse(html) do
nodes = html |> Floki.parse_document!() |> Floki.find(".pokemon-tier")
Logger.debug("[TierListScraper] Found #{length(nodes)} .pokemon-tier nodes")
case nodes do
[] ->
Logger.warning("[TierListScraper] No .pokemon-tier elements found — page structure may have changed or content is JS-rendered")
{:error, :no_pokemon_found}
_ ->
pokemon = Enum.map(nodes, &extract_pokemon/1)
valid = Enum.filter(pokemon, fn %{name: n} -> n != "" end)
Logger.info("[TierListScraper] Parsed #{length(valid)} valid pokemon (#{length(pokemon) - length(valid)} skipped with empty names)")
{:ok, valid}
end
end
defp extract_pokemon(node) do
name =
node
|> Floki.find(".name")
|> Floki.text()
|> String.trim()
tier =
node
|> Floki.find(".tier")
|> Floki.text()
|> String.trim()
%{name: name, tier: tier}
end
def write_json(data) do
json = Jason.encode!(data, pretty: true)
File.write(output_file(), json)
end
# Returns a map of %{lowercase_name => tier} for fast lookup, or an empty map if unavailable.
def load_tier_list do
with {:ok, contents} <- File.read(output_file()),
{:ok, entries} <- Jason.decode(contents) do
Map.new(entries, fn %{"name" => name, "tier" => tier} ->
{String.downcase(name), tier}
end)
else
_ -> %{}
end
end
end