akkoma/lib/pleroma/web/rich_media/parser.ex

135 lines
3.5 KiB
Elixir
Raw Normal View History

# Pleroma: A lightweight social networking server
# Copyright © 2017-2020 Pleroma Authors <https://pleroma.social/>
# SPDX-License-Identifier: AGPL-3.0-only
2019-01-01 20:26:40 +00:00
defmodule Pleroma.Web.RichMedia.Parser do
2020-09-01 16:12:45 +00:00
require Logger
2019-07-11 13:04:42 +00:00
defp parsers do
Pleroma.Config.get([:rich_media, :parsers])
end
def parse(nil), do: {:error, "No URL provided"}
if Pleroma.Config.get(:env) == :test do
2020-09-01 16:12:45 +00:00
@spec parse(String.t()) :: {:ok, map()} | {:error, any()}
def parse(url), do: parse_url(url)
else
2020-09-01 16:12:45 +00:00
@spec parse(String.t()) :: {:ok, map()} | {:error, any()}
def parse(url) do
2020-09-02 11:21:28 +00:00
with {:ok, data} <- get_cached_or_parse(url),
{:ok, _} <- set_ttl_based_on_image(data, url) do
{:ok, data}
else
{:error, {:invalid_metadata, data}} = e ->
Logger.debug(fn -> "Incomplete or invalid metadata for #{url}: #{inspect(data)}" end)
e
2020-09-02 11:21:28 +00:00
error ->
Logger.error(fn -> "Rich media error for #{url}: #{inspect(error)}" end)
error
2020-09-02 11:21:28 +00:00
end
end
defp get_cached_or_parse(url) do
case Cachex.fetch!(:rich_media_cache, url, fn _ -> {:commit, parse_url(url)} end) do
{:ok, _data} = res ->
res
{:error, _} = e ->
ttl = Pleroma.Config.get([:rich_media, :failure_backoff], 60_000)
Cachex.expire(:rich_media_cache, url, ttl)
e
end
end
end
2019-01-01 20:26:40 +00:00
@doc """
Set the rich media cache based on the expiration time of image.
Adopt behaviour `Pleroma.Web.RichMedia.Parser.TTL`
## Example
defmodule MyModule do
@behaviour Pleroma.Web.RichMedia.Parser.TTL
def ttl(data, url) do
image_url = Map.get(data, :image)
# do some parsing in the url and get the ttl of the image
# and return ttl is unix time
parse_ttl_from_url(image_url)
end
end
Define the module in the config
config :pleroma, :rich_media,
ttl_setters: [MyModule]
"""
2020-09-02 11:21:28 +00:00
@spec set_ttl_based_on_image(map(), String.t()) ::
{:ok, Integer.t() | :noop} | {:error, :no_key}
def set_ttl_based_on_image(data, url) do
case get_ttl_from_image(data, url) do
{:ok, ttl} when is_number(ttl) ->
ttl = ttl * 1000
case Cachex.expire_at(:rich_media_cache, url, ttl) do
{:ok, true} -> {:ok, ttl}
{:ok, false} -> {:error, :no_key}
end
_ ->
2020-09-02 11:21:28 +00:00
{:ok, :noop}
end
end
defp get_ttl_from_image(data, url) do
2020-09-01 16:12:45 +00:00
[:rich_media, :ttl_setters]
|> Pleroma.Config.get()
|> Enum.reduce({:ok, nil}, fn
module, {:ok, _ttl} ->
module.ttl(data, url)
_, error ->
error
end)
end
def parse_url(url) do
2020-09-01 16:12:45 +00:00
with {:ok, %Tesla.Env{body: html}} <- Pleroma.Web.RichMedia.Helpers.rich_media_get(url),
{:ok, html} <- Floki.parse_document(html) do
html
|> maybe_parse()
2020-06-09 17:49:24 +00:00
|> Map.put("url", url)
|> clean_parsed_data()
|> check_parsed_data()
end
2019-01-02 14:02:50 +00:00
end
defp maybe_parse(html) do
2019-07-11 13:04:42 +00:00
Enum.reduce_while(parsers(), %{}, fn parser, acc ->
2019-01-01 20:26:40 +00:00
case parser.parse(html, acc) do
2020-06-11 13:57:31 +00:00
data when data != %{} -> {:halt, data}
_ -> {:cont, acc}
2019-01-01 20:26:40 +00:00
end
end)
end
2019-01-02 14:02:50 +00:00
2020-06-09 17:49:24 +00:00
defp check_parsed_data(%{"title" => title} = data)
when is_binary(title) and title != "" do
{:ok, data}
2019-01-02 14:02:50 +00:00
end
defp check_parsed_data(data) do
{:error, {:invalid_metadata, data}}
2019-01-02 14:02:50 +00:00
end
defp clean_parsed_data(data) do
data
|> Enum.reject(fn {key, val} ->
2020-06-09 17:49:24 +00:00
not match?({:ok, _}, Jason.encode(%{key => val}))
end)
|> Map.new()
end
2019-01-01 20:26:40 +00:00
end