debianize-mobilizon/lib/service/rich_media/parsers/fallback.ex
Thomas Citharel 2134e7b152
Improve rich media parsers
Signed-off-by: Thomas Citharel <tcit@tcit.fr>
2022-01-18 12:52:45 +01:00

54 lines
1.3 KiB
Elixir

# Portions of this file are derived from Pleroma:
# Pleroma: A lightweight social networking server
# Copyright © 2017-2020 Pleroma Authors <https://pleroma.social/>
# SPDX-License-Identifier: AGPL-3.0-only
defmodule Mobilizon.Service.RichMedia.Parsers.Fallback do
@moduledoc """
Module to parse fallback data in HTML pages (plain old title and meta description)
"""
require Logger
@spec parse(String.t(), map()) :: {:ok, map()} | {:error, String.t()}
def parse(html, data) do
Logger.debug("Running Fallback parser")
data =
data
|> maybe_put(html, :title)
|> maybe_put(html, :description)
if Enum.empty?(data) do
{:error, "Not even a title"}
else
{:ok, data}
end
end
defp maybe_put(meta, html, attr) do
case get_page(html, attr) do
"" -> meta
content -> Map.put_new(meta, attr, content)
end
end
defp get_page(html, :title) do
html
|> Floki.parse_document!()
|> Floki.find("title")
|> List.first()
|> Floki.text()
|> String.trim()
end
defp get_page(html, :description) do
case html
|> Floki.parse_document!()
|> Floki.find("html meta[name='description']")
|> List.first() do
nil -> ""
elem -> elem |> Floki.attribute("content") |> List.first() |> String.trim()
end
end
end