git.squeep.com Git - akkoma/blob - lib/pleroma/web/rich_media/parsers/meta_tags_parser.ex

   1 # Pleroma: A lightweight social networking server
   2 # Copyright © 2017-2020 Pleroma Authors <https://pleroma.social/>
   3 # SPDX-License-Identifier: AGPL-3.0-only
   4
   5 defmodule Pleroma.Web.RichMedia.Parsers.MetaTagsParser do
   6   def parse(html, data, prefix, error_message, key_name, value_name \\ "content") do
   7     meta_data =
   8       html
   9       |> get_elements(key_name, prefix)
  10       |> Enum.reduce(data, fn el, acc ->
  11         attributes = normalize_attributes(el, prefix, key_name, value_name)
  12
  13         Map.merge(acc, attributes)
  14       end)
  15       |> maybe_put_title(html)
  16
  17     if Enum.empty?(meta_data) do
  18       {:error, error_message}
  19     else
  20       {:ok, meta_data}
  21     end
  22   end
  23
  24   defp get_elements(html, key_name, prefix) do
  25     html |> Floki.find("meta[#{key_name}^='#{prefix}:']")
  26   end
  27
  28   defp normalize_attributes(html_node, prefix, key_name, value_name) do
  29     {_tag, attributes, _children} = html_node
  30
  31     data =
  32       Enum.into(attributes, %{}, fn {name, value} ->
  33         {name, String.trim_leading(value, "#{prefix}:")}
  34       end)
  35
  36     %{String.to_atom(data[key_name]) => data[value_name]}
  37   end
  38
  39   defp maybe_put_title(%{title: _} = meta, _), do: meta
  40
  41   defp maybe_put_title(meta, html) when meta != %{} do
  42     case get_page_title(html) do
  43       "" -> meta
  44       title -> Map.put_new(meta, :title, title)
  45     end
  46   end
  47
  48   defp maybe_put_title(meta, _), do: meta
  49
  50   defp get_page_title(html) do
  51     Floki.find(html, "html head title") |> List.first() |> Floki.text()
  52   end
  53 end